Modelling the impacts of climate change on tree growth

 

Models of one form or another have been used to predict forest yield in the UK since the 1920s^,, and are now essential for forest management, particularly in the area of production forecasting. Today's models incorporate data from sample plots going back to the inception of the Forestry Commission. They are consequently highly developed, with accurate predictions of timber yield, but are static and unable to account for changes in climate or atmospheric composition at a given site. A different modelling approach is now required, not just for assessing the impacts of climate change, but for many other reasons which include the prediction of canopy and forest structure, landscape visualisation for planning and predictions of the ecological and hydrological impacts of forest management. 

Stand-based empirical models

Empirical production forecast models are based upon mensurational variables (top height, basal area, and diameter at breast height) recorded for single species, even-aged stands under specific management options, including different spacing and thinning histories. They are therefore only available where specific sample plot data have been collected. Examples include the UK Forest yield tables. Yield models of this type are only valid for projecting forward under constant environmental conditions (climatic and nutritional). This type of model is therefore unsuitable for the projection of the impacts of climate change on forest yield, and cannot be applied to species or management prescriptions which are not included in the sample plot network. Although traditional yield models of this type are unable to assess climate change impacts properly, they provide a foundation to more complex modelling approaches. Models of this type have been further developed through the use of expansion factors for non-merchantable fractions (branch, bark, root) to provide an estimate of carbon stocks in biomass.

Physiological process-based models

The approach described above relies on empirical mensuration data to provide the information on which the models are based. In contrast, process-based models only use empirical yield data to constrain outputs, and to provide allocation parameters. Most of these models rely on relationships between climate, nutrient supply, other site factors as well as physiological processes including photosynthesis and evapo-transpiration to provide detailed information on the many factors which result in tree growth. The time-frame of input variables can vary between one year and 30 minutes, while the model itself can operate at the individual tree, stand or regional level. Because these models are based on physiological processes, they provide the opportunity to incorporate the results from impact studies, where changes in model parameters have been observed in response to enhanced CO2 levels and other environemntal variables. Process-based models can also be linked to soil modules which include nutrient pools. The resultant ecosystem model is then self-perpetuating and able to model the impacts of climate change in the long-term. Process models are particularly good at representing change as compared to absolute values. Linked process-empirical models such as ForestGrowth are likely to represent the most suitable approach to modelling the impacts of climate change.

There are limitations to process-based models, and currently, they are not widely available for British forest conditions, because of the difficulties and uncertainties of parameterising for the environmental variables and tree species grown in Britain. More seriously, the data required for running process-based models is often not available at the national scale. These limitations therefore strengthen the case for an alternative approach to climate change impact modelling.

Knowledge-based modelling approaches

The Ecological Site Classification (ESC) decision support system relates spatial distribution patterns of productivity to related key climatic variables, thus providing the opportunity to model the impacts of predicted climate change. There are, however, two drawbacks. Firstly, they do not account for the predicted rise in atmospheric CO2 or for the effect that this environmental variable may have on tree physiology. Secondly, as for any predictive model, the use of ESC is reliant on the availability of productivity or distribution data for the climatic conditions predicted for the UK in the future. Putting these two drawbacks aside, ESC does provide a spatial representation of how species suitability or distribution may change as a result of a changing climate.

A similar approach is adopted in the SPECIES model which has been developed in the MONARCH project to predict the impacts of climate change on a range of native fauna and flora, based on their current European distributions. In both this model and ESC, future distributions and suitability are based on the link between current climatic conditions and distribution.