Rob Wilson
Dendroclimatology is a subfield of dendrochronology where tree-ring parameters (such as ring-width, density, blue intensity, isotopes etc) are transformed to estimates of past climate. Traditional dendroclimatology, which generally uses ring-width, density, or blue intensity parameters, is based on strategic sampling of woodland sites where the climate signal is maximised to ideally a single variable – for example where the growth of the trees is limited by either hydroclimate or temperature. If the aim of the study is to reconstruct past temperatures, one needs to target woodland sites where the growth of trees is limited by growing season temperatures. Examples of such environments are woodlands growing close to high elevational or latitudinal treelines. Trees do not grow above these limits as it is simply too cold for cell formation and ring development. Likewise, if the aim of the project is to reconstruct past hydroclimate (precipitation or drought), then woodlands are generally targeted at low elevational or latitudinal treelines where moisture availability is the dominant climate factor controlling growth although site ecology (ie rocky substrate and/or thin soils) can lead to some degree of moisture limitation.
The use of the stable isotopes of carbon and oxygen (see Section 8) for dendroclimatology are less constrained by woodland location, and sites do not need to be targeted for specific climate limitation so substantial potential exists from mid-latitude regions where traditional dendroclimatological approaches are less reliable (McCarroll and Loader 2004; Loader et al 1988, 2008, 2020; Young et al 2015; Büntgen et al 2021). The tree physiology that mechanistically underpins the fractionation of the (carbon, oxygen and hydrogen) isotopic signal in trees is relatively well-established, but potential remains to further explore the complexities of their relationships with extreme climate, disturbance and elevated atmospheric CO2 across a range of different species and ecotones, and is being increasingly used in dendroecological studies of tree response to extreme events.
Dendroclimatology is a mature scientific discipline based on at least 50 years of research (Hughes 2002, Sheppard 2010, He et al 2019; Sánchez-Calderón et al 2022), and local, regional and even hemispheric scale reconstructions exist for both temperature (Rydval et al 2017; Fuentes et al 2018; Cook et al 2013; Wilson et al 2016) and hydroclimatic parameters (Wilson et al 2013; Cook et al 2015; Steiger et al 2018; Loader et al 2020) as well as reconstructing synoptic climate phenomena such as the North Atlantic Oscillation (Linderholm et al 2008; Cook et al 2019) and the El Nino Southern Oscillation (Li et al 2013).
Ring Density and Blue Intensity
Ring density’s predominant use in dendrochronology has been for reconstructing past climate (mainly summer temperatures) from conifer tree-ring samples (Briffa et al 1988; Wilson et al 2016). The anatomy of a single conifer ring expresses low wood density values in the earlywood, due to large lumen (cell space) and thin cell walls. As the conifer ring transitions from earlywood to latewood, the lumen space reduces, and the cell walls become thicker resulting in an increase in density. It is the maximum density value within the latewood that has been a hugely important parameter for dendroclimatology. However, X-ray densitometers, which are the main method to measure wood density, are very expensive and require substantial technical training. Despite the importance of maximum latewood density (MXD) in dendrochronology, relatively few X-ray densitometer systems exist in tree-ring labs across the world.
Over the last couple of decades, a new cheaper approach to measuring relative wood density from conifer wood samples, blue intensity (BI), has been developed (McCarroll et al 2002; Campbell et al 2007; Björklund et al 2014; Rydval et al 2014; Kaczka and Wilson 2021). BI measurements are made from scanned images of polished wood samples and utilise the reflectance properties of wood to provide estimates of relative wood density. In the earlywood of a conifer ring, when density values are low and cell lumen size is large, the intensity of light reflectance is high. In the latewood, when cell walls are thicker and density is high, the light reflectance is low.
Experiments with different colours in the visible spectrum (red, green and blue – McCarroll et al 2002) identified that the blue part of the spectrum provides the strongest proxy of relative wood density. Although early hypotheses stated that blue was the optimal colour to focus on as lignin (a dominant constituent of cell walls) readily absorbs blue light more than red and green, later research has shown that the blue range of the spectrum simply accentuates the dark and light pixels better than other colours in the visible spectrum and that BI is simply measuring the ratio between total cell wall area and total cell area (cell wall + lumen area – Björklund et al 2021). Measurement of the intensity of reflectance within the blue part of the visible spectrum therefore provides a proxy of relative density that is inversely correlated with actual ring density. It is now standard practice for these reflectance data to be inverted so they represent the wood properties (ie density) we are trying to measure. This inverted state of the data is now formally defined as blue intensity (Björklund et al 2024).
BI is a very appealing tree-ring parameter for dendrochronology as the costs of scanners are much cheaper than an X-ray densitometer. The required training for sample preparation and measurement is relatively simple and, in some cases, one could argue that BI data can be generated at no additional cost to standard TRW as both the TRW and BI data can be measured at the same time. The potential limitations of BI however mostly reflect the fact that the reflectance intensity values can be biased by any discolouration of the sample. Sample discolouration can come in many forms such as differences in heartwood and sapwood colour with the sapwood wood often being lighter than heartwood. Other forms of discolouration due to fungal discolouration of the wood, wood decay, resins and even wood treatments for preservation can all systematically bias the BI measurements (Björklund et al 2024).
For dendro-archaeological dating, these issues can generally be minimised through standard flexible detrending procedures and the relative deviations from one year to the next can still be used for dating purposes (Wilson et al 2017a – see also Scot2K Case Study). However, for dendroclimatology, when capturing long term trends of potential climate change, any systematic colour bias can severely affect the climate estimates if the samples have not been prepared appropriately to minimise such colour biases. Despite these potential issues, many studies over the last decade have shown that robust estimates of past climate can be derived from BI data which are comparable to results obtained from using traditional density parameters (Björklund et al 2014; Rydval et al 2017; Wilson et al 2014; 2019; Fuentes et al 2018; Reid and Wilson 2020; Heeter et al 2023).
Reconstructing past climate in Scotland from tree-rings
It has been 40 years since the first tree-ring based reconstruction for Scotland was published (Hughes et al 1984). In this ground-breaking study, Hughes et al (1984) utilised both ring-width and maximum latewood density data from a small network of five pine chronologies across the Scottish Highlands to develop a summer temperature reconstruction (1721-1975 CE). Since that time, the tree-ring laboratory at the University of St Andrews has expanded the living pine network considerably and the current network represents over 90 sampled woodland sites (more than 4000 trees sampled) from sea-level up to the high elevation treeline (around 600 metres above sea level) across the Highlands. Of the higher elevation semi-natural pine woodlands, the average tree age is around 200 to 250 years old, representing regrowth since substantial timber clearance in the 18th and 19th centuries (Fish et al 2010; Rydval et al 2016), although some individual trees have been found that germinated in the 15th century. The mean age of living pine trees results in a fundamental limitation on how far back in time a climate reconstruction can go. However, extension is possible by using preserved pine material from either sub-fossil sources (such as in nearshore lake sediments) or historical structures (Wilson et al 2012 – see also section 4).
The Scottish Pine Project (funded between 2010 and 2020 by the Leverhulme Trust, Carnegie Trust and UK Natural Environment Research Council – see SCOT2K Case Study) aimed to explore and utilise both sub-fossil and historical sources to facilitate the extension of the living pine network to derive a substantial update of the original Hughes et al (1984) summer temperature reconstruction. Although substantial success in dating several vernacular structures was made (Wilson et al 2017; Mills et al 2017 ), specific provenancing of the historical material to appropriate high elevation locations was not possible at that time. The climate drivers of tree-growth changes as a function of elevation (see section 1 Introduction to Dendrochronology) means that near upper tree-line trees experience more temperature limitation and are sensitive to growing season temperatures. Therefore, for the new updated reconstruction, sub-fossil samples that had been extracted from lakes between 270 and 400 metres above sea level in the northern Cairngorms were used to extend the living dataset.
The new reconstruction utilised more than 400 living and more than 100 sub-fossil samples, from which both ring-width and latewood blue intensity data were measured. These data were combined to derive a highly robust (56% explained variance) July-August mean temperature reconstruction back to 1200 CE (Rydval et al 2017), and the combined tree-ring and instrumental data indicate that the 2013-2022 decade is the warmest period for the last eight centuries. Although periods of warm summers are also expressed in the period prior to the 1500s, caution is advised as the number of trees representing this period is substantially less than recent centuries, so uncertainty is greater (see 6.1 Research Recommendations). Such dendroclimatic reconstructions not only provide information on past temperatures changes and the drivers of variability (Rydval et al 2017), but they can also help inform on wider societal issues related to the drivers of famine (D’Arrigo et al 2020) and even productivity changes in agricultural crops systems over time (Martin et al 2023).
The International Tree-Ring Databank was established in the 1970s with data now archived for more than 5000 sites with many tree-ring datasets existing across the mid-to-high latitudes of the Northern Hemisphere. For those datasets representing woodlands where growth is limited by growing season temperatures, these data, especially using density parameters, can be combined to create regional and even hemispheric estimates of past temperature.
Climate change can often be expressed quite differently locally, but when multiple records are averaged over large regions, the resultant reconstructions express the large-scale forcing factors influencing climate such as cooling from major volcanoes and warming from anthropogenic emissions of carbon dioxide (Wilson et al 2016). The figure above compares these different spatial scales of past summer temperatures based on tree-ring data from Scotland, western Eurasia and the Northern Hemisphere. Although the Northern Cairngorms (NCAIRN) record expresses recent warming, the overall local scale variability expressed in this record is quite different compared to the larger scale records for western Eurasia and the Northern Hemisphere where the recent marked warming related to CO2 emissions is extremely stark. Although this framework document focusses mainly on Scotland, it must be emphasised how important such local/regional datasets are for large scale assessment of climate and ecological change.