The browser you are using is not supported by this website. All versions of Internet Explorer are no longer supported, either by us or Microsoft (read more here:

Please use a modern browser to fully experience our website, such as the newest versions of Edge, Chrome, Firefox or Safari etc.

Johannes Edvardsson

Johannes Edvardsson


Johannes Edvardsson

A statistical model for the prediction of the number of sapwood rings in Scots pine (Pinus sylvestris L.)


  • Johannes Edvardsson
  • Kári Rögnvaldsson
  • Elín Þóra Helgadóttir
  • Hans Linderson
  • Birgir Hrafnkelsson

Summary, in English

Dendrochronology is a well-established dating method for wooden objects, but due to surface processing of construction timber or natural degradation the dating of historical wood often relies on a prediction of the number of missing rings based on sapwood statistics. Since Scots pine (Pinus sylvestris L.) is one of the most common tree species in north-western Europe, the absence of reliable sapwood statistics and models for the prediction of missing sapwood rings for pine samples is remarkable. We have therefore produced sapwood statistics based on data from 776 pine trees with ages from 15 to 345 years. The material consists of both living trees and historical timber, with varying growth rates, geographic settings, and from different soil types. When the whole material is considered, the average age of the trees is 103 years, and the number of sapwood rings is 54 ± 15 (1 SD), but range from 18 to 129. Trees less than 100-years in age contained 46 ± 11 (1 SD) sapwood rings and had an average tree-ring width (TRW) of 1.76 mm. With increasing age, the average TRW decreased while the number of sapwood rings increased. The average TRW of 101–200-year-old trees is 0.99 mm while the samples contained 63 ± 12 (1 SD) sapwood rings. For trees older than 201 years, the average TRW is 0.64 mm while the number of sapwood rings increased to 85 ± 16 (1 SD). The two most important factors in determining the number of sapwood rings for a given tree when only heartwood statistics are available proved to be (i) the number of heartwood rings and (ii) the average TRW of the heartwood rings. For incomplete samples, we have therefore developed a statistical model based on the sample's heartwood rings (number and average width) to compute a prediction interval for the total number of rings. The sapwood and heartwood statistics suggest a statistical model for the number of sapwood rings with mean that increase with the number of heartwood rings. Furthermore, the average number of sapwood rings decreases with the mean width of the heartwood rings. However, the predictive power of the mean width is limited when the number of heartwood rings has already been taken into account. Thus, we suggest making predictions for the number of sapwood rings using only the number of heartwood rings. Predictions of the number of sapwood rings based on the statistical model where convincing in the case of the three different datasets that were analysed. The certainty in these predictions was such that the width of the 80% and 95% prediction intervals ranged 28–34 and 45–52 sapwood rings, respectively. Additionally, we demonstrate how make predictions when there is information about the number of remaining sapwood rings in a given sample. To make the sapwood model available, we present a free online R package for fitting our models and an online software dashboard.


  • Quaternary Sciences
  • BECC: Biodiversity and Ecosystem services in a Changing Climate

Publishing year








Document type

Journal article




  • Wood Science


  • Bootstrap methods
  • Dendrochronology
  • Heartwood
  • Sapwood
  • Sweden




  • ISSN: 1125-7865