Using Optimal Carbon Benefit Hypothesis To Predict Leaf Phenology Of Deciduous Trees

Leaf phenology is the biological event of periodic leaf-out and leaf fall that is driven by external climate factors and internal growth strategies.It has a profound effect on ecological processes such as resource acquisition,vegetation productivity,and the ecosystem carbon cycle,which is a sensitive indicator of climate change and plant adaptation strategy.Therefore,investigating phenology mechanisms and models will help us clarify the future plant response to climate change and its influence on the ecosystem.Current models including statistical and processed models focus on describing the relationship between phenology and external climate factors,while few take internal growth strategies into account,which limits the model performance in predicting within-site interspecific variations.To address this issue,we proposed predicting the leaf phenology of deciduous trees based on the optimal carbon benefit hypothesis,i.e.plants have evolved the optimal leaf phenology strategy under selective pressure that maximizes the average or total net carbon gain during phenology.This study aims to investigate the relationship between leaf phenology and carbon benefit-related leaf traits by observing 33 deciduous species at East China Normal University,Shanghai,in 2021.Meanwhile,we built the carbon benefit model to test whether leaf phenology follows the optimal carbon benefit hypothesis.The main results are as follows:（1）Among all species,the leaf-out dates were about DOY 99±13 in 2021,and the duration was generally short about 25±10 days.The leaf-out dates varied slightly with phenology sequences,that the flower-before-leaf and flower-with-leaf species leafing out earlier than the flower-after-leaf species.A large interspecific variation was in leaf fall dates,more than 100 days between the earliest and the latest leaf fall dates,and the duration was generally long about 104±10 days.Overall,the correlations between leaf fall dates and benefit-related traits were stronger than leaf-out.There was a significant negative relationship between leaf fall dates and leaf mass per area LMA（r=-0.37,P<0.05）,and a significant positive relationship between leaf fall dates and total nitrogen content（r=0.38,P<0.05）,while leaf-out had no significant relationships with these two traits.（2）Two ways of quantifying carbon benefit were applied to test the optimal carbon benefit hypothesis for deciduous trees,i.e.maximal average net carbon gain and total net carbon gain,and we set three simulation scenarios:1)assume that both leaf-out and leaf fall phenology follow the hypothesis,and the model predicts the leaf-out and leaf fall dates simultaneously;2)assume only leaf-out phenology follows the hypothesis,and the model predicts leaf-out dates given observed leaf fall dates;3)assume only leaf fall phenology follows the hypothesis,and the model predicts leaf fall dates given observed leaf-out dates.Comparing the model performance based on two benefit quantifying ways under scenario 1,it showed,that the model has no predictive power of leaf-out and leaf fall dates when the carbon benefit was the average net carbon gain.The predicted leaf-out dates（DOY 112–117）were later than observations（DOY 69–128）,while the predicted leaf fall dates（DOY 255–264）were earlier than observed dates（DOY 226–359）.When considering the total net carbon gain as the carbon benefit,the model could roughly predict leaf fall dates,especially for flower-after-leaf specie（slope=0.51,R~2=0.72,P<0.01）.（3）By result（2）,it’s more accurate for predicting leaf phenology dates when the carbon benefit was the total net carbon gain,but the model performance varies between leaf-out and leaf fall.To further investigate the role of leaf-out and leaf fall in the carbon benefit strategy,we simulated scenarios 2 and 3.In scenario 2,there was no significant improvement in predicting leaf-out when inputting observed leaf fall dates into the model,compared with scenario 1.Whereas,in scenario 3,the model had significant improvement（slope=0.25,R~2=0.14,P<0.05）,especially for flower-after-leaf species（slope=0.77,R~2=0.86,P<0.001）.In conclusion,the optimal carbon benefit hypothesis based on maximal total net carbon gain has the potential to predict interspecific variation in leaf fall dates of deciduous trees.It implies leaf fall phenology is significantly influenced by carbon benefit strategy while leaf-out is not.