Predicting Urban Plant Diversity Dynamics

To compare UPD of different types of primary and secondary UFUs, we drew boxplots with AGB and the number of trees, shrubs, herbs, and total UPD in each UFU with the R package “ggplot2” [25 (link)]. We calculated z-scores for all variables using IBM SPSS Statistics 23.0 and removed outliers (z-score greater and lower than 3 and −3, respectively) [26 (link)]. The mean and standard deviation of AGB and the number of trees, shrubs, herbs and total UPD in each UFU were calculated using IBM SPSS Statistics 23 Chicago, USA. Then, we used AGB and UPD of trees, shrubs, and herbs species as dependent variable, and socioeconomic (construction age, housing price, population density), geographical factors (longitude, latitude,) and management factors (maintenance frequency, watering frequency, fertilization frequency) as independent variables. We used stepwise regression analysis and Akaike Information Criterion (AIC) to output the best model information. We first built two models using data from 2015 and 2021. Then, we built a third model to determine if UPD and AGB responses to socioeconomic and management factors had a time lag effect by comparing whether past socioeconomic factors better explain UPD than present socioeconomic factors. We built this model using UPD and AGB data from 2021 and socioeconomic and management factors from 2015.