The study of Earth’s past temperatures plays an important role in our understanding of how to contextualize modern warming. But thermometer records tend to be short and so in order to build records of temperature that span hundreds or thousands of years we must turn to indirect methods of estimating past temperature. Tree-rings feature strongly in reconstructions of past temperatures because trees tend to grow in most locations where humans live, and because of trees leave well-dated records of their growth in each year in the form of annual ring growth. In cold locations, temperature stress may limit ring growth, such that in warmer years, when temperature stress is weaker, more growth may occur, with conversely less growth occurring in cold years. Measurement of this growth is central to our ability to speak about the terrestrial temperature variability over the last 1000 years.

Although this temperature information stored in annual tree ring growth is of immense value in reconstruction of past temperatures, trees are not thermometers. The central goal of this project was to identify and test for the effects of processes which give rise to systematic non-temperature dependent growth in trees whose variability is otherwise primarily controlled by temperature. An understanding of such processes should help us improve our estimates of Earth’s past surface temperatures. This study concentrated on two classes of mechanisms that may give rise to differences in temperature estimates derived from thermometers and tree rings.

First, a number of lines of evidence that we have developed with the support of this grant now support the argument that changes in the amount of light reaching the tree canopy can give rise to separation between temperature and tree­ ring archives. This have been shown most clearly by the result that biases exist in the tree­ring response to volcanism. Early pyrheliometer which record the reduction of light following the reduction in light availability following the Krakatoa (1883) and Novarupta (1912) show that when the temperature and light anomalies following an eruption diverge strongly the tree ­ring anomalies more closely resemble the light anomalies, strengthening the attribution of this bias to light reductions. Furthermore the spatial pattern of volcanic divergence is strikingly similar to the pattern that would be predicted based on light limitation. Taken together, these results support the hypothesis that global dimming contributed to a separation between tree ­ring based estimates of surface temperature and instrumental observations of surface temperature.

The second class of explanations that we have pursued is the argument that choices made in the preparation of tree ­ring chronologies result in the loss of climate information that existed in the raw measurements. We have shown that this is a predictable result of a basic ecological principle, Liebig’s Law of the Minimum. This principle states that, when more than one factor limit growth, plants will respond to the strongest limiting factor. This idea is central to the overall project of climate reconstruction from tree rings, in that it forms the basic argument for why trees in a particular site may respond primarily to one overarching limiting factor on growth, such as temperature or precipitation. We have thus demonstrated a new application of this old principle.

These results also suggest an improved method for reconstructing climate from tree ring records. We find that this approach increases the climate signal recovered from tree rings from all regions of the Earth that support tree growth. Both temperature and rainfall reconstructions are improved, and improvement in climate reconstruction skill happens at all frequencies.

Last Modified: 07/04/2019
Modified by: Alexander Stine