A pan-European summer teleconnection mode recorded by a new temperature reconstruction from the northeastern Mediterranean ( ad 1768

Abstract

The dominant atmospheric circulation pattern that governs European summer climate is a blocking-like pattern over the British Isles that co-occurs with a low over southeastern Europe. The meridionally oriented configuration of this circulation pattern favours the intrusion of warm air over the northeastern Mediterranean during one mode and over northwestern Europe during its opposite mode. We present a summer temperature reconstruction (1768–2008) for the southeastern node of this teleconnection pattern. This reconstruction is based on maximum latewood density (MXD) measurements of 23 Pinus heldreichii trees from a high-elevation stand in the Pirin Mountains in Bulgaria. The temperature signal in trees from high-elevation, summer-dry regions is stronger in MXD measurements compared with tree-ring width data and our reconstruction reflects well interannual- to decadal-scale summer temperature fluctuations in the Balkans as instrumentally recorded over the twentieth century. Fluctuations in our Bulgarian reconstruction correspond to summer temperature variability in the Balkans, Italy, and the southern Carpathians, but opposing temperature variability patterns are manifested over the British Isles and southern Scandinavia. The strong and consistent anti-phase relationship between our reconstruction and a reconstruction of the summer North Atlantic Oscillation (sNAO) suggests that the sNAO pattern is a main driver of the teleconnection between summer temperatures in southeastern versus northwestern Europe. This teleconnection is most pronounced on interannual timescales and has been stable over the last two centuries.

Keywords

Balkan Bulgaria climate reconstruction climate response dendroclimatology east Atlantic pattern eastern Mediterranean maximum latewood density North Atlantic Oscillation teleconnection tree-ring

Introduction

The contributions of warm temperatures and dry climate during the summer to a high potential evapotranspiration make societies in the Mediterranean region highly dependent on water resources. Current vulnerabilities in the region include agricultural production (Schroter et al., 2005) and fire risk (Moriondo et al., 2006) that depend upon both seasonal climatic extremes driven by synoptic settings as well as low-frequency changes in climate. Predictions of future climate do not forecast significant alleviation of these susceptibilities, and the Mediterranean Basin has accordingly been identified as a climatic change hotspot (Giorgi, 2006): general circulation models predict high rates of future warming (Giorgi and Lionello, 2008; Onol and Semazzi, 2009) and a decrease in precipitation (Gao and Giorgi, 2008; Hertig and Jacobeit, 2008). Summer droughts in particular are predicted to occur more frequently, but may likely be partially compensated by increased winter precipitation (Giannakopoulos et al., 2009; Onol and Semazzi, 2009). Climate extremes have already increased over the 20th century (Kostopoulou and Jones, 2005; Kuglitsch et al., 2010) and are predicted to become even more frequent in this region in the future (Meehl and Tebaldi, 2004). If these projections for future changes in water availability are accurate, the socio-economic consequences will be significant (Chang et al., 2002).

Climatic variability in the Mediterranean region is strongly linked to synoptic-scale atmospheric circulation patterns (Dünkeloh and Jacobeit, 2003; Xoplaki et al., 2003a, 2004). The North Atlantic Oscillation (NAO; Hurrell, 1995), the Arctic Oscillation (AO; Thompson and Wallace, 1998), the Eastern Atlantic pattern (EA; Wallace and Gutzler, 1981) and the Eurasian pattern (EU; Barnston and Livezey, 1987) primarily influence winter precipitation (Cavazos, 2000; Felis and Rimbu, 2010; Founda and van Loon, 2008; Quadrelli et al., 2001; Turkes and Erlat, 2003, 2008), whereas summer air temperatures are modulated by blocking conditions over the Basin and by an east–west geopotential height dipole (Kostopoulou and Jones, 2007; Xoplaki et al., 2003a).

These atmospheric circulation patterns do not only influence climatic variability on seasonal to interannual scales (Onol and Semazzi, 2009; Xoplaki et al., 2004), but also on decadal to centennial timescales (Luterbacher et al., 2010). Thus, to quantify and understand low-frequency trends in natural climate variability, long climatic time-series are necessary. Thermal and orographic forcing (Fernandez et al., 2003) further complicate the Mediterranean climatic system and local-to-regional scale studies of long-term climate variability are therefore needed to adequately and explicitly capture spatio-temporal variability in the Mediterranean Basin.

Long instrumental records and proxy data have contributed to an understanding of long-term climate variation in the Mediterranean region including reconstructions of century-scale climate variability on basin-wide (Luterbacher et al., 2002; Nicault et al., 2008; Pauling et al., 2006) as well as regional scales. Well-replicated tree ring records have been developed for the Pyrenees (Büntgen et al., 2008, 2010; Tardif et al., 2003), northern Africa (Esper et al., 2007; Glueck and Stockton, 2001; Touchan et al., 2008a, 2008b), northeastern Italy (Serre-Bachet, 1994), and the southeastern Mediterranean (Akkemik and Aras, 2005; Griggs et al., 2007; Touchan et al., 2005a, 2005b, 2007). Historical documentary data are also widely available for the Mediterranean region (Camuffo et al., 2010) and have been used to reconstruct climatic conditions in the Iberian Peninsula (Barriendos and Rodrigo, 2006; Dominguez-Castro et al., 2008) and Sicily (Piervitali and Colacino, 2001). Additional proxies such as speleothems (Fleitmann et al., 2009) and isotopic measurements on annually banded reef corals (Felis and Rimbu, 2010) have also provided additional high-resolution information about past climatic information.

While general progress has been made in understanding long-term variations in the Mediterranean Basin, there are particular regions for which reliable proxy records are scarce. The Balkan Peninsula, for instance, is located at the pivot point for the above-mentioned east–west synoptic dipole. The treeline forests that occur at multiple high-elevation locations within the Balkan region are often dominated by centuries-old trees with high dendroclimatological potential (Panayotov and Yurukov, 2007; Panayotov et al., 2010; Seim et al., 2012). Nonetheless, high-resolution proxy climate data for this region are sparse (Vakarelov et al., 2001; Xoplaki et al., 2001). Tree growth at high-elevation Mediterranean sites is typically influenced by a combination of temperature and drought factors and correlations between tree ring width (TRW) measurement series and individual climatic parameters at these sites are often too weak for the use of TRW series as climate proxies (Büntgen et al., 2010; Panayotov et al., 2010). At these sites, maximum latewood density (MXD) measurements can contain a much stronger climatic signal than tree ring width (Büntgen et al., 2008, 2010). It is worth noting that at the driest sites, even at high elevation, only drought variability is recorded in TRW (Esper et al., 2007) and at these sites TRW can be a more successful climate proxy than MXD (Esper et al., 2006).

In this study, we explore the dendroclimatic potential of a Pinus heldreichii (PIHE), also known as Pinus leucodermis Antoine, MXD data set from a high-elevation site in the Pirin Mountains in southwestern Bulgaria. We use this MXD chronology to reconstruct regional summer temperature variability and investigate the atmospheric circulation patterns that drive this variability on interannual to decadal timescales.

Material and methods

Study area

PIHE tree-ring samples were collected in an open forest at a site on the eastern slope of Vihren Peak in the Pirin Mountains of southwestern Bulgaria (41.7N, 23.5E; Figure 1). The study site is located at an elevation of 1950–2150 m a.s.l. in the subalpine zone at the local treeline. Anthropogenic influence has been limited by the steepness of the site slope (30–50°). The PIHE trees grow on Rendzic Leptosols and Regosols formed on marble bedrock.

Figure 1.

Location of the study site and of the meteorological stations.

The Pirin Mountains are located at the transition between the Mediterranean and continental climate zones (Grunewald et al., 2009), with precipitation peaks occurring from late autumn to late spring (Tran et al., 2002). A substantial fraction of the annual precipitation falls as snow in winter and deep snow covers of more than 2 m occur frequently. Average annual temperature (1954–1979) at the Vihren chalet (1970 m a.s.l.) is 3.5°C and average annual precipitation is 1378 mm.

Tree ring analysis

We collected two core samples per tree from 15 living PIHE trees at the study site in summer 2009. Trees ranged in age between 47 and 253 years with a median segment length of 187 years. MXD was measured on individual samples using a WALESCH 2003 x-ray densitometer with 0.01 mm resolution, and brightness variations were transferred into g/cm³ using a calibration wedge (Eschbach et al., 1995). Measurements were then corrected based on the relationship between absolute (volume and weight) and radiographic (x-ray) wood density. During the MXD measurement procedure, amongst other wood parameters, also TRW is obtained. To this recent collection of data, we added TRW and MXD measurements from 11 PIHE trees sampled in the Vihren National Park (at 1920 m a.s.l.) by Fritz Schweingruber in 1981. These data are available in both the Swiss Federal Research Institute WSL’s data bank and the International Tree-Ring Data Bank (ITRDB; Grissino-Mayer and Fritts, 1997; shttp://www.ncdc.noaa.gov/paleo/treering.html). The MXD and TRW chronologies based on these 11 samples covered the period 1763–1981 (>5 trees) and correlated strongly over this period with temporary TRW and MXD chronologies based on the newly collected samples (r=0.85; p<0.01 for MXD and r=0.65; p<0.01 for TRW). We therefore decided to combine the two data sets and develop two new MXD and TRW chronologies.

Cross-dating of the MXD and TRW series was checked using COFECHA software (Grissino-Mayer, 2001) and graphical analyses. Age-related trends, represented by a general negative exponential curve, were conservatively removed from the raw MXD and TRW measurements by calculating residuals using ARSTAN software (Cook, 1985). Chronologies were then developed by averaging the detrended series using a robust mean (Cook and Peters, 1997). The final chronologies were truncated at a minimum sample replication of ten series to ensure sufficient signal and prevent biases to the variance structure (Wigley et al., 1984). Interseries correlation (RBAR) and the expressed population signal (EPS) were calculated for 30 yr windows lagged by 15 years to assess the signal strength of the chronologies (Wigley et al., 1984).

In total, 42 samples from 23 trees were cross-datable and incorporated in the final chronologies with mean RBAR values of 0.32 (MXD) and 0.34 (TRW). The resulting chronologies covered the period 1768–2008 with ten series (from six trees) and mean EPS-values were 0.93 for both chronologies and were higher than the commonly utilized, but arbitrary, threshold of 0.85 from 1765 onwards (see Figure 3b).

Climate data, analysis, and reconstruction

Three meteorological stations were considered for the climate/tree growth analysis: Vihren hut (41.8N, 23.5E; 1970 m a.s.l.; 1954–1979), Bansko (41.8N, 23.5E; 935 m a.s.l.; 1933–present), and Sofia (42.45N, 23.2E; 550 m a.s.l.; 1887–present). The Vihren hut station is located at only 1 km from the study site and at roughly the same elevation, but the 25 yr long time-series for this station is not long enough for climate reconstruction purposes. The Sofia station is at a further distance (approx. 100 km) from the study site than the Bansko station (approx. 12 km), but we decided to use the Sofia climate data for analysis and reconstruction because of the length of the Sofia time-series and the reliability of the data. Correlations between monthly time-series from Sofia and Bansko are high for temperature (ranging from 0.81 for June to 0.96 for February; 1933–2005; p<0.01) and lower for precipitation (0.43 for June to 0.77 for October; 1933–2005; p<0.01). The climatic data were provided by the Bulgarian National Hydrometeorological Institute and were verified for outliers, but not homogenized. We used Rotated Principal Component Analysis (RPCA) based indices of NAO and EA as provided by the NOAA Climate Prediction Center (CPC) for the period 1950–2010.

Spatial correlation analyses between the MXD chronology and climatic data sets were based on gridded seasonal (June–August) air temperature and precipitation fields for the period 1901–2006 (CRU T.S3.0; Mitchell and Jones, 2005) and on seasonal (June–August) 500 hPA geopotential height fields for the period 1878–present (Twentieth Century Reanalysis; Compo et al., 2011). For the pre-instrumental period (1768–1900), we used seasonal historical gridded reconstructions of air temperature (Luterbacher et al., 2004), precipitation (Pauling et al., 2006), and 500 hPA geopotential height (Luterbacher et al., 2002). Spatial correlation and composite maps were generated using the KNMI explorer (van Oldenborgh and Burgers, 2005; http://climexp.knmi.nl).

We reconstructed August temperature by linearly transforming (also referred to as scaling) the MXD values to match the mean and variance of the August temperature data for Sofia over the 1887–2005 period. For this purpose, instrumental temperatures were calculated as anomalies from the 1961–1990 reference period. While the exact form of the calibration and reconstruction methodologies has been the subject of considerable discussion (e.g. Christiansen, 2011; Esper et al., 2005b; Lee et al., 2008; Mann et al., 2005; Rutherford et al., 2005; von Storch et al., 2004), a primary advantage of the so-called composite plus scaling technique with variance matching (e.g., Jones et al., 1998) is a more accurate estimate of the amplitude of past temperature change (e.g. Esper et al., 2005a; Frank et al., 2010), but at the price of increased error variance (Kutzbach et al., 2011). We note that the variety of techniques involving a linear transformation of a single proxy data set does not alter the shape of the reconstruction, but that centennial-scale trends are not preserved in our chronology because of its limited mean segment length (187 years), which is likely to affect amplitude estimates (Cook et al., 1995). Calibration/verification tests were calculated on the scaled reconstruction over two subperiods (1887–1947 and 1948–2005) (Cook et al., 1994).

Uncertainty in the August temperature reconstruction arises from unexplained variance in the scaling model (calibration error) and the decreasing number of tree ring series back through time (chronology error; Esper et al., 2007). We estimated the calibration error as two standard error (SE) 95% confidence intervals (CI) for the full calibration period 1887–2005. We used ARSTAN software (Cook, 1985) to estimate the chronology error by bootstrapping (Briffa et al., 1992). Standardized tree ring measurements for every year were sampled with replacement 1000 times and arithmetic means were calculated. Two-tailed 95% CI were estimated based on the distribution of the bootstrapped mean. The composite plus scaling technique was then applied to the upper and lower limits of the CI for the tree-ring chronology and the overall error for the temperature reconstruction was estimated as the square root of the summed and squared calibration and chronology error terms.

Results

Climate–growth relationships

The two tree-ring parameters MXD and TRW show a distinct response pattern to interannual variability in temperature and precipitation (Figure 2). These correlation patterns are found to be generally time stable with no significant differences found when computed over the full (1887–2005), early (1887–1947), or recent (1948–2005) periods. MXD responds positively to warm season temperatures with highest correlations to temperatures averaged over July to September (r=0.71; p<0.001). Monthly correlations peak during August (r=0.65; p<0.001), and for this month there is the highest evidence for signal stationarity, with split period correlations always exceeding the 99% confidence interval. Inverse correlations between MXD and precipitation are found and likely reflect the negative correlation between temperature and precipitation during the summer months.

Figure 2.

Correlation of PIHE MXD (top) and TRW (bottom) chronologies with monthly (October of the previous year until September of the current year) and seasonal (July–September and December–April) temperature (left) and precipitation (right) values. Correlations were calculated over the full instrumental period (1887–2005) and two split periods (1887–1947 and 1948–2005). Horizontal lines indicate the 99% confidence limits for the full period (dashed lines) and for the split periods (dotted lines).

TRW, on the other hand, is positively correlated with winter (December to April) temperature (r=0.35; p<0.001), but negatively correlated with temperature variations from May to September (r=−0.3; p<0.01 for June). Correlations between TRW and precipitation are generally weak, but are highest and positive from May until July (r=0.3; p<0.01 for June).

August temperature reconstruction

Based upon the climate response of the MXD measurements, we performed calibration and verification trials between the PIHE MXD series and August temperatures and July–September temperatures (Table 1).

Table 1.

Calibration and verification statistics between the PIHE MXD series and August (Aug) and July–September (JAS) temperatures.

Period	Calibration	Verification
	R ²	R ²	RE	CE
	Aug/JAS	Aug/JAS	Aug/JAS	Aug/JAS
1887–1947	0.45/0.5	0.41/0.35	0.16/−0.005	0.15/−0.008
1948–2005	0.41/0.35	0.45/0.5	0.31/0.44	0.31/0.44
1887–2005	0.42/0.42

For August temperatures, 41% to 45% of the variance was explained in the verification procedure, indicating strong reconstruction skill and a stable strength of the climate signal in the tree ring record. This stability and the ability to capture lower frequency trends was confirmed by RE and CE statistics (Cook et al., 1994), with positive values for both statistics during the early and late verification periods. This was not the case for July–September temperatures: whereas 50% of the variance was explained in the earlier period, only 35% was explained in the later period. This temporal discrepancy was reflected in the RE and CE statistics, which were negative (reflecting lack of reconstruction skill) during the early verification period. We therefore reconstructed August temperatures (Figure 3) over the period 1768–2008 and the final reconstruction explains 42% of the instrumental variance based upon full period (1887–2005) correlations. It is worth noting, however, that calibration/verification tests are influenced by the choice of reconstruction methodology: RE and CE statistics generally have lower values for scaled compared with regressed reconstructions. We thus selected a strict calibration/verification test, which might explain failure of the tests for the July–September temperature reconstruction. Furthermore, a Durban-Watson test shows that the scaled residual MXD chronology is not positively autocorrelated (d=1.76, p<0.01) and thus that low-frequency persistence in the chronology and reconstruction is limited.

Figure 3.

Annual and decadally smoothed (5 yr cubic spline) reconstructed and instrumental (1887–2005; a) and reconstructed (1768–2008; b; 10 yr smoothing spline) Sofia August temperatures. Combined error (chronology and calibration) is shown (shaded) for the full reconstruction in (b). The 12 (13) warmest (coldest) years representing the 80th (20th) percentile that are common to both records are indicated by black stars (open circles) in (a). Thirty-year EPS values and sample depth are shown in (b).

Reconstructed and instrumental temperatures exhibit common inter-annual (r=0.65 for n=116; p<0.01) to decadal-scale (calculated as the Pearson correlation coefficient between time-series of sequential 5 yr non-overlapping means, n=21, r=0.68, p<0.01) variability (Figure 3a). Lower-frequency (centennial-scale) trends are likely not preserved in our reconstruction for the Pirin Mountains (hereafter called PIMO) because of limited segment length and individual series data adaptive detrending. The instrumental period features three distinct cold periods that are common to both records: the late 1890s (~1894–1902), the early 1910s (~1909–1916), and the late 1930s (~1935–1939). The prolonged cold phase in the 1970s and 1980s and the subsequent positive trend that is recorded in the instrumental data is not well captured by PIMO, which exhibits a shorter and earlier cold phase (~1968–1976) and generally warmer temperatures in the late 1970s and 1980s. Warm periods include the late 1920s (~1927–1931), the mid 1940s (1942–1947), and the late 1990s (1997–2001). A contingency analysis (Table 2) reveals that approximately two-thirds of the warmest and coldest individual years in the instrumental data are recorded in PIMO: 12 (13) out of 23 very low (high) temperatures are common in both records. Only 11% of the 46 warmest and coldest years in the instrumental record (4 warm years and 1 cold year), however, fall outside of the error margins calculated for PIMO (Figure 3a and b).

Table 2.

Contingency table of observed and reconstructed August temperatures (1887–2005). Years were ranked and divided into five equal 23 yr classes (26 yr for the moderate class). The class of the observed August temperature for each year was then compared with the class for the reconstructed temperature.

	Observed
	Very low	Low	Moderate	High	Very high
Reconstructed
Very low	12	5	5	1	0
Low	7	7	6	3	0
Moderate	2	8	6	8	2
High	2	3	7	3	8
Very high	0	0	2	8	13

Prior to the instrumental period, our reconstruction provides new evidence for temperatures back into the 18th century, with extremely cold summers in 1808, 1814, 1819, 1857, and 1884, and anomalous warmth in 1822, 1828, 1844, 1855, and 1856 (Figure 3b). The largest deviations from the long-term mean (0.53) occurred during the late 1700s and early 1800s. A 25 yr period with predominantly warm temperatures (~1782–1807) was followed by a remarkably cold period of comparable length (~1808–1827). The peak of the warm phase occurred around 1796–1797, the peak of the cold phase around 1812–1814. Both these reconstructed warm and cold periods exceed the decadal-scale temperature variability reconstructed for the instrumental period in amplitude, although caution is warranted in this conclusion because of possible loss of low-frequency information caused by the chronology development methods utilized.

Atmospheric circulation patterns over the eastern Mediterranean region

Analyses with instrumental data have demonstrated that atmospheric circulation patterns modulate climatic conditions in the Pirin Mountains with most pronounced influence during winter (Brown and Petkova, 2007; Tran et al., 2002). Above-normal snow cover and extreme winter precipitation events in Bulgaria are associated with negative phases of NAO, the EA (Brown and Petkova, 2007; Kostopoulou and Jones, 2007), and the Arctic Oscillation (AO; Cavazos, 2000).

In summer, the dominant atmospheric circulation pattern that governs European climate is a blocking-like pattern over northwestern Europe that co-occurs with a negative geopotential height (from 300 to 1000 hPA) anomaly over southeastern Europe (Casty et al., 2007). The meridionally oriented configuration of this circulation pattern favours the intrusion of warm air over the northeastern Mediterranean during one mode, when a low dominates the British Isles, and over the British Isles during its opposite mode, when blocking occurs over the British Isles. This atmospheric circulation pattern has previously been identified as the dominant mode of summer air temperature variability in Greece on interannual as well as decadal scales (Xoplaki et al., 2003a) and as the primary driver of extreme dry periods over the Eastern Mediterranean (Oikonomou et al., 2010). It explains 24% of the variance in summer air temperature (Xoplaki et al., 2003b) and close to 30% in summer precipitation (Dünkeloh and Jacobeit, 2003) over the entire Mediterranean region.

The results in Figure 4 confirm the importance of this circulation pattern for Bulgarian climate: hot and dry (cold and wet) summers in Bulgaria are characterized by strongly negative (positive) geopotential height anomalies centered over the British Isles and anomalously positive (negative) geopotential heights over the Balkans. Two summer teleconnection patterns can be related to this meridional circulation pattern: the summer NAO (sNAO) and the summer EA (sEA) pattern. During summer, the two nodes dominating the sign of the NAO, the Icelandic Low and the Azores High, move northward and negative sNAO phases are characterized by a trough over the British Isles (rather than a ridge as is the case in winter) and by positive pressure anomalies over the Azores and over southeastern Europe (Figure 4d). As a result, negative sNAO phases invoke wet and cold summers in northwestern Europe and dry, warm conditions in the northeastern Mediterranean (Barnston and Livezey, 1987) and thus show an opposite pattern as the negative winter NAO phase. For example, precipitation in Sofia correlates negatively with the NOAA CPC NAO index (1950–2005) in winter (December–February; r=−0.32; p<0.05), but positively in summer (June–August; r=0.26; p<0.1). The sEA pattern (Barnston and Livezey, 1987) strongly resembles the sNAO pattern, albeit with an opposite sign and a relatively stronger anomaly over the northeastern Mediterranean and the Azores, and a weaker anomaly over the British Isles and southern Scandinavia in particular (Figure 4c). Summer precipitation in Sofia is thus positively correlated with the sNAO index, but negatively with the sEA index (r=−0.4, p<0.01). Previous studies of Mediterranean climate have found that EA explains 19% of the total 500 hPa geopotential height variance over the Mediterranean (Quadrelli et al., 2001) and influences temperature variability in Italy in all seasons but autumn (Toreti et al., 2010).

Figure 4.

Pearson correlation maps of summer (June–August) Sofia temperature (a), precipitation (b) and sEA (c) and sNAO (d) indices (NOAA CPC) with 500 hPa geopotential height anomaly fields over the period 1950–2005. Correlations are significant (p<0.05) for |r|>0.27. The location of Sofia meteorological station and the study site is indicated by a black star on the maps.

Spatial correlations of the Sofia temperature reconstruction PIMO

PIMO is significantly positively correlated with instrumental summer temperatures in the Balkan region and in southern Italy (Figure 5a). Inverse correlations are found with summer temperatures in the British Isles, in southern Scandinavia, and in the coastal regions of western and northern Europe. The strong anti-correlation between temperature and precipitation in summer explains the correlation pattern with European summer precipitation (Figure 5b): PIMO is negatively correlated with precipitation patterns in the Balkans and southern Italy and positively with precipitation in the British Isles and Scandinavia. The geopotential height correlation map shows a blocking over the Balkan Peninsula during warm summers that co-occurs with a low over the British Isles (Figure 5c). As described in the previous section, this pattern resembles the negative phase of the sNAO and can explain the teleconnection between northwestern and southeastern European temperature and precipitation patterns. PIMO thus correlates significantly with a RPCA-based sNAO index (1950–2008; r=−0.49; p<0.001). The correlation maps prior to the 20th century (1768–1900) show striking similarities in correlation extent with the instrumental period maps, but the strength of the correlations is weaker over this earlier (and longer) time period (Figure 5d–f). The strength of the climatic signal in PIMO, as well as the general stability of the described teleconnection patterns, is thus confirmed on an interannual scale back to 1768.

Figure 5.

Correlation maps of PIMO with instrumental (1901–2005; Compo et al., 2011; Mitchell and Jones, 2005; top) and reconstructed (1768–1900; Luterbacher et al., 2002, 2004; Pauling et al., 2006; bottom) fields of summer (June–August) temperature (a, d), precipitation (b, e), and 500 hPA geopotential height (c, f). Absolute r-values higher than 0.2 (coloured regions) are significant at the 95% confidence level (for n=100). The location of our study site is indicated on the maps by a black star. Spatial correlation and composite maps were generated using the KNMI explorer (van Oldenborgh and Burgers, 2005; http://climexp.knmi.nl).

To further investigate the stationarity of the teleconnection patterns with temperature, we compiled 90 MXD records from across Europe from the ITRDB. All chronologies were 55 years or longer after truncation at a sample replication of ten series. We were primarily interested in the high-frequency synoptic relationships across Europe and therefore the so-called residual chronologies (with low-order autocorrelation removed from individual series before chronology compilation) were used for correlation analyses (Figure 6a). We found overall positive correlations with MXD time series in the Balkans, southern and central Italy, and the Carpathians. Correlations were particularly strong for other PIHE chronologies and ranged from 0.58 for Crispo, Italy to 0.72 for Mount Olympos in Greece. PIMO also correlated significantly positively (r=0.25; p<0.01) with a documentary-based reconstruction of summer temperature in southern Italy (Camuffo et al., 2010; 1791–2006). It is interesting that PIMO did not correlate with MXD chronologies in Cyprus, the western Mediterranean, or in the Alps and central Europe, but showed significantly negative correlations with MXD chronologies from Scotland and England (r=−0.16; n.s. for Mallaig to r=−0.47; p<0.001 for Dimmie). These correlation patterns with the MXD records across Europe match the teleconnections derived from instrumental data, between tree ring and instrumental data, and between tree ring and multiproxy reconstructions across Europe. As the MXD records were independent from the above, this gives further confidence in the robustness and importance of this pattern.

Figure 6.

(a) Pearson correlation coefficients r between PIMO and other MXD chronologies in Europe. MXD chronologies include all chronologies for Europe that date back prior to 1900 and are available in the International Tree-Ring Data Bank (ITRDB; Grissino-Mayer and Fritts, 1997; shttp://www.ncdc.noaa.gov/paleo/treering.html). PIHE MXD chronologies are indicated by squares. Absolute r-values higher than 0.2 are significant at the 95% confidence level (for n=100); (b) annual and decadally smoothed (10 yr smoothing spline) reconstructions of August temperature anomalies in Bulgaria (black) and of summer NAO (Folland et al., 2009; grey, inversed).

Folland et al. (2009) combined two of the MXD records from Scotland (Coulin and Inverey) with MXD and TRW records from western Norway for their reconstruction of the sNAO and PIMO correlates significantly negatively with this sNAO reconstruction over its full length (1768–1976; r=−0.43; p<0.001; Figure 6b). The correlation is predominantly expressed in the high (interannual) frequency domain (r=−0.48; p<0.001) and diminishes on lower (decadal) frequencies (r=−0.05; n.s. for 10 yr spline-smoothed series). The correlation is fairly stable over time: a running correlation analysis (31 yr moving windows calculated over the central year) showed that high-frequency correlations were negative at all points in time and statistically significant (p<0.05) over the full time-period apart from the 1920s (1922–1932) and the early 1800s (1796–1831).

Discussion

In this paper, we present the first PIHE-based reconstruction of summer temperature for Bulgaria that extends back to 1768. This reconstruction is based upon a compilation of new and existing MXD measurements from high-elevation PIHE trees in the Pirin Mountains. The climatic signal in trees from high-elevation, summer-dry regions is generally much stronger in MXD measurements compared with TRW data (Büntgen et al., 2010) in line with the increased climatic sensitivity of MXD (Frank and Esper, 2005). PIHE TRW variability at our sites is weakly sensitive to previous winter temperatures and to early summer drought (Figure 2). The positive response of PIHE tree growth to mild winter and spring conditions has been described in other studies in the Balkans (Panayotov et al., 2010) and in earlywood measurements from southern Italy (Todaro et al., 2007) and has been attributed to an earlier onset of cambial activity and other differentiation phases, resulting in an increased duration of xylogenesis (Deslauriers et al., 2008). Other potential explanations include increased soil moisture availability after warm winter snowmelt and fine root damage during unusually cold winters (Panayotov et al., 2010). Given the location of our study site on a steep, sun-exposed slope with shallow rooting, the sensitivity of PIHE growth to summer drought does not come as a surprise, despite the high elevation. Latewood formation in high-elevation PIHE trees occurs between mid to late July and early to late September (Deslauriers et al., 2008; Rossi et al., 2006) matching well with MXD sensitivity derived from analysis with monthly instrumental data (Figure 2). August temperatures in particular influence wall thickening and lignification processes in PIHE and thus latewood density (Deslauriers et al., 2008). The lignification of the outmost row of latewood tracheids extends considerably beyond cessation of cambial activity, which could explain the relation between MXD and late summer temperature (Gricar et al., 2005).

TRW and MXD time-series generally differ in carry-over effects from previous year growth conditions and thus in autocorrelation strength (Briffa et al., 2002; Frank et al., 2007b). Earlywood cell formation, for instance, a dominant contributor to TRW, can depend on carbohydrate storage during the previous growing season as well as longer lasting effects related to climatic influences on needle length and retention, root growth, and so on (Fritts, 1976; Kozlowski and Pallardy, 1997). These autocorrelation processes tend to give the TRW a ‘redder’ spectrum in comparison to the climatic drivers (Frank et al., 2007a; Meko, 1981). TRW in PIHE trees from a site very similar to our study site in the Pirin mountains was thus found to be sensitive to previous years summer temperature and precipitation (Panayotov et al., 2010). MXD data on the other hand are generally less strongly correlated with prior seasons and have an autocorrelation structure that more closely resembles that of the climatic forcing (Frank et al., 2007a). For this reason, and because age-trends are often not as variable in MXD time-series compared with TRW series (Schweingruber et al., 1978), we applied a general negative exponential detrending in this study, which retains the strongest signal in the interannual frequency. As a result, our MXD-based reconstruction mainly reflects interannual temperature variability and decadal-scale variations through smoothing, but because of the limited mean segment length of our chronology, longer-term (centennial-scale) trends are not preserved (Cook et al., 1995).

On a decadal scale, our reconstruction reflects well summer temperature fluctuations in the Balkans over the instrumental period and particularly over the first half of the 20th century (Figure 3a). The cold periods in the 1890s and the 1910s and the summer warming in the 1920s and 1940s are not only evident from the Sofia instrumental data, but also from summer temperature records for the eastern Mediterranean (Repapis and Philandras, 1988; Xoplaki et al., 2003a), and in Mediterranean sea surface temperatures (Metaxas et al., 1991). The 1940s in particular have also been identified as a period of widespread summer drought in Bulgaria (Alexandrov et al., 2004; Koleva and Alexandrov, 2008; Tran et al., 2002) and throughout Europe (Briffa et al., 1994). Interdecadal correspondence between the reconstructed and instrumental records is less close after approximately 1970. The warming trend since the early 1980s that is apparent in the Sofia record and throughout the eastern Mediterranean (Alexandrov et al., 2004; Giles and Flocas, 1984; Philandras et al., 2008; Toreti et al., 2010; Xoplaki et al., 2003a) is interrupted in our reconstruction by warm/dry conditions recorded for the 1980s. It is worth noting that this summer warming trend in the eastern Mediterranean is lagged compared with general Northern Hemisphere warming (Feidas et al., 2004; Giles and Flocas, 1984; Maheras and Kutiel, 1999). The discrepancy between reconstructed and instrumental temperatures starting in the 1970s can be caused by the lack of homogenization of the Sofia instrumental data (and thus the lack of removal of urbanization effects, Peterson et al., 1998), by low-frequency differences between lowland (Sofia) and mountainous (PIMO) temperatures, or by end-effects in the low-frequency time-series caused by the selected smoothing function (D’arrigo et al., 2008).

Despite the reduced correspondence between instrumental and reconstructed temperatures since the 1970s on an interdecadal scale, potentially because of overly sensitive smoothing of the data, warm as well as cool events throughout the instrumental period are well represented in our reconstruction on an interannual scale. Several of the warm extremes in our reconstruction correspond to years of extreme droughts in the Bulgarian lowlands (amongst others 1928, 1945, 2000; Koleva and Alexandrov, 2008) and to heat waves in Serbia (amongst others 1952, 1994, and 2003; Unkasevic and Tosci, 2009). The 1959 and 1976 cold extremes correspond to extremely cool summers throughout the Mediterranean (Xoplaki et al., 2003b).

It is worth noting that this same year 1976 was one of the driest summers on record in Central and Western Europe (Zaidman et al., 2002), and in the British Isles in particular (Jones and Conway, 1997). An explanation for this seeming discrepancy can be found in Figure 4a, which shows a seesaw pattern between summer temperature variability in the Balkans versus northwestern Europe. A similar pattern can be found when comparing summer precipitation patterns over both regions (Figure 4b) and is associated with a prominent, zonally organized atmospheric circulation pattern with strongly opposing geopotential height nodes over the British Isles and the Balkans and a third centre over the Azores (Figure 4c). This pattern strongly resembles the sNAO pattern, described by Barnston and Livezey, (1987) as the dominant mode of North Atlantic variability and the primary driver of European summer precipitation variability (Zveryaev and Allan, 2010). When the sNAO is in a negative mode, a low centered over the British Isles is accompanied by convection and cloud formation, whereas a warm high over the northeastern Mediterranean leads to increased stability, clear skies, and maximum insolation. At the upper levels (300–500 hPa), this configuration facilitates a strong Polar Front Jet and anomalous northeasterly to easterly flow that brings warm continental air masses to the Mediterranean. At the surface level, the high pressure system over the eastern Mediterranean gives way to the British Isles Low. This cyclonic pattern is the most frequent summer circulation pattern in the Eastern Mediterranean (Kostopoulou and Jones, 2007) and it favours warm winds from North Africa and the Sahara desert (Maheras and Kutiel, 1999). High summer temperatures in the Balkans during this negative phase can thus be contributed to a combination of subsidence, anomalous easterly, continental flow at upper levels, and increased advection of warm air coming from North Africa (Xoplaki et al., 2003a). During a positive sNAO mode, opposite conditions occur and a ridge over the British Isles coincides with a trough over the northeastern Mediterranean, resulting in anomalously cool summers in the Balkans.

A comparison of PIMO with reconstructed fields of temperature, precipitation, and geopotential height shows that teleconnections of the northeastern Mediterranean climate with the previously described atmospheric circulation pattern were stable on interannual scales over the last ~ 240 years (Figure 5d–f). PIMO is positively related to temperature reconstructions in the Balkans, Southern and Central Italy, and Western Turkey, whereas opposing temperature variability patterns are manifested over the British Isles, southern Scandinavia, and the Western European North Sea and Atlantic Coast. This spatial pattern is mimicked in a correlation map of PIMO with other MXD tree-ring records over Europe (Figure 6a). PIMO shows similar temporal variability patterns as other MXD records throughout the Balkans, Italy, and even Romania and Poland and the similarity is particularly strong for the three other PIHE records from Greece and Italy. The majority of MXD records from the UK and southern Scandinavia are in anti-phase with PIMO and it thus comes as no surprise that PIMO correlates negatively with a sNAO reconstruction based on two of these records (Figure 6b). The relationship between the two reconstructions confirms the sNAO pattern as the main driver of the teleconnection between summer temperatures in southeastern versus northwestern Europe and this teleconnection is most pronounced on interannual (rather than decadal or lower frequency) timescales. It is unlikely that this lack of low-frequency correspondence is an artefact of the reconstruction method, because the two reconstructions were detrended using similar negative exponential functions which preserve decadal-scale variability and both of them calibrate well against instrumental targets on interannual as well as decadal scales (Figure 3a and Folland et al., 2009).

The influence of the sNAO on Sofia summer temperature appears to be stable in time, with the exception of the period between ~1795 and 1830. PIMO shows a decrease in EPS during this period (Figure 3b), which might explain reduced coherence with the sNAO. Also Linderholm et al. (2009) found a weakening of the relationship between the sNAO record and a summer drought record for east-central Sweden prior to 1835. Folland et al. (2009)’s reconstruction shows extremely negative sNAO values for the period 1790–1810 and they found a loss of coherency between the sNAO and summer temperature records in England and Stockholm over this time period (Folland et al., 2009). The two decades following this period were characterized by long and cold winters in Bulgaria (Nikolva, 2004) and are well-documented as also having been anomalously cold throughout Europe (Battipaglia et al., 2010; Büntgen et al., 2006) owing to extraordinary volcanic activity (Cole-Dai et al., 2009). The eruption of the Krakatau volcano in Indonesia in 1883 (Self and Rampino, 1981) is also marked by an anomalously narrow ring in the following year in the PIMO reconstruction. It is possible that external, volcanic forcing overrides sNAO influence causing temperatures during the 1810s to be anomalously cold in southeastern as well as in northwestern Europe. It is likely that other periods of reduced coherency between PIMO and the sNAO record (e.g. the 1920s) are related to strong external forcing or to the dominant influence of different atmospheric circulation patterns (D’Arrigo and Wilson, 2006; Greatbatch and Rong, 2006). The strong relationship between tree-ring records from the Balkans and the British Isles suggests that the development of a sNAO reconstruction based on a network of tree-ring data that reflect this teleconnection pattern is feasible on interannual timescales. The coherence between teleconnected records, however, is less reliable on lower (decadal to centennial) timescales and caution should be advised for the interpretation of this teleconnection pattern on decadal to centennial timescales.

Conclusion

Here we present a MXD-based summer temperature reconstruction for the Central Balkans extending back to 1768. The climatic signal in MXD measurements of PIHE trees growing at high-elevation Mediterranean sites is strong and distinct compared with TRW measurements from the same trees and this parameter is therefore recommended for tree-ring based temperature reconstructions for these areas. Our summer temperature reconstruction calibrates well over the instrumental period and reliably reflects region-wide temperature variability back to the late 18th century. Furthermore, a teleconnection pattern between summer temperatures in southeastern versus northwestern Europe is revealed in this paper. This teleconnection pattern is persistent over approximately the last two centuries and can be linked to the summer NAO pattern. However, strong external climatic forcing (e.g. volcanic activity in the 1810s) appears to over-ride this teleconnection pattern. Given the longevity of PIHE trees in the Balkans (up to 1000 years in Bulgaria and Albania), temperature reconstructions based on this species could shed light on the climate dynamics (external versus internal forcing) behind large-scale climate transition periods such as the ‘Medieval Climate Anomaly’–‘Little Ice Age’ transition in the 14th–15th centuries.

Footnotes

Acknowledgements

The authors are grateful to U Buentgen and K Treydte, for helpful comments and F Schweingruber for his data contribution. For assistance in the field we thank F Herzig, N Nikolova, M Russeva, W Tegel, N Tsvetanov, and J Van Moerkerke. We also thank Pirin National Park administration for providing permission to work in Bunderitsa valley.

Funding

This study was supported by MedCLIVAR, by NCCR Climate, University of Forestry project 115/2008, Velux Foundation Project 414, and project DTK 02/2/2010 of the National Science Fund of Bulgaria.

References

Akkemik

Aras

(2005) Reconstruction (1689–1994 AD) of April–August precipitation in the southern part of central Turkey. International Journal of Climatology 25: 537–548.

Alexandrov

Schneider

Koleva

. (2004) Climate variability and change in Bulgaria during the 20th century. Theoretical and Applied Climatology 79: 133–149.

Barnston

Livezey

(1987) Classification, seasonality and persistence of low-frequency atmospheric circulation patterns. Monthly Weather Review 115: 1083–1126.

Barriendos

Rodrigo

(2006) Study of historical flood events on Spanish rivers using documentary data. Hydrological Sciences Journal-Journal Des Sciences Hydrologiques 51: 765–783.

Battipaglia

Frank

Buntgen

. (2010) Five centuries of Central European temperature extremes reconstructed from tree-ring density and documentary evidence. Global and Planetary Change 72: 182–191.

Briffa

Jones

Bartholin

. (1992) Fennoscandian Summers from AD 500 – Temperature changes on short and long timescales. Climate Dynamics 7: 111–119.

Briffa

Jones

Hulme

(1994) Summer moisture variability across Europe, 1892–1991 – An analysis based on the Palmer Drought Severity Index. International Journal of Climatology 14: 475–506.

Briffa

Osborn

Schweingruber

. (2002) Tree-ring width and density data around the Northern Hemisphere: Part 1, local and regional climate signals. The Holocene 12: 737–757.

Brown

Petkova

(2007) Snow cover variability in Bulgarian mountainous regions, 1931–2000. International Journal of Climatology 27: 1215–1229.

10.

Büntgen

Frank

Grudd

. (2008) Long-term summer temperature variations in the Pyrenees. Climate Dynamics 31: 615–631.

11.

Büntgen

Frank

Nievergelt

. (2006) Summer temperature variations in the European Alps, AD 755–2004. Journal of Climate 19: 5606–5623.

12.

Büntgen

Frank

Trouet

. (2010) Diverse climate sensitivity of Mediterranean tree-ring width and density. Trees – Structure and Function DOI: 10.1007/s00468-009-0396-y10.1007/s00468-009-0396-y.

13.

Camuffo

Bertolin

Barriendos

. (2010) 500-year temperature reconstruction in the Mediterranean Basin by means of documentary data and instrumental observations. Climatic Change 101: 169–199.

14.

Casty

Raible

Stocker

. (2007) A European pattern climatology 1766–2000. Climate Dynamics 29: 791–805.

15.

Cavazos

(2000) Using self-organizing maps to investigate extreme climate events: An application to wintertime precipitation in the Balkans. Journal of Climate 13: 1718–1732.

16.

Chang

Knight

Staneva

. (2002) Water resource impacts of climate change in southwestern Bulgaria. GeoJournal 57: 159–168.

17.

Christiansen

(2011) Reconstructing the NH mean temperature: Can underestimation of trends and variability be avoided? Journal of Climate 24: 674–692.

18.

Cole-Dai

Ferris

Lanciki

. (2009) Cold decade (AD 1810–1819) caused by Tambora (1815) and another (1809) stratospheric volcanic eruption. Geophysical Research Letters 36: 6.

19.

Compo

Whitaker

Sardeshmukh

. (2011) The twentieth century reanalysis project. Quarterly Journal of the Royal Meteorological Society 137: 1–28.

20.

Cook

(1985) A Time-series Analysis Approach to Tree-ring Standardization. University of Arizona, 171 pp.

21.

Cook

Peters

(1997) Calculating unbiased tree-ring indices for the study of climatic and environmental change. The Holocene 7: 361–370.

22.

Cook

Briffa

Jones

(1994) Spatial regression methods in dendroclimatology – A review and comparison of 2 techniques. International Journal of Climatology 14: 379–402.

23.

Cook

Briffa

Meko

. (1995) The segment length curse in long tree-ring chronology development for paleoclimatic studies The Holocene 5: 229–237.

24.

D’Arrigo

Wilson

(2006) On the Asian expression of the PDO. International Journal of Climatology 26: 1607–1617.

25.

D’Arrigo

Wilson

Liepert

. (2008) On the ‘divergence problem’ in northern forests: A review of the tree-ring evidence and possible causes. Global and Planetary Change 60: 289–305.

26.

Deslauriers

Rossi

Anfodillo

. (2008) Cambial phenology, wood formation and temperature thresholds in two contrasting years at high altitude in southern Italy. Tree Physiology 28: 863–871.

27.

Dominguez-Castro

Santisteban

Barriendos

. (2008) Reconstruction of drought episodes for central Spain from rogation ceremonies recorded at the Toledo Cathedral from 1506 to 1900: A methodological approach. Global and Planetary Change 63: 230–242.

28.

Dünkeloh

Jacobeit

(2003) Circulation dynamics of Mediterranean precipitation variability 1948–1998. International Journal of Climatology 23: 1843–1866.

29.

Eschbach

Nogler

Schär

. (1995) Technical advances in the radiodensitometrical determination of wood density. Dendrochronologia 19: 155–168.

30.

Esper

Buntgen

Frank

. (2006) Multiple tree-ring parameters from Atlas cedar (Morocco) and their climatic signal. Tree Rings in Archaeology, Climatology, and Ecology (TRACE) 4: 46–55.

31.

Esper

Frank

Buntgen

. (2007) Long-term drought severity variations in Morocco. Geophysical Research Letters 34(17): L17702 10.1029/2007gl03084410.1029/2007gl030844.

32.

Esper

Frank

Wilson

RJS

. (2005a) Effect of scaling and regression on reconstructed temperature amplitude for the past millennium. Geophysical Research Letters 32(7): L07711 10.1029/2004gl02123610.1029/2004gl021236.

33.

Esper

Wilson

RJS

Frank

. (2005b) Climate: Past ranges and future changes. Quaternary Science Reviews 24: 2164–2166.

34.

Feidas

Makrogiannis

Bora-Senta

(2004) Trend analysis of air temperature time series in Greece and their relationship with circulation using surface and satellite data: 1955–2001. Theoretical and Applied Climatology 79: 185–208.

35.

Felis

Rimbu

(2010) Mediterranean climate variability documented in oxygen isotope records from northern Red Sea corals – A review. Global and Planetary Change 71: 232–241.

36.

Fernandez

Saenz

Zorita

(2003) Analysis of wintertime atmospheric moisture transport and its variability over southern Europe in the NCEP reanalyses. Climate Research 23: 195–215.

37.

Fleitmann

Cheng

Badertscher

. (2009) Timing and climatic impact of Greenland interstadials recorded in stalagmites from northern Turkey. Geophysical Research Letters 36: L19707 10.1029/2009gl04005010.1029/2009gl040050.

38.

Folland

Knight

Linderholm

. (2009) The summer North Atlantic Oscillation: Past, present, and future. Journal of Climate 22: 1082–1103.

39.

Founda

van Loon

(2008) Circulation anomalies associated with winter temperature extremes in Athens during the period 1900–2004. Meteorologische Zeitschrift 17: 55–59.

40.

Frank

Esper

(2005) Characterization and climate response patterns of a high-elevation, multi-species tree-ring network for the European Alps. Dendrochronologia 22: 107–121.

41.

Frank

Buntgen

Bohm

. (2007a) Warmer early instrumental measurements versus colder reconstructed temperatures: Shooting at a moving target. Quaternary Science Reviews 26: 3298–3310.

42.

Frank

Esper

Cook

(2007b) Adjustment for proxy number and coherence in a large-scale temperature reconstruction. Geophysical Research Letters 34(16): L16709 10.1029/2007gl03057110.1029/2007gl030571.

43.

Frank

Esper

Raible

. (2010) Ensemble reconstruction constraints on the global carbon cycle sensitivity to climate. Nature 463: 527–530.

44.

Fritts

(1976) Tree Rings and Climate. London: Academic Press.

45.

Gao

Giorgi

(2008) Increased aridity in the Mediterranean region under greenhouse gas forcing estimated from high resolution simulations with a regional climate model. Global and Planetary Change 62: 195–209.

46.

Giannakopoulos

Le Sager

Bindi

. (2009) Climatic changes and associated impacts in the Mediterranean resulting from a 2 degrees C global warming. Global and Planetary Change 68: 209–224.

47.

Giles

Flocas

(1984) Air-temperature variations in Greece. 1. Persistence, trend, and fluctuations. Journal of Climatology 4: 531–539.

48.

Giorgi

(2006) Climate change hot-spots. Geophysical Research Letters 33(8): L08707 10.1029/2006gl02573410.1029/2006gl025734.

49.

Giorgi

Lionello

(2008) Climate change projections for the Mediterranean region. Global and Planetary Change 63: 90–104.

50.

Glueck

Stockton

(2001) Reconstruction of the North Atlantic Oscillation, 1429–1983. International Journal of Climatology 21: 1453–1465.

51.

Greatbatch

Rong

(2006) Discrepancies between different Northern Hemisphere summer atmospheric data products. Journal of Climate 19: 1261–1273.

52.

Gricar

Cufar

Oven

. (2005) Differentiation of terminal latewood tracheids in silver fir during autumn. Annals of Botany 95: 959–965.

53.

Griggs

DeGaetano

Kuniholm

. (2007) A regional high-frequency reconstruction of May–June precipitation in the north Aegean from oak tree rings, AD 1089–1989. International Journal of Climatology 27: 1075–1089.

54.

Grissino-Mayer

(2001) Crossdating accuracy: A manual and tutorial for the computer program COFECHA. Tree-Ring Research 57: 205–221.

55.

Grissino-Mayer

Fritts

(1997) The International Tree-Ring Data Bank: An enhanced global database serving the global scientific community. The Holocene 7: 235–238.

56.

Grunewald

Scheithauer

Monget

. (2009) Characterisation of contemporary local climate change in the mountains of southwest Bulgaria. Climatic Change 95: 535–549.

57.

Hertig

Jacobeit

(2008) Assessments of Mediterranean precipitation changes for the 21st century using statistical downscaling techniques. International Journal of Climatology 28: 1025–1045.

58.

Hurrell

(1995) Decadal trends in the North Atlantic Oscillation – Regional temperatures and precipitation Science 269: 676–679.

59.

Jones

Conway

(1997) Precipitation in the British Isles: An analysis of area-average data updated to 1995 International Journal of Climatology 17(4): 427–438.

60.

Jones

Briffa

Barnett

. (1998) High-resolution palaeoclimatic records for the last millennium: Interpretation, integration and comparison with General Circulation Model control-run temperatures. The Holocene 8: 455–471.

61.

Koleva

Alexandrov

(2008) Drought in the Bulgarian low regions during the 20th century. Theoretical and Applied Climatology 92: 113–120.

62.

Kostopoulou

Jones

(2005) Assessment of climate extremes in the Eastern Mediterranean. Meteorology and Atmospheric Physics 89: 69–85.

63.

Kostopoulou

Jones

(2007) Comprehensive analysis of the climate variability in the eastern Mediterranean. Part I: Map-pattern classification. International Journal of Climatology 27: 1189–1214.

64.

Kozlowski

Pallardy

(1997) Growth Control in Woody Plants. San Diego: Academic Press.

65.

Kuglitsch

Toreti

Xoplaki

. (2010) Heat wave changes in the eastern Mediterranean since 1960. Geophysical Research Letters 37: 5.

66.

Kutzbach

Thees

Wilmking

(2011) Identification of linear relationships from noisy data using errors-in-variables models-relevance for reconstruction of past climate from tree-ring and other proxy information. Climatic Change 105: 155–177.

67.

Lee

TCK

Zwiers

Tsao

(2008) Evaluation of proxy-based millennial reconstruction methods. Climate Dynamics 31: 263–281.

68.

Linderholm

Folland

Walther

(2009) A multicentury perspective on the summer North Atlantic Oscillation (SNAO) and drought in the eastern Atlantic Region. Journal of Quaternary Science 24: 415–425.

69.

Luterbacher

Dietrich

Xoplaki

. (2004) European seasonal and annual temperature variability, trends, and extremes since 1500. Science 303: 1499–1503.

70.

Luterbacher

Koenig

Franke

. (2010) Circulation dynamics and its influence on European and Mediterranean January– April climate over the past half millennium: results and insights from instrumental data, documentary evidence and coupled climate models. Climatic Change 101: 201–234.

71.

Luterbacher

Xoplaki

Dietrich

. (2002) Reconstruction of sea level pressure fields over the Eastern North Atlantic and Europe back to 1500. Climate Dynamics 18: 545–561.

72.

Maheras

Kutiel

(1999) Spatial and temporal variations in the temperature regime in the Mediterranean and their relationship with circulation during the last century. International Journal of Climatology 19: 745–764.

73.

Mann

Rutherford

Wahl

. (2005) Testing the fidelity of methods used in proxy-based reconstructions of past climate. Journal of Climate 18: 4097–4107.

74.

Meehl

Tebaldi

(2004) More intense, more frequent, and longer lasting heat waves in the 21st century. Science 305: 994–997.

75.

Meko

(1981) Applications of Box-Jenkins methods of time-series analysis to reconstruction of drought from tree rings. Unpublished PhD Dissertation, University of Airzona, 149 pp.

76.

Metaxas

Bartzokas

Vitsas

(1991) Temperature fluctuations in the Mediterranean area during the last 120 years. International Journal of Climatology 11: 897–908.

77.

Mitchell

Jones

(2005) An improved method of constructing a database of monthly climate observations and associated high-resolution grids. International Journal of Climatology 25: 693–712.

78.

Moriondo

Good

Durao

. (2006) Potential impact of climate change on fire risk in the Mediterranean area. Climate Research 31: 85–95.

79.

Nicault

Alleaume

Brewer

. (2008) Mediterranean drought fluctuation during the last 500 years based on tree-ring data. Climate Dynamics 31: 227–245.

80.

Nikolva

(2004) Times of Plague (1700–1850). Sofia: IF94(in Bulgarian).

81.

Oikonomou

Flocas

Hatzaki

. (2010) Relationship of extreme dry spells in Eastern Mediterranean with large-scale circulation. Theoretical and Applied Climatology 100: 137–151.

82.

Onol

Semazzi

FHM

(2009) Regionalization of climate change simulations over the eastern Mediterranean. Journal of Climate 22: 1944–1961.

83.

Panayotov

Yurukov

(2007) Tree ring chronology from Pinus peuce in Pirin Mts and the possibilities to use it for climate analysis. Phytologia Balcanica 13: 313–320.

84.

Panayotov

Bebi

Trouet

. (2010) Climate signal in tree-ring chronologies of Pinus peuce and Pinus heldreichii from the Pirin Mountains in Bulgaria. Trees – Structure and Function 24: 479–490.

85.

Pauling

Luterbacher

Casty

. (2006) Five hundred years of gridded high-resolution precipitation reconstructions over Europe and the connection to large-scale circulation. Climate Dynamics 26: 387–405.

86.

Peterson

Easterling

Karl

. (1998) Homogeneity adjustments of in situ atmospheric climate data: A review. International Journal of Climatology 18: 1493–1517.

87.

Philandras

Nastos

Repapis

(2008) Air temperature variability and trends over Greece. Global Nest Journal 10: 273–285.

88.

Piervitali

Colacino

(2001) Evidence of drought in western Sicily during the period 1565–1915 from liturgical offices. Climatic Change 49: 225–238.

89.

Quadrelli

Pavan

Molteni

(2001) Wintertime variability of Mediterranean precipitation and its links with large-scale circulation anomalies. Climate Dynamics 17: 457–466.

90.

Repapis

Philandras

(1988) A note on the air-temperature trends of the last 100 years as evidenced in the eastern Mediterranean time-series. Theoretical and Applied Climatology 39: 93–97.

91.

Rossi

Deslauriers

Anfodillo

. (2006) Conifers in cold environments synchronize maximum growth rate of tree-ring formation with day length. New Phytologist 170: 301–310.

92.

Rutherford

Mann

Osborn

. (2005) Proxy-based Northern Hemisphere surface temperature reconstructions: Sensitivity to method, predictor network, target season, and target domain. Journal of Climate 18: 2308–2329.

93.

Schroter

Cramer

Leemans

. (2005) Ecosystem service supply and vulnerability to global change in Europe. Science 310: 1333–1337.

94.

Schweingruber

Fritts

Braeker

. (1978) The X-ray technique as applied to dendrochronology. Tree-Ring Bulletin 38: 61–91.

95.

Seim

Buntgen

Fonti

. (2012) The paleoclimatic value of a millennium-long tree-ring chronology from Albania. Climate Research in press.

96.

Self

Rampino

(1981) The 1883 eruption of Krakatau. Nature 294: 699–704.

97.

Serre-Bachet

(1994) Middle-ages temperature reconstrucctions in Europe, a focus on northeastern Italy. Climatic Change 26: 213–224.

98.

Tardif

Camarero

Ribas

. (2003) Spatiotemporal variability in tree growth in the Central Pyrenees: Climatic and site influences. Ecological Monographs 73: 241–257.

99.

Thompson

DWJ

Wallace

(1998) The Arctic Oscillation signature in the wintertime geopotential height and temperature fields. Geophysical Research Letters 25: 1297–1300.

100.

Todaro

Andreu

D’Alessandro

. (2007) Response of Pinus leucodermis to climate and anthropogenic activity in the National Park of Pollino (Basilicata, Southern Italy). Biological Conservation 137: 507–519.

101.

Toreti

Desiato

Fioravanti

. (2010) Seasonal temperatures over Italy and their relationship with low-frequency atmospheric circulation patterns. Climatic Change 99: 211–227.

102.

Touchan

Akkemik

Hughes

. (2007) May–June precipitation reconstruction of Southwestern Anatolia, Turkey during the last 900 years from tree rings. Quaternary Research 68: 196–202.

103.

Touchan

Anchukaitis

Meko

. (2008a) Long term context for recent drought in northwestern Africa. Geophysical Research Letters 35(13): L13705 10.1029/2008gl03426410.1029/2008gl034264.

104.

Touchan

Funkhouser

Hughes

. (2005a) Standardized precipitation index reconstructed from Turkish tree-ring widths. Climatic Change 72: 339–353.

105.

Touchan

Meko

Aloui

(2008b) Precipitation reconstruction for Northwestern Tunisia from tree rings. Journal of Arid Environments 72: 1887–1896.

106.

Touchan

Xoplaki

Funkhouser

. (2005b) Reconstructions of spring/summer precipitation for the Eastern Mediterranean from tree-ring widths and its connection to large-scale atmospheric circulation. Climate Dynamics 25: 75–98.

107.

Tran

Knight

Wesner

(2002) Drought in Bulgaria and atmospheric synoptic conditions over Europe. GeoJournal 57: 149–157.

108.

Turkes

Erlat

(2003) Precipitation changes and variability in Turkey linked to the North Atlantic Oscillation during the period 1930–2000. International Journal of Climatology 23: 1771–1796.

109.

Turkes

Erlat

(2008) Influence of the Arctic Oscillation on the variability of winter mean temperatures in Turkey. Theoretical and Applied Climatology 92: 75–85.

110.

Unkasevic

Tosci

(2009) An analysis of heat waves in Serbia. Global and Planetary Change 65: 17–26.

111.

Vakarelov

Mirtchev

Kachaunova

. (2001) Reconstruction of summer air temperatures by dendrochronological analysis of Macedonian pine (Pinus peuce Griseb.) in Pirin mountains (Southeast Bulgaria). Forestry Ideas 1–4: 16–26.

112.

van Oldenborgh

Burgers

(2005) Searching for decadal variations in ENSO precipitation teleconnections. Geophysical Research Letters 32(15): L15701 10.1029/2005gl02311010.1029/2005gl023110.

113.

von Storch

Zorita

Jones

. (2004) Reconstructing past climate from noisy data. Science 306: 679–682.

114.

Wallace

Gutzler

(1981) Teleconnections in the geopotential height field during the Northern Hemisphere Winter. Monthly Weather Review 109: 784–812.

115.

Wigley

TML

Briffa

Jones

(1984) On the average value of correlated time-series, with applications in dendroclimatology and hydrometeorology. Journal of Climate and Applied Meteorology 23: 201–213.

116.

Xoplaki

Gonzalez-Rouco

Gyalistras

. (2003a) Interannual summer air temperature variability over Greece and its connection to the large-scale atmospheric circulation and Mediterranean SSTs 1950–1999. Climate Dynamics 20: 537–554.

117.

Xoplaki

Gonzalez-Rouco

Luterbacher

. (2003b) Mediterranean summer air temperature variability and its connection to the large-scale atmospheric circulation and SSTs. Climate Dynamics 20: 723–739.

118.

Xoplaki

Gonzalez-Rouco

Luterbacher

. (2004) Wet season Mediterranean precipitation variability: Influence of large-scale dynamics and trends. Climate Dynamics 23: 63–78.

119.

Xoplaki

Maheras

Luterbacher

(2001) Variability of climate in Meridional Balkans during the periods 1675–1715 and 1780–1830 and its impact on human life. Climatic Change 48: 581–615.

120.

Zaidman

Rees

Young

(2002) Spatio-temporal development of streamflow droughts in northwest Europe. Hydrology and Earth System Sciences 6: 733–751.

121.

Zveryaev

Allan

(2010) Summertime precipitation variability over Europe and its links to atmospheric dynamics and evaporation. Journal of Geophysical Research-Atmospheres 115: D12102 12110.11029/12008jd01121312110.11029/12008jd011213.

A pan-European summer teleconnection mode recorded by a new temperature reconstruction from the northeastern Mediterranean ( ad 1768–2008)

Abstract

Keywords

Introduction

Material and methods

Study area

Tree ring analysis

Climate data, analysis, and reconstruction

Results

Climate–growth relationships

August temperature reconstruction

Atmospheric circulation patterns over the eastern Mediterranean region

Spatial correlations of the Sofia temperature reconstruction PIMO

Discussion

Conclusion

Footnotes

Acknowledgements

Funding

References