Merge pull request #1 from qdread/revision2

Revision2

README.md

@@ -1,6 +1,8 @@
 # Biodiversity: Farm2Fork
 
-Code to reproduce all analysis presented in the manuscript **Biodiversity effects of food system sustainability actions from farm to fork**, and the data for the associated Shiny app ([temporary link](https://shiny.sesync.org/apps/biodiversity-farm2fork)).
+[![DOI](https://zenodo.org/badge/DOI/10.5281/zenodo.5949590.svg)](https://doi.org/10.5281/zenodo.5949590)
+
+Code to reproduce all analysis presented in the manuscript **Biodiversity effects of food system sustainability actions from farm to fork**, and the data for the associated [Shiny app](https://qdread.shinyapps.io/biodiversity-farm2fork).
 
 **Citation (to be updated when final)**: Read, Quentin D., Kelly L. Hondula, and Mary K. Muth. Biodiversity effects of food system sustainability actions from farm to fork. *PNAS*, in preparation.
 
@@ -46,4 +48,4 @@ In addition, a few functions are called from an R package `Rutilitybelt` that I
 this repo, called `Rutilitybelt_0.0.0.9000.tar.gz`. It is installed from source as part of script `00`. It can also be installed from source using the
 command `R CMD INSTALL Rutilitybelt_0.0.0.9000.tar.gz`.
 
-*last modified by QDR, 16 August 2021*
+*last modified by QDR, 02 February 2022*

final_output/25_create_summary_data.R

@@ -67,16 +67,15 @@ write_csv(totaldemand_sums, file.path(final_output_path, 'totaldemand_sums_all_s
 
 # Goods flows by county ---------------------------------------------------
 
-options(mc.cores = 8)
-plan(multicore)
+plan(multicore(workers = 8))
 
 fp_goods <- file.path(intermediate_output_path, 'county_consumption_csvs')
 
 scenario_combos <- expand_grid(diet = c('baseline','planetaryhealth','medstyle','usstyle','vegetarian'),
                                waste = c('baseline','preconsumer','consumer','allavoidable'))
 
-sum_goods_flows_county <- function(diet, waste) {
-  flows <- read_csv(glue::glue('{fp_goods}/D_{diet}_WR_{waste}_wide.csv'))
+sum_goods_flows_county <- function(diet, waste, file_path) {
+  flows <- read_csv(glue::glue('{file_path}/D_{diet}_WR_{waste}_wide.csv'))
   message(glue::glue('{diet} by {waste} read'))
   # Outbound: each row represents origin county, sum across all destination counties (columns)
   flows_outbound <- cbind(flows[, 1:3], flow_outbound = apply(flows[, -(1:3)], 1, sum)) %>%
@@ -97,7 +96,7 @@ sum_goods_flows_county <- function(diet, waste) {
 
 }
 
-county_goods_flows <- future_pmap_dfr(scenario_combos, sum_goods_flows_county)
+county_goods_flows <- future_pmap_dfr(scenario_combos, sum_goods_flows_county, file_path = fp_goods)
 
 write_csv(county_goods_flows, file.path(final_output_path, 'goodsflows_county_sums_all_scenarios.csv'))
 
@@ -108,8 +107,8 @@ fp_landcounties <- file.path(intermediate_output_path, 'county_land_consumption_
 scenario_combos <- expand_grid(diet = c('baseline','planetaryhealth','medstyle','usstyle','vegetarian'),
                                waste = c('baseline','preconsumer','consumer','allavoidable'))
 
-sum_land_flows_county <- function(diet, waste) {
-  flows <- read_csv(glue::glue('{fp_landcounties}/D_{diet}_WR_{waste}_landconsumption.csv'))
+sum_land_flows_county <- function(diet, waste, file_path) {
+  flows <- read_csv(glue::glue('{file_path}/D_{diet}_WR_{waste}_landconsumption.csv'))
   message(glue::glue('{diet} by {waste} read'))
   # Longform data so outbound and inbound sums are grouped separately
   flows_outbound <- flows %>%
@@ -132,6 +131,6 @@ sum_land_flows_county <- function(diet, waste) {
 
 }
 
-county_land_flows <- future_pmap_dfr(scenario_combos, sum_land_flows_county)
+county_land_flows <- future_pmap_dfr(scenario_combos, sum_land_flows_county, file_path = fp_landcounties)
 
 write_csv(county_land_flows, file.path(final_output_path, 'landflows_county_sums_all_scenarios.csv'))

final_output/26_compile_Shiny_data.R

@@ -147,8 +147,8 @@ replace_na_dt(county_land_flow_sums)
 
 # Load foreign goods data -------------------------------------------------
 
-import_crop <- fread(file.path(intermediate_output_path, 'fao_VLT_provisional_crops_disaggregated.csv'))
-import_animal <- fread(file.path(intermediate_output_path, 'fao_VLT_provisional_animalonly.csv'))
+import_crop <- fread(file.path(intermediate_output_path, 'fao_VLT_crops_disaggregated.csv'))
+import_animal <- fread(file.path(intermediate_output_path, 'fao_VLT_animalonly.csv'))
 
 import_crop <- tidyr::separate(import_crop[scenario != '', ], scenario, into = c('D', 'scenario_diet', 'W', 'scenario_waste'), sep = '_')
 import_crop[, c('D','W') := NULL]

final_output/figs/fig1_revision.R

@@ -94,3 +94,23 @@ dev.off()
 pdf(file.path(fp_fig, 'pdfs/fig1.pdf'), height = 9, width = 6)
 grid.arrange(p_consprod_top, p_consprod_bottom, nrow = 2, heights = c(1, 2.6))
 dev.off()
+
+
+# Supplemental figure: absolute values ------------------------------------
+
+# Absolute values
+p_totaldemand_sums <- ggplot(totaldemand_sums %>% filter(scenario_waste %in% c('baseline', 'allavoidable')), aes(x = short_name, y = demand/1e9)) +
+  geom_col(aes(fill = kingdom), color = 'black', size = 0.25) +
+  facet_grid(scenario_waste ~ scenario_diet, labeller = scenario_labeller) +
+  scale_x_discrete(name = 'food category') +
+  scale_y_continuous(expand = expansion(mult = c(0, 0.03)), name = 'production (billion USD)') +
+  scale_fill_manual(values = setNames(okabe_colors[c(8, 4)], c('animal', 'plant'))) +
+  theme(panel.grid = element_blank(),
+        axis.text.x = element_text(angle = 45, hjust = 1),
+        legend.position = 'none')
+
+p_totaldemand_sums <- ggdraw(p_totaldemand_sums + theme(plot.margin = unit(c(25, 25, 5.5, 5.5), 'points'))) +
+  draw_label(label = 'diet scenario', x = 0.5, y = 0.97) +
+  draw_label(label = 'waste scenario', x = 0.99, y = 0.5, angle = -90)
+
+ggsave(file.path(fp_fig, 'total_consumption_10_scenarios.png'), p_totaldemand_sums, height = 9*.75, width = 12, dpi = 400)

final_output/figs/fig3.R

@@ -173,3 +173,6 @@ png(file.path(fp_fig, 'fig3.png'), height = 7, width = 7, res = 400, units = 'in
   grid.draw(gridExtra::gtable_rbind(ggplotGrob(p_top), ggplotGrob(p_bottom)))
 dev.off()
 
+pdf(file.path(fp_fig, 'pdfs/fig3.pdf'), height = 7, width = 7)
+  grid.draw(gridExtra::gtable_rbind(ggplotGrob(p_top), ggplotGrob(p_bottom)))
+dev.off()

final_output/figs/fig4_and_supplemental_figs_paneledscenariomaps.r

@@ -276,6 +276,21 @@ make_20panel_map_v2(map_panel_data = county_extinction_map_panels[land_use %in%
                     output_type = 'pdf'
 )
 
+# Fig 4 as png
+make_20panel_map_v2(map_panel_data = county_extinction_map_panels[land_use %in% 'total' & taxon %in% 'total' & scenario_waste %in% c('baseline', 'allavoidable')],
+                    base_map = county_map,
+                    region_type = 'county',
+                    variable = 'extinction_outbound_vs_baseline',
+                    file_name = 'fig4',
+                    scale_name = 'Change vs.\nbaseline',
+                    scale_factor = 1,
+                    scale_trans = 'identity',
+                    scale_type = 'divergent',
+                    ak_idx = county_ak_idx,
+                    hi_idx = county_hi_idx,
+                    add_theme = theme_void() + theme(legend.position = 'none'),
+                    n_waste = 2
+)
 
 ###
 # For each taxon, total land use, raw extinction outbound.

final_output/figs/load_data_all_figs.R

@@ -42,10 +42,6 @@ county_extinction_flow_sums <- read_csv(file.path(final_output_path, 'species_lo
 # Flows of species extinctions between states (raw pairwise)
 state_extinction_flows <- read_csv(file.path(final_output_path, 'species_lost_state_x_state_all_scenarios_med.csv'), col_types = 'cccccd')
 
-# Foreign virtual land transfers into USA, by country x tnc combination exporting them and by county importing them
-foreign_vlt_export <- read_csv(file.path(final_output_path, 'foreign_VLT_by_country_x_TNC.csv'))
-foreign_vlt_import <- read_csv(file.path(final_output_path, 'foreign_VLT_to_counties.csv'))
-
 # Foreign flows of species extinctions into USA, by country x tnc combination exporting them and by county importing them
 foreign_extinction_export <- read_csv(file.path(final_output_path, 'foreign_species_lost_by_export_country_x_tnc.csv'))
 foreign_extinction_import <- read_csv(file.path(final_output_path, 'foreign_species_lost_by_import_county.csv'))
@@ -54,6 +50,10 @@ foreign_extinction_import <- read_csv(file.path(final_output_path, 'foreign_spec
 foreign_animal_export <- read_csv(file.path(intermediate_output_path, 'fao_production_trade_animals.csv'))
 foreign_crop_export <- read_csv(file.path(intermediate_output_path, 'fao_production_trade_crops.csv'))
 
+# Foreign virtual land transfers into USA, by country x tnc combination exporting them and by county importing them
+foreign_vlt_export <- read_csv(file.path(final_output_path, 'foreign_VLT_by_country_x_TNC.csv'))
+foreign_vlt_import <- read_csv(file.path(final_output_path, 'foreign_VLT_to_counties.csv'))
+
 # Map of counties in AEA
 county_map <- st_read(file.path(spatial_output_path, 'USA_county_2014_aea.gpkg'))
 # Map of TNC ecoregions in AEA

final_output/figs/load_data_foreign_maps.R

@@ -18,13 +18,6 @@ foreign_vlt_countries <- foreign_vlt_export[, lapply(.SD, sum, na.rm = TRUE),
                                             by = .(scenario_diet, scenario_waste, country_name, ISO_A3, region_UN, subregion_UN), 
                                             .SDcols = patterns('^V')]
 
-foreign_vlt_countries[scenario_diet %in% 'baseline' & scenario_waste %in% 'baseline'][order(-VLT_pasture)]
-foreign_vlt_countries[scenario_diet %in% 'baseline' & scenario_waste %in% 'baseline'][order(-VLT_annual)]
-foreign_vlt_countries[scenario_diet %in% 'baseline' & scenario_waste %in% 'baseline'][order(-VLT_permanent)]
-
-# Top exporters of virtual land to the United States, baseline case.
-foreign_vlt_countries[scenario_diet %in% 'baseline' & scenario_waste %in% 'baseline']
-
 ### Compare domestic and foreign VLT into the USA.
 
 # Sum the foreign vlt across all countries.

final_output/figs/supplemental_figs_paneledforeignmaps.R

@@ -4,6 +4,8 @@
 source(file.path(code_path, 'final_output/figs/load_data_all_figs.R'))
 source(file.path(code_path, 'final_output/figs/load_data_foreign_maps.R'))
 
+library(scico)
+
 # Only make figures for animals, plants, and total
 # In total 16 figures will be generated. 
 # For extinction, 3 maps (animals, plants, total) x 2 (absolute and relative to baseline) = 6
@@ -224,3 +226,148 @@ for (landtype in landtypes_to_plot) {
   )
 
 }
+
+
+# Alternative version with vertical orientation ---------------------------
+
+make_world_panel_map_vertical <- function(map_panel_data, base_map, region_type, variable, file_name, n_waste = 4, percent_scale = TRUE, ...) {
+
+  args <- list(...)
+
+  # Calculate global range for fill scale for all plots and insets.
+  vals <- unlist(map(map_panel_data$data, function(x) x[[variable]]/args$scale_factor))
+  if (args$scale_trans == 'log10') vals <- vals[vals > 0]
+  global_scale_range <- range(vals, na.rm = TRUE)
+
+  # Create dummy plot with a legend so it can be extracted
+  scale_labs <- if (percent_scale) scales::percent else waiver()
+
+  if (args$scale_type == 'divergent') {
+
+    # Remap scale range to 0-1
+    global_scale_range_remap <- scale_begin_end(vals)
+
+    scale_main <- scale_fill_scico(palette = 'vik', begin = global_scale_range_remap[1], end = global_scale_range_remap[2], na.value = 'gray75', name = args$scale_name, trans = args$scale_trans, labels = scale_labs, limits = global_scale_range, guide = guide_colorbar(direction = 'horizontal'))
+  } else {
+    global_scale_range_remap = NULL
+    scale_main <- scale_fill_viridis_c(na.value = 'gray75', name = args$scale_name, trans = args$scale_trans, labels = scale_labs, limits = global_scale_range, guide = guide_colorbar(direction = 'horizontal'))
+  }
+
+  plot_leg <- get_legend(ggplot(mtcars, aes(x=cyl,y=hp,fill=mpg)) + geom_point() + scale_main + theme(legend.key.width = unit(1.5, 'cm')))
+
+  # Draw plots
+  maps_list <- map(map_panel_data$data, function(x) ggplot(left_join(base_map, x)) +
+                     geom_sf(aes_string(fill = variable), size = 0.1) +
+                     geom_sf(data = st_geometry(poly_countries), fill = NA) +
+                     scale_main +
+                     args$add_theme)
+
+  waste_labels <- waste_long_names$long_name
+  if (n_waste == 2) waste_labels <- waste_labels[c(1, 4)]
+
+  maps_laidout <- panel_plot(plots = maps_list, 
+                             y_labels = diet_long_names$long_name, 
+                             x_labels = waste_labels,
+                             y_title = 'diet scenario',
+                             x_title = 'waste scenario',
+                             global_legend = plot_leg,
+                             label_fontsize = 7,
+                             title_fontsize = 14,
+                             panel_width = 60,
+                             panel_height = 35,
+                             label_width = 5,
+                             title_width = 10,
+                             legend_height = 15)
+
+  png(glue('{fp_foreignmaps}/{file_name}.png'), width=16,height=24,res=200,units='cm')
+  grid.draw(maps_laidout)
+  dev.off()
+
+}
+
+# Extinction maps vertical ---------------------------------------------------------
+
+taxa_to_plot <- c('animals', 'plants', 'total')
+
+for (tax in taxa_to_plot) {
+  dat <- foreign_extinction_country_panels[taxon %in% tax & land_use %in% 'total']
+  dat[, scenario_diet := factor(scenario_diet, levels = diet_levels_ordered)]
+  dat[, scenario_waste := factor(scenario_waste, levels = waste_levels_ordered)]
+  dat <- dat[order(scenario_waste, scenario_diet)]
+  message(paste('Drawing maps for', tax))
+
+  # Outbound, relative, 10 panels
+  make_world_panel_map_vertical(map_panel_data = dat[scenario_waste %in% c('baseline', 'allavoidable')],
+                                base_map = country_map_toplot,
+                                region_type = 'country',
+                                variable = 'species_vs_baseline',
+                                file_name = glue('10scenariosvertical_country_extinction_{tax}_outbound_vs_baseline'),
+                                scale_name = 'Change vs.\nbaseline',
+                                scale_factor = 1,
+                                scale_trans = 'identity',
+                                scale_type = 'divergent',
+                                add_theme = theme_void() + theme(legend.position = 'none'),
+                                n_waste = 2
+  )
+
+  # Outbound, absolute, 10 panels
+  make_world_panel_map_vertical(map_panel_data = dat[scenario_waste %in% c('baseline', 'allavoidable')],
+                                base_map = country_map_toplot,
+                                region_type = 'country',
+                                variable = 'species_lost',
+                                file_name = glue('10scenariosvertical_country_extinction_{tax}_outbound'),
+                                scale_name = 'Virtual extinction\nexport',
+                                scale_factor = 1,
+                                scale_trans = 'log10',
+                                scale_type = 'sequential',
+                                percent_scale = FALSE,
+                                add_theme = theme_void() + theme(legend.position = 'none'),
+                                n_waste = 2
+  )
+
+}
+
+
+
+# Land maps vertical ---------------------------------------------------------------
+
+
+landtypes_to_plot <- unique(foreign_vlt_country_panels$land_use)
+
+for (landtype in landtypes_to_plot) {
+  dat <- foreign_vlt_country_panels[land_use %in% landtype]
+  dat[, scenario_diet := factor(scenario_diet, levels = diet_levels_ordered)]
+  dat[, scenario_waste := factor(scenario_waste, levels = waste_levels_ordered)]
+  dat <- dat[order(scenario_waste, scenario_diet)]
+  message(paste('Drawing maps for', landtype))
+
+  # Outbound, relative, 10 panels
+  make_world_panel_map_vertical(map_panel_data = dat[scenario_waste %in% c('baseline', 'allavoidable')],
+                                base_map = country_map_toplot,
+                                region_type = 'country',
+                                variable = 'VLT_vs_baseline',
+                                file_name = glue('10scenariosvertical_country_{landtype}_outbound_vs_baseline'),
+                                scale_name = 'Change vs.\nbaseline',
+                                scale_factor = 1,
+                                scale_trans = 'identity',
+                                scale_type = 'divergent',
+                                add_theme = theme_void() + theme(legend.position = 'none'),
+                                n_waste = 2
+  )
+
+  # Outbound, absolute, 10 panels
+  make_world_panel_map_vertical(map_panel_data = dat[scenario_waste %in% c('baseline', 'allavoidable')],
+                                base_map = country_map_toplot,
+                                region_type = 'country',
+                                variable = 'VLT',
+                                file_name = glue('10scenariosvertical_country_{landtype}_outbound'),
+                                scale_name = 'Virtual land\nexport (ha)',
+                                scale_factor = 1,
+                                scale_trans = 'log10',
+                                scale_type = 'sequential',
+                                percent_scale = FALSE,
+                                add_theme = theme_void() + theme(legend.position = 'none'),
+                                n_waste = 2
+  )
+
+}