Query Movebank for taxonomic coverage

movebankstudies.R

library(move2)
library(ritis)
library(sf)
library(tidyverse)

# Download movebank study-level information
movebank_studies <- movebank_download_study_info(
  # Filter to studies where data are *visible* (not necessarily downloadable)
  i_can_see_data = TRUE,
  # Specify the list of attributes we want to query
  # See: https://github.com/movebank/movebank-api-doc/blob/master/movebank-api.md
  attributes = c("name", 
                 "id", 
                 "taxon_ids",        # comma-separated, if multiple
                 "sensor_type_ids",  # comma-separated, if multiple
                 "number_of_deployments",
                 "timestamp_first_deployed_location", 
                 "timestamp_last_deployed_location",
                 "main_location_lat", 
                 "main_location_long",
                 "license_type",
                 "i_have_download_access")
)
# Convert to long format. When studies contain multiple taxon_ids and/or
# sensor_type_ids, convert them to multiple rows.
movebank_long <- movebank_studies %>% 
  separate_longer_delim(taxon_ids, ",") %>% 
  drop_na(taxon_ids) %>% 
  separate_longer_delim(sensor_type_ids, ",") %>% 
  select(name, id, taxon_id = taxon_ids, sensor_type_id = sensor_type_ids)


# Utility functions

# Wrapper-function for ritis::itis_search() to get the taxonomic hierarchy from
# a scientific binomial
# Side note: goodness gracious the ITIS API is frickin' indecipherable
get_rank_hierarchy <- function(taxon) { 
  itis_search(q = sprintf("nameWOInd:/%s/", taxon))$hierarchySoFarWRanks
}

# Pull out the specific rank (class, order, family, etc) from the string
# returned by ITIS
extract_rank <- function(rank, hierarchy) {
  str_extract(hierarchy, sprintf("%s:([A-z]+)", rank), group = 1)
}


# Query ITIS

# Fetch class, order, and family for every taxa
# Note: this queries ITIS seperately for every row. It takes forever.
movebank_taxa <- movebank_long %>% 
  filter(str_detect(taxon_id, "[A-z]+[ ]{1,1}[A-z]+")) %>% 
  mutate(rank_hierarchy = map_chr(taxon_id, get_rank_hierarchy),
         class = extract_rank("Class", rank_hierarchy),
         order = extract_rank("Order", rank_hierarchy),
         family = extract_rank("Family", rank_hierarchy))


# A few summary tables

# Studies by class. 3x more birds than mammals.
movebank_taxa %>% 
  group_by(class) %>% 
  summarize(n_studies = n_distinct(id), .groups = "drop") %>% 
  arrange(desc(n_studies))

# Studies by order. Top 3 bird orders: Charadriiformes (shorebirds, gulls,
# auks), Accipitriformes (hawks, eagles, kites), Anseriformes (ducks, geese,
# swans). Top 3 mammal orders: Carnivora (carnivores), Artiodactyla (even-toed
# ungulates), Chiroptera (bats).
movebank_taxa %>% 
  group_by(class, order) %>% 
  summarize(n_studies = n_distinct(id), .groups = "drop") %>% 
  arrange(desc(n_studies))


# A few simple figures

# Number of studies started per year, color-coded by class
movebank_taxa %>% 
  left_join(select(movebank_studies, id, timestamp_first_deployed_location),
            by = "id") %>% 
  mutate(first_year = lubridate::year(timestamp_first_deployed_location)) %>% 
  group_by(class, first_year) %>% 
  summarize(n_studies = n_distinct(id), .groups = "drop") %>% 
  drop_na() %>% 
  ggplot(aes(first_year, n_studies)) +
  geom_line(aes(color = class)) +
  xlim(2008, 2024) +
  theme_bw()

# Map of studies
movebank_taxa %>% 
  left_join(select(movebank_studies, id, main_location),
            by = "id") %>% 
  rename(geom = main_location) %>% 
  st_as_sf() %>% 
  ggplot() +
  geom_sf(data = rnaturalearth::countries110, fill = NA) +
  geom_sf(aes(color = class), size = 0.5, alpha = 0.5) +
  theme_minimal()