SpeciesGeoCoder: Fast Categorization of Species Occurrences for Analyses of Biodiversity, Biogeography, Ecology, and Evolution

Abstract

Understanding the patterns and processes underlying the uneven distribution of biodiversity across space constitutes a major scientific challenge in systematic biology and biogeography, which largely relies on effectively mapping and making sense of rapidly increasing species occurrence data. There is thus an urgent need for making the process of coding species into spatial units faster, automated, transparent, and reproducible. Here we present SpeciesGeoCoder, an open-source software package written in Python and R, that allows for easy coding of species into user-defined operational units. These units may be of any size and be purely spatial (i.e., polygons) such as countries and states, conservation areas, biomes, islands, biodiversity hotspots, and areas of endemism, but may also include elevation ranges. This flexibility allows scoring species into complex categories, such as those encountered in topographically and ecologically heterogeneous landscapes. In addition, SpeciesGeoCoder can be used to facilitate sorting and cleaning of occurrence data obtained from online databases, and for testing the impact of incorrect identification of specimens on the spatial coding of species. The various outputs of SpeciesGeoCoder include quantitative biodiversity statistics, global and local distribution maps, and files that can be used directly in many phylogeny-based applications for ancestral range reconstruction, investigations of biome evolution, and other comparative methods. Our simulations indicate that even datasets containing hundreds of millions of records can be analyzed in relatively short time using a standard computer. We exemplify the use of SpeciesGeoCoder by inferring the historical dispersal of birds across the Isthmus of Panama, showing that lowland species crossed the Isthmus about twice as frequently as montane species with a marked increase in the number of dispersals during the last 10 million years.

ConR: An R package to assist large‐scale multispecies preliminary conservation assessments using distribution data

“…It is important to stress that the accuracy of the georeferencing in such datasets is crucial for estimating risk. Two recently released R packages, Biogeo (Robertson, Visser, & Hui, ) and speciesgeocode R (Töpel et al., ), are designed to help curate and clean large datasets, and are thus complementary to ConR .…”

“…); and (3) extreme fluctuation of certain aspects of the taxon's distribution (Table ). Calculation of the two key range parameters, EOO and AOO, can be easily automated either using a taxon‐by‐taxon approach, as provided for by the web service GeoCAT (Bachman, Moat, Hill, de la Torre, & Scott, ), or in batch mode, for example in other R packages such as speciesgeocodeR (Töpel et al., ) or RED (https://CRAN.R-project.org/package=red; see Table ).…”

The grass subfamily Pooideae: Cretaceous–Palaeocene origin and climate‐driven Cenozoic diversification

“…Georeferenced records for Pooideae and outgroup taxa were downloaded from the Global Biodiversity Information Facility (GBIF.org, , ) using the rgbif package in R (Chamberlain & Boettiger, ). To exclude unreliable records, we discarded coordinates with fewer than three decimals and used additional filtering implemented in the SpeciesGeoCoder package (Töpel et al, ) in R. In short, we discarded records where coordinates: (a) were not valid (not part of the coordinate system, or marine coordinates); (b) were exactly or close to zero (threshold 0.5); (c) were the same for latitude and longitude; (d) had the same values as the capital of the country; (e) lay outside the polygon of the country; or (f) had the same value as the GBIF institutions. Taxa with synonymous names were merged using taxonomic information from GBIF.…”

Historical Biogeography of endemic seed plant genera in the Caribbean: Did GAARlandia play a role?

“…We defined five biogeographic operational areas: (A) Antilles; (B) Central America; (C) South America; (D) North America; and (E) rest of the world (Figure ). Species distributions were coded using the R implementation in the software package SpeciesGeoCoder v.1.0‐4 (Töpel et al., ). The output presence–absence matrix was visually inspected and corrected manually for erroneous assignments.…”