Extended Data Fig. 3: Sensitivity analysis of the relationship of CO2 assimilation and autumn phenology from ground observations.
From: Acclimation of phenology relieves leaf longevity constraints in deciduous forests
Partial relationships of multiple LMMs, where EOS is the response variable and (A, D, G) the long-term trend (year) and (B, E, H) Anet are treated as fixed effects. Anet estimates are simulated by the P-model and considering (B) a daylength threshold of 10.0 hours, (E) a fixed DOY cut-off on the 23rd of September, and (H) a fixed DOY cut-off on the 21st of June. (C, F, I) EOS versus Anet based on an LMM with Anet as a single fixed effect. In all bivariate and univariate models, site and species are treated as grouping variables of random intercepts. Black lines represent the expected mean values from LMMs and grey ranges their 95% confidence intervals. Colour hexagonal heatmap represents the observed data adjusted for the effects of the covariates.
