Discussion
This
section discusses the results found, the methodology applied and the
uncertainties of these two, as well as perspectives is drawn.
Results
The results indicate that it is hard to
make a reliable model to project the spread of the MPB. This is mainly due to
the poor availability of sufficient data (see uncertainty section) as well as
the binary nature of our model. Real world ecosystems rarely show abrupt
boundaries but rather a smooth continuum. No parameter stands out significantly
as the corrupting factor; instead the limiting factor was different each year,
so there are no obvious conclusions to be drawn from this.
Generally we did not predict the spread of the MPB to a satisfying extend, although it must be said, that it is hard to judge the model strictly on the comparison with the HectaresBC projection as it is not obvious from the metadata, how they calculated their susceptibility.
Generally we did not predict the spread of the MPB to a satisfying extend, although it must be said, that it is hard to judge the model strictly on the comparison with the HectaresBC projection as it is not obvious from the metadata, how they calculated their susceptibility.
Methodology
In retrospect our model could most likely have been improved significantly if we had constructed susceptibility classes allowing for less than 100% fit of all the parameters. The MPB is part of a dynamic ecosystem where abrupt boundaries are rare, so an index value indicating the likelihood of susceptibility instead of a strict binary model would probably have produced a more reliable output.
Uncertainty
Climate data
A major source of uncertainty in this
project is the temporal resolution of our data. Climate BC only offers mean
values at the four temporal resolutions monthly, seasonal, annual and normal
period.
Minimum coldest month temperature <-40°C:
For this parameter we needed the absolute minimum annual temperature instead of the mean minimum annual temperature. This has most likely increased the area susceptible according to this parameter, as the absolute minimum temperature is probably significantly lower than the mean minimum temperature.
Maximum temperature above 18.3°C at least 5 % of August:
For this parameter we needed daily data to get the ratio of days above 18.3°C, but instead we used mean maximum August temperature as a proxy. This has most likely decreased the area significantly.
Spring precipitation lower than long term average:
To let this parameter be a binary layer is probably too conservative, as this parameter is only included because the resistance of the lodgepole pine to MPB attacks decreases if the water supply is scarce. Even though the water supply is sufficient the species cannot resist an epidemic attack, so this parameter should have been a factor enhancing the susceptibility to the MPB rather than a binary condition. Furthermore, this parameter had to be excluded from the future projections (2020 scenarios), as the entire study area would have been classified as not suitable due to the projected increase in spring precipitation.
Finally, the suitable circumstances should be present for two consecutive years for the beetle to develop properly. This is not incorporated in our model.
Secondly, another major source of uncertainty is that the projections made by Hectares BC are cumulative while our model only predicts the suitable habitats for one particular year. To improve the model it should be cumulative.
Minimum coldest month temperature <-40°C:
For this parameter we needed the absolute minimum annual temperature instead of the mean minimum annual temperature. This has most likely increased the area susceptible according to this parameter, as the absolute minimum temperature is probably significantly lower than the mean minimum temperature.
Maximum temperature above 18.3°C at least 5 % of August:
For this parameter we needed daily data to get the ratio of days above 18.3°C, but instead we used mean maximum August temperature as a proxy. This has most likely decreased the area significantly.
Spring precipitation lower than long term average:
To let this parameter be a binary layer is probably too conservative, as this parameter is only included because the resistance of the lodgepole pine to MPB attacks decreases if the water supply is scarce. Even though the water supply is sufficient the species cannot resist an epidemic attack, so this parameter should have been a factor enhancing the susceptibility to the MPB rather than a binary condition. Furthermore, this parameter had to be excluded from the future projections (2020 scenarios), as the entire study area would have been classified as not suitable due to the projected increase in spring precipitation.
Finally, the suitable circumstances should be present for two consecutive years for the beetle to develop properly. This is not incorporated in our model.
Forest data
The overall most limiting factor in our model is the extent of lodgepole pine trees in the right age (80-160 ys). The metadata at HectaresBC does not say whether it counts only live trees or takes dead stands into account too. The metadata from Hectare BC does not clearly state how their projection is made, but they indicate that they only use the volume of pine, without consideration to the age of the stand. This has been a major restriction in our model.MPB data
First, it must be clarified that what is called observed pine killed in this report is also projected. The projection used by HectaresBC is BCMPB, ver.4 for the years 1999-2002 and BCMBP, ver. 5 for the future projection (Hectares BC, metadata).Secondly, another major source of uncertainty is that the projections made by Hectares BC are cumulative while our model only predicts the suitable habitats for one particular year. To improve the model it should be cumulative.
Spatial resolution
The cell size of the data from Hectares BC is 100x100m, while the cell size for the climate data is 1000x1000m. As the lodgepole pine data was obtained from HectaresBC and was the most restrictive, the area predicted by our model as being suitable might have been different if the cell size of this layer was bigger.Perspectives
An outbreak the size
of the current MPB outbreak has severe costs, not only to the ecosystem
of the forest, but also to the humans that use this forest for
commercial or recreational value. Kurz et al. (2008: 987) estimates
that timbers losses in commercial forests are more than 435 million
cubic metres. The forest sector has responded by increasing the pine
portion of the total amount harvested from 31% in 2001 to 45% in 2004,
as well as reallocating harvest. It seems like they are trying to "log"
their way out of the problem, which is most likely a rather unfortunate
approach, as the associated tree-planting creates stands composed by
almost pure lodgepole pine, which will become very susceptible to MPB
attacks in the future.
Also, as noted earlier the decomposition of such a significant amount of dead trees has a huge impact on the amount
of carbon dioxide released to the atmosphere. In the worst case scenario, the emissions
from the dead trees could be bigger than the anthropogenic emissions in
BC (Kurz et al., 2008).