Seeing the wood for the trees. Rethinking 700 years of vegetation change in Iceland using meta-analysis of palaeoecological datasets and landscape scale model reconstructions

  • Willem Wilmer Koster

Student thesis: Doctoral Thesis (PhD)


The colonisation of Iceland around 870 A.D. saw the influx of Norse settlers to a previously uninhabited island, resulting in large-scale ecological changes. Human impacts on the landscape vary in time and are spatially complex, making it difficult to accurately assess how Iceland was affected.
To improve our knowledge about the spatial and temporal patterns of landscape changes in Iceland, this thesis uses quantitative approaches to analyse existing palaeoenvironmental data from Iceland. A meta-analysis was used to determine the spatial bias in and the temporal quality of Icelandic pollen sites. Relative pollen productivity (RPP) estimates are calculated for seven ecologically important taxa, by analysing pollen-vegetation relationships at eighteen sites. The RPPs serve as input for quantitative pollen-based reconstruction models, the Multiple Scenario Approach. Landscape reconstructions totalling 3825 km² were generated for three sites (Mývatn, Reykholtsdalur, and Skálholt) and three time slices (577-877 CE, 877-1077 CE, and 1077-1277 CE).
The analysis of spatial bias shows that pollen sites in Iceland are closer to farms and have a higher annual mean temperature and precipitation than a sample of random sites. The RPPs of Icelandic taxa are lower than RPPs in comparable areas. Different spatial patterns in vegetation cover emerge from the reconstruction model outputs, Mývatn remains wooded throughout the time slices, while Reykholtsdalur becomes mostly deforested from 1077 CE onwards.
The spatial analysis shows the need for thorough analysis of the representativeness of current palynological datasets, which helps identify where future sampling should take place to create more representative datasets. The characterisation of pollen-vegetation relationships shows that there is much more to be understood about the mechanics behind differences in RPP estimates in different
locations. Quantitative reconstruction methods provide a useful tool for gaining insight into differences in land cover changes in Iceland, and to visualise different spatial patterns in the landscape.
Date of Award28 Nov 2023
Original languageEnglish
Awarding Institution
  • University of St Andrews
SupervisorRichard Thomas Streeter (Supervisor)


  • Iceland
  • Palynology
  • Pollen modelling
  • Multiple scenario approach
  • Vegetation reconstruction model
  • Spatial bias model
  • Chronological quality
  • Pollen productivity estimates
  • Relative pollen productivity

Access Status

  • Full text embargoed until
  • 26 July 2026

Cite this