Reconstruction of Summer Rainfall over the Last Five Centuries Based on Oak Chronology (Western Pomerania, Poland)

Abstract

The quantity and distribution of summer rainfall in Poland is one of the main elements of weather and makes a strong impact on the economy, mostly agriculture, stockbreeding, and forestry and the associated industries. Droughts or heavy rains, occurring increasingly more frequently in summer, are a threat for human activity. This study presents a nearly 500-year-long reconstruction of precipitation in the June–July period for northwestern Poland based on an analysis of tree-ring widths in the native oak species (Quercus spp.) from 24 study plots located in Western Pomerania. Due to the frequent droughts occurring in the last four decades, and due to strong air pollution in the 1960s–1990s, we chose the period 1901–1941 as the calibration period. The performed reconstruction is characterized by a high annual variability in precipitation sums for June and July of the year of the tree-ring formation from 1565 to 2020, with an average rainfall sum for this period equal to 131.7 mm (standard deviation: 28.4 mm). Periods with rainfall shortages in summer occurred in the following years: 1579–1596, 1629–1637, 1650–1655, 1669–1672, 1703–1706, 1739–1748, 1757–1760, 1765–1768, 1808–1811, 1838–1841, 1856–1859, 1958–1961, 1965–1968, 1981–1983, and 2002–2006. Periods with higher than average rainfall in June and July occurred in the following years: 1573–1578, 1605–1609, 1613–1617, 1638–1642, 1694–1698, 1776–1780, 1791–1794, 1828–1831, 1852–1855, 1863–1866, 1877–1879, and 1944–1946. Our reconstruction was verified based on the historical records, available from the XVI century onward. Not all the reconstructed summer precipitation values, however, were confirmed by historical records. Notably, however, the historical data are often incomplete and imprecise. Further, the tree-ring width in the oak depends not only on the precipitation in June and July but also, e.g., on the pluvial conditions late in the previous growth season.

1. Introduction

By definition, climate varies both spatially and temporally. Individual climate elements may display multi-year trends, mid-term fluctuations, and annual variability. Distinctive climatic phases are widely recognized, such as the Holocene Climatic Optimum, the warm Middle Ages, or the Little Ice Age cooling [1,2]. The contemporary changes, however, differ from those observed over the last several millennia. Individual weather elements and weather phenomena have been observed and recorded since about 200 years ago [3,4]. These results can be used to monitor climate changes [5,6,7]. The evident global climatic changes observed through the last century, attributed by many researchers to human activity, are detrimental to the environment, causing habitat destruction, shifts in plant and animal species occurrence ranges, and disruptions to energy and matter cycling in the geosystem [8,9,10,11]. For these reasons, climatic changes are observed, and their impact on ecosystems may be studied, while environmental change scenarios should be prepared based upon these observations and studies. Such forecasts are founded on the recognition of the response of selected ecosystem elements, e.g., trees, to past changes. This is why climate reconstructions based on the analysis of tree-ring widths, resulting from the response to changing growth conditions, including climatic conditions, are a valuable source of information on past climate changes. Air temperature is frequently reconstructed based on dendrochronological time series [12,13,14,15,16]. Coniferous trees growing outside their ecological optimum are especially well-suited for such studies [17,18]. Precipitation reconstructions are undertaken considerably less frequently [1,19,20] because this element is characterized by high temporal and spatial variability, which makes it much more difficult to analyze and reconstruct. Reconstructions of precipitation for various geographic regions are therefore very valuable from the perspective of ecological requirements of numerous taxa, for which rainfall is the main growth-limiting factor. Most precipitation reconstructions are limited to an analysis of changes in a multi-year perspective [21,22,23], without verifying the obtained results against historical records [24,25,26]. Thus, studies that include historical record queries are especially valuable.

This study aimed to (i) identify those climatic elements to which the native species of oak (Quercus robur L. and Q. petraea (Matt.) Liebl.) displayed growth sensitivity in Western Pomerania (NW Poland); (ii) assess the temporal stability of the sensitivity of the oak to main climatic elements shaping its growth reactions; (iii) reconstruct the identified climatic elements determining tree-ring width in the oak over the last five centuries in Western Pomerania; and (iv) verify the correctness of the reconstruction against historical records.

Considering the high sensitivity of the oak to precipitation as the main factor determining the annual variability in its tree-ring width, as is well known from the available literature [27,28,29,30], we assumed that precipitation should be the aim of the reconstruction.

2. Materials and Methods

2.1. Study Area

Western Pomerania is located in the northwestern part of Poland. It is bordered by the Baltic Sea to the north and by Germany to the west, across the Odra River. It is a young glacial area that was occupied by ice sheets several times during the last hundred thousand years. The last glaciation took place about twenty-two thousand years ago, and the ice sheet retreated about fourteen thousand years ago. Terrain forms that resulted from ice sheet melting are well-preserved in the surface relief and form a typical post-glacial landscape with numerous lakes, frontal moraine belts, outwash plains, plateaux, kame terraces, and eskers. The bedrock is dominated by tills, fluvioglacial sediments, and ice-dammed lake clays. Some sediments (mostly in ice marginal valleys) have undergone aeolian transport and now form extensive dune fields [31]. The studied oak trees are growing in fresh mixed forests and fresh forests, on cambisols and chromisols, often in multi-species forests. The oldest trees (>400 years old) are growing on the Wolin Island (northern part of the study area), within the Wolin National Park (Figure 1). This area is a frontal moraine, characterized by considerable differences in elevation, and bordered by a high cliff on the seaward side.

The analyses were performed on native oak species: Quercus robur L. and Q. petraea (Matt.) Liebl. Both species occur in large numbers in Western Pomerania, and their hybrids are also recorded (Q. x rosacea Bechst.). Q. robur has greater habitat and humidity requirements than Q. petaea; it is also a long-lived tree and is one of the most powerful trees in Poland. Q. petraea occupies drier and warmer locations, tolerates acidic soils well, and is one of the main forest-forming species in the region [32,33]. The oldest oak trees are few specimens growing as a remainder within a considerably younger stand. From the remainder of the study area, we selected old stands characterized by good health and growth dynamics. These oaks are growing both in continuous stands and in isolated mid-field enclaves.

2.2. Climate and Climate Data

Air temperature and precipitation variability for Szczecin have been compiled in an annual and multi-year perspective, considering the rate of change, determined using a linear trend coefficient. The air temperature dataset is based on monthly mean values from the years 1931–2023 (i.e., a 93-year-long period) from the IMGW Hydro-Meteorological Station in Szczecin (53°24′ N, 14°37′, 1 m a.s.l.). Monthly mean rainfall sums were compiled for the years 1861–2023 (163 years). Meteorological data for air temperature and rainfall sums come from various sources. Those from before World War 2 come mostly from Klimakunde des Deutches Reiches, Bd. II (1939) [34], referenced also in Polish literature [35,36]. Data from the period after World War 2 were compiled from annals and archival materials of the Institute of Meteorology and Water Management (IMGW).

In the Köppen-Geiger climate classification [37,38], Poland, including Szczecin, is classified within the humid continental climate, with a mild summer and precipitation through the whole year (Dfb). Notably, in Szczecin, the climate features are modified by the influence of the Baltic Sea and the Atlantic Ocean and by the circulation from above the Szczecin Lagoon. Air masses from over the ocean, sea, and continent collide in this area. As a result, the weather conditions are highly variable.

2.2.1. Air Temperature

Mean annual air temperature in Szczecin equaled 8.8 °C and varied through the study period, from 6.3 °C (in l940) to 10.8 °C (in 2019) (Figure 2). A clear cooling occurred from 1940 to 1988. Up to the year 1988, in 34 years (36.5% years), mean annual air temperature was lower than the multi-year mean, and after 1988, the same was true only for 3 years. Mean annual air temperature exceeded a value of 10 °C in 12 years, including 9 years after 2007. The warmest month in Szczecin was July (18.3 °C). In individual years, a temperature maximum also occurred in August and in June. The coldest month was January (−0.5 °C) (Figure 3). A mean monthly minimum also occurred in February and in December. In order to quantify the temporal air temperature variability, the linear trend coefficient was computed. In the studied period 1931–2023, the linear trend coefficient value equaled y = 0.0201 °C year⁻¹, which means that mean annual air temperature in Szczecin rose by 1.9 °C over 93 years.

2.2.2. Precipitation

From 1861 to 2023, mean annual rainfall sum in Szczecin equaled 552.5 mm. The highest annual sum (867 mm, 157% of the multi-year mean) occurred in 1939. The lowest (351.0 mm, 63% of the multi-year mean) occurred in 1982. In the whole study period, in 51% of years, precipitation exceeded the multi-year mean (Figure 4). In an annual perspective, the mean multi-year rainfall maximum occurred in July (72.7 mm), while the minimum occurred in February (30.2 mm) (Figure 3).

In Szczecin, apart from high rainfall in summer (191.7 mm), high rainfall occurred also in autumn (130.8), caused by an advection of warm and humid air from over the Atlantic Ocean and the Baltic Sea, frequently occurring in this season (cyclonal activity is especially intensive over the Southern Baltic in autumn). Additionally, high rainfall occurs in spring (118.9—in all the spring months), despite the fact that the cooling influence of the adjacent aquatic bodies does not favor high precipitation. This means that local conditions occur in Szczecin, favoring an increase in precipitation, especially in May (47.8 mm). At this time, there is a clear increase in air temperature, which enhances thermal contrasts. Warm substratum may favor the development of convection, especially when cooler air inflows from over the Atlantic or the Baltic Sea which displays high instability in this period. An analysis of seasonality in the annual sum corroborates high variability in a multi-year perspective.

In Szczecin, the contribution of the warm half-year (May–October) precipitation to the annual sum from the whole multi-year period equals 59.7%. This varied in individual years from 38.2% (1949) to 84.2% (1996). In 21 years (12.9%) of the discussed period, the cool half-year rainfall was higher, which means that in these years, the rainfall variability displayed outstanding oceanic features. As the rainfall sums from the summer months were the highest, their contribution to the annual sum was determined at 34.5% (June—10.3%, July—13.2%, August—11.3%). The range of rainfall sum variability in the studied 163 years was large. In June, rainfall sums varied from 8.0 to 153.0 mm. In July, they varied from 5.0 to 203 mm. In August, they varied from 7.0 to 62.3 mm. In the study period, the rainfall sums lower than the multi-year mean represented 50.5% of years for June, 45.7% of years for July, and 51.6% of years for August.

In Poland, precipitation patterns display a high spatial and temporal variability due to both oceanic and continental influences (in the north of Poland, additionally, due to the influence of the Baltic Sea). The period 1861–2023 displays an increasing trend in the precipitation sum y = 0.1471 mm year⁻¹, equal to an increase in mean annual rainfall in Szczecin by about 24.0 mm over the period of 163 years (Figure 4). The strongest increasing trend occurred for winter (y = 0.1045 mm year⁻¹). A high increase in precipitation totals in winter was due to an increase in precipitation in all winter months, especially in January and February. A considerable increase in precipitation occurred also in spring (y = 0.0473 mm year⁻¹) and was caused by an increase in rainfall sums in May (y = 0.0630 mm year⁻¹). In March and April, however, decreasing trends occurred. An increasing trend occurred also for autumn (0.0196 mm year⁻¹). Among autumn months, however, a decreasing trend occurred for October. For summer, however, a negative trend occurred (y = −0.0245 mm year⁻¹). This was caused by decreased rainfall sums for July (y = −0.0370 mm year⁻¹) and August (y = −0.0452 mm year⁻¹). An increasing trend occurred for June, however (y = 0.0577 mm year⁻¹). Thus, an analysis of linear trend coefficients revealed that—from a monthly perspective—the strongest increase in precipitation occurred for May (rise of 10.3 mm over the period of 163 years). From a seasonal perspective, the highest increase in precipitation occurred for winter (rise of 17.0 mm/163 years). August was the month for which the highest decrease was observed (drop of −7.4 mm/163 years), and summer was the season for which we noted the greatest reduction in precipitation totals (drop of −4.0 mm/163 years).

2.3. Dendrochronology Data and Dendroclimatic Analyses

Samples were collected from individuals of native oak species using Pressler borers, at 1.3 m above ground, at 24 study plots located in Western Pomerania (a total of 358 trees), between 1997 and 2021 (Figure 1). In the laboratory, samples were mounted on boards, dried, and sliced using a surgical blade in order to obtain a clear view of the tree-rings. Tree-ring widths (TRW) were measured with an accuracy of 0.01 mm under a stereomicroscope, using the TREE RINGS [39] and LDB_Measure (ver. 1.0) [40] programs. All samples were subsequently dated using standard cross-dating methods. Student’s t-test and Gl coherence coefficient (Gleichläufigkeitswert) were computed for the dendrochronological sequences, and visual similarity of the sequence variability was assessed.

Based on this, we selected 72 sequences (from various sites in Western Pomerania), characterized by the highest statistical indices (t > 4.5 and Gl > 65%) and a high visual similarity. These were used to assemble a regional chronology (QUSP_WP, Quercus spp._Western Pomerania [41]), whose quality was tested using Cofecha, part of the DPL software package version 2023 [42]. Age trend and autocorrelation were removed from the sequences included in the regional chronology via an indexing process (a two-phase detrending technique, by fitting either a modified negative exponential curve or a regression line with a negative or zero slope, Arstan program). RES chronology, which emphasizes high-frequency variations, was selected for further analyses. The expressed population signal (EPS) and the mean inter-series correlation coefficient (Rbar) were computed for 50-year moving windows with 25-year overlaps. Rbar is a measure of the degree of homogeneity of annual growth reactions of trees and the strength of the high-frequency signal characterizing individual series. Higher Rbar values indicate a stronger shared signal. EPS shows to what degree a given chronology, assembled from these series, represents the entire population. In this study, we follow the recommended minimum threshold for the EPS index, equal to 0.85 [43], which ensures a sufficiently strong shared signal and provides justification for reconstructing climate based on these data.

For examining the growth–climate relationship, we used Respo, part of the DPL software package—correlation and response function [42]. The analyses were carried out separately for air temperature (T), precipitation (P), and insolation (IN) over the period of data availability: for T from 1931 to 2021, for P from 1861 to 2021, and for IN from 1965 to 2021 (for the entire time span and for 40-year periods). For the climate–growth relationship analysis, we used mean monthly values of temperature, monthly precipitation sums, and monthly sums of hours with sunshine. The analyses were performed for 16-month periods (from June of the preceding year to September of the growth year). In order to identify the key climate elements determining the radial growth of trees, we computed correlation (r), regression (b), and multiple regression determination (r²) coefficients, which determine the direction and strength of influence of a given climatic element on tree-ring width variability.

In order to verify the stability of the relationships among the climatic elements determining the oak growth, we performed an analysis of correlation in a 31-year window, systematically shifted by one year. Additionally, in order to assess the stability of climate–growth reactions in time, the study period for which climate data were available (1861–2020) was divided into 40-year subperiods: 1861–1900, 1901–1940, 1941–1980, and 1981–2020. For each subperiod, a tree-ring chronology was compared to the variability of the identified climatic element, and the relationship between the two data series was assessed. Next, based on the regression line parameters, the subperiod 1901–1941 was selected as the most appropriate for the description (calibration) of the growth–climate relationship. This was also because this subperiod preceded a period of dynamic increase in air pollution, which could significantly modify the growth reactions of the studied trees. Verification of the resultant linear regression model was based on the remaining three subperiods. For each subperiod, the significance of differences from zero of the obtained model residuals was verified with the help of one sample two-sided t-test. Based on the established linear regression model, we subsequently reconstructed the values of the identified climatic element for those years for which oak radial growth data were available (from 1564).

For precipitation reconstructions, we used a linear regression model for the interval 1901–1940, expressed by the following formula:

$$P_i=175.40\cdot TRW{I}_i-43.81$$

(1)

where Pi—precipitation sum in the June–July period of the year i of tree-ring formation and TRWI_i—indexed tree-ring width in year i.

The resultant reconstruction was verified based on an analysis of years indicated in historical records as extreme with respect to weather conditions. In order to emphasize the extreme reconstructed values of the identified climatic element, a z-score value was computed for each year chosen based on historical records. The Z-score value expresses the difference between the value in a given year and a multi-year mean in standard deviation units according to the following formula:

$$Z_i={\displaystyle \frac{P_i-P_{mean}}{P_{SD}}}$$

(2)

where Z_i—z-score value for year i, P_i—value of the reconstructed climatic element in year i, P_mean—mean value of the climatic element for the years 1861–2020, and P_SD—standard deviation of the climatic element value for the years 1861–2020.

3. Results

3.1. Regional Chronology

The regional chronology QUSP_WP is based on 72 individual growth curves derived from oak trees growing at 24 sites in Western Pomerania. The assembled chronology spans 458 years, representing the period from 1564 to 2021 (Figure 5). The mean tree-ring width equals 1.02 mm/year. The QUSP_WP chronology displays long-term fluctuations (Figure 5). Until about 1750, there is a clear decreasing trend, which is replaced by an increasing trend that lasts until about 1920. After that, another gradual, long-term decrease in tree-ring width is observed. Notably, a clear increase in tree-ring width is also observed in the last 20 years. The observed fluctuations are likely due to the diverse ages of the oak trees, whose tree-ring series were included in the assembled chronology, and from the impact of conditions within the habitats in which the trees grow. Having eliminated the observed long-term fluctuations, we obtained an indexed chronology, which highlights a year-to-year variability in tree-ring widths, determined mostly by the climatic factor.

3.2. Climate–Growth Relationships

Figure 6 shows the relationship between tree-ring widths and temperature (T) for the period 1931–2021 (91 years), precipitation (P) for the period 1861–2021 (161 years), and sunshine duration (IN) for the period 1965–2021 (57 years). We found that the clearly positive relationship between tree-ring width and rainfall sum (positive correlations) and temperature and sunshine duration (negative values of correlation and response function) in the period from June to July of the growth year is the most stable in time. We observe similar growth–climate relationships in August of the previous year. Since the available data on precipitation in the study area go as far back in time as 1861 and reveals a high variability in rainfall over the last 160 years, we performed further analyses of the relationship between June–July rainfall and tree-ring width in four 40-year intervals. The first interval (1861–1900) is a time of relatively small climatic changes and weak anthropogenic impact. The second interval (1901–1940) is a period of noticeable climate changes and an increasing level of industrial pollution emissions. The third interval (1941–1980) is mostly a time of strong impact of industrial pollution on the environment, including trees [44,45,46,47]. Finally, the last interval (1981–2020) is, on the one hand, a time of a clear reduction in industrial emission levels, but on the other hand, it is a time of very clear climatic changes [7]. The results of these changes, linked mostly to the observed increase in greenhouse gas concentrations, include temperature increase and extended vegetation season duration [48,49,50], but also an increase in the frequency and duration of drought periods [51,52], to which the oak is especially sensitive [53,54].

3.3. Climate Reconstruction

The variability of the indexed chronology is highly consistent with the variability in precipitation sums for the period June–July in each of the four intervals distinguished in our study period spanning 1861–2020 (Figure 7). The relationship between tree growth and precipitation sum in the June–July period may be regarded as constant through time (Figure 8 and Figure 9), and mean values of the residuals for the distinguished test intervals did not significantly differ from zero (

Table 1. Results of one sample two-sided t-test reflecting the difference of the residuals from zero for three verification periods. Estimate—residual mean of precipitation sum estimation in a period, Statistic—t test statistic value, p—value expressing the probability of making an error when rejecting the null hypothesis of test t, 95% CI—95% confidence interval for residual mean of precipitation sum estimation in the studied period.

Period (Years)	Estimate (mm)	Statistic	p	95% CI
1861–1900	3.69	0.55	0.588	[−9.96, 17.34]
1941–1980	3.68	0.60	0.555	[−8.82, 16.19]
1981–2020	−2.65	−0.30	0.763	[−20.31, 15.00]

), which justifies the use of the calibration period 1901–1940 as the basis for the performed reconstruction of precipitation in the June–July period.

The positive relationship indicates that a higher precipitation sum in the June–July period is required to form a broader than average tree-ring. The observed temporal stability of these relationships corroborates the possibility of reconstructing changes in precipitation in specific years based on the oak chronology.

The reconstruction performed here (Figure 10) displays a high annual variability of precipitation in the June–July period from 1565 to 2020, with the mean precipitation sum for this period equal to 131.7 mm and standard deviation equal to 28.4 mm. Furthermore, there are clear multi-year periods with excess rainfall and rainfall shortages (red line, Figure 10). Such periods are especially numerous between 1560 and 1870, and in the early 21st century. In the years 1573–1578, 1605–1609, 1613–1617, 1638–1642, 1694–1698, 1776–1780, 1791–1794, 1828–1831, 1852–1855, 1863–1866, 1877–1879, and 1944–1946, the reconstruction indicates higher than average rainfall for the months June and July. Conversely, in the years 1579–1596, 1629–1637, 1650–1655, 1669–1672, 1703–1706, 1739–1748, 1757–1760, 1765–1768, 1808–1811, 1838–1841, 1856–1859, 1958–1961, 1965–1968, 1981–1983, and 2002–2006, the reconstruction shows rainfall shortages in summer.

In order to verify the correctness of the reconstruction, we performed a case analysis. By cases, we understand the years characterized by especially high decreases or increases in our reconstruction, described as anomalous with respect to rainfall in historical records (

Table 2. Remarks on weather anomalies in the Pomeranian historical records in selected years, along with reconstructed values of precipitation sums for the June–July period and the corresponding z-score value. Heavy font face indicates z-score values lower than −1 or higher than +1.

June–July Precipitation Sum (Z-Score)	Year	Description of the Phenomena	Additional Background Information	Source
138.6 (+0.2)	1571	Cold, rainy spring, cold, rainy summer.		Eine Sammlung unterschiedlicher die Historie der Stadt Cöslin betreffende Sachen, Johan David Wendland, Koszalin 2006.
117.6 (−0.5)	1572	Dry, sunny summer.		Eine Sammlung unterschiedlicher die Historie der Stadt Cöslin betreffende Sachen, Johann David Wendland, Koszalin 2006.
180.7 (+1.7)	1586	Long, harsh winter with abundant snow. The Odra River froze overnight. Cold, rainy spring, cool, rainy summer.	Severe grain crop failure. Very high prices and hunger in Pomerania, severe grain shortages.	Gustav Henric Schwallenber, Historia Pomeraniae Pragmatica, no date (early 18th century) in: APS, RiS, 451, Geschichte von Belgard, 1938 in: APS, RiS, 1236, Historische Beschreibung der Stadt Alten Stettin in Pommern, durch Paulum Friedeborn, Altes Stettin 1613.
108.8 (−0.8)	1587	Long, harsh and snowy winter. Cold, rainy spring, cool, rainy summer.	Severe grain crop failure. Very high prices and hunger in Pomerania, severe grain shortages.	Gustav Henric Schwallenber, Historia Pomeraniae Pragmatica, no date (early 18th century) in: APS, RiS, 451, Geschichte von Belgard 1938 in: APS, RiS, 1236, Historische Beschreibung der Stadt Alten Stettin in Pommern, durch Paulum Friedeborn, Altes Stettin 1613.
171.9 (+1.4)	1589	Very dry and hot summer.	There are no previous data from 1588. In 1589, the heat made breathing difficult. Forests spontaneously ignited, rivers dried, mills stopped, and grain dried.	Historische Beschreibung der Stadt Alten Stettin in Pommern, durch Paulum Friedeborn, Altes Stettin 1613.
228.1 (+3.4)	1604	Mild, rainy winter, warm, rainy spring, warm summer.	There are no previous data from 1603. In 1604, abundant grains. First favorable year after 1600.	Historische Beschreibung der Stadt Alten Stettin in Pommern, durch Paulum Friedeborn, Alten Stettin 1613.
136.9 (+0.2)	1609	Long, harsh and snowy winter, rainy, cool spring.	There are no previous data from 1608. In 1609, grain crop failure.	Eine Sammlung unterschiedlicher die Historie der Stadt Cöslin betreffende Sachen, Johan David Wendland, Koszalin 2006.
175.4 (+1.5)	1617	Cold, rainy winter, cool, rainy summer.	There are no previous data from 1616. In 1617, bread shortages, high food prices, scarce fruit crops.	Filipa Hainhofera dziennik podróży zawierający obrazki z Frankonii, Saksonii, Marchii Brandenburskiej i z Pomorza w 1617 roku, analysis by Krzysztof Gołda 2nd edition, edited by R. Skrycki, Szczecin 2020.
150.9 (+0.7)	1628	Long, harsh winter, rainy, cold spring.	Grain crop failure.	LAG, Rep. 40, Handschriften, II/20
163.2 (+1.1)	1668	Mild, rainy winter, rainy, warm spring, good, mild summer.	Abundant grain crops, good crops also in Poland.	APS, Collection of Evangelical Parish Books from Pomerania and Neumark, 2.
131.6 (0.0)	1691	Long, harsh winter, cold and rainy spring, cold and rainy summer.	In 1690, rainy spring and summer. In 1691, grain crop failure.	APS, AKS, I/5320, I/5321.
136.9 (+0.2)	1692	Long, harsh winter, cold and rainy spring, cold summer.	Grain crop failure. Subjects complaining about bread shortages and great poverty. Reports of grain imports to Pomerania from other provinces of the Brandenburg state.	APS, AKS, I/5320, I/5321.
168.4 (+1.3)	1708	Long, harsh and snowy winter, cold, rainy spring.	Grain crop failure.	Eine Sammlung unterschiedlicher die Historie der Stadt Cöslin betreffende Sachen, Johan David Wendland, Koszlin 2006.
175.4 (+1.5)	1725	Cold, rainy spring and cool, rainy summer.	Severe grain crop failure.	Eine Sammlung unterschiedlicher die Historie der Stadt Cöslin betreffende Sachen, Johan David Wendland, Koszalin 2006.
164.9 (+1.2)	1733	Long and harsh winter, cold, rainy spring, cold summer.	In 1732, a peculiar year, various diseases and plagues, mild winter, frost only in February, then a lot of rain. In 1733, severe grain crop failure.	APS, AKS, I/5556, Eine Sammlung unterschiedlicher die Historie der Stadt Cöslin betreffende Sachen, Johan David Wendland, Koszalin 2006.
168.4 (+1.3)	1734	Long and harsh winter, cold, rainy spring, cold summer.	Severe rye crop failure.	APS, AKS, I/5556.
140.4 (+0.3)	1735	Long and harsh winter, rainy, cold spring.	Grain crop failure, grain shortages in the market.	Eine Sammlung unterschiedlicher die Historie der Stadt Cöslin betreffende Sachen, Johan David Wendland, Koszalin 2006.
149.1 (+0.6)	1738	Long, harsh winter, temperatures dropping to −32 degrees, cold, rainy spring.	In 1737, there was a huge shortage of food, people ate anything, hunger, rainy spring and summer. In 1738, severe grain crop failure, soaring prices.	Eine Sammlung unterschiedlicher die Historie der Stadt Cöslin betreffende Sachen, Johan David Wendland, Koszalin 2006.
131.6 (0.0)	1743	Rainy, cold spring, rainy, cool summer.	There are no previous data from 1742. In 1743, severe grain crop failure, soaring prices.	Eine Sammlung unterschiedlicher die Historie der Stadt Cöslin betreffende Sachen, Johan David Wendland, Koszalin 2006.
115.8 (−0.6)	1747	Rainy winter, cold, rainy spring, cool summer.	Severe grain crop failure, soaring prices.	Eine Sammlung unterschiedlicher die Historie der Stadt Cöslin betreffende Sachen, Johan David Wendland, Koszalin 2006.
171.9 (+1.4)	1752	Long, harsh winter, cold, rainy spring, cold, rainy summer.	In 1751, severe frosts winter, cold wet spring. In 1752, heating was required as late as St. John’s Day. Strong frosts lasted until late March. Grain crop failure.	Eine Sammlung unterschiedlicher die Historie der Stadt Cöslin betreffende Sachen, Johan David Wendland, Koszalin 2006.
186.0 (+1.9)	1754	Long, harsh winter, long, rainy and cold spring, cold summer.	In 1753, only mention of a harsh winter. In 1754, fields were covered by snow for a long time. Grain crop failure, soaring prices.	Eine Sammlung unterschiedlicher die Historie der Stadt Cöslin betreffende Sachen, Johan David Wendland, Koszalin 2006.
189.5 (+2.0)	1755	Long winter with a lot of snow, blizzards, strong frosts, cold, rainy spring, cold, rainy summer.	Severe grain crop failure (especially rye) caused by very low temperatures and a durable snow cover.	APS, AKS, I/5557, Eine Sammlung unterschiedlicher die Historie der Stadt Cöslin betreffende Sachen, Johan David Wendland, Koszalin 2006.
201.8 (+2.5)	1761	Short winter, rainy spring, warm summer.	Regular transports of grain to Szczecin from 28 January to 30 December 1761. Good conditions on waterways and roads.	Wochentlich—Stettinische Frag—und Anzeigungs—Nachrichten, 1761, No 1–52.
156.1 (+0.9)	1769	Mild, warm winter, summer with good crops.	Ship traffic in the harbor from 11 January throughout the year until January 1770, uninterrupted grain transports to the city.	Wochentlich—Stettinische Frag—und Anzeigungs—Nachrichten, 1769, No 1–52.
166.7 (+1.2)	1770	Mild, warm winter, warm summer with good crops.	Ship traffic in the harbor nearly throughout the year, except for a break from 3 January to 14 February, regular transports of grain to the city.	Wochentlich—Stettinische Frag—und Anzeigungs Nachrichten, 1770, No 1–52.
168.4 (+1.3)	1778	Harsh, short winter, warm, rainy summer, warm autumn.	Ship traffic ceased from late December 1777 to early April 1778. Intense ship traffic and grain transport throughout the summer. Ship traffic ended in late December 1778.	Stettiner Intelligenz Zettel, 1778, No 1, 3, 5, 7, 9, 13, 15, 17, 19, 21, 23, 25, 27, 29, 53, 55. 57, 59, 61, 63, 65, 67, 69, 71, 89, 91, 93, 95, 97, 99, 101, 103.
156.1 (+0.9)	1792	Short, relatively mild winter, warm, rainy spring, warm summer.	Ship traffic suspended from 21 December 1791 to 8 February 1792 and again from 22 February to 14 March 1792. Following that, both ship traffic and grain transports were regular. Due to positive temperatures, ship traffic in the harbor lasted until early January 1793.	Sttetiner Intelligenz Zettel, 1791, No 100, 102, 104, 1792, No 2, 4, 6, 8, 10, 12, 14, 18, 20, 22, 24, 54, 56, 58, 60, 62, 64, 66, 68, 70, 72, 88, 90, 92, 94, 96, 98, 100, 102, 104.
164.9 (+1.2)	1793	Short, mild winter, warm, rainy summer, warm, rainy autumn.	Ship traffic ceased on 2 January 1793 and resumed after 27 February. Following that, ship traffic was regular and vast quantities of grain were delivered to the city. The harbor functioned throughout December 1793 until 8 January 1794.	Stettiner Intelligenz Zettel, 1793, No 1, 3, 5, 7, 9, 11, 13, 15, 17, No 19, 53, 55, 57, 59, 61, 63, 65, 67, 69, 71, 73, 91, 95, 97, 99, 101, 103, 1794, No 1, 3, 5.
196.5 (+2.3)	1796	Mild, warm winter, mild, warm summer, abundant grain crops, warm autumn.	Ship traffic in the harbor continued throughout December 1795 and the winter months of 1796. In the summer months, the ship traffic in the harbor was intense, and large quantities of grain were delivered to the city. Ship traffic ceased on 7 December 1796.	Stettiner Intelligenz Zettel, 1795, No 95, 97, 99, 101, 103, 1796 No 3, 5, 7, 9, 11, 13, 15, 17, 19, 21, 53, 57, 59, 61, 63, 65, 67, 69, 71, 91, 93, 95, 97, 99, 101.
175.4 (+1.5)	1801	Harsh, relatively short winter, warm summer, mild, long autumn.	Navigation season began in mid-March 1801. Intense ship traffic in the summer months. In summer, variable grain imports. In autumn, high grain imports. Ship traffic lasted until late December 1801.	Stettiner Intelligenz Zettel, 1800, No 101, 103, 1801, No 1, 3, 5, 7, 9, 11, 13, 15, 17, 19, 21, 23, 53, 55, 57, 59, 61, 63, 65, 67, 69, 71, 73, 91, 93, 95, 97, 99, 101, 103.

).

3.4. Analysis of Historical Records

Our analysis of historical records mostly involved the linguistic method, based on a careful scrutiny of literature kept at the national archives in Szczecin and Greifswald. We also conducted an inner critique of the analyzed documents. The linguistic method was also employed while analyzing regional chronicles, also included in the research. Additionally, we used a comparative method when confronting the written records included in the analyzed sources with those indicated by statistical methods. The latter were used for an analysis of statistical materials included in the 18th century press published in Szczecin.

3.5. Verification of the Reconstruction Based on Historical Records

Essentially from the start of the development of scripture, people recorded information on extreme weather conditions threatening the existence of entire societies. In the case of Pomerania, such records are known since mid-13th century. Notes on weather anomalies from the Middle Ages and the early modern era (13th–16th centuries) are found predominantly in the Pomeranian chronicles. From the end of the 16th century, numerous notes on unusual weather phenomena may be found also as glosses in parish records kept by the Pomeranian ministers. In the 17th century, institutional information and analyses, often supplementing earlier records, become more frequent. These were kept by the local authorities: initially under the Pomeranian Dukes (until 1637) and subsequently the Swedish (1630–1721), Brandenburg (1658–1701), and finally Prussian rulers (since 1701) [55,56]. Notes on the weather, published in the Szczecin newspapers (18th century), are also notable. The records of interest to us, from the 16th, 17th, and partly also from the 18th century, focused mostly on picturing the situation faced by the population of Pomerania [57,58,59,60,61,62,63,64,65]. Their structure may be simplified to the following patterns: crop failure—famine—death threat—harsh, long winter, cold and rainy/dry summer, or abundant crops—abundance of food—warm and long spring and a similar summer.

Records of this type survive from the years 1572, 1586, 1668, and 1725. The chronicle of Johann David Wendland [62] states that in 1572, the crops were very poor and the prices very high, which is why the less wealthy were at risk of hunger. All this was caused by a harsh, snowless winter and a long, cold spring, followed by a rapid warming, which started in the second half of May and lasted until late autumn. The hot and dry summer resulted in poor grain crops, as most of the grains dried. The value of the index for the oak chronology for 1572 equals 0.92, the reconstructed rainfall for June and July is 117.6 mm (with the mean for the June–July period equal to about 132 mm), and the z-score equals −0.50 (Table 2). In 1586, another Pomeranian chronicler, Paul Friedeborn (P. Friedeborn 1613, book 2, pp. 130–131), recorded a severe deficit of grain caused by crop failure and the resultant widespread famine. The underlying causal mechanism can be reconstructed as follows: a snowy and cold winter caused the waters surrounding Szczecin to freeze, including the Szczecin Lagoon. The spring was long, cool, and rainy, and similar conditions prevailed in summer, which together caused the grains to rot in the fields [63,64,65]. The indexed tree-ring width for 1586 is 1.28, the reconstructed rainfall sum for June–July is over 180 mm, and the z-score is +1.73 (Table 2). Abundant crops or food shortages were also recorded by ministers in the Pomeranian parish records. This was the case in 1668 (APS, ZEKMzPiNM, 2, p. 15), when grain crops were especially good because of a mild winter, warm and humid spring, and mild and long summer. For this year, the tree-ring width index equals 1.18, reconstructed precipitation is over 163 mm, and z-score equals +1.11 (Table 2). Conversely, in 1725 and 1726 [57], subsequent great crop failures were recorded. Grain prices soared, causing famine among the less wealthy Pomeranians. This was caused by long, cool, and rainy springs and cold and rainy summers, especially in 1725. The tree-ring width index for 1725 equals 1.25, reconstructed precipitation for the summer months is over 175 mm, and z-score equals +1.54 (Table 2). The following are examples from institutional records. In 1691–1692, harsh winters followed by cool and rainy springs and summers caused a food crisis in Pomerania. The Brandenburg administration, identifying the weather as the reason of the disaster mentioned above (APS, AKS, I/5320, I/5321), monitored the holdings of grain storages in the entirety of Pomerania and organized grain transports from the neighboring provinces of Brandenburg in order to mitigate the grain deficit in Pomerania to the maximum. The tree-ring width indexes for these years equal 1.00 and 1.03, respectively, the reconstructed precipitation oscillates at the mean level for June and July (130 mm), and z-score is close to 0 (Table 2).

Overall, we conclude that the Pomeranian historical records include numerous remarks on extreme weather phenomena. A comparison of records from various sources lends credibility to the recorded narratives. However, only their verification by means of dendrochronological research may show how weather/climate changes progressed in the past.

4. Discussion

Tree-rings are used predominantly for reconstructing air temperatures. These reconstructions rely on tree-ring widths (TRW) [16,21,22,25,66,67,68], maximum latewood density (MXD) [14,17,21,37,68], or the distribution of isotope concentrations in tree-rings [19,66,67]. Reconstructions are most commonly based on coniferous trees, especially those growing in the mountains or close to the forest boundary (including both high mountain areas and the forest–tundra boundary) [17,18]. In Europe, reconstructions focus mostly on Fennoscandia. Here, one notable example is a reconstruction of summer air temperatures from 500 AD, including an identification of warm and cool periods [14,21,23]. Reconstructions are also based on trees growing in the Alps, e.g., a reconstruction of summer air temperatures from 755 AD [17]. Temperature reconstructions are also performed for Central Europe, including Poland [22,25,26,68,69]. Temperature reconstructions are also based on other species of coniferous trees, e.g., those growing in North America [70,71] or Asia (especially in the Himalayas, e.g., [16,72]).

However, the information recorded in tree-rings is also used to reconstruct other weather and climate elements. Based on TRW, Cook et al. [66] reconstruct the NAO index values from 910 AD for Europe and the Mediterranean Sea. Przybylak et al. [26] reconstruct droughts in Poland based on TRW of pines, oaks, and firs from 996 AD. Droughts are also reconstructed by Brazdil et al. [73] for the Czech Republic. The drought periods determined here based on the SPEI index for the June–July–August period coincide with the years with rainfall deficits in Western Pomerania—1590, 1630, 1631, and 1746—and dry vegetation periods—from April to September—1590, 1631, and 1706 [73]. Freud et al. [19] reconstruct the hydroclimate based on isotope analysis performed on various species of conifers and on oaks. Drought periods and wet periods (PDSI index) for the East European Plain to the Ural Mountains for the period 1400–2016 based on almost 700 chronologies, including 9 oak chronologies, are reconstructed by Cook et al. [74]. The periods determined by this team coincide only a few times with the periods determined for Western Pomerania, e.g., dry years in 1747, 1748, 1757, 1759, and 1841 and wet years in 1607, 1614, 1641, 1695, and 1696 [74]. The reason for such slight convergence is certainly the different reconstruction area and different climatic factors affecting the rainfall distribution.

Precipitation is rather rarely the subject of reconstruction [20,75,76,77,78,79]. Li et al. [79] reconstruct precipitation based on TRW of Pinus koraiensis in Lesser Khingan Mountains, Northeast China over a period of 270 years, and Chen et al. [78] over a period of more than 250 years based on TRW from the spruce. Zhang et al. [20] reconstruct precipitation over a period of 217 years in NE China, based on TRW of various tree species (Picea, Larix, Betula). In Japan, a similar study by Li et al. [75] was based on TRW of pines and oaks. Dobrovolný et al. [76] reconstruct summer precipitation (May–July) over the last millennium in the Czech Republic, based on TRW from oak chronologies. Pechtl and Land [77] present a spring–summer precipitation reconstruction (5700–4800 B.C.E.) for southern Germany.

The above compilation shows how various tree species and various parameters can be used to reconstruct weather and climate elements. Oak trees are used for reconstructions and prove to be very good indicators of climatic changes, especially precipitation sums [53,76,77,80]. The performed analysis indicates that precipitation is the dominant factor determining the growth condition of this genus. Notably, it is challenging to select a calibration and verification period due to temporally variable relationships between growth and climatic conditions, which may result from the age of the trees, climatic changes, and other factors, especially those that are anthropogenic, e.g., rapacious forest management, vermin gradations, and the impact of industrial pollution. Most reconstruction authors base the choice of calibration period on the availability of weather data and on high correlation coefficients between tree-ring width and the reconstructed element. The temperature reconstruction for Scandinavia involves a 146-year-long calibration period (1860–2006), further subdivided into the following periods: late calibration (1934–2006), early calibration (1860–1932), and extra calibration (1816–1859) [14]. Gouiraud et al. [23], in their temperature reconstruction for Fennoscandia, use three different calibration periods for three different chronologies: 85-, 113- and 129-years-long. For the Alps, the calibration period spanned 1818–1910, and the verification period spanned 1911–2003, in addition to an extra verification up to 1760 [17]. In Poland, for a pine chronology that was the basis for winter and early spring temperature reconstruction, 50-year-long calibration (1921–1970) and verification (1871–1920) periods are used [25]. Briffa et al. [21] use a fitting period that is as long as 100 years (1876–1975). The calibration period for 697 chronologies of various species (including oak) used in the reconstruction of the PDSI index for the European area of Russia is 1931–1983 [74]. The reconstruction of weather conditions in the summer season (June–July–August (JJA)) is the subject of a study by Cook [81]. Based on chronologies from Europe (from Fennoscandia to the Mediterranean) using the self-calibrating Palmer Drought Severity Index (scPDSI), they reconstruct droughts from the beginning of our era. They appoint, among others, the occurrence of an extraordinary megadrought in the mid-15th century. For precipitation reconstructions, calibration and verification periods are shorter, e.g., 30-year-long (calibration period 1956–1986, validation 1987–2017 [76]) or over 50-year-long (same calibration and validation period: 1958–2016 [20], and 1958–2013 [75]).

Climate reconstructions performed to date included only mid-term variability in tree growth and were not verified against historical data. An attempt at such an approach applied in our analysis revealed frequent disagreements between reconstructed pluvial conditions in early summer, which are critically important for radial growth in the oak, and tree growth. Apart from the reasons enumerated above, this may be influenced also by other climatic elements occurring in the prior months, as indicated by the results of the correlation and regression analysis. In evergreen gymnosperm trees, growth depends mostly on photosynthetic activity in the year of the tree-ring formation. Conversely, in some angiosperms, including the oak, growth depends on sugar reserves accumulated in the preceding year in order to initiate growth in early spring, even before the assimilation apparatus is fully formed [81,82]. The results of our study corroborate these statements. Tree-ring widths were significantly positively correlated with the June and July precipitation sum of the year of the tree-ring formation, but also positively correlated with the precipitation sum and negatively with temperature of August of the previous year. This observation is also confirmed by other studies [27,28,29,30,83,84]. A cool and rainy August causes the trees to store photosynthesis products to a higher degree in the parenchymal cells [82]. The fact that there are two different factors to which the oak is sensitive and which affect the width of tree-rings an oak tree forms introduces an inherent error into the oak chronology-based reconstructions of pluvial conditions for the summer of the year of the tree-ring formation [27,28,29,30].

Historical data are rarely used for verifying reconstructions. Most frequently, authors present temporal variability in the reconstructed parameter, identify periods characterized by decreased or elevated values of such parameter, or determine characteristic years (usually deep minima or maxima). Few papers are also based on historical records [25,26,80,85]. Such analyses enable the credibility of the reconstruction to be tested. Not all types of events, however, can be verified in such manner: historical records contain remarks only on extreme weather states; the records frequently refer to economic or social results caused by these events, and the data are incomplete and often imprecise. Overall, the records of weather events concerning a given region represent a very valuable source of information and enable a verification of the performed reconstruction.

5. Conclusions

This study presents a chronology of tree-ring widths for the native oak species (Quercus robur and Q. petraea) from northwestern Poland, spanning the last five centuries. The growth–climate relationship analysis revealed that the tree-ring width in the oak depends predominantly on precipitation in summer (June–July). However, the demonstrated influence of rainfall in the summer season of tree ring formation may be modified by the important role of rainfall in August of the preceding year, which may to some extent weaken the quality of the reconstruction. Based on this relationship, we reconstructed the precipitation sums for the last 457 years. The reconstruction revealed humid periods (1573–1578, 1605–1609, 1613–1617, 1638–1642, 1694–1698, 1776–1780, 1791–1794, 1828–1831, 1852–1855, 1863–1866, 1877–1879, and 1944–1946) and periods with rainfall shortages in summer (1579–1596, 1629–1637, 1650–1655, 1669–1672, 1703–1706, 1739–1748, 1757–1760, 1765–1768, 1808–1811, 1838–1841, 1856–1859, 1958–1961, 1965–1968, 1981–1983, and 2002–2006). The reconstruction was based on a calibration period spanning 1861–1900 because earlier periods were characterized by a strong influence of human impact (strong air pollution) or frequent droughts associated with contemporary climate change. The identified periods and years with higher than normal rainfall in summer or with summer rainfall deficits were verified based on the historical records available for northwestern Poland. Future studies should be expanded to include a pointer year analysis and extended to span a longer period in order to help improve our understanding of long-term links between precipitation totals and distribution and the regional climate, ocean–atmosphere circulation, and human impact on climate.