Cover Image

Not for Sale



View/Hide Left Panel
Click for next page ( 65


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 64
64 Topography is mountainous; elevations range from 1,300 m (Odocoileus hemionus) herd, which travels across the highway to over 3,400 m; and valley floor width varies from 2 to 5 km. during seasonal migrations. The highway also bisects the Highways in the study area traverse montane and subalpine home ranges of numerous resident deer and is used by forest ecoregions through four major watersheds in the region. carnivores and amphibians. Caltrans has consistently col- Table 26 describes the location and general characteristics of lected deer carcass data on this highway from June 1979 to the five segments of highways that were included in this October 2005. The research team used 849 deer carcass loca- study. Vegetation consisted of open forests dominated by tions collected along 33 mi of SR 89. The data were collected by Douglas fir (Pseudotsuga menziesii), white spruce (Picea maintenance supervisors and vary in spatial accuracy (1.0-mi, glauca), lodgepole pine (Pinus contorta), Englemann spruce 0.1-mi, and 0.001-mi). (P. englemannii), aspen (Populus tremuloides) and natural grasslands. The research team obtained the UTM coordinates using a GPS unit for over 500 spatially accurate carcass loca- Findings and Results tions (< 3 m error) of WVCs between 1997 and 2004 in the Hotspot Identification and Patterns for Canadian Rocky Mountains. The research team used the One Species and Landscape UTM Nad 83 location to plot all of the WVC data within ArcGIS 9.0 on the highway network. The hillshade raster Simple graphic techniques, one dataset. A total of 546 dataset was derived from the digital elevation model (DEM: WVC observations were recorded between August 1997 and Parks Canada, GIS data management) and used as a backdrop November 2003 on all five highways in the study area. Deer layer for visual interpretation. composed 58% of the kills, followed by elk (27%), moose (7%), Rocky Mountain bighorn sheep (3%), and other un- Northern California. This study took place in Sierra gulate species (5%). The majority of WVCs occurred on the County, California, in the Sierra Nevada Mountains (Figure TCH east of Banff National Park in the province of Alberta 11). California State Route (SR) 89 runs along the east side of (46%), followed by Highway 93 South in Kootenay National the Sierra Nevada Mountains from the towns of Truckee to Park (22%), Highway 40 in Kananaskis Country (12%), the Sierraville. The highway is a two-lane undivided highway with TCH in Yoho National Park (10%), and the TCH in Banff an AADT of 2,250, peaking at 3,300 in the summer months. National Park (10%). Elevation ranged from 6,150 ft ( 1,875 m) surrounding the A simple visual analysis of the WVC locations is shown in southern most section of the highway (mile-marker 11.0) to Figure 12. A simple plotting of all WVC locations along high- 5,081 ft ( 1,549 m) in the northern section. The dominant ways does not clearly identify key areas where WVCs vegetation was ponderosa pine (Pinus ponderosa), bitterbrush occurred or areas where higher than average densities of (Purshia tridentata), and sagebrush (Artemisia tridentata). The collisions occurred--at least in this type of mountainous area is located in a rainshadow and receives relatively less landscape that characterized the study area. Simple plotting precipitation than more westerly locations. Snowfall can resulted in carcass locations being tightly packed together, in accumulate 2 to 3 ft ( 0.6 to 0.9 m) during winter (Sandy some cases directly overlapping with neighboring carcass Jacobson, personal communication). SR 89 bisects an impor- locations, thus making it difficult to identify distinct clusters, tant migration route for the Loyalton-Truckee mule deer i.e., where the real high-risk collision areas occurred. The use Table 26. Characteristics of the major highways in the study area. Posted Road Traffic Vehicle Highway Watershed Province Length Volume Speed (km) (ADT) (km/h) Trans-Canada Alberta, East of Banff Bow River 37 16,960a 110 Highway National Park Trans-Canada Banff National Park, Bow River 33 8,000a 90 Highway Alberta Trans-Canada Yoho National Park, Kicking Horse River 44 4,600a 90 Highway British Columbia Highway 93 Kootenay National Kootenay River 101 2,000a 90 South Park, British Columbia Highway 40 Kananaskis River Alberta 50 3,075b 90 a 2005 annual average daily traffic volume. Data from Parks Canada; Banff National Park; and Alberta Transportation, Edmonton, Alberta. b 1999 summer average daily traffic volume. Data from Alberta Transportation, Edmonton, Alberta.

OCR for page 64
65 Figure 11. Location of study area in northern California in Sierra County, CA. of a DEM and/or land cover map overlay does provide read- Analytical techniques, one dataset. The research team ily available information on the juxtaposition of WVCs to ter- used a linear nearest neighbor analysis, cluster analysis, Rip- rain features (e.g., lowlands, lakes, steep terrain, vegetation ley's K analysis, and density measures to identify collision cover types). hotspots at different scales of application. A visual analysis can provide some cursory conclusions about why and where WVCs tend to occur most. However, a Linear nearest neighbor analysis. All WVCs were plotted more rigorous spatial analysis can be carried out to summarize for each highway on the highway network layer in ArcGIS 9.0. or test statistically the "why and where" questions. Terrain and The research team used the Hawth's Analysis Guides24 exten- habitat are often key factors influencing where WVCs occur sion to generate the same number of "random WVCs" as (see Section 3.2).52,15,158,25 Type of terrain and the nature of the there were actually observed on each highway. A first order landscape mosaic likely influence WVC hotspot clustering pat- linear nearest neighbor index (NNI) was then used to evalu- terns. For example, landscapes with homogeneous cover types ate if the distribution of the observed WVCs in each region of and with little topographic relief (i.e., flat terrain) would likely the Canadian Rocky Mountains differed from a random result in a more random pattern of movement across a high- distribution. The NNI is a ratio between the mean nearest dis- way, and thus a more dispersed pattern of collision locations tance to each WVC [d(nn)] and the mean nearest distance on a given stretch of highway. Contrarily, a highly heteroge- that would be expected by chance [d(ran)]. Hawth's Analysis neous landscape with dissected topography is more likely to re- Guides were used to calculate d(nn) and d(ran). sult in more clearly defined crossing locations and collision NNI = d(nn)/d(ran) hotspots. The factors that contribute to these collisions will be different in both landscapes. More simplistic models with If the observed mean distance is smaller than the random fewer explanatory variables could possibly be used to charac- mean distance, then the WVCs occur closer together than ex- terize the level, more homogeneous landscape, but more com- pected by chance and NNI < 1. Once tabulated, the data were plex models with numerous variables may work better in the imported into Microsoft Excel and a Z-statistic adapted from more diverse landscape. Landscape diversity may well influ- Clark and Evans50 was calculated to test if there were signifi- ence the causes and spatial distribution of WVCs. cant differences between random and observed distances.

OCR for page 64
66 Figure 12. Spatially accurate locations of WVC locations on each road in each of the watersheds.

OCR for page 64
67 Table 27. Descriptive statistics of the nearest neighbor clusters and high-kill zone aggregations. Mean cluster No. Mean high-kill zone Cluster No. Region length high-kill aggregation length overlap clusters a SD (km) zones SD (km) index TCH-Yoho (YNP) 6 1.00 0.32 7 2.80 1.33 1.00 TCH-Banff (BNP) 6 0.92 0.38 11 4.40 3.98 0.64 TCH-Alberta (AB) 7 1.32 0.14 12 3.84 1.63 0.90 Hwy 40-Kananaskis 6 0.96 0.42 13 4.16 3.29 0.85 Hwy 93-Kootenay (KNP) 17 0.86 0.39 19 3.49 2.34 0.56 a (cluster lengthhigh kill zone /cluster length) The nearest neighbor index showed clustering (NNI < 1) determine whether a 1-mi buffer was a high- or low-kill zone for all highway regions except for the TCH in Yoho, which (see "Density measures: WVCs per mile segment" below). A showed evidence of dispersion (Table 27). The Z-statistic was convex hull was used as the cluster output; it draws a polygon significant (P < 0.05) for the TCH in Alberta and marginally around the WVCs in the cluster. Because roadkills occur in a significant (P = 0.066) for Highway 93 South. The NNI used one dimensional plane, a line was drawn from the two outer- in this analysis is only an indicator of first order spatial ran- most points along the road within the convex hull for visual domness; a K-order nearest neighbor distance (e.g., second or display and to calculate the length of each WVC cluster. third order) would likely better describe the overall spatial The nearest neighbor CrimeStat analysis produced a total distribution of WVCs.145 Sample sizes were small on the TCH of 42 WVC clusters along 41 km of highway in the study area in Yoho and Banff, and on Highway 40 in Alberta (n < 100), (Figure 13). Compared to the simple visual analysis of making overall spatial distributions of WVCs in these regions WVCs, the CrimeStat modeling technique effectively difficult to describe. reduced the blurring of WVC hotspots on long stretches of The linear NNI is a quick and easy statistical test of spatial highway. As mentioned earlier, simple plotting of WVC lo- distribution of WVCs to initially determine whether colli- cations tends to result in tight groupings of collision points sions are distributed randomly across a stretch of highway or that often overlap with other WVC locations, making it a larger highway network (e.g., a DOT district or region). If the challenge to identify where the really high-risk collision areas test indicates WVC clustering (NNI < 1.0), then the subse- actually occur. The location and number of WVC hotspots quent step would be to identify where the clusters occur using generated by the CrimeStat technique are clearly defined and a GIS-based spatial analysis. Some spatial analysis techniques can be identified with associated landscape or road-related include cluster analyses using a GIS-based NNI,25 mapping features in each highway area. roadkill densities using a "moving window" analysis,213 or a road segment approach to mapping roadkill densities.89,136 Ripley's K analysis. Ripley's K statistic describes the dis- One approach that has great promise and is user-friendly is persion of data over a range of spatial scales.200,67 Ripley's K the CrimeStat program developed by Levine.146 statistic was calculated for all WVC mortalities in each region. The research team used the K statistic as defined by Cluster analysis: nearest neighbor hierarchical technique. Levine,145 but modified it for points distributed in one The research team used CrimeStat version III146 to determine dimension (e.g., along a line or road network). The result- the location of high-kill zones or WVC hotspots within each ing algorithm was coded in AvenueTM and run in ArcView of the five highways of the Canadian Rocky Mountains study GIS.77 The algorithm counted the number of neighboring area. CrimeStat is a nearest neighbor hierarchical technique, WVCs within a specified scale distance (t) of each WVC, which identified a series of points that are spatially close based and these counts were summed over all WVCs. The research on a predefined set of criteria.146 The clustering is repeated team standardized the WVC totals by sample size (N) and until either all points (WVCs) are grouped into a single clus- highway length (RL) to allow for comparison between each ter or else the clustering criterion fails. A fixed threshold dis- highway region. The process was repeated for incrementally tance (800 m) was used for the search radius to determine the larger scale distances up to RL for all five highways. The K inclusion of a WVC in a cluster. This threshold distance statistic (adapted from Levine and O'Driscoll)77,185 was (800 m) is the same radius used in the mile-marker density defined as: analysis (see "Density measures: WVCs per mile segment" below). The criterion for the minimum number of points RL N N required to define a cluster was the mean number of WVCs K(distance)obs = I(dij ) N 2 i =1 j =1 per mile for each highway region, the same criterion used to i j

OCR for page 64
68 Figure 13. Clusters or hotspots derived from CrimeStat III software on each road in each of the watersheds in Alberta, Canada.

OCR for page 64
69 where dij is the distance from WVC i to WVC j and I(dij) is an The distribution of WVCs was heterogeneous and signif- indicator function that returns 1 if dij < _ distance and returns 0 icantly more clustered or dispersed than would be expected 185 if otherwise. A distance increment of 280 m was used for all by chance over a wide range of scales (P < 0.05, Figure 14). five highway regions to allow for a minimum of 100 ds bins on In all highway regions there was significant clustering of the shortest section of highway (i.e., the TCH in Alberta). WVCs and some significant dispersion. The TCH in Yoho To assess the significance of K-values, the research team ran had a small degree of clustering from 1 to 2 km at an inten- 50 simulations of the above equation based on random distri- sity of 0.3 km, and significant dispersion at spatial scales butions of points for each of the five categories. Figure 14 from 3 to 12 km and 18 to 45 km. This dispersion peaked displays the results as plots of L versus distance, where L is the at an intensity of 7 km. Neighbor K statistics are well suited difference between the observed K-value and the mean of the for the description of one-dimensional spatial distribu- K-values for the 50 simulations.185 Positive values of L indicate tions.200,104,192 The range of scales over which clustering crowding and negative values indicate dispersion. Figure 14 appears significant is dependent on the intensity of the dis- also presents the 95% confidence limits, calculated as the tribution of roadkills.52,192 Peaks in L(t) (i.e., the intensity of upper or lower 95th percentile of the random simulations clustering) occurred between km 4 and 5 for the TCH in minus the mean of the random simulations.185 The research Alberta and the TCH in Banff, which means there was an team defined significant crowding as any value of L above the average of 4 to 5 extra neighbors within the scale distance of upper confidence limit and significant dispersion as any value 0 to 10 km on the TCH in Banff and 0 to 12 km on the TCH of L below the lower confidence limit. in Alberta. Both these aggregations can be seen in Figure 14. A) TCH in Yoho NP B) TCH in Banff NP 6 6 4 5 2 4 3 0 2 -2 0 10 20 30 40 50 1 -4 0 -6 -1 0 10 20 30 40 -8 -2 -3 C) TCH in Alberta D) Highway 40 Alberta 5 7 4 6 5 3 4 2 3 2 1 1 0 0 -1 0 10 20 30 -1 0 10 20 30 40 50 -2 -2 -3 -3 -4 E) Highway 93 South Kootenay NP 30 25 20 15 10 5 0 -5 0 20 40 60 80 100 120 Scale Distance t (km) Figure 14. Plotted values of L statistic for the Ripley's K statistic of WVCs from five highways in Canadian Rocky Mountain study area. Ordinate axis is L(distance) for all 5 graphs.

OCR for page 64
70 In Banff they correspond with the section of the TCH that bi- (indicated by the star symbol) is due to the presence of 4.5 km sects a North-South aligned major drainage. At large scale of fenced highway with one underpass, while the second gap distances, the TCH in Banff National Park and Alberta show in WVCs is due to a large lake and river system on the north a random distribution with small scales of dispersions. On side of the TCH. Highway 93 South there is a large peak (27 extra neighbors) For the second analysis, termed high kill and low kill, the re- in WVC clustering at a scale distance of 0 to 80 km. This peak search team categorized each mile-marker segment as a high- corresponds to the bulk of the WVCs that occurred at the kill or low-kill zone by comparing the summed number of southernmost section of Highway 93 in low-elevation mon- WVCs associated with a single mile-marker segment to the av- tane habitat. Further, the highway bisects a key ungulate erage number of WVCs per mile for the same stretch of road, movement corridor in this area. for each of the five highways in the study area. If the summed The Ripley's K analysis clearly shows the spatial distribu- number of WVCs associated with a single mile-marker seg- tion of WVCs along each segment of highway. The large- ment was higher than the average calculated per mile for the scale aggregation evident on Highway 93 South in Kootenay same highway, that mile-marker segment was considered a shows the importance of broad-scale landscape variables high-kill zone. Similarly, if the summed number of WVCs such as elevation and valley bottoms in a mountain envi- within a mile-marker segment was lower than the average for ronment. The scale extent of WVC aggregations in each that highway, the mile-marker segment was listed as a low-kill study area can be used to help determine the scale extent zone. Each low- and high-kill zone (buffer) was color-coded and type of variables to be used in explaining the occurrence and displayed on each highway segment along with the asso- of road mortality of wildlife. Further, the locations of high- ciated lakes layer. Other features in the landscape, such as intensity roadkill clustering within each area can help to human use and rivers, were not displayed because they were focus or prioritize the placement of mitigation activities, not available at the correct scale resolution. The lakes layer was such as wildlife crossings or other countermeasures, on each digitized from 1:50,000 topographic maps and only displayed highway segment. with an 800 m buffer around each highway in each region. To compare the level of aggregation of high-kill zones between Density measures: WVCs per mile segment. For the next highway regions, the research team measured the mean length two analyses, the mile-marker data generated from the study of each high-kill aggregation. A high-kill aggregation was de- described in Section 3.2 was used. The research team divided fined as a high-kill zone (buffer) with at least one neighboring each of the five highways in the Canadian Rocky Mountain high-kill zone. study area into 1.0-mile-marker segments and plotted all spa- When standardized for roadway length, the majority of tially accurate WVC carcass data onto each road network. WVCs occurred on the TCH in Alberta (13.5 roadkills/mi), The research team then moved each carcass location point to followed by the TCH in Banff (2.6 roadkills/mi), the TCH the nearest mile-marker reference point. The research team in Yoho (2.1 roadkills/mi), Highway 40 (2.1 roadkills/mi), recorded the UTM coordinates of each mile-marker location and Highway 93 South (1.8 roadkills/mi). These rates of and summed the number of WVCs in that mile-marker seg- WVC were used to determine high- and low-kill segments ment, defined as 800 m (~ 0.5 mi) on either side of the given in each highway region. This analysis produced 97.6 km of mile-marker location. high-kill zones on all highways in the study area (Figure 16). For the first analysis, termed the graduated or weighted mile In 52% of the cases, a high-kill zone had a neighboring high- kill, the research team weighted each mile-marker by the kill zone. Highway 93 South had the most high-kill zones; summed number of WVCs associated with it and used grad- however, the TCH in Banff had the highest mean length of uated symbols in ArcView 3.3 to display WVCs along each aggregated high-kill zones, while the TCH in Yoho had the highway region. A 1:50,000 DEM with a pixel size of 30 m lowest mean length of high-kill zones (Table 27). The stan- 30 m was used to derive the hillshade (GIS database manage- dard deviations on TCH-BNP were high, indicating that the ment, Banff National Park) for the highways in the study area size of aggregations fluctuated highly. Figure 16 shows one and used as a backdrop for visualization. Figure 15 effectively main aggregation and a few single high zones on the TCH in shows where the WVCs occurred in relation to the valleys and Banff. In both the mile-marker visualizations (Figures 15 rugged terrain of the Rocky Mountain landscape. The black and 16), the DEM backdrops clearly show that high-kill arrows in the figures indicate where there was a large cluster- zones are associated with valleys moving perpendicular to ing of WVCs, which generally was where the highway the direction of the highway. For example, there is a large bisected a valley bottom. The TCH in Alberta has a consistent aggregation (~13 km) of high-kill zones on Highway 93 stretch of WVCs (14 to 24 roadkills at each mile-marker) South in Kootenay National Park that bisects key ungulate from the Banff National Park east boundary to just west of ranges in the valley bottoms of the montane region, at an Highway 40. The first westernmost gap in mortality numbers elevation less than 1,240 m.

OCR for page 64
71 Figure 15. Weighted mile-markers derived from summed collisions by mile-marker on each road in each of the watersheds. for roadkill points to overlap and visually mask the impor- Comparison of Hotspot Identification Techniques tance of segments of highway that have a high density of Visual analysis and observation versus analytical WVCs. Modeling or analytical techniques permit a more de- techniques. The pros and cons of the simple visual analysis tailed assessment of where WVCs occur, their intensity, and of WVC versus more complex or analytical methods were the means to begin prioritizing highway segments for poten- discussed earlier ("Simple graphic techniques, one dataset"). tial mitigation applications. Last, the identification and de- Essentially, with simple plotting of WVCs there is a tendency lineation of WVC clusters, which often vary widely in length

OCR for page 64
72 Figure 16. Density of kills at each mile marker on each road in each of the watersheds. depending on distribution and intensity of collisions, facili- produced and together occupied a total of 41 km (15%) of tates between-year or multiyear analyses of the stability or highway in the study area. The nearest neighbor CrimeStat dynamics of WVC hotspot locations. technique was more conservative compared to the mile-marker density analysis; it identified less length of highway as a WVC CrimeStat versus density-based techniques. Using the hotspot. Additionally, the average length of WVC clusters was nearest neighbor CrimeStat analysis, 42 WVC clusters were shorter than the density-based high-kill aggregations; however

OCR for page 64
73 the CrimeStat analysis produced clusters that were not con- WVCs will tend to be greater in number and more uniformly tinuous (Table 27). If the research team had selected a larger distributed than on the Yoho highway. Cluster definition will search radius for inclusion of roadkill points, fewer clusters tend to diverge, and clusters from the two approaches will be- would have been identified. CrimeStat also consistently pro- come spatially isolated. The reason is that the density-based duced fewer clusters of WVCs than the mile-marker density method has a tendency to accommodate outlying or marginal analysis. WVCs that normally would not cluster using CrimeStat. Use of either technique for identifying WVC or roadkill hotspots may depend on the management objective. The Hotspot Identification and Patterns CrimeStat approach is useful for identifying key hotspot areas for Different Species and Landscapes on highways with many roadkills because it filters the road- kill data to extract where the most problematic areas lay. The For this analysis, the research team selected one clustering mile-marker density analysis results in identifying more technique (CrimeStat) and conducted a hotspot analysis for hotspot clusters on longer sections of highway. Although this two different datasets: WVC carcass data from Canadian approach appears to be less useful to management, it may be Rocky Mountains and Caltrans DVC carcass data for North- a preferred option where managers are interested in taking a ern California. The data for Northern California was broader, more comprehensive view of wildlifevehicle con- described previously in the "Study Area" section and shown flicts within a given area. This broader view may be necessary in Figure 17. CrimeStat version III 146 was used to determine not only to prioritize areas of conflicts but also to plan a suite the location of DVC carcass hotspots along SR 89 in Sierra of mitigation measures. The location of the larger clusters County, California, and the five highways in the Canadian produced by the density analysis could be tracked each year Rocky Mountains. For visual comparisons, the research team to determine how stable they are or whether there is a notable plotted all DVC data along SR 89 in Sierra County, Califor- amount of shifting between years or over longer time periods. nia. The following paragraphs describe the hotspot patterns This type of information will be of value to managers in ad- and configurations, and examine how they may differ by dressing the type of mitigation and intended duration (e.g., species and the two landscape types. short-term vs. long-term applications). The mean number of DVCs along California SR 89 was The nearest neighbor CrimeStat clusters followed a spatial 25.7 kills/mi for the 26-year period and equates to roughly 1 distribution similar to the mile-marker high-kill zones (Fig- kill recorded per mile per year. The simple plotting of carcass ure 13). The degree of overlap between the two techniques was high for three of the five highways. For example all the clusters on the TCH in Yoho fell within high-kill zone aggre- gations (Table 27). Similar patterns of overlap were found for the TCH in Alberta and Highway 40 in Kananaskis Country. Less overlap of clusters defined by the two techniques was found for Highway 93 South and the TCH in Banff. These results pose the questions: What mechanisms influence the spatial patterns of clusters derived by both techniques? Why is cluster overlap high in some areas, but low in others? Both techniques coincided perfectly on the TCH in Yoho (100% overlap), whereas they were most divergent on Highway 93 South in Kootenay National Park (roughly 50% overlap). The overlap of clusters on the other three highways was aligned with either one of the two endpoints above. From inspection of the WVC data on all five highways, the research team suggests that the amount of WVC cluster overlap from the two techniques is likely influenced by the density and distri- bution pattern of WVCs. High overlap was found on the TCH inYoho, where steep terrain dictates more or less where animals can cross the highway. There are few suitable loca- tions where wildlife can cross the TCH; thus, roadkills occur in clearly defined sections. Clusters will naturally overlap or be in proximity because collisions rarely occur outside the key Figure 17. Spatially accurate locations of highway crossing areas. On highways that have less topo- carcasses from deervehicle collisions on graphic constraints and more dispersed wildlife habitat, State Route 89 in Sierra County, California.