Taxonomic distinctness and conservation of a new high biodiversity subterranean area in Brazil

Subterranean environments, even though they do not possess a primary production (photosynthesis), may present high biodiversity, faunistic originality, endemism, phylogenetic isolations and unique ecological and/or evolution events, in addition to rare taxa. Studies investigating the biological diversity in Neotropical caves are relatively rare and recent, and most of them have been conducted in Brazil. We sampled caves from the state of Bahia, northeastern Brazil, and through sampling sufficiency tests and richness estimators, we demonstrate that the normatization for the Brazilian cave laws is not adequate for its conservation and that only α diversity index is not enough to verify faunistic patterns. We suggest that a phylogenetic diversity index be more robust and accurate for conservation purposes, particularly the Taxonomic Distinctness index. Moreover, we propose that the sandstone complex caves from Chapada Diamantina National Park need to be classified as being of high subterranean biodiversity in a global scope.


INTRODUCTION
The subterranean environment comprises interconnected subsurface spaces that allow for the dispersion of species that inhabit there (Juberthie 2000).The hypogean realm presents faunistic originality and high endemism, as well as phylogenetic isolation (Gibert and Deharveng 2002) due to unique evolutionary events (Culver and Pipan 2009).Thus, some authors consider some subterranean systems as hotspots of biodiversity (e.g., Culver andSket 2000, Deharveng 2005).
Nevertheless, Myers (1990) proposed the term "hotspots" based on floristic diversity.The term is not used here due to its falsifiability in environments that lack primary producers, which may include subterranean systems.Therefore, here we used a new term: "High Biodiversity Subterranean Area (HBSA)".
Cave populations can be divided into three ecological-evolutionary categories.Troglophiles (populations well established inside and outside of the caves); trogloxenes (populations inhabiting caves but must go out to complete their life cycles) and troglobites (exclusively and obligatory subterranean populations) (see Schiner 1854, Racovitza 1907, Barr 1968, Poulson and White 1969).
Troglobites generally present autapomorphic character states (troglomorphisms) related to the life in permanent darkness (Trajano 1993).
The most known troglomorphisms are reduced or lack of eyes, elongated appendages and lighter coloration (Christiansen 1962).In general, troglobitic populations are specialized and fragile (Trajano and Bichuette 2006), and they need to be carefully considered in conservation programs.
Regarding the protection of Brazilian speleological patrimony, the current legislation (DECREE 6640 from November 7 th 2008) regulates the use of natural subterranean cavities for environmental bids, classifying them by relevance level: maximum, high, medium and low.Only caves with maximum relevance are entirely protected from suppression.
To regulate this decree, Normative Instruction (NI) N. 2 from August 20 th 2009 was published.In article 16 (paragraph 2) of this decree, some requirements of speleobiological studies were established, including the definition of the study period: "Art.16.The approval of speleological studies for classifying the relevance degree of natural subterranean caves is conditioned on the submission of sufficient information for the comprehension of the cave ecosystem.§ 2º The biological surveys must be completed in at least an annual cycle with at least two samplings per year, one occurring during the rainy season and the other during the dry season, aiming to minimally reveal the characteristics resulting from climatic seasonality." In addition to the proposal of only two samplings during one annual cycle, many biologists use only alpha diversity indexes such as Shannon (H') in the environmental impact reports to determine the amount of protection, or total destruction, of Brazilian caves (M.E.Bichuette and J.E. Gallão pers.obs.).Although these indexes are useful in some ecological studies, they must be used with extreme caution and parsimony (Trajano et al. 2012).
Considering these issues, we tested some ecological concepts applied to biological cave conservation.We chose a region in northeastern Brazil with sandstone caves and a high diversity of cave animals.We attempted to answer the following questions: (1) Are two samplings enough to access the minimum richness of subterranean fauna and to detect faunistic patterns?(2) Are α diversity indexes suitable for cave conservation purposes/decisions?

MATERIALS AND METHODS
We conducted the study in 11 sandstone caves from Chapada Diamantina, central Bahia northeastern Brazil, belonging to Serra do Espinhaço and cut by the Paraguaçu rivers basins (Fig. 1).
Rio dos Pombos cave (RPc), Ressuegência do Morro de Alvo cave (RMAc), and Torras System (TS) were sampled five times, and Parede Vermelha cave (PVc) was sampled six times.The samples were collected through an extensive visual inspection in different substrates (humid places, walls and ceiling, soil, guano, under rocks and trunks), covering the maximum extension possible in all habitats.
We did not use pitfall traps because we did not wish to impact the subterranean populations (Sharratt et al. 2000) and because in these sandstones caves, the rocks are exposed and do not allow for the installation of pitfall traps.Regarding the aquatic fauna, we used hand nets and Surber traps.The collection effort totaled approximately eight hours per cave, always in three or four sessions.Chiropteran fauna were not studied.
In order to determine if the sampling was enough in relation to the given access, i.e the minimum species richness, Mao Tao and Jackknife first order sampling rarefaction curves were calculated for those caves in which we had more than three samples.The analyses were performed using EstimateS version 9.1 software (Cowell 2013).
We also estimated the traditional Shannon (H') α diversity index, which is influenced by the sampling effort and requires the number of specimens (Clarke and Warwick 1998).These were calculated only for localities with more than three samples.
In addition, we calculated the Taxonomic Distinctness (TD) index and created a respective funnel graph for all sandstone caves because it is totally independent from the sampling effort (Warwick and Clarke 1998).All indexes were measured to determine which one would be the most robust in terms of cave conservation.The Species that showed morphological troglomorphisms (mainly reduced or lack of eyes and/ or melanic pigmentation) but we were unsure as to whether they were real troglobites were left here only as troglomorphic species (TM in Appendix S1) (Supplementary Material).Of course, some of these troglomorphic species could change their status to troglobitic species, if we confirm its occurrence exclusive to caves.

RESULTS
We recorded an abundance of 1,307 individuals distributed in 162 taxa (Appendix 1).Among them, 23 were troglobites and another 14 species exhibited some type of morphological troglomorphism.Besides 20 new species confirmed already (Appendix S1).PVc was the richest cave, with 13 troglobites.
The lower relative proportion of collected species after the second collection occasion was 40.0% for TS and, for the other three caves, more than half of the species were collected after the second sampling (Table I).Shannon's index (H') ranged from 2.753 (TS) to 3.459 (RPc), and the richest cave (PVc) had the third lowest H' of 2.983 (Table I).
The addition of species at each collection was almost a constant (Fig. 3), including the addition of troglobitic species after the second sampling, which is the limit in the current Brazilian legislation regarding Environmental Impact Assessments/ Environmental Impact Reports (EIA/RIMA from Portuguese).In addition to these species, we noted that in none of the cases, even in those cases with six samples (such as in the case of PVc), was the asymptote reached, indicating that the sampling is far from adequate with only two collections.

DISCUSSION
Diversity in subterranean environments is better expressed regionally than locally because many passages are accessible only to fauna (Sket 1999).Considering Igatu, there are 162 registered species and 23 are troglobites, which are distributed in a small area of only 25 km 2 .Among troglobites, this region is the type-locality of four troglobites (Discocyrtus pedrosoi Kury 2008.Glaphyropoma spinosum Bichuette, de Pinna and Trajano 2008.Troglorhopalurus translucidus Lourenço, Baptista and Giupponi 2004 and the first Brazilian troglobitic Mygalomorphae, Tmesiphantes hypogeus Bertani, Bichuette and Pedroso 2013), and it holds endemismics at a local, regional and zoogeographical scale as a new species of the troglobitic Verhoeffiella genus (Collembola), which previously only had a Paleartic distribution.However, the total number of troglobitic species may rise because some species exhibit some type of troglomorphic trait, which corroborates this hypothesis.The troglobitic catfish is already included on the IUCN Red List (Gallão and Bichuette 2012).
If we consider a local approach, PVc presented 13 troglobitic species, and there is a great possibility that this number will increase, which renders this cave of extreme importance for the Brazilian subterranean fauna because the Areias System, Alambari de Cima and Olhos d'Água caves Furthermore, 25 troglobitic and troglomorphic species would have been neglected if the localities were sampled only twice (occasionally the same species occurred in different caves) (Table I).
Regarding the taxonomic distinctness, the expected Δ + was 67.985 (Fig. 2), and only two caves were below expectations; however, they were still inside the confidence interval.Six localities were above the expected Δ + and were in the funnel graph, and two caves were above the funnel graph: MAc (Δ + 85.286) and PVc (Δ + 72.062).comprise 20, 10 and eight troglobites, respectively, and are considered Neotropical caves of high biodiversity relevance (Deharveng 2005, Trajano andBichuette 2010).
Therefore, the sandstone caves from Chapada Diamantina represent a new Neotropical HBSA, under a local and regional focus.
As a consequence of incomplete sampling, species may be neglected, leading to erroneous estimations and resulting in incorrect decisions for management and conservation plans (Conroy andNoon 1996, Nichols et al. 1998).Two samples in an annual cycle, as proposed by Normative Instruction N. 2 for environmental bids, are absolutely insufficient to demonstrate faunistic patterns or even to determine the minimum species richness.
Corroborating this inadequacy, the rarefaction Mao Tau curves reveal that the asymptote is far from being achieved and that the addition of species in each collection event is a constant, including troglobitic or troglomorphic representatives (Fig. 3).This is in addition to the fact that at least 40% of species were registered starting from the third sampling occasion, as occurred in TS.However, it is worth noting that this cave is a huge duct tunnelshaped cave and is entirely washed out during each rainy season.For the other three tested caves, the percentage of species collected starting from the third sampling occasion was higher than 50%.
Jackknife first order richness estimators were elevated, even for PVc, the only cave with six samples (Fig. 3).Thus, the survey of the Igatu caves is still not completely inclusive and two samplings are insufficient to assess the richness of a cave and to consequently classify its relevance levels.
Despite being elevated, these indeces, as Jackknife first order used here, may underestimate the real diversity.Any estimation that involves time and space does not reveal the differences in species composition in relation to time, collection period and different localities in the same habitat (Novotny 1993).Furthermore, propositions for habitat conservation must include both the observed richness and the number of species still undiscovered (Santos 2003).
According to the new Brazilian laws, troglobitic species are crucial to cave preservation or suppression.In this study, 25 troglobitic plus troglomorphic species would have been neglected (Table I) if only two samplings were used for environmental bids, which would certainly result in a decrease in the relevance of the sampled caves.Consequently, in this hypothetical situation, these caves could be suppressed.Trajano (2010) states that at least 10 samplings would be necessary, in distinct annual cycles, to begin to characterize subterranean environments; even after 20 samplings, new troglobitic species may still be found.To illustrate this requirement, in six karstic European regions, 190 samples were collected on average, and even after 100 samplings, troglobitic fauna was still being found (Culver and Pipan 2009).
The alpha diversity indeces were similar for the four localities (Table I), which do not reveal a faunistic singularity.Additionally, PVc had the third lowest H' of the four caves.If we took only this index as a basis for conservation purposes, PVc would most likely be neglected.This cave had a higher number of total and troglomorphic species, and its singularity was not reflected in the H' alpha diversity index.
Studies based only on alpha indeces that may or may not be mandatory for cave conservation must be used with extreme caution.Information about the identity of species, their phylogenetic relationships and other characteristics are lost due to the species' abundance and richness as well as their faunistic singularities, which are the main justification for the conservation of subterranean environments; these factors are minimized in numerical diversity values (Trajano et al. 2012).Cianciaruso et al. (2009) affirm that the traditional indeces hide more than they reveal due to a great loss of information, and that such indeces are not sufficient for selecting locations, in this case caves, for conservation purposes (Sarkar 2006).Thus, it is inadvisable to use only α diversity indeces for biodiversity purposes, as occurs in advisory studies for cave suppression, which are not published because confidential contracts (M.E.Bichuette and J.E. Gallão pers.obs.).
We verify the high dispersion of points on the funnel graph (Fig. 2), which corroborates the faunistic singularity and the great phylogenetic diversity that exists in the Igatu caves.
Environments that are too degraded or polluted present a low TD, which may produce points below the 95% trust interval on the funnel graph (Warwick and Clarke 1998).Many caves of Chapada Diamantina were subjected in the past to heavy diamond mining, which continues in a residual, clandestine fashion at present (Bichuette et al. 2008).This activity could explain why the results were below expectations for some caves (Fig. 2).
Both caves that had points outside of the trust intervals were MAc and PVc, with six and 63 species, respectively.Despite its low richness in a single sample, MAc presented high phylogenetic diversity because all six species were distributed in four distinct classes, which explains the point outside of the trust interval.On the other hand, PVc presented a higher richness and a higher number of troglobites, items that also corroborate the presence of this cave outside the trust interval.All of the other caves are in the funnel graph and above expectations and corroborate the faunistic singularities found in the cave's biodiversity.
In conclusion, the Igatu region of central Bahia represents a HBSA for Brazil, as well as a Neotropical region and must be protected.
Two samplings, as stated in Brazilian laws for environmental bids, are insufficient for accessing the minimum richness of subterranean species and for predicting any pattern for subterranean fauna.

H
' diversity index was calculated in Past version 2.13 (Hammer et al. 2001) and the TD index was calculated in R version 2.13.1 (2011).

Figure 3 -
Figure 3 -Sampling rarefaction curves from caves with five or more collections and Jackknife first order estimator.PVc -Parede Vermelha cave; RPc -Rio dos Pombos cave; RMAc -Ressurgência do Morro de Alvo cave; TS -Torras System.For each cave, solid lines are Mao Tau sampling rarefaction curves and dashed lines are Jackknife first order richness estimator.Horizontal numbers means samples and vertical numbers means species.