A spatially explicit modeling analysis of adaptive variation in temperate tree phenology

The geographic applicability of most phenological models is limited because of a lack in accounting for plant genotypic variation over space. This limitation may be partly addressed by quantifying plant adaptation patterns as revealed by common garden/provenance trial research. This study delineated adaptive patterns of a widely distributed tree species in North America—white ash (Fraxinus americana) using multi-year common garden observations of leaf out and leaf senescence phenology. Geographically varied phenology-climate (i.e., phenoclimatic) relationships of tree provenances were investigated both with the aid of interannual temperature variations and using process-based models. Interannual weather fluctuations likely led to varied gradients of spring phenological timing by tree origin latitude as influenced by interactions of chilling and forcing, while the latitudinal gradient of autumn phenology consistently followed a photoperiod-driven pattern. Fitted models revealed latitudinal gradients of chilling requirement (for dormancy release), forcing requirement (for bud break), and critical day length requirement (for leaf senescence) for the tree provenances. When these genotype-specific phenoclimatic relationships were accounted for in spring models, predictions closely matched the latitudinal gradient of USA-National Phenology Network (NPN) observations. On the other hand, average (non-spatial) model predictions of bud break tended to be biased in the species’ northern and southern ranges. This finding shows that introducing genotypic differences to phenological models is necessary for accurate prediction of temperate tree phenology over broad geographic regions.