Evaluation and mechanism exploration of the diurnal hysteresis of ecosystem fluxes

Diurnal environmental variations can be dynamically strong (driven by solar radiation) and can lead to hysteretic plant responses. The diurnal hysteresis patterns of ecosystem fluxes and meteorological variables are important to understand the relations and interactions between vegetation and its surrounding environment. To provide new insights on hysteresis at the ecosystem scale, we systematically evaluated the differences in ecosystem fluxes between morning (time of day before the highest potential shortwave incoming) and afternoon (time of day after the highest potential shortwave incoming), as well as both biotic and abiotic driving mechanisms, using half-hourly and hourly data from 82 eddy-covariance sites in the FLUXNET2015 Tier 1 dataset. Results show that gross primary production (GPP) is slightly higher in the morning for most plant functional types (PFTs), with less water stress, yet with higher, compensating, CO₂ concentration. Conversely, evapotranspiration (ET) is larger in the afternoon, with higher atmospheric vapor pressure deficit (VPD) and a tendency for decreased stomatal opening, in response to daytime boundary layer warming and drying. The higher ET reduces the afternoon increase of the leaf-scale VPD compared to the atmospheric VPD. A GPP decomposition analysis suggests that changes in light use efficiency (LUE) are the main factor explaining the difference in GPP between the morning and afternoon. Additionally, simulations from an ecosystem conductance model show that plant responses to VPD are higher and interactions with the atmosphere are stronger in the afternoon. Both evaluations of LUE and responses to VPD along a soil-plant-atmosphere continuum model suggest a systematic drop in leaf water potential in the afternoon affecting ecosystem stress responses. Accordingly, the hysteresis effects ultimately result in daytime variations in ET and GPP, through changes in ecosystem conductance coupled to the atmosphere.