Adapting a process based cold hardiness model to conifers

Abstract

Plant phenology has been and continues to be impacted by climate change. Process-based modeling of phenology reveals biological characteristics through interpretation of model results and parameter values. This paper aims to implement a cold hardiness model using historical Douglas-fir (Pseudotsuga menziesii (Mirb.) Franco) data, determine if model results improve with data clustering by seed source or growing location, and interpret model results into biological meaning. These interpretations have applications for reforestation and studies of plant phenological response to climate change.Cold hardiness data were compiled through literature review. A process-based cold hardiness model using daily temperature inputs was applied to multiple data clustering scenarios. Model results were analyzed for goodness of fit to determine error, efficiency, and bias. A sensitivity analysis and cross validation were performed to determine parameter sensitivity, model bias, and variance. Data clustering by seed source improved fit compared to clustering by growing location or no clustering when applied to the full dataset. Using only temperature inputs, model results had low error when data modeled were similar. Results show that for cold hardiness acclimation a linear growing degree function with a threshold of 10°C was adequate across testing data, as was a maximum cold hardiness temperature of -3°C. Interpretation of model results show that both acclimation to growing sites and seed source genetics impact cold hardiness response, though clustering by seed source improved model performance. This model can be applied to mitigate cold related risks to seedlings during production and establishment, and can be a template for predicting phenological responses in simulated future climate scenarios.