49 TRACKING MOOSE- AND DEER-VEHICLE COLLISIONS USING GPS AND LANDMARK INVENTORY SYSTEMS IN BRITISH COLUMBIA Caleb Sample1, Roy V. Rea1, and Gayle Hesse2 1University of Northern British Columbia, 3333 University Way, Prince George, British Columbia, Canada V2N 4Z9; 2Wildlife Collision Prevention Program, British Columbia Conservation Foundation, 4431 Enns Road, Prince George, British Columbia, Canada V2K 4X3. ABSTRACT: Vehicle collisions with moose (Alces alces) and deer (Odocoileus spp.) pose a serious threat to all motorists travelling highways traversing habitats of these two ungulates. In British Columbia, mitigation measures to reduce such collisions are based on spatially-accurate records of collisions involving moose and deer that are collected by the province’s highway maintenance con- tractors. To date, the British Columbia Ministry of Transportation and Infrastructure (BC MOTI) uses the paper-based Wildlife Accident Reporting System (WARS) established in 1978 to maintain carcass records. We compared carcass location data collected in 2010 to 2014 by BC MOTI using WARS to that collected by Northern Health Connections bus drivers using a newly developed GPS-based sys- tem (Otto® Wildlife device). In total, 6,929 carcasses (1,231 moose, 5,698 deer) were recorded using WARS and 474 (167 moose, 410 deer) using the Otto® Wildlife device. We compared data collected along 2,800 km on the same highways in the same seasons of the same years. We found more carcass locations were identified with the WARS method, but that in certain geographic regions, the Otto® Wildlife system identified several unique locations. We contend that more complete and finer-scale carcass location data is possible using a GPS-based system such as Otto® Wildlife, than currently collected solely with the paper-based WARS method. ALCES VOL. 56: 49–61 (2020) Key Words: Alces, car, collision, deer, GPS, moose, Odocoileus, roadkill, survey, vehicle British Columbia is a province richly inhabited by large mammals and where populations are abundant and highways traverse their habitat, wildlife vehicle collisions (WVC) are a management concern. Between 2013 and 2017, annual collision records from the Insurance Corporation of British Columbia (ICBC) averaged 11,000 animal-related crashes, with ~700 human injuries and 3 fatalities (ICBC 2018). In addition to the safety threat resulting in $41 million in ICBC insurance claims, the annual value of wildlife-specific mortality is estimated at $466 million (Sielecki 2010). Spatially-accurate, comprehensive WVC records are critical to limit threats to motorists (Huijser et al. 2007) and to provide valuable insights about spatial and temporal WVC patterns useful to implement specific WVC mitigation measures. The British Columbia Ministry of Transportation and Infrastructure (BC MOTI) uses the Wildlife Accident Reporting System (WARS) to document when and where wildlife carcasses occur and are collected throughout the province. Highway maintenance contractors are required to remove carcasses from numbered highways and submit a monthly, paper-based report to BC MOTI through WARS (Sielecki 2010). Maintenance contractors use date-of- carcass retrieval and the Landmark MOOSE AND DEER-VEHICLE COLLISIONS – SAMPLE ET AL. ALCES VOL. 56, 2020 50 Kilometre Inventory (LKI) (BC MOTI 2018b) to record the removal locations. At the time of our study, the carcass reporting methodology in WARS had remained largely unchanged since its inception in 1978. Several weaknesses have been identified previously including imprecise collision locations and incorrect and incomplete reporting of wildlife species (Sielecki 2010). To determine how WARS data col- lected under the current system might dif- fer from a GPS-based electronic record-keeping system, we partnered with PerSen Technologies Inc. (PERSENTECH) in Winnipeg, Manitoba to develop the Otto® Wildlife GPS unit (Fig. 1; Hesse et al. 2010). This device was specifically designed to capture real-time GPS coordi- nates, and the time and date of sightings of dead and live moose (Alces alces) and deer (Odocoileus spp.; mule, white-tailed, and black-tailed deer all referred to as “deer”), which together comprised 83.1% of WVCs in 2003–2007 (Sielecki 2010). Units were designed for dash-mounting with push button controls to enter data (including sound replay) without requiring the opera- tor of the device to stop their vehicle. We partnered with the Northern Health Authority and Diversified Transport of Prince George and installed Otto® Wildlife devices on the dashboards of Northern Health Connections buses for drivers to collect data on deer and moose throughout the province. To determine the usefulness of GPS technology in WVC mitigation planning, we compared the similarity of moose and deer WVC data collected with WARS and Otto Wildlife on selected highways. Our null hypothesis was that both methods produced similar temporal and spatial patterns for moose and deer WVCs. STUDY AREA Our study area was located in British Columbia, Canada, extending from Abbotsford in the south to Fort St. John in the north (Hwys 1 and 97) and from Prince Rupert on the Pacific west coast to Valemount (Hwys 5 and 16) in east-central British Columbia near the Alberta border, a total of 2,798 km of highway. It is recog- nized that variable widths in right-of-ways likely influenced detection rates at certain locations on all highways. The north and east sections of the study area are character- ized by rugged, mountainous terrain with deeply incised valleys (Child 1992), with terrain to the south and west flat to rolling with hundreds of small lakes and wetlands (Heard et al. 1997). Although mostly an homogeneous unit on a drumlinized till pla- teau surrounding periglacial lake deposits, it is dissected by many rivers, lakes and wetlands (Child 1992) and divided by the Rocky Mountains in the north and east. The landscape is dominated by coniferous Fig. 1. The Otto® Wildlife GPS device used for capturing location data for live and dead deer/ moose. The unit is powered by two AA batteries or can be plugged into the vehicle’s accessory receptacle. ALCES VOL. 56, 2020 MOOSE AND DEER-VEHICLE COLLISIONS – SAMPLE ET AL. 51 forests of hybrid white spruce (Picea engel- mannii x glauca) and subalpine fir (Abies lasiocarpa). Lodgepole pine (Pinus con- torta var. latifolia) and trembling aspen (Populus tremuloides) pioneer secondary successional sites (Meidinger and Pojar 1991), as do many species of willows (Salix spp.) and other woody browse plants used by moose and deer. Moose densities in the core of our study area were estimated at 0.63–0.78 moose/km2 (Cadsand et al. 2013); deer densities were unknown. METHODS Otto® Wildlife System The Otto® Wildlife device provided GPS locations of carcasses and additional information about the animal. To record a live sighting, the appropriate species but- ton (Fig. 1) was pressed to activate a coloured LED and a vocal playback of “deer” or “moose” to confirm that the cor- rect species was recorded by the operator. To catalog a carcass, the “dead button” was pressed immediately after the species button. Pushing the “dead button” 3 times allowed the driver to indicate that a record was in error. Latitude, longitude, time of day, and date were recorded when any of three buttons designed to collect data were pushed. Previous to our study, and to verify that the Otto® Wildlife devices were recording accurate locations of carcasses and live sighting points of interest (POI), Hesse et al. (2010) compared Otto® Wildlife POI loca- tions to existing government GIS layers and found that only 1.5% of Otto® Wildlife loca- tion data were ≥10 m from the Digital Road Atlas (DRA) layer (https://www2.gov.bc.ca/ gov/content/data/geographic-data-services/ topographic-data/roads). This study also confirmed (via exit interviews), that Otto® Wildlife did not pose a safety concern to vehicle operators and that the units were per- forming as per their intended design. The University of Northern British Columbia partnered with the Northern Health Authority to dash-mount Otto® Wildlife devices in 10 Northern Health Connections buses. Data were subsequently collected by bus drivers for live and dead moose/deer and reported to bus dispatchers from 10 June 2010 to 15 July 2014. We col- lected data from dispatchers every 3–6 months and converted Otto® Wildlife and WARS data from the same time period to KMZ (Keyhole Markup Language Zipped) map files to compare moose and deer carcass data from the two methods. Because the WARS data only contained carcass records, we made no comparison of live sightings. An inevitable source of error associated with this method of comparison was that bus drivers operated on set schedules, driving certain sections of highway once daily, while maintenance contractors patrolled highways and responded to WVC reports 24 h/d, 7 days a week. Therefore, from the outset of the study, we acknowledged maintenance contractors patrolling routes driven by bus- ses would record more carcasses than bus drivers. Carcass data were sorted and organized using Microsoft Excel. Records within 500 m of each other were identified using ArcGIS in a BC Albers 1983 coordinate system and Quantum GIS Desktop distance matrix tool, with a linear output matrix type (set within the QGIS Distance Matrix Tool; QGIS 2.4 Development Team 2014) and were deemed to be possibly referencing the same carcass (Hyrcha and Rea, unpublished). All carcass records within 500 m of each other were fur- ther evaluated for duplication using the fol- lowing criteria: carcasses recorded <2 min apart of each other by the same driver on the same date were considered individual sight- ings of distinct animals, and both records https://www2.gov.bc.ca/gov/content/data/geographic-data-services/topographic-data/roads https://www2.gov.bc.ca/gov/content/data/geographic-data-services/topographic-data/roads https://www2.gov.bc.ca/gov/content/data/geographic-data-services/topographic-data/roads MOOSE AND DEER-VEHICLE COLLISIONS – SAMPLE ET AL. ALCES VOL. 56, 2020 52 were retained. It was considered highly prob- able that records of the same species at the same location occurring between 2 min and 24 h of each other were recorded by different busses; therefore, these records were classi- fied as the same carcass (recorded twice) and one record was discarded. We accounted for location and time to determine whether same-species records occurring 1–7 days apart were duplicate sightings. For example, two records in close proximity on a stretch of highway with a very low WVC rate were deemed likely as duplicates. A spatial map of carcass locations created with Google Fusion Tables identified concentrations or hotspots of MVCs (Fig. 2). WARS We obtained WARS data for the same period and along the same highways where Otto® Wildlife devices were used. These data were not georeferenced, but provided written records of carcass locations using a series of established landmarks along high- ways. Therefore, these locations had to be georeferenced in Google Earth using the Landmark Kilometer Inventory (LKI; BC MOTI 2018b). We created KMZ mapping files for both WARS and Otto® Wildlife data and overlaid them for comparison. We then visually inspected and analyzed the area around each Otto® Wildlife and WARS record to identify matches. Comparison of Otto® Wildlife and WARS We first considered the spatial proximity of the WARS and Otto® Wildlife records. Because the majority of LKI landmarks occurred within 1 km (BC MOTI 2018b), our first sorting criteria was a maximum sep- aration of 1 km. However, because it was possible that the WARS spatial data included errors > 1 km, we evaluated matching of the datasets using maximum possible separation distances of 1, 3, and 5 km (S1, S3, S5). Separation in recording dates between WARS and Otto® Wildlife records was also used to identify matched records. For each of the 3 spatial sortings, we allowed maxi- mum temporal separations of 1, 3, and 5 days (T1, T3, T5) between matching records from the two databases. We combined the S and T values to simultaneously specify a record’s spatial and temporal sorting. For example, a spatial and temporal range of 3 km and 5 days had the sorting criterion S3T5. Comparative analyses were per- formed separately for the 9 different sorting criteria (S1T1–S5T5). Moose and deer car- cass records from the Otto® Wildlife devices were also categorized by 4 seasons of 3-month increments: spring (March, April, May), summer (June, July, August), fall (September, October, November), and win- ter (December, January, February). After comparing the Otto® Wildlife and WARS data to identify matching records, the data were sorted by time of year, latitude, highway, and separation of carcass location and time. To supplement the numerical data, geographic illustra- tions of carcass and live sighting locations (for Otto® Wildlife only) were constructed using the online application “MapMaker” (MapMaker.com 2018). RESULTS A total of 167 moose and 410 deer car- casses were recorded using Otto® Wildlife devices, and 1,231 moose and 5,698 deer carcasses with WARS. With the most relaxed sorting criterion of up to 5 km and a 5-day separation in reporting (S5T5), 20% and 27% of Otto® Wildlife moose and deer were classified as having a match with WARS data; with the strictest criterion (S1T1) these matches were 15% and 10%, respectively. http://MapMaker.com ALCES VOL. 56, 2020 MOOSE AND DEER-VEHICLE COLLISIONS – SAMPLE ET AL. 53 The averages across all sorting criteria were 16% and 21% of moose and deer records, respectively (Table 1). Moose carcasses were most commonly recorded during winter (n = 48) and fall (n = 45) with fewer in summer and spring. Fig. 2. A mapped sample of data from the middle of the study area (north-central BC) using S5T5 (spatial separation and temporal separation of 5 km and 5 days) Otto® Wildlife deer carcass records, with circles indicating Otto® Wildlife records with a corresponding WARS record (matching), and triangles indicating unique, unmatching Otto® Wildlife records (MapMaker.com, 2018). http://MapMaker.com MOOSE AND DEER-VEHICLE COLLISIONS – SAMPLE ET AL. ALCES VOL. 56, 2020 54 The number of Otto® Wildlife moose car- casses with a matching WARS record varied by season. The average matching rate was much higher in winter (26%) and fall (19%) than in spring (3%) and summer (7%) (Table 2). Most deer carcasses were recorded in fall (n = 131), winter and spring had similar levels of reporting, and summer had the low- est level (n = 82) (Table 3). As with moose carcasses, the rate of matching varied by sea- son with highest matching in spring (26%) and lowest in winter (15%) (Table 2). Matching moose and deer Otto® Wildlife carcasses and proportions were summarized based on the highway traveled (Tables 4 and 5). Carcasses of both species were most frequently recorded along Highway 97 (103 moose, 296 deer) and Highway 16 Table 1. The number of matched/unmatched records of deer and moose carcasses by sorting criteria as collected with the Otto® Wildlife and WARS systems in 2010–2014, British Columbia, Canada. The sorting criteria indicate spatial (S; 1, 3, 5 km) and temporal separations (T; 1, 3, 5 d) used to determine matches. Sorting criteria Matched # Otto® deer Unmatched # Otto® deer Matched # Otto® moose Unmatched # Otto® moose Matching (%) with WARS (deer, moose) S5T5 110 300 33 134 27, 20 S5T3 95 315 31 136 23, 19 S5T1 87 323 25 142 22, 15 S3T5 104 306 32 135 25, 19 S3T3 92 318 30 137 22, 18 S3T1 83 327 25 142 20, 15 S1T5 68 342 22 145 17, 13 S1T3 65 345 20 147 16, 12 S1T1 60 350 16 151 15, 10 Ave. 21, 16 Table 2. The number of matched/unmatched records of moose carcasses by sorting criteria and season as collected with the Otto® Wildlife and WARS systems in 2010–2014, British Columbia, Canada. The sorting criteria indicate spatial (S; 1, 3, 5 km) and temporal separations (T; 1, 3, 5 d) used to determine matches. Values are shown separately for records with and without matching WARS records. Sorting criteria Matched (n = 16–33) Unmatched (n = 134–151) Spring Summer Fall Winter Spring Summer Fall Winter S5T5 2 4 16 11 31 37 29 37 S5T3 2 4 14 11 31 37 31 37 S5T1 1 2 13 9 32 39 32 39 S3T5 2 4 15 11 31 37 30 37 S3T3 2 4 13 11 31 37 32 37 S3T1 1 2 13 9 32 39 32 39 S1T5 1 4 10 7 32 37 35 41 S1T3 1 4 8 7 32 37 37 41 S1T1 1 2 8 5 32 39 47 43 Ave. 3 7 26 19 ALCES VOL. 56, 2020 MOOSE AND DEER-VEHICLE COLLISIONS – SAMPLE ET AL. 55 (55 moose, 97 deer); lower numbers were recorded on Highways 2, 5, and 27. The pro- portion of matching Otto® Wildlife records were similar on Highways 16 and 97, with average matching rates of 13% and 17% for moose (Table 4), and 18% and 21% for deer (Table 5). Certain areas along Highway 97, such as between Quesnel and Williams Lake had a large proportion of matched deer carcasses; conversely, other areas had low matching rates. A low proportion of Otto® Wildlife and WARS records were matched on Highway 16 (Fig. 3), where – between Table 3. The number of matched/unmatched records of deer carcasses by sorting criteria and season as collected with the Otto® Wildlife and WARS systems in 2010–2014, British Columbia, Canada. The sorting criteria indicate spatial (S; 1, 3, 5 km) and temporal separations (T; 1, 3, 5 d) used to determine matches. Values are shown separately for records with and without matching WARS records. Sorting criteria Matched (n = 60–110) Unmatched (n = 300–350) Spring Summer Fall Winter Spring Summer Fall Winter S5T5 34 22 38 16 69 60 93 78 S5T3 28 20 31 16 75 62 100 78 S5T1 27 16 29 15 76 66 102 79 S3T5 32 21 35 16 71 61 96 78 S3T3 28 19 29 16 75 63 102 78 S3T1 26 15 27 15 77 67 104 79 S1T5 25 13 18 12 78 69 113 82 S1T3 22 13 18 12 81 69 113 82 S1T1 21 11 17 11 82 71 114 83 Ave. 26 21 21 15 Table 4. The number of matched/unmatched records of moose carcasses by sorting criteria and highway (H) as collected with the Otto® Wildlife and WARS systems in 2010–2014, British Columbia, Canada. The sorting criteria indicate spatial (S; 1, 3, 5 km) and temporal separations (T; 1, 3, 5 d) used to determine matches. Values are shown separately for records with and without matching WARS records. Sorting criteria Matched (n = 16–33) Unmatched (n = 134–151) H 2 H 5 H 27 H 16 H 97 H 2 H 5 H 27 H 16 H 97 S5T5 1 1 0 8 23 1 5 1 47 80 S5T3 1 1 0 8 21 1 5 1 47 82 S5T1 1 1 0 7 16 1 5 1 48 87 S3T5 1 1 0 8 22 1 5 1 47 81 S3T3 1 1 0 8 20 1 5 1 47 83 S3T1 1 1 0 7 16 1 5 1 48 87 S1T5 0 1 0 6 15 2 5 1 49 88 S1T3 0 1 0 6 13 2 5 1 49 90 S1T1 0 1 0 5 10 2 5 1 50 93 Ave. 50 17 - 13 17 MOOSE AND DEER-VEHICLE COLLISIONS – SAMPLE ET AL. ALCES VOL. 56, 2020 56 Topley and Vanderhoof – there were 37 unmatched records and none of deer. Similarly, there were no matching records of 10 deer carcasses on Highway 97 between Cache Creek and Hope. Multiple stretches of highways with clustered deer carcasses were identified: Highway 16 from Sinkut Falls Road to Vanderhoof (0.84 deer/ km), Highway 97 between Chetwynd and Taylor (0.83 deer/km), and Highway 97 between Quesnel and Clinton (0.46 deer/ km) in the northern most parts of the province. The average matching rate of Otto® Wildlife and WARS records for both Table 5. The number of matched/unmatched records of deer carcasses by sorting criteria and highway (H) as collected with the Otto® Wildlife and WARS systems in 2010–2014, British Columbia, Canada. The sorting criteria indicate spatial (S; 1, 3, 5 km) and temporal separations (T; 1, 3, 5 d) used to determine matches. Values are shown separately for records with and without matching WARS records. Highway 27 had only one recorded carcass. Sorting criteria Matched (n = 60–110) Unmatched (n = 300–350) H 5 H 16 H 27 H 97 H 5 H 16 H 27 H 97 S5T5 7 19 1 83 9 78 0 213 S5T3 7 18 1 69 9 79 0 237 S5T1 7 17 1 62 9 80 0 244 S3T5 6 18 1 79 10 79 0 217 S3T3 6 17 1 68 10 80 0 220 S3T1 6 15 1 56 10 82 0 232 S1T5 5 16 1 46 11 81 0 250 S1T3 5 15 1 44 11 82 0 252 S1T1 5 15 1 39 11 82 0 257 Ave. 38 18 - 21 Fig. 3. An example from north-central BC of a particularly high density of unmatching Otto® Wildlife deer carcass records is found on Highway 16 between Topley and Vanderhoof. Triangles and circles indicate unmatching and matching records, respectively (MapMaker.com, 2018). http://MapMaker.com ALCES VOL. 56, 2020 MOOSE AND DEER-VEHICLE COLLISIONS – SAMPLE ET AL. 57 species ranged from 13 to 50% (Tables 4 and 5). DISCUSSION Our data indicate that the majority of Otto® Wildlife moose (80–90%) and deer carcass records (73–85%) had no matching WARS record, with the relative level of dis- crepancy dependent on our sorting criteria (days and km separating carcass records). The greater than expected proportion of unmatched records can be partially explained by car- casses removed by other agencies (e.g., Ministry of Environment, RCMP), the public (such as trappers for baiting traps), or scaven- gers. Furthermore, carcasses concealed by snow plowing before the animal was recorded by a maintenance contractor would influence both availability and matching frequency. Conversely, the proportion of WARS records with a matching Otto® Wildlife record was <3% for both moose and deer; albeit, a low percentage was not unexpected since buses travel stretches of highway only once or less daily, whereas maintenance contractors patrol the same roads several times daily, every day of the week. If maintenance contractors use the LKI as intended to record carcass locations in WARS, then S1 should be a sufficient crite- rion, allowing for a spatial recording error of ~1 km on either side of the carcass. Depending on the highway classification, a carcass reported to, or detected by highway maintenance contractors must be removed as soon as possible or within 3 days (Hesse and Rea 2016). Almost all Otto® Wildlife records collected for this study were on primary highways (BC MOTI 2018a), so presumably most carcasses would be removed quickly. Therefore, a 5-day separation (T5) between matching records is a generous matching cri- terion, and a 3-day separation (T3) should be sufficient assuming protocols are followed. Nevertheless, only 20% of moose and 27% of deer carcasses in Otto® Wildlife were classified as matching with the most relaxed sorting criteria (S5T5). These large discrep- ancies are unexplained and require further study. We found that WVCs with moose and deer are most likely to occur in fall and win- ter, similar to findings by ICBC (Rea 2006, O’Keefe and Rea 2012) and BC MOTI (Sielecki 2010). These seasonal peaks in WVCs are similar to those identified in Alaska (Garrett and Conway 1999) and northern Sweden (Neumann et al. 2012). An unexpected finding was the seasonal relationship in the proportion of matching between the Otto® Wildlife and WARS records. The matching of moose carcasses between the two databases was higher in fall and winter than spring and summer. It is pos- sible that maintenance contractors and bus drivers can more easily distinguish carcasses in fall and winter when the contrast between a dark-bodied moose and the lighter land- scape makes a carcass easier to spot. Most carcass matches for deer occurred in spring with fewest in winter. Beyond the limited sample size, a possible explanation is that carcass reporting by maintenance contrac- tors is less of a priority when crews are pre- occupied with plowing and salting roads. Further, plowed snow could reduce detec- tion and/or bury deer carcasses and lower the probability of matching. The probability of an Otto® Wildlife car- cass record having a corresponding WARS record was influenced by region as certain areas had very low rates of matching. Highways in British Columbia are main- tained by contractors in 28 service areas throughout the province (BC MOTI 2019), which suggests varied efficiency among maintenance contractors at reporting and removal of carcasses, or that predators or some other agent (e.g., conservation offi- cers, trappers, or the general public; Hesse MOOSE AND DEER-VEHICLE COLLISIONS – SAMPLE ET AL. ALCES VOL. 56, 2020 58 and Rea 2016) may remove carcasses more frequently. Furthermore, the variation in width of road shoulders and highway right- of-ways, ditch depth, and areas which require brush-cutting may also contribute to differences in detection and matching rates. The GPS-based, Otto® Wildlife system records sightings of live moose and deer that can be mapped and displayed visually (Fig. 4). Unfortunately, we had no basis to compare between the systems because WARS does not provide these data, but encourage mapping of similar data for miti- gation planning. Even though clusters of live animal locations may not necessarily corre- spond to locations of potential high WVC risk per se (Neumann et al. 2012), these data are useful to road safety planners to Fig. 4. Map of study area in BC showing live moose sightings recorded by Northern Health Authority bus drivers during the study period (MapMaker.com, 2018). http://MapMaker.com ALCES VOL. 56, 2020 MOOSE AND DEER-VEHICLE COLLISIONS – SAMPLE ET AL. 59 determine what engineering or environmen- tal factors might explain differences between highway segments with and without WVCs. Another limitation was the higher prob- ability of observing carcasses with the Otto® Wildlife devices in daylight hours. Northern Health buses operate on set weekly sched- ules, mainly between 0630 and 2100 hr (Northern Health Connections 2018), whereas highway maintenance contractors are required to remove carcasses all hours of the day. For example, if a bus departs in the morning from a northern centre en route to southern British Columbia, drivers are likely to observe and record MVCs from the previ- ous night (most collisions are nocturnal) near the bus’s point of origin; conversely, maintenance contractors would likely have removed carcasses as the bus nears its termi- nus farther south. It would be useful to col- lect Otto® Wildlife records 24 h daily and incorporate daily and random route start- times (Hesse et al. 2010). In some cases, Northern Health bus drivers may have pressed Otto® Wildlife but- tons too early or too late to pinpoint a car- cass location or missed carcasses; both would increase the occurrence of unmatched records. Although enthusiasm was high when this project launched, it is possible that drivers became less keen on spotting and recording carcasses as the novelty of the project diminished, as occurred during an earlier pilot study (Hesse et al. 2010). It should be noted that most maintenance con- tractors are trained to “keep an eye out for carcasses” while bus drivers understandably may not have carcasses as their primary search image. Having bus drivers ride with maintenance contractors (and vice versa) could provide for a standardized car- cass-spotting protocol and reduce possible biases. As discussed by Hesse et al. (2010), several modifications could increase the ease of utility of the Otto® Wildlife device. For instance, bigger buttons with different textures for different species would allow drivers to locate the desired button more quickly, increasing the locational accuracy of records. A button to erase the last key- stroke would allow drivers to quickly and easily correct entry mistakes, and additional buttons to record animal behaviour might provide unique and valuable data (Hesse et al. 2010). The use of a smartphone application like that developed in Alberta (Alberta Ministry of Transportation 2017) would help update the WVC record-keeping system in British Columbia. A smartphone application can be easily designed to utilize GPS services and capture the latitude/longitude of collision locations, eliminating the need to reference carcass locations to established roadside landmarks in the LKI system. Not only would a GPS-based record-keeping system facilitate more accurate locations, it could also provide supplementary data including photos and videos if combined with dash- cam technology. In summary, data from the Otto® Wildlife units provided for a useful compar- ison of WVCs collected with the traditional WARS system. While temporal and spatial patterns between moose and deer-vehicle collisions were mostly similar for the two systems as expected, matching of WVC data was low. Overall and as expected, more WVCs were recorded with WARS, but many unique carcasses were recorded with the Otto® Wildlife system. We recommend the use of GPS-enabled data collection devices by highway maintenance contrac- tors to provide accurate location data for moose and deer WVCs to promote road safety for motorists and wildlife alike. POSTSCRIPT: In 2018, the British Columbia Ministry of Transportation and MOOSE AND DEER-VEHICLE COLLISIONS – SAMPLE ET AL. ALCES VOL. 56, 2020 60 Infrastructure began the staged rollout of a new policy that mandates the use of GPS technol- ogy by all highway maintenance contractors in all Service Areas for the purposes of collecting more accurate WARS data; all carcasses will be identified with a GPS location by 2023. ACKNOWLEDGEMENTS We thank Dr. D. Bowering, the Northern Health Authority the Ministry of Transportation and Infrastructure, and Diversified Transportation for partnering with us and installing the Otto® Wildlife devices in their buses. We especially thank the drivers who eagerly worked with us to collect these data. We would like to thank S. Emmons, J. Svendson, and S. O’Keefe for helping with logistics and managing the data during the project and F. Franczk with PERSENTECH Industries for working with us to design the Otto® Wildlife. We thank L. Sielecki for a review of a previous draft of this manuscript. Lastly, we thank our two reviewers for valu- able recommendations that led to improve- ments on an earlier draft of our manuscript. REFERENCES AlbertA Ministry of trAnsportAtion. 2017. Alberta Wildlife Watch Program. (accessed January 2019). british ColuMbiA Ministry of trAnsp­ ortAtion And infrAstruCture. 2018a. Highway Classification. (accessed March 2019). _____. 2018b. Landmark Kilometre Inventory July 2018 Version (201807). (accessed March 2019). _____. 2019. Highway Maintenance. < h t t p s : / / w w w 2 . g o v . b c . c a / g o v / content/transportation/transportation- i n f r a s t r u c t u r e / c o n t r a c t i n g - t o - t r a n s p o r t a t i o n / h i g h w a y - b r i d g e - maintenance/highway-maintenance.> (accessed August 2019). CAdsAnd, b., d. C. heArd, J. Courtier, A. bAtho, and G. s. WAtts. 2013. Moose density and composition in the southern Omineca Region, winter 2011–2012. Regional Report for the Ministry of Forest, Lands, and Natural Resource Operations. 35p. Child, K. n. 1992. Moose Management in the Omineca Sub-region. In D. F. hAtler, editor. Proceedings of the 1991 Moose Harvest Management Workshop, November 5–7, 1991. Kamloops, British Columbia. Wildlife Branch, British Columbia Environment, Victoria, British Columbia, Canada. GArrett, l. C., and G. A. ConWAy. 1999. Characteristics of moose-vehicle colli- sions in Anchorage, Alaska, 1991–1995. Journal of Safety Research 30: 219–223. doi: 10.1016/S0022-4375(99)00017-1 heArd, d., s. bArry, G. WAtts, and K. Child. 1997. Fertility of female moose (Alces alces) in relation to age and body com- position. Alces 33: 165–176. hesse, G., and r. V. reA. 2016. Quantifying wildlife vehicle underreporting on northern British Columbia highways 2004–2013. Report prepared for the British Columbia Ministry of Transportation and Infrastructure, Northern Region, Prince George, British Columbia, Canada. _____, _____, n. KlAssen, s. eMMons, and d. diCKson. 2010. Evaluating the Potential of the Otto® Wildlife GPS Device to Record Roadside Moose and Deer Locations for Use in Wildlife Vehicle Collision Mitigation Planning. Wildlife Biology in Practice. 6: 1–13. doi: 10.2461/wbp.2010.6.1 http://www.transportation.alberta.ca/Content/docType253/Production/Alberta​Wildlife​WatchProgramPlan.pdf http://www.transportation.alberta.ca/Content/docType253/Production/Alberta​Wildlife​WatchProgramPlan.pdf http://www.transportation.alberta.ca/Content/docType253/Production/Alberta​Wildlife​WatchProgramPlan.pdf http://www.transportation.alberta.ca/Content/docType253/Production/Alberta​Wildlife​WatchProgramPlan.pdf https://www2.gov.bc.ca/gov/content/transportation/transportationinfrastructure/transportation-planning/highway-classification https://www2.gov.bc.ca/gov/content/transportation/transportationinfrastructure/transportation-planning/highway-classification https://www2.gov.bc.ca/gov/content/transportation/transportationinfrastructure/transportation-planning/highway-classification https://www2.gov.bc.ca/gov/content/transportation/transportationinfrastructure/transportation-planning/highway-classification https://www2.gov.bc.ca/gov/content/transportation/transportation-infrastructure/engineering-standards-guidelines/landmark-kilometre-inventory https://www2.gov.bc.ca/gov/content/transportation/transportation-infrastructure/engineering-standards-guidelines/landmark-kilometre-inventory https://www2.gov.bc.ca/gov/content/transportation/transportation-infrastructure/engineering-standards-guidelines/landmark-kilometre-inventory https://www2.gov.bc.ca/gov/content/transportation/transportation-infrastructure/engineering-standards-guidelines/landmark-kilometre-inventory https://www2.gov.bc.ca/gov/content/transportation/transportation-infrastructure/engineering-standards-guidelines/landmark-kilometre-inventory https://www2.gov.bc.ca/gov/content/transportation/transportation-infrastructure/contracting-to-transportation/highway-bridge-maintenance/highway-maintenance https://www2.gov.bc.ca/gov/content/transportation/transportation-infrastructure/contracting-to-transportation/highway-bridge-maintenance/highway-maintenance https://www2.gov.bc.ca/gov/content/transportation/transportation-infrastructure/contracting-to-transportation/highway-bridge-maintenance/highway-maintenance https://www2.gov.bc.ca/gov/content/transportation/transportation-infrastructure/contracting-to-transportation/highway-bridge-maintenance/highway-maintenance https://www2.gov.bc.ca/gov/content/transportation/transportation-infrastructure/contracting-to-transportation/highway-bridge-maintenance/highway-maintenance ALCES VOL. 56, 2020 MOOSE AND DEER-VEHICLE COLLISIONS – SAMPLE ET AL. 61 huiJser, M. p., J. fuller, M. e. WAGner, A. hArdy, and A. p. CleVenGer. 2007. Animal-Vehicle Collision Data Collection: A Synthesis of Highway Practice. National Cooperative Highway Research Program (NCHRP) Synthesis 370. Transportation Research Board, Washington, D. C., USA. ICBC. 2018. Quick Statistics for the Media Manual. (accessed January 2020). MApMAKer.CoM. 2018. MapMaker. (accessed January 2018). MeidinGer, d., and J. poJAr. 1991. Ecosystems of British Columbia. Special Report No. 6. Ministry of Forests Research Branch, Victoria, British Columbia, Canada. neuMAnn, W., G. eriCsson, h. dettKi, n. bunnefield, n. s. Keuler, d. p. helMers and C. rAdeloff. 2012. Difference in spatiotemporal patterns of wildlife road- crossings and wildlife-vehicle collisions. Biological Conservation 145: 70–78. doi: 10.1016/j.biocon.2011.10. 011 northern heAlth ConneCtions. 2018. Northern Health Connections Route Schedule. (accessed January 2019). o’Keefe, s., and r. V. reA. 2012. Evaluating ICBC Animal–Vehicle Crash Statistics (2006–2010) for Purposes of Collision Mitigation in Northern British Columbia. Report for the Insurance Corporation of British Columbia, North Vancouver, Canada. reA, r. V. 2006. Elucidating temporal and species-specific distinctions in patterns of animal-vehicle collisions in various communities and regions of northern British Columbia. In Using Collision Data, GPS Technology and Expert Opinion to Develop Strategic Countermeasures Recommendations for Reducing Animal–Vehicle Collisions in Northern British Columbia. Research Report, Road Health-University Wildlife Collision Mitigation Research Team, Prince George, British Columbia, Canada. sieleCKi, l. e. 2010. WARS 1988–2007: Wildlife Accident Reporting and Mitigation in British Columbia: Special Annual Report. Environmental Management Section, Engineering Branch, British Columbia Ministry of Transportation and Infrastructure, Victoria, British Columbia, Canada. (accessed January 2018). https://www.icbc.com/about-icbc/newsroom/Documents/crashes-involving.pdf https://www.icbc.com/about-icbc/newsroom/Documents/crashes-involving.pdf https://www.icbc.com/about-icbc/newsroom/Documents/crashes-involving.pdf http://MapMaker.com https://www.darrinward.com/lat-long/ https://www.darrinward.com/lat-long/ https://nhconnections.ca/Portals/12/Documents/Bus-Schedule-NH-Connections.pdf https://nhconnections.ca/Portals/12/Documents/Bus-Schedule-NH-Connections.pdf https://nhconnections.ca/Portals/12/Documents/Bus-Schedule-NH-Connections.pdf https://www2.gov.bc.ca/gov/content/transportation/transportation-infrastructure/engineering-standards-guidelines/environmental-management/wildlife-management/wildlife-accident-reporting-system/wars-1988-2007 https://www2.gov.bc.ca/gov/content/transportation/transportation-infrastructure/engineering-standards-guidelines/environmental-management/wildlife-management/wildlife-accident-reporting-system/wars-1988-2007 https://www2.gov.bc.ca/gov/content/transportation/transportation-infrastructure/engineering-standards-guidelines/environmental-management/wildlife-management/wildlife-accident-reporting-system/wars-1988-2007 https://www2.gov.bc.ca/gov/content/transportation/transportation-infrastructure/engineering-standards-guidelines/environmental-management/wildlife-management/wildlife-accident-reporting-system/wars-1988-2007 https://www2.gov.bc.ca/gov/content/transportation/transportation-infrastructure/engineering-standards-guidelines/environmental-management/wildlife-management/wildlife-accident-reporting-system/wars-1988-2007 https://www2.gov.bc.ca/gov/content/transportation/transportation-infrastructure/engineering-standards-guidelines/environmental-management/wildlife-management/wildlife-accident-reporting-system/wars-1988-2007