I am developing a FSPM for dwarf tomatoes in a controlled environment (specifically a vertical farm, where the light, if it is on, is always from one direction and constant). Temperature and VPD are supposed to be stable during the day and night. Plants will be well fertilized and non water limited.
I will work with 3 cultivars and 2 densities. So a total of 6 scenarios.
The goal is to simulate carbon assimilation at the leaf level and, from there, calculate plant carbon assimilation.
I often see that the Farquhar, von Caemmerer and Berry (FvCB) biochemical model for C3 photosynthetic rates is used.
To calibrate and validate my FSPM model, I will have to measure many other things beyond photosynthesis; therefore, I am trying to understand what is “good enough” in terms of FvCB model parameter estimation.
Which parameters are “safe” to take from literature and which need to be measured?
I can use a Licor LI-6800: I an A-Ci curve sufficient for extracting the parameters I need? How many samples of the A-Ci curve per scenario are needed?
I understood that measuring nitrogen levels in leaves is beneficial for the estimation of some parameters, but is it necessary? Considering that in my casethe plants grow under non-limiting conditions.
The kinetic constants of Rubisco and the parameters related to responses to temperature are usually taken from the literature. The general belief is that they are highly conserved (not exactly true, especially for the temperature responses) but they are hard and time-consuming to estimate properly. If VPD and temperature will be constant then you should not worry too much about those parameters, just get a consistent set (e.g., the ones for tobacco by Bernacchi in the early 2000s).
It depends what the goal is. An A-Ci curve is designed to estimate the different limiting factors at saturating light and it is very useful to connect photosynthesis to the underlying biochemistry (and nitrogen). However, assuming CO2 will be constant in your growth conditions, if you want to simulate growth I would consider measuring light response curves, since light is the only factor that will vary (even if the source is constant you still have self shading and leaf angles). You can use a modified version of the FvCB model (ignore TPU and Vcmax and assume the entire LRC is limited by electron transport) which has the advantage of correcting for CO2 diffusion through stomatal conductance. Or just go for the classic non-rectangular hyperbola (which might be good enough in tomato since stomatal conductance is quite high and I assume the VPD will be moderate).
Nitrogen is not necessary but photosynthetic rates can vary significantly within and across plants (within and across cultivars). Under constant growth conditions that variation is mostly explained by nitrogen content per unit of leaf area (Narea) and possibly leaf age (the two are correlated in a dense canopy). It is very useful to establish regressions between different photosynthetic traits and Narea because the latter is more high throughput and cheaper and it will help you with scaling from leaf to plant/canopy (e.g., scaling with a vertical nitrogen profile).
a. When the light is on, VPD will be stable in a range (still to be defined) between 0.5 and 0.8 kPa —> is this good for a non-rectangular hyperbola? What do I need to measure to calibrate it?
b. In the case of light response curves, how many light response curves per plant should I take to have sufficient information to calibrate a FvCB model for one treatment?
How do you suggest doing, in practice? If I measure around 10 leaves per 3 plants per treatment, - would that be good? Do I need to measure the light response curves on the same leaves - or just measures of the photosynthetic rates are sufficient? The following image is a fair visual description of the dwarf tomato architecture I am modeling
You could probably make a pre-experiment where you would measure ASat for all your treatments, on different leaf ages and lighting conditions (high vs low light, e.g. top and bottom of the canopy). The idea is that because your plants are not stressed for water or nitrogen, only leaf age and exposition to light can influence your parameters (but maybe not).
So if you measure the assimilation at saturated light for many leaves, you’ll get an idea of the variability in the photosynthetic capacity between leaves / treatments / etc… Then you could target your A/Ci curves only on the leaves that would help you characterise this variability.
Another idea to get things faster is to use the RACIR method from the LiCOR 6800 (or whatever the new method is called).
You can also couple your gas exchange measurements with SPAD measurements to see if it can explain your variability. SPAD can be used as a proxy for N concentration, which can change due to allocation (leaves under high levels of light have more N) or reallocation (old leaves have less N). But it may not work because it saturates pretty quickly. I never worked on tomato so I have no idea, but I’m pretty sure you can find some refs on that.
You could also add a Gs~VPD response curve to parameterize a stomatal conductance model such as the one from Medlyn et al. (2011).
Then you can use any tool to fit the FvCB parameters to each A/Ci curve (and for Gs~f(A,VPD) too), e.g. plantecophys or PlantBiophysics (disclaimer here, you know I develop the package).
Bit late to the party, but hope this is still somewhat helpful Please note that I’m currently very busy, but would be happy to discuss further in September if you’d like more information.
We did a global sensitivity analysis of an FSP model of maize a while ago, and included a number of (C4) FvCB-parameters. This work is currently in preparation, so I cannot share the full manuscript yet. Generally, we found that most of the FvCB-parameters we included were unimportant (i.e. likely do not require very accurate estimation) for the outputs under consideration (LAI, aboveground biomass, yield). See e.g. the figures below (parameters 35-43).
Thank you @pmxrr3 , really fascinating work you are doing - I look forward to see the published manuscript!
We recently decided to opt for a more simplistic model of photosynthesis (for now). I will use the Thornley photosynthetic model since I work at low light intensity and all my climate variables are stable and not used as a treatment.
However, @julia.winkeler might really get some interesting insight from your work (she is calibrating a FvCB model for FSPM cucumbers).
Please note that I’m currently very busy, but would be happy to discuss further in September if you’d like more information.
