Blanding’s turtle (Emydoidea blandingii) select populations COSEWIC assessment and status report 2016: appendix 3
Appendix 3. Estimating the effect of road mortality on Blanding’s Turtles across Ontario
A report in support of the COSEWIC Update Status Report and Status Assessment in progress
David C Seburn1 & Kari Gunson2
1 Seburn Ecological Services, Ottawa, Ontario
2 Eco-Kare International, Peterborough, Ontario
The effect of road mortality on the Ontario Blanding’s Turtle population was estimated based on documented road kill rates from standardized surveys along four roads in the province. There were 1,328.8 km of road within 1x1 km grid squares with recent observations of Blanding’s Turtles. Using two different methods, the observed annual road kill rate varied from 0.2 – 0.3 Blanding’s Turtles/km on surveyed roads. An estimated 265.8 – 398.6 Blanding’s Turtles are killed on roads each year in Ontario based on this range of road kill rates. At the lowest kill rate, the population would decline by 50% in three generations (120 years). The population would decline by 50% in 65 years based on a starting population of 25,000 adults, or in 117 years given a starting population of 45,000 adults.
Traffic mortality is widely acknowledged as a significant threat to turtle populations. Modelling studies predict that populations of semi-terrestrial species such as Blanding’s Turtles (Emydoidea blandingii) likely experience road mortality rates that exceed sustainable levels in many areas with high road density (Gibbs and Shriver 2002). High adult survivorship rates are a key trait of Blanding’s Turtle biology and even modest increases in adult mortality rates can lead to population declines (e.g., Congdon et al. 1993). In addition, road mortality may preferentially affect females who must leave wetlands to find nesting sites (Steen et al. 2006). Road mortality is likely a serious issue for Blanding’s Turtles in Ontario, where the road network has expanded from approximately 7,000 km to over 35,000 km of paved roads from 1935-1995 (Fenech et al. 2001). Compounding this problem, approximately 2.7% of drivers will deliberately run over reptiles (Ashley et al. 2007).
Recently, a few studies in Ontario have documented rates of road mortality of turtles across a few different roads in different parts of the province. Using the data from these studies, we have endeavoured to estimate the annual number of large juvenile or adult Blanding’s Turtles that are killed on the roads each year based on a low, medium and high kill rate per kilometre of road. Using these annual estimates of mortality, we estimated the effect of this level of mortality over the next three generations, or 120 years, on Blanding’s Turtle population size in Ontario.
The area of occupancy of Blanding’s Turtles in Ontario was defined by using 1x1 km grid squares. A 1x1 km grid square was selected because Blanding’s Turtles can easily travel 1 km or more (Joyal et al. 2001; Edge et al. 2010). Given that Blanding’s Turtles can move >3 km (Grgurovic and Sievert 2005), it can be argued that a larger grid square should be selected. Blanding’s Turtles do not move uniformly across the landscape, however, and selecting a larger grid size (e.g., 2x2 km) substantially increases the likelihood of including areas, and therefore roads, not within a population’s typical movement patterns. Selecting a smaller grid square (e.g., 1x1 km) provides a conservative estimate of the effect of roads on Blanding’s Turtles.
The Ontario Reptile and Amphibian Atlas provided us with a GIS shapefile of all Blanding’s Turtle observations records partitioned into 1x1 km grid squares. We included only squares with the most recent record occurring within the last 20 years (1995 or more recently). The amount of road in each recent 1x1 km grid square was determined using the 2013 OMNR roads data from the Ontario Geospatial Data Exchange network. Seven broad classes of road types were defined: arterial, collector, freeway/highway, local, ramp, resource/recreation and service. We included all road types in the analyses.
Next we examined the road kill data from road survey studies. Because adults are more important to the viability of turtle populations than juveniles (Congdon et al. 1993), we only wanted to include road mortality of adults. Not all turtles killed on roads are adults, but unfortunately some dead-on-road (DOR) turtles were in poor condition and size data were not available, but turtles could be reliably assigned to either juvenile or adult age classes. We filtered out all clearly juvenile turtles (e.g., < 15 cm PL) but included all others, providing one kill rate for subadult and adult turtles combined.
Road kill rates for Blanding’s Turtles were calculated using data from pre-existing surveys on the following roads:
- #7, ~80 km surveyed in 2012-2014 (Lesbarrères et al. 2013)
- #41, ~20 km surveyed in 2012-2014 (Lesbarrères et al. 2013)
- #17, ~100 km surveyed in 2014 (Seburn 2014)
- #69, ~12 km surveyed in 2012-2014 (Morin, Riley and First Nations of Georgian Bay unpublished data)
For each surveyed road section, we divided the road into 1 km sections and counted the number of dead Blanding’s Turtles within each section. All sections of road with no observations of Blanding’s Turtle during all years of surveys were excluded from the kill rate calculation to ensure areas without the species were not included in the analysis. For #7, #41 and #69, standardized road survey data were collected over multiple years, allowing for the determination of an average number killed each year in each 1 km section of road, for a more robust estimate of annual kill rate/km. Once a road kill rate or average road kill rate was determined for each 1 km section where Blanding’s Turtles had been detected, we then calculated the median and modal road kill rates across these sections of road.
A second method was also used to calculate road kill rates for comparison purposes. For each road, the number of dead Blanding’s Turtles detected was divided by the total surveyed road length to provide the number of turtles found per kilometre. Using this method includes stretches of roads with no road kill and hence reduces any bias in the first method of excluding road segments where road kill may occur but was not detected. For roads where surveys were conducted over multiple years, a kill rate was calculated for each year, along with an overall average kill rate for each road.
The road kill rates from these two methods were multiplied by the number of km of road in the 1x1 km grid squares with recent observations of Blanding’s Turtles to determine the estimated number of Blanding’s Turtles killed each year on Ontario. Using these estimates of annual road kill, we then calculated the estimated percentage of the Ontario population that is killed each year based on the high and low population size estimates provided in the COSEWIC report. These percentages were then used to calculate the effect on the Ontario population over three generations, or 120 years, as defined in the COSEWIC report.
Over the course of 120 years, recruitment into the adult population, as well as natural adult mortality, are significant factors. Turtle populations generally demonstrate very slow population growth rates, and recruitment and mortality are often essentially balanced (e.g., Shoemaker et al. 2013). We have therefore modelled the effect of annual road mortality on the Ontario Blanding’s Turtle population with the assumption of a stable population (recruitment = natural mortality). Although this assumption may not be true, it allows for examining the effect of road mortality independent of other factors, and does not assume an arbitrary population growth rate. The population decline over time was modelled using a constant annual percentage road kill rate over time in the compound interest equation:
A = P (1 + r/n)nt
A = the future size of the population
P = the original size of the population
r = the annual interest, or kill rate (as a negative decimal)
n = the number of times the rate is compounded per year (once in this case)
t = the number of years
Because the kill rate is only applied once per year, the value n in the equation is equal to 1 and the equation simplifies to:
A = P (1 + r)t
To calculate how many years it would take for the population to decline by 50% the equation was re-arranged to solve for t, setting A/P (the future population size divided by the present population size) to 0.5:
t = ln (A/P)
ln (1 + r)
Results & discussion
Amount of road in Blanding’s Turtle range in Ontario
A total of 2,020 1x1 km grid squares contained observations of Blanding’s Turtles. Only 64% (1,283) of those squares had recent observations (1995 or more recent). This likely underestimates the number of grid squares where the species is currently present as the Blanding’s Turtle can live far more than 20 years. In addition, lack of recent records in the Ontario Reptile and Amphibian Atlas does not confirm that a species is absent (Seburn and Mallon 2015). The recent grid squares contained 1,328.8 km of road. Almost 60% of roads consisted of local roads and 90% of roads were in three categories: arterial, freeway/highway and local (Table 1). Average road length within a grid square was 1.04 km. Given that over 700 grid squares were excluded because of a lack of recent records, the total length of road within the Blanding’s Turtle range may have been underestimated by hundreds of kilometres.
|Road type||# of km||% of total km|
Road kill rate – Method 1
Using the first method, the observed kill rate was determined for 66 sections of 1 km road segments along # 7, 17, 41 and 69. The annual kill rate in each section of road varied from 0.25 - 4.0 Blanding’s Turtles/km (Figure 1). Determining an accurate measure of kill rates/km is difficult. Inevitably some turtles killed on roads will be missed as the road surveys did not always span the entire active season and it is known that predators can remove turtle carcasses from roads and the median persistence time of turtles may be a little as 3 days (Santos et al. 2011). This will underestimate some kill rates. In contrast, some roads were only surveyed for one year, which likely overlooks stretches of road where road kill does not occur annually. Along #7, for example, with four years of surveys, many stretches of road had only one Blanding’s Turtle killed every four years.
The median kill rate across these roads was 0.50 turtles/km, while the modal kill rate was 0.25 turtles/km. The modal kill rate was selected for further analyses to provide a more conservative estimate of Blanding’s Turtle road mortality.
Road kill rate – Method 2
The second method for calculating kill rates indicated that 0.07 to 1.25 turtles/km were killed across the surveyed roads (Table 2). The road kill rate along #17 is quite low, likely for two reasons. First of all, large sections of the road surveyed had no observations of Blanding’s Turtles, there were no previous records of the species in these areas in the Ontario Reptile and Amphibian Atlas, and the roadside habitat appeared inappropriate. Including these large 20-30 km sections of road that likely do not have Blanding’s Turtles would artificially reduce the kill rate for this road. Second of all, only eight road surveys were conducted during this study and so few surveys would undoubtedly underestimate actual levels of road kill.
In contrast, the road kill rate for #69 is quite high. This may be driven by selecting a short section of road (only 12 km) that was a known area of road kill. The road kill rates along #7 and #41 may be more meaningful as they were both determined from longer sections of road. The average road kill rates along these two roads (0.2 - 0.3 turtles/km) are quite similar to the modal kill rate determined using the first method.
|Road||Km surveyed||# of years||Range of annual road kill rates||Mean road kill rate (if more than one yr of data)|
|7||80||4||0.175 – 0.225||0.20|
|41||20||4||0.15 – 0.45||0.3|
|69||12||3||1.08 – 1.58||1.25|
Effect on population
The modal kill rate from the first method (0.25 turtles/km) and the kill rates along #7 and #41 (0.2 – 0.3 turtles/km) were used to examine the effect of road mortality on the Ontario population. Based on these three road kill rates, an estimated 265.8 to 398.6 Blanding’s Turtles are killed on roads in Ontario each year (Table 3). The current draft COSEWIC report estimates that the Ontario adult Blanding’s Turtle population is between 25,000 and 45,000 adults (T. Piraino personal communication). Based on a population of estimate of 25,000 adults, an estimated 1.06 – 1.59% of the population is killed each year (Table 3). Assuming a population of 45,000 adults, then an estimated 0.59 – 0.89% of the population is killed each year.
|Road kill rate (turtles/km)||Total annual kill||% of 25,000 population||% of 45,000 population|
Assuming a current population of 25,000 Blanding’s Turtles, all three road kill rates result in significant declines over the next three generations, or 120 years (Figure 2). Even taking a road kill rate of 0.2 turtles/km, the most conservative estimate, the population would decline by 50% in 65 years (Table 4).
|Road kill rate (turtles/km)||Time to reduce population of 25,000 adults by 50% (years)||Time to reduce population of 45,000 adults by 50% (years)|
Increasing the initial population size estimate to 45,000 adult Blanding’s Turtles still results in a significant decline at all three road kill rates (Figure 3). The lowest road kill rate, 0.2 turtles/km, would cause the population to be reduced by 50% in 117 years, just slightly less than three generations.
Possibly a buffer of 1 km around recent Blanding’s Turtle observations has captured more roads than the turtles will likely encounter. If that is the case, our results overestimate the number of turtles killed each year. Given that Blanding’s Turtle home range lengths >3 km (Grgurovic and Sievert 2005) have been documented, it seems, if anything, we have underestimated the amount of roads Blanding’s Turtles will encounter.
The calculated road kill rates in this study are based on only a few roads in the province. All of these roads are major roads and may not be representative of all roads included in the provincial range of the species (see Table 1). While #7 is well known as a hotspot for turtle mortality by Ontario turtle biologists, it is noteworthy that #41 had a higher average kill rate (Table 2), despite the fact that #7 has approximately 30% higher traffic volumes than #41 (MTO 2012). Traffic volume alone does not determine the amount of road kill. Given that turtles will often become immobile when a car passes nearby (Seburn, personal observation), road mortality can result even under lower traffic volumes. We have conservatively estimated the amount of road Blanding’s Turtles are exposed to as well as the road kill rates on the surveyed roads. This may compensate for the fact that road kill rates may be lower on some other roads, but we acknowledge this uncertainty.
One variable we have not included in our modelling is the growth rate of the Ontario population over time. A ten year study of Bog Turtles (Glyptemys muhlenbergii) found that the recruitment of new adults into the population each year approximately equalled the number of mortalities (Shoemaker et al. 2013). In our study, net annual adult recruitment (recruitment – natural mortality) would have to be at least 1.06% to balance the effect of road mortality based on an estimated 25,000 adult population and the lowest road kill rate (Table 3). Although such growth rates may be possible in some areas, it seems likely that many Ontario populations demonstrate limited growth once natural mortality is included. In addition, our modelling does not take into account any increase in the amount of roads over time, or any increase in the number of vehicles on roads. For example, there were 6.9 million vehicles registered in Ontario in 2005 (NRC 2007) but this number increased to 8.2 million by 2014 (Statistics Canada 2015). In conclusion, the best evidence at hand suggests that road mortality is a widespread threat to the Blanding’s Turtle and that the Ontario population will decline by more than 50% over the next three generations from this threat alone.
Teresa Piraino was immensely helpful in facilitating data access, administering the work on this project, and providing review comments. Jacqueline Litzgus coordinated the project and provided valuable guidance and review comments throughout all stages. The Ontario Reptile and Amphibian Atlas graciously and quickly provided the required Blanding’s Turtle location data. Ruben Boles, Ron Brooks, Justin Congdon, and Joe Crowley provided valuable review comments. We are grateful to the Toronto Zoo for providing financial support to conduct this work.
Ashley, P.E., A. Kosloski and S.A. Petrie. 2007. Incidence of intentional vehicle-reptile collisions. Human Dimensions of Wildlife 12: 137-143.
Congdon, J.D., A.E. Dunham and R.C. van Loben Sels. 1993. Delayed sexual maturity and demographics of Blanding’s turtles (Emydoidea blandingii): Implications for conservation and management for long-lived organisms. Conservation Biology 7: 826-833.
Edge, C.B., B.D. Steinberg, R.J. Brooks and J.D. Litzgus. 2010. Habitat selection by Blanding’s Turtles (Emydoidea blandingii) in a relatively pristine landscape. Écoscience 17: 90-99.
Fenech, A., B. Taylor, R. Hansell and G. Whitelaw. 2001. Major road changes in southern Ontario 1935 – 1995: Implications for protected areas. In Proceedings of the Fourth International Conference on the Science and Management of Protected Areas, S. Bondrup-Nielsen, N.W.P. Munro, G. Nelson, J.H.M. Willison, T.B. Herman, and P.F.J. Eagles (eds.). University of Waterloo, Waterloo, Ontario, Canada. Pp. 365-383.
Gibbs, J.P. and W.G. Shriver. 2002. Estimating the effects of road mortality on turtle populations. Conservation Biology 16: 1647-1652.
Grgurovic, M. and P.R. Sievert. 2005. Movement patterns of Blanding’s turtles (Emydoidea blandingii) in the suburban landscape of eastern Massachusetts. Urban Ecosystems 8: 203-213.
Joyal, L. A., M. McCollough and M.L. Hunter. 2001. Landscape ecology approaches to wetland species conservation: a case study of two turtle species in southern Maine. Conservation Biology 15: 1755-1762.
Lesbarrères, D., K. Gunson and D.C. Seburn. 2013. Monitoring turtle movements across highways 7 & 41: Final report. Unpublished report to Highway Infrastructure Innovation Funding Program, Ministry of Transportation of Ontario.
MTO. 2012. Traffic volumes 2012. Report published by the Ministry of Transportation of Ontario. Website accessed 26 June, 2016.
National Research Council (NRC). 2007. Canadian Vehicle Survey 2007. Website accessed 2 December, 2015.
Santos, S. M., F. Carvalho and A. Mira. 2011. How long do the dead survive on the road? Carcass persistence probability and implications for road-kill monitoring surveys. PLoS One 6:e25383.
Seburn, D.C. 2014. Identification of species at risk turtle hotspots along highway 17 -- Final report. Unpublished report to the Canadian Wildlife Federation.
Seburn, D.C. and E. Mallon. 2015. Have salamanders declined in Ontario? Unpublished report to Ontario Nature.
Shoemaker, K. T., A.R. Breisch, J.W. Jaycox and J.P. Gibbs. 2013. Reexamining the minimum viable population concept for long-lived species. Conservation Biology 27: 542-551.
Statistics Canada. 2015. Motor vehicle registrations, by province and territory (Quebec, Ontario, Manitoba). Website accessed 2 December, 2015.
Steen, D.A., M.J. Aresco, S.G. Beilke, B.W. Compton, E.P. Condon, C. Kenneth Dodd Jr, H. Forrester, J.W. Gibbons, J.L. Greene, G. Johnson, T.A. Langen, M.J. Oldham, D.N. Oxier, R.A. Saumure, F.W. Schueler, J.M. Sleeman, L.L. Smith, J.K. Tucker and J.P. Gibbs. 2006. Relative vulnerability of female turtles to road mortality. Animal Conservation 9: 269-273.
Report a problem or mistake on this page
- Date modified: