Among the phytophysiognomies of the Cerrado domain, the cerrado rupestre subdivision is the rarest, least known, and at the highest risk of disappearing in the coming decades, owing mainly to human pressure (Fernandes et al. 2016; Silveira et al. 2016; Pereira and Fernandes 2022). Cerrado rupestre is characterised by mainly shrub-like vegetation, on slopes and rocky outcrops above 800 m elevation on soils originating from the decomposition of sandstone, ironstone, and quartzite (Pereira and Fernandes 2022). Despite some differences such as floristic composition and tree cover, the shrub-tree cerrado rupestre and the grassland campo rupestre are located side by side. They share similar edaphic and climate conditions, and are often confused (Pereira and Fernandes 2022). In the Brazilian central plateau, one of the most extensive areas of cerrado rupestre is found along the highlands of the Chapada dos Veadeiros region (Lima 2008; Nascimento and Sano 2010; Pereira and Fernandes 2022).

The Chapada dos Veadeiros is considered an exceptional area regarding biodiversity and has a high level of endemism. For instance, there are several species of Arecaceae belonging to several genera in this region, including acaulescent species (Mendonça et al. 2007; Noblick and Lorenzi 2010; Noblick 2017; SpeciesLink Network 2023). In addition, the region is a recognised origin of rare acaulescent Syagrus species (Noblick and Lorenzi 2010; Noblick 2017). As stated by Cássia-Silva et al. (2022), acaulescence likely promotes speciation in neotropical seasonally dry neotropical habitats, which probably explains the occurrence of these rare, acaulescent, and microendemic species of Syagrus (see Noblick and Lorenzi 2010; Noblick 2017). Some voucher specimens collected in this region correspond mainly to an acaulescent Butia species (SpeciesLink Network 2023). Thus, for this study, it was of great interest to focus sampling efforts in this region and to search for specimens in situ to investigate them further.

Over the last two decades, most of the acaulescent species of Butia were transferred from the genus Syagrus. Such was the case for Butia campicola (Barb.Rodr.) Noblick, Butia leptospatha (Burret) Noblick, Butia exospadix Noblick, and Butia marmorii Noblick (Noblick 2004, 2006). During this time, the genus Butia comprised only five acaulescent species, including Butia microspadix Burret (Noblick 2006). Lastly, Butia pubispatha Noblick & Lorenzi (Noblick 2010) was described, and Butia poni Hauman ex Burret (Deble et al. 2017) has been resurrected and epitypified. Most recently, Butia buenopolensis B.F.Sant’Anna-Santos was discovered (Sant’Anna-Santos 2021). So, there are currently 22 accepted species of Butia (Noblick 2010; Soares 2015; Deble et al. 2017; Sant’Anna-Santos 2021), and more than a third of the genus is composed of acaulescent species. However, it is likely that new species of Butia are to be found (Noblick 2014). For instance, at least one new acaulescent species of Butia is known and remains to be described (Soares 2015: fig. 17).

The Neotropical genus Butia is a well-defined monophyletic genus ranging from Central Brazil to Uruguay, Paraguay, and Argentina (Meerow et al. 2009; Noblick 2010, 2014; Soares 2015). The genus is composed of taxa showing complex taxonomic delimitations due to the extremely variable morphology (Noblick 2010, 2014). So, ambiguous species attributions, dubious names, unsolved species complexes, and hitherto misidentified species are commonly pointed out in the literature (see Henderson et al. 1995; Noblick 2010, 2014; Soares 2015; Deble et al. 2017; Sant’Anna-Santos 2021). Morphologically, the Butia genus can be recognised by the pinnate leaves; the pinnae forming a “V” with each other in the leaf rachis, frequently with clustered ramenta on the lower vein; the peduncular bract is generally smooth or superficially striate; inflorescences solitary and interfoliar, strongly protandrous; the filaments are often inflexed at the tips; petal tips of pistillate flowers are only slightly valvate; endocarp with 3 pores equatorial or subequatorial (Silberbauer-Gottsberger 1973; Henderson et al. 1995; Dransfield et al. 2008; Noblick 2010, 2014; Soares 2015). However, even the most experienced specialists find it difficult to differentiate some species of Butia and, furthermore, Butia can be easily confused with Syagrus. In this context, leaf and flower microscopic characters associated with morphological characters have been used to corroborate the identification of Butia and the limits of several species (Glassman 1979; Noblick 2010, 2014; Tomlinson et al. 2011; Martel et al. 2013; Sant’Anna-Santos et al. 2015, 2018; Noblick and Sant’Anna-Santos 2021; Sant’Anna-Santos 2021). This approach helps to better understand the evolutionary and the auto-ecology of the genus (Sant’Anna-Santos 2021).

During fieldwork in the Chapada dos Veadeiros region, the discovery of unrecognised acaulescent specimens of Butia raised the possibility of whether they represent a new species. After morphological and anatomical analysis collected in situ, it was revealed that the new specimens indeed constitute the first new species of Butia endemic to Chapada dos Veadeiros. Because the conservation unit of that region has been the target of a bill being processed in the Brazilian Parliament to reduce its area by 73%, it is essential that all botanical novelties, especially of rare endemic species are brought to the attention of the scientific community and present evidence to strongly justify and support the conservation efforts being made for this vulnerable habitat.

This study was based on field observations in a cerrado rupestre area in the municipality of Alto Paraíso de Goiás, Goiás State, Brazil. This study also includes information from the specific literature (Glassman 1979; Henderson et al. 1995; Noblick 2004, 2006, 2010, 2011, 2014; Deble and Marchiori 2006; Soares and Longhi 2011; Deble et al. 2011, 2012a, 2012b, 2017; Marchiori 2015a, 2015b; Sant’Anna-Santos et al. 2015; Soares 2015; Noblick and Sant’Anna-Santos 2021; Sant’Anna-Santos 2021; Ressel and Guilherme 2022), herbaria (DIAM, IBGE, Rio Verde Herbarium (Instituto Federal Goiano, campus Rio Verde), MBM, UPCB, SPF, UFG, HEPH, HDCF, RB, Montes Claros Herbarium (Universidade Estadual de Montes Claros, Montes Claros)), and online databases (UB, NY, US, HUFU, MBML, CEN, HUEFS). Morphological and anatomical terminology follows Dransfield et al. (2008), Tomlinson et al. (2011), and Noblick and Sant’Anna-Santos (2021). The circumscription adopted here for Butia is based on Soares (2015), Deble et al. (2017), and Sant’Anna-Santos (2021).

Qualitative and quantitative morpho-biometric data and images were recorded in situ for the new species. Data were taken from 20 randomly chosen individuals of the new species. For Butia archeri (Glassman) Glassman, B. buenopolensis, and Butia purpurascens Glassman, the morphological and anatomical data were retrieved from the literature (Glassman 1967,1979; Noblick 2010, 2014; Soares 2015; Sant’Anna-Santos et al. 2018; Sant’Anna-Santos 2021). For B. buenopolensis, presently considered the most closely allied species of the new species, the known population was visited to record images to show the distinctive features that contrast with the new species in Fig. 7. Samples of flowers of the new species were collected in the field and immediately analysed while fresh, and also stored in ethyl alcohol for further stereomicroscopic analysis (Bioptika L60T) coupled with a CMOS 12mp PLUS camera.

The distribution map was plotted using QGIS v.3.22.13 (QGIS Development Team 2022), using the following data sources: states (IBGE 2020) and elevation (Farr et al. 2007). The distribution data for B. archeri, B. buenopolensis, and B. purpurascens were retrieved from the following herbaria vouchers: DIAM3156 and IBGE84505 for B. archeri, IBGE83300 for B. buenopolensis, RB745893 and SPF140854 for B. purpurascens. The conservation status of the new species was evaluated according to the IUCN guidelines and criteria (IUCN 2022) using the GeoCAT tool (Bachman et al. 2011).

The pinna anatomical study was based on 15 samples—collected from the middle of the central pinnae—from herbarium specimens of the collections used to describe the new species and specimens in the field following Sant’Anna-Santos (2021). After rehydration following Meira and Martins (2003), freehand sections of the midrib, intermediate region, and margin were directly mounted in water and observed under a light microscope (Bioptika B20+) coupled to a digital camera (CMOS 12mo PLUS). To better visualise the raphides, the intermediate region of the pinnae was also longitudinally sectioned. For B. archeri, B. buenopolensis, and B. purpurascens, the leaf anatomy data was retrieved from Sant’Anna-Santos et al. (2015, 2018) and Sant’Anna-Santos (2021). The staminodes were subjected to histochemical testing for lipophilic compounds with Sudan (O’Brien and McCully 1981). To better visualise the xylem endings, the staminodial ring was submitted to polarised light and sectioned longitudinally. Pinnae and flower samples of the new species were also fixed in Karnovsky’s solution (Karnovsky 1965) for SEM analysis. After dehydration in ethyl alcohol, the samples were dried in a critical point drier (Balzers CPD 030) and sputter-coated with gold (Balzers SCD 050). The samples were observed using a JEOL JSM-6360-LV scanning electron microscope.

The diversity of the Butia genus in Chapada dos Veadeiros is still an open question. Current species of Butia are divided into two groups (Noblick 2014): the small acaulescent species and those that are caulescent (with above-ground stems). In Chapada dos Veadeiros, both Butia groups have been recorded (Mendonça et al. 2007; SpeciesLink Network 2023). The caulescent Butia species were first studied in this region by Martins (2012), who recorded B. archeri and B. purpurascens. The first species is usually caulescent (Soares 2015: fig. 1; Sant’Anna-Santos 2021: fig. 2) and has the widest distribution in the Brazilian central plateau (Noblick 2010, 2014; Soares 2015). The second species is also caulescent, larger in height (Soares 2015: fig. 14; Ressel and Guilherme 2022: fig. 1), and restricted to the southwestern Goiás State and surrounding localities (Noblick 2010; Soares 2015) – far away from Chapada dos Veadeiros. Moreover, neither species seems to occur in Chapada dos Veadeiros according to the distribution maps of the most recent revision of Butia (Soares 2015). Nevertheless, Noblick (2010) considered Chapada dos Veadeiros as an occurrence of B. archeri. Despite the similarity of some voucher specimens to B. archeri or B. purpurascens, the occurrence of these species in the region remains uncertain. My preliminary field observations (not documented) show that these caulescent specimens are not as tall as B. purpurascens, and their heads are not as compact as B. archeri. Thus, a broad sampling of specimens, especially during flowering, will be essential to solving the issue of the caulescent Butia in the region.

What about the hitherto unknown acaulescent Butia species from Chapada dos Veadeiros? At first sight, the acaulescent Butia soffiae resembles a dwarf form of B. archeri. Thus, some collectors have misidentified B. soffiae as B. archeri or only identified it to the genus level. Justifiably, the occurrence on the Brazilian central plateau and the lack of spines or teeth on the pseudopetiole pointed to B. archeri (see Noblick 2010; Soares 2015). Additionally, B. buenopolensis had not yet been described by Sant’Anna-Santos (2021), which explains the identification at the time. So, from a morphological standpoint, Butia soffiae is more closely allied morphologically to B. buenopolensis.

Despite the noteworthy similarities between B. soffiae and B. buenopolensis, such as the lack of spines or teeth on the pseudopetiole, congested purple flowers that differ in size (staminate vs pistillate), small size and acaulescence, there is enough difference to separate them. Morphologically, the size of the leaf rachis (Fig. 7A–B) and the colour of the pseudopetiole fibres (Fig. 7C–D) are vegetative characters useful to distinguish the two species. Regarding inflorescences, reliable characters to recognize B. soffiae as a distinct species from B. buenopolensis are: the rachilla number (Fig. 7E–F), peduncular bract size (Fig. 7E–F), peduncle colour (Fig. 7G–H), the size difference between inflorescence and peduncular bract (Fig. 7I–J) and the staminodes (Fig. 7K–JL). The detailed comparison between the two species is shown in Table 1 and the identification key. Therefore, despite the resemblance to B. buenopolensis, there is strong evidence supporting B. soffiae as a new species.

Regarding the habitat, both species are from the highland cerrado rupestre. However, the closest population of B. buenopolensis is at least 545 km from the Chapada dos Veadeiros (Fig. 4B). Furthermore, like the Serra do Cabral massif, the Chapada dos Veadeiros present a strong insular component, even compared to neighbouring areas (Felfili et al. 2007; Firmo et al. 2021; Sant’Anna-Santos 2021). So, the region is a cradle of endemics, including palms, and new endemics of several botanical families continue to be discovered (Noblick and Lorenzi 2010; Silva and Sodré 2014; Silva and Souza 2014; Noblick 2017; Pastore 2018; Villarroel et al. 2018; Santos et al. 2022; Silva and Santos 2023).

The pinna anatomy also supports Butia soffiae within the genus and as a new species. The mirrored mesophyll is the most striking pinnate character for determining the genus of the new species. Including B. soffiae, all the 19 valid Butia species anatomically studied possess this feature (Tomlinson et al. 2011; Noblick 2014; Sant’Anna-Santos et al. 2015, 2018; Noblick and Sant’Anna-Santos 2021; Sant’Anna-Santos 2021). Thus, this character is probably a synapomorphy for Butia (Noblick 2014; Sant’Anna-Santos et al. 2015, 2018; Noblick and Sant’Anna-Santos 2021). The raphides are also important to differentiate between species within the genus (Sant’Anna-Santos et al. 2015, 2018). Nine out of 19 valid species anatomically studied, including B. soffiae, have raphides on the pinnae (see Sant’Anna-Santos et al. 2015, 2018; Sant’Anna-Santos 2021). Among the four species compared here (B. archeri, B. buenopolensis, B. soffiae, and B. purpurascens), only B. soffiae has raphides. Thus, this character is reliable for identifying B. soffiae – even in non-reproductive specimens from the Chapada dos Veadeiros.

Raphides on flowers are also important for supporting the cohesive unit of Butia since they are lacking in Syagrus and Jubaea (Martel et al. 2013; Sant’Anna-Santos 2021). Epidermal raphide-containing idioblasts were first shown by Martel et al. (2013) on petals of staminate flowers of 13 Butia species. The same character was recently found in sepals and petals of staminate and pistillate flowers of B. buenopolensis (Sant’Anna-Santos 2021) and now in B. soffiae. Therefore, 15 Butia species have raphides in the perianth (Martel et al. 2013; Sant’Anna-Santos 2021), another possible synapomorphy of the genus. Among the microscopic characters, the presence of osmophores in Butia is confirmed here for the first time. Recently, similar structures were shown in B. buenopolensis (Sant’Anna-Santos 2021). Osmophores were also reported in some other neotropical Arecaceae, such as Acrocomia aculeata (Jacq.) Lodd. ex R.Keith and Syagrus cabraliensis (Noblick and Lorenzi) Sant’Anna-Santos (Mazzottini-dos-Santos et al. 2015; Sant’Anna-Santos et al. 2023a). Despite the low potential for taxonomical purposes, osmophores are important for elucidating the auto-ecology of the species, as showed in Fig. 5C–L. SEM analysis of Butia soffiae flowers showing squamiform trichomes below the staminodial ring (Fig. 3P, R) – also the first report for the genus. Unfortunately, these trichomes were not noticed in the field and so fresh flowers were not analysed in histochemical tests – an important tool to clarify their function. Similar trichomes were shown in Syagrus inajai (Spruce) Becc. by SEM analysis (Genovese-Marcomini et al. 2013: fig. 5), but, similarly, their function was not investigated. Nevertheless, these trichomes might be taxonomically relevant for Syagrus, as they have not been observed in Syagrus aristeae B.F.Sant’Anna-Santos, Syagrus cabraliensis, and Syagrus carvalhoi B.F.Sant’Anna-Santos (Sant’Anna-Santos et al. 2023a, 2023b, 2023c). Hence, evaluating the taxonomical potential of this trait for Butia will be important.

The newly discovered endemic acaulescent Butia from the highlands of the Brazilian central plateau display unprecedented characters with taxonomic, evolutionary, and ecological potential. Butia soffiae is the first endemic species of Butia formally described for the Chapada dos Veadeiros. The type population of B. soffiae is located near Chapada dos Veadeiros National Park, the target of a bill being processed in the Brazilian Parliament to reduce its area by 73%. Thus, the discovery of B. soffiae reinforces the uniqueness of the palm flora and the importance of maintaining or even increasing the park area.