Roads negatively affect animal populations by presenting barriers to movement and gene flow and by causing mortality. We investigated the impact of a secondary road on a population of black ratsnakes (Elaphe obsoleta) in Ontario, Canada by radio-tracking 105 individuals over 8 years. The road was not a significant barrier to movement and none of the reproductive classes examined (male, non-reproductive female, reproductive female) avoided crossing the road. However, the road was a significant source of mortality. From a total of 115 road crossings by radio-implanted snakes, 3 individuals were killed by cars, resulting in a mortality rate of 0.026 deaths per crossing. We multiplied this mortality rate by the total number of expected road crossings by all individuals in the population in an active season (340) to estimate the number of road kills (9 individuals) each year. This estimate was higher than the actual number of road kills found, but half the number estimated from road kill models. Population viability analysis revealed that our estimate of road mortality was enough to increase the extinction probability for this population from 7.3% to 99% over 500 years. Road mortality of more than 3 adult females per year increased the extinction probability to >90%. Our results strengthen the view that road mortality can have a pronounced negative effect on populations of long-lived species.
Study area and species
We conducted this study from April 1996 to October 2004 at the Queen’s University Biological Station (QUBS), 100 km south of Ottawa, Ontario, Canada. The study area was approximately 2000 ha and comprised mainly of rolling terrain covered with second-growth deciduous forest, intermixed with numerous old fields, rocky outcrops, lakes, and marshes. One secondary road (Opinicon Road) bisects the study area from northeast to southwest and services numerous smaller cottage roads (Fig. 1 in Blouin-Demers and Weatherhead, 2002). Here we consider only the effect of Opinicon Road. Because the road is used principally to access summer cottages, traffic is heaviest during the summer months, coincident with the snakes’ active season. The road is approximately 6 m wide including the shoulders, with a narrow (4 m) margin of grass and herbaceous vegetation. Beyond this margin the dominant habitat is mixed deciduous forest. During the study, the road surface was primarily gravel. Although the speed limit was 80 km/h, vehicles generally traveled somewhat more slowly (∼60 km/h) because of the numerous curves and hills.
Our ratsnake population has been studied continuously since 1981 (Weatherhead et al., 2002). Here we refer to the study species as the black ratsnake (E. obsoleta), but note that recent genetic analyses (Gibbs et al., 2006) create some uncertainty regarding the eventual taxonomic designation.
Within the study area, we captured individuals opportunistically throughout the active season and at hibernacula during spring emergence (Blouin-Demers et al., 2000, Row and Blouin-Demers, 2006). We implanted radio-transmitters (Model SI-2T, battery life of 24 months at 20 °C, Holohil Systems Inc., Carp, Ontario) in 105 individuals (25 males, 52 non-reproductive females, and 28 reproductive females) for periods ranging from a few weeks to a few years. Individuals with radio-transmitters were located approximately every second day. We recorded the UTM coordinates with a GPS at each location to map the movement paths of individuals.
For the road avoidance analyses, we used individuals that were tracked for a complete active season (May–September) only and divided them into 3 groups based on their reproductive status: males (N = 15), reproductive females (N = 17), and non-reproductive females (N = 34). Female ratsnakes rarely reproduce every year. Thus, we considered individuals tracked in multiple years (N = 14) independently each year they were tracked. We calculated the degree of road avoidance for each individual by comparing the actual number of road crossings made by the individual to the number of road crossings it would have made if it moved randomly with respect to the road. We determined the number of crossings made by a randomly moving snake by generating 20 ‘random walk’ movement paths for each individual (Klingenbock et al., 2000, Koenig et al., 2001) in ArcView 3.2 (Environmental Systems Research Institute, Redlands, California). Each random movement path started in the same location as its paired snake movement path and had the same chronological series of distances moved, but we used a randomly determined bearing between each move. For each individual, we took the difference between the mean number of road crossings for the 20 random movement paths and the actual number of road crossings made by the individual. We then compared these values between the reproductive classes. Finally, we determined overall road avoidance for each reproductive class by testing the distribution of differences between real and random crossings against 0.
We determined the effect of road mortality on this population by 1) determining if the risk of mortality was the same for all reproductive classes, and 2) by calculating the mortality rate of radio-implanted individuals and using this rate to estimate the total number of adult roads kills expected each year. We estimated the risk of road mortality for each reproductive class by comparing the total number of road crossings made between classes and months. We used the same classes and individuals as in the road avoidance analysis. The test, however, was different from our measure of road avoidance because a particular reproductive class may have a greater road mortality risk simply because it moves more. Because we compared the number of road crossings to a paired randomly moving snake in the road avoidance analysis, this greater risk would not have been detected. We expected the road mortality risk to be higher for snakes that hibernated closer to the road and, therefore, measured the distance from each individual’s hibernaculum to the road and included it as a covariate in the analysis.
We calculated road mortality rate by dividing the total number of radio-tracked individuals that were killed by vehicles by the total number of road crossings for all 105 individuals (including individuals not tracked for a full active season). We then estimated the expected number of road kills each year by multiplying the road mortality rate by the expected number of road crossings for all adults in the population. To determine if our estimate of the total number of road kills was reasonable, we compared the results of our calculations with the actual number of road kills found on the road each year. In all years, the whole length of the road within our study area (10 km) was usually driven at least once a day in both directions and any road kill encountered was collected, but we did not quantify search effort formally. The exception was 1997 when an extensive road kill study was conducted and the road was driven slowly once a day and biked once a week.