The karst area of São Domingos and its surroundings represents one of the regional formations of the Bambuí geomorphologic unit, which constitutes the largest set of limestone areas favourable to caves development in Brazil. The region has an approximate area of 105, 200 km². In the region there are five large systems of caves, with up to 23 km of development (Figure 2).

Superficial rivers belonging to the Paranã Basin (Alto Tocantins) penetrate the limestone layers after draining an extensive arenitic region, forming large cave systems (Karmann and Setúbal 1984, Auler and Farrant 1996). As part of these, there are upper tributaries in the vadose zone fed mainly by infiltrating waters from the surface.

The area is part of a state-level Conservation Unit (Terra Ronca State Park - TeRSP), created 17 years ago but still with diverse land ownership problems. One aggravating factor is that some river sources that cross the cave systems are located outside the park. There are anthropic activities as livestock, extensive plantations and clandestine mining which on long-term are responsible for the adverse effects on local drainage systems, which include the subterranean domain.

The studied region belongs to the morphoclimatic Dominion of the Cerrados (Ab’Saber 1977), with a dry season between May and September, sometimes extending through October (Nimer 1979). Heavy showers characterize the rainfall regime during the rainiest periods (October through March), transporting a large amount of organic matter in the caves (Trajano and Bichuette 2010).

The sampling area embodies subterranean systems and epigean rivers, sampled differently regarding frequency and volume since in some cases the availability of water was restricted. Abbreviations and characteristics of each sampled point are described in Table 1.

The region was sampled on five occasions: April and October 2011, and February, June and October 2012. The subterranean stretches included drips, subterranean rivers (base level) and water pools (Figure 3). The sampled caves and rivers are: Angélica, Bezerra, Buraco das Araras, São Bernardo, São Mateus and Revolucionários caves; the epigean Palmeiras, Angélica, da Lapa, São Domingos Rivers, and the Resurgence of Terra Ronca (Table 1). Revolucionários Cave and São Domingos River are outside the park area, but inserted in the same limestone lens. The Angélica Cave is the only locality sampled on every trip, as a long-term monitoring base.

Samples were taken with plankton nets (20µm mesh), which remained installed for periods varying from twenty minutes to three hours, according to the local conditions. Nets, installed under dripping water sites based on the device proposed by Brancelj (2004) were used to sample the fauna percolating through the unsaturated zone for similar periods. For the shallow stretches, the flow (used to calculate the filtered volume) was estimated from the velocity of the water surface. This velocity was determined by the time it took for a small plastic disc to travel a pre-determined distance. In the case of percolating water and pools, the volume was directly estimated using a graduated plastic container. The material was subsequently fixed (formaldehyde 4%) and packed. Afterwards, the samples were analysed in the laboratory of Limnology in the Biology Department of FFCLRP/USP (Ribeirão Preto campus), and the organisms were identified to the lowest possible taxonomic level.

The following physical and chemical variables of water were measured by using a Horiba multiprobe (model U50G): temperature (°C), pH, ORP redox potential (mv), conductivity (mS/cm), dissolved oxygen (mg/L), oxygen saturation (%), total of dissolved solids TDS (g/L), turbidity (NTU), depth (m) and salinity (%).

Relative taxa richness among sites was assessed. Furthermore, the richness of the taxa in epigean and hypogean environments was taken into account separately for comparison. Similarity in composition of aquatic invertebrates in the different habitats was analysed by cluster analysis (Sorensen, single linkage).

In order to study the species-environment relationship, Canonical Correspondence Analysis (CCA) was used, associating the measured environment variables with the matrix of taxa present at eleven sampled sites.

Finally, by means of simple linear regression (Pearson correlation), we verified the relationship between the sampling effort and the richness of taxa, as the volumes sampled in each segment varied. Analyses of similarity and CCA were performed using the software PAST (version 2.13) and R language (version 3.0.1), respectively.

In this preliminary approach, we studied the richness of taxa in habitats of karstic systems, which represents new data on tropical subterranean domain species. The fauna belonging to samples of percolating water were remarkable and responsible for the high number of recorded species in the studied habitats

Rotifers were the richest group in our study, representing more than half of the total richness recorded. The dominance of rotifers was associated with the increase in the trophic state of a system due to short life cycles and rapid reproduction, which favour this group in more dynamic, competitive and selective environments. Presently, it is known that rotifers are also dominant in several other aquatic environments, independent of the trophic state (Rocha et al. 1995). This group species have greater tolerance to different environmental conditions, and present short reproductive cycles and high renewal rate, providing an advantage for the formation of well-established populations (Neves et al. 2003, Segers 2008).

Diversity in subterranean water courses is frequently limited by competition, and especially limited by energetic resources and space (Culver 1994). In Brazilian karst, the subterranean aquatic microfauna certainly configures an important source in the diets of larger taxa, mainly for fish populations and aquatic invertebrate groups that inhabit the most inaccessible points of the unsaturated zone. Rotifers are considered a source highly rich in energy, as they make available a large quantity of renewable food through conversion of primary production in tissue that is easily assimilated by consumers (Esteves 1998). In dry periods, when the level of the phreatic zone tends to suffer brisk reductions, some fish become imprisoned in pools, without record of populations in base-level streams (e.g. populations of catfishes Ituglanis epikarsticus Bichuette & Trajano, 2004 of São Mateus Cave). Therefore, it is possible that rotifers act as an essential component of the food chain, offering rapid renewal of nutrients for the other trophic levels.

The low number of Copepoda occurrences in our samples must be due to environmental factors and/or interspecific relationships in the communities. In dry periods, the water volume flowing through the system can decrease, as well as the access to aquatic communities from unsaturated zone habitats. In fact, some of our sampling points could not be efficiently accessed in all periods, in part because of the reduced local rain rates. Besides, sampling frequency could have also affected our results to the extent that it did not offer access when abundant taxa of these communities occurred (i.e. different life cycles). We also have to consider that benthonic fauna was not well assessed.

The fauna distribution depends on the scale of the karst system and generally species change in number from the surface to the saturated zone (Gibert et al. 2009). The most relevant factors to the composition of aquatic biota consist of physical and chemical variables of water, the thickness of rocks and the rainfall regime, as they directly influence waters properties and volume of the karst systems (Pipan et al. 2006, Kogovšek 2010, Meleg et al. 2011, Pipan and Culver 2012). The most relevant physical and chemical factors explaining the composition of aquatic biota were considered the thickness of rocks and the rainfall regime, as they directly influence waters properties and volume of the karst systems (Pipan 2006, Kogovšek 2010, Meleg et al. 2011, Pipan and Culver 2012). The environmental variables used in our correlation analysis have low significance in explaining the different faunistic distributions. Certainly, there are other more important factors that have not been considered, such as the dissolved organic carbon and concentration of ions (e.g., sodium, nitrate and potassium). Until present, few studies have considered the levels of carbon, phosphorous and nitrogen in the waters (e.g., Pipan et al. 2006, Simon et al. 2007), even if the dissolved organic carbon is an important limiting factor (Simon and Benfield 2002). Even if we did not measure these values, we can infer the availability of organic carbon in percolation water for the high number of taxa found in drips.

We observed that drips samples have the most unique fauna of all the other subterranean habitats. In our study, it is important to highlight that the water volumes filtered in drips were much smaller than the volume of water filtered in other habitats. However, we encountered elevated richness and occurrence of taxa exclusive to these samples. Thus, we recorded the greatest richness of taxa from the least voluminous samples (drips of Angélica Cave and epigean resurgence), while some of the most voluminous ones had inferior richness (São Domingos, Lapa and Angélica epigean rivers). The total of taxa encountered in drips outnumbered all others samples, emphasizing the complexity of communities from the unsaturated zones.

For the environmental parameters considered, we have observed that there was a separation between epigean and hypogean sites in the diagram generated by CCA as well as in the dendrogram of faunistic similarity (except for SDEp and RevS sites). Both demonstrated that there were clear differences between the biota of epigean and groundwater environments while distinctions of hypogean habitats among themselves are less evident. When comparing each sample individually, we noticed faunistic distinction between them.

Epikarst and sinkholes are both responsible for supplying the aquatic hypogean realm, but they differ in their capacities to redirect and/or store water through the karst (Juberthie 2000). The small size of voids of the epikarstic and the vadose zone restricts the transport of particles and microorganisms, and also represents a much slower flow. However, the total area occupied by the unsaturated zone is much more ample and complex, permitting innumerable vertical and lateral connections. These factors interact in a heterogeneous way, generating various microhabitats and promoting the formation of distinct communities (Moldovan et al. 2012).

The waters that percolate the subterranean systems are filtered by soils on the karst surface and tend to suffer reduction of dissolved oxygen. Organic material, carbon, and diverse ions suffer reduction of dissolved oxygen, depending on the manner in which they are used and transformed through chemical processes by the interstitial and/or unsaturated zone biota. Water circulation in surface rivers promotes higher oxygenation rates than in the subterranean environment, which generally are formed by smaller and less dynamic or lentic systems (Malard and Hervant 1999, Pipan and Culver 2005).

Redox potential, the second most important variable in the analysis, has negative correlation to concentrations of oxygen, suggesting that reductive processes predominate in the absence of oxygen. Low levels of ORP can be associated with environmental pollution and/or high concentrations of ions, as occurs in waters with karstic origins (Drew and Hötzl 1999). The uncontrolled anthropic activities in the surroundings of the studied area can promote silting and pollution in drainages systems, including the subterranean domains. Most taxa recorded in subterranean habitats were rotifers, which appeared associated with alkaline pH and elevated conductivity. However, the other registered taxa were not strongly related to the measured parameters.

Fong and Culver (1994) found for some caves of the United States that the first order watercourses fed by percolation water have higher diversity than others of higher order. Generally, the water that drains directly from the ceiling can supply a better approximation of physical and chemical characteristics of the upper horizon. However, once in water pools, it frequently presents more different environmental conditions than in the epikarst.

Comparing the results obtained with studies on the unsaturated zone in temperate regions (Brancelj 2002, Pipan and Brancelj 2004a, Pipan et al. 2006; Moldovan et al. 2007, 2012, Meleg et al. 2011, Cottarelli et al. 2012), the tropical aquatic subterranean biota (mainly from the unsaturated zone) has a distinct composition. The richness demonstrated was very high compared to records from more extensive approaches, with a sampling effort along a couple of years (e.g., Pipan and Brancelj 2004a, Pipan et al. 2006). Furthermore, Copepoda was not the group with the highest diversity, as in samples of percolating water from many temperate regions, mainly in Europe. However, none of these studies have used nets with such thin mesh (smaller than 40µm), so the richness of Rotifera may have been underestimated. We also have to emphasize that the sampling method we used could have underestimated the number of benthonic taxa, such as Harpacticoida, commonly find in samples of groundwater (Pipan et al. 2008, Moldovan et al. 2012).

The presence of exclusive taxa at some sites emphasizes the singularity of the studied systems. There are faunistic differences between sites belonging to the same limestone lens and geographically neighbouring. Many species inhabiting the void network of the unsaturated zone are known to have a linear distribution of only a few hundred metres (Pipan et al. 2006, Moldovan et al. 2012), and a discontinuous distribution, caused by different environmental condition (Brancelj 2002, Meleg et al. 2011, Moldovan et al. 2012).

The structure of the unsaturated zone can vary significantly in different regions according to its lithology, geomorphic history, as well as environmental parameters and seasonal variables (Klimchouk 2004). The wide variability of rainfall regime in tropical zones is expected to impact on the tropical subterranean fauna and to offer different models of distribution and diversity than in temperate karst zones.