DISCUSSION

As southern Ontario becomes increasingly urban, planning departments in the Region of Waterloo need to consider the ways in which roads can destroy and fragment wildlife habitat and reduce biodiversity (Baxter et al. 2015). According to the 2011 Census by Statistics Canada, 86% of Ontario residents live in urban areas, a number that has most likely increased by 2016. This is especially pertinent in the case of the snapping turtle, whose seasonal trips to find nesting sites can sometimes put them in direct encounter with high-use, multi-lane roads. In Southern Ontario, the total length of major roads has increased 5-fold between 1935 and 1995 (Gunson and Shueler 2012). While snapping turtles are not an endangered species, they are very vulnerable to disturbances in the mortality rate and population recovery can sometimes be difficult (Canada Species at Risk Profile Website). Furthermore, habitat loss is pushing turtles further and further into more fragmented areas, “primarily due to conversion of wetlands to agriculture and urban development” (Canada Species at Risk Registry). Through a weighted Multi-Criteria Analysis we aimed to find “hotspot zones” of likely snapping turtle habitat and identify if and where these habitats intersected with high-use roads.

 

ERROR AND UNCERTAINTY

1. Data: As with any analysis, the way that data is sampled can introduce error into the process. For our analysis we primarily used information from Land Information Ontario, and Region of Waterloo Open Data, but the metadata did not include detailed information on how the spatial information was obtained. As some of the data was very spatially specific and detailed, errors in accuracy and precision on the ground could have detrimental effects on the rendering of the data layer. In addition, the data layers of land cover and soil type are susceptible to the Modifiable Areal Unit Problem, in that pixels with a certain attribute could actually include a significant amount of another attribute (or the pixel could just be wrongly attributed altogether). Furthermore, we ensured that our analysis was as temporally relevant as possible by using only datasets from the last 10 years, recognizing that this is still a relatively long amount of time in which significant changes may have occurred.
2. Data Manipulation and Reclassification: As some of our data was projected in the wrong coordinate system, re-projection introduces measurement uncertainty in the resulting projected features. In addition, before we performed our MCE, we reclassified each dataset into 5 classes, a way of generalizing that can mask or create misleading patterns in the data. Furthermore, as our final habitat clusters were aggregated into 1000×1000 pixels, this introduces more generalization into the data, but the goal of our analysis was not to find sites at an extremely fine level of detail, rather, the goal was to identify broader areas in which are likely conducive to snapping turtle habitat. Of course, these areas may also include areas that were identified by our MCE as being less favourable to snapping turtle habitat, but the mobility and range of snapping turtles means that one small sliver of unacceptable habitat within a larger area of suitable habitat is not very likely to inhibit their use of the area.
3. Estimating/Assuming Suitable Factors: Finding relevant data was sometimes a problem. For example, we were not able to find turtle sighting data with an appropriate density of observation, so we used preferable habitat factors instead. Using the literature, we established relevant parameters to run a Multi-Criteria Evaluation, but of course, this introduces uncertainty in that it does not measure the phenomenon of turtle prevalence directly. In actuality, turtles may choose to inhabit or not inhabit areas for reasons outside of those included in our analysis. Furthermore, while GIS analysts will usually work with a biologist to identify habitat analysis factors, we had to rely on the estimations from the literature. This necessarily included a certain amount of guesswork.
4. Nature of the Subject: Snapping turtles are fairly resilient, adaptable animals with few specific habitat needs. This means that we cannot say that the presence or absence of one particular factor can accurately predict their habitat preferences, but we believe that combining multiple habitat predictors together gave us a more relevant representation of their potential habitat.

 

RECOMMENDATIONS FOR ANALYSIS IMPROVEMENT 

The primary way that this analysis could be improved, or verified, would be to test it with actual data on turtle individuals, comparing areas identified as highly suitable with areas identified as having a low suitability for habitat. Without this information, we have no way of knowing whether our analysis was an accurate predictor.

In our results, we only examined the intersection of highest-use roads and with the most favourable habitat zones. This left us with 6 zones of high importance. However, this analysis could be expanded to include zones with lower importance or less busy roads, as even these pose threats to turtle mobility.

Finally, another factor to consider in snapping turtle mortality is the fact that they are hunted for game. We could not include this in our analysis due to lack of spatial data, but it should be considered in further studies on snapping turtle habitat.

 

RELEVANCE OF ANALYSIS (HOW THE INFORMATION CAN BE USED)

Ryan et al (2013) argue that there while it is understood that processes associated with urbanization, such as increasing road networks, can affect snapping turtle population, there is “a lack of understanding of their spatial ecology in urban areas where these threats are the greatest” (614). Our analysis identified key snapping turtle habitat hotspots within the region of Waterloo within which turtles are at risk of road mortality. With this information, planners can identify ways of mitigating this threat to snapping turtle population health. Gunson and Shueler (2012) identify two main types of intervention to address habitat connectivity: 1) Structural interventions, such as underpasses, fencing, or wildlife bridges, or 2) Behavioural/public awareness interventions, such as wildlife crossing signs (which are cheaper and easier to implement). It is recommended that crossing signs are placed 1 km apart, and paired over the continuous hotspot for effective motorist awareness (Gunson and Shueler 2012). We recommend that in identified hotspot zones 1-3 and 6, the most effective mitigation measure would be turtle crossing signs, as these intersect with country and residential roads in which drivers can be made effectively aware of the dangers. However, hotspot zones 4 and 5 intersect with multi-lane freeways, and the high speeds and structural barriers of these roads may be more effectively addressed through structural interventions such as fencing which funnels turtles towards a road crossing underpass.

Another use for this information is in the creation of turtle conservation areas, such as the aforementioned Turtle Park. This analysis can be used to identify areas which are likely to have higher concentrations of turtles already, and in which turtle habitat could be enhanced by creating designated conservation areas.