The birds of Reserva Extrativista Chico Mendes , South Acre , Brazil Aves da Reserva

This paper describes the avifauna sampled at Reserva Extrativista Chico Mendes, Acre, Brazil, during October and November, 2008. We recorded 344 bird species of 17 orders and 57 families through point counts, mist-nets and general observations. The most prevalent families were Tyrannidae, Thamnophilidae and Thraupidae with 53, 36 and 22 species, respectively. We recorded some range restricted, little know, and habitat specialists birds exemplified by Crypturellus atrocapillus, C. strigulosus, Primolius couloni, Aulacorhynchus prasinus, Drymophila devillei, Simoxenops ucayalae, Cnipodectes superrufus, Hemitriccus flammulatus, Percnostola lophotes, Xiphorhynchus chunchotambo, and Conioptilon mcilhennyi. Although we surveyed only during the dry season, the rarefaction curves indicate a satisfactory sampling effort. The data show that the Chico Mendes reserve holds a unique Amazonian bird community, which is influenced by the presence of bamboo and second growth vegetation. The results of this paper reinforce the biological importance of the RESEX and highlight the need for more inventories and bird studies at this isolated and little known region of the Brazilian Amazon.


InTRoduCTIon
The western Amazonian bird community is very poorly known, and the Acre avifauna is still being discovered and described.Despite the recent history of ornithological surveys of the State of Acre (less than 60 years of studies), up to 655 bird species have been recorded, and around 75 new species are likely to be included in the State list in the near future (Guilherme, 2009).Given the fast rate of change in our knowledge, all possible information describing the avifauna of this State are significant for the future management and conservation of this important and remote part of the Amazon.
Since the 1950's some researchers have collected and surveyed birds in the State of Acre.One of the first bird collections in this State was led by P. E. Vanzolini in 1951(Vanzolini, 1952), who confirmed the first 140 bird species for Acre State, and four new species for Brazil (Pinto & Camargo, 1954).A few other bird surveys took place in Acre in late 1950's and 1960's (Novaes, 1957(Novaes, , 1958)), but ornithologists only published the first bird checklists about the avifauna of this region in the 1980's and 1990's (for a detailed description see Guilherme, 2009).Most of the published bird surveys were carried out in the Juruá river Basin in northern Acre (i.e.Novaes, 1957Novaes, , 1958;;Whittaker & Oren, 1999;Whittaker et al., 2002).Other papers include inventories from the Acre river basin (i.e.Pinto & Camargo, 1954;Guilherme, 2001).This study describes and highlights the important observations of the avifauna sampled at Reserva Extrativista Chico Mendes (RESEX Chico Mendes), a federal extractive reserve located further south in the Acre river basin, almost at the Brazilian-Peruvian-Bolivian border.

METhodS
This study was conducted in RESEX Chico Mendes (10° to 11° S, 68° to 70° W), in Acre State, Southwest Amazon, Brazil (Figure 1).This reserve has approximately 1-milion hectares of 'terra firme', non-flooded, openedcanopy forests, with understory dominated by lianas, small palms and/or bamboos (Costa, 2000).The Chico Mendes Reserve is located in a region with annual average temperature around 24 ºC, and a mean precipitation of 2,000 mm (Costa, 2000).The rainy period is from December to March and it is drier from June to August (Costa, 2000).After an extended drought in the region, combined with slash-and-burn techniques used by most of the reserve inhabitants, about 300,000 ha of forest in the reserve were accidentally burned during August and October 2005 (Brown et al., 2006;INPE, 2010).For this reason, the study site is a mosaic of burned and unburned primary Amazonian 'terra firme' forests with patches dominated by bamboos.Most of the burned areas were forested and had been regenerating for three years at the time of the study in 2008.We conducted samples in the end of the dry season, during October 5 th and November 10 th 2008, in the center of the reserve and approximately 100 km from Bolivian-Peruvian border, in Xapuri municipality (Figure 1).By unburned forests, we refer to sites that had not burned in the 2005 droughtbut the presence of bamboo in the region may indicate a relatively recent history of fire (Nelson & Irmão, 1998).
We sampled twelve 550 m trails that were all located at least 1 km apart from one another.We used mist-nets, point counts and random observations (between the coordinates 10° 13' S, 68° 44' W and 10° 32' S, 68° 42' W) (Figure 1).We performed the random observations at other sites, not just along the trails, including main roads and plantations.For each avifaunal sampling trail, 28 mist-nets (12 x 2.5 m; mesh size 36 mm) were erected in four groups of seven nets along 550 m transects.Each group created a netline of 7 x 12 m extending for 90 -100 m.Groups were separated by an open space of 50 m.We opened the 28 nets for two days, from sunset (6:30 h) to about 13:30 h, accruing 2,050 mist-net hours in total (a capture effort of 11,720 hm 2 ).We checked the nets hourly and closed it during periods of heavy rain.All birds captured were identified to species level, weighed, and measured (standard measurements included wing, tail, bill, and total length) and, Figure 1.Map of Brazil (A), the State of Acre (B), and RESEX Chico Mendes (C).Arrows and dots showing approximate location of sampling trails (map adapted from Costa, 2000).whenever possible, were aged, sexed and photographed.The captured birds were banded with a numbered metal ring obtained from Centro Nacional de Pesquisa para Conservação de Aves Silvestres (CEMAVE) -Instituto Chico Mendes de Conservação e Biodiversidade (ICMBio) (The Brazilian Bird Banding Laboratory in the Brazilian National Institute of Environment).All recaptures from the same sampling period and from the same net line were excluded from the analysis to avoid double counting.
In addition to the mist-netting, Luiz Mestre carried out 96 point counts along the trails.The point counts were accomplished two times at the same point on two different days, but not the same days as the mist-netting.First sampling was conducted between 6:30 h and 7:30 h, and the second between 7:30 h and 9:00 h.Each point count consisted of ten minutes of bird observations and recording, spaced 150 m each other (50 m, 200 m, 350 m and 500 m along each transect).All bird registrations were recorded using a digital recorder and a directional microphone during point counts and confirmed (if possible) visually by binoculars.Unknown vocalizations were subsequently checked against known calls and, if necessary, confirmed by consulting with other experienced ornithologists (i.e. A. Aleixo, S. Dantas, E. Guilherme).The distance from the observer and the height at first detection were also noted.We excluded from the richness and abundance analysis all the birds flying or registered outside the range of 50 m radius of each point count.We also did not include birds that came from directions of other sampling points (primarily wide-ranging canopy species such as Pscitacidae, Embereziidae and Thraupidae), to avoid double counting.Individuals of most species are unlikely to have been registered on consecutive days, because sampling was always carried out at different times of the day.However, we may have inflated numbers of individuals for some territorial species.For this reason, we compared the cumulative curves in three different ways.First, we plotted the curves using only first day sampling separately (Figure 2, PC1).Second, we plotted both days together including all registrations (Figure 2, PC2).Third, we only included the first detection event of a species at a point, using day two only to accumulate species we did not observe on day one.If a species was recorded on the first day and then again on day two, we excluded the day two observation (Figure 2, PC3).
We analyzed the three methods used separately (mist-nets, point counts, general observations) based on number of individuals and percentages.We used percentages of individuals classified separately in the levels of order, family and species.For this reason, the sum of percentages of different categories (taxonomical level) does not reflect the total of upper taxonomical level.We used point counts and mist-net data to calculate rarefaction species curves based upon number of individuals.We calculated these curves with incidence data from point counts and mist-nets separately.We analyzed the patterns of species richness between different methods comparing individual-based rarefaction curves constructed using the analytical formulae from EstimateS v.7 (Colwell, 2004).The comparisons were standardized by of the number of individuals, as we were interested in patterns of species richness and not species density (Gotelli & Colwell, 2001).An estimate of the 'true' species richness in each quantitative method (mist-nets and point counts) were calculated using the software EstimateS v.7, using the mean of the four commonly employed abundancebased estimators (ACE, CHAO1, JACK1 [Jackknife], and BOOTSTRAPPING).We followed the CBRO (2009) list for nomenclature and taxon ordinance.

RESulTS And dISCuSSIon
We registered a total of 344 bird species at RESEX Chico Mendes.These bird species were included in 17 orders and 57 families.The most representative families were Tyrannidae, Thamnophilidae and Thraupidae, with 53, 36 and 22 species respectively.Mist-nets captured 868 individuals comprising 137 species from 33 families, and point counts recorded 2,380 individuals from 186 species in 38 families.Mist-nets added 55 species, and general observations added 105 species to the total number of species.
The rarefaction curves from mist-nets, point-counts or both methods together did not reach an asymptote.However, point counts and both methods summed, showed a possible beginning of stabilization (Figure 2).The point count estimate of the mean 'true' richness was 213.1 (± 13.6) species, using the four predicted abundancebased estimators' means (see methods).Mist-net estimated richness was 169.5 (± 9.3) species using the mean of the four estimators.It is possible to observe a better-estimated richness on associating both sampling methods, resulting in 288.3 (± 13.8) species (Table 1).However, it is important to consider that these numbers were from limited samplings based only on point counts and mist-nets and for only one specific season during one year.This result shows the importance of supplementing standardized methods with general observations when a complete species list is the desired outcome.The random observation method added 106 bird species to this checklist, permitting us to estimate for this period about 400 species in the area if combining this method, point count estimate, and mist-net estimate (106 + 288.32 = 394.32).
We highlight important observations of restricted range and little known southeast Amazon birds (Table 2).Relevant examples include the following species.Crypturellus atrocapillus, a species limited to the western Amazon (Inambari endemism range).In Brazil, it was only recorded in Acre State; in Parque Nacional da Serra do Divisor (Guilherme, 2009) and Reserva Extrativista do Alto Juruá (Whittaker & Oren, 1999).We recorded C. atrocapillus mostly in second growth and post-burned sites.Crypturellus strigulosus is a species with a restricted Amazonian range, apparently using mostly Campinas and Campinaranas habitats (Guilherme, 2009).We recorded this rare species in three sampling sites both in pristine and post-burned areas.Jabiru mycteria is included in a secondary list and has few records for Acre State (Guilherme, 2009); we saw a dead juvenile hunted by a local resident.We have no evidence of this registration (picture or skin) but the easy identification of this distinctive species means we were able to include this species in the list.Primolius couloni is a species that is globally vulnerable due to habitat degradation and illegal trade (Birdlife International, 2010;IUCN, 2000); and the limited range in east Peru, north Bolivia and southwest Brazil, also contributes to its vulnerability (Tobias & Brightsmith, 2007).In Brazil, P. couloni was mainly observed in Acre State (Whittaker & Oren, 1999;Guilherme, 2009).In the study sites, we observed the species flying and perched in pairs, mostly in pristine forests, and on few occasions, we registered larger  , 1997;Whittaker & Oren, 1999).
We captured and recorded few individuals of this species in pristine and second growth areas.Percnostola lophotes is a species limited to the western Amazon (Hoyo et al., 2003).
In Brazil, it was only recorded in Acre State (Whittaker & Oren, 1999;Guilherme, 2009).We captured two individuals of P. lophotes in pristine and post-burned sites.Hypocnemis peruviana and H. subflava are in their range limit and were recorded and photographed in the sites.Their ranges were also confirmed in Guilherme (2009) and Isler et al. (2007).
Xiphorhynchus chunchotambo occurs in northwest South America, and in Brazil, this species has only been registered in Acre State (Guilherme, 2009).We registered about ten individuals of this species both in pristine and second growth sites.Conioptilon mcilhennyi occurs in east Peru, north Bolivia and southwest Brazilian Amazonia.This species has only been recorded in Acre State within Brazil, but despite being historically rare, has been associated with second growth and post-burned forests (Guilherme, 2009;Mestre et al., 2009).Additionally, we underline the importance of this region for the conservation of some North American migratory birds.During this study, we recorded Pandion haliaetus, which uses most of the large Amazonian rivers and breeds in northeastern USA and southern Canada (Hoyo et. al, 1994;Mestre & Bierregaard, 2009).Tringa solitaria which breeds in north of North America (Alaska, Canada, USA) and migrates to the southern United States, Central and South America (Hoyo et al., 1996).Contopus cooperi that occurs from Alaska (USA) to Southeast Brazil (Hoyo et al., 2004).Contopus virens, which breeds in eastern North America, migrating to north South America (Sick, 1997;Hoyo et al., 2004).Tyrannus tyrannus occurs from Canada to northern Argentina (Hoyo et al., 2004).Progne subis breeds in North America and north Central America migrating to South America during the Neartic winter (Hoyo et al., 2004).Catharus ustulatus breeds in North America and migrate to South America (Hoyo et al., 2005).Finally, Coccyzus americanus which, despite having Acre within its proposed range (Erize et al., 2006, Restall et al., 2007, Infonatura, 2007), had not previously been registered in the State (Table 2).We clearly observed this species in a forested site, near post-burned forests, perched in the midstory.Although we do not have a documented register of this species, our observation can be considered a first important clue regarding its occurrence in these sites.
The RESEX Chico Mendes forests are also clearly important for avian bamboo specialists, once we recorded 17 species within bamboo formations.Based on Kratter (1997) classification, we found two species that are considered to be obligate bamboo specialists (Drymophila devillei and Hemitriccus flammulatus), seven species of near obligate specialists (Celeus spectabilis, Simoxenops ucayalae, Automolus melanopezus, Percnostola lophotes, Myrmeciza goeldii, Ramphotrigon megacephalum and Ramphotrigon fuscicauda), and three facultative bamboo specialists (Campylorhamphus trochilirostris, Epinecrophylla ornata, and Microrhopias quixensis).We found D. devillei, H. flammulatus and P. lophotes only in sites with abundant bamboo formations, including some second growth and post-burned areas.For this reason, our data also confirm the restriction of these species to bamboo forests as Table 2. Bird species list of RESEX Chico Mendes, South Acre, Brazil.The list follows the CBRO bird checklist from 2009 and specify the taxon (including bird species, family and order); Portuguese and English names of each species; the status (St) of the bird species as resident (R), North migrant (VN) and if the species was registered in point counts (PC), mist-nets (MN) or by general observations (GO) (# status not confirmed CBRO).Habitats where species were registered 1: Primary Forests; 2: Second growth sites including post burned; 3: Plantations, open and/or aquatic habitats; 4: Flying.CA, Categories of Abundance in the sampling sites (c: common, u: uncommon, r: rare); TE, Types of Evidence (r: recording, p: picture, c: collected and deposited in Museu Paraense Emílio Goeldi, o: only observed).

Taxon
Portuguese  (1996) and Hoyo et al. (2003).The species S. ucayalae is considered near threatened by IUCN (2000), and in some cases can be considered as restricted to bamboo forests as well (Guilherme, 2009).We captured this species mostly in sites dominated by bamboos, including second growth and post-burn sites.Despite its near threatened status, S. ucayalae populations may be using these human-influenced environments.It is also relevant to describe the registration of Cnipodectes superrufus, an uncommon bamboo specialist from southwest Amazonia, which may be least abundant and perhaps the most threatened of all bamboo specialists in the Amazonian forests (Lane et al., 2007;Tobias et al., 2008).We captured only one individual in a bamboo/ second growth forest in a post-burned area (10° 20' S, 68° 40' W).Finally, it is important to cite five other species registered in our study (Crypturellus atrocapillus, Nonnula ruficapilla, Monasa flavirostris, Lophotriccus eulophotes and Neopelma sulphureiventer) which apparently have a preference of bamboo habitats as well (Kratter, 1997), however, the few registrations in literature and in our field samplings do not permit generalizations (Table 2).The results of this paper showed that the RESEX Chico Mendes holds a unique Amazonian bird community, influenced by bamboo and second growth vegetation.Although the avifauna described here is far from a complete census of the RESEX, and needs to be resampled in other seasons, the rarefaction curves and estimators indicate a good sampling effort.These data, regarding the local avifauna, reinforce the biological importance of the RESEX and highlight the need for more inventories and bird studies at this isolated and little known region of the Amazon.

ACKnowlEdgEMEnTS
We thank to local people from Reserva Extrativista Chico Mendes who welcomed and worked with us.We also thank Sidnei Dantas, Alexandre Aleixo and Edson Guilherme who gently helped us to identify the bird songs, pictures and specimens.We also thank to two anonymous referees who significantly helped to improve this manuscript.The data presented here are part of the project "Biodiversity implications of forest disturbance and related landscape dynamics in the Brazilian Amazon"

Figure 2 .
Figure 2. Rarefaction curve from EstimateS Software based on mist-nets (MN), point-counts (PC1, only first day sampling each trail considered; PC2, all samples considered; PC3 using the second day only to accumulate species not seen in the first day), and the two methods together (MN+PC2), where estimates are solid lines and 95% confidence interval dotted line.

Table 1 .
The mean of the 'true' species richness, estimated by point-counts (PC), mist-nets (MN), and both together (PC+MN) by the four richness estimators calculated in EstimateS v.7 software; ACE, Chao, Jack (Jackknife), Bootstrapping, and total mean of the estimators.In parentesis are standard deviations of these means.groups with about eight individuals.Aulacorhynchus prasinus, despite occurring from Central America to much of western South America, has only been registered a few times in Brazil, and only in Acre State (Sick