UC Berkeley

Urban forests have aesthetic, environmental, human health, and economic benefits that motivate tree planting programs. Realizing these benefits depends on tree survival. Cost-benefit analyses for urban forest ecosystem services are sensitive to mortality rate assumptions and associated population projections. However, long-term mortality data is needed to assess the accuracy of these assumptions. Analytical tools from demography, such as life tables, mortality curves, and survival analysis, can improve our understanding of urban tree mortality. Demographic approaches have been widely used in forest ecology to quantify population dynamics and project future changes in wildland systems. However, to apply demographic techniques to urban forests, longitudinal data is needed, with repeated mortality observations on individual trees. In this dissertation, I analyzed five years of longitudinal data from two Northern California studies: street trees in Oakland and yard trees in Sacramento. These field projects are complemented by a conceptual overview of demographic approaches to urban tree mortality (Chapter 1), and an investigation of practitioner-based tree monitoring programs.

For the Oakland study (Chapter 2), I documented tree mortality and planting rates, net population growth, and assessed selected risk factors for survival. I monitored the entire street tree population in a small plot for five years after an initial inventory (2006). I adapted the classic demographic balancing equation to quantify annual inputs and outputs to the system, tracking pools of live and standing dead trees. There was a 17.2% net increase in live tree counts during the study period, with 3.7% overall annual mortality. However, population growth was constrained by high mortality of small/young trees. Size-based mortality rates followed a Type III curve, with highest mortality for small trees, and lower for mid-size and large trees. I used multivariate logistic regression to evaluate the relationship between 2011 survival outcomes and inventory data from 2006. Significant associations were found for size class, foliage condition, planting location, and a multiplicative interaction term for size and foliage condition.

For the Sacramento study (Chapter 3), I assessed tree losses during the establishment phase for a residential tree give-away program. A cohort of young trees distributed in 2007 was monitored for five years. I used Random Forests to identify the most important risk factors at different life history stages, and survival analysis to evaluate post-planting survivorship. Analysis included socioeconomic, biophysical, and maintenance characteristics. In addition to field observations of tree planting status, survival, and maintenance, I also collected property ownership information (renter vs. owner-occupancy, homeowner change, and foreclosure) through the Multiple Listing Service and neighborhood socioeconomic characteristics from the U.S. Census. I found that 84.9% of trees were planted, with 70.9% survivorship at five years post-planting. Planting rates were higher in neighborhoods with higher educational attainment, and on owner-occupied properties with stable residential ownership. Five-year survival was also higher for properties with stable homeownership, as well as for tree species with low water use demand. When I incorporated maintenance characteristics from the first year of field observations, factors related to tree care were important to survival. Many residents did not adhere to recommended maintenance practices. These results illustrate the critical role of stewardship and consistent homeownership to young tree mortality on residential properties, and suggest that survival assumptions in urban forest cost-benefit models may be overly optimistic.

To learn more about practitioner-driven monitoring efforts, I surveyed 32 local urban forestry organizations across the United States about the goals, challenges, methods, and uses of their monitoring programs (Chapter 4). Non-profit organizations, municipal agencies, state agencies, and utilities participated. Common goals for monitoring included evaluating the success of tree planting and management, taking a proactive approach towards tree care, and engaging communities. Challenges included limited staff and funding, difficulties with data management and technology, and field crew training. Programs used monitoring results to inform tree planting and maintenance practices, provide feedback to individuals responsible for tree care, and manage hazard trees. Participants emphasized the importance of planning ahead: carefully considering what data to collect, setting clear goals, developing an appropriate database, and planning for funding and staff time. Urban tree monitoring partnerships between researchers and local organizations should be developed, with standardized protocols and clear research questions. Such partnerships would provide urban forestry professionals with improved mortality information to evaluate the success of planting programs, while expanding the data sets available to researchers. The Oakland and Sacramento studies (Chapters 2 and 3) offer examples of demographic approaches to urban tree mortality that can be replicated and expanded as more longitudinal data becomes available from both researchers and practitioners.