The following is the established format for referencing this article:Zulian, V., A. R. Norris, K. L. Cockle, A. N. Porter, L. G. Do, and K. L. De Groot. 2023. Seasonal variation in drivers of bird-window collisions on the west coast of British Columbia, Canada. Avian Conservation and Ecology 18(2):15.
We examined the effects of façade-level building and vegetation features on bird-window collision risk, and how these effects varied across seasons at a Pacific coastal campus with mild winters, abundant evergreen vegetation, and seasonally varied bird communities. We searched for bird carcasses at 57 façades of 8 buildings at the University of British Columbia (UBC) over 155 days between January 2015 and March 2017 (total: 8835 façade surveys). Collision monitoring occurred across five equal sampling periods that represented stages of the annual cycle of the bird community, including the fall and spring migratory periods, the breeding season, and the long overwintering period. For each season, we compared logistic regression models predicting the odds of a collision from different sets of façade and vegetation characteristics expected to influence collisions: façade area, area of glass, porous surface cover (ground and shrub vegetation, soil, leaf litter), tree cover, and the number of building stories reflecting vegetation. Consistent with other studies, area of glass had a positive influence on collision probability in all seasons; however, the effect was strongest during the fall migratory period, when daily collision mortality rates peaked at UBC. The number of stories reflecting vegetation also increased collision probability, but only in the fall, indicating that the vertical extent of vegetation and reflective glass may affect collision risk differently as bird communities change across seasons. Façade area increased collision probability only in the winter (a long and lethal period for bird collisions at UBC), reflecting different risk factors associated with the species most vulnerable to collisions in this season. Our results highlight the need to measure building and vegetation effects across the longest and most lethal stages of the annual cycle of birds, both to predict the impact of proposed buildings and to prioritize mitigation strategies that will result in the greatest conservation benefits.
The built environment, specifically glass on buildings and other structures, results in the direct mortality of an estimated 16 to 42 million birds annually in Canada and 365 to 988 million birds in the USA (Machtans et al. 2013, Loss et al. 2014). Birds collide with transparent glass because it is effectively invisible to them (i.e., it does not alert them to the obstruction in their airspace) and because they interpret reflections of vegetation or sky as true airspace or habitat (Klem 1989, Klem et al. 2009, Altshuler and Srinivasan 2018). Given that the physical extent of human development is expected to expand even more rapidly than the human population, and much of this expansion is occurring in regions of high avian diversity, avian collision mortality will likely increase across the planet unless effective mitigation measures can be implemented (Seto et al. 2013). There is a growing interest in designing bird-friendly cities (Beatley 2020), but to do so requires an understanding of the design factors that influence collision risk, factors that can vary widely across regions and across seasons within the same region (Hager et al. 2017, Loss et al. 2019, Riding et al. 2020).
A range of building façade features as well as local- and landscape-scale habitat variables interact with bird behavioral dynamics to create a building’s “mortality signature,” (Hager et al. 2013) or building-specific collision risk profile (Klem et al. 2009, Cusa et al. 2015, Riding et al. 2020). Fatal bird collisions with glass have been shown to increase with area of glass (Hager et al. 2013, Cusa et al. 2015, Elmore et al. 2021a), reflections of vegetation (Klem et al. 2009, Gelb and Delacretaz 2009, Borden et al. 2010), building size (Hager et al. 2017, Elmore et al. 2021a) and the presence of attractants that draw birds close to windows, such as vegetation (Borden et al. 2010, Hager et al. 2017), bird feeders (Bayne et al. 2012), and water (Klem 1989). Additionally, bird-window collisions are typically higher at buildings embedded within or adjacent to greenspace or surrounded by less intensive urban development (Cusa et al. 2015, Hager et al. 2017).
The façade and vegetation factors that influence collision risk are expected to vary geographically and seasonally because of variation in bird abundance, community composition, and vulnerability of species across different ecological contexts. Bird behavior, morphological traits, flight velocity, and seasonal- and regional-specific habitat use and diet can all interact with façade and vegetation characteristics to influence collision risk (Cusa et al. 2015, Wittig et al. 2017, Nichols et al. 2018, De Groot et al. 2021, Elmore et al. 2021a, Samuels et al. 2022; E. K. Jackson, J. A. Elmore, S. R. Loss, B. M. Winger, and R. Dakin, 2020, unpublished manuscript). For example, in the Eastern USA during the peak fall migration (September/October), after accounting for relative abundance, mid-upper canopy foragers were more susceptible to collisions compared to groups that foraged on the ground or in shrubs (Wittig et al. 2017). However, in the Toronto region (Eastern Canada), during the migratory seasons (April-May, September-October), susceptibility of foraging guilds depended on landscape context, such that ground foragers were likely to collide with buildings in highly urbanized areas, and foliage gleaners were more likely to collide in areas surrounded by abundant greenery (Cusa et al. 2015). In Stillwater, Oklahoma (USA), collision correlates varied seasonally and by bird species; for example, lawn cover increased collision risk in summer and fall, but not in spring, and for American Robins (Turdus migratorius) but not for any other species (Riding et al. 2020).
In North America, building and vegetation features influencing bird-building collisions have primarily been studied in the central and northeastern USA, and less is known about the factors that influence bird-building collisions in the Pacific Northwest, which may differ for several reasons. In central and northeastern USA, winters are cold, most vegetation is deciduous (Kuchler 1964), and collision rates are relatively low during the overwintering period, except at homes with feeders (Hager et al. 2013, Dunn 1993, Klem et al. 2004). In contrast, in the Pacific Northwest, a higher proportion of ground cover, trees and shrubs are evergreen, retaining their leaves throughout the year, and the highest density of birds occurs throughout the prolonged overwintering period between early November and mid-March (Butler et al. 2021). Moreover, Western North America supports bird species with strictly western ranges, as well as distinct subspecies and subpopulations of continentally distributed species. The ecological requirements, migratory routes, diet, and behavior of these species, subspecies, and subpopulations can differ from their eastern counterparts, and hence influence collision risk (Kelly and Hutto 2005).
We previously studied seasonal variation in collision mortality at the University of British Columbia, within a temperate rainforest biome on the west coast of Canada (De Groot et al. 2021). Daily collision mortality was highest in the fall migratory period. In contrast to most results from central and eastern North America, collision mortality at UBC was as high during the overwintering period as during the spring migratory period. The spring migratory season is relatively short (~2 months; April and May) compared to the prolonged overwintering period which, for most species, lasts ~5 months (early November to end of March). Therefore, cumulative collision mortality is likely higher in winter compared to the spring migratory period. To develop and prioritize mitigation strategies that effectively reduce bird collision mortalities in the Pacific Northwest, it is critical to identify risk factors year-round (Riding et al. 2020), and especially in the seasons with the greatest cumulative mortality.
We examine how façade-level features of buildings and surrounding vegetation influence bird-window collisions at UBC, and how these effects change across seasons. We predict that the occurrence of collisions will increase with increasing façade size, area of glass, vertical extent of reflections of vegetation, and extent of vegetation and other porous surfaces within 50 m of each façade, but that the magnitude of effects will vary across the seasons.
Our study was conducted at the Vancouver campus of the University of British Columbia (UBC; 49°15′37.73″ N, 123°14′45.54″ W; elevation 95 m; https://www.ubc.ca/our-campuses/vancouver/) in Point Grey, which is bordered by the Strait of Georgia and the north arm of the Fraser River, on unceded territory of the xʷməθkʷəy̓əm (Musqueam) First Nation of the Pacific coast of British Columbia, Canada. The 420-ha campus is composed of 58% park-like vegetation, with grass lawns and abundant planted evergreen coniferous trees, deciduous broadleaf trees, shrubs, and native forest remnants dominated by evergreen conifers (Dyck 2016). The campus falls within the Coastal Western Hemlock biogeoclimatic zone (Pojar et al. 1987, 1991, UBC Centre for Forest Conservation Genetics 2021), a temperate rainforest biome with a local mean annual precipitation of 1427 mm and mean annual temperature of 9.3 °C. The University of British Columbia (UBC) is surrounded by 70-130-year-old second-growth forest, including the 763 ha Pacific Spirit Regional Park (http://pacificspiritparksociety.org/about-the-park/), which separates the campus from the City of Vancouver to the east. To place the UBC campus within the continuum of urbanization described by Hager et al. (2017), we used ArcMap 10.3 to derive a minimum convex polygon that enclosed all buildings on campus and created a 5-km buffer around the polygon. We then used the 2010 North American Land Change Monitoring System (NALCMS) database (250 m resolution; Latifovic et al. 2012) to calculate that the landscape within 5-km of our study buildings includes 73% ocean, 12% non-urban cover (forest, wetland, grass, and cropland), and only 15% urban cover.
Bird richness and abundance vary seasonally at UBC. The highest densities occur throughout the avian overwintering period between late October and early March, and during the fall and spring migratory periods (Butler et al. 2021). The majority of altitudinal and latitudinal migrants arrive or move through the region in late August to late October in the fall, and migratory movement in spring peaks between mid-March and the end of May (Campbell et al. 1997, 2001, WildResearch 2015).
We chose 8 study buildings using a randomized design stratified by building height and percent cover of vegetation within 50 m of buildings, following recommendations outlined in Hager et al. (2017). The resulting 8 buildings chosen were between 2 and 12 stories tall (mean = 5) with an average footprint of 772 m² (range: 133-2537 m²). All buildings were constructed of concrete, stone, or brick and glass, with no other reflective surfaces, other than metal window frames. Building names and model covariates can be found in Appendix 1. See De Groot et al. (2021) for additional details on selection methods and characteristics of the buildings.
To assess how façade and vegetation characteristics influence bird-window collisions, we monitored collision mortality by searching for carcasses at all façades (total: 57 façades across 8 buildings; range: 4-15 façades per building). We analyzed data at the façade-level instead of the building-level because of the high variation in vegetation and façade features within buildings (Appendix 1). We defined a façade as the exterior face of a building, generally facing the same direction, with two exceptions. First, we considered several portions of a building footprint as one façade for the purpose of analysis if we could not reliably discern which façade a bird may have struck (e.g., adjacent or parallel sides of a small alcove). Second, a few façades included overhangs or ledges that would catch carcasses from bird-window collisions above, preventing searchers from detecting the collision. In such cases, we considered only the stories or areas of the façade that we were effectively able to monitor for collisions (i.e., not above a ledge or overhang).
Bird collisions data
We conducted bird collision surveys on 155 survey days over 5 45-day sampling periods between 22 January 2015 and 15 March 2017. Surveys were conducted during daylight hours in early afternoon, following the sampling protocol outlined by S. B. Hager and B. J. Cosentino (2014, unpublished manuscript) and Hager et al. (2017). Each survey was conducted by two observers walking in opposite directions, searching for whole and scavenged bird carcasses within 2 m of the study building façades. We performed a full clean-up to remove all carcasses and scavenged remains the day before each 45-day sampling period to ensure we were not counting mortalities that occurred outside of our seasonal sampling periods. We performed surveys daily for 21 consecutive days, followed by surveys every second day in the subsequent 24 days, removing evidence during each survey to avoid double counting collisions in subsequent days within each sampling period. Sampling effort was consistent across all seasonal sampling periods, with the exception of winter 2015, during which we performed surveys every second day across the entire 45-day sampling period. A heavy snowfall mid-way through the winter 2017 sampling period prevented us from reliably finding carcasses, therefore surveys were halted for several days, an additional clean-up was conducted, and the collision survey period resumed the following day and was extended accordingly to ensure constant search effort relative to fall, spring, and summer. We conducted a total of 1881 façade-level collision surveys in September-October 2015, covering the peak fall migratory period. We conducted 3192 façade-level collision surveys in January-March 2015 and 2017, sampling a portion of the overwinter period between November and mid-March, when resident birds are joined by overwintering latitudinal and altitudinal migrants. We conducted 1881 façade-level collision surveys in April-May 2016, covering peak spring migration. We conducted 1881 façade-level collision surveys in June-July 2016 (hereafter "summer"). Bias-corrected mortality estimates from these surveys were presented in De Groot et al. (2021).
Previously, we showed that searcher efficiency was constant across seasons, whereas carcass persistence varied by season; being the longest in summer (median 6.54 days) and shortest in the fall (0.81 days; De Groot et al. 2021). Searcher efficiency and carcass persistence trials were conducted to estimate season-specific bias-corrected estimates of collision mortality in De Groot et al. (2021), but we did not have sufficient data on searcher efficiency and carcass persistence to calculate bias-corrected estimates of collision mortality at the façade-level. We rarely found more than 1 carcass during the same façade survey (2 carcasses were detected on 10 occasions at 3 buildings), so in the present study we considered the presence (1) or absence (0) of a bird carcass as our response variable for each survey at each façade and conducted separate analyses by season to avoid confounding effects of seasonal variation in carcass persistence.
Characteristics of building façades and surrounding vegetation
We calculated façade area (FacArea, m²) using a map layer of all buildings on the UBC campus as of January 2015 (UBC 2015) to obtain façade lengths (m), and the associated attribute table to determine façade heights (m; excluding areas above ledges or overhangs that would catch carcasses, as described above). To obtain an estimate of total glass area for each façade (AreaGlass, m²), we estimated the percentage of glass area at each façade and then multiplied these estimates by FacArea.
We used a combination of ArcGIS Pro 2.6.0 (ESRI 2015) and Google Earth Pro 220.127.116.1176 historical imagery (Google 2015, 2017) to digitize the 8 UBC buildings, outline their 50-m buffers, and calculate vegetation cover and porous surface cover (VegPerc: ground and shrub vegetation, open soil, bark mulch) vs. impervious ground cover for each building. Photos were used to ensure accuracy and guide digitization under overhangs and tree canopy. Then, for each building façade, we calculated the proportion of the 50-m buffer area with tree canopy cover (TreeCov) using the land cover classification dataset associated with the Ministry of Forests, Lands, Natural Resources Operations and Rural Development Land Cover Atlas for Urban and Rural BC - Metro Vancouver to Hope Summary Report (Government of British Columbia 2022, Fraser Valley Regional District and Metro Vancouver Regional District Land Cover Classification, unpublished data, BC Ministry of Forests, Lands, and Natural Resource Operations 2022, Land cover atlas for urban and rural BC - Metro Vancouver to Hope summary, unpublished report).
Given that reflection of vegetation in glass is considered one of the main causes of bird-window collisions (Klem 1989, Gelb and Delacretaz 2009), and that collisions increase with a greater proportion of glass and with vegetation height (Klem et al. 2009), we quantified the vertical extent of reflection of vegetation at each façade by counting the number of stories with glass on a given façade for which there was adjacent unobstructed vegetation (ground cover, shrubs, or tree canopy) at the corresponding building level. Barriers that could obscure reflections included adjacent buildings or other solid structures. In most cases, we considered vegetation within 50 m of façades, however, for a few façades with no adjacent buildings or nearby obstructions, e.g., at the periphery of campus or large gardens, we included any vegetation up to 100 m away that could reflect in glass of our study building façades. For simplicity, hereafter we refer to this variable as “number of stories reflecting vegetation” (StorReflVeg). It was not appropriate to simply measure vegetation height to capture the height of reflections because many trees are taller than study buildings, some façades had tree canopy but no ground vegetation or shrubs, and some entire stories of individual façades contained no glass. We also chose not to use image processing software, such as ImageJ (Schindelin et al. 2012), because reflections are highly dependent on the angle of approach to the glass façade.
To assess how bird collisions were influenced by characteristics of building façades and surrounding vegetation, we used a model selection approach with generalized linear mixed-effects models (binomial family, logit link) that predicted the presence (1) or absence (0) of bird carcasses (Table 1). We built separate models for each season because carcass persistence varies by season (De Groot et al. 2021). The separate models also allowed us to assess seasonal variation in drivers of collision risk without examining statistical interactions between each potential driver and season. The full model included all building and vegetation covariates, and an interaction between façade area and area of glass. We included this interaction to test the prediction that façade area would influence the positive effect of glass area on collision rates, or, that a percent of the façade area with glass would increase collision rates, as shown in a similar study at Duke University campus, in North Carolina, USA (Ocampo-Peñuela et al. 2016). The “building only” and “vegetation only” models were subsets of the full model and included, respectively, only the building covariates and the vegetation covariates. We nested façade identity within building identity as a random effect in all models to account for the hierarchical error structure due to multiple measurements within façades and buildings. To account for variation in exposure time (i.e., the time between surveys, during which carcasses could accumulate below a façade), we used the log of the number of hours since the previous survey as an offset for all models. We also fit a “null model,” which included only the intercept, random effects, and the offset (Table 1). Before performing the analyses, all predictor variables were scaled to have a mean of 0 and a standard deviation of 1.
After ensuring that covariates were not strongly correlated (Appendix 1, Fig. A1.1), we fit models using maximum likelihood estimation within the R package glmmTMB (Brooks et al. 2017). We compared the four models within each season using Akaike’s Information Criterion corrected for small sample sizes (AICc). We considered models to be plausible if deltaAICc values were within 2 AICc of the lowest AICc model (Burnham and Anderson 2002). To understand the change in odds of bird collisions per unit increase in building and vegetation covariates, we calculated the odds ratios for each covariate of the best-ranked model for each season. The odds ratio corresponds to the exponential function of the estimated effect (b) of each parameter. Covariates were interpreted as affecting bird collisions when the 95% confidence intervals on their odds ratios did not overlap 1. We conducted a residual analysis, a zero-inflation, and dispersion tests of the best-ranked model in each season to verify that models fit the data using the DHARMa R package (Hartig 2022).
Across our 57 façades, FacArea had a mean of 342 m² (range: 46-1101); AreaGlass had a mean of 50 m² (range: 1-354); VegPerc (permeable surface) had a mean of 41% (range: 0-93%); TreeCov had a mean of 22% (range: 0-82%); and StorReflVeg had a mean of 2.2 (range: 0-8).
We detected 152 bird-window collisions from 23 species during 142 of our 8835 façade surveys over the 5 seasonal sampling periods. Thrushes (particularly Varied Thrush, Ixoreus naevius) and sparrows were the most common groups of species detected in mortality counts in spring, fall, and winter, and kinglets (Regulidae) comprised 24% of total mortalities in fall at our study buildings (see De Groot et al. 2021 for a detailed account of collision vulnerability in winter, and mortality by species across seasons). Raw collision rate, i.e., the percent of surveys during which a collision was detected, was 1.63% overall, but varied > 10-fold across study buildings, ranging from 0.35% to 3.85%. The best-ranked model and influential explanatory variables were different for each season.
For the fall migration period, the building-only model received the most support from the data (Table 2). Area of glass and number of stories reflecting vegetation had positive and significant effects on the occurrence of bird collisions (Table 3). The odds of a collision in the fall migration period increased by a factor of 2.56 (odds ratio for AreaGlass; Table 3 and Appendix 2, Fig. A2.1) for each increase of 58 m² of glass (calculated by back-scaling from the scaled predictor used in the model). The odds of a collision in the fall migration period also increased by a factor of 1.92 for each increase of 2.6 in the number of stories reflecting vegetation (Table 3 and Appendix 2, Fig. A2.1).
For the overwinter period, the building only and the full models received the most support from the data (Table 2). Among the covariates included in these models, area of glass and façade area had positive effects in both models (Table 3). The odds of a collision in the overwinter period increased by a factor of 1.47 (for the building-only model) or 1.41 (for the full model) for each increase of 58 m² of glass, and by a factor of 1.57 (building) or 1.48 (full) for each 235 m² increase in façade area (Table 3, Fig. 1).
During the spring migration and breeding period, the building only and the full models received the most support from the data (Table 2). In both models, only the area of glass had a positive effect on collisions. The odds of a bird collision increased by a factor of 2.30 (building-only model) or 1.99 (full model) for each 58 m² increase in area of glass (Table 3, Fig. 1).
For the summer post-breeding period, the null and the building-only models received the most support from the data (Table 2). Area of glass showed a positive effect in the building-only model. The odds of a collision increased by a factor of 1.85 for each 58 m² increase in area of glass (Table 3).
De Groot et al. (2021) previously demonstrated strong seasonal variation in the rates of bird-window collisions at UBC; here, we showed that season also influenced how building and vegetation characteristics affected bird-window collision mortality. Collisions increased with the area of glass on building façades across all seasons, but the effect was largest during the fall when collision rates are highest at UBC, and negligible in summer when collision rates are very low (De Groot et al. 2021). Additionally, collisions increased with the number of stories reflecting vegetation during fall and with total façade area in the overwintering period. Other than number of stories reflecting vegetation, which is a variable that integrates the effect of height of glass and corresponding adjacent vertical vegetation, none of the vegetation-related variables strongly predicted collision risk at our study façades.
Our result that collisions increased with the area of glass is consistent with findings from most other studies (Klem et al. 2009, Hager et al. 2013, Cusa et al. 2015, Loss et al. 2019, Elmore et al. 2021a) and the fact that this was the only factor that consistently increased collision risk across seasons, suggests that it is a primary driver of collision mortality at our study site. Several studies have confirmed that the proportion of glass reflecting vegetation is a good predictor of bird collisions (e.g., Klem et al. 2009, Gelb and Delacretaz 2009, Borden et al. 2010), and Rebolo-Ifrán et al. (2019) found that vegetation reflection was one of the main drivers of bird-window collisions in Argentina, doubling the annual collision rate when compared to windows that lacked reflection of vegetation or other attractors. Klem (2009) and Rebolo-Ifrán et al. (2019) also found that increasing vegetation height can also increase collision risk, but to our knowledge, no other study has attempted to quantify the combined effect on collision risk of the vertical extent of vegetation (ground cover to tree canopy) with the vertical extent of glass on façades. Our result that collision mortality increased with the number of stories reflecting vegetation in fall may be explained by migratory stopovers of a diverse community of bird species that forage across a broad vertical stratum of the tree canopy (including kinglets that comprised 24% of collision mortalities), and shrub and ground feeding birds (that accounted for 50% of detected collisions; De Groot 2021) during the fall period at UBC campus. De Groot et al. (2021) only collected species abundance data in winter and therefore were not able to calculate relative species vulnerability to collisions in other seasons. Future work should examine the relationship between the physical factors that increase collision risk and species’ vulnerabilities across seasons to better understand how seasonal variation in species’ abundance and habitat use contribute to collision risk.
As we expected, façade area had a positive effect on collisions, but we only observed this effect in the overwintering period. Façade area is expected to increase the occurrence of collisions by restricting the flight path and potentially funneling birds toward the glass (Klem et al. 2009). Previous studies have shown that building area can increase collision mortality, particularly in areas of low urbanization (Hager et al. 2017). Many of the species most vulnerable to collisions in winter at our study site (De Groot et al. 2021) are highly frugivorous in winter (e.g., Varied Thrush; George 2020, American Robin; Vanderhoff et al. 2020). During winter, these birds’ ability to navigate around obstacles may be impaired due to intoxication from foraging on fermented fruits (Fitzgerald et al. 1990, Kinde et al. 2012, De Groot et al. 2021). The collision mortality rate in winter was high at UBC and equivalent to the collision rate during the spring migratory period (De Groot et al. 2021). Because most overwintering bird species are present in the area for up to five months, risk factors affecting this bird community may result in cumulatively more collision deaths over the entire winter compared to most other seasons. Although there is growing anecdotal evidence that birds are vulnerable to collisions with glass after consuming fermented fruits in winter (Brown et al. 2019, De Groot et al. 2021), the effect of fruiting trees and shrubs on collision risk should be tested empirically for overwintering birds in temperate climates.
We predicted, and several studies showed, that bird-window collision risk increases with the amount of vegetation near buildings (Klem et al. 2009, Gelb and Delacretaz 2009, Borden et al. 2010, Gómez-Martínez 2019, Rebolo-Ifrán et al. 2019); however, area of tree cover, ground and shrub vegetation, soil, and other porous surfaces within 50-m buffers were not good predictors of collision risk in our models. There are important reasons for caution when interpreting these results. First, the factors influencing bird-window collision risk can vary depending on the spatial scale of the analysis (Hager et al. 2013, Loss et al. 2019), with collision mortality tending to increase at buildings within less urbanized landscapes (low urbanization within ~5 km radius; Hager et al. 2017). Our study only examined variation in collision risk relative to vegetation variables at the smallest scale (within 50 m) adjacent to façades. However, at a broader scale, only 15% of the landscape within a 5-km radius of our study buildings is urbanized, with lush remnant native and planted deciduous and evergreen vegetation cover on campus, adjacent to forest and ocean. These features can attract high densities of birds, particularly during the migratory and overwintering months. Therefore, vegetation variables at greater scales may be more influential in affecting collision risk at our study site. Although such context is not the norm for all urban areas, it is common to many university campuses, “green” cities, national parks, resorts, and similar recreational areas in which buildings may have a disproportionate and underappreciated effect on wild birds. Second, given our finding that number of stories reflecting vegetation increased collision risk in the fall, the vertical extent of vegetation reflection may be more relevant to collision risk in our context than the horizontal extent of vegetation surrounding building façades. Third, it is possible that our measurements still did not capture the most biologically relevant information about the vegetation surrounding buildings. Because much of the tree canopy at UBC is above the height of buildings, the upper canopy of trees (e.g., where we observed large flocks of Pine Siskins, Spinus pinus, foraging in winter) may occur above the range of collision risk.
It is important to recognize a few additional caveats to our results. First, similarly to other studies that have examined the effects of vegetation and building features on collision mortality (Klem et al. 2009, Gelb and Delacretaz 2009, Borden et al. 2010, Hager et al. 2017), we could not account for removal of carcasses by scavengers or detectability of carcasses by observers, which may have especially biased against finding carcasses at façades surrounded by thick ground cover or shrubs, or in wet conditions. Although we conducted carcass persistence and searcher efficiency trials to correct mortality estimates by season (reported in De Groot et al. 2021), these trials were not conducted at an appropriate scale to correct mortality estimates at the façade level. Further, our study cannot account for risk factors causing collisions that were not detected because they were not immediately fatal and/or had sub-lethal effects (Hager et al. 2013, Samuels et al. 2022). Second, our models did not include all factors that can independently affect collision risk and/or interact with building and vegetation characteristics to influence collision probability. Additional factors include façade shape (e.g., alcoves, corners; Riding et al. 2020); species differences in behavior, foraging guild, as well as sex and age of individuals (Loss et al. 2019, Winger et al. 2019, Riding et al. 2020, Adad Fornazari 2021, Colling et al. 2022); weather (Lao et al. 2023); migratory traffic rate (Elmore et al. 2021b); and anthropogenic light (Lao et al. 2020, Van Doren et al. 2021). As with the building and vegetation variables that we studied, the degree to which these factors influence collision mortality likely varies seasonally. For example, anthropogenic light can attract and disorient nocturnally migrating birds and increase risk of daytime and nocturnal collisions in fall and spring (La Sorte et al. 2017, Van Doren et al. 2017, 2021, Lao et al. 2020, Adams et al. 2021). However, the effects of light on collision risk for diurnal species outside of the migratory period is presumably lower compared to the risks of daytime collisions due to clear and reflective glass. Future research should test assumptions and differentiate among the relative threats associated with daytime and nocturnal collisions across all seasons and across a range of geographical regions. In addition, understanding species- and age-specific effects is particularly important for assessing the possible implications of alternative building and landscape proposals across the full annual cycle for steeply declining birds and species-at-risk.
Cities are increasingly encouraged to use tree planting and other Nature-based solutions to increase climate resiliency, improve human well-being, and benefit biodiversity (e.g., Frantzeskaki et al. 2019). However, during the period of highest daily collision mortality (fall migration) at our study site, collision risk increased with the vertical extent of reflections of surrounding vegetation. To ensure that tree planting does not lead to increased collision mortality, bird-friendly building design and collision mitigation must be adopted alongside urban greening efforts. Our results also support recent changes to bird-friendly design guidelines that increase the height requirement of bird-friendly mitigation on façades to at least the height of adjacent mature tree canopy (Canadian Standards Association 2019).
Although our results corroborate evidence that façade size and area of glass positively increase collision mortality, we caution that conservationists should continue to advocate for bird-friendly design in smaller buildings such as houses. Small buildings cumulatively account for more collisions, because of their relative abundance on the landscape, compared to larger building classes (Machtans et al. 2013, Loss et al. 2014). Future building design should employ synergistic, cost-effective architectural design features that address multiple objectives and benefits for birds and humans. For example, reducing glass area on building facades, and other design elements such as external vertical sunshades, grills, or perforated metal or wooden screens, may reduce bird collision risk by reducing the threats posed by clear and reflective glass and by reducing light emission from structures (City of Toronto 2016, Canadian Standards Association 2019). These same design features can also provide occupant privacy, improve energy efficiency, and reduce building overheating in a warming world (BC Housing Research Centre 2019).
For existing buildings, many bird-friendly mitigation strategies are available to reduce collision risk, including hanging closely spaced cords in front of windows, or applying adhesive markers, films, or artistic designs in a dense pattern to the outside surface of glass (Brown et al. 2019, 2020, 2021, De Groot et al. 2022, Riggs 2022, American Bird Conservancy 2023, FLAP Canada 2023). Considering the highest collision rates during the short fall migratory period and also the high collision rates during the prolonged overwintering period, our results underline the importance of measuring building and vegetation effects across the most lethal and lengthy stages of the life cycle for birds, both to predict the impact of proposed buildings and for guiding mitigation strategies and priorities that will result in the greatest conservation benefits.
RESPONSES TO THIS ARTICLE
Responses to this article are invited. If accepted for publication, your response will be hyperlinked to the article. To submit a response, follow this link. To read responses already accepted, follow this link.
Viviane Zulian: data analysis and visualization; writing: original draft, review, and editing; Andrea R. Norris: conceptualization, data curation, methodology, data analysis; writing: original draft, review, and editing; Kristina L. Cockle: data analysis and visualization; writing: original draft, review, and editing; Alison N. Porter: conceptualization, methodology, investigation, data curation, project administration; Lauryn G. Do: data curation, data analysis; Krista L. De Groot: conceptualization, methodology, investigation, supervision, project administration, funding acquisition; writing: original draft, review, and editing.
We thank Vanessa Cheung (UBC SEEDS), Ruth Fitzell, Nicholas Froese, Alessandra Gentile (UBC SEEDS), Andrew Huang, Kat McGrath, Dominique Melançon, Sarah Parker, Cadi Schiffer, and Sally Taylor for assisting the authors in collecting field data. Kathleen Moore provided additional GIS expertise and advice on data layers. We thank Penny Martyn (UBC Campus and Community Planning) for her tireless support of bird collision prevention and mitigation on campus. Liska Richer supported data collection during the winter 2017 field season through the UBC SEEDS program, and Sally Otto (Biodiversity Research Centre) supervised SEEDS students. Ildiko Szabo (Beaty Biodiversity Museum) provided additional support through provision of and assistance with identification of carcasses. Funding and staff support for data collection, project and volunteer coordination, and GIS analysis was provided by Environment and Climate Change Canada. We wish to acknowledge that the UBC Point Grey (Vancouver) campus occupies unceded territory of the xʷməθkʷəy̓əm (Musqueam) First Nation.
Our data and code are included in the paper's appendices.
Adad Fornazari, G., A. Saldanha, R. Ribas Lange, T. Froes, D. Klem, Jr., B. A. Moore, and F. Montiani-Ferreira. 2021. Window collisions by birds in Brazil: epidemiologic factors and radiographic and necropsy assessments. Journal of Avian Medicine and Surgery 35:313-324. https://doi.org/10.1647/20-00009
Adams, C. A., E. Fernández-Juricic, E. M. Bayne, and C. Cassady St. Clair. 2021. Effects of artificial light on bird movement and distribution: a systematic map. Environmental Evidence 10:37. https://doi.org/10.1186/s13750-021-00246-8
Altshuler, D. L., and M. V. Srinivasan. 2018. Comparison of visually guided flight in insects and birds. Frontiers in Neuroscience 12:157. https://doi.org/10.3389/fnins.2018.00157
American Bird Conservancy. 2023. Bird-strike prevention: how to stop birds from hitting windows. American Bird Conservancy, The Plains, Virginia, USA. https://abcbirds.org/glass-collisions/stop-birds-hitting-windows/
Bayne, E. M., C. A. Scobie, and M. Rawson-Clark. 2012. Factors influencing the annual risk of bird-window collisions at residential structures in Alberta, Canada. Wildlife Research 39:583-592. https://doi.org/10.1071/WR11179
BC Housing Research Centre. 2019. BC energy step code: design guide supplement S3 on overheating and air quality. BC Housing Research Centre, Burnaby, British Columbia, Canada. https://www.bchousing.org/publications/BC-Energy-Step-Code-Guide-Supplemental.pdf
Beatley, T. 2020. The bird-friendly city. Creating safe urban habitats. Island, Washington, D.C., USA.
Borden, W. C., O. M. Lockhart, A. W. Jones, and M. S. Lyons. 2010. Seasonal, taxonomic, and local habitat components of bird-window collisions on an urban university campus in Cleveland, OH. Ohio Journal of Science 110:10. https://kb.osu.edu/bitstream/handle/1811/52787/OJS110N3_44.pdf?isAllowed=y&sequence=1
Brooks, M. E., K. Kristensen, K. J. van Benthem, A. Magnusson, C. W. Berg, A. Nielsen, H. J. Skaug, M. Mächler, and B. M. Bolker. 2017. glmmTMB balances speed and flexibility among packages for zero-inflated generalized linear mixed modeling. R Journal 9:378-400. https://doi.org/10.32614/RJ-2017-066
Brown, B. B., L. Hunter, and S. Santos. 2020. Bird-window collisions: different fall and winter risk and protective factors. PeerJ 8:e9401. https://doi.org/10.7717/peerj.9401
Brown, B. B., E. Kusakabe, A. Antonopoulos, S. Siddoway, and L. Thompson. 2019. Winter bird-window collisions: mitigation success, risk factors, and implementation challenges. PeerJ 7:e7620. https://doi.org/10.7717/peerj.7620
Brown, B. B., S. Santos, and N. Ocampo-Peñuela. 2021. Bird-window collisions: mitigation efficacy and risk factors across two years. PeerJ 9:e11867. https://doi.org/10.7717/peerj.11867
Burnham, K. P., and D. R. Anderson. 2002. Model selection and multimodel inference: a practical information-theoretic approach. Second edition. Springer, New York, New York, USA.
Butler, R. W., D. W. Bradley, and J. Casey. 2021. The status, ecology, and conservation of internationally important bird populations on the Fraser River Delta, British Columbia, Canada. British Columbia Birds 32:1-52. https://bcbirds.bcfo.ca/wp-content/uploads/2021/04/bc-birds-volume-32-fraser-complete-web.pdf
Campbell, R. W., N. K. Dawe, I. McTaggart-Cowan, J. M. Cooper, G. W. Kaiser, M. C. E. McNall, and G. E. J. Smith. 1997. The birds of British Columbia. Vol. 3. Passerines: flycatchers through vireos. UBC Press, Vancouver, British Columbia, Canada.
Campbell, R. W., N. K. Dawe, I. McTaggart-Cowan, J. M. Cooper, G. W. Kaiser, A. C. Stewart, and M. C. E. McNall. 2001. The birds of British Columbia. Vol. 4. Passerines: wood-warblers through old world sparrows. UBC Press, Vancouver, British Columbia, Canada.
Canadian Standards Association. 2019. Bird-friendly building design. Publisher location https://www.csagroup.org/store/product/CSA%20A460:19/
City of Toronto. 2016. Bird-friendly guidelines. City of Toronto, Toronto, Ontario, Canada. https://www.toronto.ca/city-government/planning-development/official-plan-guidelines/design-guidelines/bird-friendly-guidelines/
Colling, O. M., C. G. Guglielmo, S. J. Bonner, and Y. E. Morbey. 2022. Migratory songbirds and urban window collision mortality: vulnerability depends on species, diel timing of migration, and age class. Avian Conservation and Ecology 17(1):22. https://doi.org/10.5751/ACE-02107-17012
Cusa, M., D. A. Jackson, and M. Mesure. 2015. Window collisions by migratory bird species: urban geographical patterns and habitat associations. Urban Ecosystems 18:1427-1446. https://doi.org/10.1007/s11252-015-0459-3
De Groot, K. L., A. N. Porter, A. R. Norris, A. C. Huang, and R. Joy. 2021. Year-round monitoring at a Pacific coastal campus reveals similar winter and spring collision mortality and high vulnerability of the Varied Thrush. Ornithological Applications 123:duab027. https://doi.org/10.1093/ornithapp/duab027
De Groot, K. L., A. G. Wilson, R. McKibbin, S. A. Hudson, K. M. Dohms, A. R. Norris, A. C. Huang, I. B. J. Whitehorne, K. T. Fort, C. Roy, J. Bourque, and S. Wilson. 2022. Bird protection treatments reduce bird-window collision risk at low-rise buildings within a Pacific coastal protected area. PeerJ 10:e13142. https://doi.org/10.7717/peerj.13142
Dunn, E. H. 1993. Bird mortality from striking residential windows in winter. Journal of Field Ornithology 64:302-309. https://sora.unm.edu/sites/default/files/journals/jfo/v064n03/p0302-p0309.pdf
Dyck, C. 2016. Mapping the biodiversity potential on the University of British Columbia Campus. University of British Columbia, Vancouver, British Columbia, Canada. https://www.sustain.ubc.ca/sites/default/files/seedslibrary/Habitat%20Mapping%20Report_Caylee%20Dyck_Apr%207_0.pdf
Elmore, J. A., S. B. Hager, B. J. Cosentino, T. J. O’Connell, C. S. Riding, M. L. Anderson, M. H. Bakermans, T. J. Boves, D. Brandes, E. M. Butler, M. W. Butler, N. L. Cagle, R. Calderón-Parra, A. P. Capparella, A. Chen, K. Cipollini, A. A. T. Conkey, T. A. Contreras, R. I. Cooper, C. E. Corbin, R. L. Curry, J. J. Dosch, K. L. Dyson, E. E. Fraser, R. A. Furbush, N. D. G. Hagemeyer, K. N. Hopfensperger, D. Klem, Jr., E. A. Lago, A. S. Lahey, C. S. Machtans, J. M. Madosky, T. J. Maness, K. J. McKay, S. B. Menke, N. Ocampo-Peñuela, R. Ortega-Álvarez, A. L. Pitt, A. Puga-Caballero, J. E. Quinn, A. M. Roth, R. T. Schmitz, J. L. Schnurr, M. E. Simmons, A. D. Smith, C. W. Varian-Ramos, E. L. Walters, L. A. Walters, J. T. Weir, K. Winnett-Murray, I. Zuria, J. Vigliotti, and S. R. Loss. 2021a. Correlates of bird collisions with buildings across three North American countries. Conservation Biology 35:654-665. https://doi.org/10.1111/cobi.13569
Elmore, J. A., C. S. Riding, K. G. Horton, T. J. O’Connell, A. Farnsworth, and S. R. Loss. 2021b. Predicting bird-window collisions with weather radar. Journal of Applied Ecology 58:1593-1601. https://doi.org/10.1111/1365-2664.13832
Environmental Systems Research Institute (ESRI). 2015. ArcGIS desktop: release 10.3.1. Environmental Systems Research Institute, Redlands, California, USA.
Fitzgerald, S. D., J. M. Sullivan, and R. J. Everson. 1990. Suspected ethanol toxicosis in two wild Cedar Waxwings. Avian Diseases 34:488-490. https://doi.org/10.2307/1591442
Fatal Light Awareness Program (FLAP) Canada. 2023. BirdSafe©. FLAP Canada, Toronto, Ontario, Canada. https://birdsafe.ca/
Frantzeskaki, N., T. McPhearson, M. J. Collier, D. Kendal, H. Bulkeley, A. Dumitru, C. Walsh, K. Noble, E. van Wyk, C. Ordóñez, C. Oke, and L. Pintér. 2019. Nature-based solutions for urban climate change adaptation: linking science, policy, and practice communities for evidence-based decision-making. BioScience 69:455-466. https://doi.org/10.1093/biosci/biz042
Gómez-Martínez, M. A., D. Klem, Jr., O Rojas-Soto, F. Gonzáles-García, and I. MacGregor-Fors. 2019. Window-strikes: bird collisions in a Neotropical green city. Urban Ecosystems 22:699-708. https://doi.org/10.1007/s11252-019-00858-6
Google. 2015, 2017. Google Earth Pro 18.104.22.16876. Google, Mountain View, California, USA. https://www.google.com/earth/about/versions/
Gelb, Y., and N. Delacretaz. 2009. Windows and vegetation: primary factors in Manhattan bird collisions. Northeastern Naturalist 16:455-470. https://doi.org/10.1656/045.016.n312
George, T. L. 2020. Varied Thrush (Ixoreus naevius). Version 1.0. In A. F. Poole and F. B. Gill, editors. Birds of the world. Cornell Lab of Ornithology, Ithaca, New York, USA. https://doi.org/10.2173/bow.varthr.01
Hager, S. B., B. J. Cosentino, K. J. McKay, C. Monson, W. Zuurdeeg, and B. Blevins. 2013. Window area and development drive spatial variation in bird-window collisions in an urban landscape. PLoS ONE 8:e53371. https://doi.org/10.1371/journal.pone.0053371
Hager, S. B., B. J. Cosentino, M. A. Aguilar-Gómez, M. L. Anderson, M. Bakermans, T. J. Boves, D. Brandes, M. W. Butler, E. M. Butler, N. L. Cagle, R. Calderón-Parra, A. P. Capparella, A. Chen, K. Cipollini, A. A. T. Conkey, T. A. Contreras, R. I. Cooper, C. E. Corbin, R. L. Curry, J. J. Dosch, M. G. Drew, K. Dyson, C. Foster, C. D. Francis, E. Fraser, R. Furbush, N. D. G. Hagemeyer, K. N. Hopfensperger, D. Klem, Jr., E. Lago, A. Lahey, K. Lamp, G. Lewis, S. R. Loss, C. S. Machtans, J. Madosky, T. J. Maness, K. J. McKay, S. B. Menke, K. E. Muma, N. Ocampo-Peñuela, T. J. O’Connell, R. Ortega-Álvarez, A. L. Pitt, A. L. Puga-Caballero, J. E. Quinn, C. W. Varian-Ramos, C. S. Riding, A. M. Roth, P. G. Saenger, R. T. Schmitz, J. Schnurr, M. Simmons, A. D. Smith, D. R. Sokoloski, J. Vigliotti, E. L. Walters, L. A. Walters, J. T. Weir, K. Winnett-Murray, J. C. Withey, and I. Zuria. 2017. Continent-wide analysis of how urbanization affects bird-window collision mortality in North America. Biological Conservation 212:209-215. https://doi.org/10.1016/j.biocon.2017.06.014
Hartig, F. 2022. DHARMa: Residual diagnostics for hierarchical (multi-level/mixed) regression models. R package version 0.1.6. R Foundation for Statistical Computing, Vienna, Austria. https://cran.r-project.org/web/packages/DHARMa/vignettes/DHARMa.html
Kelly, J. F., and R. L. Hutto. 2005. An east-west comparison of migration in North American Wood Warblers. Condor 107:197-211. https://doi.org/10.1093/condor/107.2.197
Kinde, H., E. Foate, E. Beeler, F. Uzal, J. Moore, and R. Poppenga. 2012. Strong circumstantial evidence for ethanol toxicosis in Cedar Waxwings (Bombycilla cedrorum). Journal of Ornithology 153:995-998. https://doi.org/10.1007/s10336-012-0858-7
Klem, Jr., D. 1989. Bird-window collisions. Wilson Bulletin 101:606-620. https://sora.unm.edu/sites/default/files/journals/wilson/v101n04/p0606-p0620.pdf
Klem, D., C. J. Farmer, N. Delacretaz, Y. Gelb, and P. G. Saenger. 2009. Architectural and landscape risk factors associated with bird-glass collisions in an urban environment. Wilson Journal of Ornithology 121:126-134. https://doi.org/10.1676/08-068.1
Klem, D., D. C. Keck, K. L. Marty, A. J. Miller Ball, E. E. Niciu, and C. T. Platt. 2004. Effects of window angling, feeder placement, and scavengers on avian mortality at plate glass. Wilson Bulletin 116:69-73. https://doi.org/10.1676/0043-5643(2004)116[0069:EOWAFP]2.0.CO;2
Kuchler, A. W. 1964. Potential natural vegetation of the conterminous United States. Special Publication No. 36. American Geographical Society, New York, New York, USA.
Lao, S., A. W. Anderson, R. B. Blair, J. W. Eckles, R. J. Turner, and S. R. Loss. 2023. Bird-building collisions increase with weather conditions that favor nocturnal migration and with inclement and changing weather. Ornithological Applications 125:duac045. https://doi.org/10.1093/ornithapp/duac045
Lao, S., B. A. Robertson, A. W. Anderson, R. B. Blair, J. W. Eckles, R. J. Turner, and S. R. Loss. 2020. The influence of artificial light at night and polarized light on bird-building collisions. Biological Conservation 241:108358. https://doi.org/10.1016/j.biocon.2019.108358
La Sorte, F. A., D. Fink, J. J. Buler, A. Farnsworth, and S. A. Cabrera-Cruz. 2017. Seasonal associations with urban light pollution for nocturnally migrating bird populations. Global Change Biology 23:4609-4619. https://doi.org/10.1111/gcb.13792
Latifovic, R., C. Homer, R. Ressl, D. Pouliot, S. N. Hossain, R. R. Colditz, I. Olthof, C. P. Giri, and A. Victoria. 2012. North American land-change monitoring system. Pages 296-305 in C. P. Giri, editor. Remote sensing of land use and land cover: principles and applications. First edition. CRC Press, Boca Raton, Florida, USA. https://doi.org/10.1201/b11964-24
Loss, S. R., S. Lao, J. W. Eckles, A. W. Anderson, R. B. Blair, and R. J. Turner. 2019. Factors influencing bird-building collisions in the downtown area of a major North American city. PLoS ONE 14:e0224164. https://doi.org/10.1371/journal.pone.0224164
Loss, S. R., T. Will, S. S. Loss, and P. P. Marra. 2014. Bird-building collisions in the United States: estimates of annual mortality and species vulnerability. Condor 116:8-23. https://doi.org/10.1650/CONDOR-13-090.1
Machtans, C. S., C. H. R. Wedeles, and E. M. Bayne. 2013. A first estimate for Canada of the number of birds killed by colliding with building windows. Avian Conservation and Ecology 8(2):6. https://doi.org/10.5751/ACE-00568-080206
Musqueam Indian Band. 2023. Place names map. Musqueam Indian Band, Vancouver, British Columbia, Canada. https://www.musqueam.bc.ca/our-story/our-territory/place-names-map/
Nichols, K. S., T. Homayoun, J. Eckles, and R. B. Blair. 2018. Bird-building collision risk: an assessment of the collision risk of birds with buildings by phylogeny and behavior using two citizen-science datasets. PLoS ONE 13:e0201558. https://doi.org/10.1371/journal.pone.0201558
Ocampo-Peñuela, N., R. S. Winton, C. J. Wu, E. Zambello, T. W. Wittig, and N. L. Cagle. 2016. Patterns of bird-window collisions inform mitigation on a university campus. PeerJ 4:e1652. https://doi.org/10.7717/peerj.1652
Pojar, J., K. Klinka, and D. A. Demarchi. 1991. Coastal western hemlock zone. Pages 95-111 in D. V. Meidinger and J. Pojar, editors. Ecosystems of British Columbia. BC Ministry of Forests, Victoria, British Columbia, Canada.
Pojar, J., K. Klinka, and D. V. Meidinger. 1987. Biogeoclimatic ecosystem classification in British Columbia. Forest Ecology and Management 22:119-154. https://doi.org/10.1016/0378-1127(87)90100-9
Rebolo-Ifrán, N., A. di Virgilio, and S. A. Lambertucci. 2019. Drivers of bird-window collisions in southern South America: a two-scale assessment applying citizen science. Scientific Reports 9:18148. https://doi.org/10.1038/s41598-019-54351-3
Riding, C. S., T. J. O’Connell, and S. R. Loss. 2020. Building façade-level correlates of bird-window collisions in a small urban area. Condor 122:duz065. https://doi.org/10.1093/condor/duz065
Riggs, G. J., C. M. Barton, C. S. Riding, T. J. O’Connell, and S. R. Loss. 2022. Field-testing effectiveness of window markers in reducing bird-window collisions. Urban Ecosystems 26:713–723. https://doi.org/10.1007/s11252-022-01304-w
Samuels, B., B. Fenton, E. Fernández-Juricic, and S. A. MacDougall-Shackleton. 2022. Opening the black box of bird-window collisions: passive video recordings in a residential backyard. PeerJ 10:e14604. https://doi.org/10.7717/peerj.14604
Schindelin, J., I. Arganda-Carreras, E. Frise, V. Kaynig, M. Longair, T. Pietzsch, S. Preibisch, C. Rueden, S. Saalfeld, B. Schmid, J.-Y. Tinevez, D. J. White, V. Hartenstein, K. Eliceiri, P. Tomancak, and A. Cardona. 2012. Fiji: an open-source platform for biological-image analysis. Nature Methods 9:676-682. https://doi.org/10.1038/nmeth.2019
Seto, K. C., S. Parnell, and T. Elmqvist. 2013. A global outlook on urbanization. Pages 1-12 in T. Elmqvist, M. Fragkias, J. Goodness, B. Güneralp, P. J. Marcotullio, R. I. McDonald, S. Parnell, M. Schewenius, M. Sendstad, K. C. Seto, and C. Wilkinson, editors. Urbanization, biodiversity and ecosystem services: challenges and opportunities. Springer Dordrecht, The Netherlands. https://doi.org/10.1007/978-94-007-7088-1
University of British Columbia (UBC). 2015. UBC 3D smart map. University of British Columbia, Vancouver, British Columbia, Canada.
UBC Centre for Forest Conservation Genetics. 2021. Subzone/variant climate data. UBC Centre for Forest Conservation Genetics, Vancouver, British Columbia, Canada. https://cfcg.forestry.ubc.ca/resources/cataloguing-in-situ-genetic-resources/subzonevariant-climate-data/
Van Doren, B. M., K. G. Horton, A. M. Dokter, H. Klinck, S. B. Elbin and A. Farnsworth. 2017. High-intensity urban light installation dramatically alters nocturnal bird migration. Proceedings of the National Academy of Sciences 114:11175-11180. https://doi.org/10.1073/pnas.1708574114
Van Doren, B. M., D. E. Willard, M. Hennen, K. G. Horton, E. F. Stuber, D. Sheldon, A. H. Sivakumar, J. Wang, A. Farnsworth, and B. M. Winger. 2021. Drivers of fatal bird collisions in an urban center. Proceedings of the National Academy of Sciences 118:e2101666118. https://doi.org/10.1073/pnas.2101666118
Vanderhoff, N., P. Pyle, M. A. Patten, R. Sallabanks, and F. C. James. 2020. American Robin (Turdus migratorius). Version 1.0. In P. G. Rodewald, editor. Birds of the world. Cornell Lab of Ornithology, Ithaca, New York, USA. https://doi.org/10.2173/bow.amerob.01
WildResearch. 2015. Iona Island bird observatory 2010-2015 cumulative report. WildResearch, Vancouver, British Columbia, Canada. http://wildresearch.ca/wp-content/uploads/2016/04/WildResearch_IIBO_Cumulative_Report_2010-2015_FINAL.pdf
Winger, B. M., B. C. Weeks, A. Farnsworth, A. W. Jones, M. Hennen, and D. E. Willard. 2019. Nocturnal flight-calling behaviour predicts vulnerability to artificial light in migratory birds. Proceedings of the Royal Society B: Biological Sciences 286:20190364. https://doi.org/10.1098/rspb.2019.0364
Wittig, T. W., N. L. Cagle, N. Ocampo-Peñuela, R. S. Winton, E. Zambello, and Z. Lichtneger. 2017. Species traits and local abundance affect bird-window collision frequency. Avian Conservation and Ecology 12(1):17. https://doi.org/10.5751/ACE-01014-120117
Table 1. Models used to assess how bird collisions are influenced by façade-level characteristics of building and vegetation at the University of British Columbia. The response variable “collisions” is binomial and represents the presence (1) or absence (0) of carcasses at each façade during each survey. The full model included all building and vegetation covariates: the façade area (FacArea), the area of glass in each façade (AreaGlass), the percentage of ground and shrub vegetation, soil, and other porous surfaces in a 50-m buffer in front of the façade (VegPer), the percent cover of trees (including evergreen and deciduous; TreeCov), and the number of stories reflecting vegetation (StorReflVeg). We also included an interaction between FacArea and AreaGlass, random effects of the building and façade identity, and an offset that accounts for the time between surveys. The “building only” and the “vegetation only” models are subsets of the full model, which included, respectively, the building- and vegetation-related covariates. The four models were applied for each season separately.
|Full model||Collisions ~ FacArea + AreaGlass + VegPer + TreeCov + StorReflVeg + FacArea * AreaGlass + offset (Meth) + (1| Build:Fac)|
|Building only||Collisions ~ FacArea + AreaGlass + StorReflVeg + FacArea * AreaGlass + offset (Meth) + (1| Build:Fac)|
|Vegetation only||Collisions ~ VegPer + TreeCov + StorReflVeg + offset (Meth) + (1| Build:Fac)|
|Null model||Collisions ~ offset (Meth) + (1| Build:Fac)|
Table 2. Models predicting bird collisions with building façades at the University of British Columbia, compared within seasons and ranked by Akaike information criteria corrected for small samples sizes (AICc). For each model, we additionally indicate degrees of freedom (df), log-likelihood (logLik), the difference in AICc between the given model and the best ranked model (ΔAICc), and the Akaike weight, which indicates the relative support for each model given the data (weight). Fall migration period includes surveys from 9 September to 23 October (timed to cover the period of peak migration); the overwinter period extends from early November until mid-March but surveys occurred 22 January–15 March; the period of spring migration and breeding includes surveys from 15 April to 29 May, timed to coincide with peak migration and the early breeding period; the post-breeding summer period included surveys from 16 June to 30 July.
|Fall migration period (n = 1881)|
|Overwinter period (n = 3192)|
|Spring migration and breeding period (n = 1881)|
|Summer post-breeding period (n = 1881)|
|† Show models within ΔAICc < 2.|
Table 3. Odds ratios with 95% confidence intervals (95% C.I.) for each parameter according to the best ranked models in each season (fall = fall migration period, winter = overwintering period, spring = spring migration and early breeding period, summer = breeding period). Odds ratios correspond to the exponential function of each parameter estimate and show the increase in odds of a detected bird collision occurrence per unit increase in the scaled value of the covariate.
|Season||Model||Covariates||Odds ratio||95% C.I.|
|† Highlights the covariates for which the confidence interval on the odds ratio did not overlap 1.|