Summit vascular flora of Serra de São José , Minas Gerais , Brazil

The campos rupestres form a mosaic of rocky savannas concentrated mainly along the Espinhaço chain, on the Brazilian shield. Though the Serra de São José lies over 100 km to the south of the Espinhaço chain, the campo rupestre flora of this small range harbors several endemic plant taxa. The provided checklist is the result of two decades of floristic research complemented with data from herbaria and literature. The flora is compared with the results of several other pertinent surveys. A total of 1,144 vascular plant species, representing 50.3 species/km, were documented to date in the São José range, representing a species-richness per unit area over five times greater than other known campo rupestre floras. The most species-rich families were the Asteraceae (126 species), Orchidaceae (106), Melastomataceae (63), Leguminosae (60), Cyperaceae (45), Poaceae (41), Rubiaceae (37), Myrtaceae (28), Bromeliaceae (27), Eriocaulaceae (23), Lamiaceae (23), and Malpighiaceae (22). Introduction The Serra de São José (21°3-7' S, 44°6-13' W) is a small quartzite mountain range just northeast of the town of Tiradentes and west of Prados, state of Minas Gerais, Brazil (Figure 1). The local climate is characterized by rainy summer and arid winter, as exemplified the climate diagram from São João del Rey (6 km from the range, Figure 2). The range is recognized as an area of extreme importance for nature conservation within the state of Minas Gerais (PROBIO 2004). Apart from their relatively large species richness, the campo rupestre and forest biotas of the Serra de São José harbor many endemic and endangered species, and this fact alone justifies the range as a priority area for conservation at both the state and federal levels. Historically the Serra de São José range lay in the path of several field naturalists who travelled into the hinterland and collected botanical material on their expeditions in the late 18th and early 19th Centuries. Several plant names were dedicated to these collectors. Langsdorffia hypogaea and Eugenia langsdorffii were dedicated to Georg Heinrich von Langsdorff, who collected on the Serra de São José between 1774 and 1852. Figure 1. Location of the São José range in Minas Gerais and in Brazil.


Introduction
The Serra de São José (21°3-7' S, 44°6-13' W) is a small quartzite mountain range just northeast of the town of Tiradentes and west of Prados, state of Minas Gerais, Brazil (Figure 1).The local climate is characterized by rainy summer and arid winter, as exemplified the climate diagram from São João del Rey (6 km from the range, Figure 2).The range is recognized as an area of extreme importance for nature conservation within the state of Minas Gerais (PROBIO 2004).Apart from their relatively large species richness, the campo rupestre and forest biotas of the Serra de São José harbor many endemic and endangered species, and this fact alone justifies the range as a priority area for conservation at both the state and federal levels.
Historically the Serra de São José range lay in the path of several field naturalists who travelled into the hinterland and collected botanical material on their expeditions in the late 18th and early 19th Centuries.Several plant names were dedicated to these collectors.Langsdorffia hypogaea and Eugenia langsdorffii were dedicated to Georg Heinrich von Langsdorff, who collected on the Serra de São José between 1774 and 1852.Friedrich Sellow (collection years: 1818-1820), the latter of which was awarded with at least the following taxa, all of which occur in the Serra de São José: Miconia sellowiana, Mandevilla sellowii, Anthurium sellowianum and Stachytarpheta sellowiana (Figure 3).João Barbosa Rodrigues collected in the nearby Lenheiro range, only 6 km away, between 1842 and 1909, but it is not certain whether he collected in the Serra de São José as well.One of the founders of the Museu Nacional herbarium in Rio de Janeiro, Heinrich August Ludwig Riedel (collection years: 1790-1861), collected in the range between 1824 and 1825; Gaylussacia riedelii and Rhynchospora riedeliana were dedicated to him.Auguste François Marie Glaziou (collection years: 1833-1906) apparently collected in the range, but there are many erroneous and dubious data in collections atributed to him.The topotypic population of Croton josephinus (Figure 4) is restricted to the summit of the highest peak.The São José range is the type collection locality for several of the aforementioned species, plus, Genlisea filiformis and Vellozia crinita.
------------Figure 2. Climate diagram from the São João Del Rey station, only 6 km SW of the São José range.The most frequent period of savannic fires is also indicated.------------At the end of the 19th Century, Astolpho Álvaro da Silveira, a very active field naturalist, visited and studied many mountain ranges of Minas Gerais.He wrote several books about various natural aspects of these mountains (for instance Silveira 1920;1921;1928;1931), but all his writings are currently rare and very hard to find.Silveira was very interested in the Eriocaulaceae, a family which has its center of diversity in the campos rupestres -the main type of summit vegetation of the Serra de São José.Silveira´s official collection, including material from the Serra de São José (located numbers range from 149-1,257), was acquired by the herbarium of the Museu Nacional in Rio de Janeiro (R), but several of his specimens sporadically appear in other herbaria in Brazil and abroad.
In the 20th Century, most collections in the Serra de São José were rather sporadic, on occasions when naturalists visited the historic towns of Tiradentes, Prados or São João del Rey.Two of these naturalists collected more intensely, but their findings were never systematized: In 1936 the visit of Henrique Lahmeyer de Mello Barreto rendered collections (4,765-5,090 are the numbers located so far), with one probably mistyped collection number: 17,537 from the same date as his number 5,090 (NY, RB).In the 1950s and 60s, Apparicio Pereira Duarte collected intensely in the Range and his material is in many herbaria including the NY, R and RB.His located collection numbers (2,5469,033) are mostly in R and there is even a very large number (34,971) which may be yet another mistyping.So far we were able to locate only 104 of his specimens from the range.
Our study area is indicated in Figure 1.In 1986, as a student, the first author undertook an almost accidental visit to the summit above Tiradentes and, in the following year, began two decades of relatively intense botanizing, especially in the species-rich campo rupestre and on the cliff walls of the range.One of the first long trips was accompanied by Marcos Valério Peron, at that time a promising Myrtaceae taxonomist, who numbered several of our collections (currently in RB).The first author began by studying the Orchidaceae of the range (Alves 1991a), was compelled to describe several new plant species (Alves 1990a;1991b;1992a) and ended up overwhelmed by the species richness in other plant groups.In 1989 Ruy J. V. Alves began his post graduate studies of the range under the guidance of the second author, Jiří Kolbek, at the Institute of Botany, former Czechoslovak Academy of Sciences.
Vellozia kolbekii (Figure 5) was probably a significant early contribution (Alves 1992a).Since then some taxa described from the Range were renamed.For instance Pelexia phallocallosa Alves was renamed Pachygenium phallocalosum (R.J.V. Alves) Szlach., R. González and Rutk.(Szlachetko et al. 2001), but Leslie Garay (pers.comm.)considered it a synonym of P. orthosepala (Rchb.f. and Warm.)Schltr.Others ended up as synonyms (Sarcoglottis caudata Alves = S. simplex (Griseb.)Schltr.(Stannard et al. 1995).After publishing the topotypic rediscovery of the endangered Stachytarpheta sellowiana Schau. in the Serra de São José (Atkins et al. 1996), we also found small populations in the Carrancas, Lenheiro and Ouro Branco ranges; at the time we did not know about the collection of Mello Barreto from 1936 (NY).For about a decade, the first author spent most of his time visiting the Range, mailing duplicates to specialists, and photographing over 5,000 slides (most of which are currently damaged by mold due to the humidity of Rio de Janeiro).However, there were unforgettable exceptions, exemplified by the kind treatment and keen sense of humor of Elsie Franklin Guimarães from the Rio Botanic Garden and Margarete Emmerich from the Museu Nacional: no matter how busy they were, they always put our plants under the binocular and accompanied us along the tortuous road towards determination.------------In 1996, when the first author assumed his current post at the Rio National Museum, collecting intensified slightly, because several students and a few colleagues accompanied some of the field trips and field courses of phytocoenology annually held in the Serra de São José.The exuberant diversity of the vegetation inspired some of these scholars to such an extent that they undertook formal studies in the Range: In 2001, Andréia Lúcia Pereira de Oliveira listed most of the Bromeliaceae from our collections in the Range as part of her Bachelor´s degree in Biology, and this listing was incorporated herein.Débora Medeiros concluded her MSc in 2002, listing the Euphorbiaceae from cerrado and campo rupestre of the range, which lead to the description of a curious new species of Croton with entirely glabrous leaves: C. arlineae (Medeiros et al. 2002).This also inspired the development of her doctoral thesis: a revision of the Sect.Medea of Croton (Medeiros 2007;Medeiros et al. 2008), during which two further new species were found.In her MSc.thesis in 2005, Rosana Augstroze Rutter Drummond listed 56 species of Melastomataceae (Drummond et al. 2007) from the range, which also resulted in the discovery of a new species of Cambessedesia (Figure 6) described with the valuable cooperation of Angela Borges Martins and named after the town of Tiradentes (Alves et al. 2008).This spectacular plant is known only from two tiny populations adding to, at the most, fifty individuals, and is probably narrowly endemic.The PhD thesis of the first author included a floristic checklist of the Serra de São José with 595 species including lichens, bryophytes and vascular plants.This was published in a limited series (Alves 1992).Since then, several other researchers have worked in the Range, and some of them published their results.This work is neither complete nor final.We have concentrated mainly on the campo rupestre vegetation on the range summits, the vascular flora of which can be expected to harbor between 1,200 and 1,400 species.
In the survey of forest tree species from the southern foot of the São José range, by Oliveira-Filho and Machado (1993), sampling was limited to individuals with a basal trunk diameter >= 5 cm (the material is in the ESAL herbarium).Analysing tree species with trunk diameter at breast height >5 cm in 18 evenly distributed belt transects totalling 0.9 ha, Gonzaga et al. (2008) found 130 species in the forests at the southern foot of the Serra de São José.Comparison of this partial flora with another 23 forest fragments in the region revealed affinities with forests of both higher and lower altitudes.Gavilanes et al. (1995) listed 966 plant binomials belonging to 126 families for the Serra de São José, but collection numbers were not provided.In 2007, not even a single voucher specimen from the Range could be located in the PAMG/EPAMIG herbarium, where the material was mentioned to have been deposited.Furthermore the listing has some odd binomials: for instance among the ferns, Anemia chlupata and A. striata, which are coincident with codenames, written in Czech, from our field notebooks from September 1989 and accidentally printed in the PhD thesis of the first author.Our collections of these were determined later, respectively by Jefferson Prado and John T. Mickel, as Anemia villosa and A. imbricata.
In 1996 we guided Lea Scheinvar into the range in search of Cactaceae.The resulting collections in the RB herbarium should be viewed with caution: several species were collected in gardens in the town of Tiradentes and not in the Serra de São José as their tags indicate.The Cactaceae listed herein are exclusively from the range.
Since the very beginning, our collections were intended to subsidize descriptions of vegetation and did not concentrate exclusively on Angiosperms.Hundreds of algal, moss, and lichen specimens were collected and sent to specialists.In 1990, when the lichen specimen Alves 1,118 (PRA) reached Teuvo Ahti in Helsinki, it turned out to be the second known collection of a new species, Cladonia bahiana Ahti, at the time still in press (Antonín Vězda, Czech Republic, personal comm.).
Lichenologists and bryologists from the Botanical Institute in São Paulo collected intensely in the range in 1993, but this expedition unfortunately never produced a lichen checklist.The actual lichen richness of the Range is considerable, and the few odd names we were able to compile make no justice to the actual lichen richness, and we list them herein with this caveat.Yano and Peralta (2008, in edit.)listed 114 Bryophyte species from the Serra de São José, including several new records for Brazil and for the state of Minas Gerais.There were several other scientific studies directly inspired by the Serra de São José floristic and vegetation project.Valéria Cid Maia accompanied us on several field trips, and produced a survey of insect gall diversity from the range (Maia and Fernandes 2004).

Material and methods
The study area is a small range which consists of several mountains with elevations from 900 to 1,430 m a.s.l. and passes, aligned from SSW to NNE, being less than 2 km wide and roughly 15 km long.The top of the range and a large part of its flanks are covered by campo rupestre vegetation, occasionally interrupted by gallery forest and small pockets of latossol with cerrado.A roughly 1 km wide belt of dry cerrado forest runs along the southern foot of the range, while the northern side is flanked mostly by plantations in previously cerrado area.Mean annual temperatures measured in São João Del Rei (6 km from the western extremity of the range, at a similar mean altitude) vary from 14 to 26° C (with absolute temperatures reaching 36° C and 1° C respectively.Winter temperatures atop the range can reach below freezing point (minus 3° C was measured at 1250 m a.s.l. by Alves 1992) and frost is quite common.Rainfall averages during winter reach a minimum of 8 mm in July, while in the summer they can exceed 300 mm in January.Localized seasonal fires, both natural and provoked, are frequent in the area, and the summits are often grazed and trampled by cattle.
Our collections in the Serra de São José span two decades, and approximately half of the voucher specimens (collected prior to 1996) are deposited in the herbarium of the Rio de Janeiro Botanical Garden (RB).Subsequent collections are in the National Museum in Rio de Janeiro (R).In order to minimize impact on rare species such as several Orchidaceae, we only preserved flowers in alcohol for determination by floral analysis (Alves 1990b;c;1998).Though our efforts were concentrated mainly on the summit rock-dwelling flora, about fifteen complementary collection excursions were also done in the forest at the southern foothills of the range, along its entire length.
Due to the intensity and frequency of changes in higher taxonomic ranks imposed by recent developments in systematics, we have chosen to list the vascular plants in alphabetical order by family, genus and species.As some families still do not have stable circumscriptions assigned by the Angiosperm Phyllogeny Group, we follow Cronquist (1988) for convenience of comparison to older floristic lists, and Rapini (2000) for joining the Asclepiadoideae under the Apocynaceae.
About half the determinations were provided by the specialists listed after each family.When several specialists are mentioned, they contributed at distinct times along the two decades.For a few of the specimens in the RB and foreign herbaria, we transcribed the determinations in institutional databases (JABOT 2008, Tropicos.org 2008, New York Botanical Virtual Herbarium 2008).The remaining determinations were done by comparison in herbaria, use of keys and published descriptions.The most recent names in the Tropicos.org(2008) database were adopted (for instance by Harold Robinson for the Asteraceae).For the nomenclature of Drosera, however, field observations and cultivation have lead us to maintain the distinctions based on field observations of Saint Hilaire, disregarded by many latter authors.There are some cases, however, when the names in the Tropicos.orgdatabase are mistyped, or known synonyms are not referred to, and in these we did our best to use the original specific epithets.Apart from this checklist, detailed treatments of some taxonomic groups like the Droseraceae, Lentibulariaceae, Melastomataceae and Bryophytes from the range are currently being edited as chapters of a book by our colleagues.A detailed treatment of the Euphorbiaceae and Phyllanthaceae by Débora Medeiros had been accepted earlier by a journal, and is currently in print, hence only a simple checklist is provided herein.Though we concentrated on the summits, as we passed through the forested areas on our way up, we did collect sporadically, and we also registered peculiar vegetation on shaded quartzite boulders.The classification of life-forms follows the main subdivisions of Raunkiaer (1934).For species which can develop more than one life form, only the predominant one was computed.
Distinct yet indetermined morphospecies within a given genus are marked as sp.1, sp.2 etc.Though we initially intended to provide all collection data, observations, determining specialists etc., the resulting list was 100 pages long.Hence the format of the current checklist is simplified, including only the names of taxa, a code for the predominant life form, and the first or the most representative specimen (collector and number, with herbarium acronyms after Holmgren et al. 1990).The current vascular flora was compared with floristic data from the Serra do Cipó (Giulietti et al. 1987) and Pico das Almas (Stannard et al. 1995), followed by the calculation of Sørensen similarity indices.

Results and Discussion
Current knowledge of species richness values also reflects collection effort, and their comparison is not always a simple matter.There are 3,200 collections in the database, with 1,258 species of vascular plants and bryophytes distributed over less than 25 km 2 .Hence, the São José range has 50.3 species/km 2 , while other much larger campo rupestre localities studied for comparable time periods rendered comparably poorer species richnesses per unit of area (Table 1).Of 1,144 vascular morphospecies, 957 are determined to or beyond species level (with 76 as "cf." or "aff."), and 181 are determined to genus (of which 9 are in "cf.").Even if all the taxa with uncertainties of infrafamilial determination were not computed (leaving 885 species and subspecies of vascular plants and 114 bryophytes), this would still result in an astonishing species richness of 35.4 species/km 2 .Though the differences in Table 1 may partly be caused by distinct collection efforts, the Serra de São José still has a notoriously high species richness.
- ---------------Table 1. Floristic composition, collection efforts and species richness in three campo rupestre localities with relatively complete surveys.Data sources: Serra do Cipó, northern Minas Gerais (Giulietti et al. 1987), Pico das Almas, Bahia (Stannard et al. 1995) The distinct selective method used in the forest by Oliveira-Filho and Machado (1993), along with the fact that we concentrated on campo rupestre, may partly explain why only 58 of their 277 species are also listed herein.Only 35 species in the present survey are also reported by Gonzaga et al. (2008).It is a rather curious fact that, in the higher parts of the studied forest where several transects were executed (especially near the southern wall above Tiradentes), several very common tree species such as Clusia arrudea, Dyssochroma viridiflora, Maclura tinctoria, Myrciaria tenella, Podocarpus sellowii were not sampled by the latter authors.We speculate that this may possibly be explained by limitations of using symmetrical distribution of sample plots.A systematic survey of all vascular species including herbs and shrubs in these same forests may easily lead to the discovery of several hundred species currently unregistered for the locality.
Though both aforementioned surveys are essentially from the same forest and used comparable selection criteria of trunk diameter, they have only 92 species in common, while respectively 185 and 38 exclusive species were sampled.Even when combined into one floristic matrix, both forest surveys are very dissimilar to the campo rupestre studied herein (Figure 7).Even if only the 210 campo rupestre tree species are considered, the resulting Sørensen similarity index is still only 0.262.Only 24 species registered herein are also common to the forest surveys by Oliveira Filho and Machado (1993) and Gonzaga et al. (2008) Two relatively large campo rupestre floras were compared with the checklist herein: the Serra do Cipó (Giulietti et al. 1987) and Pico das Almas (Stannard et al. 1995).As expected, the Sørensen similarity index is slightly lower between the farthest localities (Figure 8).------------Despite the extrazonal nature and small areas occupied by outcrop vegetation in general, the species richness of campo rupestre floras seems to follow the first and second biocoenotic principles of Thienemann (1920;1954): "the more variable the habitat conditions, the higher the diversity in a biocoenosis" and "in a community very rich in species, the incidence of the different species is usually low, whereas in a community poor in species, the single members often occur in large numbers, such a community or biocoenose being less stable than the former."(Klötzli 1992).
Amidst the current shiftings of entire groups of species from one genus to another (Alves and Vianna Filho 2007), there seems to be no taxonomic consensus in sight (examples are the Acianthera-Pleurothallis-Specklinia and the Hadrolaelia -Hoffmannseggella -Laelia -Sophronitis complexes among the Orchidaceae).
The presence of lithophytes and a higher proportion of epiphytes apparently distinguish campo rupestre from surrounding zonal vegetation.Phanerophytes make up 40.3 % of the vascular flora of the São José range, followed by chamaephytes and hemicryptophytes (Figure 9).Among the phanerophytes, we registered 210 species of trees and tall shrubs (18.5 % of the summit flora).
The hydrophytes have not been systematically collected.Of 446 woody species in the flora, 291 species are trees or tall shrubs and 155 are subshrubs.Thickened underground organs such as bulbs, corms, rhizomes, tubers, lignotubers and tuberous roots were found in 446 species belonging to 59 families of the São José range flora.These organs are important adaptations to natural fires.Because not every species encountered in the range was uprooted, the real percentage is probably even higher.The largest numbers of subterranean organs were found among the Asteraceae (94 species), Melastomataceae and Leguminosae (each with 33), Orchidaceae (30), Apocynaceae (including Asclepiadoideae, with 26), Lamiaceae (21), Malpighiaceae (16), Euphorbiaceae (15), and Rubiacae (13).
As the surrounding savannas, the campo rupestre vegetation is also well adapted to natural disturbances by fire, though the ideal periodicity still remains to be established.The most serious issue in "campo rupestre" is the presence of grazing cattle, which fertilize originally oligotrophic soils, and allow the establishment of invasive plant species.Among these, the molasses-grass (Melinis minutiflora P. Beauv.) is apparently of most concern, since it promotes the spread of fire to areas previously free from it.As the "campo rupestre" in the São José range (and probably many other areas with similar rock-dwelling vegetation) exhibit unprecedented species-richness and occupy extremely reduced geographical areas, the exclusion/removal of grazing domestic animals from these is the most urgent and imperative conservation measure.
- ---------- - ---------------Environment, Tiradentes) for their frequent logistic support to our expeditions.This checklist was only possible due to the contributions, comments and determinations of approximately one hundred people and institutions, support by local people and enterprises.Special thanks are due to many taxonomic specialists cited as collaborators in Appendix 1, and to Antonín Vězda (PRA, Lichenes), Jiří Váňa (PRC, Bryopsida and Hepaticopsida), Marcelo Marcelli (SP, Lichenes), Mariângela Menezes (R, Algae and Cyanobacteria), Olga Yano (SP, Bryopsida and Hepaticopsida), Renato Rodrigues Cabral Ramos (Museu Nacional, Geology) and Teuvo Ahti (H, Lichenes) for suggestions and valuable comments, Ivana Rajznoverová for technical assistance with the manuscript, the Instituto Estadual de Florestas (IEF/MG), especially Ana Paula Pinheiro and Janaína Aguiar, and the Floresta Nacional de Ritápolis (FLONA/IBAMA) for occasional logistic support in the field.This project was partly supported and authorized by the Instituto Estadual de Florestas.The participation of the second author was possible due to Czech research project number AVOZ 60050516.The scholarships (1988)(1989)(1990)(1991)(1992)(1993)(1994)(1995)(1996) from the Brazilian Council of Research (CNPq) were also essential to these accomplishments.A notable Brazilian collaborator was Wanderbilt Duarte de Barros, who became a close friend while he presided the Brazilian Foundation for the Conservation of Nature (FBCN), and who, as Director of the Rio Botanical Garden, helped the present Flora effort immensely by inviting the first author to work as a visiting researcher after having concluded his PhD studies in Prague, Czech Republic.Thanks to his support it was possible to compare many collected species in the RB herbarium.

Figure 1 .
Figure 1.Location of the São José range in Minas Gerais and in Brazil.------------Carl Friedrich Philippe von Martius collected in the Serra de São José during January of 1818, and his specimens are deposited in Vienna, St. Petesburg, London, Leiden, and Leipzig.Several species which occur in the range were described in Martius´ monumental Flora Brasiliensis and other works, for instance Psittacanthus robustus, Barbacenia (Aylthonia) tomentosa, and

Figure 6 .
Figure 6.Cambessedesia tiradentensis Alves, Drummond & A. B. Martins (Melastomataceae), a new species named after the town of Tiradentes, is known only from two small campo rupestre populations in the São José range.The holotype is from this specimen.Photo: R. J. V. Alves.

Figure 7 .
Figure 7.Comparison of the floristic surveys of the Serra de São José campo rupestre (top) and forest checklists by Oliveira-Filho and Machado (1993) and Gonzaga et al. (2008).Numbers represent species or varieties.Is = Sørensen similarity index.

Figure 8 .
Figure 8.Comparison of the vascular floras of the Pico das Almas in Bahia (Stannard et al. 1995), Serra do Cipó, northern Minas Gerais (Giulietti et al. 1987) and the Serra de São José.Numbers represent species or varieties.Is = Sørensen similarity index.