Tapina villosa (Gesneriaceae) was published by George Gardner in 1842, based on material he collected in Serra de Natividade (Tocantins, Brazil) in 1840. The species is now recognized as Goyazia villosa (Gardner) R.A. Howard. Since Gardner’s travels in Central Brazil, this species had not been collected again and the taxon was considered as possibly extinct. After a long time, we report the discovery of two new populations of G. villosa in the municipality of Palmas and in the Estação Ecológica Serra Geral do Tocantins, ca. 200 and 100 km north of the Serra de Natividade, respectively. The newly collected materials allow us to better characterize the morphology and infer the phylogenetic placement of this poorly-known species. Here we demonstrate that G. villosa is closely related to G. rupicola in a clade including Goyazia and Mandirola species, and provide an updated description of the species, including field images, photographic plate, information on its distribution and habitat, and a taxonomic key for the species of Goyazia.

Cerrado, Gloxiniinae, protected area, savanna

George Gardner (1812–1849), a Scottish botanist, intensively explored the Brazilian Cerrado between the years 1836 and 1841, collecting thousands of plant specimens. He also described and published many new plant species from this phytogeographic region, most of them published in the illustrated botanical magazine named Icones Plantarum (Gardner 1842). Between October 1839 and February 1840, Gardner stayed at Vila de Natividade, a locality today known as the county town Natividade in Tocantins State (Fig. 1), to explore and describe the flora and geology of the mountains east of the town (Gardner 1942: 283). It is where Gardner collected plant specimens that he later described as Tapina villosa Gardner. This species is currently classified as Goyazia villosa (Gardner) R.A. Howard and was until now only known by Gardner’s type collection.

Figure 1. 

Location of the type collection of Goyazia villosa made in 1840 (circle) and new records made in the municipality of Palmas in 2017 (diamond) and Estação Ecológica Serra Geral do Tocantins (EESGT) in 2021 and 2022 (triangle). Brazilian states: BA (Bahia), MA (Maranhão), PI (Piauí), TO (Tocantins). Track of G. Gardner based on the Atlas dos Viajantes no Brasil (https://viajantes.bbm.usp.br/). Map produced with QGIS (QGIS Development Team 2024).

Since the first description of Tapina villosa, this species was subsequently classified into different genera: Tapeinotes villosa (Gardner) Walp., Ligeria villosa (Gardner) Hanst., Anetanthus villosus (Gardner) Benth. & Hook.f. ex B.D. Jacks., and Gloxinia villosa (Gardner) Wiehler. The latter is an illegitimate name because Gloxinia villosa (Gardner) Wiehler is a later homonym with Gloxinia villosa (Lindl.) Mart., which is currently placed in Sinningia villosa Lindl. The current classification of Tapina villosa in Goyazia Taubert was proposed by Howard (1975), based on characters such as “shape and lobing of the corolla, adherence of all four anthers, pubescent ovary, and the scaly rhizomes”. The genus Goyazia also includes two other species endemic to Brazil (G. rupicola Taub. and G. petraea (S.M. Phillips) Wiehler) that are also found on rocky outcrops within the Cerrado phytogeographic domain (Araujo 2007, 2024). The genus belongs to the subtribe Gloxiniinae and has strong morphological and phylogenetic affinities with Mandirola Decne. (Araújo et al. 2010). Currently, Goyazia villosa is an accepted name in the Flora e Funga do Brasil (Araujo 2024), although Roalson et al. (2005) treated the species as incertae sedis and possibly related to Phinaea Benth. Therefore, the taxonomic placement of G. villosa was ambiguous and needed to be reexamined based on additional morphological and phylogenetic data.

Recent floristic exploration in the savannas of the Jalapão region (Tocantins) has resulted in a large number of new species described, as well as new records of poorly known taxa (e.g. Yamamoto et al. 2008; Araújo et al. 2016; Borges and Antar 2016; Mendes et al. 2017; Antar et al. 2018; Pastore and Antar 2021; Santana and Simon 2022). However, most of this vast area remains poorly collected (Antar and Sano 2019; Santana and Simon 2022). To fill gaps in floristic knowledge, we organized expeditions to different localities in the state of Tocantins in 2017, and also carried out a survey of the vascular flora of the Estação Ecológica Serra Geral do Tocantins (EESGT) between 2018 and 2022. Several floristic novelties that resulted from these expeditions have already been published (Alves et al. 2018; Guimarães et al. 2021; Mendes et al. 2022; Schindler et al. 2023).

In the context of these expeditions, we discovered two populations of G. villosa in the central and eastern parts of Tocantins, more than a century and a half after Gardner first collected this species. In this paper, we present an updated and detailed morphological description of G. villosa including field images, distribution maps, habitat preferences, as well as a taxonomic key for the species of Goyazia. Phylogenetic analyses were also conducted using newly obtained nrDNA ITS sequences to verify the placement of G. villosa within the Gloxiniinae.

In 2005 and 2013, expeditions aiming to collect G. villosa were carried out in the type locality at Serra de Natividade and surrounding areas by two of the authors (AOA and AC). However, no plants of G. villosa were located in this area. Subsequent field expeditions were carried out in different parts of the state of Tocantins in 2017. Field expeditions to survey the flora of EESGT were conducted by three of the authors (BS, MF, and MFS) between 2018 and 2022 (Schindler et al. 2023). EESGT is a large protected area (716,306 hectares) located in the eastern portion of the state of Tocantins and the extreme west of the state of Bahia (Fig. 1). According to Köppen’s classification, the climate corresponds to a tropical climate with dry winter, with rain from November to April and marked dry season from May to October (Aw) (Alvares et al. 2014), and altitudes range from 400 to 730 m. Vegetation types in EESGT include savannas and grasslands, which dominate the landscape, as well as riverine forests and swamps along water courses (Franke et al. 2018).

Observations and photographic records of the species were carried out in the field in 2017, 2021, and 2022. Collections were deposited in the CEN, G, R, RB, and SORO herbaria (acronyms according to http://sweetgum.nybg.org/science/ih/). Identification of the specimens collected was based on the protologue (Gardner 1842) and comparisons with digital images of the types deposited at BM, CGE, E, FI, G, K, P, and W. Morphological terminology used in the description follows Beentje (2010) and Font Quer (2001) for general morphology, and Ellis et al. (2009) for leaf venation. The assessment of the species conservation status followed the IUCN guidelines (IUCN 2024) using criterion B (geographic range). Estimates of Area of Occupancy (AOO) and Extent of Occurrence (EOO) were calculated using QGIS, version 3.28 (QGIS Development Team 2024). We used land cover maps from the Mapbiomas platform (Projeto MapBiomas, 2024) to estimate G. villosa habitat availability and reduction based on the variation of native vegetation and anthropic areas within the EOO of the species between 1986 and 2021.

The ribosomal DNA internal transcribed spacer (ITS) of a newly collected sample of G. villosa (B. Schindler et al. 41) was sequenced to estimate the phylogenetic placement of this species within the Gloxiniinae. Leaves from a plant collected in the field were immediately dried in silica gel. DNA was extracted from tissues samples using the SILEX method (Vilanova et al. 2020). Amplification and sequencing of ITS follow the procedures described in Araújo et al. (2010). The newly acquired sequence was deposited in EMBL/GenBank (no PP468351) and was added to the ITS alignment generated by Araújo et al. (2010), which includes 13 Gloxiniinae genera. Maximum likelihood analysis of this dataset was conducted using the software RAxML (Stamatakis 2014) through the web-server of the Swiss Institute of Bioinformatics (Kozlov et al. 2019). The unpartitioned sequence alignment was analyzed using the GTR + GAMMA model. The robustness of the tree was evaluated with bootstrap resampling and 1000 replicates.

The rediscovery of G. villosa was made possible by active search and botanical exploration of moist cliffs, and its very specialized habitat. These cliffs are known for hosting vascular plant assemblages, which typically include small species of Gesneriaceae, Pitcairnia (Bromeliaceae), Anemia (Anemiaceae), and liverworts. Gardner’s description of the locality of G. villosa as ‘dry’ may refer to the rocky clefts providing protection from direct insolation, thereby maintaining a relatively humid microclimate. The populations rediscovered between 2017 and 2022 were growing in very humid conditions. These occurrences confirm that the species still persists in the wild and calls for more intensive field surveys of the Cerrado’s endemic flora. Analysis of the recent collections of G. villosa allowed a better understanding of the morphological traits that characterize this species. The presence of a scaly rhizome is illustrated here for the first time (Fig. 2B). This feature, together with the small habit of the plant, the coherent four anthers, and the pubescent ovary are congruent with the placement of this species in the genus Goyazia (Howard 1975). The species can be easily differentiated from the two congeners (G. petraea and G. rupicola) by serrate leaf margin and long (1.8–3.5 cm) pedicels (Figs 2, 3D).

Recent collections also provided material for ongoing studies on pollen micromorphology that preliminarily show that G. villosa shares many characters with the two other species of Goyazia and the three species of Mandirola (Souza et al., in prep.). In agreement with morphology, our preliminary phylogenetic analysis based on a single nuclear marker (ITS) places G. villosa together with G. rupicola, the type species of the genus, in a well-supported clade (BS = 100%) including species of Goyazia and Mandirola (Fig. 4). A natural intergeneric hybrid was recently described under the name Goydirola A.O. Araujo & M. Peixoto, indicating the close relationships between the two genera (Araujo et al. 2021). However, a more thorough study of the phylogenetic relationships based on a larger number of genes (Ogutcen et al. 2021) and a larger infraspecific sampling (Fiorini et al. 2020) is still needed to clarify the generic circumscription of Goyazia and Mandirola and to define the closest relative species of G. villosa.

We thank EESGT managers (ICMBio), Ana Carolina S. Barradas, Marco A. Borges, and Maximo M. Costa, for logistical support; all EESGT rangers, especially Walter B. Silva, who accompanied us on all expeditions; Valdeci F. Gomes (Dudu), Sérgio E. Noronha, and Raiana R. Souza, for assistance in the field and conservation assessment analysis; and the curator and staff of herbarium CEN for sending duplicates of Goyazia villosa. We also thank Gustavo Shimizu and John L. Clark for carefully reviewing the manuscript. RJVA and NGS thank the Lira family from Rio Sono for logistic support.