Our MCE analysis highlights regions in which social benefits, in terms of health and recreation, can be optimized through planting cherry blossom trees – regions that we believe should be prioritized in municipal planting decisions. The rationale behind doing so is to maximize the effects of urban trees in improving general well-being, encouraging recreation, and fostering social cohesion (Ulmer et al., 2016).

Uncertainties

Our statistical analysis, intended to gauge relationships between cherry blossom trees and social indicators, showed no correlations between cherry tree density and neither property value, income, nor population density. However, a number of external factors influencing these social variables, such as historical context, infrastructure, and other types of trees, could have equivocated meaningful analysis of the variables we isolated. These externalities could perhaps explain our observed absence of correlation, particularly between property value and tree density – an unexpected result, since increased property value is a benefit to which urban trees are often ascribed (McPherson & Schroeder, 1992).

Given the lack of correlation between cherry blossom density and property value and population density, we did not incorporate these factors into our MCE analysis. Although income was similarly uncorrelated with tree density, we nevertheless integrated it into our analysis because income is widely acknowledged as a social determinant of health (WHO, 2016). However, the strength of this assumption is uncertain due to the amount of additional, albeit related, variables such as race, gender, and education level, that influence health. Since urban trees have been found to improve general well-being, we identified low-income, hence presumably unhealthy, areas as  regions that could benefit from planting. Because we also wanted to take population density and tree density into consideration, we conducted a grouping analysis to identify Census Tract neighbourhoods characterized by low income, low tree density, and medium population density to be used as one of our MCE analysis criteria. Our use of Census Tracts, which encompass large parcels of land, may have obscured necessary variability within these parcels. Additionally, since we only accounted for the density of cherry blossom trees, rather than total trees, we also do not know if the planting areas we identified already have a high density of other types of trees that would otherwise provide social benefits.

Our other criteria for planting included healthcare centres, on the basis of the health benefits of trees, as well as greenways and cultural landmarks – tourism areas to which cherry blossoms could add recreational value. Since we do not know how the City of Vancouver defined these areas, there may be additional regions informally considered as healthcare (i.e. addiction centres) or recreational regions (i.e. makeshift skateboard parks) that are not recognized on our map.

The degree of location accuracy for our map features is unknown because the City of Vancouver does not provide uncertainty for their longitude/latitude coordinates, and in the case of street tree locations, for which only addresses were provided, additional uncertainty was introduced through the process of geocoding address data, which generated strange outliers.

Areas of Further Research

Because our research did not take other types of ornamental trees into account due to time constraints, further research could explore the social benefits of a comprehensive range of ornamental trees. Broader still, types of trees could be expanded to also include fruit provisioning, ecosystem service (i.e. stormwater filtration, air pollution reduction), and wildlife habitat utility, using MCE to determine where to plant trees in a dynamic city with differential, at times competing, arboreal needs. A weighted MCE analysis could be used to prioritize tree function objectives, although such prioritization should take community values into careful consideration. Since tree roots can be problematic for roads or underground structures such as pipes, proximity to these structures should also be assessed, with an appropriate buffer applied.