A Database for Exploring Rarity Patterns and Processes

Please refer to Maciel and Nurk (2025) in order obtain a full understanding of the process involved in the building of this database.

This database comprised 662 woody species from 101 savanna locations. Ten open savanna sites measuring between 0.15 and 1 hectare, sixty-five woody savanna sites ranging from 0.18 to 1.8 hectares, and twenty-six savanna forest sites spanning from 0.1 to 1 hectare.

The species growth forms were obtained from BIEN and and Flora R package. All species are classified in growth forms (shrubs, trees, and trees/shrubs). A shrub is a tiny woody plant with several stems or branches growing from its base, whereas a tree is a tall plant with an extended stem or trunk that supports branches and leaves. When the species were classified in both growth forms on different sites, we classified them as shrubs/trees. The information on vegetation types was collected from the articles themselves, but when the study lacked this information, it was obtained from the map of the IBGE—Instituto Brasileiro de Geografia e Estatística. Each site has a vegetation type (open savanna, woody savanna, and savanna woodland) based on IBGE maps. In the IBGE maps, Open Savanna (OS) is defined by trees that are less than 7 meters tall and have a crown cover of up to 10%. Woody Savanna (WS) trees can grow to be 14 meters tall and have a crown cover of 40% to 60%. Savanna Woodland (SW) features trees taller than 15 meters and a crown cover greater than 70%. Sites that were located in places with different vegetation types were removed from further analysis steps. After filtering in this way, our database consisted of 60 studies published between 1986 and 2002, encompassing 101 sites.

The rgbif package was used to obtain the Global Biodiversity Facility (GBIF) occurrences for each species in January 2024. The occ_search() function was used to search for occurrences only in the Neotropical region. Next, the function myspecies_coords() was used to create a dataframe with the list of species and geographic coordinate information for each species and its synonyms. The flora package was accessed to standardize the names of the species according to the Flora of Brazil. The GBIF records with site records for each species were merged because the combination reduces the spatial gap and yields a database comprising 5,773 geographic occurrences. The geographic coordinates of each species were used to extract data on their occurrence in Olson ecoregions.

The rrindex package was used to estimate species rarity. The rrindex package creates continuous values for the Rabinowitz rarity dimensions: Geographic Range Index (GRI), Habitat Specificity Index (HSI), and Population Size Index (PSI). Each index has a range that goes from 0 to 1. Neotropical GRI was calculated using latitude and longitude. The HSI was calculated using ecoregions because they can capture species' ecoregion preference, provide a more precise indicator of species endemism, and inform global conservation policies. PSI was calculated using the number of sites where a species occurred in our dataset as a proxy for global abundance. The total of GRS, HSS, and PSI determined a species' rarity continuum, which ranged from 0 (common) to 3 (rare). The conclusion included two variables: species rarity continuum and species richness (rare and common) per site.

Data on aboveground and belowground biomass was collected from a global atlas for species and sites. The aboveground and belowground biomass were normalized to range between 0 and 1 for each species and site.