The average an-The snake assemblage ( Squamata : Serpentes ) of a Cerrado-Caatinga transition area in Castelo do Piauí , state of Piauí , Brazil

This study records and analyzes the diversity and structure of a snake assemblage in a transition area between Cerrado and Caatinga, in the municipality of Castelo do Piauí, state of Piauí, comparing the distribution and similarity of the species composition with other open localities already studied in Brazil. We used three complementary sampling methods: time constrained search (TCS), pitfall traps with drift fences (PFT), and incidental encounters (IE). During the TCS and PFT, 912 hours/observer and 6,468 days/trap were used, respectively. We estimated 23 species of snakes for the locality, although only 19 species were recorded. Philodryas nattereri Steindachner, 1870 (n = 10), Liophis poecilogyrus (Schlegel, 1837) (n = 9), Liophis viridis Günther, 1862 (n = 8) and Thamnodynastes sp. (n = 8) were the most abundant species. Terrestrial, cryptozoic, and diurnal snakes predominated in the assemblage (Boidae = 2 species, Dipsadidae = 12, Colubridae = 2, Elapidae = 1, Viperidae = 2). The results indicate that the fauna of the locality is similar with that of other open formations, especially the Caatinga, corroborating previous floristic studies. Comparisons between snake assemblages analyzed by different authors suggest structural differences between the assemblages of the Cerrado and the Caatinga, contradicting the hypothesis of mixed composition of fauna in these biomes.

The high diversity of reptiles in Brazil is associated with the presence of different biomes, resulting in a gradient of regions with open and xeric formations (such as the Caatinga) to regions with extremely humid forest formations (like the Amazon Forest) (RODRIGUES 2003(RODRIGUES , 2005)).Despite the great diversity of snakes (371 species) recorded for Brazil (BÉRNILS 2010), information on distribution and natural history in this group is still unsatisfactory, especially for the Cerrado and Caatinga, biomes with the lowest known diversity (SAZIMA & HADDAD 1992, RODRIGUES 2005, SAWAYA et al. 2008).
The main landscapes in the state of Piauí, northeastern Brazil, are associated with Cerrado sensu lato (30.0% of the total area of the state; CEPRO 1992, CASTRO et al. 1998, FARIAS & CASTRO 2004) and Caatinga (28.4%), as well as transitional areas or ecotones between these two biomes (CASTRO et al. 1998, CASTRO 2003).Few herpetological studies have been conducted in the state in localities of Caatinga (VANZOLINI et al. 1980, ARAUJO et al. 1998), Cerrado (COIMBRA FILHO & MAIA 1979), or in transitional areas between these two biomes (VANZOLINI et al. 1980, ROCHA & SANTOS 2004, ROCHA & PRUDENTE 2010).Herpetological studies in these and in other localities in northeastern Brazil are mostly restricted to species lists, descriptions of new taxa, and basic information on their biology, with scant informa-tion on factors structuring assemblages, as well as on their structure.
Nevertheless, recently ROCHA & PRUDENTE (2010) presented data on the composition and aspects of the natural history of the snake assemblage in a transition area between Cerrado and Caatinga in the state of Piauí.
Herein we present the results of an inventory of a snake assemblage from the Municipality of Castelo do Piauí, another transitional area between Cerrado and Caatinga in the state of Piauí.We provide information on composition, estimates of species richness and abundance, and a comparison with snake assemblages from others localities localized on open areas already studied in Brazil.

MATERIAL AND METHODS
The study area of about 3,000 ha belongs to ECB Rochas Ornamentais do Brasil and is located in the Municipality of Castelo do Piauí, northeast of the state of Piauí, Brazil (05°19'20"S, 41°33'09"W, Fig. 1).The vegetation is typical of a transition area between the Caatinga and Cerrado, with a predominance of a low altitude Cerrado Rupestre (rocky grasslands) (CASTRO & COSTA 2007, FARIAS & CASTRO 2004).The average an- Fieldwork was conducted in three physiognomically distinct areas, with the following characteristics: Area 1 -low altitude Cerrado Rupestre (rocky grasslands), characterized by superficially rocky soil, with rocky outcrops and with poorly developed plants dispersed between rocky outcrops and between trenches (CASTRO & COSTA 2007) (Figs 2 and 3); Area 2 -Campo Cerrado (open Cerrado), characterized by a predominantly herbaceous-shrub component (rather than a shrub-arboreal) and fields that remain flooded during the rainy season (Figs 4 and 5); and Area 3 -secondary forest of typical Cerrado, composed of mainly shrubs and small trees, with a predominance of secondary vegetation and absence of herbaceous com-ponents.In this last area, pioneer species such as Mimosa sp. and other Mimosaceae have replaced the native vegetation, resulting in transitional and altered vegetation (Figs 6 and 7).
Seven sets of traps were installed in each of the three areas.Each set contained four 60 L buckets, arranged in a "Y" configuration, 10 m apart from one another and connected by 100 cm high drift-fences, totaling 84 buckets and 630 m of driftfences.In each of the three areas, the traps remained open during 77 days, spread over four expeditions, two in the dry season and two in the rainy season.This effort resulted in 2,156 days/ trap in each area, totalizing 6,468 days/trap performed.
Time constrained search (TCS) consisted of walking slowly along pre-existing tracks within the areas, searching actively (visually) in burrows, termite nests, shrubs, grasses tussocks, etc.This procedure was performed for 38 days in each of the three areas.The search lasted four hours per day (two hours in the morning and two hours at night), and was carried out by two observers.Thus, each area had the same sampling effort of 304 hours/observer, totaling 912 hours/observer for the three areas.
We considered an encounter incidental (IE) when snakes were found alive or dead in the study area and surrounding region during other activities in which the collection time was not being computed, as well as instances when snakes were found and collected by workers and local residents.
The specimens collected were deposited in the Herpetological Collection of the Museu Paraense Emílio Goeldi (MPEG), Municipality of Belém, state of Pará, Brazil (Appendix).
For the species accumulation curves, rarefaction methods and richness estimates, we used the program "EstimateS" version 7.5 (COLWELL 2005a) from the samples obtained by the methods TCS and PFT.For TCS, a sample corresponded to 8 man-hours (114 samples) and for PFT it corresponded to a collection day for seven sets of traps per area (231 samples).To estimate richness we used the first order Jackknife estimator (Jack 1) (COLWELL & CODDINGTON 1994, COLWELL 2005b).For the construction of the accumulation curve and estimation of species richness, the number of species was associated with the number of individuals, using the obtained samples.We performed a hierarchical relationship between relative abundance and dominance ratios between species.For classification of the macrohabitats and daily activity we used the categories proposed by CADLE & GREENE (1993), with modifications by STRÜSSMANN (2000).
To evaluate the similarity between the composition of the snake assemblage of Castelo do Piauí and those of other open and transitional areas of Brazil, we used data from 10 localities (Fig. 1), divided into three groups as follows: Cerrado-Caatinga   Nacional Serra da Capivara (Capivara), state of Piauí (ARAUJO et al. 1998).Only localities with minimum sampling efforts of at least 80 collection days and/or 200 field hours in total of TCS were used for this comparison.We performed a Principal Coordinates Analysis (PCoA) for the species recorded in these localities (PIELOU 1969, MANLY 1994, KOVACH 1999), and a cluster analysis using the weighted pair-group method of arithmetic averages (WPGMA, indicated for samples of groups with different richness or differences in sampling, according to KOVACH 1999).Both analyses were run by using Gower's general similarity coefficient (SNEATH & SOKAL 1983, KOVACH 1999).Analyses were performed using the program MVSP 3.1 (KOVACH 1999).The distribution map was produced by using ArcView GIS version 3.3.

Habitat
The first two axes of the Principal Coordinates Analysis (PCoA) together explained 36.6% of the data variance (Axis 1: eigen value = 1.573 and 21.9% of the variance; axis 2: eigen value = 1.05 and 14.6% of the variance).Axis 1 ordered the assemblages into two main groups: group 1 formed by assemblages from the Caatinga (Exú, Apodi, Capivara and Dunas SF), ecotones (Castelo, JFreitas and PNSC), and northeastern Cerrado (USantos); group 2 was formed by assemblages from Midwestern (Manso) and Midwestern Cerrado (Emas and Itirapina) (Fig. 10).Axis 2 of the PCoA ordered the assemblages into two other groups: group 1 was formed by the assemblages from the Cerrado (Manso and USantos) and from two transition areas (PNSC and J Freitas); group 2 was composed by the assemblages from other open formations of Northeastern Brazil, including Castelo (Fig. 10).Similarly to the PCoA results, the cluster analysis showed two groups according to vegetation characteristics.The group 1 was formed by the Cerrado assemblages (Manso, Emas and Itirapina) and the northeastern Cerrado (USantos).The group 2 was formed by the Dunas SF and a subgroup composed by assemblages from the Caatinga (Capivara, Exu and Apodi), and Cerrado-Caatinga transition areas (PNSC, Castelo and JFreitas) (Fig. 11).PNSC (53.6%) and J Freitas (37.0%) are the most similar assemblages to Castelo (Tab.II).

DISCUSSION
The low diversity of snakes observed in the assemblages studied in Castelo and in other two Cerrado-Caatinga transition localities (PNSC, JFreitas) is probably a pattern of the transitional formations, since localities with a predominance of open vegetation types tend to have relatively lower diversity when compared to forested formations (VANZOLINI 1974, 1976, VITT & VANGILDER 1983, SILVA & SITES 1995, STRÜSSMANN 2000, BERNARDE & ABE 2006, SAWAYA et al. 2008).Additional studies are necessary to elucidate this hypothesis.
The combination of different capture methods is the best way to sample an area since it makes it possible to access the faunas from distinct habitats (GREENBERG et al. 1994, SAWAYA et al. 2008, RIBEIRO-JÚNIOR et al. 2008).Nevertheless, in spite of the fact that three capture methods were employed in our study area, both rarefaction and species accumulation curves did not stabilize, indicating the possibility of additional records for Castelo do Piauí in further studies.
The snake assemblage from Castelo is characterized by the prevalent abundance of specimens of Dipsadidae (representing 83.1% of the total number of records) and by the low abundance of Boidae, Colubridae, Elapidae and Viperidae specimens.The predominance of terrestrial, diurnal snakes was also observed in other communities from open areas or ecotones, such as PNSC (ROCHA & PRUDENTE 2010), Exu (VITT & VANGILDER 1983), Capivara (ARAÚJO et al. 1998) and JFreitas (ROCHA & SANTOS 2004).Although not tested, the few records of arboreal snakes are tentatively attributed to the low height and density of the vegetation, and the absence of aquatic species in the studied area can be attributed to the lack of water bodies.Eight of the 19 species (42.1%) recorded for the locality present wide geographic distributions, occurring in different biomes (such as the Cerrado, Caatinga, Pantanal, Amazon and Atlantic forests), while other seven species (36.8%) occur exclusively in Cerrado and Caatinga.Additionally, two species are found exclusively in the Caatinga -Epicrates assisi Machado, 1944 andL. viridis, andtwo in the Cerrado -Liophis paucidens (Hoge, 1953) and Bothropoides lutzi (Miranda-Ribeiro, 1915).Although B. lutzi can occur at the limits of the Caatinga, the species is considered typical of the Cerrado, according to CAMPBELL & LAMAR (2004).The presence of shared species with Cerrado and Caatinga, associated with the occurrence of transitional vegetation species (CASTRO & COSTA 2007), indicate that the locality of Castelo do Piauí represents an ecotonal area, influenced by both types of biomes.
The similarity between assemblages from the Cerrado-Caatinga transition localities (Castelo, JFreitas and PNSC), the northeastern Cerrado (USantos), and the Caatinga (Apodi, Exu and Capivara) indicate that the geographical proximity of adjacent biomes have an influence on the composition of the fauna from Castelo.This fact can be corroborated by the presence of species in common with Cerrado and Caatinga assemblages, besides the presence of species with wide geographic distribution.However, species richness, faunal composition, and number of shared species in the above mentioned localities, as well as the formation of two distinct groups observed in our analysis -Cerrado and Caatinga -provide evidence that they have their own faunal identities, as previously argued by authors such as COLLI et al. (2002), RODRIGUES (2003), andSAWAYA et al. (2008).
Figures 2-7.Sampling areas.Area 1: low altitude Cerrado Rupestre, (2) detail of the vegetation and (3) detail of the stony soil typical of the area; Area 2: open Cerrado, (4) detail of the vegetation and (5) detail of the soil; Area 3: secondary forest of typical Cerrado, (6) detail of the vegetation and soil during the rainy season, and (7) detail of the vegetation and soil during the dry season.