BETA DIVERSITY OF BIRDS ( Passeriformes , Linnaeus , 1758 ) IN SOUTHERN AMAZON DIVERSIDADE BETA DE PÁSSAROS ( Passeriformes , Linnaeus , 1758 ) NA AMAZÔNIA MERIDIONAL

The importance of estimating the biological diversity and understanding how ecological specialization of species changes with spatially-organized habitats in undeniable. High beta diversity between neighboring places means an elevated number of species living within small distances, which usually are adapted to local conditions and highly vulnerable to anthropogenic actions as deforestation and burning. We investigated beta diversity of birds from the order Passeriformes in Southern Amazon, within landscapes with a large heterogeneous vegetation cover (habitats with flooded forest, dry forest, and marsh palm) through sound, observational, and photographic censuses. We marked 126 points in equidistant transects. A total of 148 species of birds were identified, distributed in 27 families. We found that 97% of the species foraged in flooded forest, 77% in dry forest and 19% in marsh palms, and only 18% foraged in the three habitats. An ordination analyses revealed which species showed the strongest preference to each habitat. The analysis for the Global beta diversity showed that this value is high (Whittaker: 7.7405), and the same pattern was obtained with the measure of pairwise dissimilarity. An influence of spatial distance was clearly observed in the cluster analysis and confirmed with a partial Mantel analysis; however, this was not observed at points that coincided with the transition and substitution of species. The influence of spatial distance in the dissimilarity index (beta diversity) was significant (r: 0.0608, p: 0.0049). The assemblage of species in small local populations with high beta diversity may be at risk if deforestation, selective logging, and poaching continue. In summary, this study provides important information on specific habitats with high beta diversity that may be at risk of destabilization of its populations due to continuing environmental changes imposed by humans.


Introduction
Ecologists have long been curious to understand how biological diversity changes with the environment, in particular species specialized in different ecologies in spatially-organized habitats (1) .The smaller fraction of regional gamma diversity corresponds to the local alpha diversity and represents the number of species within small areas of similar and uniform habitats.Beta diversity represents the rate of species variation (turnover rate) between habitats (2) .Beta diversity frequently measures the substitution of species without considering the relative abundance (3) .However, the inclusion of the relative abundance or frequency index allows a more informative evaluation of species diversity, especially when it varies between habitats (4) .Studies that compare variation in the number of species between local habitats are necessary to determine patterns, particularly in places in the world that are enriched in bird species (5) .In these areas, deforestation and controlled burnings for agricultural expansion along with hunting pressure are decreasing species richness.
High beta diversity along an environmental gradient reflect the specialization of species in habitats.The quantification of this parameter can be used to design strategies for protection of bird diversity in these areas (6) .A high beta diversity within small distances reflects the presence of abundant species between neighboring habitats, which generally are locally specialized and highly susceptible to anthropogenic action (3) .The spatial variation of birds composition with types of habitats is rarely measured, but it has shown to be important to define the area needed to protect some species (3) .There are some studies on the changing of the composition of animal communities along altitudinal gradients, but beta diversity in tropical systems is particularly misunderstood (7,8) .According to these studies, a gradual change in composition occurs with altitude, although some results may have been affected by the sampling method (9) .Instead of using the total animal community, quantification of beta diversity using a specific taxonomic group, such as birds (order Passeriformes), may be easier in terms of research planning (10) .Despite the apparent capacity for dispersion, many factors seem to influence variation of bird diversity or delimit their geographic distribution, for example the presence of rivers (4) .Habitats created by rivers, such as floodplain forest, flooded habitats, and habitats with fluctuations in vegetation, present about 15% regional birds, suggesting that watercourses may have had a relevant role in the origin of different bird species in the current region of Southern Amazon.The present study investigated the change in the composition of bird species of the order Passeriformes along gradients in the Alto-Guaporé region, Southern Amazon, a region that presents an elevated richness of birds and has a constant substitution of species due to a high diversity of environments.We aimed to answer two questions: Are spatially closed habitats more similar in species composition (beta diversity)?Are there differences in species richness among the type of sampled habitats (flooded forest, dry forest, and marsh palm with palm tree fruits)?We used the observed patterns of variation in species composition obtained within a bird community and explored possible implications for conservation planning.

Methods
This study was performed from December 2011 to September 2012, in three large localities (A,B,C) in the region of Alto-Guaporé in Southern Amazon, a region characterized by a heterogeneous vegetation cover (180879.94S,8331318.14W between 780708.00S,8507364W; see Fig. 1) Beta diversity of birds (Passeriformes, Linnaeus, 1758) in southern Amazon We sampled three types of local habitats: marsh palm, dry forest, and flooded forest.The marsh palms are habitats of planes in Southern Amazon with riparian areas dominated by Mauritia flexuosa, flooded during the rainy period and permanently humid in the dry periods (n: 18 points).In the dry forest there is no rainfall, the height of trees fluctuates between 30 and 60 m, with closed canopy and thick bush (n: 72 points).Flooded habitats occur along the banks of the rivers Alegre, Verde, and Guaporé.These habitats are always flooded and the trees present a maximum height of 20 m with lianas and aquatic plants, and without sub-woods (n: 36 points).The census procedures included the most efficient hearing census (11) , and observational census with the utilization of binoculars and photographic equipment.Identification of each foraging bird species was accomplished with the collected photographs and recordings.We also referred to specific bibliography to associate the diet of each bird species with the food available in each sampled habitat (11)(12)(13)(14)(15)(16)(17)(18) , so that we could make inferences on each foraged habitat per bird species.We used the quantitative lifting method (adapted from Blondel et al.) (19) .We sampled points in transects equidistant 5 km from each other.Each transect had six points located 200 m apart (this distance was adequate to avoid juxtaposition of territory between most species and allowed us to cover all habitats).In total, 126 points were sampled.The recordings in each point were done using audio digital equipment (96KHz), for 15 minutes during the morning (period of the day when birds are most active), three times in different days.The sampling effort consisted of 63 days of census and 31 hours and 50 minutes of recordings covering approximately 700 km in length, in total.The vocalizations were then edited and the species identified, and finally revised by a bird specialist (records were deposited in the Laboratory of Mammalogy, UNEMAT, Brazil).
To investigate the presence of clustering of species relative to the type of sampled habitat, the PCA method was applied (Principal Components Analysis is a statistical procedure that uses an orthogonal transformation to convert a set of observations of possibly correlated variables into a set of values of linearly uncorrelated variables called principal components).We calculated the global β diversity to measure diversity between the sampled habitats (2) .The Index quality of Jaccard (cc), a matrix of presence and absence for a dataset, represents the macrostructure of species distributed between the habitats, although application is restricted to habitats with more than seven species.This Index was calculated with the following formula: cc= c/ (a+b+c)*100 (1) , or cc=c/ (A+B-c)*100 (2) where: "a" is the number of species unique to sample 1, "b" is the number of specie unique to sample 2, "c" is the number of species common to samples 1 and 2, "A" is the total number of species in sample 1, and "B" is the total number of species in sample 2.
The distance values are plotted as similarity dendograms, the lower the index the less the similarity, created through the UPGMA (Unweighted Pair Group Method with Arithmetic Mean).The correlation between two matrices of dissimilarity was evaluated: one matrix of spatial distance (Euclidean) and the other one of species dissimilarity (Bray-Curtis) utilized the Mantel Test (Number of permutatitons: 999) (20).All analyzes were performed using R version 2.14.0.2011 program (21) .

Results
A total of 148 species of birds (order Passeriformes) were identified and distributed in 27 families.Of these, 97% (n= 145) of the species foraged in flooded forests, 77% (n= 114) in dry forests, and 18.91% (n= 28) in marsh palm.Only 18% (n= 27) of the bird species foraged in all three habitats.The habitat with the highest species richness was the flooded forest (7) (Table 1).The value of species richness per habitat and sampling areas can be seen in Table 2.The ordination showed that most species are highly associated with a habitat, for instance, Mimus saturninus is mainly observed within the flooded forest habitat, and Empidonomus varius, Onychorhynchus coronatus, Thamnomanes caesius, Myrmotherula menetriesii, and Dysithamnus mentalis in the dry forests (Figure 2).The analysis of the global beta diversities in total showed that substitution of species is predominant (Whittaker: 7.7405, Harrison: 0.14334, Cody: 554.5, Routledge: 0.700, Wilson-Shmida: 32.97, Mourelle: 0.610, Harrison2: 0.024, Williams: 0.5714).These values indicate that beta diversity is high.As expected, we found the same pattern when applying the measure of pairwise dissimilarity.In the cluster analysis (Figure 3), the effect of spatial distance is clear because in groups 1 and 2 the habitats closer to each other are very similar (according to the Jaccard similarity index; r= 0.872, p = 0,002).This is not observed with the points of groups 3 and 4; these points corresponded to the moment of transition and substitution of species.However, in the majority of cases, the influence of spatial distance was confirmed with the partial Mantel analysis.The influence of spatial distance on the dissimilarity index (beta diversity) was significant (Mantel statistic r: 0.0608, p (value): 0.0049) (Figure 4).

Discussion
Silva (22) suggested that many amazon species penetrate the Brazilian savanna biome following the gallery forest or vice versa, both places presenting a high diversity of birds (23) .In this study, the sampled places are very close to the transition between the Amazon and the Brazilian savanna, and this fact may explain the substitution of species that we observe.The high variability in the composition of species (unequally distributed) in the sampled points also contributed to the high species richness of birds Passeriformes.As suggested by Williams (24) , the increased level of environmental heterogeneity in the sampled habitats may be very relevant in explaining the high species richness that we found.Cohn-Haft, Whitaker, and Stouffer (25) also proposed that habitat heterogeneity is the most important determinant for bird species richness inside the Amazon, and not primary productivity or level of rainfall as advocated by other authors.
Concerning the preference for habitats, this study showed that, although some species revealed to be habitat specialists, most bird species of the order Passeriformes explored and foraged many types of habitats.The habitat that presented more species was the flooded forest (Table 2).In this type of habitat, birds may find suitable conditions that directly influence their life cycle, such as, water, food, shelter, and protection from predators (26,27) .In the Amazon, many species, even residents, disperse according to water and food resources and these are abundantly present in the flooded forest habitat.Ferreira et al. (28) found more richness of species at flooded habitats, a finding that agrees with our own results.The species Mimus saturninus had the highest level of association with Beta diversity of birds (Passeriformes, Linnaeus, 1758) in southern Amazon this type of habitat, though always with open grasslands with scattered trees and shrubs close-by.We believe this may be explained by the fact that this species lives in groups of thirteen individuals and they forage the river for reproduction and food (fruits).In the dry forest habitat there are many winged insects, which could explain the presence of Empidonomus varius, Dysithamnus mentalis, Onychorhynchus coronatus, Thamnomanes caesius, and Mymotherula menetriesii in this habitat, since they primarily feed in these insects.
The high beta diversity that we found in this study is very important, particularly for strategic conservation plans.We provide specific points along an environmental gradient in the Amazon forest, where a high richness and high variation in species composition within different types of habitats may be found.More attention should be given to the importance of local conservation.High beta diversity should be taken into consideration when designing a natural reserve, particularly in including, placing and in the extension of contiguous heterogeneous habitats (29) , which contrast with lowland habitats where species are widely distributed (11,30,31) .The positive correlation between the species dissimilarity and geographic distance observed for Passeriformes in this study is in line with the Neutral Theory (32) , which states that species similarity in a community decreases with increased geographic distance between different environments.This means that because of the limitation in species distribution, the more distant the habitats are, the more different they will be in terms of species composition (33) .This author assumes that the limitation in species distribution is a function of immigration and local extinction, with species appearing and disappearing.Therefore, for species that have a small distribution, local and regional variation will impact their distribution the most.The analysis of the Jaccard similarity index revealed a major point where substitution of species is significant.Groups 3 and 4 appear on a point of transition and mixture of species close to the North Amazon, where in a small space, many different species may be found.This species composition is organized in limited local populations with high beta diversity that should be protected from the deforestation, selective logging, and poaching that have been occurring because this community is particularly vulnerable to disruption, and immediate recovery will most likely lead to reorganization of an unbalanced ecosystem.Thus, this study provides important information on specific habitats with high beta diversity whose stability is at risk in the face of continuing environmental challenges.