Spatial, Roadway, and Biotic Factors Associated with Barn Owl (Tyto alba) Mortality and Characteristics of Mortality Hotspots Along Interstates 84 and 86 in Idaho
Date of Final Oral Examination (Defense)
Type of Culminating Activity
Master of Science in Raptor Biology
James Belthoff, Ph.D.
Eric Yensen, Ph.D.
Jesse R. Barber, Ph.D.
One of the world’s highest roadway mortality rates for barn owls (Tyto alba) occurs along Interstate 84/86 (I-84/86) in southern Idaho. Although mortality occurs in numerous portions of the I-84/86 corridor, there are segments where relatively much higher numbers of owls are killed (in total comprising >20% of the corridor total, hereafter “hotspots”). My objectives were to 1) identify areas of greatest mortality (hotspots), 2) understand the spatial, roadway, and biotic factors potentially contributing to barn owl-vehicle collisions and 3) assess how mortality hotspots have changed over time. If factors contributing to barn owl mortality along highways can be identified, it may be possible to find ways to reduce barn owl-vehicle collisions in this region. To do so, I conducted road surveys to identify locations of barn owl-vehicle collisions, and quantified spatial, roadway, and biotic factors along the focal highway to examine how they related to patterns of barn owl roadway mortality. I also quantified mortality hotspots to examine temporal and spatial changes between a previous survey in 2004-2006 and this study in 2013-2015.
Standardized road kill surveys conducted by Than Boves from 2004 to 2006 located 812 dead barn owls. Between 2013 and 2015, I located another 550 dead barn owls. I characterized nine spatial, 19 roadway, and nine biotic variables that may potentially affect barn owl roadway mortality using squares of 1-, 3-, and 5-km lengths centered on 120 randomly selected sites along the I-84/86 corridor. I evaluated variables at each of the three scales in relation to the number of dead barn owls counted along 1- and 5-km highway segments to determine their respective best scales (either 1-, 3-, or 5-km) using Akaike Information Criterion (AICC). This approach produced two sets of models: the 1-km highway segment model set and the 5-km highway segment model set.
The final variable set included 14 variables for both the 1- and 5-km model sets. I assessed the potential effects of all possible combinations of these variables within each set (spatial, roadway, and biotic) on number of dead barn owls in 1- and 5-km highway segments using Generalized Linear Models within an AICC information theoretic model selection framework and combined the variables from the top models in each variable set into a final set in which I assessed all possible combinations (a total of eight variables for the 1-km set and seven variables for the 5-km set). I averaged the variables into a final model for the 1-km set, whereas model averaging was not necessary for the 5-km set. One of the variables in the final 1-km model (width of the median) was further analyzed to determine its potential correlation with percent land cover type.
In the final 1-km model set, percentage human structures, cumulative length of secondary roads (length of all roads other than I-84/86), and width of median had an inverse relationship with the number of dead barn owls/1-km segment/survey. Percent land cover type varied with the width of the median in that the median was generally wider when the highway was surrounded by shrubs (rs = 0.30, p = 0.0008) and narrower when surrounded by crops (rs= -0.24, p = 0.009). The number of dead barn owls/1-km segment/survey increased with commercial average annual daily traffic (CAADT), small mammal abundance index, and when the plant cover type in the roadside verge was grass. The final model for the 5-km model set included percentage of crops in which the number of dead barn owls/5-km segment/survey increased as the percentage of crops increased. Barn owls are associated with agricultural lands and thus less likely to occur in areas with high percentages of human structures, secondary roads, and when the median is wide in shrublands. Barn owl carcasses increased with higher small mammal abundance index values as well as when there was grass in the verges. Furthermore, the small mammal abundance index was greater in grass versus mixed shrub verges (Wilcoxon rank sum test: eastbound verge, W = 1507, p = 0.01; westbound verge, W = 2255, p
I evaluated temporal and spatial changes in hotspots between survey periods using point density estimation and KDE+. Additionally, of the 120 randomly selected sites, I calculated which fell within an area delineated as a hotspot and which did not as defined by the point density estimation analysis. I compared characteristics of the two types of sites (hotspot and non-hotspot) for the 14 spatial, roadway, and biotic variables selected for final modeling.
The area between Bliss and Hazelton was the section of I-84/86 with the highest rates of barn owl-vehicle collisions in both surveys, although particular hotspots did exhibit some expansions and contractions between 2004-2006 and 2013-2015. Two of the historical hotspots no longer appeared as hotspots in the recent surveys indicating they perhaps have shifted or were so fatal they reduced the local barn owl population and thus no longer appear as hotspots. Therefore, these historical hotspots may still be important mortality zones and important for future mitigation consideration as the hotspots potentially have reduced the barn owl population in these areas.
The most important difference between hotspots and other sites was the higher number of secondary roads (Wilcoxon rank sum test: W = 613, p = 0.001) and higher traffic volume (W = 600, p = 0.002) in hotspots. However, hotspots were also generally situated close to the Snake River Canyon and other water features which should have more prey, provide nesting and/or roosting sites, and attract owls; had low slopes (level terrain) which would allow owls to fly low to the pavement; narrow medians (correlated with cropland); and flexible rather than rigid pavement type (potentially related to noise level), and did not contain the highest number of dairies (which should attract owls to their higher rodent populations). The hotspots were also in regions of I-84/86 with moderate to high small mammal abundance and features that should correlate with higher rodent abundance: low percentages of human structures near the highway, grass cover types in the median and verges, high percentages of crops, and few obstructions to low flight.
Mortality hotspots along I-84/86 were generally devoid of low flying obstructions, so establishing barriers to low flight may be an effective technique to reduce barn owl-vehicle collisions. Reducing small mammals in verges and median vegetation could also potentially reduce barn owl mortality. Because I found fewer small mammals in areas with shrubs, establishing taller shrub vegetation may reduce small mammal habitat and reduce hunting success, encouraging owls to hunt elsewhere. Reducing wildlife-collisions involving barn owls in Idaho is important for motorist safety and would be an important step in ensuring the persistence of this avian species.
Arnold, Erin Melissa, "Spatial, Roadway, and Biotic Factors Associated with Barn Owl (Tyto alba) Mortality and Characteristics of Mortality Hotspots Along Interstates 84 and 86 in Idaho" (2016). Boise State University Theses and Dissertations. 1129.