Identifying migration corridors of mule deer threatened by highway development

Abstract

Highways are hazardous to migratory ungulates world‐wide, causing direct and indirect impacts to ungulate survival. Moreover, significant financial costs are incurred in damage from wildlife–vehicle collisions and in building and maintaining wildlife passage structures. Information is needed to link ungulate movements to collision occurrence to prioritize needed construction of wildlife crossings on highways. We simultaneously documented mule deer (Odocoileus hemionus) migration corridors and mule deer–vehicle collisions (DVCs) in South‐central Oregon, USA, over 6 years (2005–2011). We calculated Brownian Bridge Movement Models for 359 migrating mule deer equipped with Global Positioning System technology. We modeled DVC counts as functions of probability of use during migration, annual average daily traffic (AADT), and habitat characteristics. Probability of use during migration was the strongest predictor of where DVCs occurred (r = 0.93). Predicted DVCs also increased with AADT but peaked at approximately 8,000 and then decreased. Where AADT was above approximately 8,000, fewer deer attempted to cross the highway and DVCs decreased because, over time, deer either abandoned the migration route or were killed trying to cross this busy highway. Our results suggest that managers should focus on migration corridors or high‐density DVC locations to identify where fencing and under/overpasses could be most effective for maintaining migratory corridors when confronting increasing traffic and development that bisect seasonal ranges of mule deer. © 2015 The Wildlife Society.

Multiscale models of habitat use by mule deer in winter

“…We fitted adult deer with GPS collars programmed to record a location every 4 hours and self‐release after 52–72 weeks (Coe et al ). We assessed winter habitat use of mule deer using locations obtained between 1 January–31 March because this was the time period when deer were non‐migratory and concentrated on winter ranges.…”

“…Wildlife veterinarians from ODFW approved and supervised capture operations, which followed the Guidelines of the American Society of Mammalogists for the use of wild mammals in research (Sikes and Gannon 2011). We fitted adult deer with GPS collars programmed to record a location every 4 hours and self-release after 52-72 weeks (Coe et al 2015). We assessed winter habitat use of mule deer using locations obtained between 1 January-31 March because this was the time period when deer were nonmigratory and concentrated on winter ranges.…”

Effects of distribution, behavior, and climate on mule deer survival

“…Deer‐vehicle highway collisions are a concern to state agencies nationwide (Romin and Bissonette , Coe et al ). We used a Oregon state highways GIS layer from Oregon Department of Transportation land use development zones (2009) (http://www.oregon.gov/ODOT/TD/TDATA/pages/gis/odotmaps.aspx, accessed 17 Aug 2014) to calculate the number of highways an individual deer must cross during fall or spring migration (range = 0–3) to reach its seasonal range (average traffic volume = 15,200 Annual Average Daily Traffic of all vehicles for 3 highways in study area from 2005–2012).…”

“…We defined 4 seasons that reflected different stages of the mule deer life cycle: summer (1 Jun-30 Aug), fall (1 Sep-30 Nov), winter (1 Dec-28 Feb), and spring (1 Mar-31 May). We also identified spring (Apr and May) and fall (Oct and Nov) migration periods (Coe et al 2015), and the hunting season (Aug-Oct) as potentially important sources of temporal variation in survival. We plotted the GPS data for each individual deer to determine summer and winter ranges.…”

Longest sage‐grouse migratory behavior sustained by intact pathways

“…We assumed that a 95% isopleth represented a conservative boundary encompassing available migratory habitat (i.e., pathways). We collected the majority of sequential locations used in migratory analyses (>97%) at 6-hour intervals; however, when the time between locations exceeded 12 hours (>2 missing locations), we split the dataset and estimated separate BBMMs to aid in convergence while avoiding artificial inflation of Brownian-motion variance (Horne et al 2007, Coe et al 2015. We assumed a location error of 26 m, because 88% of migratory locations included a GPS fix with an estimated accuracy within 26 m, and used pathways to define the extent to which resources were available to migrating sage-grouse.…”