First avifaunal survey of a Cerrado dry forest enclave on the right bank of the São Francisco River , Minas Gerais , Brazil , with insights on geographic variation of some species

Cerrado dry forest enclaves have been treated as an endemic bird area. In the last years, some avifaunal surveys have been conducted in dry forests on the left bank of the São Francisco River, eastern Brazil. Nevertheless, there is a gap of detailed ornithological surveys in the Cerrado dry forest enclaves on the right bank of this river. Here, we present the first avifaunal survey of a Cerrado dry forest enclave on the right bank of the São Francisco River. The study area, named “Curral de Pedras”, comprises a dry forest enclave and other associated habitats in central Minas Gerais state, southeastern Brazil. We recorded 172 bird species; nine taxa have their ranges strongly associated to the Caatinga; six present their ranges associated to the Atlantic Forest and one is typical of the Cerrado. Important records are those of Campylopterus calcirupicola, a recently described species associated with dry forests of central Brazil, and Celeus ochraceus, collected in its southernmost range. Furthermore, we obtained specimens that present intermediate phenotypes between the following woodcreeper species: Lepidocolaptes squamatus / L. wagleri and Xiphocolaptes albicollis / X. falcirostris. Those records reinforce the need of further sampling and collecting activities in those dry forest enclaves aiming future researches in taxonomy, geographic variation, and phylogeography. Key-Words. Birds; Cerrado; Seasonally dry tropical forests; Geographic variation.

Small areas of SDTFs are also patchily distributed throughout the Cerrado (a biogeographic province predominantly covered by tropical savanna), usually associated to limestone outcrops and limestone derived soils (Rizzini, 1997).These "islands" were named "Cerrado dry forests" by Santos et al. (2012) and are very important from a biogeographical point of view, since they suggest historical connections among several SDTF nuclei (Werneck & Colli, 2006).
In the last years, a few avifaunal inventories have been conducted in arboreal caatingas (sensu Santos et al., 2012) on the left bank of the São Francisco River (Kirwan et al., 2001(Kirwan et al., , 2004;;Lopes et al., 2010;Dornelas et al., 2012).Nevertheless, there is a gap of detailed ornithological surveys in Cerrado dry forest enclaves on the right bank of this river.Thus, the aim of this paper is to present a bird inventory in a poorly known area located on the right bank of the São Francisco River, named "Curral de Pedras".This area was only briefly sampled by the staff of Museu Nacional, in late November 1995, based on records of K. franciscanus reported in the literature (Lima, 1999;Kirwan et al., 2004).

Study area
Curral de Pedras is a hilly site with limestone outcrops at higher elevations, located in the municipality of Jequitaí, central Minas Gerais state, southeastern Brazil, between the coordinates 17°04'36"S and 17°06'25"S, 44°32'48"W and 44°35'34"W (Fig. 1).This site lies on elevations ranging from 500 m to 700 m above sea level.Vegetation is composed mainly by deciduous forest on the slopes, interconnected to a riparian forest along the Riacho Fundo stream.Limestone outcrops occur in the highest points of the study area, with a peculiar flora composed principally by cacti (Cereus jamacaru, Tacinga saxatilis), bromeliads (Encholirium sp.), and sparse trees (Ceiba rubriflora, Ficus sp., Enterolobium contortisiliquum, and Jatropha sp.).Areas subject to human use, occupied by plantations and pastures are also found along the main stream.
The avifauna survey was based on 10-species lists, also known as MacKinnon lists (see Mackinnon & Phillips, 1993), a very useful method for rapid assessment surveys and to estimate species richness (Ribon, 2010;MacLeod et al., 2011;Cavarzere et al., 2012).We used the Chao1 estimator for a comparison to the observed richness obtained by field work, following recommendations by Herzog et al. (2002).The data were 1,000-times randomized to produce rarefaction curves of observed and estimated richness, using EstimateS 9.1.0(Colwell, 2013).We also calculated the index of frequency in MacKinnon lists (IFL) for each species, which was obtained by dividing the number of lists of 10 species in which each species occurred by the total number of lists (Ribon, 2010).Species were treated as frequent (IFL above 10%), medium frequent (IFL between 5% and 9.9%), and infrequent (IFL below 5%).
Bird species were identified by vocal recognition and by observations with binoculars.Voucher specimens were collected with an air rifle and 10 mist nets, under SISBio permanent permit number 28301-1.Sampling effort of mist nets was 5,200 m².h (following Straube & Bianconi, 2002).Specimens were prepared as study skins and have been deposited in the ornithological collection of the Museu de Ciências Naturais da Pontifícia Universidade Católica de Minas Gerais (MCNA), Belo Horizonte, Brazil.

Species accounts and analyzes of geographical variation in some taxa
We analyzed series of two species of woopeckers (Celeus ochraceus and C. flavescens) and five species of woodcreepers (Sittasomus griseicapillus, Lepidocolaptes wagleri, L. squamatus, Dendrocolaptes platyrostris, and Xiphocolaptes albicollis) that exhibit geographical variation in eastern Brazil (Appendix 1).These analyzes were based on specimens housed in MCNA, in the Coleção Ornitológica do Centro de Coleções Taxonômicas da Universidade Federal de Minas Gerais (DZUFMG), and in the Museu de História Natural e Jardim Botânico da Universidade Federal de Minas Gerais (MHNJBUFMG).Plumage color of these specimens was analyzed in order to detect patterns of geographic variation.For woodcreepers, plumage descriptions were based on Munsell soil color charts (2000).Measurements were taken using callipers and a ruler.Measurements taken (following Baldwin et al., 1931) were: length of total culmen, length of bill from nostril (only for Celeus spp.), length of closed wing, length of tail, and length of tarsus.

Avifaunal survey
We recorded 172 bird species at Curral de Pedras (Appendix 2).The rarefaction curve, based on 226 MacKinnon lists, suggests that more species would likely be added in future surveys, with an estimated richness of 196 species (Fig. 2).
Saltatricula atricollis (Black-throated Saltator) is the only savanna species, which is typical of the Cerrado province (following Silva, 1995Silva, , 1997;;Silva & Bates, 2002).This is probably related to the fact that Curral de Pedras is predominantly forested, without typical native open vegetation types.In the study area, the Black-throated Saltator was recorded in low frequency (0.9%) and only in pastures.
The following species are included in red lists of Minas Gerais state (COPAM, 2010) and/or Brazil (ICMBio, 2014): C. zabele, Spizaetus tyrannus (Black Hawk-Eagle), and S. ornatus (Ornate Hawk-Eagle).We failed to find P. roquettei, an endangered species at state, national and global levels (COPAM, 2010;ICMBio, 2014;BirdLife International, 2017).Nevertheless, given that it has been recorded in nearby areas (Raposo et al., 2002;Kirwan et al., 2004;Vasconcelos et al., 2006Vasconcelos et al., , 2008;;Lopes et al., 2008) and that Curral de Pedras habors suitable habitats to this species, such as dry and gallery forests, we suspect that further surveys will find it in the study area.

Species accounts and analyzes of geographic variation in some taxa
Below, we present comments on the distribution, geographic variation, and natural history of some species.

Campylopterus calcirupicola (Dry-forest Sabrewing)
This is a recently described species endemic to dry forests of central Brazil (Lopes et al., 2017).Curral de Pedras is located 75 km southwest from the type locality, in Montes Claros.Despite being a medium frequent species (IFL = 5.8%), it was recorded predominantly inside dry forest during the dry season (in May and July), where it was observed visiting flowers of Ruellia brevifolia (Acanthaceae) and Gurania sp.(Cucurbitaceae).During the rainy season (November), it was observed in the vegetation growing over limestone outcrops, visiting flow-ers of Tacinga saxatilis (Cactaceae).On September, one individual was observed bathing among wet leaflets of a shrubby legume (Fabaceae).
Three females were collected (MCNA 4977, MCNA 5050, MCNA 5190).Using the identification key provided by Lopes et al. (2017), these specimens match perfectly C. calcirupicola, including rectrices narrow and pointed (in comparison to C. diamantinensis), with bright bronze green basal half (Fig. 3).Measurements and body mass of these specimens are presented on Table 1.
Given the scarcity of records of this species, with only eight known collecting sites (Lopes et al., 2017), the occurrence of C. calcirupicola in Curral de Pedras is very important to the knowledge of its range, since it is located at its southernmost limits.

Trogon surrucura (Surucua Trogon)
This species, sometimes treated as an Atlantic Forest endemic (Brooks et al., 1999), was frequent in our survey  (IFL = 10.2%).All records were obtained in dry and gallery forests.Three specimens were collected: a female (MCNA 4965) and two males (MCNA 5181, MCNA 5182).They represent the nominotypical subspecies, which, in northern Minas Gerais, is the expected taxon occurring west of the Espinhaço Range, in the São Francisco River hydrographic basin (Vasconcelos & D'Angelo-Neto, 2007).In the eastern slopes of this mountain range, it is replaced by the subspecies T. s. aurantius, which lives in semideciduous (Atlantic) forests (Vasconcelos & D'Angelo-Neto, 2007).
Amongst the most important diagnostic characters of the three subspecies, in C. f. flavescens the mantle, the back, and the upperwing-coverts are black, narrowly barred white; whereas in C. f. ochraceus those parts are tinged of cinnamon-buff with black spots (occasionally absent), varying from heart-shaped to chevron-shaped (Short, 1972(Short, , 1982;;Winkler et al., 1995;Winkler & Christie, 2002).Measurements of C. f. flavescens are also larger than those of C. f. ochraceus (Short, 1972;Winkler et al., 1995).Celeus f. intercedens has dorsal white bars broader than C. f. flavescens, sometimes presenting the dorsal buffy tinge of C. f. ochraceus (Short, 1972(Short, , 1982;;Winkler et al., 1995).This subspecies is considered intermediate in size between C. f. flavescens and C. f. ochraceus (Winkler & Christie, 2002).Despite presenting variable plumage, sometimes considered intermediate in dorsal barring between C. f. flavescens and C. f. ochraceus, and with some of the rust tinge of C. f. ochraceus, these characters are considered stable in some areas (Short, 1982;Winkler et al., 1995), which led Short (1982) to recognize Celeus f. intercedens as a subspecies.Short (1982) also pointed out that an intergradation zone between C. f. flavescens and C. f. ochraceus probably occurs in southern Bahia and Espírito Santo states, but without any signal of intermediate and stable phenotypes of C. f. intercedens.Benz & Robbins (2011), based on molecular data, proposed that C. ochraceus deserves full species status instead of being regarded as a subspecies of C. flavescens.Nevertheless, these authors pointed out that the taxonomic status of C. f. intercedens still needs to be clarifyied, based on populations of C. ochraceus that resemble this taxon in suture zones located in Central Brazil.
More recently, Firme (2015) presented a systematic revision of Celeus, in which C. f. intercedens was synonymized with C. f. flavescens, based on the large varia-tion of diagnostic characters previously assigned to both subspecies.Firme (2015) also considered C. ochraceus as a full species.Thus, in the most recent revision of this complex, only two taxa are considered: C. flavescens (including intercedens) and C. ochraceus.
Celeus ochraceus was infrequent in our sampling (IFL = 4.4%), occurring exclusively in dry forest habitats.One male was collected (MCNA 5057).It presents cordiform black spots on the back and on the wings (Fig. 4), a diagnostic character of C. ochraceus (Firme, 2015).Measurements of this specimen are also close to the range of the small series (n ≤ 9) of C. ochraceus analyzed by Short (1972), which are (in mm): wing: 140.0; tail: 83.9; bill (from nostril): 21.0.The only exception is tarsus length (26.3 mm), which is larger than expected and matches the range of measurements of C. flavescens.

Lepidocolaptes cf. squamatus (Scaled Woodcreeper)
This species was frequent during our survey (IFL = 19.5%),occurring mainly in dry forest and its borders.We collected two specimens (MCNA 4978, MCNA 5237).Both exhibit plumage of crown as assigned for L. squamatus (following Silva & Straube, 1996).Nevertheless, their underparts are more similar to that of L. wagleri, showing inconspicuous dark borders on the streaks of breast and belly (Fig. 6).Other specimens also present intermediate plumage between L. wagleri and L. squamatus (Silva & Straube, 1996;Vasconcelos et al., 2012; see also  Appendix 1).Thus, it is worthy that a detailed revision coupling molecular and morphological data is conducted within this complex in order to understand a possible intergradation zone between the two taxa.
Currently, two subspecies of D. platyrostris are recognized: D. p. platyrostris and D. p. intermedius (Marantz, 1997;Marantz et al., 2003).A comprehensive revision of plumage variation of this species was recently presented by Cabanne et al. (2011), who found two areas of plumage stability: one in the Cerrado of central Brazil (population I) and another in southern Atlantic Forest (population II).They considered that population I represents D. p. intermedius and population II is represented by D. p. platyrostris, despite their type localities are out of those areas of plumage stability.
We collected four specimens of D. platyrostris in the study area (MCNA 5051,MCNA 5189,MCNA 5235,MCNA 5236).They are intermediate between both stable populations found by Cabanne et al. (2011), which also suggests geographic variation following Gloger's ecogeographic rule.This was expected because Curral de Pedras is located between the range of the two areas of plumage stability.Also, the study area represents a collection gap for D. platyrostris.Thus, a detailed description of the plumage of the four specimens is presented on Table 2, following Cabanne et al. (2011).Despite presenting small interindividual variation, the specimens' overall plumage pattern is darker and more similar to population II (southern Atlantic Forest) than to population I (central Brazil) (Table 2).Some character states from our series do not correspond to any of those described by Cabanne et al. (2011) (Table 2).

Xiphocolaptes albicollis (White-throated Woodcreeper)
This was a medium frequent species during our survey (IFL = 6.2%), recorded exclusively in dry forest.We collected a pair in the study area (MCNA 5203,MCNA 5204).The plumage features of the specimens collected in the study area matches the description of the subspecies X. a. bahiae.So far, X. a. bahiae has been reported to occur only in the state of Bahia (Cory, 1919;Marantz et al., 2003).In comparison with specimens of the nominotypical subspecies from Três Rios (Rio de Janeiro) and Bertioga, São Paulo, they are generally paler (Fig. 7), with crown and nape very dark brown (10YR 2/2) to very dark grayish brown (10YR 3/2) with streaks varying between very pale brown (10YR 7/3.5) and light pale brown (10YR 6/3).In comparison to nominotypical specimens, crown and nape contrast less with brown (10YR 4/2.5) back.Cheeks are unstreaked, varying between pale yellow (2.5Y 8/2) and light gray (2.5Y 7/2).These cheeks form a continuous moustachial stripe to a loral spot of the same color, a pattern that resembles X. falcirostris.Breast and sides varies between brown (10YR 5/3) and light olive brown (2.5Y 5/3), with streaks light yellowish brown (2.5Y 6.5/3), without dark spots on their borders.Belly is less barred in comparison to specimens of nominotypical race.Another character that approaches these specimens to X. falcirostris is their more pronounced decurved bills (Fig. 7).
It is noteworthy that in specimens collected in semideciduous forests, between areas of humid and dry forests, the plumage pattern appears to be intermediate between X. a. albicollis and X. a. bahiae.They present dark head and striped cheeks similar to specimens of the nominotypical subspecies.Nevertheless, the breast streaks are never bordered by dark spots and belly barring is less pronounced, similar to X. a. bahiae.The background color of breast and upperparts present a wide variation among specimens, with some resembling those of X. a. albicollis and others with a pattern closer to that of X. a. bahiae (Fig. 7).
We checked photographs of the type of X. a. bahiae, housed in the Field Museum of Natural History (FMNH 65976), and its overall plumage pattern is similar to spec-imens collected at Curral de Pedras, except for its more streaked back and its rump that appears more rufescent.We also analyzed a photograph of a topotypical specimen, which presents a similar plumage pattern of both specimens collected in the study area (Silva, 2015).Pinto & Camargo (1961), based on the analysis of the two specimens of X. a. villanovae used in the original description of this subspecies, suggested a relationship of this taxon to X. a. bahiae.We checked photographs of the type of X. a. villanovae, housed in the Museu de Zoologia da Universidade de São Paulo (MZUSP 7593) and it differs from X. a. bahiae by presenting a very strong streaked pattern on the back and a densely barred belly.Nevertheless, since both type localities of X. a. villanovae and X. a. bahiae are located east to the Espinhaço Range in Bahia state, it is possible that X. a. villanovae is just an extreme of the cline occurring in its northern range.
These results also suggest a geographical variation in X. albicollis following the Gloger's ecogeographic rule, with darker specimens from the humid coastal forests and paler specimens from dry forests.Intermediate specimens do occur between these regions in semideciduous forests.Also, the less contrasting head and nape with back, the conspicuous moustachial stripe, the less barred abdomen, and the more decurved bill of specimens from Curral de Pedras approach them to X. falcirostris, known to occur only in the opposite (left) bank of the São Francisco River (Silva & Oren, 1997).Cory & Hellmayr (1925) were the first to note that "by the lighter, less blackish pileum, less distinctly streaked auriculars, and absence of black barring underneath X. a. bahiae seems to form the transition to X. falcirostris, of northeastern Brazil".This suggests a possible gene flow between both species.Thus, we strongly suggest that further detailed taxonomic revisions and phylogeographic studies must be conducted to access the possible clinal variation in X. albicollis and to test possible hybridization of this species with X. falcirostris, which also shows a not well resolved pattern of geographic variation (Silva & Oren, 1997).In future studies it would be extremely important to collect specimens from intermediate localities between the known range of these taxa, including the subspecies X. a. villanovae, sometimes treated as a subspecies of X. falcirostris (Cory & Hellmayr, 1925-but see Meyer de Schauensee, 1966;Pinto, 1938), and X. f. franciscanus, previously considered closer to X. albicollis (Pinto, 1978).

Knipolegus franciscanus (Caatinga Black-Tyrant)
The Caatinga Black-Tyrant was a frequent species in the study area (IFL = 15.0%).In the same area, Marcos A. Raposo observed c. 50 individuals of this species in a single hour of field sampling in November 1995 (Kirwan et al., 2004), despite we never found it in such abundance (see below).The majority of records of habitat use were obtained in dry forests or their edges (77.1%, n = 27), despite we also observed it in the vegetation growing on limestone outcrop (20%, n = 7) and in gallery forest (2.9%, n = 1).Among all individuals observed during our sampling effort of 31 days (n = 43), the majority were Pap.Avulsos Zool., 2018;v.58: e20185815 8/18 males (90.7%), while only 9.3% were represented by females.We do not know if females are actually less abundant than males or if they present a more secretive behavior, which, together with their more cryptic plumage pattern, make them less detectable than males.In a proportional rate, we obtained eight specimens represented by seven males (MCNA 4966,MCNA 5049,MCNA 5055,MCNA 5180,MCNA 5227,MCNA 5230,MCNA 5268) and a single female (MCNA 5191).
Nothing is known about the species' breeding biology, but based on measurements of testicle size of the seven collected males, breeding appears to occur during the end of dry season and the beginning of the rainy season, between September and November (Table 3).

CONCLUSION
The data presented here shows that the avifauna of a Cerrado dry forest enclave in central Minas Gerais present a few taxa typical of the Caatinga, despite located only c. 100 km south of this biogeographic province.This suggests extinction of Caatinga species as other dry forest enclaves are far from this SDTF nucleous.Also, the size of dry forest enclaves, coupled with their distance in relation to the Caatinga province, appears to be an important factor shaping the occurrence of Caatinga birds.For example, in the Paranã River Valley, the largest Cerrado dry forest enclave (Werneck, 2011), with a considerable area of c. 13,000 km² (Bianchi & Haig, 2013), but c. 300 km west of the Caatinga, only the following taxa have been reported: Columbina picui strepitans, Icterus jamacaii, and Compsothraupis loricata (Blamires et al., 2002;Pacheco & Olmos, 2006;Martins, 2007).Nevertheless, other factors should be responsible for this pattern, since several phytophisiognomies occurring in the Caatinga do not occur in Central Brazilian dry forests (Andrade- Lima, 1981;Rizzini, 1997), probably excluding some taxa typical to this biogeographic province.Thus, we suggest future studies aiming to understand these patterns of occurrence of typical Caatinga birds in the Cerrado dry forest enclaves.
Finally, based on the specimens collected during this study, we reinforce the need of further research in taxonomy, geographic variation, and phylogeography of the two species complex of woodcreepers that were sug-gested to occur on different banks of the São Francisco River: Lepidocolaptes squamatus (right bank) / L. wagleri (left bank) and Xiphocolaptes albicollis (right bank) / X. falcirostris (left bank).In comparison to Amazonian rivers, the São Francisco River seems not to work as an effective geographic barrier to prevent gene flow between woodcreepers species (Marantz et al., 2003).Future detailed phylogeographic studies and taxonomic revisions should focus on the possible intergradation between L. squamatus / L. wagleri and X. albicollis / X. falcirostris in the dry forests occurring across this river.

Figure 2 .
Figure 2. Species rarefaction curve (continuous line) and estimation curve using the Chao1 estimator (dashed line) for the avifauna of Curral de Pedras, Minas Gerais, southeastern Brazil, using 226 MacKinnon lists.

Figure 1 .
Figure 1.Map of the study area in South America (black dot) in the Cerrado biogeographic province and its position in relation to the Caatinga and the Chaco biogeographic provinces (A); detail of the study area at Curral de Pedras with tracks of the sampled trails (white lines) (B).

Table 1 .
Morphometrics and body mass of three female specimens of Campylopterus calcirupicola from Curral de Pedras, Minas Gerais, Brazil.

Table 2 .
Plumage characters of four specimens of Dendrocolaptes platyrostris collected in Curral de Pedras, Minas Gerais, Brazil.