What are the most important factors determining different vegetation types in the Chapada Diamantina , Brazil ?

A transect was used to examine the environmental and biological descriptors of a compact vegetation mosaic in the Chapada Diamantina in northeastern Brazil, including the floristic composition, spectrum of plant life forms, rainfall, and soil properties that defined areas of cerrado (Brazilian savanna), caatinga (seasonally dry tropical forest thorny, deciduous shrub/arboreal vegetation) and cerrado-caatinga transition vegetation. The floristic survey was made monthly from April/2009 to March/2012. A dendrogram of similarity was generated using the Jaccard Index based on a matrix of the species that occurred in at least two of the vegetation types examined. The proportions of life forms in each vegetation type were compared using the chi-square test. Composite soil samples were analyzed by simple variance (ANOVA) to examine relationships between soil parameters of each vegetation type and the transition area. The monthly precipitation levels in each vegetation type were measured and compared using the chi-square test. A total of 323 species of angiosperms were collected distributed in 193 genera and 54 families. The dendrogram demonstrated strong difference between the floristic compositions of the cerrado and caatinga, sharing 2% similarity. The chi-square test did not demonstrate any significant statistical differences between the monthly values of recorded rainfall. The organic matter and clay contents of the soilsin the caatinga increased while sand decreased, and the proportions of therophyte, hemicryptophyte, and chamaephyte life forms decreased and phanerophytes increased. We can therefore conclude that the floristic composition and the spectrum of life forms combined to define the cerrado and caatinga vegetation along the transect examined, with soil being the principal conditioning factor determining the different vegetation types, independent of precipitation levels.


Introduction
The Chapada Diamantina, the northern extension of the Serra of the Espinhaço Range in Brazil, is included within the Caatinga biome of northeastern Brazil and hosts a large diversity of vegetation types associated with its physiographic spectrum, including cerrado (neotropical savanna), campo rupestre (open, rocky field grasslands), forest, and caatinga (seasonally dry tropical forest thorny, deciduous shrub/arboreal vegetation) (Harley, 1995;Juncá et al., 2005).
A number of floristic studies have already been undertaken in the Chapada Diamantina, including those considering: campo rupestre (Conceição et al., 2007;Neves and Conceição, 2010), cerrado (Costa et al., 2009), caatinga (Lima and Lima, 1998), submontane seasonal semi-deciduous forest, upper montane forest, and gallery forest (Funch et al., 2008;Nascimento et al., 2010;Couto et al., 2011).Rapid ecological surveys of diverse vegetation types have also been undertaken (Harley et al., 2005;Juncá et al., 2005;Funch et al., 2009), although most of the studies considered only be shrub-arboreal layer and/or made collections on irregular schedules.Funch et al. (2009) published a vegetation map of the Chapada Diamantina National Park and the surrounding region describing vegetation types, based on Harley (1995), and associated with different soil types distributed contiguously within a relatively limited area.In light of the intriguingly close dispositions of different vegetation types under apparently identical macro-climatic conditions, we investigated cerrado and caatinga vegetation sites and their ecotone along an E-W transect just west of the National Park to contribute to our understanding of the phytogeography of Brazil.
Our study focused on the floristic compositions of the herbaceous-subshrub and shrub-arboreal layers, life forms, rainfall, and soil properties in addressing the question of which factors are most important in defining the cerrado and caatinga vegetation along this transect.
We hypothesized that variations in the physiognomy, floristic compositions, and structures of natural ecosystems could occur due to variations in the chemical and physical properties of their soils (Haridasan, 2000;Amorim and Batalha, 2006;Juhász et al., 2007;Silva et al., 2013), and that permeability and the proportions of fine particles would be the parameters most closely associated with species richness (Medinski et al., 2010).We also considered that: 1) climatic factors have a decisive role in determining species and plant community distributions (Terradas, 2001;Lavers and Field, 2006); 2) water availability is one of the principal determinants of vegetation cover (Silva and Batalha, 2008;Cianciaruso and Batalha, 2009), and; 3) life forms are important elements in vegetation descriptions (Raunkiaer, 1934).

Study area
The Chapada Diamantina National Park has a much accentuated topography of low mountains with rocky peaks and steep slopes, narrow valleys with extensive plains to the west that are separated by long chains of hog-back mountains.Altitudes in the region vary from 400 to more than 2,000 m a.s.l.The isolated peaks and extended mountain chains have extensive rock outcrops and litholic neosols (thin, rocky soils of low fertility), while soils on the intermediary plains are predominantly latosols (deep, well-drained, acidic, with low fertility) (Juncá et al., 2005).
The regional climate is mesothermic (Cwb) (Köppen, 1948), with average monthly temperatures varying between 18 and 25 °C, with the lowest temperatures occurring during driest months.The average annual precipitation is 1,218 mm (for the period between 1951-2011).The rainy season usually begins in November and extends until March, while the dry season initiates in June and continues until October (Juncá et al., 2005).
The present study was conducted in two vegetation types: cerrado, represented by three local physiognomies (two sites of open cerrado -OP, rocky outcrop cerrado -ROC, and typical cerrado -TC) and caatinga (CAA).The cerrado-caatinga transition (CCT) between of these both vegetation types also was sampled (Table 1).These vegetation are contiguously distributed along a 10 Km E-W transect sampled (12º 26' 05" S to 12º 27' 7" S to 41º 31' 05" W to 41º 35' 52" W) just west of the Chapada Diamantina National Park in the Chapada Diamantina, Bahia State, Brazil.The phytophysiognomies of cerrado are included in cerrado stricto sensu (Ribeiro and Walter, 2008).Caatinga is characterized by a predominance of low shrubs and trees that are highly branched, often spiny, and deciduous in the dry season (Pennington et al., 2009).

Floristic composition and life forms
The vegetation types were selected along a 10 Km east-west transect based on the vegetation map of Funch et al. (2009) and identified in the field as distinct areas of well-conserved cerrado and caatinga.We established ninety 100 m 2 plots (15in each of the four vegetation types examined: the two types of cerrado, the cerrado/caatinga transition, and caatinga) for a total of 0.9 ha.Plant collections were performed monthly in all sites from April/2009 to March/2012 of both fertile and sterile plant material of herbaceous-subshrub and shrub-arboreal species (this period included three rainy seasons and three dry seasons).The voucher specimens were deposited in the Herbarium of the State University at Feira of Santana (HUEFS) and identified with the aid of specialists and by comparison with previously identified material.The classifications of the plant families followed the proposal of the APG III (2009).
The life forms of the specimens were classified according to Raunkiaer (1934), excluding epiphytes and considering succulents separately from phanerophytes.The proportions of herbaceous-subshrub/shrub-arboreal (epiphytes, succulents, chamaephytes, hemicryptophytes, and therophytes /phanerophytes) species were calculated for each of the vegetation types, for cerrado-caatinga transition, and for the different cerrado physiognomies.
The importance values (IV) of the herbaceous-subshrub species (the sum of coverage and relative frequency) and shrub-arboreal layer (the sum of the density, dominance and relative frequency) were used to identify the five most important species of each layer in each vegetation type (Neves, 2013).

Rain fall and soils properties
Rainfall gauges were installed in each vegetation type and in the ecotone.Monthly precipitations levels were measured between January/2010 and March/2012 (Figure 1).Three composite soil samples (0-20 cm) were collected in each vegetation type and in the ecotone.Each composite sample was taken by mixing soils from three random points, totaling 18 composite samples (12 in the cerrado, three in the caatinga and three in the cerrado-caatinga transition).The physical and chemical properties of these samples were analyzed in an agricultural laboratory (Embrapa Semi-Árido -PE), according Donagema et al. (2011).

Statistical analyses
To determine the floristic similarities between the cerrado, caatinga and ecotone sites (considering the different cerrado physiognomies), clustering analysis was performed based on their Jaccard similarity (JS) as derived from a binary matrix, using the Unweighted Pair Group Method with Arithmetic Mean (UPGMA) run on Past software (Hammer et al., 2001).The cut-off used in this study was 25% following Mueller-Dombois and Ellenberg (1974).We compared the proportions of species in each life form class among the different vegetation types and the three cerrado physiognomies using the chi-square test (ᵡ 2 ).The average values and standard errors of each physical and chemical component of the soils in the different vegetation types were calculated and simple analysis of variance (ANOVA) was performed to detect statistical differences between them, utilizing Sisvar software (Ferreira, 2000).The Tukey test (P<0.05)was used to compare the observed averages.The rainfall indices in the vegetation areas were compared using the chi-square test (ᵪ 2 ).

Life forms
The floristic spectra of life forms in the sampled vegetation are presented in Figure 2. Chi-Square tests indicated significant differences in the percentages of chamaephytes, therophytes and phanerophytes species between the caatinga and the cerrado, and between the caatinga and the cerrado-caatinga transition areas.Significant statistical differences (using the same test) were also noted between the cerrado (except the rocky outcrop cerrado) and caatinga, and between the cerrado and the cerrado-caatinga transition for the hemicryptophyte.

Similarities
The dendrogram demonstrated strong difference between the floristic compositions of the cerrado and caatinga, sharing 2% of similarity (Figure 3).Caatinga is the most distinct physiognomy and the cerrado-caatinga transition area is more similar with cerrado (24%).The cerrado physiognomies form a coherent group with 39% similarity.

Rainfall and soil properties
The chi-square test did not demonstrate any significant statistical differences between the monthly values of recorded rainfall.The average values and standard errors of the physical-chemical parameters of the soils of each vegetation type and the different cerrado phytophysiognomies are presented in Table 2.The cerrado soil was predominately  sandy, while cerrado-caatinga transition and caatinga areas were sandy clay loams.The soils in all of the vegetation types were dystrophic (base saturation < 50%), acidic (pH < 5, except in the open cerrado, site I) and had relatively low Al contents (Al < 1.3 cmol/dm 3 , except in the cerrado-caatinga transition) and with exchangeable retention capacities less than 13, with high H + e Al +3 loads.

Discussion
The cerrado and caatinga vegetation sites as well as their transition, were well delimited in terms of their floristic compositions, life form spectra and soil properties.The cerrado vegetation demonstrated plant communities with ample varieties of three-dimensional structures and spatial heterogeneities, as reflected in the three different phytophysiognomies identified.Pennington et al. (2009) noted that cerrado vegetation physiognomies can be found under the same (or slightly more humid) climatic conditions as seasonal tropical dry forests, and that these vegetation types are often be found relatively close to one another.
Changes in the floristic compositions of the cerrado, cerrado-caatinga transition, and caatinga were accompanied by changes in the proportions of total sand, clay, and organic matter of their soils.Although high species-richness is often found in ecotone areas due to their greater environmental heterogeneity (Rodrigues and Nave, 2000), the transition zone studied here was less richness than the cerrado and caatinga surveyed.It is probable that the environmental conditions in that area are more rigorous and limit the occurrence of larger numbers of species (Mota et al., 2011).
The cerrado-caatinga transition area shared a greater percentage of species with the cerrado (approximately 70%) than with the caatinga (approximately 16%), indicating a high floristic similarity between them.More than 50% of the life forms shared between these two sites were therophytes -thus making essentially insignificant contributions to the overall differences in physiognomies between those vegetation types.Approximately 60% of the species shared by these two vegetation types were phanerophytes, which demonstrated deciduousness as a strategy to reduce water losses during the dry period (Neves, 2013).
The best represented life-form in the cerrado vegetation studied here was that of therophytes -a situation different from other cerrado sensu stricto areas that generally show high proportions of phanerophytes (Batalha and Martins, 2004;Costa et al., 2004).Phanerophytes were the most representative life forms in the caatinga vegetation studied -which was also different from the findings of Costa et al. (2007) in shrub caatinga communities growing in "warm and  low latitude and altitude steppe climates (Bsh, according to the Köppen (1948) system)" that were well-represented by therophytes.While the principal strategy used in Caatinga phytoclimates to avoid water loss during the dry season is a predominance of therophytes, the caatinga studied here adapted a phanerophyte strategy of leaf-fall (Larcher, 2000).
The systematic collections undertaken in the present study, which covered three rainy seasons, three dry seasons, and three intermediate seasons, may have been responsible for the contrast between our life form data and those of studies by Batalha and Martins (2004) and Costa et al. (2004Costa et al. ( , 2009) ) who all made shorter and less frequent collections.Batalha and Martins (2004) noted that species whose buds are not exposed to the open air (i.e.hemicryptophytes, cryptophytes, and therophytes) tend to be under-represented among collections undertaken during just a single period.Consecutive collections during a 36 month period (in the present study) allowed for ample sampling of the floristic composition.
No significant differences were observed between the three different cerrado phytophysiognomies in terms of their soil properties (principally in relation to physical parameters) and their respective floristic spectrums of life forms.Dantas and Batalha (2011) encountered relationships between edaphic conditions and floristic composition and richness in a fine-scale study of a cerrado area.Marimon Junior and Haridasan (2005) noted that soil fertility did not support the hypothesis of the contiguous occurrence of cerradão vegetation and cerrado sensu stricto vegetation.The same authors reported that the greater percentage of clay in the cerradão site allowed the establishment of greater densities of trees due to greater soil-water availability.Ruggiero et al. (2002) reported that campo cerrado, cerrado sensu stricto and cerradão could not be distinguished based on soil chemical parameters.The greater species richness and densities of shrub-arboreal individuals in the typical cerrado site in the present study appeared to be associated with the greater proportions of clay and organic material there that would presumably increase water retention, aid the growth of soil microorganisms, and increase nutrient availability for plant growth (Motta et al., 2002).The presence of blocks of stone in the rocky outcrop cerrado, the occurrence of wild-fires in the typical cerrado, and the densities of arboreal individuals and the interactions between different species in each phytophysiognomy appear to be factors generating the distinct cerrado phytophysiognomies.
While low precipitation levels (less than 700 mm/year) concentrated into as few as three consecutive months during the summer in the southern hemisphere and associated with average temperatures of 26 °C (Nimer, 1989) are considered typical conditions for the establishment of caatinga vegetation (Andrade- Lima, 1981;Sampaio, 1995), rainfall was not a useful factor for delimiting cerrado and caatinga vegetation, such as their transition, in the present study.The soil-water relationships in these vegetation sampled may be influenced by their respective proportions of total sand and clay, because although the cerrado-caatinga transition and caatinga sites had rainfall levels slightly less than that of the cerrado, the water retentions of their soils would presumably be greater due to their larger proportions of clay and organic matter -thus favoring the establishment of greater plant abundances and species richnesses in the shrub-arboreal layer (Motta et al., 2002).
This study demonstrates that the floristic compositions, floristic life form spectra, and the heterogeneity of the soil were elements that delimited cerrado, cerrado-caatinga transition, and caatinga.The soil was the principal conditioning factor determining the different vegetation types, independent of precipitation levels.As the organic matter and clay contents of the soil increased and the percentages of sand decreased the proportions of therophytes, hemicryptophytes and chamaephytes diminished, while phanerophytes increased.The importance of edaphic conditions, and not just the scarcity and irregularity of rainfall, as is usually supposed, must be considered as a factor in establishing caatinga vegetation.
It must also be emphasized that no single variable determined the occurrence of each vegetation type, but rather the interactions between them.Species interactions themselves likewise influence the establishment of vegetation formations, as species richness depends on a mix of factors that can alter the effects of other variables.
Appendix.Species list in the vegetation types and cerrado physiognomies, such as in the cerrado-caatinga transition, Chapada Diamantina, NE Brazil.

Species
Voucher

Table 1 .
Descriptions of vegetation types physiognomies and vegetation types in the mosaic parameters cerrado-caatinga, Chapada Diamantina, NE Brazil.

Table 2 .
Averages and standard deviations of the physical and chemical characteristics of the soils in the different vegetation type areas, Chapada Diamantina, NE Brazil.