The cemeteries of Chicago (IL, USA) comprise only a small part of the city’s urban forest (approximately 700 ha in total), but to wildlife they offer unique resources, such as dying trees, that are difficult to find in the surrounding hardscape. The purposes of this study were to characterize this understudied component of Chicago’s urban forest, and to gain insight into how cavity-nesting birds select urban habitat. To approach the latter question, we used Classification and Regression Tree (CART) models to predict each species’ presence or absence based on our local and landscape variables.
We conducted fieldwork in 34 points in 18 of the 22 cemeteries within Chicago’s city limits. Within 50 m of each point, we identified and measured trees, recorded tree cavities, classified trees by life stage (e.g. alive and healthy, declining, dying, or dead), counted deciduous and evergreen shrubs, and counted the numbers of headstones or other monuments ≥ 2 m tall (potential perches for insectivorous birds). We also conducted fixed-radius point counts to identify birds using the cemeteries, and we measured canopy cover, distance from roads, and distance from a body of water using ArcGIS. We used the R package “rpart” to create our CART models.
Results/Conclusions
Median tree basal area from all 34 points was 8.0 m2/ha (min. = 1.9 m2/ha, max. = 19.4 m2/ha), while median density was 25.5 trees/ha (min. = 6.4 trees/ha, max. = 42.0 trees/ha). Silver maples (Acer saccharinum) were found in the greatest proportion of points. Oaks (Quercus spp.) were only found in 13 of the 34 points, yet they dominated the cemetery forest in terms of basal area, comprising nearly 20% of the total. We detected 44 tree species and 42 bird species, including 12 native cavity-nesting species.
The CART models identified important variables in a species’ habitat selection, and also identified potential threshold values for those variables. For example, eastern bluebirds (Sialia sialis) were more likely to be detected at points where there were > 15 headstones or other monuments ≥ 2 m tall in the sampling area, suggesting that the availability of foraging perches improves habitat quality for this species. Downy woodpeckers (Picoides pubescens) were more likely to be detected at points with > 0.06 m2/ha basal area of dead or dying trees. This relatively low threshold could easily be met by the intentional creation of snags from invasive tree species such as Norway maple (Acer platanoides).