DESCRIBING FLOWERING SCHEDULE SHAPE THROUGH MULTIVARIATE ORDINATION

Premise of research.Variation in the overall shape (e.g., skew, modality) of the flowering schedule can affect intra- and interspecific interactions, but full appreciation of these effects is hindered by the difficulty of describing and comparing schedule shapes. We propose a novel approach to schedule description and comparison based on multivariate ordination. Unlike other methods of describing shape, ordination does not require a priori assumptions about the nature of schedule variation and so may detect variation otherwise unseen.Methodology.Using illustrative data to develop the approach, we investigated chord distances and Kolmogorov-Smirnov distances as measures of pairwise differences in schedule shape in principal coordinates analysis. We applied this same technique to an empirical data set and used constrained principal coordinate analysis to determine whether variation in schedule shape in the empirical data set can be explained by other phenological variables.Pivotal results.Principal coordinates analysis identified biologically meaningful variation in schedule shape in both the illustrative and empirical data sets. For both distance measures and both data sets, the first ordination axis arranged individuals by the skew of their flowering schedule, and the second axis separated bimodal schedules from unimodal schedules, particularly at intermediate regions on axis 1. Constrained ordination of the empirical data set revealed that schedule skew tended to vary with flowering duration and timing of flowering onset and that plants producing more flowers tend to be more unimodal in their flowering schedule.Conclusion.Multivariate ordination successfully separated individuals according to the shape of their flowering schedule and identified relationships between phenological variables of potential biological importance. The relationship between onset and skew, for example, could affect the realized strength of phenological assortative mating. Ordination is a departure from function-fitting schedule description. It is broadly applicable to other types of phenological data sets, offering, for example, a new tool for tracking phenological changes over years.