The Chacoan bat fauna identity : Patterns of distributional congruence and conservation implications

The aim of this paper is to characterize the Chaco thorn-scrub savanna as a unit, biologically defi ned by means of its identifi cation as a regional pattern of distributional congruence, through the evaluation of the utility of bat assemblages as biogeographical indicators of the Chaco at regional scale, and to then evaluate this characterization in terms of the conservation value of this habitat. We have considered the whole territory of Paraguay and Bolivia, and the northern of Argentina; that is the entire Chaco biogeographical province, and part of the surrounding biogeographical provinces. We conducted an exhaustive search and constructed a database consisting in 9509 geo-referenced records showing that the bat fauna of the Chaco is richer than previously reported. We used an optimality criterion to identify patterns of distributional congruence determined by the distribution of bats. We recovered several areas that correspond with the Chaco. The areas that are coincident with the Chaco were supported by eight species: Eptesicus brasiliensis, E. diminutus, Eumops dabbenei, Histiotus velatus, Promops centralis, P. nasutus, Pygoderma bilabiatum, and Tonatia bidens. These results show that bat assemblages are useful as characterizing of regional patterns of distributional congruence. These patterns provide fi rst-step testable hypotheses of areas of endemism for future analyses of neighboring regions or analyses at more inclusive scales. Bat assemblages may be used to characterize the Chaco despite the fact that it was thought this habitat lacked an identity in relation to its bat fauna. The Chacoan nature as a biologically distinctive natural unit is clear now also in terms of regional patterns of distributional congruence of bat species, which provides further evidence for intensifying efforts to protect this endangered habitat.


INTRODUCTION
P r o t e c t e d a r e a s s y s t e m s s h o u l d b e implemented considering endemism, species richness, uniqueness of the ecosystem, and the degree of species extinction risk (Parker et al. 1993).Areas containing high concentrations of endemic taxa are important in conser vation irrespective of the overall richness of the area (Mares 1992).However, geographic patterns of species richness are greatly better known than geographic patterns of endemism (Kerr 1997).Endemism means restriction of the geographic range of a taxon to a defined area (Anderson 1994, Gaston 1994).An area of endemism (AE), in turn, is defi ned by the coincident restrictedness of two or more taxa to a geographic area (Platnick 1991).The geographic congruence of species distributional ranges has been interpreted as the result of allopatric speciation in biological communities isolated by the appearance of a biogeographical barrier (Espinosa-Organista & Llorente-Bousquets 1993, Crisci et al. 2000).Thus, AEs refl ects past, but also potentially future, speciation events (Fa & Funk 2007).Therefore, AEs are considered the fundamental units of analysis in evolutionary biogeography (Mor rone 2009).Besides, because of its geographical isolation, AEs are associated with the evolution of new adaptations in species and communities (Er win 1991, Vane-Wright et al. 1991, Spector 2002).
Several methods have been proposed over the last years to identify AEs (Morrone 1994, Linder 2001, Szumik et al. 2002, Szumik & Goloboff 2004, Deo & DeSalle 2006, Giokas & Sfenthourakis 2007, Dos Santos et al. 2008).The development of such formalized methods allow biogeographers, not only to evaluate if the traditionally proposed phyto and zoogeographical regions represent coherent units in terms of spatial overlap, but also to discover previously unrecognized regions.Purposely, Szumik et al. (2002) and Szumik & Golobof f (2004) formalized a method that explicitly evaluates the superposition between species geographic ranges by applying an optimality criterion (Szumik & Roig-Juñent 2005).
Being the ultimate goal the identifi cation of AEs, often regional analyses have been performed as fi rst approaches to the delineation of patter ns of distributional congr uence.
The amplitude of full distributional ranges of species of dif ferent taxa is highly variable.Accordingly, at regional scale, it is likely to find species with restricted ranges, which may determine AEs in strict sense (e.g., those included in the analysis of Díaz Gómez 2007), in coexistence with widespread species, whose ranges exceed that region, and which do not determine AEs in strict sense (e.g., some of those included in the analyses of Aagesen et al. 2009, Navar ro et al. 2009, Sandoval et al. 2010, Nori et al. 2011, Sandoval 2012, and Szumik et al. 2012).Mammals and par ticularly bats are good biogeographical indicators and they are important for defi ning faunal regions (Koopman 1976, 1981, 1982, Proches 2005).Bat distributional records in Argentina are reasonably well documented, thus allowing meaningful biogeographical analysis.Furthermore, Barquez et al. (1999) noted that the Argentinean Chaco has the richest bat fauna when compared with other biogeographical provinces in the country.But, regionally almost all bat species are widely distributed.Bat species have mostly ver y wide distributional ranges, which may include records in almost entire continents.For this reason, the performing of partial analyses may serve as a fi rst-step to establish biogeographical hypotheses.Never theless, although in the case of partial analysis it is possible to assess the degree of overlap between the portions included in the study area of distributional ranges of species, it is important to highlight that some of the areas resulting from the analysis constitute only regional patterns of distributional congruence and do not constitute AEs in strict sense.Yet, all of the obtained areas are equally interesting.While areas of endemism in strict sense provide strong basis for biogeographical regionalization, regional patterns of distributional congruence provide first-step testable hypotheses of areas of endemism for future analyses of neighboring regions or analyses at more inclusive scales (Szumik et al. 2012). As Aagesen et al. (2009) highlighted, we use the optimality criterion to analyze species distribution in a regional context while relaxing the criterion of endemism.Thus, as a result of our analysis, we expect to obtain characterizing species of certain areas (but not endemic to such areas).However, strictly, due to the characteristics of the distributional ranges of these species (exceeding the limits of the study area), such species should not be considered individually as indicators of the area.But in this paper we propose to evaluate the possibility of characterizing areas that were obtained from the analysis of portions of distributional ranges of species widely distributed, not by individual species, but the by species assemblages that are characterizing these areas.That is, the identity of certain areas in regional analyzes obtained may be determined by the association of cer tain species, which although widely distributed, are part of a unique assemblage in that part of its range.
The Chaco, as one of the most endangered habitats of the world (Bucher & Huszar 1999, Noss et al. 2002, Zak et al. 2004) has been identifi ed as an area of global importance for conservation (Silva & Guevara 2004).Over the last century human activities have transformed the Chaco into a dense and unproductive thorny scrubland (Bucher & Huszar 1999).The region has been par ticularly disturbed by severe logging and ranching, in such a way that in many areas, not only the forest has disappeared but also the herbaceous stratum, leaving only thorny bushes and cacti.The Chaco is still being impacted by uncontrolled forest exploitation (associated with timber harvesting, charcoal production, and tannin extraction), overgrazing by uncontrolled livestock, and over-exploitation of wildlife (Morello & Saravia Toledo 1959a, 1959b, Cabrera & Willink 1973, Bucher & Huszar 1999, Zak et al. 2004, The Nature Conservancy et al. 2005).The numbers are alarming: according to Zak et al. (2004), almost 85 % of the original Chacoan forest has been adversely disturbed.Additionally, the Chaco is currently being exposed to massively destructive activities such as extensive soybean crops.Conversely, only 20 % of the territor y is included in any kind of reser ve.The largest protected areas are in Bolivia (Kaa-Iya National Park) and Paraguay (Chaco Biosphere Reser ve).Argentina, in spite of being the country with the largest portion of Chacoan territor y, has protected only 9 % of this habitat (Pacheco et al. 1994, Yahnke et al. 1998, Andelman & Willig 2002, Nauman 2006).The situation of these dry forests is even more fl imsy because it occurs at the limits of their potential occurrence due to low precipitation (Hueck 1978).
Recently, in an attempt to recognize patterns of distributional congr uence in nor ther n Argentina (from 21º S to 32º S), Szumik et al. (2012) applied the optimality criterion to analyze records of more than 800 species of plants, insects, amphibians, reptiles, birds, and mammals.This analysis, the most complete made up to date, has allow the identifi cation of the Argentinean Chaco as a patter n of distributional congruence.But so far the Chaco as a whole has not been identifi ed as a regional pattern of distributional congruence through the use of a spatially explicit quantitative analysis.
In this paper we analyze bat distributional records from northern Argentina, Bolivia and Paraguay, in an attempt to identify the whole Chaco thor n-scr ub savanna as a regional pattern of distributional congruence, through the evaluation of the utility of bat assemblages as biogeographical indicators of the Chaco at a regional scale, and to assess the importance of the Chacoan bat fauna in terms of the conservation value of this endangered habitat.

Study area
The entire Chaco biogeographical province is included within the political limits of four countries: Paraguay, Bolivia, Brazil and Argentina (Cabrera & Willink 1973, Cabrera 1976).For the purposes of this analysis we included information from the whole territory of Bolivia and Paraguay, and from all northern Argentina provinces (Catamarca, Chaco, Córdoba, Corrientes, Entre Ríos, Formosa, Jujuy, La Rioja, Misiones, Salta, San Juan, Santa Fe, Santiago del Estero, and Tucumán) containing portions of Chaco (Fig. 1).Also other vegetational units, that surround the Chaco in these Argentinean provinces, and in Bolivia and Paraguay, were considered.
The Chaco is a very interesting ecosystem, biogeographically and evolutionarily, because it is one of the few places in the world where the transition between the tropics and the temperate zone is not a desert, but a transition of landscapes dominated by dry forests and woodlands (Daly & Mitchell 2000).
Briefl y, the Chaco lies east of the Andes in northern Argentina, southern Bolivia, and western Paraguay (plus a small portion in Mato Grosso do Sul, Brazil;Prado et al. 1992), ranging from tropical (17° S) to subtropical environments (33° S).With a latitudinal extension of more than 1500 km, an average longitudinal extension of 750 km, and an area of over 1000000 km 2 , the Chaco is the second largest natural biome in South America, exceeded only by the Amazon region.It is an open, xerophytic forest over a vast alluvial plain that rises gradually toward the west, with most elevations below 500 m, but reaching 1800 m in the western mountains.The climate is strongly continental.To the west, the rainfall is mainly concentrated in summer (about 350 mm per year, with more on the mountains), whereas to the east, it rains throughout the year (approximately 1200 mm per year).Temperatures decrease from north to south, and east to west.The dominant trees are species of Schinopsis Engl.("quebrachos colorados"), Prosopis L. ("algarrobos"), as well as Aspidosperma quebrachoblanco Schltr.("quebracho blanco"), Ziziphus mistol Griseb.("mistol"), and others.Some cacti as Opuntia quimilo K. Schum.and Stetsonia coryne (Salm-Dick) Britton & Rose are very common, as are the "palo santo" (Bulnesia sarmientoi Lorentz ex Griseb.) and the palm trees Copernicia alba Morong and Trithrinax campestris (Burmeist.)Drude & Griseb.
The Chaco can be divided into longitudinal and latitudinal sectors.According to Cabrera &Willink (1973) andHueck (1978), there are four such sectors: an eastern district or 'Chaco Oriental' (humid and with a predominance of Schinopsis balansae Engl.), a western district or 'Chaco Occidental' (dry and dominated by Schinopsis lorentzii (Griseb.)Engl.), a mountainous district or 'Chaco Serrano' (on the mountains of the western edge of the vegetal formation, Sierras Subandinas and Sierras Pampeanas, where Schinopsis haenkeana Engl. is predominant) and a southern district or 'Chaco Austral' (characterized by savanna grasslands and the absence of trees).Morello & Adámoli (1968) and Bucher (1982) proposed a different scheme, used by Barquez & Ojeda (1992) in their analysis of the Chacoan bats, with three longitudinal sectors, the Eastern, the Central and the Western Chaco.Later, when defi ning the ecoregions, Burkart et al. (1999) divided the Chaco into only two sectors: the Humid (eastern) and the Dry (western) Chaco.We used these last two schemes to discuss our results.Latitudinally, the Chaco can be divided into a northern sector, or 'Chaco Boreal', and a southern sector or 'Chaco Austral' (Myers & Wetzel 1983, Barquez & Ojeda 1992).We also considered these sectors in our discussion. Most

Data source
The taxonomy used herein is in accordance to Barquez (2006) for Argentinean bats and to Gardner (2007) for  other bat species.We used information from records of occurrence for 131 species of bats (Appendix), this is, all bat species reported to date for Paraguay, Bolivia, and northern Argentina.The main sources of data were Anderson (1997), López-González (1998), and Barquez et al. (1999), but the distributional information has been updated taking account of other publications (Emmons 1997, López-González et al. 1998, 2001, Willig et al. 2000, Ten et al. 2001, Aguirre 2002, Siles et al. 2003, Gorresen & Willig 2004, Stevens et al. 2004, Vargas-Espinoza et al. 2004, 2008, Azurduy & Emmons 2005, Acosta & Venegas 2006, Dick et al. 2007, Flores-Saldaña 2008, Barquez et al. 2009, 2011, Díaz & Barquez 2009, Sandoval et al. 2010).Some species recently added (Acosta & Aguanta 2005, Azurduy & Emmons 2005, Emmons et al. 2006) or described from Bolivia (Pacheco et al. 2004, Solari & Baker 2006) were also considered in our analyses although they were cited from only one point of occurrence in that country, and as far as we know they are absent in Argentina and Paraguay (Gardner 2007).In total, our database consisted in 9509 geo-referenced records (Fig. 2).We agree with Tabeni et al. (2004) in the use of geo-referenced records to optimize the quality of the analysis, contrary to the use of distributional maps for the species, which tend to overestimate species diversity in the areas under study (but see Soberón et al. 2000).

Distributional analysis
The method used to identify the regional patterns of distributional congruence, determined by the distribution of bats in northern Argentina, Bolivia and Paraguay, is that proposed by Szumik et al. (2002) and Szumik & Goloboff (2004).This method implements an optimality criterion that explicitly considers the spatial position of the species in the study region.The study region is divided into cells and the groups of cells (= areas) are evaluated by means of an index which is calculated for each species and depends on the adjustment of the distribution of the species to an area.The obtained values are combined for all the species that contribute to the area to obtain the value of endemicity (score) of the area.Only the areas with the maximum endemism values (scores) are retained.This optimality criterion is implemented in the computer program NDM and its viewer VNDM (Goloboff 2005, available at http:// www.zmuc.dk/public/phylogeny).
We analyzed the matrix of georeferenced data using grids with cells of different sizes (0.5, 0.75, 1, 1.25, 1.5, 1.75, and 2 degrees of latitude-longitude), and considering different values of fi ller for assumed and inferred presences (0 and 0, 20 and 40, 40 and 60, 60 and 80, 80 and  Search parameters (see Materials and Methods) and patterns of distributional congruence obtained that correspond with the Chaco biogeographical province (and that are discussed here).Abbreviations are as follows: CS: cell size (the side size in latitude-longitude degrees is given); A-I: values for assumed and inferred presences.Parámetros de búsqueda (ver Materiales y Métodos) y áreas de endemismo obtenidas que corresponden con la provincia biogeográfi ca Chaqueña de manera más o menos precisa (y que son discutidas en el presente trabajo).Las abreviaturas son las siguientes: CS: tamaño de celda (se presenta el tamaño de lado en grados de latitud-longitud); A-I: valores para presencias asumidas e inferidas.

Search parameters
Chacoan regional patterns of distributional congruence We carried out the endemicity analysis by means of a heuristic search and default parameters: founding groups of cells by adding/eliminating one cell at a time, saving groups defi ned by two or more endemic species, and with scores higher or equal to 2.0.Groups of cells with more than the 50 % of species in common were ruled out, obviously retaining those with highest score, and groups of cells with a score up to 1 % inferior were stored in memory.We obtained consensus areas using 50 % of similarity in species against any of the other areas.The results were mapped using the Global Mapper v11.02 program.

RESULTS
We were able to obtain patterns of distributional congr uence with all the grid sizes used.With cells of 0.5º x 0.5º, 0.75° x 0.75°, 1° x 1° and 1.25° x 1.25°, we obtained only local patterns of distributional congruence, limited to Bolivian territory, which will not be presented or discussed here.With other cell sizes, we obtained remarkable results, i.e., regional patterns of distributional congruence (some of them presented in Table 1).So, for this spatial scale all grids with cells larger than 1.5º latitude-longitude are useful to obtain regional patterns of distributional congruence.Several of the obtained patterns are ver y interesting because they constitute areas that are congruent with conventional vegetational units, which were not defi ned based either in a numerical or explicit criterion, or in their bat fauna.
We found areas that correspond to the Chaco with grids with cell sizes of 1.5°, 1.75º, and 2º latitude-longitude (Table 1).Areas corresponding to more humid forests were also recovered with grids with cell sizes of 1.5º, 1.75º, and 2º latitude-longitude, but these will be discussed elsewhere.

Chacoan patterns of distributional congruence
The areas coincident with the Chaco (Fig. 3) were recovered ten times (Table 1) and are suppor ted by eight species, Eptesicus brasiliensis (Desmar est), E. diminutus Osgood, Eumops dabbenei Thomas, Histiotus velatus (I.Geof froy St.-Hilaire), Promops centralis Thomas, P. nasutus (Spix), Pygoderma bilabiatum (Wagner), and Tonatia bidens (Spix).Eptesicus brasiliensis is an almost exclusively Chacoan species in Argentina and Paraguay, where there are few records outside the Chaco (Fig. 4); until now, there are no records of E. brasiliensis in Bolivia (Davis & Gardner 2007).Eptesicus diminutus, E. dabbenei, and P. nasutus have most of their occurrence points in Argentina in the Chaco, but they were also recorded in other vegetational units; however, most of their non Chacoan occurrence points are peripheral to the Chaco (Fig. 4).In Paraguay, most records of E. diminutus and P. nasutus, and all records of E. dabbenei, are from the Chaco (Fig. 4).Neither E. diminutus nor E. dabbenei were recorded in Bolivia (Davis & Gadner 2007, Eger 2007), although the Argentinean records of these species are very close to the political boundaries between the two countries (Fig. 4).Promops nasutus has most of their occurrence points in Bolivia in the Chaco, but it was also rarely recorded in other vegetational units; however, these non Chacoan records are peripheral to the Chaco (Fig. 4).In Paraguay, Bolivia, and Argentina, H. velatus was barely recorded and has some Chacoan records, but it was also recorded in forested natural units others than the Chaco, although Mapa de la región de estudio en el que se esquematiza uno de los diez patrones de congruencia distribucional obtenidos que son coincidentes con la provincia biogeográfi ca del Chaco (tamaño de celda: 1° latitud-longitud).
almost all these records are peripheral to the Chaco (Fig. 5).Promops centralis was recorded mainly in Chacoan localities, mostly in the Paraguayan Humid and Dry Chaco and less in the Bolivian Dry and Argentinean Humid Chaco (Fig. 5).Pygoderma bilabiatum was recorded mostly in the Paraguayan Humid Chaco, and in the adjacent Argentinean Paranean forests, and less in the Bolivian Dr y Chaco and the adjacent Bolivian and Argentinean Yungas forests; apparently, its presence in the Chaco is only marginal: all the points of occurrence in forested habitats others than the Chaco (Paranean and Yungas forests) are marginal to the Chaco (Fig. 5).In Argentina, T. bidens was recorded in Chacoan localities, but also in Yungas and Paranean habitats, been all the non Chacoan records very close to the Chaco.In Paraguay, this species was recorded almost exclusively in Chacoan localities.However, in Bolivia all the known records are from the north of the country (Fig. 5).
Histiotus velatus contributes to only one of the ten equivalent recovered areas coincident with the Chaco (with an individual score of 0.662; Table 1).Eptesicus diminutus, E. dabbenei and Pygoderma bilabiatum contributes between 0.332 (individual score of P bilabiatum) and 0.82 (individual scores of E. diminutus and E. dabbenei) to the score of four of the recovered areas (Table 1).Promops centralis and P. nasutus contribute between 0.624 (P.nasutus) Mapa de la región de estudio en el que se indican los registros de ocurrencia conocidos de cuatro de las ocho especies que caracterizan los patrones de congruencia distribucional chaqueños: E. brasiliensis, E. dabbenei, E. diminutus y P. nasutus.Los rangos distribucionales de estas cuatro especies en la región de estudio presentan un ajuste notable con la provincia biogeográfi ca del Chaco.and 0.85 (P.centralis) to the score of fi ve of the recovered areas (Table 1).Finally, E. brasiliensis and T. bidens contribute between 0.622 (individual score of E. brasiliensis) and 0.831 (individual score of T. bidens) to the score of six of the recovered areas (Table 1).

Species richness
In order to build up the database that was later analyzed using the optimality criterion, it was necessar y not only to review specimens deposited in museums, but also to see previously published work.As a result, information that was outdated and/or scattered was updated and compiled, contributing to the current knowledge of species richness in the whole Chaco.

Portions of the Chaco
America (Gardner 2007).The Argentinean Chaco contains 44 species of bats, making this the most diverse region of Argentina with regard to bats (Barquez et al. 1999, Barquez 2006).Twenty four of these species were also recorded in the Bolivian Chaco (not considering eight of probable presence) and 37 in the Paraguayan Chaco (Anderson 1997, López-González 1998).
In their study, Barquez & Ojeda (1992) established that the Western and Eastern Chaco holds the largest number of species of the whole Chaco in Argentina.They have pointed out that 31 species are present in each portion, whereas the Central Chaco contains only 14 (Barquez & Ojeda 1992).They assumed that differences in richness are likely related to the contact of the Chaco with the humid and transitional habitats that occur adjacent to the Western and Eastern sectors and to the contributions of these habitats to the bats assemblages of those portions.(Barquez & Ojeda 1992).Today, with the new subdivision of the Chaco into two ecoregions, the Dry Chaco (roughly equivalent to the Western Chaco), holds 35 species, whereas the Humid Chaco (roughly equivalent to the Eastern Chaco), holds 29 species (Table 3).Species B&O 1992 B 2006 M&W 1983 B&O 1992 L-G 1998 B&O 1992 A 1997  Bat species recorded in the sectors of the Chaco as ecoregion according with updated records.X indicates that reliable records of the species exist; X* indicates that the author has mentioned that unconfi rmed data make the species presence probable; P indicates that the species would be present according with the area considered by the author, but there are no punctual records confi rming its presence.

Distributional analysis
Especies de murciélagos registradas en los sectores del Chaco como eco-región de acuerdo a registros actualizados.X indica que existen registros fehacientes de la especie; X* indica que el autor ha mencionado que datos no confi rmados hacen la presencia de la especie probable; P indica que la especie estaría presente de acuerdo al área considerada por el autor, pero no hay registros puntuales que confi rmen su presencia.

Sectors of the Chaco Dry Chaco Humid Chaco
Species Argentina Paraguay Bolivia Argentina Paraguay patterns.He sustained that the Argentinean bat fauna could be almost all included within the zoogeographic "Patagonian subregion", and just a little part would be included in the Eastern Brazilian Highlands and Coast subregions, in northeastern Argentina.Barquez et al. (1999) considered that Koopman's division is no longer plausible, and pointed out that the Chacoan element is pronounced and forms a great faunal ecotone between the subtropical forests, being distinctive from the Patagonian subregion of Koopman (Barquez et al. 1999).According to their analysis, however, at that time Barquez et al. (1999) maintained the idea that bats are not of great utility to delineate faunal regions.They considered that the Argentinean Chaco cannot be considered as a faunal region for bats, and treated it as an ecotonal, southern, semiarid extension of the Yungas rainforests to the west and the Paranean forests to the east (Barquez et al. 1999).
G a l l a r d o ( 1 9 7 9 ) d e m o n s t r a t e d t h a t amphibian and snakes show a high level of endemism in the Chaco.For birds, as for individual species of bats, it seems that there is a low level of endemicity (Straube & Di Giacomo 2007).The species of birds that inhabit the Chaco tend to be widely distributed over South America (Short 1975), as well as the bats from the Chaco.Most species of bats from the Chaco biogeographical province have broad distributions in South America and occur in diverse habitats (Koopman 1982, Myers & Wetzel 1983).As noted by Myers & Wetzel (1983), this apparently common pattern of low endemicity rates between birds and bats may be the result of the increased vagility of volant vs. non volant ver tebrates and the absence of barriers (Straube & Di Giacomo 2007), or more precisely, the ineffi ciency of barriers that may isolate populations of species of other taxa less vagile.Several authors (Koopman 1982,  Myers & Wetzel 1983, Barquez & Ojeda 1992) have mentioned that the Chaco seems to have a lack of endemic species of bats.Barquez et al. (1999) pointed out that three species are restricted to the Chaco in Argentina: Eptesicus brasiliensis, Cynomops paranus, and Promops centralis, but later Barquez (2006) included the distribution of the former two to other regions in the countr y.Promops centralis remains as endemic, but only for the eastern portion of the Chaco in Argentina, and not for the entire region (Barquez et al. 1999) and is not exclusive of the Chaco in Paraguay or Bolivia (Anderson 1997, López-González 1998).Briefl y, as mentioned by Myers & Wetzel (1983), the unique composition of the Chacoan bat fauna seemed to result from the deletion of species from neighboring areas, rather than from the differentiation or inclusions of new elements.
However, according to our analysis, bat assemblages are good indicators of regional patterns of distributional congruence that are coincident with the Chaco biogeographical province.We have been able to identify the Chaco as a regional pattern of distributional congruence analyzing bat records.Although this biogeographical province seems not to have exclusive species, it does have a set of typical species allowing defi ning it as a natural unit.This faunal assemblage, although not as rich in species as its tropical rainforest counterparts, is distinctive in species composition and has its own identity.
The reasons for this apparent contradiction between exclusive and typical species may be those pointed out by Barquez & Ojeda (1992) and Willig et al. (2000).Although there are not exclusive species in the Chaco, Barquez & Ojeda (1992) found a group of species whose optimum abundance seems to correspond well with this region in Argentina.According to them, the central portion of the Argentinean Chaco reflects the optimum occurrence for 11 species, which can be rated as common or abundant.We have identifi ed regional patterns of distributional congruence that are coincident with the Chaco and these areas are supported by eight species (Table 1).These species are not the same as listed by Barquez & Ojeda (1992), but, like those, all are widely distributed species.According to Barquez & Ojeda (1992), the important thing for species being considered as characteristic of the Chaco would be their relative abundance.The eight species that are indicators of Chacoan regional patterns have records outside the Chacoan habitat, but these are mostly marginal records at the limits of the main distributional area resulting in high values of individual endemicity scores.Thus, we could refer to this group of species as "Chacoan species", despite its presence in surrounding areas.In turn, Willig et al. (2000) found that strong dif ferences exist in the Paraguayan chiropteran fauna between sites East (mesic) and West (xeric) of the Rio Paraguay.They argued that these differences are related to dominance by molossids and vespertilionids in dry regions, versus phyllostomids in mesic regions.Three of our eight Chacoan indicator species are molossids (Eumops dabbenei, Promops centralis, and P. nasutus), three are vesper tilionids (Eptesicus brasiliensis, E. diminutus, and Histiotus velatus), and there are only two phyllostomids (Pygoderma bilabiatum and Tonatia bidens).Of our eight indicator species, molossids plus vespertilionids represent 75 % of the species.
Although in our study region the identifi ed characterizing species have a distributional area more or less restricted to the Chacoan biographic province, in a larger spatial context this is not so.Because distributional ranges of most bat species included in this analysis continue outside the study region (i.e., extending over other American countries), it seems reasonable to criticize the present study arguing that the study region is inadequate or not natural, or that the selected taxon (bats) is not appropriate for this kind of analysis.However, as stated by Szumik et al. (2012), this would be equivalent to criticize phylogenetic analyses for dealing with possibly incomplete monophyletic groups.Our analysis aims to hypothesize relationships between cells in the study area and "nothing is stated or implied about cells that would occupy an extended grid" (Szumik et al. 2012, p. 319) Besides, although all the eight species that characterize areas tightly related to the Chaco (Eptesicus brasiliensis, E. diminutus, Eumops dabbenei, Histiotus velatus, Promops centralis, P. nasutus, Pygoderma bilabiatum, and Tonatia bidens), have ranges that exceed to lesser or greater extent the Chacoan biogeographical province when the entire South American continent is considered the complete assemblage of species is characteristic only of areas closely related to the Chaco.Undoubtedly, inclusion of other biomes will led to obtain other results, but will no necessarily affect the ability to delineate the Chaco as a pattern of distributional congruence or most probably as part of a composite area of endemism (e.g., the recently proposed new biogeographical province known as "Seasonally Dr y Tropical Forests" of South America, which currently would include the Eastern Chaco, among other forest formations; Prado 2000).Thus, a priori, we think that the results are robust despite the selection of outlier areas.Nevertheless, it would be very interesting to include all biomes in South America and to formally test the robustness of these results.
Therefore, it is impor tant to note that the species that determine the patterns of distributional congruence identified through the analysis are characterizing species and not necessarily endemic species of those areas.A characterizing species may or may not be endemic to an area.Therefore, although there are not bat species endemic to the Chaco, it is possible to consider some bat species assemblages as characterizing of the Chaco.An endemic species means that the species is native to a region and occurs there and only there.A species that is endemic to an area is found nowhere else.According to this, it is well known that the Chaco lacks endemic bat species.On the other hand, the term "characterizing species" is not equivalent in the sense that it is not expected that these species to be endemic to the area they defi ne.The method chosen to test the hypothesis of bat assemblages as indicators of regional patter ns of distributional congr uence was developed within the context of areas of endemism.We used this method because it allows assessing, through an optimality criterion, the degree of overlap between the distributional areas of species, but it is important to note that the areas resulting from the analysis constitute only regional patterns of distributional congruence and not areas of endemism in strict sense, but possibly patches of a bigger area of endemism.Only those areas determined by the distributional areas of species of restricted ranges would constitute areas of endemism in strict sense.The NDM/VNDM program, actually widely used for biogeographical regionalization (e.g., Carine et al. 2009, Casagranda et al. 2009), has consistent theoretical basis and is well grounded in a valid and mostly accepted concept of areas of endemism, and can be used not only to identify areas of endemism in strict sense but also to identify regional patterns of distributional congruence determined by widespread species whose ranges exceed the study region (Aagesen et al. 2009).

Conservation priorities
Conservation assessment is a rapidly evolving discipline whose goal is the design of networks of protected areas that represent and ensure the persistence of nature (i.e., species, habitats, and environmental processes) by separating priority areas from the activities that degrade or destroy them (Knight et al. 2008).The number of species at risk of extinction continues to increase ever y year, and the extinction rates have increased to mass extinction proportions (Hughes et al. 1997, May & Tregonning 1998, Chapin et al. 2000, Kerr & Deguise 2004).
Species extinctions in tropical South America in general, and in the Amazon rainforest in particular, have attracted almost all the attention and efforts of the conservationists (Mann 1991).However, Redford et al. (1990), Mares (1992), andWillig et al. (2000) have suggested that others areas than tropical rainforests deser ve the same allocation of time and resources.In fact, as Mares (1992) has shown, for mammals the drylands of South America suppor t more endemic species, genera, and families than the Amazon rainforest.So, conser vation of mammalian diversity requires increasing the number of protected areas in the extensive dr ylands of South America.Unfor tunately, there is an inadequate number of protected areas in Northwestern Argentina.Ojeda et al. (2003) noted that while the Puna and Yungas biomes are rather well protected, the arid and semiarid Monte and Chaco are not.Besides, Ojeda et al. (2003) pointed out that large areas in the Chaco have never been sampled.Therefore, additional faunal surverys are badly needed in this biome, which is under strong pressure by humans (Ojeda & Mares 1984, Ojeda 1999, Ojeda et al. 2002, 2003).
Conser vation strategies that consider b i o g e o g r a p h i c a l u n i t s a t t h e s c a l e o f ecoregions are ideal for protecting a full range of representative areas, conser ving special elements, and ensuring the persistence of populations and ecological processes, par ticularly those that require the largest areas or are most sensitive to anthropogenic alterations (Olson et al. 2001).Although c o n s e r v a t i o n e f f o r t s s h o u l d f o c u s o n maximizing the protection of as many species as possible (Andelman & Willig 2002), the conser vation of adaptive variation of species may be as impor tant as the conser vation of par ticular species (Araújo 2002).Thus, protection should not only maximize the number of protected species but should maximize variation in the attributes of species as well (Stevens et al. 2004).In this sense, even though bat species that determine the patterns of distributional congruence that are equivalent to the Chaco are mostly widespread species, it is important to note that identifi ed species assemblages could be and should even be considered a conser vation priority.The identification of species assemblages that characterize those patterns is important in a regional context and to protect these assemblages would imply the protection of particular attributes associated with particular ecological and evolutionar y processes in which bats are involved.At regional level, Paraguay, Bolivia, and Argentina should ensure the protection of representative areas of all different landscapes, even those that apparently do not have their own specifi c identity, and apparently are nothing more than impoverished versions of neighboring habitats, as the Chaco was regarded in relation to its chiropteran fauna.
H e r e , w e w a n t t o e m p h a s i z e t h e conser vation value of the Chaco in terms of patter ns of distributional congr uence of its bats, and to highlight its nature of being a biologically defi ned unit with its own assemblages of species and ecological and evolutionar y processes that it is necessar y to protect.Studies like this one are useful in defining priority conser vation areas, which may consider not only endemism in their determination and their defi nition (Myers 1988, 1990, Bibby 1994, Baquero & Tellería 2001, Kerr & Burkey 2002, Silva & Guevara 2004; but see Dinerstein & Wikramanayake 1993, Pressey et al. 1993, Ceballos & Brown 1995).
of the Chacoan territory lies within Argentina, where the Chaco thorn scrub savanna extends from the border with Paraguay and Bolivia southward to San Luis province.It occupies an extension of about 675000 km 2 , representing about 60 % of the total area of the Chaco and 25 % of the territory of Argentina.The Paraguayan and Bolivian portions, together, are slightly larger than half of the extension of the Argentinean Chaco.In Paraguay it includes approximately 260000 km 2 (~ 25 % of the total area of the Chaco and 60 % of the area of the country), and 153000 km 2 in Bolivia (representing 15 % of the total area of the Chaco, as well as of the country) (Nauman 2006).

Fig. 1 :
Fig. 1: Map of the study area showing the Chaco biogeographical province and its two sectors delineated in the Burkart et al. (1999)'s ecoregions scheme (see text).Mapa de la región de estudio en el que se indica la provincia biogeográfi ca del Chaco y sus dos sectores delineados en el esquema de eco-regiones deBurkart et al. (1999; ver texto).

Fig. 2 :
Fig. 2: Map of the study area, which includes Bolivia, Paraguay and northern Argentina, showing the distribution of the geo-referenced records of the 131 bat species that constituted the analyzed database.

Fig. 3 :
Fig. 3: Map of the study area schematizing one of the ten obtained patterns of distributional congruence that are coincident with the Chaco biogeopraphical province (cell size: 1° latitude-longitude).

Fig. 4 :
Fig. 4: Map of the study area showing the known occurrence records of four species of the eight that characterize the Chacoan pattern of distributional congruence: E. brasiliensis, E. dabbenei, E. diminutus, and P. nasutus.Distributional ranges of these four species in the study area have a remarkable fi t with the Chacoan biogeographical province.

Fig. 5 :
Fig. 5: Map of the study area showing the known occurrence records of four species of the eight that characterize the Chacoan pattern of distributional congruence: T. bidens, P. bilabiatum, P. centralis, and H. velatus.Distributional ranges of these four species in the study area have a number of records outside the Chacoan biogeographical province.
K o o p m a n ( 1 9 8 2 ) d e l i m i t e d n i n e biogeographical areas for South America based on distribution of bats and vegetation