Altitudinal distribution and species richness of herbaceous plants in campos rupestres of the Southern Espinhaço Range , Minas Gerais , Brazil

The variation in the species richness of herbaceous plants along an altitudinal gradient was analyzed in the Serra do Ouro Branco and Serra do Ribeiro, in the townships of Ouro Branco and Ouro Preto, respectively. Plant similarity between both serras was also assessed. Twenty spots were sampled along a 440 m (1105 m to 1545 m) altitudinal range; at each site, herbs were collected within ten 1 m plots, totaling 200 m2. We found 101 species distributed in 59 genera and 25 families. The richest plant families in Serra do Ouro Branco were Poaceae (22 spp.), Asteraceae (14 spp.) and Cyperaceae (10 spp.), while in Serra do Ribeiro, they were Poaceae (17 spp.), Cyperaceae (12 spp.) and Asteraceae (8 spp.). Variation between the number of species and altitude was not significant. The higher number of species in Serra do Ouro Branco may be due to different local environmental factors and to the occurrence of grazing and fires. The serras presented a high similarity value (J = 0.44), but cluster and ordination analysis indicated the formation of two distinct groups, reflecting the importance of local factors to determine the floristic composition of neighboring areas of campos rupestres.


Introduction
Although Brazilian mountain ecosystems are of great ecological and economic importance, they are threatened by human activities and their biology is poorly known (Martinelli 2007).Studies on tropical mountain ranges are important to understand the processes and mechanisms that influence biodiversity and organism responses to environmental changes, as global warming (Gottfried et al. 1999;Lomolino 2001;Beckage et al. 2008).
Variations in species richness along altitudinal gradients are relatively well studied (Hodkinson 2005).In short, two plant species distribution patterns were described for different organisms and biogeographical regions (Lomolino et al. 2006).The first one maintains that the number of species decreases as altitude increases (Rosenzweig 1995).This negative linear relationship was reported for different organisms, as birds (Terborgh 1977), herbivore insects (Fernandes & Price 1988;McCoy 1990;Carneiro et al. 1995) and plants (Givnish 1999;Jones et al. 2003).The second one holds that the number of species diminishes as we get closer to the extremes of the altitudinal gradient, and defines a maximum value of species richness at intermediate altitudes (Gentry & Dodson 1987;Rahbek 1997;Colwell & Lewis 2000;Lomolino 2001;Grytnes 2003).It was described for different groups of plant species (Tryon 1989;Grytnes 2003;Bachman et al. 2004;Krömer et al. 2005), herbivore insects (McCoy 1990), birds (Herzog et al. 2005) and mammals (Nor 2001).
Campos rupestres are found in the states of Minas Gerais, Bahia and Goiás.They are usually constituted by mosaics of plant communities, formed by a herbaceous stratum followed by perennial and sclerophyllous bushes and subshrubs occurring at altitudes between 900 and 2070 m, on great extensions of quartzitic outcrops with shallow, compact litholic soils (Giulietti & Pirani 1988;Romero 2002).The herbaceous stratum is mainly formed by species of the families Poaceae, Cyperaceae, Eriocaulaceae and Xyridaceae.The bush stratum comprises a high number of species of Asteraceae, Melastomataceae, Lamiaceae, and Velloziaceae (Giulietti & Pirani 1988).
In the southern Espinhaço Range, the Quadrilátero Ferrífero (Iron quadrangle) stands out by its rich deposits of mineable resources and remarkable biological diversity (Drummond et al. 2005).The region is distinguished by its high diversity of habitats, which may be related to edaphic peculiarities, to the characteristic mountain relief of the region and to the fact that it is located in a transition area between the Atlantic Forest and cerrado biomes (Council & Murta 2007).Among its different phytophysiognomies, we can mention forest (e.g.seasonal forests, gallery forests, cloudy forests), savanna (cerrado sensu stricto) and grassland (campos rupestres on quartzite, campos rupestres on canga and campos limpos) formations (Viana & Lombardi 2007).
The present work surveyed herbaceous plants in two serras located in the southern Espinhaço Range and verified if species richness diminishes with altitude.The influence of the environmental heterogeneity on the species composition of each serra was also assessed by analyzing the floristic similarity of the sampled places.

Material and Methods
This work was carried out in Serra do Ouro Branco (SOB) and Serra do Ribeiro (SR), located in the townships of Ouro Branco and Ouro Preto (Fig. 1), respectively.The SOB stands out as the most significant element of the southern border of the Quadrilátero Ferrífero.The altitude of its circa 65 ha varies from 1,000 to 1,573 m (Alkmim 1987).Located approximately 10 km to the North of the SOB, with altitudes varying between 1,270 and 1,550 m, the SR comprises a set of two smaller formations.Climate is mesothermic -Cwb (Köppen 1948), with mild, rainy summers and dry, cold winters.Mean annual temperatures vary between 17ºC and 20ºC and the annual rainfall records are approximately 1,500 mm (Giulietti & Pirani 1988).In the campos rupestres of both serras, the sampling places are characterized as quartzitic grasslands with or without subshrubs (Rizzini 1979), usually next to rocky outcrops, sometimes with evidence of grazing and fires.
Field expeditions were carried out between March and July 2004.Ten collection points arbitrarily defined, in an attempt to encompass the different types of habitats along the altitudinal gradient of each serra, were sampled; geographical coordinates and spot heights (Tab. 1) were determined using an Etrex Venture (Garmin®) GPS.Sampling was performed in ten 1 m 2 plots systematically distributed, at a distance of 5 m from each other along a 50 m imaginary line (Pivello et al. 1999), so that 10 m² were sampled at each altitudinal point, totaling 200 m².Plants were identified by comparison with specimens kept at the OUPR and BHCB herbaria (acronyms according to Holmgren et al. 1990) and with the help of specialists.Only the angiosperms composing the herbaceous stratum were sampled.Fertile samples were deposited at the OUPR herbarium.Botanical families are circumscribed according to the Angiosperm Phylogeny Group -APG II (APG 2003).
Analyses of covariance were performed to determine if plant species richness diminishes with altitude (Crawley 2002).On our model, plant richness was the response variable and altitude (covariable) and the serras (categorical variable) were the explanatory variables.The analyses were performed with the statistical package R version 2.5.1 (R Development, Core Team 2005), using the 'glm' procedure, and Poisson errors were calculated through chi-square tests (χ²).Residual analyses were carried out to check error distribution and adjust the model (Crawley 2002).
We used the Jaccard index to measure similarity between the sampling points (McCune & Grace 2002) and the relationships between them were characterized by cluster and ordination analyses based on the method of unweighted means (UPGMA), using the FITOPAC software (Shepherd 1996), and on a multidimensional scale analysis (MDS) carried out with XLSTAT data analysis and statistical solution for Microsoft ® Excel 2007.

Results
One hundred and one species distributed in 59 genera and 25 families were collected in the two studied serras (Tab.2).The families with the highest number of species were Poaceae (25), Asteraceae (14), Cyperaceae (13) and Polygalaceae (9).Out of this total, 86 species grow in the SOB, 41 of which are exclusive to it, and 61 species occur in the SR, 16 of which are exclusive to it (Tab.2).In the SOB, the families with the greatest number of species were Poaceae (22), Asteraceae (14) and Cyperaceae (10), while in the SR, they were Poaceae (17), Cyperaceae (12) and Asteraceae (8).

Spermacoce neotenuis Govaerts (B 72) ------+ ---+ ---------Declieuxia cordigera
Although the MDS analysis showed an organization similar to that of the UPGMA, it revealed different relationships between some points, i.e. points 1 and 2 of the SR and point 8 of the SOB; and the points 7, 9 and 10 of the SR (Fig. 4).The highest similarity value is between points 5 and 6 of the SR (J = 0.61), while points 9 and 10 of the SR grouped outside the set of the two serras (Fig. 3 and Fig. 4).

Discussion
The total herbaceous richness recorded in this study is similar to that reported in other floristic surveys in campos rupestres (Giulietti et al. 1987;Stannard 1995;Pirani et al. 2003;Zappi et al. 2003;Conceição & Pirani 2005;Viana & Lombardi 2007).However, these results are probably underrated because the sampled area was small and the field work was only carried out in one climatic season of the year, so that it does not take seasonal variations into account.The higher species richness found for families Poaceae, Asteraceae and Cyperaceae is also corroborated by previous work (Giulietti et al. 1987;Safford 1999;Filgueiras 2002).
The absence of exotic species at collection points shows that few disturbances affect the studied place, despite of the presence of cattle and the occurrence of frequent fires, especially in the SOB, where populations of Melinis minutiflora P. Beauv.and Hyparrhenia rufa (Nees) Stapf.grow along the highway that crosses it.The occurrence of fires increases the probability of intrusion of invasive African grasses as Melinis minutiflora, Urochloa decumbens Stapf.and Megathyrsus maximum Jacq.(Pivello 1999).
Although the geographical proximity and the similar orogeny of the two serras imply similar climatic conditions, geology and biogeographical history of the species, the SOB presented higher richness of herbaceous plants than the SR.A possible explanation is the effect of grazing and fires on the plant community structure of grassland vegetation (Howe 1994), since, in the SOB, the Distribution and richness of herbs in campos rupestres Herbaceous species composition, mainly in natural grassland formations, has been found to present modifications related to the intensity and to the historical time of cattle grazing (Pucheta et al. 1998;Olff & Ritchie 1998).When comparing samples of a same area impacted by large grazers in eastern Australia, McIntyre et al. 2003 verified that those with medium disturbance intensity presented higher plant richness as compared to samples with little or much disturbance, which they explained by a decrease in competition and an increase in regeneration.
Along the altitudinal gradient, local factors can be more important than regional ones to determine the occurrence of species (Pausas & Austin 2001;Herzog et al. 2005).In this context, the facts that almost 30% of the total species occur in a single sampling point and that different species richness was found between samples at similar altitude suggest that the communities can be influenced by local factors or present a great natural variation in their species composition, with high âdiversity values (Lieberman et al. 1996).Again, one should keep in mind that the result interpretation is partial and restricted, due to the small sample size.In addition, the higher number of species of families Burmanniaceae, Cyperaceae, Eriocaulaceae and Gentianaceae in the Serra do Ribeiro is related to favorable local characteristics, as the presence of slopes with moist or soaked soils, a typical environment for given species of these families, as Burmannia bicolor Mart., Rhynchospora consanguinea Boeck.and Syngonanthus caulescens (Poir.)Ruhl.
According to Sano & Almeida (1998), campos rupestres often shelter single species clusters, whose presence is conditioned, among others factors, by soil moisture.In the Espinhaço Range, seasonality is evident, with heavy cloud cover during the winter, which causes high moisture, sporadic rains and a lot of dew, so that some regions have a waterlogged soil all year round (Giulietti & Pirani 1988).
The absence of pattern in species richness variation as altitude increases in both serras reflects the heterogeneity of the studied vegetation, which suggests the need of studies on the influence of local environmental and biological factors on the distribution of herbaceous plants, such as the availability of nutrients or water in the soil and competition or facilitation, respectively (Mallen-Cooper & Pickering 2008).The Espinhaço Range is a low altitude, very fragmented formation, which implies a small influence of macro-scale factors along the altitudinal gradient (Carneiro et al. 1995) as, for instance, climatic changes and the formation of different habitats (Whittaker et al. 2001).
Furthermore, the proximity between the sample points and the differences in elevation ranges and distances between them should also be considered to explain the absence of an altitudinal pattern.Nonetheless previous studies carried out on broader altitudinal ranges presented consistent variations in organism distribution, which resulted in a statistically significant relationship between species richness and altitude (Gottfried et al. 1999;Kessler 2000;Jones et al. 2003;Grytnes 2003;Bachman et al. 2004).
Since the serras presented a high number of species (44.5%) in common, the formation of the two groups observed in the UPGMA and MDS analyses probably reflects the different occurrence of species at the sample points of each serra, which, in turn, have peculiarities similar to insular systems in the determination of species richness (Conceição & Pirani 2007).The higher similarity found between contiguous SR samples is probably due to the low species richness variation between them, since great part of the species of this serra grow in various sample points.
Nevertheless, despite the small area sampled at each point, the marked difference between points 9 and 10, in the SR, may be related to their isolation in the landscape.In addition, the occurrence of species in specific places and the concentration of species in a single sampling point (Tab.2) and the presence of different species richness between samples at similar altitude (Fig. 2) point out the importance of local factors in the species composition found, which contribute to the formation of mosaics in campos rupestres (Conceição & Pirani 2005).

Figure 1 -
Figure 1 -Localization of the study areas in the southern Espinhaço Range, Minas Gerais, Brazil.

Table 1 -
Localization, spot height and description of the collection points in the Serra do Ouro Branco (OB) and in the Serra do Ribeiro (SR).

Table 2 -
Presence (+) and absence (-) of species of herbaceous plants along the altitudinal gradient in the serras do Ouro Branco and do Ribeiro, located in the Espinhaço Range, Minas Gerais, Brazil.Voucher material is represented by the collection number of R.A.X.Borges (B).