Dendrochronology as a Source of Evidence for Environmental History

Transcription

1 Dendrochronology as a Source of Evidence for Environmental History Introduction Dendrochronology is a method for dating events, sites, buildings and artefacts using tree-rings and other associated features within certain types of wood. It can also be used to reconstruct past environments and as an indicator of past human activity. The method was founded by Andrew Douglass, known as the father of dendrochronology, who used tree rings to quantify changes in rainfall and temperature in the south-western United States (Douglass 1919). In temperate regions some species of tree such as oak and conifer have distinctive growth periods usually during the summer months and periods of little or no growth in the winter. Tree rings are formed between these periods annually and can be counted to determine the age of a living tree or when the tree was felled to produce timber for construction or items such as furniture or musical instruments. The width of the rings depends on the amount of growth during the growth season and can therefore be used as a proxy to estimate climatic and environmental conditions during the growth season (Baillie 1995). The accuracy of this method for dating and reconstruction depends on a number of factors such as the condition and intactness of the wood, the master chronology available for an area and the number of samples suitable for analysis. Under certain ideal conditions dendrochronology can reconstruct events to within a month, but more likely to within a year (Stoffel et al 2010), however in unfavourable conditions accuracy can be limited to a range of years or decades, produce a number of different possible dates or not produce any date at all. Therefore this method cannot be used in all locations and has variable levels of accuracy. Master Chronologies In order to date a piece of wood its ring pattern needs to be compared to a master chronology for the area in which the tree grew. A master chronology is a continuous record of tree ring patterns for a particular species of tree grown in a particular region from the present day back in time. They have to be the same species as different 1

2 species react differently to the same climatic conditions, likewise with the same species and different geographic regions. They can be constructed by using living trees and working backwards by overlapping the ring pattern with timber from sources such as buildings, ships, archaeological sites, bogs and wetlands as shown in figure 1. Figure 1 Constructing a master chronology from, left to right, a wetland, archaeological site, building and living tree (Baillie 1995, p.16). Master chronologies have been created dating back to prehistoric times in some areas. Bristlecone pine, for example, has been used in the White Mountains of California to create chronologies back to 6000 BC (Baillie 1995); in the Northern Alps Hohenheim oak has produced chronologies to 8480 BC (Friedrich et al 2004) and in England chronologies back to 4989 BC have been created (Baillie and Brown 1988). However in other areas master chronologies do not go back as far, such as northeastern Qinghai in China to 515 BC (Sheppard et al 2004) and in the Southern Alps where floating chronologies could not be connected to the Northern Alp record as climatic conditions are different (Cufar 2007). Other areas have no master chronologies, for example Egypt where there is a lack of adequate recent wood to extend records to the current day as many of the species the ancient Egyptians used have become extinct. Also Italy and the Mediterranean have few master chronologies (Cufar 2007) along with much of Asia and South America. Trees growing in constantly wet tropical regions are not suitable for tree ring analysis as they have no variation in growth so rings are not very clearly defined (Eckstein 2007). 2

3 This unevenness in global distribution of master chronologies and there length along with a bias in research towards Europe, North America and Japan limits the use of dendrochronology as a dating tool on a global scale. Dating With Dendrochronology The accuracy of matching a piece of wood to a master chronology is heavily dependent on the intactness of the wood being studied. Sapwood is the layer within trees such as oak that lies between the bark and the hardwood. In order to accurately date a piece of wood the sapwood needs to be present so that all the rings can be counted. Because sapwood is relatively soft and fragile it is usually removed when making objects such as furniture, panels or instruments (Grissino- Mayer et al 2004, Läänelaid and Nurske 2006) and can be lost due to erosion or insect damage in buildings or rot if the wood is coming from a peat bog etc. (Baillie 1995). When sapwood is not present an estimate has to be made of how many rings are missing. This is done by calculating the average number of sapwood rings present on trees growing where the wood originated. The outer rings can also be missing from other trees such as pine by erosion and rot (Towner et al 2001). However even with all the rings present the date obtained is the date the tree was felled and not necessarily the date the artefact or building being studied was created (Cufar 2007). Läänelaid and Nurske (2006) had these problems when trying to date a panel of Hans van Essen s Still life with a lobster. It was made worse by them not knowing where the panel originated, at a 50% confidence level Oaks grown in Germany and The Netherlands have 13 to 23 sapwood rings, with a median of 17, and Oaks from Poland 13 to 19 with a median of 15. They assumed the wood was from Poland and obtained a felling date of 1609 to 1636, based on the 95% confidence level, 1613 to 1619 at 50% confidence and decided 1615 was the most probable. They then used the average length of time between felling and painting in the 16 th to 17 th century of 2 to 8 years to get an estimate of 1617 to 1623 of when the painting was made, this was cross checked by a art historian who dated the picture as This example shows that estimating sapwood rings is far from exact, with either a wide range of 27 years or low confidence levels of 50% affecting the accuracy of the date. It also 3

4 shows that the date of felling can be many years before the date of use and there is difficulty in determining this. Eckstein (2007) also highlights the problem of wood being stored after it is cut down, affecting the dating of buildings in the medieval town of Lübeck, Germany. He assumed that wood in a merchant s house that had 4 different felling dates came from a market, but could not determine how long they had been stored before being used. A former ecclesiastic building, however, with all the intact timbers being felled in 1317 was assumed to have been constructed from fresh timber. Eckstein admits that his findings are open to interpretation. Eckstein et al (cited in Baillie 1995) tried to date roman structures at Carlisle Castle, but because different samples from the site had different amounts of sapwood and hardwood missing it was not possible to get an exact date. A Later study by Hillam (1992) only managed to determine building phases of AD 72/3 to 85 and AD 93 to 97. This again shows that the completeness of the wood is important, however when dating older structures it is less likely to be available. Even with intact wood and a suitable local master chronology Yamaguchi (1986) found that autocorrelation prevented him from getting an exact data. Computer programs are used to match wood samples with a master chronology by comparing how closely the width measurements of each line up statistically. He found that a 290 ring Douglas fir log dated by other methods at 1482 to 1668 matched the Pacific Northwest Douglas Fir Master Growth-ring Sequence at 6 different places with a confidence level of 99.9% with the highest values at 1504, 1647 and John Fletcher also got multiple matches of 1035, 1046, 1138, and 1221 when trying to date a structure at Eynsford Castle, Kent with chronology from Westminster Abbey (Baillie 1995). Rigold (1975) picked 1138 for his research as it matched other archaeological evidence he had collected, despite the dendrochronology not being verifiable. Baillie (1995) sees this as potential problem with dendrochronology if dates are selected based on expectations rather than hard scientific evidence. Towner et al (2001) highlight another problem with the use of dead wood in construction, by studying remains at McKean Pueblito, New Mexico. The date at which a tree died can be determined, but not when it was cut down or used in 4

5 construction. This coupled with the loss of outer rings by erosion led to a large level of uncertainty over the date of 1713 that was estimated. Reconstructing With Dendrochronology Dendrochronology can also be used to reconstruct past events based on the effects they had on trees while they were growing. For example floods occurring on the Red River, Canada were reconstructed back to 1648, in order to aid flood planning in the future (St George 2010). Scars from abrasion or impacts can show up in tree rings along with flood rings in for example ash and oak. These result from restricted growth either from floodwaters defoliating trees or prolonged inundation of the roots. Tilting and uprooting can also be determined from rings, even after the tree has righted itself (St George 2010). Manabe and Kawakatsu (1968) used tilting of trees to reconstruct past typhoons in Japanese cedar back to 473 AD. Using living trees can also be problematic; Koch (2009) found that false rings, more than one growth ring in a year, can occur due to disturbances during the growth season and rings can be missing as a result of suppressed growth over periods longer than a year, affecting the ability to accurately date events. These problems can be reduced by using numerous specimens; however that may not be possible in all locations. Other major events such as volcanic eruptions, earthquakes, landslides and wildfires have also been successfully dated or reconstructed to within a year using dendrochronology. However some major events can be missing from the tree ring record if the trees in the area were not directly affected and most minor events are missing completely due to their limited impact on growth (Stoffel et al 2010). Some events can completely remove all the vegetation and create new surfaces. These surfaces can be dated by measuring the age of the trees that have grown since, this requires ecesis to be known, this is the time between the exposure of a new surface and germination of trees. This varies on factors such as the substrate, available seed sources and climatic conditions and can be several decades (Koch 2009). Dendrochronology can also be used to infer socio-economic changes. For example the Black Death in the 14 th century, where around a third of the British population died, is marked by a period of woodland regeneration as building declined (Hollstein 1980, cited in Baillie 1995). Therrel (2005) found that tree rings correlated well with 5

6 maize yield and consequently church tithes paid with agricultural goods, see figure 2. This was due to the affect of drought on both tree growth and maize yields. Without the historical context, however, changes in tree growth would only indicate that an event or change occurred and not what caused it. Figure 2 Tree ring reconstruction (spiked line) and maize yield anomalies (smooth line) in central Mexico (Therrel 2005). Conclusions The examples in this report show that dendrochronology has a wide range of potential applications including dating artefacts and buildings, reconstructing the severity and timing of events and indicating socio-economic changes. The accuracy of the dates it produces depends heavily on the accuracy, geographic location and length of the master chronology and the intactness of the sample being analysed. Many assumptions are made and findings are open to interpretation and biased selection by others. Despite the problems inherent in dendrochronology it provides a useful tool in piecing together history when used in conjunction with other historical and scientific data sources, however without geographical or historical context the results can be difficult to interpret. Further knowledge of processes such as ecesis, sapwood growth in different locations and what happened to wood between it being felled and used are required in order to make dendrochronology more accurate and useful as a dating and reconstruction tool. 6

7 References Baillie, M. G. L. (1995) A slice through time: dendrochronology and precision dating. London: B. T. Batsford Ltd. Baillie, M. G. L. and Brown, D. M. (1988) An overview of oak chronologies. British Archaeological Reports 196, Čufar, K (2007) Dendrochronology and past human activity - a review of advances since Tree-Ring Research 63 (1): Douglass, A. E. (1919) Climatic cycles and tree-growth. Washington: Carnegie Institution of Washington. pp 9-14 (available at: Eckstein, D. (2007) Human time in tree rings. Dendrochronologia 24: Friedrich, M., Remmele, S., Kromer, B., Hofmann,J., Spurk, M., Kaiser, K. F., Orcel, C. and Kuppers, M. (2004) The 12,460-year Hohenheim oak and pine tree-ring chronology from central Europe - a unique annual record for radiocarbon calibration and paleoenvironment reconstructions. Radiocarbon 46 (3): Grissino-Mayer, H. D., Sheppard, P. R., and Cleaveland, M. K. (2004) A dendroarchaeological re-examination of the Messiah violin and other instruments attributed to Antonio Stradivari. Journal of Archaeological Science 31(2): Hillam, J. (1992) The dating of archaeological sites in the United Kingdom. Lundqua 34: Koch, J. (2009) Improving age estimates for late Holocene glacial landforms using dendrochronology - Some examples from Garibaldi Provincial Park, British Columbia. Quaternary Geochronology 4: Läänelaid, A. and Nurkse, A. (2006).Dating of a 17th Century painting by tree rings of Baltic oak. Baltic Forestry 12 (1): Mannabe, D. and Kawakatsu, K. (1968) Chronological investigations on the annual ring and typhonic patterns of the Yakushima cedar. Reports of the Kyushu University of Forestry 22: Rigold, S. E. (1975) Structural aspects of medieval timber bridges. Medieval Archaeology 19: Sheppard P. R., Tarasov, P. E., Graumlich, L. J., Heussner, K-U., Wagner, M., Österle, H. and Thompson, L. G. (2004) Annual precipitation since 515 B.C. reconstructed from living and fossil juniper growth of northeastern Qinghai Province, China. Climate Dynamics 23: St George, S. (2010) Dendrohydrology and extreme floods along the Red Rvier, Canada, in: Stoffel, M., Bollschweiler, M., Butler, D. R. and Luckman, B. H. eds. Tree rings and natural hazards: a state of the art. London: Springer. pp

8 Stoffel, M., Bollschweiler, M., Butler, D. R. and Luckman, B. H. eds. (2010) Tree rings and natural hazards: a state of the art. London: Springer. pp 3-26 Therrell, M. D. (2005) Tree rings and El Anõ del Hambre in Mexico. Dendrochronologia 22 (3): Towner, R., Grow, D., Psaltis, J. and Falzone, A. (2001) The importance of sample context in dendroarchaeological interpretation: An example from northwestern New Mexico, USA. Tree-Ring Research 57 (1): Yamaguchi, D. K. (1986) Interpretation of cross correlation between tree-ring series. Tree- Ring Bulletin 46: