SPOTTED TURTLE USE OF A CULVERT UNDER RELOCATED ROUTE 44 IN CARVER, MASSACHUSETTS

Transcription

1 Introduction SPOTTED TURTLE USE OF A CULVERT UNDER RELOCATED ROUTE 44 IN CARVER, MASSACHUSETTS Delia R. J. Kaye (Phone: , dkaye@vhb.com), Senior Environmental Specialist, Vanasse Hangen Brustlin, Inc. Box 9151, 101 Walnut Street, Watertown, Massachusetts 02471; Kevin M. Walsh, Massachusetts Highway Department; Eric L. Rulison, Hofstra University; and Christopher C. Ross, Massachusetts Department of Environmental Protection Abstract: A new highway alignment for relocated Route 44 in Carver, Massachusetts, resulted in the direct alteration of 2.5 acres and indirect alteration of 3.9 acres of habitat for three statelisted turtle species: the wood turtle (Clemmys insculpta), spotted turtle (Clemmys guttata), and eastern box turtle (Terrapene c. carolina). As part of the mitigation requirements for impacts to rare species habitat, the Massachusetts Highway Department (MassHighway) conducted a twoyear preconstruction study to determine the habitat preferences and seasonal movements of the statelisted species. The study determined that no wood turtles were present in the study area, that there was a large but declining population of box turtles, and that two highly used spotted turtle habitats would be bisected by the proposed highway entrance ramp. An intermittent stream channel proposed to be piped under the new entrance ramp was identified as a primary travel corridor between the two habitats. Based on the findings of the preconstruction study, MassHighway identified a simple solution to allow the stream channel to continue to provide a migratory corridor for spotted turtles. To achieve this goal, MassHighway increased the proposed culvert size from a 24inch pipe to a 6foot by 6foot box culvert. In the spring and summer of 2004, postconstruction monitoring was conducted to determine the effectiveness of the culvert as a spotted turtle crossing structure. Nine turtles were fitted with radio transmitters and thread bobbins and followed three times per week in the spring and early summer, and once per week in the late summer to determine culvert effectiveness. Direct evidence (thread trails, visual observation) was documented for seven turtles, and indirect evidence (radio telemetry points on both sides of the culvert, visual observation) was documented for 13 turtles, confirming the use of the culvert as a crossing structure. A future study is recommended to document potential effects of traffic and noise on the spotted turtle population, continued use of the culvert, and potential changes to rare species habitat from the highway construction. In 2002, the Massachusetts Highway Department (MassHighway) began construction of relocated U.S. Route 44 between Carver and Plymouth. Construction of the new highway entrance ramp for Route 44 in Carver resulted in the direct alteration of 2.5 acres of prime rare species habitat and indirect alteration of 3.9 acres of prime rare species habitat. MassHighway conducted extensive mitigation measures to compensate for unavoidable impacts to rare species habitat, including acquisition of 27.8 acres of prime and suitable rare species habitat (a mitigation ratio of 6.4:1) and commitment to complete a twoyear preconstruction study developed by the Natural Heritage and Endangered Species Program (NHESP) to determine the seasonal movements and habitat preferences for wood turtles, spotted turtles, and eastern box turtles. Field work for the preconstruction study was conducted in 1998 and Results of the study indicated that wood turtle habitat did not occur in the area, that there was a healthy population of spotted turtles in the emergent wetland adjacent to Route 58 (Wetland 17), and that there was a large but declining population of box turtles. The study also concluded that spotted turtles used a stream channel tributary to the Winnetuxet River and adjacent wetlands to travel to Turtle Pond, a vernal pool located approximately 550 feet east of Wetland 17. The results of the study determined that spotted turtle habitat would be bisected by the new Route 44 entrance ramp. All environmental clearances and permits were already acquired for the project, and there was no requirement or obligation to further mitigate impacts to the turtle habitat. However, based on the findings of the study, MassHighway identified a simple solution to not only address the direct impacts to habitat, but to provide a critical wildlife passage link that would preserve the greater habitat area and prevent isolation and eventual elimination of 6.5 acres of prime spotted turtle habitat in Wetland 17. To achieve this goal, MassHighway increased the size of the proposed culvert, from a 24inch reinforced concrete pipe to a 6foot by 6foot concrete box culvert, which would not only convey the stream but provide migratory passage for spotted turtles under the highway interchange ramp. Study Area The main population of spotted turtles is located in an open emergent marsh, Wetland 17, containing soft rush (Juncus effusus), common cattail (Typha latifolia), common reed grass (Phragmites australis), steeplebush (Spiraea tomentosa), and tussock sedge (Carex stricta). Two intermittent streams traverse Wetland 17 and converge approximately 100 feet west of the newly constructed Route 44 entrance ramp to form the main tributary to the Winnetuxet River. The main intermittent stream (north channel) is fed by runoff from Route 58 to the west, and flows into the Winnetuxet River approximately 1,300 feet east. A second intermittent stream (south channel) also collects runoff from Route 58. Both channels are approximately three to five feet wide, with depths ranging from 0 to 2.5 feet, depending on the season. The substrate consists of a deep organic layer. The banks are heavily vegetated with overhanging soft rush and tussock sedge. Chapter ICOET 2005 Proceedings

2 The new Route 44 entrance ramp constructed between Wetland 17 and the forested wetland east of Wetland 17 conveys the tributary to the Winnetuxet River under the entrance ramp via a 6foot by 6foot concrete box culvert. The 60foot long culvert was constructed below the streambed elevation, providing approximately four to six inches of organic substrate on the culvert bottom. In the vicinity of Wetland 17, the entrance ramp is 10 to 15 feet above the wetland, with 2:1 riprap sideslopes. During construction, the culvert was cut to match the 2:1 sideslopes, which shortened the effective culvert width from approximately 68 feet to 44 feet. Based on a culvert length of 44 feet, the openness ratio (OR: cross-sectional area of the culvert divided by its length) is 0.8. Although the culvert was designed and constructed prior to publication of the Draft Massachusetts River and Stream Crossing Standards: Technical Guidelines (University of Massachusetts-Amherst 2004), the culvert exceeds the recommended openness ratio of 0.75 for new crossing structures. Approximately 200 feet south of the culvert, the ramp is approximately three feet above the replacement wetland elevation. The steep, riprapped sideslopes that support the entrance ramp effectively serve as a barrier to turtles, and turtles are directed to the 6foot by 6foot box culvert if they intend to cross. South of the main tributary to the Winnetuxet River, the red maple forested wetland ponds in several areas during the spring, and contains many braided, slowmoving channels that outflow from Turtle Pond and drain into the main tributary. A large portion of this wetland is paludal woodland, exhibiting mound and pool topography, with many sphagnum moss mats in the wetter areas. Red maple dominates the tree layer, with black gum (Nyssa sylvatica) as a common representative, and white pine (Pinus strobus) on hummocks within the wetland. In the lower forest layers, highbush blueberry (Vaccinium corymbosum) and northern arrowwood (Viburnum dentatum) are common. A dense growth of cinnamon fern (Osmunda cinnamomea) covers the forest floor. Turtle Pond is a vernal pool east of Wetland 17 and south of the tributary to the Winnetuxet River. This abandoned cranberry bog is partially vegetated with shrub vegetation, including highbush blueberry, arrowwood, winterberry holly (Ilex verticillata), and cranberry (Vaccinium macrocarpon). The pool also contains swamp loosestrife (Decodon verticillatus), pondlily (Nuphar variegatum), three-way sedge (Dulichium arundinaceum), and large mats of sphagnum moss (Sphagnum sp.). Depths in the pool range from 2 inches to 2.5 feet. Methods Capture and characterization Turtle capture was conducted daily between April 8 and April 29, Turtles were handcaptured by visually searching suitable habitat, and trapped using two 10inch diameter minnowstyle traps constructed of hardware cloth with escape hatches, funnel shaped entrances, and sized for spotted turtles. During the weeks of April 20 and 28, two unbaited hoop traps were set in Wetland 17 to capture two radio tagged turtles whose transmitters appeared to be failing (Turtles 30A and 50A). Captured turtles were aged, sexed, measured, weighed, and notched for individual identification. Each captured turtle was also numbered with a bright orange nontoxic paint pen on its carapace for easier visual observation. Age was determined by counting annuli on each right plastral plate, which is a reasonable estimate of age for many turtle species (Sexton 1959). Turtles were sexed based on eye color, jaw color, vent location, and plastral concavity. Turtles were marked by notching marginal scutes, modified from Cagle (1939). A triangular metal file was used instead of the square file described by Cagle because the triangular file is less intrusive and produced equivalent notches. Ernst et al. (1994) reported studies that found sexual maturity of spotted turtles was attained by the time they grow to a carapace length of 80 millimeters. We attempted to classify males and females at all ages, but classified spotted turtles with a carapace length of less than 80 millimeters as juveniles. Monitoring Thread trailing and radio telemetry were used to monitor the movements of the sample population. Radio telemetry was used to provide snapshots of movement, while thread bobbins were used to show actual movement trails. By using these monitoring techniques, effectiveness of the culvert could be determined. Radiotelemetry Nine turtles were fitted with radio transmitters (AVM model SM 1H; 164 MHz) between April 8, 2004, and May 19, Six radiotagged turtles (3 males; 3 females) that were documented to occur on both sides of the culvert during one or both years of the preconstruction study were a priority for the postconstruction study. These turtles were: 6A, 8, 25, 30A, 50A and 60A. It was theorized that these turtles would attempt to maintain the same home range and movement patterns that they exhibited prior to the highway construction, and that tracking these turtles would yield more useful information on whether and how they crossed than for those turtles that were either known to not cross, or did not have preconstruction data recorded. Three additional turtles (2 females, 1 male) were also fitted with radio transmitters for a total of nine radio tracked turtles. These turtles were: 13, 52 and 550. Radio transmitters were glued to the right rear side of each turtle s carapace with a fastsetting, twopart epoxy. Turtles were released at their point of capture within one hour. Spotted turtles were tracked three times a week through June 30, and once a week from July 1 through September 30. Radio transmitters were removed from all but one turtle during On the Road to Stewardship 427 Wildlife Crossing Structures

3 the first week of October before the turtles entered dormancy (the signal on one turtle was lost before its transmitter could be removed). Each turtle was tracked to within one foot of its location, and point coordinates were recorded with a handheld Global Positioning System (GPS) unit (Garmin GPS II Plus). Coordinates were transferred to an aerial photo base and used to determine turtle home ranges. Five females and 4 males were fitted with radio transmitters and tracked for the field season or until the signal was lost. Turtle 6A was initially fitted with a radio transmitter but did not move from its original capture location for 4 weeks of tracking, so her transmitter was removed and placed onto another female (Turtle 52) that had been observed passing through the culvert. Thread trailing From the date of capture through June 30, the nine radiotagged turtles were also fitted with thread bobbins to assist in locating travel corridors and provide direct evidence of culvert use. Bobbin attachment was modified from methodology employed by Wilson 1994 with bobbins obtained from Coats North America (formerly Barbour Threads). Bobbins were approximately three grams and contained 200 yards of thread. Bobbins were placed in 3 4inch diameter heat-shrinkable tubing (Russell industries, Inc., HUG-34-4PB) then heated so that the tubing encased the thread bobbin. The encased thread bobbin was then epoxied onto the turtle s carapace and edges were smoothed over with caulking to prevent tangling. Bobbins were tied to vegetation and unraveled as turtles moved. Expelled thread was followed each day for three consecutive days to determine individual movement paths. A sketch of the thread trail was drawn, illustrating the movement. Expelled thread was collected daily and removed to minimize the possibility of tangling. Thread bobbins were removed from all turtles during the week of June 30, Data and analyses Based on radio tracking and thread trailing data gathered over the field season, maps of each turtle s home range and movements were generated. Home ranges were determined by the minimum convex polygon method for ArcView 3.x, v. 1.2 (Jenness 2004). Turtles with at least 10 observations over a sixweek period were included in the home range analyses. Because the home range maps contain locationspecific data for a statelisted species, the home range analyses are provided descriptively and not graphically. Results and Discussion Spotted turtles were captured from April 8 to July 7, This section provides the results of the 2004 captures, with comparisons to the preconstruction study, as well as a description of the effectiveness of the culvert as a connectivity link between Wetland 17 and Turtle Pond. Home range estimates are also provided for radio tracked individuals, with comparisons to the preconstruction study. Population dynamics Fifty-six individual spotted turtles were captured during the 2004 study. In 2004, 35 of the 56 turtles (63%) were recaptures, while 21 were new captures (37%). Four turtles captured for the first time in 2004 had been notched by others prior to As with previous years, most turtles (41 individuals, representing 73%) were captured in April. The majority of the captures (46 turtles; 82%) occurred in Wetland 17, followed by seven captures in Turtle Pond (12%), and three turtles in the red maple forested wetland (6%). As with previous years, most turtles (55 individuals; 98%) were captured by hand. Six turtles were caught in traps in Wetland 17, five of which were 2004 recaptures. The population of spotted turtles appears to be healthy. To date, 81 spotted turtles have been captured and marked. Forty spotted turtles were captured and marked during the 1998 field season, 20 turtles were captured and marked in the 1999 field season, and 21 during Of the 81 turtles, 13 were previously captured and marked by others. In 2004, new captures were skewed towards females (9 individuals) and juveniles (8 individuals), with four male captures (table 1). This ratio is similar to 1999 captures. Over the three field seasons, juveniles comprised greater than onethird of the population (31 individuals; 38%), indicating substantial juvenile recruitment into the population. Overall, females were captured at slightly greater ratio than males (1.2:1), although in 1998, males outnumbered females by a 1.3:1 ratio. This trend is not similar to what has been reported in the literature, where several studies have documented a male bias in turtle populations when there is an adjacent major roadway such as Route 58 (Aresco 2005, Steen and Gibbs 2004, Gibbs and Shriver 2002). Some researchers theorize that populations contain higher numbers of males because females are physiologically required to seek open, sandy upland areas for nesting, which may involve higher roadcrossing frequency than males, thereby exposing females to greater vulnerability from vehicle mortality. Chapter ICOET 2005 Proceedings

4 Table 1. Age and Sex Ratio of Spotted Turtle Captures * Juveniles were classified as those individuals with a carapace length less than 80 millimeters In the study area, the slight female bias may be caused by nest site selection. Like many chelonians, sex determination in spotted turtles is temperaturedependent. Temperatures at or above 86 degrees Fahrenheit during egg maturation produce all females (Ernst et al. 1994). Roadways can provide seemingly suitable nesting sites as they often have higher soil temperatures, lack canopy cover, and exhibit higher ambient temperatures because of the heat that pavement absorbs and attracts (Aresco 2005). If nesting is occurring adjacent to Route 58, it is likely that more females than males are being produced. Preliminary assessments of juvenile sex, while unreliable because sexual differentiation is not well established in juveniles, indicates that many more females are present in the study area than males. Of the 31 juvenile captures, 23 were female, 4 were male, and 4 were undetermined. Eight dead spotted turtles were encountered during the three field seasons (five in 1998, two in 1999, and one in 2004). Two adult females (Turtles 43 and 50) and four juveniles were found dead on the Route 58 shoulder (east side), and two dead juveniles were found in Wetland 17. It is likely that the females were either attempting to cross Route 58 to nest in uplands on the west side of the road, or were attempting to nest on the roadway sideslopes. However, the high number of recaptures observed between the postconstruction and preconstruction study appears to indicate that the sexually mature individuals are experiencing relatively low mortality rates. Based on the slight female bias in the population, and the large number of recaptures observed in 2004, Route 58 does not appear to cause additive mortality. The proportional ratio could mean that the Route 58 and 44 are not a large source of additive mortality. Females may also be nesting in close proximity to Wetland 17 and/or Turtle Pond in areas that do not require roadway crossings. Home range analysis In, 2004, radio tracked turtles consisted of five females and four males. Five spotted turtles tracked during both preconstruction seasons (Turtle 5A, 6A, 8, 30A, and 60), and one individual tracked in 1999 (Turtle 25) was radio tracked in Two turtles captured in 1999 but not previously tracked (Turtles 52 and 550) and one 2004 capture (Turtle 13) were tracked in Originally, four females and four males were fitted with radio transmitters. After four weeks of tracking, Turtle 6A s signal was weak, and movement appeared to be minimal based on thread bobbin tracking. Her radio transmitter and thread bobbin were removed, and a new transmitter and thread bobbin was fitted to another female (Turtle 52). Mean home range size for males was 3.1 acres in 2004 and 3.3 acres over the three years (table 2). Female mean home range was slightly smaller (2.6 acres in 2004 and 2.1 acres between years). Male home ranges varied from 1.4 to 4.9 acres in 2004, and female home ranges varied from 1.7 to 3.6 acres. Our data are consistent with Graham s (1995) result of 1.98 acres. His average, however, may be low because only three individuals were tracked. In western Massachusetts, Milam and Melvin (2001) found home ranges varied from 0.5 acres to 85 acres, with a mean home range of 8.9 acres in a study involving 26 individuals. Larger home ranges were attributed to the longer tracking period of study and inclusion of all data points in the analyses. The smaller home ranges observed in this study may be attributed to higher quality habitat in a smaller area. Table 2: Spotted Turtle Home Range Sizes On the Road to Stewardship 429 Wildlife Crossing Structures

5 Home ranges for five turtles were analyzed during the pre and postconstruction study. Most turtles showed large yeartoyear variations, with no consistent differences between pre and postconstruction. Turtle home ranges vary year to year, based on numerous factors such as climate and food resources. Aquatic turtles such as spotted turtles are often found in welldefined populations (Gibbons 1968), and this was also observed for spotted turtles in the study area. Spotted turtles showed a great deal of overlap in their home ranges with other radioed turtles, as well as with other captured turtles. In 2004, six of the nine radioed turtles spent time in both Wetland 17 and Turtle Pond. All turtles, except Turtle 13, spent a portion of their time in Wetland 17. All turtles, except Turtle 6A and Turtle 60, spent a portion of their time in Turtle Pond. Culvert effectiveness Past studies have indicated that spotted turtles appear to maintain their corridors and movement patterns between years (Perillo 1997, Klemens 2000). In order to determine if the installation of a 6foot by 6foot culvert was an effective tool to maintain habitat connectivity used by this population of turtles, radio transmitters and thread trailing devices were attached to nine turtles. Direct Evidence. Direct evidence such as thread trail or visual observation of a turtle moving through the culvert was used to confirm use of the 6foot by 6foot box culvert. Seven of the nine tracked individuals had been followed either in 1998 or in On average, bobbins were attached for nine weeks, collecting 27 days of movement throughout the late spring, when movement is typically at its highest. Turtle 30A had its bobbin attached for the longest time (11 weeks), while Turtle 52 had its bobbin attached for the shortest time (3 weeks). Thread trailing was used to show actual movements by the turtles, compared to telemetry, which provided movement snapshots. The benefit of thread trailing is that it allows researchers to directly observe individuals movement patterns. A list of the turtles and dates of tracking is provided in table 3. Table 3. Spotted Turtles Tread Trailing Dates Bobbins were only functioning 3 consecutive days per week Thread trailing provided direct evidence that the culvert was effective as a crossing structure and that the construction of the Route 44 access ramp will not negatively impact the ability of the turtle population to access Wetland 17, Turtle Pond, and other habitats. Seven individuals (3 male, 3 female, 1 unknown) used the culvert seven times (table 4). A thread trail was observed five times indicating movement; two turtles that were not being tracked were visually observed moving through the culvert, and one unidentified individual (eluded capture) was visually observed moving through the culvert. Four of the nine turtles with bobbins used the culvert. Turtle 30A provided the most direct evidence of use, leaving a thread trail on two occasions. Females appeared to move in relation to nesting (mid-june), while males movement appeared to correlate to mating (May). Turtles 7, 54, and 70A were all observed at the culvert entrances multiple times but never observed traveling through or on both sides of the culvert. The culvert may provide additional usefulness other than maintaining connectivity. The culvert has 4 to 6 inches of organic substrate that may be used in thermoregulation as well as the shade it provides. The culvert may also provide foraging opportunities because of the different environment it provides for food resources. Chapter ICOET 2005 Proceedings

6 Table 4. Spotted Turtles Culvert Use (Direct Evidence) Indirect Evidence. Thirteen individuals were observed on both sides of the culvert during the field season. This determination was made using radio telemetry and from direct observation. It is likely that the turtles used the culvert because other pathways are improbable. The highway entrance ramp is elevated 10 to 15 feet above the wetland with steep, riprap slopes. The bottom three feet to five feet of the ramp sideslopes consist of twofoot to threefoot boulders, which would make it extraordinarily difficult for turtles to climb. Individuals would have to travel several hundred feet to the south with little cover to access a flatter portion of the ramp that could be more easily ascended. Thirteen individuals (8 male, 5 female) were observed on both sides of the culvert. Seven of the 13 individuals had transmitters. The culvert may have been used a total of 31 times by these 13 individuals (table 5). Overall, between direct and indirect observations, 14 different individuals possibly used the culvert for a total of at least 39 times. Table 5. Spotted Turtles Culvert Use (Indirect Evidence) On the Road to Stewardship 431 Wildlife Crossing Structures

7 Biographical Sketches: Delia R. J. Kaye, CWB, senior environmental scientist, Vanasse Hangen Brustlin, Inc. Ms. Kaye received a B.S. degree in wildlife biology from the University of Vermont and is pursuing a master s degree in biology from Harvard University Extension School. Her current work includes rare species habitat evaluations, wildlife studies, and wetland and stream restoration. Ms. Kaye served as field supervisor for the Route 44 turtle study. Kevin M. Walsh, assistant deputy chief engineer, Environmental Section, Massachusetts Highway Department. Mr. Walsh has over 16 years of experience in environmental permitting of highway projects. As the environmental project manager for the Route 44 Relocation Project, he oversaw all aspects of environmental permitting and analysis for the project including coordination with the State, Federal, and local environmental agencies. He holds a bachelor s degree in environmental design. Eric L Rulison, graduate student, Hofstra University. Mr. Rulison completed a B.S. degree in wildlife management at the University of New Hampshire and is pursuing a master s of science degree in biology from Hofstra University, where his research interests focus on effects of nuisance mammals on native rare species. While at VHB, he conducted much of the field work for the Route 44 turtle study as well as other studies involving rare reptiles and amphibians. Christopher M. Ross, environmental analyst, Department of Environmental Protection. Mr. Ross is a graduate of Northeastern University with a bachelor s degree in civil engineering. Mr. Ross was the regional lead engineer for the permitting of the Route 44 project. He has over 17 years experience with DEP in the permitting and review of complex engineering issues associated with permits issued under the Wetlands Protection Act. Mr. Ross is also the regional liaison to MassHighway on interagency issues. References Aresco. M.J The effect of sex-specific terrestrial movements and roads on the sex ratio of freshwater turtles. Biological Conservation. 123: Belzer, W.B. and D. A. Reese Radio transmitter attachment for turtle telemetry. Herpetological Review. 26(4): Cagle, F.R A system of marking turtles for further identification. Copeia. 1939(3): Ernst, C.H Biological notes on the bog turtle, Clemmys muhlenbergii. Herpetologica. 33: Ernst, C.H., J.E. Lovich, and R.W. Barbour Turtles of the United States and Canada. Smithsonian Institution Press, Washington D.C. Forman, R.T.T. and L.E. Alexander Roads and their major ecological effects. Annual Review of Ecological Systematics. 29: Gibbs, J.P. and W.G. Shriver Estimating the effects of road mortality on turtle populations. Conservation Biology. 16(6): Gibbons, J. W Population structure and survivorship in the painted turtle, Chrysemys picta. Copeia. 1968: Graham, T.E Habitat use and population parameters of the spotted turtle, Clemmys guttata, a Species of Special Concern in Massachusetts. Chelonian Conservation and Biology. 1(3): Heyer, R.W., M.A. Donnelly, R.W. McDiarmid, L.C.Hayek, and M.S. Foster (eds.) Measuring and monitoring biological diversity: Standard methods for amphibians. Smithsonian Institution Press, Washington. Jenness, J Convex hulls around points (conv_hulls_pts.avx) extension for ArcView 3.x, v Jenness Enterprises. Available at: Klemens, M.W. (ed) Turtle Conservation. Smithsonian Institution Press, Washington. Milam, J.C. and S. M. Melvin Density, habitat use, movements, and conservation of spotted turtles (Clemmys guttata) in Massachusetts. Journal of Herpetology. 35(3): Perillo, K Seasonal movements and habitat preferences of spotted turtles (Clemmys guttata) in north central Connecticut. Chelonian Conservation and Biology. 2(3): Sexton, O.J A method of estimating the age of painted turtles for use in demographic studies. Ecology. 40(4): Steen, D.A. and J.P. Gibbs Effects of roads on the structure of freshwater turtle populations. Conservation Biology. 18(4): University of Massachusetts Amherst Massachusetts River and Stream Crossing Standards: Technical Guidelines. Downloaded from: guidelines_river_stream_crossings.pdf. Wilson, D.S Tracking small animals with thread bobbins. Herpetological Review, 25(1): Chapter ICOET 2005 Proceedings

8 Introduction USE OF HIGHWAY UNDERPASSES BY LARGE MAMMALS AND OTHER WILDLIFE IN VIRGINIA AND FACTORS INFLUENCING THEIR EFFECTIVENESS Bridget M. Donaldson (Phone: , Virginia Transportation Research Council, 530 Edgemont Road, Charlottesville, VA Abstract: The rapid increase in animal-vehicle collisions on U.S. roadways is a growing concern in terms of human safety, property damage and injury costs, and viability of wildlife populations. Wildlife crossing structures are gaining national recognition by transportation agencies as effective measures to reduce animal-vehicle collisions and connect wildlife habitats across transportation corridors. In Virginia, white-tailed deer and black bear pose the highest risk. This one-year study was conducted to monitor various underpass structures in Virginia to determine the structural and location attributes that make a crossing successful in terms of use by large mammals. The underpasses, most of which were not specifically designed as wildlife crossings, consist of box culverts and bridges of varying sizes. Remote cameras installed at seven underpass sites in Virginia have recorded more than 2,700 wildlife photographs and documented 1,107 white-tailed deer crossings in the most heavily used structures. Underpasses with a minimum height of 12 ft were successful at facilitating deer passage. Such structures were also heavily used by a variety of wildlife species, including coyote, red fox, raccoon, groundhog, and opossum. Structures with drainages that mimic natural waterways can encourage use by a diversity of terrestrial, semi-aquatic, and aquatic species. This report provides guidance in choosing cost-effective underpass design and location features that are necessary to consider to increase motorist safety and habitat connectivity. The findings also demonstrate that if only a minimal number of deer-vehicle collisions is prevented by an effective underpass, the savings in property damage alone can outweigh the construction costs of the structure. The increasing frequency of animal-vehicle collisions in the United States is taking an enormous toll in terms of wildlife viability and driver safety. For species that commonly attempt to cross roads, the numbers of animals killed can have a devastating effect on their populations. Roads and highways act as barriers for other species, isolating populations and increasing the chance of local extinction. For humans, more than $1.1 billion in vehicle damage is caused in the United States from an estimated 1.5 million traffic accidents involving deer alone (Hedlund et al. 2003). In Virginia, the white-tailed deer (Odocoileus virginianus) population has increased 400 percent since 1968, and Virginia s human population has increased 61 percent. As a result of these drastic increases, the number of reported deer-vehicle collisions (DVCs) in the state has increased nearly eight-fold in the last 35 years. In areas where black bear attempt to cross roads, road-mortality has significantly affected black bear (Ursus americanus) populations in the southern Appalachians. As roads are upgraded to accommodate greater traffic volumes, the rate of successful black bear crossings in the Appalachians decreases significantly, and black bears become reluctant to cross roads (Brody and Pelton 1989, Virginia Dept. of Game and Inland Fisheries 2002). This avoidance of roads can isolate wildlife populations, and ultimately reduce biodiversity and genetic variability. Wildlife crossings, or passages beneath or above a roadway, are a form of mitigation designed to facilitate safe wildlife movement across a transportation corridor. In a literature review of 16 mitigative techniques to reduce DVCs, the only measures consistently found to achieve DVC reductions were the installations of exclusionary fencing and wildlife crossing structures (Knapp et al. 2004). Similarly, the 2003 report issued by the Insurance Institute for Highway Safety regarding methods to reduce DVCs concluded: Fencing, combined with underpasses and overpasses as appropriate, is the only broadly accepted method that is theoretically sound and proven to be effective (Hedlund et al. 2003, p. 14). Because many states have relatively few structures designed to facilitate wildlife passage, monitoring for wildlife use is often limited to underpasses that were designed for other purposes. Structures such as bridges or culverts that were constructed to span streams and rivers, to protect wetlands, or to provide access for farm animals or equipment may function as wildlife crossings. Virginia has multiple structures throughout its roadway system that are likely used by wildlife. The Virginia Department of Transportation (VDOT) has constructed two structures designed for large mammal passage in northern Virginia, and others are currently under construction on Route 17 through the Great Dismal Swamp National Wildlife Refuge. However, VDOT has no information regarding the performance of any of its structures in terms of facilitating animal movement. The purpose of this study was (1) to determine the effectiveness of VDOT s existing large mammal crossings, (2) to determine the design and location attributes that make a wildlife crossing successful in terms of use by Virginia s large mammals and the associated influence on animal-vehicle collisions, and (3) to analyze the costs of wildlife crossing construction relative to the potential savings in property damage resulting from a reduction in animal-vehicle collisions. Methods Underpass study sites Seven underpass structures were monitored over a 12-month period, from June 1, 2004, to May 31, These sites were chosen in order to obtain a representative sample of structures beneath Virginia roadways that potentially function as deer and black bear crossings. Five of the structures (Sites 1, 2, 4, 5, and 6) were not constructed for the purpose of wildlife movement, and two structures (Sites 3A and 3B) were installed specifically for animal passage. Because most of the structures were not designed as wildlife crossings, study sites are generally referred to as On the Road to Stewardship 433 Wildlife Crossing Structures

9 underpasses or structures rather than wildlife crossings in this report. Sites 3A and 3B were constructed with a section of grating 45 ft by 10 ft (18 m by 3 m) centered in the ceilings (in the highway median above) to allow in sunlight. Most structures convey water (generally a narrow stream or creek) but also offer ample space for animal movement. Eleven variables including structural, landscape, and human activity attributes were measured at each site (table 1). The openness factor, a structural variable used as a measurement of ambient light in a structure, was calculated by the equation (width x height)/length. Openness has been found to be a significant factor in determining relative effectiveness of structures in terms of use by deer and other species (Reed et al. 1975). Other attributes, including structure ground covering and frequency of human visits to the structures, were also recorded. The description of the deer habitat suitability indices are described below. Table 1. Attributes of underpass structures in Virginia monitored from May 2004 through May 2005 a Openness was determined by calculating (height x width)/length (Reed et al., 1975). b Because the ceilings of structures 3A and 3B have a grated center section that allows in light, the openness value was calculated using 3 4 of the length. C Based on average annual daily traffic of High (10,000-49,999), Medium (1,000-9,999), and Low (0-999). d Based on a scale of 1 to 100. Land use was similar in the immediate surroundings of study sites, reflected by the proximities of the indices. Underpass monitoring Data from monitoring animal movements were obtained from Game-Vu (Nature Vision, Inc.) and Stealth Cam digital scouting cameras. These remote cameras photograph images based on infrared heat and motion sensors. Game-Vu Digital Trail cameras use undetectable infrared illumination at night rather than a flash and were, therefore, installed at sites where human visitation was a concern. Stealth Cam cameras were used at the other sites because of their slightly longer range at night. Two cameras were installed at each site. For box culvert monitoring, one camera was attached to a tree, to a wooden post, or near the ground at both of the structure entrances. For bridge monitoring, two cameras were attached to trees on either end of the bridge. Because cameras could not capture the entire range beneath the Site 2 bridge, sand beds at each end of the bridge were checked weekly for large mammal tracks to supplement camera monitoring. Structures were visited once every week during the 12-month period to download photographs from the cameras and to replace batteries. Data recorded from photographs at each site included the date, time, number of photographs of each species, and direction of travel. The number of complete passages through the structure by deer and black bears, the number of turn-around events (approaches to an underpass with incomplete crossings), and the number of hesitancy behaviors by deer (indicated by muzzles lowered to the ground) were determined (Reed et al. 1975, Gordon and Anderson, 2003). Chapter ICOET 2005 Proceedings

10 On some occasions, camera battery power depleted one to two days prior to replacement. In order to account for site differences in camera operative days, crossing frequency indices were calculated by dividing the number of crossings by deer and black bears at each site by the respective number of camera operative days. Development of deer habitat suitability indices In order to make valid comparisons of underpass use by deer, it was necessary to quantify either deer population or deer habitat suitability in the vicinity of each underpass. Given the spatial and geographic variation among the seven sites, it could not be assumed that deer populations and habitat suitability were uniformly distributed around the sites. Because the size of the deer population immediately surrounding each site was unavailable, deer habitat suitability indices were developed. The higher the index, the higher the relative deer habitat suitability was surrounding the underpasses. These indices were later used in the statistical analyses of underpass use. A geographic information system (GIS) is a widely used tool for modeling habitat suitability for a variety of applications. Much of the development of deer habitat suitability indices for this study was adapted from the work of Clevenger et al. (2002) and Clevenger and Waltho (2005). They found that a GIS-generated model, using habitat information derived from expert literature to perform pairwise comparisons, most closely approximated an empirical model for identifying black bear habitat. For the development of the deer habitat suitability indices for this study, a similar methodology was applied with the use of ArcGIS (Environmental Research Institute, Redlands, California) and a pairwise comparison matrix (Saaty 1977). National land cover data were obtained from the U.S. Geological Survey and imported into ArcGIS. This dataset included 13 habitat types within the areas surrounding the underpass sites. Because the home range of deer is generally no larger than one mile (Severinghaus and Cheatum 1956), a one-mile buffer was generated around each of the seven underpass sites. The ArcGIS Spatial Analyst extension was then used to determine the percentage of each habitat type within each one-mile radius. For the pairwise comparison, each of the 13 habitat types was rated against the other in terms of relative importance of the habitat for white-tailed deer. Ratings were based on a nine-point continuous scale: 9, extremely more important; 7, very strongly more important; 5, strongly more important; 3, moderately more important; 1, equally more important; 1/3, moderately less important; 1/5, strongly less important; 1/7, very strongly less important; and 1/9, extremely less important (Saaty 1977). Information on which to base deer habitat ratings was obtained from four sources of information (Harlow 1984, Newson 1984, Shrauder 1984, Whittington 1984); the latter three were directly relevant to deer in Virginia. Two individuals, including the author, used the information from these sources to complete the pairwise comparison matrix. The completed pairwise comparisons resulted in weights for each habitat type. A consistency ratio was calculated to ensure consistency in rating development. Pairwise comparison matrices with consistency ratios greater than 1.0 were reevaluated (Saaty 1977). The percentage of each habitat type (derived from the GIS analyses) within the one-mile radius of each site was then multiplied by its weight (derived from the pairwise comparison). The weighted values for the site were summed to derive a deer habitat suitability index. This method was repeated with each of the seven underpass sites. Indices are listed in table 1. Underpass evaluations The numbers of deer and black bear crossings, turn-arounds, and hesitancy behaviors were compared among all underpass sites. Sites 3A and 3B are located approximately 0.25 mi (400 m) from one another and potentially facilitate the movement of the same animal populations. Because they primarily differ in structural attributes, the use of these structures by all species was compared to help provide valuable information on animal preferences for crossing structures. Data analyses Multiple regression analyses were performed to determine the influence of underpass attributes on deer crossing frequency, adjusting for differences in deer habitat suitability between sites. Statistical analyses were performed with the assumption that the crossing frequency was a measure of the quality of the underpass as sensed by wildlife. For all analyses, differences were considered statistically significant when p < Criteria for success In order to evaluate an underpass in terms of its effectiveness in reducing the barrier effect of roads and reducing animal-vehicle collisions, it was necessary to specify the criteria for success. Goals and criteria, adapted from those of Forman et al. (2003), include (1) maintain habitat connectivity (determined by a minimum passage of animals detected), (2) maintain genetic interchange (determined by a passage of adults), and (3) allow for dispersal (determined by passage of juveniles). Determination of deer-vehicle collisions relative to underpass locations Using available information obtained from Virginia s Highway Traffic Records Information System and Fairfax County police records, the number and locations of DVCs reported from 1997 through 2001 and 1995 through 2004, respectively, were recorded within several miles of each monitored site used by deer. Because only reported collisions are included in these records, however, a potentially large percentage of the actual collisions that occurred was unavailable for analysis. On the Road to Stewardship 435 Wildlife Crossing Structures

11 Cost analysis To transportation agencies, cost is often the largest deterrent to constructing wildlife crossings. Because there are currently no regulatory directives or guidelines pertaining to wildlife crossings, the decision by transportation agencies to construct them is often based on the expected return in investment. This may be in the form of ecological benefits and increased driver safety. Ecological benefits include the creation of wildlife corridors, reduced effects of fragmentation (Forman et al. 2003), and reduced road mortality. Driver safety includes a reduction in animal-vehicle collisions and the corresponding reduction in deaths, injuries, and property damage, which translates into savings for taxpayers. With regard to taxpayer savings, one human fatality from a DVC can result in a loss of millions of dollars in damage, hospital costs, and lost wages. Property damage costs alone comprise a substantial taxpayer cost. Since assigning a monetary value to the ecological benefits is difficult, the economic benefits solely in terms of a reduction in property damage were analyzed. Property damage values were derived from the 2003 average cost in property damage from DVCs in Virginia ($2,530). Construction costs for two effective underpasses in this study, Sites 1 and 3A, were used for the analyses. Annualized costs, or the equivalent uniform annual costs, were calculated for these underpasses for comparison with annual DVC incidents. Annualized costs in these examples are the yearly costs of an underpass as if they were uniform throughout the service life of the structure (estimated at 70 years; Blackwell and Yin 2002). Although these sites do not have fencing designed to funnel animals toward the structures, fencing prices ($125,000 for each structure) were added to the underpass construction costs to represent more realistic examples of wildlife crossing scenarios. Results Deer and black bear activity A total of 2,702 photographs of wildlife were captured at the seven sites over the 12-month period (fig. 1). Six of these photographs were of black bear, and 1,040 were of white-tailed deer. Figure 1. Black bear approaching entrance of Site 1 (A), deer in Site 1 (B), red fox in Site 3A (C). Black bears No black bears crossed through any of the monitored underpasses, although they approached the northern entrance of Site 1 on three occasions. On two occasions, a bear remained facing the culvert entrance for two minutes before turning and leaving the area. On September 20, a bear approached the entrance a second time 38 minutes after the first approach. Deer A total of 1,107 deer crossings occurred through four of the seven underpass sites in the 12-month period (x= 277, N = 4). Sites 1, 2, and 3A received the heaviest use, and Site 3B the least. There were no crossings or visits by deer in Sites 4, 5, and 6. Although Site 1 received relatively high use by deer, it was associated with the highest number of turn-around events and hesitancy behaviors (fig. 2). Figure 2. Number of white-tailed deer crossings (A) and number of turn-arounds and hesitancy behaviors (B) for underpasses visited by deer from June 2004 through May Chapter ICOET 2005 Proceedings

12 The crossing frequency index was highest and most consistent at Site 2, with 1.34 crossings per day. Site 1 averaged 1.1 crossings per day, Site 3A averaged 0.91 crossing per day, and Site 3B received relatively little use, at 0.02 crossing per day. The monthly crossing frequency indices at each site were highest in the autumn months, dropped steeply in winter, and rose in late spring. During the period of heaviest activity in the fall, Site 1 received the heaviest use, reaching an average of 4.7 crossings per day in September (fig. 3). Figure 3. Average number of white-tailed deer crossings per day each month, from June 2004 through May Other wildlife activity Although the focus of this project was large mammal use of underpasses, the number and species of other wildlife were also recorded. Each underpass site was used by a minimum of two species. Other species detected included opossum, squirrel, house cat, bobcat, red fox, coyote, raccoon, groundhog, mice species, amphibian species (southern leopard frog and American toad), black rat snake, at least two bird species (songbird and great blue heron), and fish species. Because amphibian, reptile, mouse, and fish use of the underpasses was observed but not detected by cameras, these species were not included in the analyses. Nocturnal species used the underpasses between dusk and dawn, with daytime use generally limited to deer and groundhog. Cameras at Site 3A photographed a coyote with a small mammal in its mouth (species cannot be determined). Because of the proximity and similar landscape attributes of Sites 3A and 3B, the sites were useful for comparing use by species. Activity was greater for all species in Site 3A, with 1,177 photographs compared to the 708 photographs at Site 3B (fig. 4). Underpass evaluations Figure 4. Number of photographs in Site 3A (A) and 3B (B) taken by two cameras at each site. Data analysis A large discrepancy in deer crossing frequency was apparent between structures with a height greater than or equal to 12 ft (3.7 m) and those with a height less than 12 ft. To represent this distinction, height values were differentiated into these height groupings. Adjusting for deer habitat suitability at each site, an underpass height greater than 12 ft was significantly correlated to crossing frequency (Beta = 0.78 ± 0.20, P = 0.047). Landscape and human activity variables were not significantly correlated to crossing frequency. Criteria for success Underpasses were evaluated according to whether they met the predefined goals (table 2). Because no black bears crossed during the monitoring period, evaluations were based solely on white-tailed deer. In terms of meeting all three underpass goals for deer, Sites 1, 2, and 3A were determined to be effective overall. On the Road to Stewardship 437 Wildlife Crossing Structures

13 Table 2. Goals and successes (in terms of white-tailed deer use) of seven underpasses monitored from June 2004 through May 2005 a Based on a comparison of the number of crossings at Site 3B relative to that at Sites 1, 2, and 3A, it is unlikely that 7 crossings over a 1-year period adequately met the goal of maintaining habitat connectivity. It is possible that the proximity of 3B to 3A influenced the crossing frequency at 3B. Deer-vehicle collisions relative to underpass locations Small sample sizes restricted the statistical analyses of the number of DVCs immediately surrounding the underpasses relative to segments with no underpasses. These data were, therefore, depicted graphically to illustrate the number of DVCs adjacent to underpass locations (fig. 5). For Site 1, there were no reported DVCs 0.75 mi to the west and 1.25 mi to the east of the underpass within a five-year period ( ) for which data were available. At the section east of the underpass, a high ridge prevents deer from entering the highway. Within the 2.5-mile road segment (flanked by two perpendicular roads) under which Sites 3A and 3B lie, there were five DVCs within a 10-year period (1995 through 2004). Figure 5. Reported deer-vehicle collisions in relation to location of underpasses used by white-tailed deer. DVC data for I-64 (A) include a 5-year period ( ), and data for the Fairfax County Parkway include a 10-year period ( ). Cost analysis The annualized costs were $6,600 for Site 1 and $23,000 for Site 3A (based on total costs of $257,000 and $585,000, respectively). Underpasses at these prices are cost-effective in terms of property damage savings when they prevent a minimum of 2.6 DVCs per year and 9.1 DVCs per year, respectively (fig. 6). Chapter ICOET 2005 Proceedings

14 Because the Site 1 and Site 3A underpasses were constructed when the road was constructed, pre-construction DVC data are unavailable. However, considering that the numbers of deer crossings in Site 1 and Site 3A were 319 and 293, respectively, it is probable that more than 2.6 and 9.1 DVCs, respectively, were prevented that year. If this was the case, then the savings in property damage alone outweighed the annualized cost of the underpasses. Discussion Figure 6. Cost savings in property damage resulting from reduction in deer-vehicle collisions. The results of this research concur with those of studies that have found that properly sized and located structures receive heavy use by wildlife (Foster and Humphrey 1995, Clevenger and Waltho 2005, Mata et al. 2005), thereby reducing the ecological effect of roads and reducing animal-vehicle collisions. For large mammals in Virginia, appropriate structure design is essential for maximizing the benefit from wildlife crossing construction. Sites 1, 2, and 3A were determined to be effective road-crossing mechanisms for deer. Crossing frequencies were highest during late summer and fall. Crossing use reflected seasonal activity, with deer averaging nearly five crossings per day at Site 1 in the fall months. This corresponds to greater periods of movement associated with mating and feeding activities. The other sites were ineffective in terms of facilitating deer passage. For most sites, this was a result of the structure s small size and corresponding low openness factor. For the bridge of Site 6, this was not the case. At this site, the structure size was adequate, but the uneven approach and lack of visibility from one end to the other likely discouraged large mammal use. At the western opening of this bridge, a four-foot ledge slightly impeded access to the area beneath the bridge, and the ledge and a rock cliff obstructed views of the habitat on the far side of each entrance. Effective underpasses were easily accessible with level approaches and had clear lines of site to the habitats on the far side. Black bears approached but did not cross through any of the underpass sites. Because of annual fluctuations in food availability, environmental conditions, and inter- and intra-specific interactions, however, one year of data collection is insufficient to allow the conclusion that the structures are unsuitable for bears (Manen and Pelton 1995). Attributes of effective underpasses White-tailed Deer Underpasses with a minimum height of 12 ft (3.7 m) were significant determinants of deer crossing frequency. The bridge (Site 2) had the highest deer passage rate and lowest number of incidents of hesitancy behavior and turnarounds. This was expected because of its large size and lack of walls, unlike the Site 1 and 3A box culverts. Although the second highest number of crossings was at Site 1, 19 percent of the approaches to Site 1 were associated with turn-arounds. The high number of crossings at this site was likely influenced by the structure s position in the landscape. The southeastern borders along the underpass openings slope to a high ridge. This ridge functioned as a barrier to deer movement across the highway (as evidenced by no DVCs within 1.25 mi [2,012 m] east of the underpass), and the surrounding hillsides served as a guideway for deer toward the underpass (fig. 7). Although the optimal placement of Site 1 undoubtedly contributed to its high use by deer, the high number of incidents of hesitation and turn-arounds is likely explained by its relatively low openness index (0.19). Figure 7. Aerial view of topographic features surrounding Site 1. On the Road to Stewardship 439 Wildlife Crossing Structures