A DENDROCLIMATOLOGICAL ANALYSIS OF FIR (Abies Alba MILL.) GROWTH IN THE BORJA MOUNTAIN AREA OF BOSNIA AND HERZEGOVINA

: This paper presents the results of a dendroclimatological analysis of the growth of fir trees in the Borja mountain area, Bosnia and Herzegovina. The required data were taken in 2 sites, and the sample covered a total of 60 series of radial increment (15 dominant trees per site with two cores). Standard dendrochronological processing and data analyses using the Coffecha and Arstan software were performed. The preliminary 137-years long regional fir chronology for the Borja mountain area was obtained. The obtained results of the correlation analysis between the radial increment index of the defined chronology, on the one hand, and the monthly precipitation and temperature, on the other, showed that higher summer pre cipitation significantly increased the increment of fir, and that the increase in temperature in these months did not not cause a significant increment reduction. However, the results of the FAI drought index application for the stated purposes have shown that more intense droughts in the summer months can cause a significant reduction in the increment of fir in a given area.


A DENDROCLIMATOLOGICAL ANALYSIS OF FIR (Abies Alba MILL.) GROWTH IN THE BORJA MOUNTAIN AREA OF BOSNIA AND HERZEGOVINA INTRODUCTION
Although they give an impression of expressed staticity and invariability, the trees in the forests, suburban and urban zones are subject to constant changes, which are expressed in changes in the dimensions of trees or their elements of growth (diameter, height, volume, crown, etc.). According to S ta j i ć and Vu č ko v i ć (2016), irreversible increase in the size of a tree or some part of it in a certain period of time is called growth (total time period) or increment (growth in the period of 1 year or more, and less than the total age).
In the context of the analysis of changes in the conditions of the environment and the consequences of these changes on people and their surrounding area, the research of the link between climate impact, as a factor of the environment, and growth and the increment of forest trees and forests has a special place. Because of their speci-ficity (exactness of measurement, close connection with climatic and other ecological factors, the ability to collect data and several centuries back, etc.), research of the growth and diameter increment of trees are of particular importance for analyzing and defining the nature of the reaction of trees to changes in the basic conditions for growth (Stajić et al. 2014(Stajić et al. , 2015. The research of diameter increment, its annual, mid-term and long-term variations, as well as the impacts of various environmental factors that determine its flows, is concerned with researchers in the field of forest growth, but also various researchers from related disciplines, such as dendrochronology and dendroclimatology, as well as dendroecology, dendrogeomorphology, etc. (S t a j i ć 2010). According to Vučković et al. (2005) "...dendrochronology was created by the application of dating markant, mostly narrow tree-rings, the so-called characteristic ring, using the chronological order of annual rings typical of species, region, age, and origin... ". S ta j i ć (2010), quoting papers by Schweingruber (1983) and C h e r u b i n i et al. (2004) states that dendrochronology "...is concerned with the study of the growth of trees and its relation to environmental conditions, depending on space and time...". S m i t h and L e w i s (2007) pointed out that dendrochronology is a science that deals with dating and studying annual growth layers (increments) in forest trees and shrubs.
In the framework of dendrochronology, a (sub) discipline called the dendroclimatology is differentiated, which is based on determining the relationship between the structure and the size of radial increment and climatic elements and their variations. Speer (2010) notes that dendroclimatology is part of a wider scientific discipline called paleoclimatology, which, apart from annual rings, uses glacial glossary, deposition of sediments, coral and individual rocks to study the climate, indicating its progressively greater significance due to the actualization and strengthening of global awareness of climate change and their manifestation through the presence of various extreme natural phenomena, with devastating effects on people and their lives.
Fir was the subject of research in a number of dendrochronological and dendroclimate studies on the Balkan peninsula and Europe. In Bosnia and Herzegovina and Serbia, fir has not been much explored in this regard. In Bosnia and Herzegovina, the influence of variation of climatic elements on the growth of fir was investigated in mixed beech, fir and spruce stands in the forest of Lom (C a sta g n e r i et al., 2014), while in dendroclimatic sense fir in Serbia is unexplored. According to preliminary data of the Second Forest Inventory on large areas, the wood volume in high forests is 38.780.852 m 3 or 18% of the total wood volume in the territory of the Republic of Srpska. In Bosnia and Herzegovina, fir is represented in mixed forests of beech and fir and in beech, fir and spruce forests and very rarely in pure forests of fir (B al l ian, Halilović 2016). Fir in the mountainous zone forms forest communities with spruce and sometimes with Scots pine and beech, making it the most important forest community in Bosnia and Herzegovina -a community of beech-fir forests (S tefa n o v i ć 1977, S tefa n o v i ć et al., 1983al., , Beu s 1984.
Bearing in mind that fir is one of the most common and most important species of trees in the Republic of Srpska and Bosnia and Herzegovina, research of the diameter increment of fir trees, as a bioindicator of their vitality and reactions to the change of environmental factors, has an emphasized economic and ecological significance. Therefore, the aim of this paper is to obtain the basic -preliminary knowledge about the dendroclimatological response of fir in 2 sites within the area of Borja mountain and the potential of creating a regional chronology for the given area.

MATERIAL AND METHOD
The area of research is the Borja Mountain in the western part of the Republic of Srpska, Bosnia and Herzegovina (Figure 1). Its highest peak is Velika Runjevica (1,078 m) and is located in the northwestern part of the mountain. Borja is mostly built of magmatic rocks. It is dominated by mixed stands of fir and beech, and clean stands of fir are also present. The presence of three associations was established: Blechno-Abietetum (Ht. 1938) Ht. and Cestar 1967, Galio rotundifolii-abietetum M. Wra b e r 1959 and Rusco hypoglossi-Abietetum Bru j i ć 2004. It should also be noted that the Borja mountain area represents the border area of fir in the north.
Samples (cores) were taken from two stands on Borja Mountain (Figure 1), in which two sample plots -SPs (50 m x 50 m) were established in their preserved and homogeneous parts. On the first sample plot, in the association Rusco hypoglossi-Abietetum Brujić 2004 on the dystric cambisol, the determined number of trees per hectare is 732, the basal area is 45.7 m 2 /ha and the volume is 590.2 m 3 /ha. In the second sample plot, in the association Galio rotundifolii-Abietetum M. Wrab er 1959 on eutric cambisol, the number of trees per hectare is 588, the basal area is 28.7 m 2 /ha and the volume is 273.7 m 3 /ha.
To study the influence of climate on fir increment, Presler's borer was used for drilling 15 dominant trees of fir (30 trees in total) with two cores (total of 60 series of radial increment). The required number of drilling trees is determined based on the recommendations of a number of authors, according to which the minimum sample size for establishing a quality individual chronology is 15 trees, if two cores are taken from each tree (L u c k m a n , 2007, F r i tt s 1976, A c c e tto 1977, L eva n i ch , 1996).
After collection, the necessary samples were forwarded to the laboratory, placed and glued to the wooden carriers. After the glue was dried, the processing of the samples, i.e. the preparation of the core with sand paper (grinding) was initiated, with the aim of making an accurate measurement of the radial increment. The processed samples were scanned using high performance scanners (Epson Perfection V30 Photo). The measurement of the width of radial increment was made using licensed software CooRecorder 7.6 and CDendro 7.6.
Fir chronology by localities was developed based on empirical radial increment data. Checking the correct dating of each increment value, that is, the accuracy of adding the given increment in the "right" calendar year, was carried out by the Crossdating procedure. Crossdating was done visually and using statistical methods within the specialized statistical program "Cofecha" (H o l m e s , 1983, G r i s s i n o -M aye r, 2001), which provides qualitative control of the collected series of the radial increment (tree-rings width) and points to possible problems in dating and measurement (S taj i ć 2010).
Standardization of the radial increment series was made using the Arstan program, by applying smoothing cubic splines. The result of processing in the Arstan program is the radial increment index series, i.e. three versions of the radial increment index chronology (standard -STD, residual -RES and Arstan -ARS chronology) and a comprehensive overview of the results of statistical analysis. In this program, the average chronologies of the sites are formed by averaging the individual series into a single chronology using biweight robust mean procedure, designed to reduce the influence of isolated values outside the data interval (Cook, 1985, C o o k, Ho l mes , 1986. For the comparison with climatic parameters, the ARS chronology of the radial increment index was used. The correlation coefficient (r) and the coefficient of convergence -GLK (H u b e r, 1943; Ec kstei n , Bau ch 1969), which are a measure of the degree of agreement of series sequences or chronologies in the observed interval, were used for estimating the possibility of merging the obtained local increment chronologies into the a regional chronology for the Borja mountain area.
For the analysis of the influence of climatic elements on the radial increment, data from the regional weather station "Doboj" were used for the period from 1951 to 2014. This impact was analyzed by applying the correlation (1) between monthly temperatures and precipitation and the ARS chronology of the radial increment indices and (2) the FAI index of drought and ARS chronology of the radial increment indices.

RESULTS
The basic dendrochronological-statistical parameters of the analysed empirical radial increment values for the SP are shown in Table 1.  The total number of measured radial increments (No) is 1017 (SP1) and 1592 (SP2). The arithmetic mean (Mean) of individual series of empirical increment values is 3.54 mm (SP1) and 1.92 mm (SP2), and standard deviation (SD) is 1.667 mm (SP1) and 0.961 mm (SP2). Serial correlation (SC) between the empirical increment values and the master chronology of radial increment is 0.444 (SP 1) and 0.434 (SP 2). The values of the mean sensitivity (MS) are from the category of "medium" according to the guidelines of G r i s s in o -M aye r (2001) and amounts to 0.228 (SP1) and 0.241 (SP2). The number of "problematic" (flagged) segments" (NPS) with an insufficient degree of agreement between the series (low correlation) in the total number of segments (NS) in series is from 2 to 4 (Table 1).
Diagram 1 shows sample depth by SPs. The results show the sample depth of the longest radial increment series in SP dates back to 1849 (SP2) and to 1870 (SP1).
The averaged empirical radial increment series (Ir) of both SP are presented in Diagram 2. A significantly larger average value of increment in the second half of the 20th century could be seen compared to the previous period. Based on the Diagram 1. The sample depth per SP Diagram 2. The averaged empirical series of increment by SP obtained growth rates and previous knowledge, it can be assumed that significantly smaller radial increment in the first 100 years of the tree's life results from the unfavourable position of the observed trees in the stand in that period (e.g. stagnation stage) rather than from the clear influence of the climate factor.
As a result of the application of the Arstan program, the ARS chronology of the radial increment indices (I) was defined (Diagram 3). This chronology will be further used for comparison with climate parameters, i.e. for analysing the impact of analysed climate parameters on the forest growth.
In the next phase, the analysis of the relationship between the defined chronologies in the SP was initiated, with r and GLK being determined ( Table 2). The r-value between defined chronologies (0.515) is high and statistically significant. The value of the GLK coefficient (0.848) is also statistically significant. Starting from the previously presented analysis, whose results showed a good matching of the defined chronologies of both locations, the prepa-ration of the preliminary regional chronology of fir for the Borja mountain region began. Diagram 4 shows the ARS version of this chronology. The first year of the chronology of the increment was 1849, and last one 2014, which means that the chronology is 166 years (line "a" of Diagram 4). If the beginning of the chronology takes years in which there are at least 4 series of radial increments, then the first year of chronology is 1878, and the length then is 137 years (line "b" of Diagram 4). The average size of the radial increment index is very close to 1, the standard deviation is 0.130, the skewness coefficient is 0.36 and the kurtosis coefficient is 5.058. The mean sensitivity and the serial correlation are 0.138 and -0.106, respectively.
In order to further assess the quality and reliability of the obtained chronology for dendroclimatology researches EPS was calculated. The value of this indicator in the first years and decades of chronology (from 1878 to 1950) is below the lower limit of acceptance and trust chronology, which according to Wi gl ey et al. (1984) is 0.85. However, since the 1950s (since precipitation and temperature data exist), the EPS is higher than 0.85, indicating the possibility of partial use of the above defined chronology for dendroclimatic analysis.
To detect the period of a year during which precipitation and temperature correlated best with the radial increment, correlation analyses were performed between the radial increment indices and precipitation and temperature for August, September and October of the previous year (Avg p, Sep p and Oct p) and April to October of Diagram 3. ARS version of the chronology the current year (Apr t -Oct t). The calculated correlation coefficients between the radial increment index and monthly precipitation data are shown in Table 3. A statistically significant positive correlation between the analyzed variables at the significance level p<0.05 was found for September of the previous year and July of the current year.
The determined correlation coefficients between the radial increment index and monthly temperatures are shown in Table 4. A statistically significant negative correlation between the analyzed variables at significance level p <0.05 was established for September of the previous year.
In order to further analyze the relationship between the climate and fir increment as well as to minimize the effect of multi-correlation of precipitation and temperature, the relationship between the preliminary regional chronology and the FAI index of drought was investigated. The determined coefficient of correlation between the analysed variables for the Borja mountain range is -0.27 and it is statistically significant (p = 0.033).
Diagram 4. Regional chronology of fir for the Borja mountain area

DISCUSSION
Based on the obtained dendrochronological-statistical parameters of the series of radial increment, it can be concluded that the used increments series from the Borja Mountains represent a good material for the development of chronologies per localities. At both locations, the number problematic segments, i.e. segments in which the radial increment series do not have such a pronounced degree of ''coincidence in flows'' are within the allowed range. MS values indicate that the series of increments are "medium-sensitive" to the effect of the environmental factors. The ACI coefficient of row chronology was high, indicating that it is necessary to "clear" the increment size of the current year from the "part of the increment size" caused by the influence of different factors from the previous year(s). By developing the chronology and the procedures of standardization, averaging of increments values and autoregressive modelling, autocorrelation as an adverse effect was almost completely eliminated from the increment values, but the average level of sensitivity was also reduced.
The checking of the level of radial increment agreement into the defined two chronologies by locations was made using the correlation coefficient (r) and the coefficient of convergence (GLK). The obtained values of these coefficients showed that there is a significant matching, or similarity of the defined chronologies of growth between the localities provided the possibility of producing a regional chronology for the Borja mountain area. This defined chronology has a preliminary character, bearing in mind that, as already mentioned, the aim of this paper was to obtain basic knowledge about the dendroclimatic response of fir in two locations within the area of Borja Mountain. In that context, the obtained basic dendrochronological-statistical indicators of preliminary fir chronology suggested that fir trees were able to respond similarly to the variations of climate in this area. According to Bu tl er et al. (2013), EPS is one of the basic indicators of a chronology sample depth quality and a parameter that quantifies the level of common signal among the increment series used to obtain a chronology. The empirically determined and accepted limit value of the EPS coefficient of 0.85 represents the border below which the observed chronology begins to lose a coherent common signal and begins to dominate the signal of individual trees. Therefore, the chronology of radial increment with lower boundary values can be used for dating (for example, in dendroarheological researches), but due to variability, they are not reliable for calibration with climatic data (B ri ffa, Jo n es 1990; S p eer, 2010). The results obtained here signalled that, according to the EPS sizes, the strength of the common signal in the first few decades was unsatisfactory. This can be explained by the nature of this indicator, but also by the fact that it is closely related to sample depth or the number of increment values determined in each calendar year. Namely, sample depth in all series in chronology is not the same There are some series with a quite small number of data in the first portion of chronology, and because of that such series have low calculated cross-correlation between the series in the first decades. However, the EPS values for chronology segments 1950-1990 and 1975-2014 are above 0.85, which coincides with the periods for which climatic data exist. Therefore, this preliminary chronology can be used for preliminary conclusions about the ''nature'' of the relationship between fir growth and climate of the given area. However, new research of climate-fir growth are necessary in this area, which will allow for a more representative sample and increase the strength of the common signal (variance) between the series in chronology.
The results of the conducted climate-growth analysis showed that the increased precipitation in summer months (July) positively resulted in statistically larger fir increment. This stems from a logical fact about the need of each tree species for high water in the summer part of the vegetation period, especially in the hottest month of this period, when precipitation is at least available. The pronounced effect of precipitation is to be expected because fir is a tree species that requires a lot of moisture for growth (D i zd a rev i ć et al., 1987) and does not tolerate dry soil and air (C v j et i ćanin et al., 2016). The average amount of rainfall in the area of the fir distribution range in Bosnia and Herzegovina is 1000 -2000 mm, so that the distribution of fir in a large number of cases is caused by a lack of sufficient moisture.
In addition, a statistically significant correlation was found between the radial increment and precipitation (positive) and temperature (negative) in September of the previous year. This practically means that fir is not able to suffer more temperatures and less precipitation in the late summer of the previous year, which can cause a significant reduction in the increment in the next year. Generally, the influence of temperature variation on fir increment variation is far less pronounced than the influence of precipitation. This is somewhat understandable, bearing in mind that the sample plots are at a relatively low altitude for fir (an average of about 840 m). Starting from a commonly accepted s principle in dendrochronology, the socalled principle of limiting factors (F r i tt s 1976, Cook et al., 1990), it is expected to detect a significant increase in the influence of temperature on fir growing at higher altitudes, where temperature is the dominant limiting growth factor. Bearing in mind that only 12% of the area of pure and mixed forests of fir and spruce and mixed forests of fir, spruce and beech in Bosnia and Herzegovina are at an altitude lower than 800 m (Matić et al., 1971), it is also necessary to conduct such research in the predominant part of fir forests at higher altitudes, which, according to the mentioned authors, are at an average altitude of 1100 m.
In order to eliminate the consequences of the so-called multicolinearity, or mutual inter-correlation between temperature and precipitation (higher temperatures also cause less precipitation and vice versa, which can impede a valid conclusion about the individual (separated) influence of these two climatic elements on increment) for the assessment of the relations between climate and growth drought indexes should be used, in which data on precipitation and temperature are "combined" at the same time. Of the numerous drought indexes used in dendroclimatological studies, the FAI (Forest Aridity Index) drought index was used for this research. This index represents the ratio of average temperatures in July and August, and rainfall from May to August corresponding to the most intensive period of importance for growth and production of organic matter of forest trees (Führer et al., 2011, Moricz et al., 2018. For the research area, the average annual value of the FAI index in the period 1953-2014 is 4.86. The results of the correlation between the FAI drought index and the radial increment indices (r = -0.34) showed that the increase of the drought index causes statistically significant decreases in the radial increment of fir.
The obtained results on the nature of the dendroclimtological reaction of fir in the Borja mountain region coincide with the results of the research by Castagneri et al. (2014), carried out in the area of the Lom virgin forest. These authors, similar to our results, found that higher precipitation in summer months was "very appealing" for fir, and caused a significant increase in increment. Additionally, C astagn eri et al. (2014) specifically pointed out the fact that the observed increase in temperature in the last three decades did not have a reflection on the significant reduction of increment values.

CONCLUSIONS
In order to determine the character of climatic influence on radial increment, a dendroclimatic analysis of fir growth was carried out in the region of Borja mountain in Bosnia and Herzegovina. The results of growth trend analysis and dendrochronological-statistical parameters of radial increment series reveal that the collected and processed sample represents suitable and statistically sufficient quality material for the development of local (master) chronologies of radial increment at the investigated locations (sites). It was determined that the growth increment trends in the defined local chronologies were quite similar, and therefore there was a statistically significant agreement between the established chronologies, which enabled the preparation of a preliminary regional chronology of fir in the Borja mountain area. The defined chronology of radial increment, with sufficient statistical reliability and the strength of a common signal in growth, has a total length of 137 years (1878-2014 years).
The conducted mathematical-statistical analyses of the relationship between the preliminary regional chronology (series of annual tree rings indices) and the monthly sum of precipitation and temperature show that higher amounts of precipitation in late summer of the previous year and during the current year cause a significant increase in the increment of the fir. On the other hand, higher summer temperatures do not result in a significant decrease in the current year's increment. However, this happens only due to the increase in temperature during September of the previous year. Having in mind these results, it can be preliminarily concluded that the eventual decrease in the amount of precipitation caused by the temperature rise in summer months would not cause a significant decrease in the radial growth of fir, suggesting that fir in this area may endure some level of precipitation reduction and temperature increase without a significant reduction in the radial increment. However, the results of the applied correlation analysis between the radial increment index and the FAI index of drought showed that in the case of more severe drought in summer months, fir may experience significant reductions in increment size.
In general, the obtained results suggest that the growth of fir in the Borja mountain area is more dependent on the variation of precipitation than the temperature. This can be partly expected, bearing in mind the relatively low altitude of these two sites where fir grows in comparison with the whole altitudinal range of fir in Bosnia and Herzegovina, but also the fact that it is a species of trees that require a lot of moisture. Since it is necessary to collect a series of preliminary findings for the planning and implementation of serious and multidisciplinary research on the reaction of fir to changes in climatic conditions in the analyzed area, the initially obtained results are very illustrative and practical. Nevertheless, it is necessary to conduct new research of the relation between the climate and the growth of fir in several other locations in this region, as well as to apply a number of other methodological procedures that additionally "illuminate" the nature of the increment reaction to a multidecedal temperature and precipitation changes, as the most important climatic elements.