Study Area

Two protected areas and their surroundings within the Plateau were chosen for our study: PERP and PNSV (Figures 1 and 2). PERP has an area of 10,700 hectares and is in the municipality of São Gonçalo do Rio Preto (18°04′12″ to 18°14′15″ S, and 43°23′34″ to 43°18′18″ W). PNSV occupies an area of ca. 124,555 hectares in the municipalities of Olhos D’Água, Bocaiúva, Buenópolis, and Diamantina (17°44′11″ to 17°59′28″ S and 43°35’50″ to 43°59′33″ W).

Figure 1.

Study areas within the Cadeia do Espinhaço, Minas Gerais, Brazil. Larger inserted map: Espinhaço Mountain range (gray), Diamantina District Plateau (box), Parks (hatched). Smaller inserted map: Brazil (pale gray), Minas Gerais (dark gray), Bahia (black). PNSV = Sempre Vivas National Park; PERP = Rio Preto State Park.

Figure 2.

Habitats in the Diamantina Plateau. (a and b) Sempre Vivas National Park, RAPELD module area PNSV I with campos rupestres and cerrado savannas; (c and d) Rio Preto State Park, RAPELD module area PERP I with campos rupestres, cerrado savannas, and transitional areas; (e and f) Rio Preto State Park, RAPELD module area PERP II with campos rupestres and gallery forests.

Floristic Survey

Botanical collections were made (SISBIO Licence 40530–1 issued to the first author on August 6, 2013) using two different methods, the RAPELD protocol (Magnusson et al., 2005), with a reduction to 7 ha per module and some modifications to suit the campo rupestre environment (see Chaves et al., 2019), and traditional floristic collecting, between August 2013 and October 2014. For the RAPELD protocol, three sample modules were established, each of which with 21 sample units. Two sample modules were set up in PERP (PERP I and PERP II) and one in PNSV (PNSV I). Sample units were demarcated along two 5 km trails perpendicular to the isoclines (topographic contours) of the terrain. Sample units measured 250 m × 40 m (=10,000 m² = 1 ha) and were demarcated at 10 m increases in altitude in PNSV and 25 m increases in altitude in PERP (1,000, 1,025 m, etc.; Figure 1). Each 10,000 m² unit was subdivided into 25 subunits of that measures 10 m along the isocline × 40 m across that were easy to visualize and uniform in altitude, topography, and soil type. RAPELD collections were made within sample units, and both fertile and sterile specimens were collected. Even if additional specimens were added to the collection from the immediate vicinity of the unit, this was still considered a RAPELD collection since the motivation for the collection was its occurrence in the sample unit. Traditional floristic collecting (Mori et al., 1985; Ter Steege, Harispersaud, Bánki, & Schieving, 2011) was done in and out of the two protected areas, within a 117 ha polygon area of the Plateau whenever an Asteraceae was found either flowering or fruiting; sterile specimens were not collected. Total collecting effort was 32 collecting days (1–5 collecting days per month) with teams of 1 to 4 collectors (average team = 2.1 collectors) between August 2013 and October 2014.

The PERP I module ran through savanna (cerrado) as well as campos rupestres, while the PERP II and PNSV I modules were in an area mainly of campos rupestres (Figure 2). All campos rupestres areas showed high heterogeneity of microhabitats with differences in soil, humidity levels, altitude, and floristic composition. In PERP I altitude varied between 760 and 940 m on poorly drained and well-drained soils, while in the savanna areas soils were deep with a high proportion of clay. In PERP II, altitude varied between 1,350 and 1,525 m, with prevalent rocky outcrops and some gallery forests, wet and well-drained fields (campos). In PNSV I, there was little variation in altitude (1,220–1,290 m), with campos rupestres dominating the vegetation but with pockets of savanna and gallery forest.

Taxonomic Identification

Authors J. B. A. B. Jr., J. N. N., and V. L. R. are Asteraceae specialists and identified most of the species but other specialists were consulted (see Acknowledgments). Herbarium vouchers were deposited in the Universidade de Brasília and in the Universidade Federal do Vale do Jequitinhonha e do Mucuri (see also Species Link 2017a, 2017b, 2017c, 2017d). Species with doubtful identifications (affinis or conferatum) and potential new species were listed in Table 1 but not counted and considered in the comparative analysis. Species’ geographic distributions were compiled from the literature (Suppl. Information) and on-line databases (Herbário Virtual Reflora, 2017; Species Link, 2017a). Scientific names were updated following Flora do Brasil 2020 em construção (2017a).

Table 1.

List of Asteraceae Species Collected in Two Conservation Units and Their Surroundings in the Campos Rupestres of the Diamantina Plateau in the Espinhaço Mountain Range.

Richness and Endemism

A total of 613 collections of Asteraceae were made that resulted in 172 published, named species (Table 1); five morpho-species that are either undescribed species or doubtful identifications are not included in this tally. The RAPELD survey recorded 115 species in a sample of 12,775 individuals of Asteraceae. Forty-seven species (27%) are endemic to campo rupestre vegetation, and 38 species (22%) are found only in the southern “Minas Gerais” region of the CdoE mountain range (Figure 1; Table 1). Furthermore, 13 of these species (7.5%) are endemic to the Plateau. The true number of endemic species of the Plateau may be even higher as collections of three possible new species were made. Lychnophora sp. 1 (Table 1) is an unpublished species recognized by Semir (1991).

PERP had two RAPELD modules located at different altitudes, one through campos rupestres and one through campos rupestres grading into savanna (cerrado); PNSV had only one module mainly through campos rupestres. This resulted in 226 RAPELD collections in PERP versus 83 RAPELD collections in PNSV. In PERP, with two sample modules, there was more diversity (127 vs. 76 species in PNSV) and 57 species collected were exclusive to PERP and were not found in PNSV or the park’s surrounding areas. On the other hand, in PNSV, only 16 exclusive species were found. Interestingly, 21 species were collected in the surroundings of both parks but recorded in neither of them (Table 1).

Endangered Species and Conservation Status

A total of 38 endangered species of Asteraceae, representing 16% of the Asteraceae on Brazil’s official list (Ministério do Meio Ambiente, 2014), are known to occur in the Plateau (Table 2). We recollected 30 of these (Figure 3; Tables 1 and 2), of which 26 species were recorded in one or both parks and 4 species were only recorded by traditional floristic collecting out of the park areas. The status of the recollected species was one critically endangered, 19 endangered, and 10 vulnerable (Martinelli & Moraes, 2013; Ministério do Meio Ambiente, 2014; Table 2). The number of endangered species recorded in PERP (18 spp.) and in PNSV (13 spp.) was similar but only five species were common to both parks (Table 1). Eight endangered species with known distribution within the Plateau were not recollected. Only one of the six critically endangered species from the Plateau was found (Piptolepis leptospermoides recollected from PNSV by traditional floristic collecting).

Table 2.

List of Endangered Asteraceae Species (numbered) From the DD Plateau and Species Whose Distributions Expanded (Not Numbered).

Figure 3.

Asteraceae from the Diamantina Plateau. (a) Wunderlichia senaeii Glaz. ex Maguire & G.M. Barroso ‘E’; (b) Heterocoma erecta (H. Rob.) B. Loeuille, J.N. Nakaj. & Semir ‘E’; (c) Lychnophora cf. humillima Sch. Bip.; (d) Aspilia andersonii H. Rob.; (e) Aspilia diamantinae J.U. Santos ‘E’; (f) Verbesina macrophylla (Cass.) S.F. Blake; (g) Lychnophora triflora (Mattf.) H. Rob.; (h) Minasia scapigera H. Rob. ‘E’; (i) Lychnophora martiana Gardner ‘E’; (j) Piptolepis leptospermoides (Mart. ex DC.) Sch. Bip. ‘E’; (k) Minasia alpestris (Gardner) H. Rob. ‘E’; (l) Lychnophora uniflora Sch. Bip. ‘E.’ ‘E’ = endangered species, for category see Table 2.

Endangered Species and Conservation Status

For 20 Diamantina Plateau endangered species (53% of the known total of 38 species), our study has brought qualitative gain: they are now known to occur in one or both parks, and in many cases, frequency data have been obtained from RAPELD modules (see Chaves et al., 2019). Five such endangered species have been recorded from protected areas for the first time (species marked with ^b in Table 2). For the other 18 species (47%), there has been no qualitative change in conservation unit occurrence. However, only for five species does this ‘no change’ represent continued failure to record the species in a conservation unit. Most qualitative ‘no change’ means the species had already been recorded from the parks by previous collectors.

Our negative results from the three RAPELD modules, which sampled 12,775 individuals of Asteraceae and found only 27% of endangered species known to occur in the Plateau, are possibly as significant as the positive results (recollections of endangered species known to occur there). If endangered species were uniformly distributed in different habitats, a high percentage of endangered species recovery might have been expected from the RAPELD modules. Closer investigation of the endangered species that were not collected showed that two of them are common either in the northern “Bahia” CdoE or in the more westerly Serra Geral de Goiás. For species where wide-scale distributional data could not explain absence, microdistributional differences or microhabitat specialization, as proposed by Chaves et al. (2019), stochastic distribution, or a combination of these are the most likely explanations. Percentages of gravel, sand, and silt in the soil explained 67% of the difference between mycorrhizal communities (Carvalho et al., 2012). Substrate (quartzitic, arenitic, or ironstone) was the most important campo rupestre driver of floristic diversity (Zappi, Moro, Meagher, & Nic Lughadha, 2017). These findings suggest that these endangered species were absent because they are restricted to unsampled, microhabitat communities. Dimmerostema oblonga, collected by us and only three times in 175 years in widely separate campo rupestre areas, could be such a specialist—or simply a very rare species. Another possibility is extreme geographic narrow endemism, such as has been recorded for the charismatic, unmistakable Vellozia gigantea Mello-Silva whose nine known populations are within 27 km of each other (Lousada, Borba, Ribeiro, Ribeiro, & Lovato, 2011).

Conservation strategies and the choice of priority areas are usually based on the presence of red-listed endangered species, but abundance, persistence, geographic distribution, and level of threats should also be quantified (Kakkala & Moinen, 2013; Nakajima, Junqueira, Freitas, & Teles, 2012). Campo rupestre lineages show high phylogenetic conservatism and low dispersal and colonization capabilities and are thus highly deserving of conservation priority (Conceição et al., 2016). Knowledge of species abundance allows for safer decision-making during management of protected areas (Brandon, Fonseca, Rylands, & Silva, 2005). CdoE alpha-taxonomic knowledge remains incomplete and more collecting is needed (Madeira, Ribeiro, Oliveira, Nascimento, & Paiva, 2008; U. Oliveira et al., 2016).

Although many endangered species found in this study may appear protected, the protected areas often lack effective management and physical infrastructure. National and state parks are threatened by mining, criminal fires, invasive species, and habitat destruction as population and tourism grows (Echternacht et al., 2011; Silveira et al., 2016). It has been estimated under various climate change scenarios that between 40% and 82% of the CdoE’s area of campo rupestre will be lost or fragmented and practically disappear in the northern region by 2070 (Bitencourt, Rapini, Damascena, & De Marco, 2016; Fernandes, Barbosa, Negreiros, & Paglia, 2014; Fernandes, et al., 2018). Such area reduction could mean mass extinction of endemic species, even in current protected areas, that could shrink by up to 40% in area by 2050 (Bitencourt et al., 2016; Fernandes et al., 2018). Species distribution models that predict current as well as future species-range distribution patterns under climate change are essential so conservation planning can adapt to novel scenarios.