Keywords

1 Introduction

This section presents a summary of the results of climate reconstruction for Poland. For the first time in Poland, these results were investigated by an interdisciplinary team of scientists (geographers, dendrochronologists and mathematicians) along with historians and archivists. The group made, among other things, a reconstruction of climate based on a long Scots pine chronology. Historians collected a few thousand records of source materials related to climate. These data were collected and quantified in such a manner so as to make them comparable with the time scale made by the dendrochronologists (Przybylak et al. 2005). This study presents a background to Polish dendrochronology along with the dendrochronological proxy data derived by the paleoclimatological team, focusing on the dendroclimatological potential. The main dendrochronological data in Poland consist of the results of tree rings width measurements of native species, and a number of introduced species that have become naturalised, in Poland, and chronologies built from them. Some wood density measurements were also available for dendroclimatological analyses in the mountainous areas (Büntgen et al. 2007). Nearly all the studies looked at the relationships of growth with temperature and precipitation values, and occasionally other parameters were considered too (insect outbreaks, site characteristic etc.). In many of the dendroclimatological studies “moon rings”, called also “included sapwood”, a type of discoloration in the heartwood, (Krąpiec 1999) and “frost rings” were also investigated (Cedro 2004). Such anatomical anomalies are useful for recording of the occurrence and frequency of extreme weather phenomena.

The climate of Poland is described as moderately warm, transient between Atlantic and continental. Mean summer temperature varies between 16.5°C and 20°C (except in the mountains), and mean sum of precipitation is about 600 mm per year. Relief plays a significant influence on the variation in climate; lower temperatures are characteristic of the mountainous area of the Sudety and the Carpathian Mountains. In the highlands the growth/climate relationship is quite specific. In the higher areas the influence of temperature increases and that of precipitation decreases. At elevations above 1,000 m a.s.l. the influence of sunshine as a growth-stimulating factor is higher in the summer, replacing the role of low temperatures limiting growth at such elevations.

Many other factors influence the climate-growth relationships in stands of trees, these include long-term trends connected with age, and both endo- and exogenous factors. Because of this, in dendroclimatological analysis, tree-ring data are transformed into residual chronologies, enhancing the low-frequency variance and greatly reducing any autocorrelation and thus emphasising the climate signal in the series. Dendroclimatological research has been carried out in Poland for over 60 years (Zinkiewicz 1946; Ermich 1953; Feliksik 1972; Bednarz 1976). As a result, there are numerous chronologies available for both native and naturalised species which form a basis for further research. These are summarised in Table 7.1.

Table 7.1 Climatic signal in selected tree species in Poland
Full size table

Below we present the longest chronologies for each species, significant for wood dating. The longest chronologies have been constructed for:

  1. 1.

    Oak (Quercus robur L., Q. petraea LIEBL.)

  2. –

    Gdańsk Pomerania, 996–1985 AD (Ważny 1990)

  3. –

    Szczecin Pomerania 578–1393 AD and 1554–1994 AD (Ważny 1999), 553–1340 AD Zielski and Krąpiec 2004

  4. –

    Central Poland 593–965 AD (Ważny 1999)

  5. –

    Archaeological site in Biskupin 887–722 BC (Ważny 1993)

  6. –

    Warmia and Mazury 1093–1665 AD, 1695–2003 AD (Krąpiec et al. 2005)

  7. –

    Pułtusk region 1066–1462 AD (Krąpiec 1992)

  8. –

    Kalisz-Sieradz region 422–1027 AD, 1148–1316 AD (Zielski and Krąpiec 2004)

  9. –

    Lesser Poland 910–1997 AD (Krąpiec et al. 1998)

  10. –

    Low Silesia 780–1994 AD (Krąpiec et al. 1998)

  11. –

    Great Poland 449–1994 AD (Krąpiec et al. 1998)

  12. –

    Subfossil oaks in southern Poland 1795–612 BC, 474 BC–1555 AD (Krąpiec 1996, 2001)

  13. –

    Kuyavia and Pomerania region 1085–1584 AD (Zielski and Krąpiec 2004)

  14. 2.

    Scots pine wood (Pinus sylvestris L.)

  15. –

    Vistula valley – Pomerania region 1106–1994 AD (Zielski 1997)

  16. –

    Great Poland 1153–1700 AD and 1786–2001 AD (Zielski and Krąpiec 2004)

  17. –

    Suwałki Region 1582–2004 AD (Szychowska-Krapiec and Krąpiec 2005a)

  18. –

    Masovia 1176–1408 AD and 1652–1783 AD (Krąpiec et al. 2005)

  19. –

    North-eastern Poland 1110–1460 AD (Szychowska-Krapiec and Krąpiec 2005b)

  20. –

    Warmia and Mazury 1081–1408 and 1410–2003 (Krąpiec et al. 2005)

  21. –

    Low Silesia 1062–1418 AD and 1648–1999 AD (Szychowska-Krapiec and Krąpiec 2001)

  22. –

    Lesser Poland 1091–2004 AD (Szychowska-Krąpiec 2009)

  23. 3.

    Fir wood (Abies alba Mill.)

  24. –

    In its natural range in Poland 1106–1998 AD (Szychowska-Krąpiec 2000)

  25. –

    Historical wood, Carpathian mountains 1406–1746 AD (Ważny 1999)

  26. 4.

    Norway spruce wood (Picea abies Mill.)

  27. –

    Beskid Żywiecki 1641–1995 AD (Szychowska-Krąpiec 1998)

  28. –

    Białowieża National Park 1785–1999 AD (Koprowski and Zielski 2006, 2008)

  29. –

    Babiogórski National Park 1650–1993 AD (Bednarz et al. 1998–1999)

  30. –

    Tatrzański National Park 1699–1978 AD (Feliksik and Schweingruber ITRDB)

  31. –

    Tatra region in Poland and Slovakia 1628–2004 AD (Büntgen et al. 2007).

Dated chronologies have therefore been made for most native coniferous and ring porous dicotyledonous trees in Poland, and these have proved to be very useful for dendroclimatological studies.

Dated wood has also been used for broader climatic studies. For example, similarities and differences between chronologies from living pine and spruce chronologies have been used to specify climatically different forestry regions (Zielski et al. 2001; Wilczyński et al. 2001; Koprowski and Zielski 2006). In recent studies, subfossil wood of Taxodioxylon taxodii from the Miocene period has also been used in paleoclimatological research in southern Poland (Kłusek 2006).

2 Material and Methods

This section presents the basic methodology used by the present authors. Other researchers may have used different methodologies, and these will be detailed in their publications.

2.1 Sampling

In most cases the material for dendroclimatological research is collected as cores from living trees. A basic tool for taking the samples is a Pressler borer. In some cases samples in the form of disks have been taken from the stem. Historical wood has either had cross-sectional disks removed, or cores extracted using a special drill for dry timber. A special technique is needed for taking research material from subfossil wood (for example so called black (bog) oaks) when a chainsaw is used, or from fossil (Miocene) wood where polished sections are used.

2.2 Material

For temperature reconstruction in Poland, a residual regional chronology of Scots pine tree-ring widths from the Lower Vistula region (Fig. 7.1) covering the period 1168–1994 has been used (Zielski 1997). The chronology included measurements of tree-ring widths in living trees (back to 1767) and historical timber (wooden frames and wooden roof constructions from old churches and buildings).

Fig. 7.1
figure 1_7

Location of source data used to reconstruct temperature in Poland. Long-term series of air temperature; (1) Gdańsk; (2) Bydgoszcz; (3) Warsaw; circle: area for which tree-ring widths regional chronology of the Scots pine was constructed (Przybylak et al. 2005)

Full size image

2.3 Methods

2.3.1 Measurement and Basic Statistical Methods

The samples were treated in the standard way and measured to the nearest of 0.01 mm by means of a mechanical instruments with a computer registering the ring width. A number of methods were used to assess the cross-matching between the samples:

  • Gleichläufigkeit (% GL) (Huber 1943)

  • The Student’s t-test (Baillie and Pilcher 1973)

  • Program TREE-RINGS (Krawczyk and Krąpiec 1995)

  • Program CATRAS (Aniol 1983)

  • Program COFECHA (Holmes 1986; Grissino-Mayer 2001)

  • The skeleton plot method (Douglass 1939; Schweingruber et al., 1990; Zielski and Krąpiec 2004)

The residual chronologies for dendroclimatological analysis were built by program CRONOL (Holmes 1984). The climate-growth relationships were calculated by means of program RESPO (Holmes 1984) and program PRECON (Fritts 1996). This last program applies a bootstrap response function to estimate the error using random sampling for the data (Guiot 1993). The response function methods and program PRECON have been described in detail by Fritts (1976, 1996) and Briffa and Cook (1990). Hierarchical cluster analysis (HCA, program STATISTICA) was used to distinguish regions with similar increment pattern for Scots pine and Norway spruce.

Firstly, local chronologies were established. The correctness of the construction of those chronologies was checked using program COFECHA (Holmes 1986). In the next step, these chronologies were transformed using program CRONOL (Dendrochronology Program Library – DPL, routine CRN; Holmes 1984). The detrending procedure was applied to each of the examined series. Application of the autoregressive procedure to the detrended tree-ring series produced a residual version of the chronologies.

2.3.2 Signature Years

Cropper’s (1979) algorithm was used in order to detect extreme wide and narrow rings, which identify extreme weather conditions. A five-year moving window was applied, according to formula:

$$z_i = \frac{x_i - mean[window]}{stdev[window]}$$

where:

z i – index value in the year i

x i – original value in the year i

mean [window] – arithmetic mean of the ring width within window

xi−2, xi−1, xi, xi+1, xi+2

stdev [window] – standard deviation of the ring width within the window xi−2, xi−1, xi, xi+1. xi+2

The selection of extremely wide and narrow rings demands threshold values of zi. For example all values zi < −1.25 are defined as negative pointer years and all values of zi > 1.25 are defined as positive pointer years (Meyer 1998–1999). Information about climate originating from historical (documentary) sources was used to explain the occurrence of extreme years in pre-instrumental times. The frequency of occurrence of pointer years were counted in the periods 1177–1994 for pine (Fig. 7.2) and 1843–1999 for spruce (Fig. 7.3) (Wójcik et al. 2001).

Fig. 7.2
figure 2_7

Frequency of extreme years of Scots pine for 50 year periods (Wójcik et al. 2001)

Full size image
Fig. 7.3
figure 3_7

Frequency of extreme years of Norway spruce for 50 year periods (Wójcik et al. 2001)

Full size image

2.3.3 Reconstruction

Temperature and precipitation were reconstructed from pine ring-width curves through multiple regression analysis. These studies showed that there was a strong correlation between growth and climate in particular months (Fig. 7.4), and this varied between species as a result of their different genetic make-up. The conclusions are that reliable reconstructions can only be produced for the months where the relationships are seen to be statistically significant. Temperature and rainfall reconstructions from Scots pine tree-ring series were produced for the Kuyavia and Pomerania regions of Poland (Figs 7.5 and 7.6). The highest correlations between temperature and growth were found in February and March, and those for rainfall occurred in June and July. The results of the reconstructions were smoothed using an 11-year moving window. The indexed chronology (with reduced long-term growth trends) POLSKANE (Zielski 1997) was used in these analyses. The model was calibrated using data for the period 1921–1970 and verification was carried out on data for the period 1871–1920. The temperature reconstruction was made on the basis of measurements from the meteorological station in Bydgoszcz along with series of mean values for the region, derived from four stations: Chojnice, Płock, Bydgoszcz and Toruń. The process resulted in the model between growth and climate being described by means of linear regression:

Fig. 7.4
figure 4_7

Toruń region. Climate/growth dependence for tree rings of Scots pine, period 1891–1991. Line- multiple regression, columns-linear correlation. Marks on line or grey columns mean significant dependence between ring widths and selected months (counted by RESPO programme, DPL package) (Zielski and Krąpiec 2004)

Full size image
Fig. 7.5
figure 5_7

Comparison mean monthly temperatures (from January to April in years 1861–1993) with pine residual chronology for Toruń region (Zielski and Krąpiec 2004)

Full size image
Fig. 7.6
figure 6_7

Reconstructed and smoothed mean monthly temperatures from February to March in Kujawy and Pomorze region in years 1168–1870 (Zielski and Krąpiec 2004)

Full size image

index temperature-precipitation =15.556*c + 0.278 (Zielski and Kamiński 2003).

In order to characterize the reconstructed climate and improve the comparison with historical data, categories of seasons were distinguished (cool, warm, normal, wet and dry);

  • For mean regional air temperature

Criteria

Number of points

Characteristic of season

x < m − 1 SD

1

Cool

m − 1SD < x < m + 1SD

2

Normal

x > m + 1SD

3

Warm

Explanations: x – mean air temperature of February and March in a given year, m – mean air temperature of February and March in the years 1861–1990, SD – standard deviation of mean air temperature of February and March in the years 1861–1990

  • For precipitation (June and July) from Bydgoszcz

Criteria

Number of points

Characteristic of season

y < m − 1 SD

1

Dry

m − 1SD < y < m + 1SD

2

Normal

y > m + 1SD

3

Wet

Explanations: y – mean total of precipitation from June to July in a given year, m – mean total of precipitation from June to July in the years 1861–1990, SD – standard deviation of precipitation total from June to July in the years 1861–1990.

In the next stage five categories of years were selected, each on the basis of temperature at the end of winter and the beginning of spring and precipitation levels at the end of spring and beginning of summer:

  1. 1.

    February–March cool and June–July dry

  2. 2.

    (a) February–March normal and June–July dry; or (b) February–March cool and June–July normal

  3. 3.

    (a) February–March normal and June–July normal; or (b) February–March warm and June–July dry; or (c) February–March cool and June–July wet

  4. 4.

    (a) February–March warm and June–July normal; or (b) February–March normal and June–July wet

  5. 5.

    February–March warm and June–July wet

The results from this analysis for the period 1861–1990 were entered into the calibration of the dendrochronological model in a similar way to those used previously for the temperature model alone. Like the results of the temperature analysis, the agreement of the reconstructed series produced by this means with the known meteorological data was very strong, with a coefficient of correlation of 0.702 in years 1921–1970 (Kamiński 2002; Zielski and Kamiński 2003).

2.3.4 Others

Other reconstructions from Polish pine chronologies have used the Principle Component Analysis (PCA) method and program CATRAS (Aniol 1983). Indexed chronologies were reduced to the form of a variance matrix of new variables, so-called eigenvalues. Finally, a new set of variables were obtained for each time series, the so-called principal components. The first two principal components were used for subsequent interpretation, as these contained a significant proportion of the overall variance.

3 Results and Discussion

The calculations of the statistical relationships from calibration period 1921–1970 proved that temperature in selected months explains slightly more than 50% of pine growth variation (correlation coefficient is 0.725). The verification process shows that temperature from 1871 to 1920 produced by means of the linear regression are rather very well correlated with the actual measured temperatures data (correlation coefficient is 0.49). The graph of reconstructed temperatures (Fig. 7.6) is therefore quite reliable. The y axis represents reconstructed temperatures values, and the x axis, calendar years. It is possible to read the reconstructed winter/spring temperature in degrees Celsius. This season is generally characterized by temperatures below 0 degrees. The graph highlights unusual growth periods, for example the end of the twelfth century, the sixteenth to the seventeenth centuries, and the nineteenth century.

Even during the not typical periods, some extreme years were also noted. Warmer periods (mean temperature > 0.5°C) and cooler periods (mean temperature ≦ -1°C) can be distinguished. The reconstruction data suggest periods when the winter/spring temperature was warmer than normal in 1180–1200, 1240–1270, 1340–1360, 1430–1490, 1530–1590, 1660–1680, 1820–1850, 1910–1940 and from 1985. In contrast, cooler winter/spring periods occurred in: 1290–1310, 1400–1420, 1500–1510, 1600–1650, 1750–1770, 1800–1810, 1880–1900, and 1900–1980. The coldest winter/spring periods were the first decade of the fourteenth century, the last decades of the fifteenth century and at the beginning of sixteenth century, as well as in the first decade of the nineteenth century.

Years, classified by index 4 and 5, on the graph are above value three (Fig. 7.7). Values below 3 represent time spans with prevailing indexes from 1 to 2. From the analysis of raw data results, that in years 1168–1870 point out prevailing of average years (Fig. 7.7).

Fig. 7.7
figure 7_7

Reconstructed and smoothed indexes of selected seasons categories in Kujawy and Pomorze region in years 1168–1870 (Zielski and Krąpiec 2004)

Full size image

For Scots pine (Pinus sylvestris), the most common tree species in Poland (about 70% of area of Polish forests), local chronologies were built covering the whole of Poland and from these it is possible to recognize spatial variance in the increment patterns, and hence suggest dendrochronological regions, in which the trees are responding in a similar way to environmental factors. Some 330,000 tree-ring widths were measured on 5,400 cores, representing 2,800 trees. Local chronologies were built for 136 sites distributed throughout Poland. Local chronologies (Fig. 7.8a) are characterized by many individual features resulting from differences in age and genetic features of the tree, soil conditions etc. Indexed chronologies (Fig. 7.8b) were constructed using program ARSTAN (Cook and Holmes 1996). Indexation increased the short-term (annual) variance of chronology, the main cause of which was, undoubtedly, climate conditions. As a result, Poland was divided into eight homogenous dendrochronological regions (Fig. 7.9) (Wilczyński et al. 2001; Zielski et al. 2001). The selected regions cover most climatic regions of Poland.

Fig. 7.8
figure 8_7

(a) Raw data and (b) residual chronologies of Scots pine for sites from Poland (Zielski et al. 2001)

Full size image
Fig. 7.9
figure 9_7

Sites location and established homogenous regions of Scots pine increment pattern in Poland (Wilczyński et al. 2001)

Full size image

Temperatures in the winter months and in the beginning of spring were factors which differentiate the pine chronologies in a west-east direction, whereas the summer temperature/rainfall conditions vary on a north-south axis.

Similar research was carried out on the second most important tree species in Poland, the Norway spruce (Figs. 7.10 and 7.11). Region 1a grouped trees growing outside the natural range, and region 1b, trees from the natural boreal-Baltic area. Region 2a groups spruce sites in the so called “spruceless area”, whereas 2b is a region of the Hercynian-Carpathian distribution of Picea abies. Comparison of Figures 7.9 and 7.10 shows that homogenous groups of pine overlap with homogenous spruce groups. It is therefore possible to conclude that a combination of climatic and physical-geographical features control the formation of similar growth areas for both species within the Polish lowlands.

Fig. 7.10
figure 10_7

Sites localisation (o marked), established homogenous regions, and distribution of meteorological stations (x marked) (Koprowski and Zielski 2006)

Full size image
Fig. 7.11
figure 11_7

Typical climate-growth relationships for each region. Significant values (confidence level 95%) for particular months are marked above (counted by PRECON programme) (Koprowski and Zielski 2006)

Full size image

Bednarz (1976) revealed variations in seasonal climate from Pinus cembra tree rings in the Tatra mountains (Fig 7.12). The first half of the eighteenth century stands out as a period of strong continentality, and the beginning of decline of Arolla pine (Pinus cembra) tree-ring widths, indicating an increasing drop in temperature. This fact is supported by historical data showing the spread of glaciers in the Tatra mountains at this time (Maruszczak 1991).

Fig. 7.12
figure 12_7

Rhytm of thermal changes in years 1740–1956 in Pinus cembra tree rings from Tatra mountains, dotted line – consecutive 10-years means, solid line – tree ring widths. According to Bednarz data (1976) (Maruszczak 1991)

Full size image

One of the most important developments in the dendroclimatological research has been the introduction of X-ray densitometry, which revealed the strong correlation between the latewood density of coniferous trees and summer temperatures. These results are particularly striking in the cases where trees are growing in extreme conditions, for example on the higher elevations in mountains, and in the subarctic zone, close to the polar border of forests (northern treeline), where the most important factor limiting tree growth is climate. On these sites it has been shown that summer temperature is the most important factor controlling the tree-ring widths of coniferous species. European studies have shown that summer (April to September) temperature during the period 1750–1850 were particularly significant in limiting tree-ring widths, especially in the northern half of the continent. It was shown that “summer” temperature in Europe from 1812 to 1816 and in the 1830s were lower than the long-term mean (Briffa et al. 1988). In the Tatra mountains, in Poland and Slovakia, this research was done by Büntgen et al. (2007) (Figs. 7.13 and 7.14).

Fig. 7.13
figure 13_7

Tatra June–July and April–September temperature reconstructions. The tree-ring widths (a) and the maximum latewood density (b) (Modified Büntgen et al. 2007)

Full size image
Fig. 7.14
figure 14_7

Comparison of the observed (black) and predicted (grey) temperatures, after scaling (a) the tree-ring width (TRW)-based mean chronology against June–July temperatures, and (b) the maximum latewood density (MXD)-based mean chronology against April–September temperatures (1901–2002). Temperatures are expressed as anomalies from the 1961–1990 mean. Correlations were significant at the 99.9% confidence level (Modified Büntgen et al. 2007)

Full size image

For example Figure 7.15 shows a temperature reconstruction for the months January to April for 1170–1994 created from Scots pine chronologies (Przybylak et al. 2005). Figures 7.16 (spruce) and 7.17 (pine) present indexed chronologies and extreme years, produced using a moving window method (Wójcik et al. 2001). The frequency of wide and narrow tree-rings in the chronologies show long periods of growth above the long-term mean, for example pine in the fourteenth century, the end of the nineteenth century, and below long-term mean growth, for example, in the decades after 1940. Similar results were obtained for Norway spruce.

Fig. 7.15
figure 15_7

Reconstruction of mean January–April air temperature (oC) in Poland for the period 1170–1994 using a standardised chronology of Scots pine tree-ring widths (Pinus sylvestris L.) (Modified after Przybylak et al. 2005) Rek-ll, Rek-50 - 11- and 50-year running means; reconstruction using areally-averaged air temperatures from Warsaw, Bydgoszcz and Gdańsk for calibration, Obs – mean January–April areally-averaged air temperature from Warsaw, Bydgoszcz and Gdańsk (Przybylak et al. 2005; Przybylak 2007)

Full size image
Fig. 7.16
figure 16_7

Master plot of extreme years of Norway spruce in north-eastern Poland (Wójcik et al. 2001)

Full size image
Fig. 7.17
figure 17_7

Master plot of extreme years of Scots pine in north Poland (Wójcik et al. 2001)

Full size image

4 Conclusions

The reconstruction of temperature conditions for Poland, based on tree-rings of coniferous trees (pine, spruce and fir) have been presented, and the high dendroclimatological potential of other trees growing in Poland was also shown. As yet relatively few reconstructions have been carried out. One reason for this is that during the 1980s when such research was being carried out elsewhere, long historical and archaeological chronologies were not available in Poland. Such data have commercial application in wood dating, and this sometimes leads to problems in the sharing of information.

Given that there are observed relationships between temperature variation and tree-ring chronologies, the most useful potential for climatic reconstructions may be the multi-millennial oak chronologies from natural sites such as the alluvial deposits of the River Vistula, and from archaeological sites, such as the Hallstat era settlements in Biskupin.

Analysis of climate/growth relations was also conducted with other, non-native species. It is worthwhile remember that this kind of research was carried out on Douglass fir, growing on many sites in Poland (Feliksik and Wilczyński 2004). Reaction of this tree to climate is similar to related with Douglass fir, together in Pinaceae family, Scots pine and can also proved about well adaptation of Pseudotsuga to environmental conditions of Poland. These observations can have an important bearing in discussions about the possible prediction of forest composition in the presence of prejected climate changes.