Modeling daily flowering probabilities: expected impact of climate change on Japanese cherry phenology

Abstract

Understanding the drivers of phenological events is vital for forecasting species' responses to climate change. We developed flexible Bayesian survival regression models to assess a 29-year, individual-level time series of flowering phenology from four taxa of Japanese cherry trees (Prunus spachiana, Prunus × yedoensis, Prunus jamasakura, and Prunus lannesiana), from the Tama Forest Cherry Preservation Garden in Hachioji, Japan. Our modeling framework used time-varying (chill and heat units) and time-invariant (slope, aspect, and elevation) factors. We found limited differences among taxa in sensitivity to chill, but earlier flowering taxa, such as P. spachiana, were more sensitive to heat than later flowering taxa, such as P. lannesiana. Using an ensemble of three downscaled regional climate models under the A1B emissions scenario, we projected shifts in flowering timing by 2100. Projections suggest that each taxa will flower about 30 days earlier on average by 2100 with 2-6 days greater uncertainty around the species mean flowering date. Dramatic shifts in the flowering times of cherry trees may have implications for economically important cultural festivals in Japan and East Asia. The survival models used here provide a mechanistic modeling approach and are broadly applicable to any time-to-event phenological data, such as plant leafing, bird arrival time, and insect emergence. The ability to explicitly quantify uncertainty, examine phenological responses on a fine time scale, and incorporate conditions leading up to an event may provide future insight into phenologically driven changes in carbon balance and ecological mismatches of plants and pollinators in natural populations and horticultural crops.

Timing of cherry tree blooming: Contrasting effects of rising winter low temperatures and early spring temperatures

“…However, we argue that there are no relevant evidence to clearly elucidate the physiological mechanisms of the effect of chilling requirement on the first flowering time of cherry trees. In fact, the physiological mechanisms controlling bud burst and flowering of plants in early spring are not precisely known (Allen et al, 2014). Moreover, there is a flaming dispute on whether the effects of chilling accumulation and heat accumulation are overlapped in time (Allen et al, 2014;Chuine et al, 2016;Martínez-Lüscher et al, 2017).…”

Deciphering hybrid larch reaction norms using random regression

“…Even in equatorial regions, oil palm is able to react to subtle variations in photoperiod and drought events by changing its bunch productivity (Legros et al 2009). Cherry tree phenology reacts promptly to climate, notably heat, with global warming expected to bring flowering a month forward (Allen et al 2014). Ismaili et al (2016) showed a significant genotype-byyear interaction in apricot tree, and recommend the use of mixed models for the analysis of perennial plants’ longitudinal data.…”

Beyond seasonal climate: statistical estimation of phenological responses to weather

“…This type of survival model with time-dependent covariates has also been used in the medical sciences (Sen et al 2010). Recently, similar daily survival models were used to estimate the effects of forcing and chilling on cherry tree phenology (Terres et al 2013, Allen et al 2014. These studies show that, when critical parameters of mechanistic models are known or assumed (in particular, the threshold amount of chilling and the date of transition from chilling to forcing), survival models may offer a useful bridge between mechanistic and statistical models.…”