DENDROCHRONOLOGY

            The objective of dendrochronology is to use characteristics of the tree-rings (e.g., width and maximum density) to date artifacts or buildings and reconstruct the climate of the past (Fritts, 1976) (Figure 2.7).  Dendrochronology is not a good indicator of fall drought because the trees are generally dormant then, but winter precipitation can affect growing season soil moisture status and spring growth.  Tree rings offer more reliable information about moisture conditions during spring and summer growth period (Fritts, 1976) .  Dendrochronology uses the natural clock provided in the annual growth rings of trees as both a means of dating and as a measure of climate (Baillie, 1995) .

            Dendrochronology primarily uses tree-ring series sampled through fieldwork.  The samples in this study are compiled from timbers found in gravel beds, Roman roads, Roman limes, structures, and archaeological sites (Figure 2.8).  The samples are prepared so that the rings can be seen clearly, then they are crossdated by comparing the patterns of wide and narrow rings in many different samples to assure accuracy (Figure 2.9).  Without this careful dating process, certain anomalies (e.g., missing and false rings) that occur occasionally would create chronological inaccuracies (Douglass, 1947, Cook and Kairiukstis, 1992, Schweingruber, 1966) .  After the ring series have been crossdated, they can be measured.       

An annual ring width series may contain nonclimatic trends of biological origin, such as age, slope, and competition with other trees. Each series must be standardized.  Standardization is accomplished by dividing the ring-width value by the value of a curve, negative exponential, regression line or spline fitted to the series for each year (Cook and Kairiukstis, 1992) .  This removes systematic changes associated with the increasing age of the tree and the differences in growth rate between trees.  The standardization process results in series of dimensionless indices with statistical properties (i.e., mean and variance) that are stable through time (Cook and Kairiukstis, 1992) .  In German oak and pine chronologies, the long-term climate trend can be difficult to ascertain because this process of standardization removes much of the signal, however, it is still useful for short-term climate variability (Cook et al., 1999).  The standardized series can be averaged to create a mean master chronology that represents the growth rate of the trees in region.

Standardization equalizes absolute growth differences between series, i.e., it forces all ring width series to have a uniform mean, so that tree records with larger than average growth do not disproportionately affect the chronology when multiple series are combined into a mean chronology (Fritts, 1976) .  A computer standardization program, such as ARSTAN or TSAP, eliminates variations in tree growth fluctuations due to aging and averages the multiple series into a mean or master chronology (Cook and Kairiukstis, 1992, Schweingruber, 1966) .  The availability of tree-ring sources for this particular study ranges from eight thousand samples in the last century BC to just under one hundred in the AD 300s (Becker, 1993, Becker, 1981, Friedrich and Greiner, 2001, Spurk et al., 1998) .  The more tree-ring samples responding to the same regional climate that are dated and incorporated into the chronology, the more precise the understanding of the growth potential of any given year.  It also helps to reduce the other influences that affect tree-ring growth, besides of climate.  The statistical characteristics of the resulting master chronology are examined and the relationship between the ring-width indices and climate is then modeled (Figure 2.10).  This is followed by verification from independent data sources: historical documentation, available meteorological data, or other climate indicators (Fritts, 1976) .

Dendrochronology is one of the more important of the climatic proxies, due to its ability to identify periods of growth and drought in annual, and sometimes intraannual, increments.  The ideal in dendrochronological studies is the creation of long tree-ring chronologies of a single species compiled from long, continuous samples that cover the length of the chronologies (even to thousands of years with the bristlecone pine) (Fritts, 1969) .  This, unfortunately, is not the case with the German chronologies and many others.

The long German chronologies do not have long series from individual tree-ring samples (where some individual samples may exceed four thousand years in length), but are build up from relatively short segments, most less than three hundred years long.  This procedure requires sifting through thousands of tree-ring samples from a wide variety of environmental and historical settings (from riverbanks to cathedral beams).  This process is a long and tedious endeavor.  The use of computer analysis and the availability of complete cross-sections of fallen trees allow the best samples to be included in the chronology (Figure 2.11).  With the long segments cross-dated, the other shorter segments are then added if the general date is known or removed from the chronology if there is question.

            The climate – tree-ring relationships in European tree rings can be verified, or placed in time, through documented climatic conditions.  Historical documents such as Procopius’ recording of a volcanic eruption in AD 536 or John Lydus’ recording of a drought in AD 626 have given validity, early on, to the science of dendrochronology (Maas, 1992, Procopius and Dewing, 1914) .  The main object of dendrochronology is to use characteristics of the rings to statistically reconstruct the climate of the past (Fritts, 1976) .  The ring growth is related to meteorological conditions to reveal the climate of the growing season (Fritts, 1976) .  The relationship between ring growth and quantified climate is derived through calibration of ring width or some other ring variable with current meteorological data. Usually some form of linear regression is used for the calibration.

There are several other factors that can also contribute to tree ring variability such as elevation, exposure, slope, competition from other trees and variation between the different tree species.  It may seem that there are too many variables to reveal accurate data for the paleoclimatologists, but in reality, crossdating assures chronological accuracy and is based on co-variability which guarantees the presence of climate information (Glock, 1950) .  The use of many samples averages out the many extraneous factors that are not shared as climate is and gives an accurate picture of the climatic conditions that all the trees shared. 

Tree-ring dating of individual series is confirmed by comparing series from many different trees, and comparing tree-ring chronologies throughout the region.  After non-climatic influences have been removed through standardization and replication, the magnitude of the tree-ring indices reveal the growth pattern of any given year.  This is exceptionally valuable in truly understanding both climate history and the history of humankind.  Furthermore, the present study would not be feasible without the use of dendrochronology.