Analyze alpha diversity by year from bugsigdb.org

alphadiversity.R

library(bugsigdbr)
bsdb <- importBugSigDB(version = "devel")

# Create a stacked barplot of the proportion of Pielou, Shannon, Chao1, Simpson, Inverse Simpson, and Richness as a function of year
library(tidyverse)
bsdb_by_year <- bsdb |>
  filter(Year > 2014) |>
  dplyr::group_by(Year) |>
  dplyr::summarize(
    Pielou = sum(!is.na(Pielou)) / n(),
    Shannon = sum(!is.na(Shannon)) / n(),
    Chao1 = sum(!is.na(Chao1)) / n(),
    Simpson = sum(!is.na(Simpson)) / n(),
    InverseSimpson = sum(!is.na(`Inverse Simpson`)) / n(),
    Richness = sum(!is.na(Richness)) / n()
  )
bsdb_by_year_long <- bsdb_by_year |>
  tidyr::pivot_longer(
    cols = c(Pielou, Shannon, Chao1, Simpson, InverseSimpson, Richness),
    names_to = "Metric",
    values_to = "Value"
  )
ggplot2::ggplot(bsdb_by_year_long, ggplot2::aes(x = Year, y = Value, fill = Metric)) +
  ggplot2::geom_col(position = "stack") +
  ggplot2::labs(title = "Reporting of diversity measures by year",
                x = "Year",
                y = "Sum of fractions reported") +
  ggplot2::theme(legend.position = "bottom")

# Create a line plot from bsdb_by_year_long of all numeric columns by year
png("diversity_by_year.png", width = 4, height = 3.25, units = "in", res = 600)
ggplot2::ggplot(bsdb_by_year_long, ggplot2::aes(x = Year, y = Value*100, color = Metric)) +
  ggplot2::geom_line(linewidth=1.5) +
  ggplot2::labs(title = "diversity measures over time",
                x = "Year",
                y = "% studies reporting") +
  ggplot2::theme(legend.position = "bottom")
dev.off()
system("open diversity_by_year.png")

lm(Shannon ~ Year, data = bsdb_by_year) |>
  summary()