Geo-referenced spatial data for angiosperm species endemic to Brazil

The REFLORA 2021 and BFG (BFG 2020) databases are fed new data daily and edited for changes in nomenclature. The georeferencing work carried out here was developed between August 2018 and December 2019. Thus, the difference between the number of endemic species recorded in 2020 (i.e. about 18,000 species) and the number of georeferenced species provided here (i.e. 14,992 species) is supported by the following factors:

1. In August 2018, 1,393 species had no vouchers in REFLORA.
2. In order to obtain the highest possible accuracy in species occurrence data, we established editing procedures for the use of geographical distributions from the collection records (outlined below). These procedures made it impossible to include 1,615 species with inconsistencies in the collection records.

This georeferenced occurrence dataset for endemic species provides the basis for a wide range of biodiversity studies, for example, spatial studies conducted at various hierarchical levels, i.e. family, genus, species; effects of global change; changes in distributions of species; conservation; and systematics.

Conselho Nacional de Desenvolvimento Cientıfico e Tecnológico (CNPq) and FAPERJ - Fundação de Amparo à Pesquisa do Estado do Rio de Janeiro for the postdoctoral fellowship granted to JGS. RCF received a Research Productivity Fellowship from CNPq (proc.303420/2016-2) and FAPERJ (processes n° E-26/202.778/2018) through Programa Cientista do Nosso Estado.

Brazilian angiosperms dataset

Species list compilation:

The list of species was established in two phases. First, the initial list of names of all endemic species of Angiospermae was generated through the BFG in the Brazilian Flora (BFG 2020) website (http://floradobrasil.jbrj.gov.br), edited by several taxonomic experts in each sampled family, using the following search filters: (1) Group: Angiospermae; (2) Occurs in Brazil: yes; (3) Occurrence: only occurs in Brazil; (4) Endemism: only endemic to Brazil; and (5) Origin: native (Fig. 1).

Figure 1.  

Taxonomic and geographic data refinement workflow. Steps of data filtering to obtain the endemic angiosperm species list for Brazil, based on the list available from BFG 2018 (Filardi et al. 2018), in the Brazilian Flora 2020 website and records from Reflora Herbarium Virtual (2018-2019).

Based on this list of all endemic Brazilian angiosperm species retrieved from the BFG floristic database between August 2018 and October 2019, all occurrence records were downloaded from the REFLORA 2021 virtual herbarium (www.reflora.jbrj.gov.br) from 73 herbaria (ALCB, ASE, B, BRBA, CEN, CEPEC, CESJ, CGMS, COR, CRI, DVPR, E, EAC, ECT, ESA, EVB, FIG, FLOR, FURB, GH, HACAM, HBR, HCF, HDCF, HEPH, HRCB, HSTM, HTO, HUCO, HUCP, HUEFS, HUEM, HUEMG, HUENF, HUFU, HUNEB, HUNI, HUPG, HVASF, IAN, IBGE, ICN, K, LUSC, MAC, MBM, MBML, MG, MO, MUFAL, NY, P, PEL, PMSP, R, RB, RBR, REAL, RFA, UERJ, RFFP, RON, S, SJRP, SPF, UB, UFRN, UNIP, UNOP, UPCB, US, VIES and W, the herbaria acronyms following Thiers (2020), continuously updated). After these two phases, 18,000 endemic species to Brazil were identified, corresponding to the raw database. Producing this consolidated dataset involved 1 1/2 years of detailed revision of coordinates and nomenclaturally updating of the names in these datasets, as follows:

We created a protocol to clean the datasets (Fig. 1), the data were processed carefully by checking nomenclatural status and excluding records with erroneous occurrence data. The accuracy of species identification follows the list of endemic species of BFG 2020. Four steps were conducted for cleaning the taxonomic data. In the first step, we checked and cleaned the data taxonomically and nomenclaturally; only vouchers identified to species level about which we were uncertain were removed, including ‘cf.,’ ‘aff.,’ ‘sp.,’ and ‘spp.’. In the second step, we corrected the spelling of taxon names, which, for some species, had multiple entries with different spellings. In the third step, varieties and subspecies were grouped at the species level. In the fourth step, hybrids were excluded, synonyms were checked and accepted names were adopted according to the BGF. We performed the first four steps using the “filter” tool in Microsoft Excel v. 14.5 (Microsoft Office 2010 Proofing Tools).

Subsequently, we conducted manual cleaning procedures on the records. For cleaning the records, three steps were performed on the geographic data. In the first step, records of specimens with imprecise or vague descriptions of locations (e.g. Negro River, north coast, south coast) and incomplete (e.g. Amazonia, Bahia, Brazil) or incongruent information concerning locations (e.g. with no administrative unit, location in the ocean) were excluded. In the second step, we removed the taxonomic duplicates and records of duplicate samples with the same species name and place of occurrence and voucher information. In the final dataset, each record corresponds to a single herbarium specimen for which the geographical location has been checked and is unique to that locality. Duplicates were removed from the list, based on locality, collector name, collector number and the year in which the sample was collected. After data cleaning, the total number of records dropped from 827,016 to 183,201 occurrence records with complete voucher information.

The use of GPS became more widespread in 1995-1996, but there were still few satellites at that time (Kaplan 2005). Given that the occurrence records for all species endemic to Brazil were collected mainly over the last two centuries, it was not surprising that more than 75% records were not georeferenced. Hence, in the third step, we manually edited and included the coordinates of each voucher, based on databases of localities and municipalities maintained by the Brazilian Institute of Geography and Statistics website (IBGE) (http://mapas.ibge.gov.br), for 161,563 occurrence records of 14,992 endemic angiosperm species. For 21,632 records, it was not possible to perform georeferencing due to lack of sufficient information on the voucher. In this step, we removed the complete voucher data, since the main objective concerns the use of the dataset for biogeographical analysis. We performed the three steps using the “filter” tool in Microsoft Excel v. 14.5 (Microsoft Office 2010 Proofing Tools).

The final checklist is composed of native and endemic angiosperms and includes only vouchers identified to the species level, based on the Brazilian Flora (BFG 2020) and complete records, based on REFLORA. The complete list of vouchers, including all geographical duplicates (duplicate samples for same location) and photos to check the identity of the species, is available at REFLORA 2021(http://reflora.jbrj.gov.br/).

The geographic coverage encompasses the national territory of Brazil, which extends from 5° to -34° Latitude; -34° to -73° Longitude and covers a total area of approximately 8.5 million km² (IBGE). The dataset comprised all species of Angiospermae found exclusively in Brazil and it contains occurrence records in six phytogeographic domains, i.e. Amazonia, Caatinga, Cerrado, the Atlantic Forest, Pampa and Pantanal, in Chacoan, Parana, South Brazilian and South-eastern Amazonian dominions (Fig. 2, sensu Morrone (2014)).

Figure 2.  

Spatial distribution of angiosperms for all georeferenced data available at the Reflora Herbarium Virtual after data cleaning.

-34 and -5° Latitude; -73° and -34 Longitude.

To facilitate the search for taxa at different hierarchical levels, the dataset comprises three different worksheets of specimens collected over the past two centuries organised according to APG IV classification (Chase et al. 2016) and these have been organised alphabetically, as follows:

(1st Worksheet) A total of 649 species of basal angiosperms belonging to five orders, i.e. Canellales, Laurales, Magnoliales, Nymphaeales and Piperales from 13 families and 50 genera. Number of records is georeferenced by order in Fig. 3A.

Figure 3.  

Number of records georeferenced for endemic angiosperm species in Brazil on a municipality level, by order for: A. basal angiosperms; B. monocots; and C. eudicots.

(2nd Worksheet) A total of 3,854 species of monocots belonging to nine orders, i.e. Alismatales, Arecales, Asparagales, Commelinales, Dioscoreales, Liliales, Pandanales, Poales and Zingiberales from 32 families and 370 genera. Number of records is georeferenced by order in Fig. 3B.

(3rd Worksheet) A total of 10,489 eudicots, belonging to 31 orders, i.e. Apiales, Aquifoliales, Asterales, Boraginales, Brassicales, Caryophyllales, Celastrales, Cornales, Cucurbitales, Dilleniales, Dipsacales, Ericales, Escalloniales, Fabales, Gentianales, Geraniales, Gunnerales, Lamiales, Malpighiales, Malvales, Myrtales, Oxalidales, Picramniales, Proteales, Ranunculales, Rosales, Santalales, Sapindales, Solanales, Vitales and Zygophyllales from 128 families and 1,199 genera. Number of records is georeferenced by order in Fig. 3C.