Species composition, diversity, and vegetation structure in a gallery forest-cerrado sensu stricto transition zone in eastern Mato Grosso, Brazil

We investigated the existence of an ecotone in the species composition, diversity and vegetation structure in the transition between gallery forest and cerrado sensu stricto in central Brazil. We tested two hypotheses: 1) a ecotone can be found between gallery forest and cerrado; 2) a gradient exists in the species composition of the cerrado. We established three parallel transects of 5 m × 350 m running between gallery forest and cerrado, which were divided into subplots of 5 m × 10 m, perpendicular to the margin of the Bacaba stream in Nova Xavantina (Mato Grosso, Brazil). We identified and measured the height and diameter of individual plants in the plots with a diameter of at least 3 cm at 30 cm above the ground. We recorded 140 species, of which 26 were exclusive to the gallery forest, and 95 to the cerrado. The cerrado presented higher species richness (observed and estimated) and diversity (diversity profiles) than the gallery forest. Both hypotheses were accepted: a distinct ecotone was observed between gallery the forest and cerrado, and a pronounced gradient was found in species composition among the cerrado plots, apparently in response to the variation in soil moisture content, probably related to topography.


Introduction
To date, 11,242 native species of vascular plants have been identified in the Cerrado savanna biome of central South America, making this one of the biologically richest savannas found anywhere in the world (Mendonça et al. 2008).On a local scale, high plant species richness is related to variation in the physical-chemical conditions of the soil, relief, and topography, which may result in the formation of gradients of diversity (Oliveira-Filho et al. 1989;Oliveira-Filho et al. 2004;Marimon et al. 1998;Moreno & Schiavini 2001;Carvalho et al. 2013;Loschi et al. 2013).These local factors result in the formation of a distinct mosaic of forests, savannas, and grasslands, which is typical of the Cerrado Biome (Ribeiro & Walter 2008;Reatto et al. 2008).
Gallery forests and the cerrado sensu stricto are also found within this mosaic.The gallery forest is an evergreen habitat that forms on mesotrophic soils associated with streams (Ribeiro & Walter 2008).The Cerrado is a brevideciduous savanna formation (Lenza & Klink 2006), which is generally found on distrophic soils and in interfluvial areas (Ribeiro & Walter 2008).The transition between forest and savanna vegetation types may be either abrupt, with marked changes in species composition and structure (Askew et al. 1970;Moreno et al. 2008;Marimon et al. 2010;Loschi et al. 2013) or gradual, with species being substituted over a gradient (Askew et al. 1970;Oliveira-Filho et al. 1989;Oliveira-Filho & Fluminham-Filho 1999;Moreno et al. 2008).
In the eastern region of the Brazilian state of Mato Grosso, gallery forests and cerrado can often be found in close proximity to one another, but these plant communities are characterized by a distinct transition in terms of species composition and other characteristics (Askew et al. 1970).However, more gradual variation can be observed within a given vegetation type, as found by Marimon et al. (1998) in cerrado on a scarp slope and by Marimon-Junior & Haridasan (2005) in a transition zone between cerrado (savanna) and cerradão (savanna woodland).These studies indicate that changes in the species composition and structure of the vegetation are determined by the variation in relief or edaphic characteristics.In more subtle gradients, changes in the vegetation and species composition are more difficult to observe, and can only be understood more conclusively through the integration of studies of species richness and composition, and the vertical and horizontal structure of the vegetation (Marimon et al. 1998;Cardoso & Schiavini 2002;Juhász et al. 2006;Gonçalves et al. 2011).A number of different studies in the Cerrado Biome have indicated that the differences in the composition and structure between the savanna formations and interfluvial forests (cerrado stricto sensu, savanna woodland, and mesophilic forest), and the forest formations associated with bodies of water (gallery and riparian forests) are related to changes in the physical (moisture and flooding levels) and chemical (availability of nutrients) properties of the soils, which are determined primarily by relief and the topography of the terrain (Oliveira-Filho et al. 1990;Moreno & Schiavini 2001;Carvalho et al. 2005;Gonçalves et al. 2011;Loschi et al. 2013;Carvalho et al. 2013).Given these considerations, the present study compared the variation in the structure and species diversity and composition of the woody, nonwoody and liana vegetation of gallery forest and cerrado sensu stricto in eastern Mato Grosso state, Brazil.Two hypotheses were tested: 1) there is an abrupt shift in the species composition and diversity and structure between the two types of vegetation, and 2) within the cerrado, this transition is more gradual in response to variation in the height of the terrain in relation to the level of the water in the adjacent stream.

Material and methods
The present study focused on the Bacaba Municipal Park, a conservation unit located within the transition zone between the Amazon forest and Cerrado savanna in the municipality of Nova Xavantina, in the east of the Brazilian state of Mato Grosso.The region's climate is Aw in the Köppen system, with well-defined rainy and dry seasons.Mean monthly temperature is 25°C, and annual precipita-tion is typically between 1300 mm and 1500 mm (Marimon et al. 2010).The park is dominated by cerrado strictu senso vegetation, primarily typical cerrado (Marimon et al. 1998;Marimon-Junior & Haridasan 2005;Gomes et al. 2011), but also rocky cerrado (Marimon et al. 1998;Maracahipes et al. 2011).Other forest formations, such as cerradão (savanna woodland) (Marimon-Junior & Haridasan 2005) and gallery forests (Marimon et al. 2010), can also be found within the park.
We established three parallel transects of 350 m × 5 m separated by 100 m and perpendicular to the right margin of the Bacaba stream (Fig. 1).Each transect was subdivided into 35 plots of 5 m × 10 m.All three transects traverse a road (3 m in width) perpendicularly, although this area was not included in the samples (Fig. 1).The slope of the terrain along each transect was measured using a clinometer made from a 15 m long flexible crystal quarter-inch PVC hose (1.5 mm thick) to determine the height of the center of each plot in relation to the level of the water in the Bacaba stream during the dry season (May, 2014).To investigate the possible existence of a gradient in the characteristics of the vegetation along the three cerrado sensu stricto transects, the plots were grouped into three sectors of 100 m, according to their distance from the edge of the water -sector 1 = 50-150 m from the water; sector 2 = 150-250 m; sector 3 = 250-350 m.
With regard to the vertical structure of the vegetation, we visually classified the five plots nearest to the Bacaba stream as gallery forest and the 30 more distant plots as cerrado, resulting in a total sample of 15 plots (750 m²) in the gallery forest and 90 plots (4,500 m²) in the cerrado.Eight of the cerrado plots were characterized by dense stands of bamboo Actinocladum verticillatum (Nees) McClure ex Soderstr.(Poaceae).These plots were excluded from the analyses due to the effects of this feature on the species composition and structure of the Cerrado vegetation (Silvério et al. 2010), which may have biased the results.
We measured the trunk diameter and height of all live individuals, including woody and non-woody plants, and lianas, with a diameter of at least 3 cm at a height of 30 cm above the soil (D 30cm ).We identified the species in the field and through comparisons with the specimens available in the NX herbarium of the Nova Xavantina campus of Mato Grosso State University (UNEMAT).We assigned the species to families according to the system proposed by the Angiosperm Phylogeny Group (APG III 2009) and we revised the taxon names based on the "Flora do Brasil" database, available on http://floradobrasil.jbrj.gov.br/2014/.
Species similarity between areas was evaluated using Morisita's quantitative coefficient (Magurran 2011) and Sørensen's qualitative coefficient (Brower & Zar 1984).We used an Indicator Species Analysis (ISA) to verify the existence of the species characteristic of each vegetation type (Dufrêne & Legendre 1997;McCune & Mefford 1999).We calculated the Importance Value Index (IVI) of the species (Curtis & Mcintosh 1950;Mueller-Dombois & Ellemberg 1974) found in each vegetation type to compare the horizontal structure of the vegetation between the gallery forest and cerrado.We ordinated the plots using Principal Coordinates Analysis, or PCoA (Felfili et al. 2011) in order to evaluate the existence of a gradient or ecotone in the species composition between gallery forest and cerrado, and tested the significance of the groups generated using ANOSIM with the Bonferroni correction (Clarke & Warwick 1994).
In the specific case of the cerrado, we also evaluated the influence of the topographic gradient on the substitution of species with an abundance of at least 15 individuals through the interpretation of a taxon density graph, an approach proposed by Landeiro et al. (2010).
We compared species richness between the gallery forest and cerrado using rarefaction with 1,000 randomizations, adjusting the sampling effort by the number of individuals recorded in the smallest sample (Gotelli & Colwell 2001;Magurran 2011), and by using the Mao Tau estimator.We estimated plant species richness using the Jackknife I estimator, and used diversity profiles derived from the Rényi exponential series (Tóthmérész 1995) to compare diversity between vegetation types.
We compared the density of individuals per plot and the mean height of the plants in the gallery forest and cerrado using Student's t test with unequal variance, given the lack of homogeneity in variance between the samples (Zar 2010).We applied the nonparametric Mann-Whitney test (U) to the comparison of trunk diameter between vegetation types, given that the assumption of normality was not satisfied, even after the logarithmic transformation of the data (Zar 2010).
We used linear regression to assess the possible existence of a structural gradient in the cerrado determined by the height of the plots in relation to the level of the water in the Bacaba stream (Zar 2010).For this, we used the height of the plots in relation to the level of the stream as the independent variable, and the mean height, trunk diameter and densities recorded per plot as the response variables.
The ISA and PCoA were run in PCOrd version 6.07 (McCune & Mefford 2011).Phytosociological parameters were determined in FITOPAC 2.1.2(Shepherd 2009).Species similarity coefficients, ANOSIM, rarefaction, diversity profiles, the t and Mann-Whitney tests, and linear regression, were run in PAST (PAleontological STatistics) version 2.15 (Hammer et al. 2001), while the taxon density graph was produced in R 3.0.2(R Core Team 2013).

Results
We recorded a total of 140 species, of which, only 19 (13.6%) were found in both vegetation types.We sampled 1667 individuals in the cerrado sensu stricto, representing 114 species (of which 95 or 83% were exclusive to this habitat type), 86 genera, and 40 plant families.In the gallery forest, we recorded 135 individuals belonging to 45 species (26 or 58% exclusive), 39 genera, and 26 families (Tab.1).
The similarity in the species composition between the gallery forest and cerrado was low, whether estimated by the Sørensen's qualitative index (0.24) or in particular by the quantitative index of Morisita (0.07).The analysis of indicator species (Tab. 1) identified 10 indicator species for the cerrado and 12 for the gallery forest (Monte Carlo: p < 0.001), reinforcing the lack of similarity between the vegetation types.In addition, none of the 10 species with the highest IVI scores in the cerrado (36.52% of the total) were included in the 10 most important species in the gallery forest (64.87 % of the total), further reinforcing the distinction in the species composition of the two communities.
The Principal Coordinates Analysis (PCoA) of species density and composition, in which the first (eigenvalue of 4.65) and second (eigenvalue of 2.81) axes explained 11.9% and 7.2% of the variation, respectively, differentiated four distinct groups.The first group was formed by the gallery forest plots, and the other three groups by the three sectors of the cerrado located at different distances from the Bacaba Stream (Fig. 2).The groups formed in this analysis were considered to be consistent, according to the results of the ANOSIM (R = 0.41; p < 0.001).The more ample dispersal of the cerrado plots in comparison with those of the gallery forest indicates that the cerrado encompasses a gradient and is more heterogeneous than the gallery forest.This conclusion is reinforced by the separation of the cerrado plots located at different distances from the Bacaba stream, in particular those located at distances of between 50 and 150 m (sector 1) from this watercourse (Fig. 2).
Acta bot.bras.29(3): 327-338. 2015.Table 1.Phytosociological parameters of the species and families sampled along a gradient of gallery forest (GF) and cerrado sensu stricto (CE), in decreasing order of the IVI values recorded for the cerrado sensu stricto species in the Bacaba Municipal Park in Nova Xavantina, Mato Grosso (Brazil).N = number of individuals sampled; BA = basal area; IVI = Importance Value Index; *= 10 species with the highest IVI values in the gallery forest; IC = indicator species of the cerrado sensu stricto; IM = indicator species of the gallery forest; T= Total.We also noted a gradient in the substitution of species along the cerrado transects.For example, Erythroxylum tortuosum, Syagrus flexuosa, Aspidosperma tomentosum, Salvertia convallariodora and Davilla elliptica presented high densities in the plots furthest from the stream (between 150 m and 350 m), and much lower densities in the adjacent plots (50 m to 150 m).At the opposite extreme, Buchenavia tomentosa, Callisthene fasciculata, Myrcia splendens, Astronium fraxinifolium and Magonia pubescens were found at the lowest densities in the most distant sector (3), while the highest densities were recorded in sectors 2 and 3 (Fig. 3).

S p e c i e s F a m i l y
After adjustment for sampling effort (Fig. 4), the rarefaction curve indicated higher species richness for the cerrado (55±7 species) in comparison with the gallery forest (45 species).Observed species richness in the cerrado represented 85.1% of that estimated by Jackknife 1 (134±4 species), while that of the gallery forest represented 64.3% of the estimated richness, i.e. 70±4 species.Diversity was also higher in the cerrado, irrespective of the index considered (Fig. 5).

N BA IVI S p e c i e s F a m i l y C E G F C E G F C E G F
128     The extrapolated density per hectare in the cerrado was 4066 ind.ha -1 , with a basal area of 22.1 m².ha -1 , whereas in the gallery forest, density was 1800 ind.ha -1 and basal area, 33.3 m².ha -1 .The mean abundance per plot in the gallery forest (9.0±3.9) was significantly lower (t = 6.69, d.f.= 95; p < 0.001) than that recorded in the cerrado (20.3±6.9).However, individual plants were significantly taller (t = -2.15,d.f.= 14.47; p < 0.049) in the gallery forest (mean = 7.3±1.3m) than in the cerrado (3.7±0.2 m), and trunks were significantly thicker (U = 45, p < 0.001) in the cerrado (median diameter = 6.3 cm) than in the gallery forest (median diameter = 11.6 cm).

Discussion
A number of studies have compared forest and savanna habitats in the Cerrado biome (e.g.Oliveira-Filho & Fluminham-Filho 1999;Juhász et al. 2006;Moreno et al. 2008;Carvalho et al. 2013;Loschi et al. 2013) and other South American savanas (e.g.Sarmiento et al. 1984), as well as in African (e.g.Müller et al. 2012) and Australian savannas (e.g.Taylor & Dunlop 1985).All these studies found that variation in species composition and diversity, and habitat structure are determined by local environmentl factors such as relief, topography, and the physical-chemical properties of the soil.In the Cerrado biome, the gallery forests that border watercourses are typically found in valley bottoms, where soils tend to contain more nutrients and moisture, and may even be flooded seasonally, whereas the savanna vegetation (cerrado sensu stricto) is normally found on the higher terrain, on well-drained, weathered, dystrophic soils that are never flooded (Ribeiro & Walter 2001;Oliveira-Filho & Ratter 2002;Ribeiro & Walter 2008).
These local differences in environmental conditions between the gallery forest and in the cerrado probably underpin the marked differentiation observed in the species composition of the two types of vegetation, due to the high percentage of exclusive species.In a cerrado in the Brazilian state of São Paulo, Carvalho et al. (2013) also recorded a low percentage of typical cerrado species in the gallery forest, which they related to the distinct environmental conditions found in the two types of habitat.These findings reinforce the accuracy of the preliminary visual classification of the vegetation conducted in the field, considering only the physical structure of the two habitats.A marked distinction in the species composition of the savanna and gallery forest vegetation was also recorded by Loschi et al. (2013) in the Cerrado, and by Müller et al. (2012) in Africa.In forest ecotones, however, the transition between two types of vegetation is more subtle, and demands more complex analyses, as observed by Cardoso & Schiavini (2002) and Moreno & Schiavini (2001) in a continuum of gallery forest, seasonal forest, and savanna (cerradão) woodland, and by Juhász et al. (2006) and Gonçalves et al. (2011) in gradients of savanna woodland and gallery forest.In some cases, the transition between savanna and gallery forest can be characterized as a distinct type of vegetation, known as "candeal" (see Oliveira-Filho & Fluminham-Filho (1999).
In other words, the characteristics of the transition within and between the different types of habitats found in the Cerrado biome may vary considerably among sites and regions, being related intimately to local variations in edaphic conditions.In some cases, such as that observed in the present study in the gallery forest and cerrado in the Bacaba park, there may be an abrupt ecotonal change between habitats.In other cases, the transition may be more gradual or subtle, such as that observed within the cerrado in the present study.Despite the reduced interpretational power of the first two axes of the PCoA (19%) for the separation of the gallery forest and cerrado, which indicates the need for caution in the interpretation of the data, the analyses of similarity and indicator species, as well as other phytosociological parameters, were emphatic in their differentiation of the characteristics of the two types of vegetation.
It interesting to note that the percentage of indicator species identified in the cerrado (8.8%; n = 10) was considerably lower than that recorded in the gallery forest (26.7%; n = 12).The cerrado plots were also more amply dispersed in the ordination analysis, with a marked tendency for the distance from the Bacaba stream (50-150 m, 150-250 m, 250-350 m) to affect the clustering of the plots.In addition, the taxon density graph for the cerrado plots indicated the presence of a marked gradient in the substitution of species, with some species predominating in the plots closest to the stream, and others being more abundant in the plots furthest away.A similar pattern has been observed by Oliveira-Filho et al. (1989) in interfluvial cerrado in the Brazilian state of Mato Grosso do Sul.Similarly, in their comparison of interfluvial and valley cerrado in the Brazilian Federal District, Sarmento & Silva-Júnior (2006) recorded a high number of exclusive species for both types of vegetation, and Fonseca & Silva-Júnior ( 2004) found a clear difference in their species composition.
The higher species richness and diversity recorded in the arboreal-shrubby cerrado community in comparison with that of the gallery forest is typical of the Cerrado biome.Felfili et al. (2007) recorded high plant species richness and diversity in five areas of Cerrado in comparison with six areas of floodplain gallery forest on the central Brazilian plateau.In the basins of the Araguaia and das Mortes rivers in eastern Mato Grosso, Marimon & Lima (2001) also recorded higher species richness in cerrado habitats in comparison with gallery forest.In their survey of three tracts of the same gallery forest sampled in the present study, Marimon et al. (2002) recorded between 74 and 86 species, that is, values lower than those recorded in the adjacent cerrado here studied.
In the cerrado sensu stricto of Bacaba Municipal Park, Marimon et al. (1998) recorded 103 species on a scarp slope, Maracahipes et al. (2011) identified 85 species in rocky cerrado, and Gomes et al. (2011) registered 89 species in typical cerrado one year after a bushfire.Felfili et al. (2002) concluded that areas of cerrado sensu stricto in proximity to the Amazon forest suffer the influence of this biome and are characterized by higher species richness and diversity.This does in fact appear to be the case, considering the results obtained in the present study area, and in other areas of Cerrado in eastern Mato Grosso, such as that studied by Nogueira et al. (2001) in the municipality of Canarana, approximately 130 km north of Bacaba park, which recorded 88 species, and that of Felfili et al. (2002) in Água Boa (75 km north), where 80 species were recorded.
Both the present study and that of Marimon et al. (1998), who also assessed the characteristics of the cerrado along a topographic gradient in Bacaba park, recorded the highest levels of species richness found to date in areas of Cerrado (see Lenza et al. 2011;Mews et al. 2014;Felfili & Silva-Júnior 1993).Marimon et al. (1998) concluded that this high species richness was related to the slope of the terrain and the inclusion of relatively small individuals, given that the Acta bot.bras.29(3): 327-338.2015.minimum trunk diameter (D 30cm ) in these studies (present study and Marimon et al. 1998) was 3 cm rather than 5 cm, as in Lenza et al. (2011), Mews et al. (2014), and Felfili & Silva-Júnior (1993).However, even if the individuals with a D 30cm of less than 5 cm are excluded from the database of the present study, the species richness of the cerrado (n = 104 species) would still be the highest ever recorded for this vegetation type.This permits us to suggest that the topographic gradient was the factor determining the high plant species richness recorded in the present study of the cerrado, although at the present time, we are unable to determine which edaphic factors may have produced this result.
The low species richness observed in the gallery forest may have been related not only to the reduced sampling effort (total area of 750 m² in comparison with 4100 m² in the cerrado), but also the seasonal flooding of this habitat.In the same tract of gallery forest sampled in the present study, Marimon et al. (2002) recorded 77 species (minimum D 30cm = 5 cm, total plot area = 4700 m²).While Marimon et al. (2002) surveyed a much larger plot than that of the present study, the number of species is still well below the 104 recorded here in the cerrado (minimum D 30cm = 5 cm, total plot area = 4100 m²).This is reinforced by both the Jackknife and Mao Tau estimates, which both indicated higher species richness in the cerrado, contradicting the idea that the smaller size of the gallery forest sample may have biased the results.In fact, reduced species richness is typical of floodplain forests in the Cerrado biome, given that the seasonal flooding restricts the establishment of many species (Nogueira & Schiavini 2003).This alone may account for a large part of the difference in species richness between gallery forest and cerrado found in the present study.
The difference between the cerrado and gallery forest in the height and diameter of the individuals sampled further emphasizes the formation of an ecotone, in this case based on structural features, between the two environments, which facilitated the visual differentiation of the vegetation types during the preliminary survey of the sample plots.This shift is not always so apparent, however, where changes in species composition and vegetation structure occur over a gradient or continuum (Cardoso & Schiavini 2002).
Differences in the density recorded in the gallery forest and cerrado further emphasized the existence of an ecotone in the study area.The higher densities recorded in the cerrado in comparison with the gallery forest are related to the typically arboreal-shrubby characteristics of the cerrado sensu stricto (Oliveira-Filho & Ratter 2002;Ribeiro & Walter 2008).By contrast, the gallery forest is formed primarily by arboreal individuals of larger size (Oliveira-Filho & Ratter 2002;Ribeiro & Walter 2008), which is reflected in the greater basal area recorded in this habitat in comparison with the cerrado.The application of a relatively small lower limit for trunk diameter (D 30cm ≥ 3 cm) would also have increased the possibility of the inclusion of smaller shrubs -which are more typical of the Cerrado -in the samples.
Using a more rigorous criterion of trunk size (diameter at breast height -D 130cm ≥ 10 cm), Felfili (1995) recorded densities of individuals in gallery forest almost twice as high as those recorded in the cerrado, confirming the more arboreal nature of the vegetation in the gallery forest.
A systematic gradient was also found in the structure of the vegetation along the cerrado transects, with plots nearest the stream being characterized by lower density, but larger plants, in terms of their height and diameter.This may be related to the greater availability of moisture in the lower plots (Oliveira-Filho et al. 1989;Fonseca & Silva-Júnior 2004), as well as the more clayey texture of the soil in this part of the terrain (Marimon et al. 1998).Plants in areas with less hydrological stress tend to invest more in their aerial biomass, as suggested by Marimon et al. (1998) and Oliveira-Filho et al. (1989).The reduced determinant coefficients found for the relationship between plot elevation and plant size (diameter and height) and in particular density nevertheless indicate that other factors, not analyzed here, may also affect plant size and density in Cerrado communities.In this case, the confirmation of any direct causal relationship would require more detailed studies of the physical-chemical properties of the soils on which the cerrado community analyzed in the present study is based.
In the present study, we demonstrated an abrupt ecotone in both species composition and diversity and vegetation structure between cerrado and gallery forest, which could be easily observed under field conditions, confirming our first hypothesis.The analysis of the cerrado plots nevertheless indicated the presence of a more subtle gradient in species composition and the size of individuals in relation to the distance of the plot from the body of water, confirming our second hypothesis.Both the ecotone and the habitat gradient found here appear to be related to the indirect influence of topography on the moisture content and fertility of the soil, as shown in a number of previous studies of local transitions found within and among different Cerrado vegetation types (e.g.Oliveira-Filho et al. 1994;Oliveira-Filho & Fluminham-Filho 1999;Cardoso & Schiavini 2002;Juhász et al. 2006;Gonçalves et al. 2011;Loschi et al. 2013).However, physical-chemical analyses of this soil will be needed to elucidate the edaphic factors involved in the differentiation of the two types of vegetation studied here.

Figure 1 .
Figure 1.A: Schematic diagram of the transects (T1, T2 and T3) established in an area of transition between gallery forest and cerrado sensu stricto in Bacaba Municipal Park in Nova Xavantina, Mato Grosso (Brazil).gallery forest; cerrado sensu stricto; + Road.B: Distance from the stream and height of the plots in relation to the level of the water in the stream.

Figure 2 .
Figure 2. Principal Coordinates Analysis (PCoA) of the plant species composition of the cerrado sensu stricto and gallery forest in the plots sampled in the Bacaba Municipal Park in Nova Xavantina, Mato Grosso (Brazil).GF = gallery forest; CE1 = cerrado sensu stricto 50-150 m from the stream; CE2 = cerrado sensu stricto 150-250 m from the stream, and CE3 = cerrado sensu stricto 250-350 m from the stream.

Figure 4 .
Figure 4. Rarefaction curve standardized by the sampling effort (number of individuals) of the smallest sample for the cerrado sensu stricto and gallery forest in the Bacaba Municipal Park in Nova Xavantina, Mato Grosso (Brazil).The gray area represents the 95% confidence interval.