Ant fauna in megadiverse mountains : a checklist for the rocky grasslands

The rocky grasslands, locally known as campo rupestre, are an ecosystem characterized by a montane, fire-prone vegetation mosaic, with rocky outcrops on quartzite, sandstone, or ironstone soils. They are inserted in a matrix of sandy and stony grasslands, and other vegetation types, such as Cerrado (Brazilian savanna), Atlantic Forest, and Caatinga (Giulietti & Pirani, 1997; Alves et al., 2014; Fernandes et al., 2014). Rocky grasslands occur mainly along the Espinhaço Mountains, a vast mountain range that has its southern limit in the state of Minas Gerais, southeastern Brazil, encompass important smaller ranges, such as Serra do Caraça and Serra do Cipó, and ends in Chapada Diamantina, state of Bahia, northeastern Brazil (Giulietti & Pirani, 1997). Rocky grasslands are also found in the mountains of central (e.g., Serra da Canastra) and southeastern Brazil (e.g., Serra da Mantiqueira), whose geology and flora resemble those of the Espinhaço Mountains (Giulietti & Pirani, 1997; Rapini et al., 2008; Vasconcelos, Abstract The rocky grasslands, environments locally known as campo rupestre, occur mainly along the Espinhaço Mountains and are considered local centers of biodiversity and endemism in Brazil. However, knowledge of ant species richness (Hymenoptera: Formicidae) in this kind of environment is still poor. Aiming at filling this gap, we compiled information from empirical studies and literature records. We found a total of 288 species of 53 genera and eight subfamilies recorded in rocky grasslands. Myrmicinae and Formicinae were the most representative subfamilies, with 53% and 18% of the total species richness, respectively. The genera with the largest number of species were Pheidole (41) and Camponotus (40). This large number of ant species recorded for the rocky grasslands surpasses those found in other studies conducted in several different places. Ant species richness decreased with altitude; most species occur below 800 m a.s.l. (171), and only a few species occur above1600 m a.s.l. (17). Some genera occur only at a specific altitude (e.g., Azteca and Dolichoderus at 800/900 m a.s.l.; Leptogenys and Labidus at 1400 m a.s.l.), which points out to the potential use of ants as biological indicators. Our results suggest that the rocky grasslands favor high ant diversity. The patterns of ant richness associated with the altitudinal gradient reinforce the idea of considering the rocky grasslands as priority areas for biological conservation. Moreover, we observed a lack of records on the occurrence of most ant species considered in the present study (93%), which shows that Brazilian myrmecologists need to invest more in taxonomy, management, and data sharing. Sociobiology An international journal on social insects


Introduction
The rocky grasslands, locally known as campo rupestre, are an ecosystem characterized by a montane, fire-prone vegetation mosaic, with rocky outcrops on quartzite, sandstone, or ironstone soils.They are inserted in a matrix of sandy and stony grasslands, and other vegetation types, such as Cerrado (Brazilian savanna), Atlantic Forest, and Caatinga (Giulietti & Despite the information on the flora and fauna of rocky grasslands (Giulietti et al., 1987;Lessa et al., 2008;Rapini et al., 2008;Rodrigues et al., 2011), sampling in those environments has been heterogeneous and large areas remain unexplored (Madeira et al., 2008).In addition, most of the literature on the biodiversity of rocky grasslands focus on plants and vertebrates (Silveira et al., unpublished data).Therefore other groups, such as invertebrates, remain unknown.Some of the challenges to invertebrate conservation are a scarce and underfunded basic research, and the overlooking of invertebrates in most conservation policies (Cardoso et al., 2011).
Conversely, invertebrates dominate most terrestrial environments and deliver several ecosystem services (Cardoso et al., 2011).Among invertebrates, ants (Hymenoptera: Formicidae) represent one of the most important and abundant terrestrial groups (Hölldobler & Wilson, 1990).Well known for their functional roles, ants have been used as bioindicators due to their sensitivity to environmental and climate changes (Lach et al., 2010).Previous studies reported changes in ant diversity along altitudinal gradients (Fisher, 1996;Bharti & Sharma, 2013;Bishop et al., 2014).Altitudinal gradients are excellent to model species distribution, due to differences in abiotic conditions.However, as most literature records came from temperate mountains, the lack of information on tropical mountains makes it difficult to elaborate conservation plans.
What we know about ants from rocky grasslands comes from case studies on ant species associated with a particular plant or area (e.g., Guerra et al., 2011;Viana-Silva & Jacobi, 2012;Fagundes et al., 2013).However, a complete record that comprises the whole diversity and distribution of ants is still missing.In order to fill this gap of knowledge and support invertebrate conservation in the rocky grasslands, we need a more thorough biodiversity survey.In the present study, we compiled a checklist of ant species and their occurrence from original data and published information.

Data sampling and database
We searched for studies carried out in areas of rocky grasslands that informed the geographic coordinates of their sampling sites and identified ants to the species.We compiled records from a total of eight datasets, most of which original and collected in Serra do Cipó (Lana, 2015).Seven other datasets were found through an online survey in the Web of Knowledge, other academic search engines such as Google Scholar, and Brazilian academic libraries.Among those sources is one unpublished dataset from Serra do Cipó (hereafter "Cipó") (Soares, 2003) and six published datasets from Cipó (n = 1), Serra do Rola Moça State Park (hereafter "Rola Moça") (n = 1), Itacolomi State Park (hereafter "Itacolomi") (n = 3), and Ibitipoca State Park (hereafter "Ibitipoca") (n = 1).All studied sites are located in the Espinhaço Mountains, except for Ibitipoca.
As we aimed at providing a broad inventory of the ant fauna, we used studies carried out with different sampling efforts and methods.Table 1 describes the samples collected from the literature, including information on sampling method, environment, and location.Details on species identification are given for each study.19°16'17.8'' S, 43°36'18.1'' W 19°15'50.6'' S, 43°35'10.3'' W 19°13'56.5'' S, 43°34'34.8'' W 19°17'43.0'' S, 43°33'17.4'' W 19°17'49.6" S, 43°35'28.2" W 19°16'59.3In the original dataset (Lana, 2015) ants were sampled in seven sites during the Long Term Ecological Research of the Rocky Grasslands of Serra do Cipó (PELD-CRSC, in the Portuguese acronym).Those seven sites were chosen along an altitudinal gradient in Cipó, from 800 to 1,400 m a.s.l.In each site, three transects of 200 m were set up and five sampling points were established at 50 m from each other.In 2011 and 2012, ants were sampled quarterly, mainly with pitfalls traps, but also with beating, sweep nets, and Winkler traps.Ants were identified using a taxonomic key (Fernández, 2003), by comparison with specimens deposited in the Laboratório de Mirmecologia do Centro de Pesquisas do Cacau (CEPEC/CEPLAC), and by consulting specialists (Jacques H. C. Delabie).Nomenclature followed Bolton et al. (2005), with posterior improvements made available on the Online Catalog of the Ants of the World (AntCat).
We built a complete species checklist (Appendix) with information on several ant species organized by study, sampling locality, and altitude.As it was not possible to match the morphospecies hosted in different institutions and collections, we included in the analysis only one record for each morphospecies, regardless of its potential presence in more than one study, area, or altitude.Each morphospecies (e.g., Pheidole sp.1) probably represents more than one species, as the same nomenclature was established by different authors.However, excluding those records or trying to tell them apart could interfere with the estimation of the real diversity.Although we are aware of this taxonomic limitation, in face of the difficulty of assigning names to several Neotropical ant species (e.g., Camponotus and Pheidole) and the lack of current taxonomic revisions for many species-rich genera (e.g., Brachymyrmex, Cyphomyrmex, and Solenopsis) (Lach et al., 2010), this is the most parsimonious option for a study that aimed at estimating ant species diversity on a broad scale.

Study sites
The Espinhaço Mountains are 50-100 km wide and 1,200 km long, and encompass several mountains (up to approximately 2,000 m a.s.l.) (Giulietti & Pirani, 1997).Rocky grasslands occur mostly from 900 to 2,033 m a.s.l.In the basal part of the range, at altitudes between 800-1,000 m a.s.l., we found ecotones between savanna and rocky grassland.Trees and shrubs are the most common life forms at lower altitudes, but their predominance decreases with altitude, as the soil profile also changes, and they gradually give way to outcrops and grasslands (Giulietti & Pirani, 1997;Alves et al., 2014).Similarly, together with the altitudinal gradient there is also a climate gradient, in which the mountaintop is colder and moister than the base (Giulietti & Pirani, 1988).
Serra do Cipó is located in the southeastern part of the Espinhaço Mountains, state of Minas Gerais, southeastern Brazil (Fig 1).This region has a diversified mosaic of vegetations, which varies with soil type and altitude (from 800 to 1.600 m a.s.l.).This environment is covered by a low vegetation composed of shrubs and small trees and abundant grasses and sedges.There are also several watercourses, along which gallery evergreen forests grow (Giulietti et al., 1987).
Itacolomi State Park and Serra do Rola Moça State Park are characterized by rocky grasslands that grow on ironstone, locally known as canga.Both areas are located in the Iron Quadrangle of Minas Gerais, a 7,200 km 2 region in the southern part of the Espinhaço Mountains (Fig 1).The Iron Quadrangle is geologically dominated by ironstone and represents one of the world's main mineral provinces (Jacobi et al., 2011).Itacolomi comprises an altitudinal range varying from 700 to 1,772 m a.s.l. and a mosaic of rocky grasslands, canga, semi-deciduous montane forest, and associated vegetation types (Gastauer et al., 2012).Rola Moça is located in an ecotone between the Cerrado and Atlantic Forest biomes, which comprises several vegetation types, such as Cerrado, semi-deciduous forest, riparian forest, and prominent rocky grasslands developing on canga on the mountaintops (1,200 -1,500 m a.s.l.) (Jacobi et al., 2008).
Ibitipoca State Park is a protected area located in Serra da Mantiqueira, southern Minas Gerais (Fig 1).This site is characterized by a vegetation type composed of grasses, herbs, and shrubs on outcrops of quartzite rocks associated with shallow soils and high sun incidence (Dias et al., 2002).
Among the identified species, only Camponotus crassus (Mayr 1862) occurred in all sites.Only morphospecies of Pheidole exhibited similar distribution.Thus, Camponotus and Pheidole emerged as the most widespread genera currently recorded for rocky grasslands.The dominance of those genera is consistent with the patterns suggested for other Neotropical ants (Fernández & Sendoya, 2004) and similar ecosystems, such as open Cerrado (Ribas et al., 2003;Campos et al., 2011;Pacheco & Vasconcelos, 2012).Likewise, Myrmicinae and Formicinae were also the most prominent subfamilies in ant inventories conducted in different environments, such as Cerrado (Ribas et al., 2003;Campos et al., 2011), Amazon (Miranda et al., 2012), andCaatinga (Ulysséa &Brandão, 2013).
The large number of ant species recorded for rocky grasslands (288) deserves attention, as other studies carried out in wider geographical ranges found a smaller or similar number.For example, checklists made for the Caatinga (Ulysséa & Brandão, 2013) and Amazon (Miranda et al., 2012) found 173 and 276 species, respectively.Although we did not find a comprehensive inventory for the Cerrado that could be used for comparison, studies carried out over large areas revealed about 150 species (Ribas et al., 2003;Campos et al., 2011;Pacheco & Vasconcelos, 2012).Actually, it is hard to compare number of species among studies or environments, as different studies used different sampling efforts and methods.
Nevertheless, by analyzing the map in Fig 1, we infer that the rocky grasslands have several sampling gaps (e.g., Chapada Diamantina, Serra da Canastra, northern and southern Minas Gerais).Considering this gap of knowledge, high endemism, and complex environmental mosaic found on those mountains, we expect ant diversity in the rocky grasslands to be even higher than observed in the present study (288).Moreover, the large number of unidentified species together with the inclusion of only one record per morphospecies point out to an underestimation of the number of ant species in the rocky grasslands.
We observed a decrease in ant richness along the altitudinal range (Appendix).The lowest altitudes, 800 and 900 m a.s.l., contributed with 171 and 127 ant species, respectively, whereas the highest altitude (1,600 m a.s.l.) had a smaller number of species (17).Only Solenopis occurred at all altitudes.Those findings corroborate the general diversity pattern of ants that live on mountains, in which the number of species decreases with altitude (Fisher, 1996;Brühl et al., 1999;Longino & Colwell, 2011;Bharti & Sharma, 2013).Nonetheless, very few studies have documented the altitudinal trends of ant biodiversity in Brazilian montane ecosystems (but see Araújo & Fernandes, 2003).
At 800 m a.s.l., there were 18 exclusive species, whereas at 1,400, 1,500, and 1,600 m a.s.l.there were 21, five, and two, respectively.We found some genera of dominant arboreal ants (Azteca and Dolichoderus) restricted to the mountain base (800/900 m a.s.l.), which indicates that altitude may restrict ant occurrences.Similarly, at 1,400 m a.s.l., we recorded some unique genera, such as specialized predators (Leptogenys) and legionary ants (Labidus) (Brandão et al., 2012).Those findings corroborate the potential of ants as bioindicators, especially of climate change.Similar patterns of restriction of functional groups to particular altitudes have already been observed for tropical (Brühl et al., 1999) and temperate regions (Bharti & Sharma, 2013).However, those findings may have been biased by the sampling effort used in each site.Ant responses to altitude, associated with the high richness found in a small geographic area, point to the importance of conserving the rocky grasslands.This conclusion is consistent with the strategies recommended for ant conservation, which state that efforts should be targeted to high biodiversity, high endemism, and extremely threatened areas (Alonso, 2010).We also bring to light the need for investing in ant taxonomy, database management, and data sharing, which are essential tools for biodiversity conservation, though they are neglected in most Brazilian research institutions.Those gaps of knowledge became clear when we searched for information on ant species occurrence and distribution (only those with complete taxonomic identification) in online databases (Antwiki, AntWeb, CRIA speciesLink) and specialized catalogues (Kempf, 1972;Brandão, 1991).We noticed that most ant records (94%) neither were followed by a formal record for the rocky grasslands nor were hosted in databases (Appendix, symbol * ).Approximately 5% of the records contained no information on geographic distribution and only 1% accounted this kind of information for the rocky grasslands.Therefore, myrmecology in Brazil needs to invest strongly in taxonomy and species inventories.Despite some recent advances, rocky grasslands still have many sampling gaps for ants, and, therefore, they need more efforts in conservation.
Appendix.Occurrence and distribution of ant species along the altitudinal gradient, localities, and studies carried out in the Brazilian rocky grasslands.Sampling sites: Serra do Cipó (SC), Itacolomi State Park (IT), Ibitipoca State Park (IB), and Serra do Rola Moça State Park (RM).Numbers in columns indicate the literature sources described in Table 1.
*Species with no occurrence in the sampling area; ‡ Species with no information on distribution; †Species with occurrence recorded for the sampling site.*Species with no occurrence in the sampling area; ‡ Species with no information on distribution; †Species with occurrence recorded for the sampling site.

Fig 1 .
Fig 1.The location of rocky grasslands in Brazil (A).Rocky grasslands along the Espinhaço Mountains and other Brazilian mountain ranges (circumscribed) (B).Sampling sites in the southern part of the Espinhaço Mountains and the southern mountains of Minas Gerais (C).

Fig 2 .
Fig 2. Proportion of ant species exclusive to each site and shared between at least two sites.Sampling site abbreviations: Serra do Cipó (SC), Itacolomi State Park (IT), Ibitipoca State Park (IB), and Serra do Rola Moça State Park (RM).Dark bars represent exclusive proportion and light bars correspond to the shared proportion of species.
Occurrence and distribution of ant species along the altitudinal gradient, localities, and studies carried out in the Brazilian rocky grasslands.Sampling sites: Serra do Cipó (SC), Itacolomi State Park (IT), Ibitipoca State Park (IB), and Serra do Rola Moça State Park (RM).Numbers in columns indicate the literature sources described in Table1 (Continuation).*Specieswith no occurrence in the sampling area; ‡ Species with no information on distribution; †Species with occurrence recorded for the sampling site.Occurrence and distribution of ant species along the altitudinal gradient, localities, and studies carried out in the Brazilian rocky grasslands.Sampling sites: Serra do Cipó (SC), Itacolomi State Park (IT), Ibitipoca State Park (IB), and Serra do Rola Moça State Park (RM).Numbers in columns indicate the literature sources described in Table1 (Continuation).*Specieswith no occurrence in the sampling area; ‡ Species with no information on distribution; †Species with occurrence recorded for the sampling site.Occurrence and distribution of ant species along the altitudinal gradient, localities, and studies carried out in the Brazilian rocky grasslands.Sampling sites: Serra do Cipó (SC), Itacolomi State Park (IT), Ibitipoca State Park (IB), and Serra do Rola Moça State Park (RM).Numbers in columns indicate the literature sources described in Table1 (Continuation).*Specieswith no occurrence in the sampling area; ‡ Species with no information on distribution; †Species with occurrence recorded for the sampling site.Occurrence and distribution of ant species along the altitudinal gradient, localities, and studies carried out in the Brazilian rocky grasslands.Sampling sites: Serra do Cipó (SC), Itacolomi State Park (IT), Ibitipoca State Park (IB), and Serra do Rola Moça State Park (RM).Numbers in columns indicate the literature sources described in Table1 (Continuation).*Specieswith no occurrence in the sampling area; ‡ Species with no information on distribution; †Species with occurrence recorded for the sampling site.Occurrence and distribution of ant species along the altitudinal gradient, localities, and studies carried out in the Brazilian rocky grasslands.Sampling sites: Serra do Cipó (SC), Itacolomi State Park (IT), Ibitipoca State Park (IB), and Serra do Rola Moça State Park (RM).Numbers in columns indicate the literature sources described in Table 1 (Continuation).
Occurrence and distribution of ant species along the altitudinal gradient, localities, and studies carried out in the Brazilian rocky grasslands.Sampling sites: Serra do Cipó (SC), Itacolomi State Park (IT), Ibitipoca State Park (IB), and Serra do Rola Moça State Park (RM).Numbers in columns indicate the literature sources described in Occurrence and distribution of ant species along the altitudinal gradient, localities, and studies carried out in the Brazilian rocky grasslands.Sampling sites: Serra do Cipó (SC), Itacolomi State Park (IT), Ibitipoca State Park (IB), and Serra do Rola Moça State Park (RM).Numbers in columns indicate the literature sources described in Occurrence and distribution of ant species along the altitudinal gradient, localities, and studies carried out in the Brazilian rocky grasslands.Sampling sites: Serra do Cipó (SC), Itacolomi State Park (IT), Ibitipoca State Park (IB), and Serra do Rola Moça State Park (RM).Numbers in columns indicate the literature sources described in Occurrence and distribution of ant species along the altitudinal gradient, localities, and studies carried out in the Brazilian rocky grasslands.Sampling sites: Serra do Cipó (SC), Itacolomi State Park (IT), Ibitipoca State Park (IB), and Serra do Rola Moça State Park (RM).Numbers in columns indicate the literature sources described in Occurrence and distribution of ant species along the altitudinal gradient, localities, and studies carried out in the Brazilian rocky grasslands.Sampling sites: Serra do Cipó (SC), Itacolomi State Park (IT), Ibitipoca State Park (IB), and Serra do Rola Moça State Park (RM).Numbers in columns indicate the literature sources described inTable 1 (Continuation).*Specieswith no occurrence in the sampling area; ‡ Species with no information on distribution; †Species with occurrence recorded for the sampling site.

Table 1 .
Sites where ants were sampled in the Brazilian rocky grasslands.

Table 1 (
Continuation).Species with no occurrence in the sampling area; ‡ Species with no information on distribution; †Species with occurrence recorded for the sampling site. *

Table 1 (
Continuation).Species with no occurrence in the sampling area; ‡ Species with no information on distribution; †Species with occurrence recorded for the sampling site. *
*Species with no occurrence in the sampling area; ‡ Species with no information on distribution; †Species with occurrence recorded for the sampling site.

Table 1 (
Continuation).Species with no occurrence in the sampling area; ‡ Species with no information on distribution; †Species with occurrence recorded for the sampling site. *