Virtually Tracking Planted Urban Tree Survival with Street-Level Imagery

Abstract Background Street tree plantings are common in urban greening programs, and these trees provide important ecosystem services that increase as trees survive to maturity. Field-based monitoring to understand mortality rates and causes is valuable for urban forest management but very time-consuming. Methods We used street-level imagery to virtually monitor survival for 2,884 street trees over several years postplanting in Philadelphia, Pennsylvania, United States. Results We observed similar mortality rates to other studies, with 7.5% of trees dead or removed by the first summer after planting and the mortality rate dropping to 3.5% between the third and fourth summers postplanting. Logistic regression models were constructed over various time horizons to understand which site, neighborhood, and species characteristics related to survival outcomes. These models showed that higher tree survival was associated with less impervious surface surrounding the tree; lower social vulnerability in the neighborhood; and tree planting in the fall season as opposed to spring. Conclusions Our results point to management activities that could improve survival outcomes, such as planting site enhancements and establishment maintenance, as well as the use of monitoring data to drive decisions regarding planting season. This study demonstrates the value of streetlevel imagery interpretations to provide mortality data on a large number of street trees planted over multiple years.