Tree-Ring Chronology of Pedunculate Oak (Quercus robur) and its Potential for Development of Dendrochronological Research in Croatia

We present the local tree-ring chronology of pedunculate oak (Qercus robur) from Kobiljak near Zagreb, Croatia (16o09’ E, 45o49’ N, 140 m a.s.l.). The chronology is based on 17 trees and is 127 years long and covers the period of 1883-2009. The well replicated part of the residual version of the ARSTAN chronology with SSS>0.80 (interval of 88 years, period 1922-2009) was used for dendroclimatological analysis, which showed that June precipitation has positive and temperature has negative effect on tree-ring variation. Comparison with 40 available oak chronologies from the surrounding countries confi rmed its good teleconnection with 2 local oak chronologies from Austria, 2 from Hungary, and 3 from Slovenia. It also exhibits good heteroconnection, i.e. similarity with chronologies of beech (Fagus sylvatica), from various sites in Slovenia. The similarities can be ascribed to response to common climatic factors. The results indicate that the chronology could be a good reference point for constructing a longer regional chronology in Croatia and surrounding countries, which could be used for different purposes including dating of objects of cultural heritage.


INTRODUCTION 1. UVOD
Numerous wood science laboratories in the world develop dendrochronology, which as a rule includes investigations of tree ring widths and wood structure.Oak (Quercus sp.) is considered the most important wood in European dendrochronology.It is mainly represented by the pedunculate oak (Quercus robur L.) and sessile oak (Q.petraea Liebl.), which cannot be differentiated by their wood anatomy.Despite different ecological requirements of the two species, their treering patterns usually show good agreement.Therefore, they are often treated together as European oak or simply oak (Quercus sp.).
Since the 1990s, when the construction of the fi rst multimillennial oak chronologies was completed (Baillie, 1995), oak dendrochronology has made considerable progress.The longest tree-ring chronology in the world is the oak chronology of the laboratory in Hohenheim reaching back to 8480 BC (Friedrich et al., 2004).There are several other millennial chronologies, which have been constructed all over the western and central Europe (e.g., for review see Haneca et al., 2009).Oak research has been at the same time extended also to areas, which were for a long time considered as less optimal for dendrochronology, like Flanders/ Belgium (Haneca et al., 2006).It has also been extended to the areas east, and southeast of traditional oak research (e.g.Wimmer and Grabner, 1998;Gryaneus, 1996Gryaneus, , 2003 Long tree-ring chronologies have been used for investigating the past and for predicting future changes in the climate and environment (e.g.Friedrichs et al., 2009a, b;Kern et al., 2009;Haupt et al. 2011;Levanič et al., 2011).In addition, they were used for dating of archaeological wood and artefacts from historic constructions and archaeological sites (e.g., for review see Čufar, 2007;Haneca et al., 2009).
In Slovenia, a neighbouring country of Croatia, which has similar climate regimes to some extent (Alpine, Mediterranean and continental), the fi rst local tree ring chronologies of oak were constructed in the 1990s (Čufar and Levanič, 1999), but their tree-ring patterns seemed to have no similarity with oak chronologies from other countries, such as those north of the Alps.Recently, a 548 years long regional oak chronology has been constructed in SE Slovenia, showing a good supra-regional signal refl ected in the radius of ca.500 km, which demonstrated to be climatic in its nature (Čufar et al., 2008a).It enabled reconstruction of climate for the span of the chronology, indicating that hot and dry June conditions limit the growth of oak in the area (Čufar et al., 2008b).Reconstructed years with extremely hot-dry and wet-cool conditions could be confi rmed by the reports in archived documents.Interestingly, the extreme years did not agree with those reconstructed from oak tree-rings in Western Europe (Kelly et al., 2002).The Slovenian oak chronology, which showed good teleconnection with the chronologies of the surrounding countries, has also been successfully used for dating the wood of the objects in Slovenia as well as the objects of the Croatian cultural heritage (Čufar and Šimek, 2008; Čufar et al., 2006, 2008c).
Since the tree-ring chronologies can be considered 'living organisms', it is necessary to work to improve and prolong the existing ones and to construct the new ones, especially in Croatia and neighboring countries where dendrochronological research still needs to be developed.
The objectives of this study are (1) to construct a local oak chronology for the site near Zagreb, Croatia, (2) to show how climatic factors infl uence tree-ring variation of oak in the sampling area, (3) to fi nd out if there exists teleconnection of this chronology with oak chronologies in the surrounding countries (4) and if there exists heteroconnection of Croatian oak with other tree species.All of this would provide useful information to develop a strategy to improve dendrochronological research in Croatia.

MATERIJAL I METODE
culate oak (Quercus robur L.) mixed with big greenweed on the area where the level of ground water is high.The age of trees was estimated to 140-150 years.

Dendrochronological analysis 2.2. Dendrokronološka analiza
For dendrochronological investigations, disks from 17 felled Quercus robur trees (DBH 40 ± 5 cm) were taken at 4 m above ground.The wood was polished and tree-ring widths were measured along the mean diameter, i.e. two radii, to the nearest 0.01 mm using TSAP-Win program (Frank Rinn, Heidelberg, Germany).The tree-ring series were visually and statistically crossdated and compared with each other by calculating the t-values according to Baillie and Pilcher (1973) and coeffi cient of agreement (Gleichläufi gkeit -Glk) (Eckstein and Bauch 1969) using TSAP-Win program.Tree-ring series, two per each tree, were crossdated, and series of 33 radii were found acceptable for further analyses.
Crossdated tree-ring series of individual trees were assembled into a chronology using the program ARSTAN (Holmes 1994).We calculated ARSTAN chronologies, a non-detrended -raw-data, and a detrended residual chronology.

Tree rings and climate 2.3. Godovi i klima
The climatic infl uence on tree growth was studied using the residual version of the ARSTAN chronology (expressed as tree-ring indexes vs. time), for which the original tree-ring width series were standardized in a two-step procedure.First, the long-term trend was removed by fi tting a negative exponential function (regression line) to each tree-ring series.Second, a more fl exible detrending was made by a cubic smoothing spline with a 50 % frequency response of 30 years to further reduce non-climatic variance.Subsequently, autoregressive modelling of the residuals and biweight robust estimation of the mean were applied (Cook and Peters, 1997).
The climatic data (average monthly temperatures and monthly sums of precipitation for the period 1922 to 2009) were obtained from the meteorological station Grič, Zagreb (Figure 1).The station is representative for the sampling area.
The climate/growth relationships were calculated using the program DendroClim2002 (Biondi and Waikul 2004), whereby the residual version of the tree-ring chronology was the dependent variable and the regressors were the monthly mean temperatures and monthly sums of precipitation for each biological year from the previous October to the current September over the time axis from 1922 to 2009.DendroClim2002 uses correlation functions and response functions, which are the most common statistical models used in dendrochronology.The term 'function' indicates a sequence of coeffi cients computed between the tree-ring chronology and monthly climatic variables, which are ordered in time from the previous-year growing season to the current one.In 'correlation' functions, the coeffi cients are univariate estimates of Pearson's product moment correlation, while in 'response' functions, the coeffi cients are multivariate estimates from a principal component regression model (Biondi and Waikul, 2004).The program applies a bootstrap process according to Guiot (1991) to assess the statistical significance of the correlation and response function.
The stability in time of the climate/growth relationships was checked by moving the correlation and response function, calculated for a 60-year time window, over the chronological life span from 1883 to 2009 (Biondi 1997).

Teleconnection and heteroconnection 2.4. Telekonekcija i heterokonekcija
The residual oak chronology of Kobiljak was tested for teleconnection.It was compared with oak chronologies from Austria, Hungary, Slovenia and Serbia (Čufar et al., 2014).For this purpose, we prepared residual chronologies of 40 sites (27 from Austria, 9 from Hungary, 3 from Slovenia and 1 from Serbia) according to the same procedure described above.
In addition, we also tested the residual oak chronology of Kobiljak with 15 available tree-ring chronologies of beech (Fagus sylvatica L.) (Čufar et al. 2008d) prepared according to the same procedure.
Comparison of the chronology of Kobiljak and others was made by calculating the t-values and coeffi cients of agreement (Gleichläufi gkeit -Glk) using TSAP-Win.

Kronologija
The obtained oak chronology of Kobiljak (abbreviation HR1) is based on tree-ring data from 17 trees (33 radii).It is 127 years long and covers the period 1883-2009 (Figure 2), however its optimally replicated part, based on 13 or more trees, extends from 1922 to 2009 and reaches the subsample signal strength (SSS) > 0.80.The statistics of the raw-data and residual chronologies HR1 is given in Table 1.

Climatic signal in chronology 3.2. Klimatski signal u kronologiji
Below-average temperature and above-average precipitation in June, i.e. a cool and moist June, are the most signifi cant factors favoring oak growth on the site (Fig. 3).The correlation (r) between the residual treering chronology and June precipitation is 0.333, while the correlation with June temperature series is -0.325.
In addition, both precipitation and especially temperature showed a consistent stability over a period of about 60 years (Fig. 4).The results of teleconnection of the HR1 chronology with the chronologies from Austria, Hungary, Slovenia and Serbia is given in Table 2.Although the comparison of HR1 was made with 40 residual chronologies prepared according to the same standardiza-
Out of 40 chronologies used for comparison, 7 showed signifi cant similarity (t BP ≥ 4) with HR1 chronology in its well replicated part .The chronologies showing similarity are located NE, N and NW from Kobiljak.The most distant chronology A14 is from Baumgarten in Austria ca.270 km away (as a crow fl ies), and the nearest one is SI1 from the area of Novo mesto in Slovenia ca.60 km away.The highest agreement (t BP = 6.7) was obtained with the U02 chronology of Zamárdi in Hungary, which is ca.230 km away from Kobiljak.The values of coefficients of agreement were in all cases above 64 %.It should be noted that the chronology of the surroundings of Sremska Mitrovica, Serbia, had t-value 3.8 and Glk 69.1, which is just slightly below the significance value.
The similarities among the chronologies could be ascribed to a common positive response to climate, especially June temperatures (Čufar et al., 2008, 2014).Besides the chronologies used in this study, great importance of June temperatures was also identifi ed in Turkey oak (Quercus cerris) on sites in Central Italy (Corona et al., 1995;Romagnoli and Codipietro, 1996) ca.500 km away from Kobiljak, which could indicate that even teleconnection with Italian chronologies might be possible.

Table 3
Crossdating parameters for comparison of the residual chronology of oak from Kobiljak near Zagreb, Croatia (HR1) with local residual chronologies of beech (Fagus sylvatica) from various sites in Slovenia; for the location, compare Figure 6 (parameters: overlapping -Ovl, coeffi cient of agreement -Glk% and t-value -t BP ) Tablica 3. Parametri za usporedbu rezidualne kronologije hrasta iz Kobiljaka pokraj Zagreba, Hrvatska (HR1) s lokalnim rezidualnim kronologijama bukve (Fagus sylvatica) iz različitih staništa u Sloveniji; lokacije usporediti sa slikom 6. (Ovlpreklapanje, Glk -koefi cijent slaganja, t B -t vrijednost.Croatia has rich wooden cultural heritage, which also includes archaeological wood from the distant past.In the past two decades, the Croatian archaeologists contacted wood scientists from the University of Zagreb and Ljubljana to investigate archaeological wood from their excavations.They pointed out that dendrochronological dating of their artifacts was needed. The archaeologists Tatjana Tkalčec and Tajana Sekelj Ivančan from the Institute of Archeology in Zagreb requested analyses and dating of wood from three different sites.The fi rst investigations gave no encouraging results.Poor preservation of archaeological wood and low number of tree-rings were the main obstacles for dendrochronological dating (Čufar et al., 2006).However, instructions of wood scientists as to how to properly collect and prepare the material soon led to fi rst successfully dated timbers in the old town of Varaždin (Čufar and Šimek, 2008) and in Torčec gradić (Čufar et al., 2008c).The timbers from Varaždin were dated to 1415 (terminus post quem) and those from Torčec gradić to 1263 (terminus post quem).In both cases, the Slovenian regional oak chronology was used as well as the over 800-year long regional oak chronology of the laboratory of the University of Natural Resources and Applied Life Sciences in Vienna (Wimmer and Grabner, 1998).Since the Austrian and Slovenian regional chronologies crossdate well (t BP = 9.7) (Čufar et al., 2008a), successful dating of archaeological wood showed that the chronologies from Austria and Slovenia could act as important reference points to de-velop dendrochronology in Croatia.Enhanced co-operation, also including other countries like Hungary, could help to develop a longer regional oak chronology in Croatia and to establish a chronology network in the area south and southeast of the Alps.

ZAKLJUČAK
The investigated wood from living trees of pedunculate oak in Kobiljak near Zagreb enabled us to build a 127 years long chronology spanning the period 1883-2009.The suffi ciently replicated part is 88 years long and spans the period 1922-2009.
Dendroclimatological analysis showed positive effect of June precipitation and negative effect of June temperatures on tree-ring variation.Both signals are stable over time.
Although relatively short, the chronology exhibits good teleconnection, i.e. similarity with other oak chronologies of Austria, Hungary, and Slovenia.The parameters of agreement with the chronology of the surroundings of Sremska Mitrovica, Serbia were just slightly below the signifi cance value.
Surprisingly, the oak chronology of Kobiljak also exhibits good agreement with some beech chronologies from Slovenia, which indicates that it has good potential for heteroconnection, i.e. similarity with chronologies of other species.
Good teleconnection and heteroconnection could be ascribed to a common factor -the climate.