Modelling analysis confirms the role of NPQ saturation for the divergence of the GPP–SIF relationship during heatwave

Abstract

We appreciate Antala et al. (2025) for their critical assessment of our article titled ‘Heatwave breaks down the linearity between sun-induced fluorescence and gross primary production’ (Martini et al., 2022), and support their emphasis on the importance of understanding how nonphotochemical quenching (NPQ) modulates the relationship between sun-induced fluorescence (SIF) and gross primary production (GPP). However, we believe that their Correspondence may have overlooked several crucial aspects of our work and their conclusions are, at least partially, based on some misunderstandings. We address what we believe represents their major points of criticism and reinforce the results presented in Martini et al. (2022) by means of a new modelling analysis.

In their opening paragraph titled ‘Heatwave did not break the linearity between SIF and GPP’, Antala et al. (2025) claim that a diverging, nonlinear SIF–GPP relationship is evident already in the preheatwave (pre-HW) and was not caused by the heatwave (HW) conditions, as reported by Martini et al. (2022). They arrive at this conclusion by focussing, day by day, on the subdiurnal dynamics of SIF and GPP (figs 1 and 2 of Antala et al., 2025). However, interpreting the data in this way is inherently challenging, which is why Martini et al. (2022) have refrained from doing so, for two main reasons. First and foremost, environmental conditions change dramatically during the course of any one day with ensuing effects on plant physiology. For example, during the morning hours, photochemical and fluorescence yields often exhibit a negative relationship driven by changes in photochemical quenching. Conversely, during midday hours, the relationship between photochemical and fluorescence yield is typically positive and driven by changes in NPQ (Porcar-Castell et al., 2014; Magney et al., 2020). In fact, Antala et al. (2025) confirm substantial subdiurnal changes in plant physiology themselves by invoking a dominant role of changes in stomatal conductance in shaping the subdiurnal SIF–GPP relationship during pre-HW conditions, which we do not dispute. Second, the SIF–GPP relationship on a subdiurnal basis is highly sensitive to angular effects, which have the potential to obscure the true physiological signal. While these effects were minimized by correcting SIF measurements for the escape probability, the simple correction method used may introduce additional uncertainty. Consequently, Martini et al. (2022) averaged data over certain hours of the day, and thus a given range of solar zenith angles, to obtain robust results.

In the paragraph titled ‘NPQ did not saturate’, Antala et al. (2025) put forth that ‘NPQ does not saturate with decreasing GPP’. This is indeed the case, as demonstrated in fig. 4(b) of the original manuscript where a single linear relationship is evident for both pre-HW and HW conditions, and Martini et al. (2022) never claimed the opposite.

What Martini et al. (2022) did in fact suggest is that NPQ saturates at high levels of relative light saturation of photosynthesis (x; figs 5, 6a,b in the original manuscript), leading to increased energy allocation to SIF (fig. S4b in the original manuscript), thereby altering the direction of the SIF–GPP relationship observed both at leaf and canopy scale (fig. 3 in the original manuscript).

To verify the reasoning by Martini et al. (2022), we conducted simulations using the leaf-scale physiological module of the SCOPE model (Van der et al., 2009, 2014), replacing its parameterization of the NPQ-relative light saturation of photosynthesis (x) with two separate linear relationships; one for pre-HW, where NPQ increases with x, and one for HW conditions, where NPQ is saturating/decreasing with x, based on leaf-level observations at high x values shown in Fig. 1(a). We then conducted a series of simulations by varying leaf temperature from 25 to 40°C at high radiation conditions (to ensure high x values). As shown in Fig. 1(b), these simulations reproduce the divergent relationship of the leaf-scale yields of photochemistry and fluorescence between pre-HW (positive relationship) and HW (negative relationship) conditions shown in fig. 3(b) of the original manuscript, corroborating the reasoning by Martini et al. (2022).

Additionally, the statement ‘NPQ also contains information about the fluorescence emission’ by Antala et al. (2025) is inaccurate in our view and not supported by the references cited. The confusion likely arises from the fact that NPQ is derived from active fluorescence data, but this does not imply that NPQ contains information about fluorescence emission. NPQ reflects the first-order rate constant of NPQ (Kn) (Porcar-Castell, 2011), as well as the first-order rate constants of fluorescence emission (Kf) and constitutive heat dissipation (Kd). While Kn is strongly physiologically regulated, Kf is assumed to remain constant (Clegg, 2004), and Kd is only weakly dependent on temperature (Van der et al., 2014). Therefore, NPQ primarily provides insight into the rate of the NPQ process.

Antala et al. (2025) emphasize φNPQ, noting its different patterns compared with NPQ, and claim that ‘Correlating φNPQ with a quantum yield of photochemistry further supports no saturation of φNPQ in measured data’. Again, we did not claim that photochemical vs nonphotochemical yields saturate. In fact, we do show (fig. S12 of the original manuscript) that the behaviour of the yield of NPQ differs somewhat from NPQ (fig. 6 of the original manuscript).

Finally, in the paragraph titled ‘Fluorescence dramatically decreased with severe heat’, Antala et al. (2025) mention the presence of negative SIF values in their fig. 1 and question the accuracy of retrievals when SIF is low, that is, during the heatwave. Here, we agree on the need to scrutinize fluorescence retrievals when the signal-to-noise ratio is low; however, would like to clarify that while some values approached zero, negative SIF values were not observed. Even though some retrieval methods, such as the spectral fitting method, may produce physically not meaningful negative SIF values, we suggest not omitting these, as these reflect the overall uncertainty of retrievals. In addition, the active leaf-scale and passive canopy-scale data show similar behaviours (fig. 3b,c in the original manuscript), providing an independent confirmation of the retrieved canopy-level SIF data.

Taken together, we unequivocally demonstrate using an independent process-based modelling approach that the diverging relationships between SIF and GPP during pre-HW and HW conditions are driven by the saturation of NPQ during the HW period, confirming the core message of Martini et al. (2022). We agree with Antala et al. (2025) in that further work is necessary to elucidate the processes underlying observed changes in excitation energy distribution caused by extreme environmental conditions, which is a prerequisite for correctly interpreting the SIF–GPP relationship during such periods. However, their Correspondence seems to contribute little to that end. Synthesis studies using combined active and passive fluorescence data and canopy scale fluxes collected at multiple structurally and functionally differing sites are, in our opinion, the way forward for process understanding and for improving a generalizable parameterization of models, such as SCOPE, to simulate carbon fluxes and radiative transfer under extreme environmental conditions.

None declared.

DM, MM and GW contributed to the writing of the article. GW carried out the SCOPE simulations and DM crafted the figure.

The New Phytologist Foundation remains neutral with regard to jurisdictional claims in maps and in any institutional affiliations.