RECENT CHANGES IN THE SCOTS PINE TREE-RINGS AND THEIR RELATION TO CLIMATE IN NORTHERN FINLAND
Abstract
Network of Scot pine tree-ring chronologies was studied by means of dendroclimatology in northern Finland. Ring-widths were measured, cross-dated and 17 site chronologies and a regional chronology built. Network was spatially most representative during the last 75 years and the special emphasis was thus put to analyse the recent changes in the growth characteristics and their dependence on climate. The most important climatic factors controlling the regional tree growth were the mid-summer temperatures and winter-conditions, expressed as July mean temperatures and index of North Atlantic Oscillation (NAO) during November (previous to the growing season), respectively. The influence of mid-summer temperatures on the radial growth was shown to be somewhat time-scale and time-independent, whereas the correlation between NAO and tree-rings was apparently more site-specific. Last ten years have experienced considerable shift of the growth that can be seen in the regional tree-ring chronology as a change from anomalously deteriorated to ameliorated growth, from period 1995-1999 to 2000-2004, respectively. This change could be mainly attributed to the simultaneously elevated summer temperatures. The radial growth was notably high in the year 2003.
INTRODUCTION
Annual increments of tree radial growth, tree-rings, are commonly used indicators of different factors affecting the growth variability of trees. Great benefits of tree-rings are that they form continuous annually resolvable time series providing long-term perspective on natural variability at individual-to-regional scale. In addition, great number of observational series of ring-widths can be obtained with relatively low costs. These time series can be further calibrated against any other reasonable and similarly resolved time series, most often with climatic records, in order to reveal the dependency of tree growth on these environmental factors (Fritts 1976). At the extreme conditions of northern timberlines, the primary factor known to affect the radial growth of trees is thermoclimate, directly and indirectly limiting the vital processes that control the growth. In northern Finland, the temperatures of growing season, especially during the mid-summer, are known to bear great influence on the radial tree growth (Mikola 1950; Sirén 1961; Hari and Sirén 1972; Lindholm 1996; Helama et al. 2004). However, also the climatic factors during the snow-melt, as well as during the mid-winter, have previously shown to have impact on the growth variability at these high-latitude sites (Lindholm et al. 2001; Macias et al. 2004).
The main aim of this work was to study the tree growth variability by means of tree-rings. For this purpose, network of Scots pine ring-width chronologies at the proximity of the coniferous timberlines in northernmost Finland were analyzed. Special emphasis was put on the annual growth variability during the period of meteorological observations. These two types of data, biological and meteorological, were used here in order to model the relationships between year-to-year tree growth dynamics and their relationships to prevailed climate, as well as the time-dependency of these relationships. As shown by Hallet et al. (2004), large-scale climate indices might in some cases predict the ecological processes better than local weather. For that reason, monthly and seasonal indices of North Atlantic Oscillation (NAO) were used in the present analysis along with the variables of temperature and precipitation. NAO, characterized by an oscillation of atmospheric mass between the Arctic and the subtropical Atlantic, produces large changes in the mean speed and direction over the North Atlantic. This, in turn, bears influence the heat and moisture transport between the Atlantic and the continents around it, also in the study region (Hurrell 1995).
The study in hand contributes to the general needs for understanding the sensitivity of the northern coniferous forests under the drastic influence of climate variability at annual-to-decadal basis. This study will also be a part of more comprehensive investigations on dendroclimatological and dendroecological processes in the region (e.g. Lindholm 1996; Helama et al. 2002, 2004; Macias et al. 2004), complementing the knowledge on the most recent changes in the growth of the northern Scots pine.
MATERIAL AND METHODS
Tree-ring network
Network of Scots pine (Pinus sylvestris L.) ring-width chronologies was used as an indicator of growth variability at the northern Lapland (Fig. 1). This network of 17 site chronologies is spatially well spread over the region being thus well representative of the regional growth patterns. The sample collection was done in the late fall of the year 2004 thus that the latest measured ring can be taken as entirely grown increment of the same year. Tree-ring cores were extracted by an increment borer. The ring-widths were measured from each core to the nearest one hundredth of a millimetre. Careful cross-dating, by means of the widths of the rings, was practised using the dendrochronological procedures of Holmes (1983), along with visual comparison of the produced time series. Material comprises of 50140 ring-width measurement in 386 ring-width series. Each series represents the radial growth of a tree as a function of time. It covers the last 241 years (1764-2004) with the mean segment length of 129.9 years.
Time-series analyses with tree-rings
Growth trend, expected to represent the long-term growth decline due to biological ageing of a tree, was modelled for each individual ring-width series using the approach of Fritts et al. (1969). That is, the growth trend was modelled using the negative exponential curve, linear regression with negative slope or line through the series mean. Tree-ring indices were derived from this trend line (curve) as ratios. Site chronologies were constructed averaging the annual index values by arithmetic mean. Regional chronology was calculated as the average of the site chronologies.
Analyses using meteorological data
Monthly and seasonal series of temperature and precipitation were measured at the meteorological station of Sodankylä. These were the time series of mean temperatures and precipitation sums. NAO indices were used here as computed by Hurrell (1995) from the difference of normalized sea level pressure between Ponta Delgada, Azores and Stykkisholmur/Reykjavik, Iceland. The relationships between the growth and climatic variables were estimated using Pearson correlations.
Due to relative shortness of some of the site chronologies, climatic analyses and inter-site comparisons were performed here using the time span of 1930-2004. Growth anomalies prior to that date, and their relation to climate, have been recently interpreted for example by Helama et al. (2002, 2004).
RESULTS AND DISCUSSION
General growth characteristics
Correlations between the tree-ring chronologies were fairly good, indicating the presence of common growth signal among the present sample. On average the inter-chronology correlation was 0.48. The first order autocorrelation in all 17 chronologies was on average 0.47.
Pine growth in the region since 1875 can be characterized as follows. The growth was severely deteriorated during the later part of the 19th century (Fig. 2). The growth minimum was experienced in 1903 when the regional tree-ring index dropped to its lowest value during the examined time frame. Ameliorated growth phase occurred during the 1920s and 1930s after which the growth in general decreased with trend-like fashion until the years of the early 1970s. Thereafter, slight increase did occur to the recent maximum in 2003.
Relationships between growth and climate
The most significant climatic factor bearing influence on the tree-ring growth was mid-summer (July) temperature. The correlation between the mid-summer temperatures and regional tree-ring chronology was 0.53.
In order to compared the growth variations and mid-summer temperatures at different time-scales, the comparison between the low-pass filtered and high-pass filtered time series was also performed. Decadal variations were extracted from the series using 15-years cubic spline (50 percent cutoff) and the short-term variations as the residuals from the low-pass curve (Fig. 3). Both of the correlations were relatively high, indicating the presence of the time-scale independent relationship between tree-ring growth and mid-summer temperatures (Fig. 3).
The North Atlantic Oscillation (NAO) was not very clearly imprinted in the regional tree-ring chronology. The correlation between the NAO-index of November (previous to the growth season) was 0.33. The influence of NAO is possibly more site-related and time-dependent (Macias et al. 2004). This can be partly seen in the site-specific comparison: in Pitkäjärvi and Partakko (see Fig. 1) the correlations with the November NAO-index were close to the level of 0.40 whereas in some sites the correlation was below the level of 0.10.
These features, the dependence of the tree-ring variability on the summer temperatures and the conditions during the early winter (expressed as November NAO-index), were the two most significant climatic factors seeming to bear region-wise influence on the pine growth during the exploited time frame. As the correlations between tree-rings and November temperature or precipitation did not exceed that of the November NAO-index (not shown), it could be assumed that the index of the NAO of that month probably represents both of these weather variables, and that its influence on the growth is a reflection of both, the snow conditions and decreasing temperatures, during that part of the year.
Correlations of similar sign and strength, between Scots pine ring-widths and winter-time NAO indices, have been previously drawn for example by Lindholm et al. (2001) and discussed by Macias et al. (2004) in the context of northern Finland. However, as shown previously (Macias et al. 2004), the correlations between tree-rings and differently seasonal and monthly NAO indices can be considerably higher during specific periods or at some specific sites in the region.
Extreme growth anomalies
Regional growth anomalies were determined using the 1-year and 5-year growth periods (Table 1). In general, these periods were well verified the positive growth phase during the 1930s and the relatively poor growth phase after the mid-1960s. It is noteworthy that the growth in 2003 was only slightly below the growth during the most ameliorated growth years of the 1930s. The most ameliorated growth, however, occurred in 1937 (Table 1). Both of these years, 1937 and 2003, were years around which rare climatic events took place. According to Tuomenvirta (2004), the year 1938 was warmest ever recorded by meteorological stations in Finland. In northern part of the country, however, it seemed to be the year 1937 that was relatively warmer in the context of growing season temperatures. The summer of 2003 has been estimated to be the warmest for several centuries in some parts of Europe (Chuine et al. 2004; Luterbacher et al. 2004). At the local scale, the year 2003 experienced relatively very warm mid-summer also in the study region (Table 2).
Very interestingly, the negatively anomalous growth period between the years 1995-1999 was followed by the good growth years between 2000 and 2004 (Table 1). It is evident that this positive growth phase during the most recent years has been caused by simultaneous summer-temperature rise (Table 2). The period 2000-2004 was actually the warmest such periods during the study period (1930-2004) in northern Finland (Table 2).
Not only the positive growth and climatic year and periods, but in addition the negative ones showed clear consistency of joint-occurrence. There occurred overlap between the negatively anomalous periods during both the mid-1960s and in the mid-1990s (Table 1 and 2). Yet, the single year extremes in the growth and temperatures both gave an impression of the deteriorated growth conditions during the cool 1960s. There seemed occur some mismatch between the climatic and growth years which were most probably due to physiological processes internal to tree growth, making the climatic signal a bit noisier. The use of different approaches in the tree-ring standardization, such as pre-whitening of the tree-ring indices prior to climatic comparison (see Guiot 1986; Monserud 1986) could, at least partly, overcome this issue.
CONCLUSIONS
This present study was aimed to determine the recent growth characteristics as well as their dependency on the prevailed climate in northernmost Finland. As shown in previous studies (Mikola 1950; Sirén 1961; Hari and Sirén 1972; Lindholm 1996; Helama et al. 2004), the influence of mid-summer temperatures is crucial for pine growth in the harsh northern conditions. This could be seen for example during the 1930s and during the recent years of third millennia (2000-2004) during which the positive phases in the summer-time temperatures and Scots pine growth joint-occurred. That is to say that the ameliorated and deteriorated periods of growth have varied in accordance with the summer-temperatures. The positive influence of NAO phenomenon and the related weather types during its positive and negative phases on pine tree-rings have also shown in some of the previous studies (Lindholm et al. 2001; Macias et al. 2004). As the NAO bears influence on both, characteristics of temperature and precipitation variations, it is likely that it impacts on growth conditions in an integrated fashion so that the site-specific features may largely come to amplify and nullify its local influence. The winter-time NAO index has performed a pervasive positive phase during the past 30 years or so (e.g. Hurrell et al. 2001), but such a trend cannot directly been detect in the regional tree growth even though November NAO-index was shown to bear impact on the tree-rings. This is potentially due to relatively greater influence of summer temperatures on growth.
REFERENCES
Chuine, I., Yiou, P., Viovy, N., Seguin, B., Daux, V. & Le Roy Ladurie, E. 2004: Historical phenology: Grape ripening as a past climate indicator. Nature 432: 289-290.
Fritts, H.C. 1976. Tree Rings and Climate. Academic Press. London. 567 p.
Fritts, H. C., Mosimann, J. E. & Bottorff, C. P. 1969: A revised computer program for standardizing tree-ring series. Tree-Ring Bulletin 29: 15-20.
Guiot, J. 1986: ARMA techniques for modelling tree-ring response to climate and for reconstructing variations of paleoclimates. Ecological Modelling 33: 149-171.
Hallet, T.B, Coulson, T., Pilkington, J.G., Clutton-Brock, T.H., Pemberton, J.M., & Grenfell, B.T. 2004. Why large-scale climate indices seem to predict ecological processes better than local weather. Nature 430: 71-75.
Hari, P. & Sirén, G. 1972. Influence of some ecological factors and the seasonal stage of development upon the annual ring width and radial growth index. Royal College of Forestry. Department of Reforestation. Stockholm. Research Notes 40: 1-22.
Helama, S., Holopainen, J., Timonen, M., Ogurtsov, M.G., Lindholm, M., Meriläinen, J., & Eronen M. 2004. Comparison of living-tree and subfossil ring-widths with summer temperatures from 18th, 19th and 20th centuries in northern Finland. Dendrochronologia 21: 147-154.