In maritime-influenced regions like the Pacific Northwest, the small anthropogenic temperature increase in the last century has been overlaid on significant ENSO-driven fluctuations in temperature. This complex pattern of temperature variation makes it difficult to detect anthropogenic changes in phenology. The current study contributes to the small number of analyses globally of long-duration legacy phenology data sets collected in one place by one person, in this case from 1958-2014. This dataset focuses on urban landscape plants and includes dates for start leaf, start flower, maximum flower, end flower, start fall color and fall defoliation. We tested whether phenology of this urban plant community changed over this time period and which three-month temperature windows were the best predictors of phenology. Ours is the first study we know of to apply ordination analysis to a matrix of phenology data, with years characterized by phenology of 155 phenological events per species. In univariate analyses, we also examined 743 species and tested for acceleration of one or more phenological events by spring warming and delay by lack of winter chilling using multiple regression.
Ordination of forty-two years in terms of the 155 species/event combinations with the most complete records revealed two important ways in which phenology of the urban landscape community varied. The first axis differentiated years with late start-flower and start-leaf dates from years with early phenology; early years tending to occur early in the study period and late-phenology years more recently. The second axis differentiated El Nino and La Nina years and was driven by maximum flowering dates. In contrast, ordination of almost 2000 species/event combinations over a shorter 18 year timespan revealed only El Nino/La Nina effects and not a trend over time. Number of growing degree days in fall/winter and spring three-month windows explained on average 50% of the variation in spring phenology. Seventy eight percent of models for spring or summer events included a negative effect of spring temperature and a positive effect of fall-winter temperature, suggesting the possibility that delayed flowering due to lack of winter chilling is common in these ornamental species. Distributions of change in phenology for individual species over time centered on zero; start of fall color stood out as the only event that occurred earlier over time for almost all species.