fitACi: Fit A-Ci curve
In AleMorales/photofit: Estimate parameters from photosynthesis curves

Description Usage Arguments Details See Also Examples

View source: R/FvCBAci.R

fitACi will fit a steady-state model of photosynthesis to a response curve measuring CO₂ assimilation and intercellular CO₂ molar fraction.

fitACi(
  data,
  priors = priors_ACi(),
  pars = initial_ACi(),
  fixed = c("KmC", "KmO"),
  method = "quad",
  algorithm = "BFGS",
  ...
)

`data`	Measurements as a `data.frame` with two columns called `Ci` and `A` representing intercellular CO₂ molar fraction and net CO₂ assimilation, respectively.
`priors`	List of parameters for the prior distributions as generated by `priors_ACi`.
`pars`	Vector of initial values of parameters as generated by `initial_ACi`.
`fixed`	Vector with names of parameters that will not be fitted to the data but fixed to their initial values. Same names as in `priors_ACi` and `initial_ACi` should be used.
`method`	Method to approximate the posterior distribution of parameters. See Details.
`algorithm`	Name of non-linear optimization algorithm used to calculate maximum a posteriori. See Details.

A classic FvCB model that implicitly assumes infinite mesophyll conductance is used. Net CO₂ assimilation for every measurement point is calculated as the minimum of three potential rates limited by Rubisco kinetics (Ac), electron transport (Aj) and triose phosphate utilisation (At):

Ac = Vcmax*(Ci - CiStar)/(Ci + KmC*(1 + 210/KmO))

Aj = J*(Ci - CiStar)/(4*Ci + 8*CiStar)

At = 3*TPU

The parameters that can be fitted are:

Vcmax Maximum rate of carboxylation (μmol m⁻² s⁻¹).
CiStar Intercellular CO₂ compensation point (μmol mol⁻¹).
Kmc Michaelis-Menten constant of carboxylation with respect to CO₂ (μmol mol⁻¹).
Kmo Michaelis-Menten constant of carboxylation with respect to O₂ (mmol mol⁻¹).
J Maximum rate of electron transport under measurement conditions (μmol m⁻² s⁻¹).
TPU Maximu rate of triose phosphate utilisation (μmol m⁻² s⁻¹).
Rd Mitochondrial respiration in the light (μmol m⁻² s⁻¹).
sigma Standard deviation of measurement error (μmol m⁻² s⁻¹).

The default initial values are generated by initial_ACi whereas the mean and standard deviations of Gaussian prior distributions are generated by priors_ACi.

Values allowed for the method argument are "quad" (quadratic approximation) and "sir" (sampling importance resampling).

All methods to approximate the posterior start with a calculated of the maximum a posteriori. The argument algorithm corresponds to the method argument of the function optimx (see help for this function for a list of possible algorithms).

initial_ACi, priors_ACi

data = data.frame(Ci = c(245.39, 164.70, 87.59, 50.22, 332.55, 333.15, 421.98,
                  515.53, 612.01, 707.88, 904.34, 1099.13, 1295.58, 1485.47),
           A = c(8.90, 5.72, 1.94, -0.18, 11.38, 11.49, 13.49, 14.66, 15.10,
                 15.61, 15.49, 15.43, 14.97, 15.51))

# Quadratic approximation
fitQuad = fitACi(data = data, fixed = c("KmC", "KmO"))
summary(fitQuad)
plot(fitQuad)
plot(fitQuad, type = "marginal")
plot(fitQuad, type = "pairs")

# Quadratic approximation - Avoiding TPU
fitQuad2 = fitACi(data = data, pars = initial_ACi(TPU = 100),
                 fixed = c("KmC", "KmO", "TPU"))
summary(fitQuad2)
plot(fitQuad2)
plot(fitQuad2, type = "pairs")

# SIR approximation
fitSIR = fitACi(data = data, pars = initial_ACi(TPU = 100),
                 fixed = c("KmC", "KmO", "TPU"))
summary(fitSIR)
plot(fitSIR)
plot(fitSIR, type = "marginal")
plot(fitSIR, type = "pairs")