3.1 Fitting the Curves

The following code can be used to read the data and fit each curve:

# Load required packages
library(PhotoGEA)
library(lattice)

# Define a vector of paths to the files we wish to load; in this case, we are
# loading example files included with the PhotoGEA package
file_paths <- c(
  PhotoGEA_example_file_path('c3_aci_1.xlsx'),
  PhotoGEA_example_file_path('c3_aci_2.xlsx')
)

# Load the data from each file
gasex_exdf_list <- lapply(file_paths, function(fpath) {
  read_gasex_file(fpath, 'time')
})

# Get the names of all columns that are present in all of the Licor files
columns_to_keep <- do.call(identify_common_columns, gasex_exdf_list)

# Extract just these columns
gasex_exdf_list <- lapply(gasex_exdf_list, function(x) {
  x[ , columns_to_keep, TRUE]
})

# Combine the data from all the files
gasex_data <- do.call(rbind, gasex_exdf_list)

# Define a new column that uniquely identifies each curve
gasex_data[, 'curve_id'] <-
  paste(gasex_data[, 'species'], '-', gasex_data[, 'plot'] )

# Organize the data
gasex_data <- organize_response_curve_data(
    gasex_data,
    'curve_id',
    c(9, 10, 16),
    'CO2_r_sp'
)

# Calculate the total pressure
gasex_data <- calculate_total_pressure(gasex_data)

# Calculate temperature-dependent values of C3 photosynthetic parameters
gasex_data <- calculate_temperature_response(gasex_data, c3_temperature_param_bernacchi)

# The default optimizer uses randomness, so we will set a seed to ensure the
# results from this fit are always identical
set.seed(1234)

# Fit all curves in the data set
aci_results <- consolidate(by(
  gasex_data,
  gasex_data[, 'curve_id'],
  fit_c3_aci,
  Ca_atmospheric = 420
))

#> Warning: package 'lattice' was built under R version 4.4.3

When this document was generated, evaluating this code required the following amount of time:

#>    user  system elapsed 
#>   19.92    0.21   21.19

The timing results may vary depending on the particular machine used to run the code. Nevertheless, this is a small time investment for an advanced algorithm that uses derivative-free optimizers for robust fitting and calculates nonparametric confidence intervals to determine which estimated parameter values are reliable.

This example contains 13 commands, so it certainly isn’t short; however, a close look reveals that much of the commands are general and would apply to any set of C₃ response curves. In fact, only a few parts would need to be modified, such as the list of files to read, the curve identifier, and the value of mesophyll conductance. While using PhotoGEA, you are encouraged to copy this example and any others to use as the base of your own scripts; work smarter, not harder!

3.2 Viewing the Results

Having fit the response curves, it is also possible to view the fits, the parameter estimates, and their confidence intervals. PhotoGEA provides several tools for doing this, which enable users to check the fit quality and ensure that only reliable parameter estimates are used in subsequent analysis.

We can plot the measured values of net assimilation (black circles), the fitted values of net assimilation (An), and each of the limiting assimilation rates calculated during the fitting procedure: the Rubisco limited rate (Ac), the RuBP regeneration limited rate (Aj), and the triose phosphate utilization (TPU) limited rate (Ap). This is a basic quality check where we can make sure that the fits make sense and look believable:

plot_c3_aci_fit(aci_results, 'curve_id', 'Ci', ylim = c(-10, 80))

In this figure, some curves are missing one or more of the potential limiting rates. When this occurs, it means that no points in the curve were found to be limited by that process.

Another way to check the overall quality of the fits is to plot the residuals, which should be randomly distributed:

xyplot(
  A_residuals ~ Ci | curve_id,
  data = aci_results$fits$main_data,
  type = 'b',
  pch = 16,
  grid = TRUE,
  xlab = paste0('Intercellular CO2 concentration (', aci_results$fits$units$Ci, ')'),
  ylab = paste0('Assimilation rate residual (measured - fitted)\n(', aci_results$fits$units$A, ')'),
)

For individual parameters, we can take a look at the best-fit values and the associated confidence intervals. Here is an example showing values of Tp_at_25, the maximum rate of triose phosphate utilization.

aci_results$parameters[, c('curve_id', 'Tp_at_25_lower', 'Tp_at_25', 'Tp_at_25_upper')]
#>       curve_id Tp_at_25_lower  Tp_at_25 Tp_at_25_upper
#> 1  soybean - 1       7.664314  8.167682       8.609374
#> 2 soybean - 5a           -Inf        NA            Inf
#> 3 soybean - 5b           -Inf        NA            Inf
#> 4  tobacco - 1       7.597912        NA            Inf
#> 5  tobacco - 2      12.100282        NA            Inf
#> 6  tobacco - 4       9.901601 10.282169      10.671615

Some of these estimates have an upper limit of Inf and a best estimate of NA. A comparison with the fits shown above indicates that for these curves, insufficiently many points were found to be TPU-limited, preventing a reliable estimate of Tp_at_25.

It is also possible to plot the best-fit values of a parameter averaged across subsets of the data in a bar chart, where the error bars represent the standard error of the mean. Any values of NA will be excluded. Here we plot values of Vcmax at 25 degrees C for each species.

barchart_with_errorbars(
  aci_results$parameters[, 'Vcmax_at_25'],
  aci_results$parameters[, 'species'],
  xlab = 'Species',
  ylab = paste0('Vcmax at 25 degrees C (', aci_results$parameters$units$Vcmax_at_25, ')'),
  ylim = c(0, 200)
)