Part 1
Fragstats
White spruce stands (cells with 40-100% white spruce population) took up 239906 ha in 2001 and dropped 65% to 84843.75ha in 2011 (Table 2). In 2011, White Spruce only accounted for 0.047% of the study area. The results of core area metrics are the same as the area metrics, meaning that all of the grids are regarded as core areas for white spruce. The number of white spruce stand patches had increased from 751997 to 788392 (5%) over the time period. The lowest COHESION pertains to the white spruce stands class (40-100%). The decrease in COHESION as the percentage of white spruce increase indicates that areas, where white spruces located, are less clumped or aggregated across the landscapes.
Table 2: Changes in Class level metrics for boreal forests in Ontario-Quebec, 2001-2011
Part 2
Current distribution
The result of the 10-fold MaxEnt analysis illustrates the current distribution of white spruce and which variables contributed the most to the distribution. The area under the Receiving Operator Curve (AUC) is used to evaluate model performance (Chart 1). The Maxent model for this analysis was highly accurate as it has a statistically significant AUC value of 0.946, almost 0.05 higher than the null model (0.5).
Figure 3 shows the point-wise mean of the 10 replicated analyses in the cloglog data format. The colours indicate the predicted probability that conditions are suitable for white spruce. Red illustrates the high probability of suitable conditions for the species, green illustrates conditions typical of those where the species is found, and blue illustrates the low predicted probability of suitable conditions. The model largely corresponds to the actual white spruce stands (black dots), although there are a few outliers in the northern end of the study area.
It is apparent that the current occurrences of white spruces cluster along the southern border of Ontario and Quebec. Areas northeast of Lake Superior is particularly favourable to the species. Northern parts of the study area are deemed to be most unsuitable for white spruce under the current condition.
Table 3 gives the estimates of relative contributions of the environmental variables to the Maxent model. To determine the first estimate in each iteration of the training algorithm, the increase in regularized gain is added to the contribution of the corresponding variable or subtracted from it if the change to the absolute value is negative. For the second estimate, for each environmental variable, in turn, the values of that variable on training presence and background data are randomly permuted.
8 environmental variables with >2% percentage of contributions were selected for further analysis based on the analysis of variable contributions (table 3). Together, they accounted for over 90 % of the model. However, it is important to note that annual variables may a higher influence when conducting projection analysis for a long period of time. The 8 variables are BIO5 isothermality, BIO 8 temperature annual range, BIO12 annual precipitation, BIO 14 precipitation of the driest month, BIO 17 precipitation of driest quarter, BIO 15 precipitation seasonality, BIO 16 precipitation of wettest quarter, and BIO 5 max temperature of the warmest month.
Table 3. Contribution of 19 environmental variables to MaxEnt model
* Isothermality is also the environmental variable with the highest gain when used in isolation, which therefore appears to have the most useful information by itself. The result is the same for test gain and AUC on test data as with training gain.
† Precipitation of Driest Quarter is also the environmental variable that decreases the gain the most when it is omitted, which therefore appears to have the most information that isn't present in the other variables. The result is the same for test gain and AUC on test data as with training gain.
Chart 1: Receiving Operator Curve for MaxEnt modelling of the current climate condition
Figure 3. Picture of prediction generated by MaxEnt for the current climate condition
Part 3
Future projection
Pictures of predictions are created for each climate change scenario (RCP 2.6, 4.5, 6.0, and 8.5) for the years 2050 and 2070. The pictures are imported to ArcMap as ASCII files and are overlaid by the province and white spruce stands layers. All of the maps are compiled together for easier comparison (Figure 4) Comparisons are made between the pictures under current climate conditions and that under future projections. 4 main observations are made:
I. The significant northward retreat of the suitable habitat range for white spruce.
As shown in Figure 4, areas with a high predicted probability of suitable conditions (red) and conditions where white spruce is typically found (green) shifted higher along the latitudes in all projected climate change conditions when compared to the current model. However, the shift is not uniform horizontally. Suitable ranges tend to extend more towards James Bay than to other landmasses nearby, as illustrated in Figure 4c, 4d, 4f, 4g. The shift is most apparent under RCP 8.5 in both 2050 and 2070. Light blue areas (although they indicate a relatively how predicted probability) are found above 50°N in most climate scenarios, which are previously uninhabitable for white spruce.
II. Reduction of suitable habitat in general
Areas highlighted in warm colours and green decreased remarkably (Figure 4). This means that the predicted probability of presence is drastically lower under the warmer climate, regardless of the pathway we will take in the coming decades. The higher the GHG emission, the more severe the reduction in suitable habitat range; thus, an increase in temperature and GHG have considerable negative impacts on the presence of white spruce. The effect is mostly felt in the southeastern parts of the study area in 2050 and in the central areas in 2070. This observation is best illustrated in the reclassified maps (Figure 5). Grids with a suitability score 0.5 or higher are classified as potentially suitable habitat and that with a score lower than 0.5 are deemed as unsuitable for white spruce down the road. Suitable habitat contracted significantly from RCP 2.6 to 8.5 and from 2050 to 2070.
III. Expansion of suitable habitats from current climate conditions
Despite the general reduction trend identified earlier, under RCP 4.5 in 2050 (Figure 4b) and RCP 2.6 in 2070 (Figure 4f), the sizes of suitable habitats are noticeably larger than the other scenarios in the respective years. Surprisingly, in Figure 4c and 4f, the number of cells for suitable habitats is comparable to the current occurrence.
IV. Retainment of habitat around Lake Superior
Notwithstanding the general northward migration of suitable conditions, the area north and east of Lake Superior remains to be the potential habitat for white spruce for all climate scenarios. The suitability score around that area varies the least (Figure 4). It remains yellow and red for the most part and the size of which does not change as much as in other parts of the study area. In 2070 RCP 4.5, 6, 8.5, that area is the only location that can possibly withstand the harsh conditions imposed by climate change.
Individual maps for each climate scenario can be found below.
Figure 4: Species distribution modelling under changing climate conditions
Figure 5: Changing suitable climate conditions and habitat range under different climate scenarios.
