Tree growth, post-establishment [defined as a resumption of pre-transplant growth rate (Struve and Joly 1992)], will be determined by measuring caliper of the established 48 research oak (Q. bicolor, Q. rubra) trees, annually for the next four years (preferably longer, but contingent upon funding). Annual increment will be analyzed using piecewise regression. Multiple regression and analysis of covariance will be used to investigate relationships between response variables (annual increment, crown growth) and site conditions (the volume, bulk density, penetrability, nutrients, texture, soil organic matter, and species). Soil testing will be conducted at the Soil Analysis Laboratory at UMass. Mortality will be analyzed using contingency tables and logistic regression. Soil resistance will continue to be measured. Textural analysis is expensive and need only be performed on a limited basis. If root tissue samples will be taken, sampling of lower-order, fine root tissue (2-4mm, width) will be performed to perform hydraulic conductivity measurement, and regression analysis will be performed to determine relationship of root response and nursery production method (B&B, BR, CG, IGF). Local (i.e., Amherst, MA) weather data including precipitation, temperature, and growing degree day accumulation, will be accessed through the Northeast Regional Climate Center at Cornell University for analysis including regression models in relation to tree growth. Crown volume will be measured in accordance with Troxel et al., 2013. Measurements of the caliper and crown of trees, as well as further potential root sampling, will continue to be taken annually, at three intervals during the growing season. Time (labor hours) and costs (equipment, supplies, etc.) associated with tree-related maintenance activities (pruning, tree replacement, watering, fertilizing, etc.) will be recorded and summarized by means, standard deviations, and a five-number summary (minimum, maximum and quartiles). We would first test for mean time differences associated with maintenance practices across the two species of oak trees using a two-sample t-test assuming unequal variances. Assuming no difference across species, we would pool the species data and test for differences across all four nursery production methods. Anderson-Darling tests would be used to assess normality of the maintenance time distributions. We would test for difference across tree types in maintenance time variances and mean maintenance times. Two-sample F-tests would be used to test for differences in variances and two-sample t-tests assuming different variances as well as pooled-data regression analysis would be used to test mean maintenance time differences across treatments. All tests would be completed using a 5% level of significance.
Though measurement of large numbers (i.e., many hundreds or even thousands) of urban trees is often desirable, wise-use of limited resources to examine several dimensions (i.e., growth, effects of nursery production method, root hydraulic conductivity, long-term costs of maintenance) of these locally-grown and established trees has already formed the basis for several research manuscripts (Green et al., 2015; Markarian et al., 2016; Yin et al., 2017; Allen et al. 2017) and one manuscript in-process (Bocsi et al.). Continued measurement and sharing of data/analyses will benefit both the scientific community through further research publication, and local stakeholder audience members (i.e., Town of Amherst, MA Tree Warden Assoc.) who will be regularly updated concerning this work.
