To make these models useful for decision and policy makers their performance needs to be tested and the uncertainties in their predictions clarified. Until now, the validation of models has often been insufficient, as it has been performed internally on a pseudo-independent data set sampled from the same study area, or by re-sampling the training data set. Thus currently it may be unwise to base risk assessments on such models.
A powerful approach that can to some extent overcome the above shortcomings is to validate models by projecting them into the past and comparing them with independent indicators of past conditions. Previously this has been applied to climate models; here we use the approach with models that describe ecological processes. The approach must avoid reliance on using the same dataset (for example fossil pollen values) to represent past plant communities and also to derive or validate palaeoclimatic datasets. In cold-climate regions, a new type of palaeobiological archive is now available: ancient DNA (aDNA) in cold or frozen sediments and soils, which reflects the plants growing at a location in the past. The DNA-based record of plant community composition, dated using radiocarbon techniques, promises a method of verification of ecological model predictions that removes the potential circularity described above.
In addition, it is important to evaluate temporal niche stability and the likelihood that plants can persist through time, even in adverse conditions, without migrating to more favourable habitats. Some of the uncertainty in future predictions relates to the fact that models are typically constructed on the assumption of no change in the niche over time and that there is an immediate response to climate change (no resilience). Clonal plants may be highly resilient to environmental change. For example, individuals of the high-alpine sedge Carex curvula are over 2000 years old and persisted in situ through the changes of the Little Ice Age and Medieval Optimum. DNA fingerprinting of clonal plants can be used to determine clone age, which may reach to centuries or millennia.
Project description
