Introduction
Uncertainty in Earth system model projections of our future climate is largely driven by a lack of understanding and model representation of ecosystem processes associated with CO2 assimilation and storage by the terrestrial biosphere (Friedlingstein et al ., 2014; Lovenduski & Bonan, 2017). Photosynthesis is the largest carbon flux on the planet and the gatekeeper process for an uncertain terrestrial carbon sink (Le Quere et al ., 2016). Terrestrial biosphere models (TBMs) are particularly sensitive to structural and parametric representation of photosynthesis (Friend, 2010; Bonanet al ., 2011; LeBauer et al ., 2013; Rogers, 2014; Sargsyanet al ., 2014; Rogers et al ., 2017; Ricciuto et al ., 2018). Photosynthetic capacity, specifically the apparent maximum carboxylation capacity of Rubisco at 25 °C (Vcmax,25 ), is a key model input that drives considerable uncertainty in TBM projections (Friend, 2010; Bonan et al ., 2011; LeBauer et al ., 2013; Sargsyan et al ., 2014).Vcmax,25 is further used to derive the maximum rate of electron transport (Jmax ), the triose phosphate utilization rate (when implemented) and respiration (R ) (e.g. Sitch et al ., 2003; Rogers et al ., 2014; Lombardozziet al ., 2018). If this key parameter shows significant diurnal variation, the implications for TBM projections of CO2assimilation could be notable.
Terrestrial biosphere models (TBMs) typically representVcmax as a fixed parameter (Rogers, 2014) but with some consideration of other factors such as day length (Bauerleet al ., 2012), temperature acclimation (Kattge & Knorr, 2007; Lombardozzi et al ., 2015; Smith et al ., 2015), and emerging potential for covariation with environmental drivers (Aliet al ., 2015; Smith et al. 2019). Only a few studies have investigated diurnal variation in Vcmax (Singsaaset al. , 2000; Kets et al. , 2010; Nascimento & Marenco, 2013). In vitro , Rubisco activation state changes rapidly with irradiance (Salvucci & Anderson, 1987; Parry et al ., 1997), shows diurnal variation (Sage et al. , 1993; Pérez et al. , 2005), and is subject to regulation by Rubisco activase (Portis, 2003; Sage et al ., 2008), which could contribute substantially to diurnal variations in Vcmax . However, in vitro and in vivo data on Rubisco activation state can vary markedly (Rogers et al ., 2001; Sharwood et al ., 2017), and without an easy way to assess activation state in vivo , and an approach to incorporate in vitro measurements into TBM formulations, a gas exchange approach to measuring Rubisco activity is desirable.
While there is a paucity of data on whether diurnal changes inVcmax occur (see Singsaas et al. , 2000; Kets et al. , 2010; and Nascimento & Marenco, 2013 for examples), there are many studies investigating diurnal and circadian patterns in gas exchange parameters. Net CO2 assimilation (Anet ) is known to vary diurnally (Leverenz, 1981; Epron et al. , 1992; Singsaas et al. , 2000; Panek, 2004; Leakey et al. , 2004; Harrison et al. , 2010; Kets et al. , 2010; Nascimento & Marenco, 2013; Bader et al. , 2016), which can be driven by environmental changes. There is a midday depression in net CO2 assimilation that has been related to vapor pressure deficit and water relations (Tenhunen et al ., 1984; Rodà, 1999), whereby stomatal conductance mediates a supply-driven change in carbon assimilation (Leverenz, 1981; Epron et al. , 1992; Harrisonet al. , 2010; Resco de Dios et al. , 2016a,b) and balances the need to fix carbon with preventing water loss (Matthews et al ., 2017), although circadian changes in Anetand gs are decoupled from one another (Doddet al ., 2004; Resco de Dios, 2017). Brodribb & Holbrook (2004) found that leaf hydraulic conductance declines over the day, which may mediate declines in mesophyll conductance (gm ) (Bickford et al ., 2009; Flexas et al ., 2013), reducing chloroplastic CO2 supply (and thereforeAne t) later in the day and may even mediate changes in stomatal behaviour (Sack et al ., 2016). Nardini et al . (2005) found that leaf hydraulic conductance (Kleaf ) was under circadian regulation inHelianthus annuus , which was the cause of diurnal oscillations inKleaf . There may be further changes in realized quantum yield of photosynthesis that result from delays in recovering from photoprotection or photodamage throughout the day, limiting energy available for carbon fixation (Long & Humphries, 1994; Gamon, 2015; Kromdijk et al ., 2016). As well, there are diurnal shifts in carbohydrate accumulation, and this can lead to feedback inhibition of photosynthesis (Sun et al ., 1999). In this way, diurnal changes in net CO2 assimilation can be driven by both supply and demand of substrates, and these diurnal changes in gas exchange scale up to the canopy level (Resco de Dios, 2016a). Many of the diurnal changes in Anet outlined above are due to diurnal environmental changes which models can mimic well. However, it is unclear whether there may be diurnal variation inVcmax , and if there is, diurnal variation inVcmax may present a simple way to account for the effect of diurnal changes in leaf carbon exchange in models.
Current approaches for measuring Vcmax in vivorequire measurement of the response of photosynthesis (A ) to internal CO2 concentration (Ci ) which can take over an hour for a single measurement (Bernacchi et al ., 2003). Recently, Stinziano et al . (2017) developed a method for the rapid measurement of the A-Ci response (the RACiR technique). The dramatic shortening of measurement time enables the collection of high-resolution diurnal patterns in apparentVcmax and apparent Jmax(Stinziano et al ., 2017, 2019ab; Coursolle et al ., 2019; Lawrence et al ., 2019). In addition, this new technique provides values for parameters measured in vivo that are directly applicable to current TBM model formulations and enables measurement of diurnal dynamics under circumstances where in vitro work cannot be done, e.g. where sampling constraints in remote locations are logistically challenging, or when secondary compounds limit the ability to successfully extract and measure Rubisco activity. Studies assessing TBM performance on a diurnal scale are limited (in some cases due to temporal resolution of the models). in the case of ORCHIDEE, carbon fluxes are consistently over-estimated relative to eddy covariance data later in the day across plant functional types (Krinner et al ., 2005), while the Community Atmosphere Model (CAM4 and CAM5) over-estimates latent heat fluxes later in the day at mid-latitude and boreal regions (Lindvall et al ., 2012) suggesting an over-estimation of transpiration, which would imply overestimations of stomatal conductance, Ci , andAnet . These discrepancies may be related to endogenous rhythms in ecosystem-level gas exchange (Resco de Dioset al ., 2012) that are currently unaccounted for.
Our objective was to determine whether diurnal variation inVcmax occurs, and if so, the extent to which such variation affects leaf-level gas exchange modeling. Here we show that the RACiR technique can be used to successfully measure diurnal variation in Vcmax . We demonstrate that diurnal variation in Vcmax is variable across 11 species from several plant functional types, leading to a variable effect on modelled leaf-level gas exchange, underscoring the need to consider diurnal dynamics in Vcmax when modeling.