TRENDS IN THE EXTENT OF DAMAGES FROM NATURAL HAZARDS IN FORESTS IN SERBIA AND THE INFLUENCE OF TEMPERATURE AND PRECIPITATION

: In terms of the visible effects of climate change, in Serbia as well as in other coun tries is present an increase of average annual temperature, reduced amount of precipitation, the occurrence of floods, droughts, forest fires and ice breakage. The aim of the research is to determine the pattern of the trends in the extent of damages from natural hazards in the forests in Serbia, their share in the total forest damage and the impact of temperature and pre cipitation on the extent of damage from natural hazards in the forests during the researched period. A modelling method was used as a basic scientific method, with an application of trend, regression and correlation analysis, methods of reasoning (abstraction and concreti -zation, analysis and synthesis, induction and deduction, etc.), along with other methods and techniques necessary for obtaining all needed elements (processes, absolute growth, growth rate, etc.). Regardless of the lack of significance in all models, the research results indicate that the damages from natural hazards in the forests in Serbia are expanding, with increased share in total forest damages, which might be related to significantly lower temperatures and higher rainfall (snow-breaks, snow uprooting, ice-breaks, ice uprooting, floods, etc.).

Напомена: Истраживање је реализовано у оквиру пројекта "Истраживања климатских промена и њиховог утицаја на животну средину -праћење утицаја, адаптација и ублажавање", подпројекат "Социо-економски развој, ублажавање и адаптација на климатске промене" (ев. бр. 43007, eв. бр. подпројекта 43007/16-III), финансираног од стране Министарства просвете, науке и технолошког развоја и пројекта "Истраживање социо-економских аспеката ризика услед климатских промена у шумским подручјима" (бр. 401-00-1457/2015-10), финансираног од стране Министарства пољопривреде и заштите животне средине -Управе за шуме. Abstract: In terms of the visible effects of climate change, in Serbia as well as in other countries is present an increase of average annual temperature, reduced amount of precipitation, the occurrence of floods, droughts, forest fires and ice breakage. The aim of the research is to determine the pattern of the trends in the extent of damages from natural hazards in the forests in Serbia, their share in the total forest damage and the impact of temperature and precipitation on the extent of damage from natural hazards in the forests during the researched period. A modelling method was used as a basic scientific method, with an application of trend, regression and correlation analysis, methods of reasoning (abstraction and concretization, analysis and synthesis, induction and deduction, etc.), along with other methods and techniques necessary for obtaining all needed elements (processes, absolute growth, growth rate, etc.). Regardless of the lack of significance in all models, the research results indicate that the damages from natural hazards in the forests in Serbia are expanding, with increased share in total forest damages, which might be related to significantly lower temperatures and higher rainfall (snow-breaks, snow uprooting, ice-breaks, ice uprooting, floods, etc.).

INTRODUCTION
Climate changes and their consequences significantly influence the state of the environment and endanger the surroundings in which the hu-man species are evolved over millennia (2011/а, 2014/a). Hence, there is the great interest in explaining these phenomena, understanding the mechanisms of action of all influencing factors and efforts to change the situation and permanently provide conditions for the survival of the human species.
In line with such developments, particular focus is on the forest ecosystems and their functioning in the changed climatic conditions. Thereby, one should know that there is a mutual interaction between climate and forests, whereas climate affects the forest, but almost to the same extent forests shape the climate of the land areas (Vučković, Stajić, 2003, Bolte et al., 2009, Stajić et al., 2014, Sohngen, Tian, 2016. Mentioned influence balance is very important for land life processes, because the forest is positioned somewhere between global climate influence and all other terrestrial creatures, mitigating the effects of precipitation, temperature and air movement (wind). The effects of climate impacts on forests are manifested in the form of storm breaks and uprooting, ice breaks and uprooting, floods, landslides, forest fires and other similar effects of climate change. All these factors act to reduce the number of all forest functions (production, environmental and social), thus jeopardizing the chances of survival of many plant and animal species, including the human race (2011/a, 2012/a, Houghton et al., 1996).
Hazard represents "…a dangerous phenomenon, substance, human activity or condition that may cause loss of life, injury or other health impacts, property damage, loss of livelihoods and services, social and economic disruption, or environmental damage" (2009). According to §45 of the Law on Forests (2010), natural hazard in forests represents "…a significant disturbance to the biological balance and serious damage to forest ecosystems primarily by fires, drying, plant diseases and pests, storm-breaks and storm uprooting, snow-breaks and snow uprooting, floods, torrential flows, landslides and other unforeseen factors over large areas of forests and forest lands" (2010).
Therefore, in recent years many studies related to connections between climate and forests have been initiated, with an aim to find solutions which will hold up the retrograde processes and prevent further deterioration of forests. In Serbia, the changes are also detected, whilst the negative consequences are numerous (environmental and economic). Especially in recent years, there are frequent occurrences of floods, landslides and ice breakage. The consequences are measured in thousands of cubic meters of felled or damaged wood (which has been, forced by circumstances, harvested), which were originally not yet intended for a timber market. Despite the fact that the selling of such wood is bringing financial profit, the absence of all other type of benefits provided by such trees through their physiological processes undoubtedly is categorized as damage (Baković, 2016).
A time is needed to rehabilitate forest damaged in such way. The process might be shorted by great financial means (significantly higher than those gained through sales of uprooted and damaged trees). As one of the first steps in studying these phenomena, it is important to ascertain the direction and intensity of these changes, as well as to identify the factors that cause them. Respectively, this research started by determining trends of the extent of damage from natural hazards, measured by volume of logged timber 1 , and attempts to determine the influence of the two main climate factors (temperature and precipitation) on the damage extent. Trends were conducted at four levels 2 : a) the whole of Serbia (total), b) the Belgrade region, c) the region of Šumadija and Western Serbia, d) the region of Southern and Eastern Serbia. The ratio of damage volume and temperature and precipitation was carried out only for the whole territory of Serbia. Such territorial research framework was influenced by the structure of the collected data (the manner of organization of data collection, systematization and storage in relevant institutions).
The research operates with two time frames -the period covered by the numerical data and the period when it was conducted. The period covered by the data includes information dating from 2010 to 2015. The research was conducted in the period June-August 2016. Preparation and processing of data and writing a report on the conducted research lasted for the next three months.
The aim of the research is to determine the pattern of the trends in the extent of damages from natural hazards in the forests in Serbia, their share in the total forest damage and the impact of temperature and precipitation on the extent of damage from natural hazards in the forests during the researched period. The purpose is to create the basis to undertake appropriate measures and activities to combat the negative effects of climate impacts on forest ecosystems, based on the acquired knowledge. The subject of research are the extent of damage from natural hazards in the forests, the total extent of forest damages, the average annual temperature and average precipitation.
Basic hypotheses can be set as follows: 1. there is a growing trend in the observed variables (extent of damage from natural hazards in forests in Serbia, Belgrade, Šumadija and Western Serbia Southern and Eastern Serbia) during the research period; 2. there is a growing trend in the share of damage from natural hazards in the total forest damage extent; 3. there is a strong dependence of the extent of damage from natural hazards in the forests on the temperature and size of precipitation.

METHOD AND DATA ANALYSIS
In the research were used following methods: a modelling method, trend analysis (Rankоvić, 2009(Rankоvić, , 2012, regression and correlation analysis (Jovičić, 1981, Petz, 1985. In addition, classical methods of reasoning were used (induction and deduction, analysis and synthesis, abstraction and concretization, etc.), along with other methods and techniques needed for obtaining all the necessary elements for drawing conclusions (averages, absolute growth, growth rate, etc.). In assessing the significance of individual elements of obtained regression models, the coefficient of determination (R 2 ), the correlation coefficient (R), t-statistics and F-statistic (tests the significance of the correlation coefficient) have been used. The chosen significance threshold is α=0.05 (the level of error is equal to or less than 5%), as common for this type of research.
In addition, exponential growth rate (I) has been used for the assessment of the pace of the observed variables variation. The exponential growth rate was derived from the parameter with time (t) in exponential regression model.
The basic data (Table 1) are taken from internal documents of Public enterprise "Srbijašume"   2015 (2011/c, 2012/b, 2013, 2014/b, 2015, 2016/b). All data are related to state forests, because of the difficulties in recording the damages in the private sector and fitting them into a consolidated data overview. Data on the average annual air temperature (T) and the average annual sum of precipitation (P), which were measured at 39 weather stations on the territory of Serbia, are taken from the bulletin "Meteorological Yearbook 1 -Climatological Data", annually published by the Republic Hydrometeorological service of Serbia (2011)(2012)(2013)(2014)(2015)(2016). In order to calculate the average for Serbia, the data of meteorological stations were used for obtaining arithmetic mean (Table 2) for the six observed years.

RESULTS
The research results are divided into three parts: a) trends of damages from natural hazards, b) trend of a share of damage from natural hazards in total damages and c) the impact of temperature and precipitation on the extent of damage from natural hazards. In the first case, the regression models of damages from natural hazards are shown, followed by average annual exponential growth rate (I). In the second case is given the regression model of a share of damages from natural hazards in the total forest damage. In the third case, the regression model of dependency of the extent of damage from natural hazards on changes in temperature and precipitation is shown.

Trend in forest damages from natural hazards
Following regression models were obtained by processing the collected data (Figure 1): the extent of forest damages from natural hazards in Serbia (Q S , in m 3 ) Bearing in mind the tests of statistical significance, one can conclude that the correlation coefficients are significant in the case of Serbia (total) and Šumadija and Western Serbia (<5%). The error is only slightly higher in the case of Belgrade (6%), while in the case of South and Eastern Serbia is somewhat higher (15%). All parameters with time are statistically significant (t-test), except in the case of Southern and Eastern Serbia, while the other parameters in all cases are statistically insignificant. This situation is due to relatively short series and a small number of degrees of freedom. It is expected that the future regular monitoring of these elements will enable longer series of information and more precise models.
Interpretation of parameters with time shows that in the average, damage extent has been increased (confirmed hypothesis 1), as follows: There is a high volume of damages from natural hazards, whereas southern and eastern parts of Serbia are largely affected. Around ⅔ (66.8%) of total damage extent is in this region. In this region is the largest average annual exponential growth rate of damages (77% annually), which is not comforting for the upcoming period. Hence, the damage can be considered a real catastrophe for that part of Serbia, but also for the country as a whole.

Trend in share of damages from natural hazards in the total forest damage
Trends in share of forest damages from natural hazards in the total forest damage in Serbia (U, in %) were observed by using the following regression model (Figure 2): U=5.50•t-11,029.95 (1.57) (-1.57) .38 R=0.62 F (6, 4) =2.47 (p=0.19).
The significance of the correlation coefficient and parameters is at a lower level than assumed (errors are higher than 5%, but less than 20%), due to slightly shorter data series. Despite that, it can be stated that the main trend of the share of damages from natural hazards in the total forest damages is increasing (confirmed hypothesis 2), occurring at a rate of 5.5% per year. This indicates that Diagram 1. Regression models of extent of damages from natural hazards Diagram 2. Regression model of share of damage from natural hazards in the total forest damage in Serbia these types of damages should receive more attention than before and that continuous performance of technical monitoring is needed. In such way, information which would serve as a professional basis for taking appropriate measures and activities to combat the negative consequences of natural hazards on forest ecosystems will be acquired.

Influence of temperature and precipitation on the extent of forest damages from natural hazards
The impact of the two most important climate elements (temperature and precipitation), on the extent of forest damages from natural hazards, was observed through the development and analysis of the corresponding regression model. In the regression model, the extent of damages from natural hazards is dependent variable (Q S ), while independent variables are temperature (T), precipitation (P) and time (t). The model looks like the following: Based on obtained elements for assessing the statistical significance, it can be noted that the significance on the error level of 5% is not achieved. This is probably due to the relatively short data series, the number of independent variables and a small number of degrees of freedom. Hence, further monitoring and operation with a higher number of initial data, frequent update of newly formed models, with a greater degree of significance are suggested.
However, from the obtained models it can be seen that the influence of the independent variables is in accordance with a pre-given expectation (confirmed hypothesis 3). With increasing temperature, the extent of damages from natural hazards decreases (in other words, with decreasing temperature share of the damages grows), while with increasing precipitation the extent of damages is increasing. This indicates that the extent of damage caused by precipitation (e.g. flood, snow-breaks, snow uprooting, etc.) in the reporting period has been higher than the extent of damages caused by the high temperatures (e.g. fires). The parameter with time (t) shows that forest damages from natural hazards on annual average have increased by approximately 29,040.45 m 3 . Given the values of insignificant parameters, these amounts should be considered with retention.

DISCUSSION
Previous studies are indicating that changes in temperature and precipitation from climate change might have significant effects on forests over the past century. For example, there is more evidence that the fires are controlled, at least partly, by climate (Sohngen, Tian, 2016). Also, research indicates a future reduction of the forest land values in Europe, because of "…the decline of economically valuable species in the absence of effective countermeasures" (Hanewinkel et al., 2013). Specifically, it was found that by 2100, "… depending on the interest rate and climate scenario applied -this loss varies between 14 and 50% (mean: 28% for an interest rate of 2%) of the present value of forest land in Europe, excluding Russia, and may total several hundred billion Euros" (Hanewinkel et al., 2013).
The average annual exponential growth rate of forest damages from natural hazards determined in this research is higher than 50%. Similarly, in the National Strategy for protection and rescue in emergency situations was noted that the number of natural hazards in Serbia increased considerably in the last 100 years . From 1900 to 1940 100 natural hazards were registered, continuing by the increase in the following years: 650 (1960 in 1970) to 2,000 (1980 to 1990) and 2,800 (1990-2000.) Only in 2007, 258 forest fires were registered, with the burned vegetation area of 33,000 ha, of which 16,000 ha of forests. It is expected that in the future, Serbia will be more often exposed to climate extremes (2011/b). In Europe, from the mid-20 th century, 130 storms per year have been recorded, with, in average, two destructive storms per year, causing considerable forest damages. The most destructive storms in Europe, so far recorded (Lothar and Martin), occurred in 1999. The damage costs were estimated at more than 240 million m 3 of wood in over 15 countries 4 .
The recorded damage had significant economic consequences. A series of storms affected the Scandinavian countries and Central Europe in 2007 and caused damage, which "…insured losses (…) were calculated at €4.5 billion with about €2.4 billion in Germany alone" . In addition to direct damage from the storm, the Germany forestry industry reported a revenue loss of about 1 billion € . After the floods in Serbia in 2014, "… the total recorded financial loss in the Public enterprise "Srbijašume" is estimated at a total of 269,987,832 RSD" (Baković, 2016). These damages are split into three categories: silviculture (forest seedlings in the newly established forests were uprooted or completely destroyed, with the damage estimated at 27,497,080 RSD); nurseries (damage to the seeding and planting material, infrastructure, facilities, equipment, etc., with the damage estimated at 22,912,297 RSD), forest harvesting (damage to forest roads, bridges, manufactured wood products, with recorded damage of 219,578,455 RSD) (Baković, 2016). Forest fires from 2007 caused damage of approx. 40 million €. Only the recovery amounted 24 million €, whilst the indirect damage was not estimated (2011/b).
In relation to the total extent of damages, more than ⅔ occurred in the region of Southern and Eastern Serbia. These results are partly originating from catastrophic ice breaks, which occurred in the region. In fact, in 2014 "…the ice wave that swept across Eastern Serbia (in altitude range of 300 to 1000 meters above sea level) caused the appearance of large ice breaks and uprooting, making the unprecedented damages to public and private property. Forests, orchards, electrical lines, homes, roads (public and (Baković, 2016). According to the Action Plan for the recovery of damaged forests in state and private ownership for 2015-2018 "…the necessary resources for recovery of losses by ice wave in eastern Serbia are estimated to 438,306,043 RSD or € 3,652,550 €" (Baković, 2016).
The research results indicate that the share of damage from natural hazards in total damage in forest ecosystems in Serbia is increasing by approx. 5% annually. Previous studies, which included 30 countries in Europe, concluded a growing trend of damages as a result of natural hazards (forest fires, storms, snow-breaks etc.). Approximately 0.15% of the total timber volume (about 35 mill. m 3 ) per year is damaged by natural hazards (average for the period 1950-2000). By taking into account the total felling in selected countries of around 430 million m 3 , the damage represents around 8.1% of the felling volume. Slightly more than half of the damage (53%) is from storms, 16% is from fires and about 3% of the damage originates from the snow. Overall, "…the total abiotic component is 77%" (Schelhaaset al., 2003). Local, regional, and global changes in temperature and precipitation can affect the occurrence, frequency, duration, scope, and intensity of disturbances in forest ecosystems (Dale et al., 2001;Hanewinkel et al., 2013). This research indicates that one can expect an increase in the damages from natural hazards with a decrease in temperature and an increase in precipitation. Previous studies also indicate that temperatures and rainfall are critical factors influencing the occurrence of natural hazards (Lindner, Rummukainen, 2013). Climate models indicate changes in the frequency, intensity and duration of extreme events, including more hot days and more heavy precipitation events during the 21 st century (Lindner, Rummukainen, 2013). Projected climate changes imply "…not only a change in mean climate parameters, such as temperature or precipitation, but also that such a change will result in changes in the frequency and magnitude of extreme weather events" (Schlyter et al., 2006).
The interconnection between climate conditions and abiotic damage of forest ecosystems (e.g. snow and ice breaks) is important in order to identify the possible negative effects of climate change on forests and determine the most appropriate forest management approach in the future (Kilpeläinen et al., 2010). Activities in forest management can have a noticeable impact on the vulnerability of forest ecosystems to natural hazards. Previous studies have shown that forest management practices can equally contribute to increasing damages from natural hazards as climate change . Some of the alternative ways of management "…reduction in the rotation length of harvesting and adaptation of species composition to those species that are expected to perform better under a changed climate" are leading to "…some reduction in increment, but does not influence the amount of wood removed from the forest" (Schelhaas et al., 2015).

CONCLUSIONS
Based on the obtained regression model, the following can be concluded: • the absolute average increase of forest damages from natural hazards in Serbia (total) is about 17,653.83 m 3 •year -1 , and the average annual exponential growth rate is 56.2%; • the absolute average increase of forest damages from natural hazards in Belgrade is around 32.49 m 3 •year -1 , and the average annual exponential growth rate is 36.4%; • the absolute average increase of forest damages from natural hazards in Šumadija and Western Serbia is around 5,355.94 m 3 •year -1 , and the average annual exponential growth rate is 33.2%; • the absolute average increase of forest damages from natural hazards in Southern and Eastern Serbia is around 11,790.43 m 3 •year -1 , and the average annual exponential growth rate is 77.0%; • the share of damages from natural hazards in the total forest damage in Serbia increases by an average of around 5.5% per year; • it can be expected that the damage will increase with temperature decrease by 1°C for about 63,407.61 m 3 , with an increase in rainfall of 1 mm•m -2 for about 6.6 m 3 , and the impact of other factors (not included in the model) causes an increase of approximately 29,040.45 m 3 •year -1 . Above mentioned information, regardless the lack of significance in all models, indicate increase of forest damages from natural hazards in Serbia, increase of their share in total forest damages and that the increase of these damages can be connected with the lower temperatures and greater precipitation (snow and ice breaks, snow and ice uprooting, floods, etc.). At the same time, the region of Southern and Eastern Serbia can be marked as particularly affected (the largest average annual exponential growth rate of the damages). All subsequent research should focus on this region of Serbia and work on finding possible causes of such a state.
Since all of the above apply to state-owned forests, it would be necessary to include determination, reporting and consolidation of all information related to the damage from natural hazards for private forests. Defining the right methodology for work performance would be needed, whilst for successful implementation of legislative solutions appropriate protocols should be developed, with clear task division for all the actions and actors (to the level of individuals). The existence of such a protocol would provide a uniform and easy collection, systematization and storage of all necessary information, which could be further used for analytical purposes.
Also, in order to obtain a model with higher reliability, while having relatively short data series, it is necessary to implement alternative econometric methods, specifically developed for such cases (Hsiao, 2007). Such methods combine time series and comparative data (panel data analysis). For such an approach, it is necessary to supplement and prepare existing data to be suitable for specific kind of processing.