ConclusionsI set out to develop and test a bioclimate model and use this model to project climate change impacts on the distribution of bioclimate envelopes at a regional scale. Using assembled data sources, climate normal and climate change projections, I asked two questions: 1) What are the climatic variables that separate bioclimate subzones, and 2) will the distribution of these bioclimate subzones change in the future? It is possible to use bioclimate subzone mapping to model, with reasonable accuracy, the extent and magnitude bioclimate envelopes shift over time. Though it is possible to further explore magnitude of climate change with future periods beyond 2025 – this study shows that significant climate change is no longer a theoretical future possibility. A shift in biologically meaningful climate parameters is demonstrated using two climate periods 1961-90 and 1991-2016.
Further work is recommended to reduce classification error in the BOH subzone – particularly as this zone shows the greatest potential for change in climate characteristics. There is evidence that the BOH can be split into subzones as other published bioclimate treatments of the Yukon by Strong (2013) and Jorgensen and Meidinger (2015) split the BOH into subregions. However these bioclimate delineations are driven by expert experience, not on plot data. Relevé data housed by Yukon government and private industry could be used to confirm and delineate BOH subzones. Another approach, to improve overall prediction accuracy, is to increase the number of random points used to generate climate variables. Increasing the sample size however is problematic. Random Forest will tend to over-fit the model when provided with sufficiently large number of training data – masking incorrectly assigned training data.
|
Although improvements to the underlying bioclimate map are needed, a bioclimate envelope model derived from a zonal classification system appears, in its current delineation, to be a useful tool to identify patterns of climatic change that has direct biological interpretations. Further, using this method it is possible to identify climate change trends that do not conform to the narrative that climate warming in the Yukon has equal influence on biological potential. Further research in this area could incorporate a “distance” measure to identify areas that are already on the edge of their bioclimate envelope. Such areas might be more vulnerable to climate change. Similarly, areas close to the “centre” of a bioclimate envelope may be able to absorb a greater degree of climatic shift.
Although the zonal site is well expressed in a bioclimate map, only a portion of the landscape is the zonal type and ignoring site (micro-climatic, edaphic and topographic) variability and interaction between ecological processes limits the realism of potential climate change impact assessments. Further analysis could incorporate topo-edaphic constraints to climate change to tease out land facets that are, to a degree, decoupled from the regional climate predicted by downscaled climate models.
Mapping how and where climate change have had (or may have) an influence on the distribution of bioclimate subzones has a practical management application. Areas that a predicted to shift may be targeted to explain observed direct changes of biological potential such as regeneration, mortality, and growth rate. There may also be indirect measures of change in biological potential such as shift in habitat use by wildlife or fire return interval/intensity.
|