Votre recherche
Résultats 8 ressources
-
Abstract Digital leaf physiognomy (DLP) is considered as one of the most promising methods for estimating past climate. However, current models built using the DLP data set still lack precision, especially for mean annual precipitation (MAP). To improve predictive power, we developed five machine learning (ML) models for mean annual temperature (MAT) and MAP respectively, and then tested the precision of these models and some of their averaging compared with that obtained from other models. The precision of all models was assessed using a repeated stratified 10‐fold cross‐validation. For MAT, three combinations of models ( R 2 = .77) presented moderate improvements in precision over the multiple linear regression (MLR) model ( R 2 = .68). For log e (MAP), the averaging of the support vector machine (SVM) and boosting models improved the R 2 from .19 to .63 compared with that of the MLR model. For MAP, the R 2 of this model combination was 0.49, which was much better than that of the artificial neural network (ANN) model ( R 2 = .21). Even the bagging model, which had the lowest R 2 (.37) for log e (MAP), demonstrated better precision ( R 2 = .27) for MAP. Our palaeoclimate estimates for nine fossil floras were also more accurate, because they were in better agreement with independent paleoclimate evidence. Our study confirms that our ML models and their averaging can improve paleoclimatic reconstructions, providing a better understanding of the relationship between climate and leaf physiognomy.
-
Abstract Digital leaf physiognomy (DLP) is considered as one of the most promising methods for estimating past climate. However, current models built using the DLP data set still lack precision, especially for mean annual precipitation (MAP). To improve predictive power, we developed five machine learning (ML) models for mean annual temperature (MAT) and MAP respectively, and then tested the precision of these models and some of their averaging compared with that obtained from other models. The precision of all models was assessed using a repeated stratified 10‐fold cross‐validation. For MAT, three combinations of models ( R 2 = .77) presented moderate improvements in precision over the multiple linear regression (MLR) model ( R 2 = .68). For log e (MAP), the averaging of the support vector machine (SVM) and boosting models improved the R 2 from .19 to .63 compared with that of the MLR model. For MAP, the R 2 of this model combination was 0.49, which was much better than that of the artificial neural network (ANN) model ( R 2 = .21). Even the bagging model, which had the lowest R 2 (.37) for log e (MAP), demonstrated better precision ( R 2 = .27) for MAP. Our palaeoclimate estimates for nine fossil floras were also more accurate, because they were in better agreement with independent paleoclimate evidence. Our study confirms that our ML models and their averaging can improve paleoclimatic reconstructions, providing a better understanding of the relationship between climate and leaf physiognomy.
-
Intense and frequent drought events strongly affect plant survival. Non-structural carbohydrates (NSCs) are important “buffers” to maintain plant functions under drought conditions. We conducted a drought manipulation experiment using three-year-old Pinus tabulaeformis Carr. seedlings. The seedlings were first treated under different drought intensities (i.e., no irrigation, severe, and moderate) for 50 days, and then they were re-watered for 25 days to explore the dynamics of NSCs in the leaves, twigs, stems, and roots. The results showed that the no irrigation and severe drought treatments significantly reduced photosynthetic rate by 93.9% and 32.6% for 30 days, respectively, leading to the depletion of the starch storage for hydraulic repair, osmotic adjustment, and plant metabolism. The seedlings under moderate drought condition also exhibited starch storage consumption in leaves and twigs. After re-watering, the reduced photosynthetic rate recovered to the control level within five days in the severe drought group but showed no sign of recovery in the no irrigation group. The seedlings under the severe and moderate drought conditions tended to invest newly fixed C to starch storage and hydraulic repair instead of growth due to the “drought legacy effect”. Our findings suggest the depletion and recovery of starch storage are important strategies for P. tabulaeformis seedlings, and they may play key roles in plant resistance and resilience under environmental stress.
-
Few studies have focused on the combined impact of climate change, CO2, and land-use cover change (LUCC), especially the evaluation of the impact of LUCC on net primary productivity (NPP) in the future. In this study, we simulated the overall NPP change trend from 2010 to 2100 and its response to climatic factors, CO2 concentration, and LUCC conditions under three typical emission scenarios (Representative Concentration Pathway RCP2.6, RCP4.5, and RCP8.5). (1) Under the predicted global pattern, NPP showed an increasing trend, with the most prominent variation at the end of the century. The increasing trend is mainly caused by the positive effect of CO2 on NPP. However, the increasing trend of LUCC has only a small positive effect. (2) Under the RCP 8.5 scenario, from 2090 to 2100, CO2 has the most significant positive impact on tropical areas, reaching 8.328 Pg C Yr−1. Under the same conditions, climate change has the greatest positive impact on the northern high latitudes (1.175 Pg C Yr−1), but it has the greatest negative impact on tropical areas, reaching −4.842 Pg C Yr−1. (3) The average contribution rate of LUCC to NPP was 6.14%. Under the RCP8.5 scenario, LUCC made the largest positive contribution on NPP (0.542 Pg C Yr−1) globally from 2010 to 2020.