rangeRinPA-package: Protected Area Personnel and Ranger Numbers are Insufficient...

In courtiol/rangeRinPA: Protected Area Personnel and Ranger Numbers are Insufficient to Deliver Global Expectations

Protected Area Personnel and Ranger Numbers are Insufficient to Deliver Global Expectations

Description

The goal of rangeRinPA is to reproduce most results from the paper entitled "Protected area personnel and ranger numbers are insufficient to deliver global expectations" by Appleton et al. (Nature Sustainability, 2022; https://doi.org/10.1038/s41893-022-00970-0).

Details

For reproducing the results of our paper, just follow the code presented in the section "Examples" below.

References

Protected area personnel and ranger numbers are insufficient to deliver global expectations. Appleton et al. Nature Sustainability (2022).

Examples

## Not run: 

###################################################### MAIN ANALYSIS

## Set and create all directories to store files

path <- "inst/extdata/" ## set the path where you want to store all created files
if (!dir.exists(path)) stop("You must use an existing path")

path_tables <- paste0(path, "tables/")
if (!dir.exists(path_tables)) dir.create(path_tables)

path_figures <- paste0(path, "figures/")
if (!dir.exists(path_figures)) dir.create(path_figures)

path_predictions <- paste0(path, "predictions/")
if (!dir.exists(path_predictions)) dir.create(path_predictions)


## Perform imputations and predictions

Ncpu <- 2L  ## define the number of CPUs to use
n_trees <- 2000


## Run all LMM workflows and save results (optional, see below)

LMM_100 <- run_LMM_workflow(data = data_rangers, Ncpu = Ncpu, coef = 1)
save(LMM_100, file = paste0(path_predictions, "LMM_100.Rdata"))
rm(LMM_100)
gc()

LMM_075 <- run_LMM_workflow(data = data_rangers, Ncpu = Ncpu, coef = 0.75)
save(LMM_075, file = paste0(path_predictions, "LMM_075.Rdata"))
rm(LMM_075)
gc()

LMM_050 <- run_LMM_workflow(data = data_rangers, Ncpu = Ncpu, coef = 0.50)
save(LMM_050, file = paste0(path_predictions, "LMM_050.Rdata"))
rm(LMM_050)
gc()

LMM_025 <- run_LMM_workflow(data = data_rangers, Ncpu = Ncpu, coef = 0.25)
save(LMM_025, file = paste0(path_predictions, "LMM_025.Rdata"))
rm(LMM_025)
gc()

LMM_000 <- run_LMM_workflow(data = data_rangers, Ncpu = Ncpu, coef = 0)
save(LMM_000, file = paste0(path_predictions, "LMM_000.Rdata"))
rm(LMM_000)
gc()


## Run all RF workflows and save results (optional, see below)

RF_100 <- run_RF_workflow(data = data_rangers, Ncpu = Ncpu, coef = 1, n_trees = n_trees)
save(RF_100, file = paste0(path_predictions, paste0("RF_100_", n_trees, ".Rdata")))
rm(RF_100)
gc()

RF_075 <- run_RF_workflow(data = data_rangers, Ncpu = Ncpu, coef = 0.75, n_trees = n_trees)
save(RF_075,file = paste0(path_predictions, paste0("RF_075_", n_trees, ".Rdata")))
rm(RF_075)
gc()

RF_050 <- run_RF_workflow(data = data_rangers, Ncpu = Ncpu, coef = 0.50, n_trees = n_trees)
save(RF_050, file = paste0(path_predictions, paste0("RF_050_", n_trees, ".Rdata")))
rm(RF_050)
gc()

RF_025 <- run_RF_workflow(data = data_rangers, Ncpu = Ncpu, coef = 0.25, n_trees = n_trees)
save(RF_025, file = paste0(path_predictions, paste0("RF_025_", n_trees, ".Rdata")))
rm(RF_025)
gc()

RF_000 <- run_RF_workflow(data = data_rangers, Ncpu = Ncpu, coef = 0, n_trees = n_trees)
save(RF_000, file = paste0(path_predictions, paste0("RF_000_", n_trees, ".Rdata")))
rm(RF_000)
gc()


## Load all results saved on disk
## (useful only if all objects created above are not already in memory)

files_to_load <- c("LMM_000.Rdata", "LMM_025.Rdata", "LMM_050.Rdata", "LMM_075.Rdata",
                   "LMM_100.Rdata", paste0(c("RF_000_", "RF_025_", "RF_050_", "RF_075_", "RF_100_"),
                   n_trees, ".Rdata"))

sapply(paste0(path_predictions, files_to_load), function(file) load(file, envir = .GlobalEnv))



###################################################### MAIN FIGURES

## Figure 1

plot_map_sampling(data_rangers_with_geo)
ggplot2::ggsave(filename = paste0(path_figures, "figure_main_1.pdf"),
                width = 8.8, height = 5.4, units = "cm")


## Figure 2

plot_density_panel(what = LMM_100, data = data_rangers)
ggplot2::ggsave(filename = paste0(path_figures, "figure_main_2.pdf"),
                width = 18, height = 21, units = "cm")


## Figure 3

plot_density_vs_PA_panel(data = data_rangers, coef = 1)
ggplot2::ggsave(filename = paste0(path_figures, "figure_main_3.pdf"),
                width = 18, height = 21, units = "cm")


## Figure 4 & 5 not made with R!



###################################################### MAIN TABLE

## Table 1

table_predictions_main <- table_predictions_summary(what = LMM_100, data = data_rangers)
readr::write_excel_csv(table_predictions_main,
                       file = paste0(path_tables, "table_main_1.csv"))



###################################################### EXTENDED DATA FIGURES


## Extended Data Figure 1

plot_tallies_across_continents(what = LMM_100, data = data_rangers)
ggplot2::ggsave(filename = paste0(path_figures, "figure_extended_1.pdf"),
                width = 18, height = 9, units = "cm")


## Extended Data Figure 2

plot_tallies_across_methods(list_results_LMM = list(LMM_000, LMM_025, LMM_050, LMM_075, LMM_100),
                            list_results_RF = list(RF_000, RF_025, RF_050, RF_075, RF_100),
                            data = data_rangers)
ggplot2::ggsave(filename = paste0(path_figures, "figure_extended_2.pdf"),
                width = 18, height = 9, units = "cm")



###################################################### EXTENDED DATA TABLES

## Extended Data Table 1

table_core_data_formatted <- table_core_data(data_rangers, what = LMM_100)

readr::write_excel_csv(table_core_data_formatted,
                       file = paste0(path_tables, "table_extended_1.csv"))


## Extended Data Table 2

table_predictions_main_with_PI <- table_predictions_summary(what = LMM_100, data = data_rangers,
                                                            with_PI = TRUE)
readr::write_excel_csv(table_predictions_main_with_PI,
                       file = paste0(path_tables, "table_extended_2.csv"))


## Extended Data Table 3

table_predictions_methods <- table_predictions_per_method(
                           list_results_LMM = list(LMM_000, LMM_025, LMM_050, LMM_075, LMM_100),
                           list_results_RF = list(RF_000, RF_025, RF_050, RF_075, RF_100),
                           data = data_rangers)

readr::write_excel_csv(table_predictions_methods,
                       file = paste0(path_tables, "table_extended_3.csv"))


## Extended Data Table 4

table_predictions_methods_densities <- table_predictions_per_method(
                            list_results_LMM = list(LMM_000, LMM_025, LMM_050, LMM_075, LMM_100),
                            list_results_RF = list(RF_000, RF_025, RF_050, RF_075, RF_100),
                            data = data_rangers, density = TRUE)

readr::write_excel_csv(table_predictions_methods_densities,
                       file = paste0(path_tables, "table_extended_4.csv"))


## Extended Data Table 5, 6, 7 & 8 not done with R!



###################################################### SUPPLEMENTARY DATA FIGURES

## Supplementary Data Figure 1

set.seed(123)
plot_reliability_vs_sampling(data_rangers)


## Supplementary Data Figure 2

plot_density_vs_sampling(data_rangers)


## Supplementary Data Figure 3

plot_features_selected(list_results_LMM = list(LMM_000, LMM_025, LMM_050, LMM_075, LMM_100),
                       list_results_RF  = list(RF_000, RF_025, RF_050, RF_075, RF_100),
                       data = data_rangers)


## Supplementary Data Figure 4

plot_features_selection_panel(result1 = LMM_100, result2 = RF_100, who = "rangers")


## Supplementary Data Figure 5

plot_finetuning(result = RF_100, who = "rangers")


## Supplementary Data Figure 6

plot_PA_by_data_type(what = RF_100, data = data_rangers)


###################################################### SUPPLEMENTARY DATA TABLES

## Supplementary Data Table 1

readr::write_excel_csv(table_completeness_obs(data_rangers),
                       file = paste0(path_tables, "table_SI_1A.csv"))
readr::write_excel_csv(table_completeness_km2(data_rangers),
                       file = paste0(path_tables, "table_SI_1B.csv"))
readr::write_excel_csv(table_completeness_vars(data_rangers),
                       file = paste0(path_tables, "table_SI_1C.csv"))


## Supplementary Data Table 2

table_cor <- table_correlates(data_rangers)
readr::write_excel_csv(table_cor,
                       file = paste0(path_tables, "table_SI_2.csv"))


## Supplementary Data Table 3

training_info_initial <- table_training_initial(
                       list_results_LMM = list(LMM_000, LMM_025, LMM_050, LMM_075, LMM_100),
                       list_results_RF = list(RF_000, RF_025, RF_050, RF_075, RF_100),
                       data = data_rangers)

readr::write_excel_csv(training_info_initial,
                       file = paste0(path_tables, "table_SI_3.csv"))


## Supplementary Data Table 4

training_info_final <- table_training_final(
                     list_results_LMM = list(LMM_000, LMM_025, LMM_050, LMM_075, LMM_100),
                     list_results_RF = list(RF_000, RF_025, RF_050, RF_075, RF_100),
                     data = data_rangers)

readr::write_excel_csv(training_info_final,
                       file = paste0(path_tables, "table_SI_4.csv"))


## Supplementary Data Table 5

fine_tuning_selected <- table_tuning(
             list_results_LMM = list(LMM_000, LMM_025, LMM_050, LMM_075, LMM_100),
             list_results_RF = list(RF_000, RF_025, RF_050, RF_075, RF_100),
             data = data_rangers)

readr::write_excel_csv(fine_tuning_selected,
                       file = paste0(path_tables, "table_SI_5.csv"))


## Supplementary Data Table 6

predictions_info <- table_predictions_data(
                      list_results_LMM = list(LMM_000, LMM_025, LMM_050, LMM_075, LMM_100),
                      list_results_RF = list(RF_000, RF_025, RF_050, RF_075, RF_100),
                      data = data_rangers)

readr::write_excel_csv(predictions_info,
                       file = paste0(path_tables, "table_SI_6.csv"))



###################################################### OLD TABLES AND FIGURES NOT USED

plot_projections(what = LMM_100, data = data_rangers)

plot_map_reliability(data_rangers_with_geo)

table_projections(what = LMM_100, data = data_rangers)



###################################################### COMPUTATIONS FOR TEXT

## Number of countries/territories surveyed:
nrow(table_raw_data(data_rangers))


## Rounded numbers for all personnel:
table_projections <- table_projections(what = LMM_100, data = data_rangers)
table_projections %>%
  dplyr::group_by(.data$who) %>%
  dplyr::summarize(dplyr::across(tidyselect::contains("number") |
                                 tidyselect::contains("increase"),
                   .fns = ~ 1000*round(.x/1000)))


## Rounded densities of all personnel:
table_projections %>%
  dplyr::group_by(.data$who) %>%
  dplyr::summarize(dplyr::across(tidyselect::contains("density"), .fns = ~ round(.x, digits = 2)))


## Area covered by PAs:
data_rangers %>%
  dplyr::select(area_surveyed = .data$PA_area_surveyed,
                total_area = .data$area_PA_total) %>%
  dplyr::summarise(area_surveyed = sum(.data$area_surveyed),
                   prop_total_area = area_surveyed/sum(.data$total_area))


## Percentage of land covered with PA in our study:
data_rangers %>%
  dplyr::filter(!is.na(staff_others) | !is.na(staff_rangers) | !is.na(staff_total)) %>%
  dplyr::summarise(area_total = sum(.data$area_PA_total),
                   prop_PA = .data$area_total / sum(.data$area_country)) -> coverage_PA
round(100*coverage_PA$prop_PA, digits = 2)


## Numerical example to explain how imputation is done (in methods):

data_rangers %>%
  dplyr::filter(countryname_eng == "Spain") %>%
  dplyr::select(.data$staff_rangers,
                .data$PA_area_surveyed,
                .data$PA_area_unsurveyed) -> data_spain_before_imputation
 data_spain_before_imputation
 fill_PA_area(data_spain_before_imputation, coef = 1)


## Data reliability:

summary(data_rangers$reliability)
sum(data_rangers$reliability >= 15, na.rm = TRUE)
length(na.omit(data_rangers$reliability))
round(mean(data_rangers$reliability, na.rm = TRUE), 2)
round(sd(data_rangers$reliability, na.rm = TRUE), 2)


## Info on computing time:

extract_results(list_results_LMM = list(LMM_000, LMM_025, LMM_050, LMM_075, LMM_100),
                list_results_RF = list(RF_000, RF_025, RF_050, RF_075, RF_100)) %>%
                dplyr::mutate(time = .data$run_time * .data$Ncpu) %>%
                dplyr::summarize(total_time_h = sum(.data$time))


## End(Not run)

courtiol/rangeRinPA documentation built on Sept. 29, 2022, 9:54 a.m.