Forecasting plant phenology: evaluating the phenological models for Betula pendula and Padus racemosa spring phases, Latvia

Abstract

A historical phenological record and meteorological data of the period 1960-2009 are used to analyse the ability of seven phenological models to predict leaf unfolding and beginning of flowering for two tree species-silver birch Betula pendula and bird cherry Padus racemosa-in Latvia. Model stability is estimated performing multiple model fitting runs using half of the data for model training and the other half for evaluation. Correlation coefficient, mean absolute error and mean squared error are used to evaluate model performance. UniChill (a model using sigmoidal development rate and temperature relationship and taking into account the necessity for dormancy release) and DDcos (a simple degree-day model considering the diurnal temperature fluctuations) are found to be the best models for describing the considered spring phases. A strong collinearity between base temperature and required heat sum is found for several model fitting runs of the simple degree-day based models. Large variation of the model parameters between different model fitting runs in case of more complex models indicates similar collinearity and over-parameterization of these models. It is suggested that model performance can be improved by incorporating the resolved daily temperature fluctuations of the DDcos model into the framework of the more complex models (e.g. UniChill). The average base temperature, as found by DDcos model, for B. pendula leaf unfolding is 5.6 °C and for the start of the flowering 6.7 °C; for P. racemosa, the respective base temperatures are 3.2 °C and 3.4 °C.

The interaction between freezing tolerance and phenology in temperate deciduous trees

“…This is a field to be further explored given it is not known what fluctuation it needs for deacclimation to occur ( Kalberer et al, 2006 ). Temperature fluctuations may also affect the timing of leaf-out as recently suggested by Kalvāns et al (2014) and Wang et al (2014) .…”

Photoperiod constraints on tree phenology, performance and migration in a warming world

“…; Kalvāns et al . ). Most models currently used to estimate the spring onset of bud burst or carbon uptake in temperate and boreal forests rely on temperature, although the manner of accumulating thermal units and the role of chilling requirements varies (Hunter & Lechowicz ; Migliavacca et al .…”

Testing the suitability of thermal time models for forecasting spring wheat phenological development in western Canada

“…The ability to predict a specific crop development stage relative to disease, insect, and (or) weed cycles helps in devising suitable, timely, and environmentally sustainable management strategies, and also in forecasting crop yield. Thermal time units, such as growing degree-days (GDD), are commonly used to assess the rate of plant growth and development as impacted by temperature, and can be correlated to plant development (Slafer and Savin 1991;Gordon and Bootsma 1993;Shaykewich 1995;McMaster and Wilhelm 1997;McMaster et al 2003;Yuan and Bland 2005;Kalvāns et al 2014). The standard GDD unit is calculated using average daily air temperature (T a ) and a base temperature (T b ); daily values are summed to give weekly, dekadal, monthly or seasonal totals.…”