The area is comprised of the municipalities of Águas Belas, Iati and Saloá, which are part of the southern Agreste of Pernambuco (Map 1). The municipalities of Águas Belas and Iati are part of the geomorphological unit of the Interplanaltic Depression of Alto Ipanema, while the municipality of Saloá is located in the Pernambuco - Alagoas Structural Summit (CORRÊA et al., 2010).

Expeditions were made in July 2013. The methodological procedure applied follows the recommendations of IBGE (2012). The sampling technique adopted was Stratified Randomization, which consists of stratifying a given forest area (population) into homogeneous subpopulations (strata), based on the interpretation of aerial photographs or another remote sensing.

For a better operationalization of field procedures, a complementary methodological technique was selected to adapt to the objectives and conditions presented by the study, namely Rapid Ecological Assessment (REA) (SOBREVILLA; BATH, 1992), which allows for quick qualitative data collection but requires prior analysis of the study area. The REA method is based on a previous understanding of the research area, making use of databases, aerial photographs, remote sensing images, and maps, so that field procedures can be defined. Species are collected for identification and deposition in herbaria, and the physical characteristics of the area are noted. Subsequently, all data extracted in the field are treated to provide tools for an analysis of the spatial information. Thus, the REA can generate succinctly a description of the vegetation, perform georeferencing of the vegetation patches, analyze the conservation status of the study area, as well as generate a list of species collected. This method is considered flexible, presents speed, ease usage and data analysis, and allows the study to be conducted with a reduced team of researchers.

Combining the two methodological techniques described in the spatial transect selected for this study, 10 plots of 30 × 30 m were established. These were allocated where significant changes in the vegetation composition as well as in the topographic gradient could be evaluated. It should also be noted that the plots were installed in places where the vegetation was minimally preserved. As a result of these requirements, the plots were not equidistant, which allowed for the research to document the nuances found along the landscape profile. The 10 plots installed among the municipalities can be seen in Map 1 and their locations with respective georeferencing are shown in Table 1.

The plots were distributed to perform botanical collections, to contemplate the dominant species in each selected environment. After the collections, the specimens were taken to the herbarium to be identified and deposited and to test the hypothesis of the variation of vegetation structure along the topographic profile, as well as the occurrence of a cerrado patch at the top of this profile. Also, photographic records were made to distinguish plant physiognomies, and georeferencing of collection sites was performed using a GPS receiver.

The NDVI is simple, easy to implement and can be effective in predicting surface properties when the vegetation canopy is not too dense or too sparse.

The NDVI is obtained through the equation:

Wherein NIR and R are spectral two-way reflectance factors (ratio of the radiance of the target on the surface to the radiance of a conservative, Lambertian surface) at the wavelengths of the visible (red, R) and near-infrared (NIR), respectively (MYNENI et al., 1995). These are the surface reflectance ranges in near-infrared (0.8 µm to 1.1 µm) and visible (red - 0.6 µm to 0.7 µm) (ASRAR et al., 1984).

To obtain the NDVI in the spatial transect analyzed, images of the TM sensor (Thematic Mapper) were used, which are onboard the Landsat 5 satellite, acquired from the Brazilian National Institute of Space Research - INPE. The images used for this study were obtained on December 10, 2013.

The images were stacked and orthorectified. Geometric correction (orthorectification) was performed by image-image comparison. This correction procedure takes one image whose geometry has already been corrected as a reference to correct another image. This step is of fundamental importance since the images obtained have distortions concerning the terrain. Image processing was performed using the ERDAS Imagine version 9.3 program licensed at the Geoprocessing and Remote Sensing Laboratory - SERGEO of the Geographic Sciences Department of UFPE, with the Model Maker tool.

From the analysis of the NDVI vegetation index (Map 2), it was possible to perform a series of correlations from the data collected.

We made comparisons of the NDVI values found in our study with other works, especially those examining the cerrado. In a study carried out in the municipality of Santa Rita do Passa Quatro, state of São Paulo, on the differentiation of cerrado physiognomies using the vegetation index, Bitencourt et al. (1997) found the following values: for the physiognomy campo sujo the NDVI from 0.02 to 0.20; campo úmido from 0.02 to 0.20; campo cerrado from 0.20 to 0.26; cerrado sensu stricto from 0.26 to 0.38; cerrado sensu stricto tending to cerradão from 0.38 to 0.44; cerradão from 0.44 to 0.55 and mata mesófila from 0.44 to 0.55.

Carvalho Junior et al. (2008), in a study on savanna pattern classifications using the vegetation index in Chapada dos Veadeiros, state of Goiás, found NDVI values for both dry and rainy periods in the following Cerrado features: for the cerrado denso environment a mean NDVI value of 0.8 was found for the rainy period and 0.62 for the dry period; for the predominantly campo environment, a value of 0.60 for the rainy period and 0.35 for the dry period; in the environment composed of the cerrado ralo or cerrado rupestre, values of 0.60 for the rainy season and 0.4 for the dry season.

The values of NDVI found for the cerrado patch at the top of the Brejo de Saloá, represented by plots 9 and 10, were similar to those found in areas of cerrado, including the areas of cerrado in central Brazil.

However, it is necessary to emphasize that the variation of the vegetation index values can be explained by the difference in floristic diversity and structural complexity in each physiognomy, because the correlations among the herbaceous, shrubby and arboreal strata of the cerrado trigger important variations in biophysical parameters (ASNER; WARNER, 2003). Thus, the standardization of a certain physiognomy of the cerrado or other biome according to the values obtained through the NDVI is limited, because such determination depends on several parameters such as climate, vegetation type, land use and soil exposure, among other aspects.

A very relevant analysis is the interpretation of the vegetation density values obtained from the NDVI, compared with the altimetry values of all ten plots studied (Graph 1).

The detailed and individualized correlation by parcel explains, among other aspects, the oscillations contained in the correlation of the NDVI values with the altitude values, and the influence of other physical factors. Parcel 1 corresponds to a transect in the municipality of Águas Belas, and has an altitude of 408 m.a.s.l. The area is covered by a fragment of dense arboreal caatinga located on a slope. For this area, the NDVI had a value of 0.35, a relatively high index, mainly because it is inserted in the Sertaneja Depression, where the dominant vegetation is a drier caatinga. However, the high value can be explained because the position of this plot was relatively close to the Brejo de Águas Belas, with higher humidity than the surrounding caatinga, favoring the formation of dense arboreal caatinga vegetation, with higher density values than other plant formations.

The second parcel is also located in the municipality of Águas Belas, at an altitude of 457 m.a.s.l. The plot was established in an area of dense, shrubby caatinga. However, the area is influenced by agropastoral activities. The NDVI for this plot is 0.34, relatively high for the pattern of the Sertaneja Depression. It can be explained by the proximity to the Brejo de Águas Belas, as in the case of the first plot, as well as by the density of the local shrubby caatinga itself.

The third parcel was also installed in the municipality of Águas Belas at an altitude of 396 m.a.s.l. The plant formation of the site analyzed is caatinga, with a marked presence of vegetative, shrubby and herbaceous structures. Regarding the NDVI, this plot presented a value around 0.26, which can be explained by the very physiognomy of the vegetation that does not show high density. These variations are probably observed due to the decrease in humidity to the first two plots. Although there is no microclimatic study for the area, it is possible to infer this premise, since the humidity reflects directly on the density of the vegetation.

Still in the municipality of Águas Belas, the fourth parcel was also installed. This point has an altitude of 394 m.a.s.l., which does not vary significantly from the previous plots. The NDVI for this area was 0.26, which is not high, a reflection of the vegetation coverage, a notably spaced arboreal caatinga.

In the municipality of Iati the fifth plot was installed, at an altitude of 508 m.a.s.l. The physiognomy of the vegetation cover remained as typical caatinga of a dry environment, where a shrubby stratum prevails. The vegetation of this environment is widely spaced and suffers a lot with the agropastoral activity in the region, presenting several degraded areas. This fact influenced the NDVI, which registered the lowest value to all plots (0.18), revealing that in this environment the vegetation is not very dense.

The sixth plot was installed in the municipality of Saloá, where the altitude reached 747 m.a.s.l. and a marked difference in the physiognomy of the vegetation was noted. The vegetation cover of the area is predominantly arboreal, with some elements of the caatinga and cerrado, presenting itself as a formation of montane forest. Concerning the NDVI, it had a value of 0.33. Theoretically, this plot should have a higher NDVI than the others. However, according to records of the vegetation, it is configured as a transitional area, comprising features of cerrado and caatinga. Thus, it presents areas of more widely spaced vegetation, explaining the low NDVI, when compared to plots 1 and 2. Nóbrega et al. (2012) studied the lichen Parmotrema andinum, collected in this same area, for bioprospection and biotechnology. This species was only registered for the cerrado of the Brazilian Central-west region until then, which corroborates our findings.

The seventh parcel was established in the municipality of Saloá. The altitude recorded for this point was 965 m.a.s.l., which has a decisive impact on the configuration and development of the vegetation cover. It is a formation of montane forest, standing out in its exuberance and arboreal size. The density and exuberance of the vegetation cover are reflected in the NDVI, which had a value of 0.47 for this plot.

The eighth parcel was established in the municipality of Saloá, with an altitude of 997 m.a.s.l. The vegetation cover of the area analyzed is characterized as a mixture between shrubby and tree layers, drawing attention to its exuberance and high density, responsible for the highest value of NDVI found, 0.49. The area has an ecotonal character, presenting species typical of montane forest, caatinga and cerrado. However, in the upper parts of the profile, the soil becomes increasingly sandy (quartz sands), which facilitates the penetration and percolation of water. In the uppermost parts of the brejo there is greater humidity, however, a soil that facilitates percolation can promote drought-like conditions, which is typical of edaphic cerrados.

The last two plots (9 and 10) deserve an integrated analysis because they are the same environment. These plots were installed at the top of the topographic gradient, in the municipality of Saloá, with altitudes of 1090 and 1052 m.a.s.l., respectively. The vegetation of these plots represent a true disjunction of the cerrado, with species typical of that biome, whose altitude added to high humidity, a mild climate and soil characteristics, provide conditions for such a patch of cerrado to establish itself in that environment. However, not only do cerrado species occur in this area, but elements of the caatinga are also present, forming an ecotonal environment. The terrestrial lichens of the Cladoniaceae family, which occur naturally on sandy soils of climatic and edaphic cerrado (PEREIRA et al., 2019), including in the Amazon, have also been recorded in this area (MOTA-FILHO et al., 2007). This evidence, together with new data from the NDVI analysis, reinforce the idea that this is an ecotonal area resulting from the conjunction of the action of physical factors on a topographic profile variation between a dry and semi-arid depression and a humid, elevated area in the interior of the semi-arid Northeast (SANTOS et al., 2014). As such, the only relevant difference between these two plots is that the value of the NDVI of plot 9 (0.27), is below what is expected for the environmental conditions of the area. However, this is explained by clearings present in the plot, exposing the sandy soil, making the value of this index decrease. For plot 10, the value of NDVI was 0.45, returning to the pattern of the area.

Regarding the classification of the Cerrado patch found at the top of the Brejo de Saloá, it is important to observe two parameters mentioned by Ribeiro and Walter (2008). The first one concerns the physiognomy, defined by the structure and dominant forms of growth. The other criterion concerns the physical factors of the environment and also the floristic composition. In this sense, the enclave of cerrado found in the study area fits in the physiognomy of a cerrado senso stricto, due to its structure, as well by its floristic composition because species such as Stigmaphyllon paralias, Tocoyena formosa, Tibouchina multiflora, Lantana camara, and L. canescens, among others, have been collected, which are part of the flora of the cerrado senso stricto, and have been cited in the literature (BASTOS; FERREIRA, 2010; MEDEIROS, 2011; MENDONÇA et al., 2008; RIBEIRO; WALTER, 2008).

The sector where the cerrado enclave is located, at the top of the topographical profile, is quite differentiated from the portion of the Sertaneja Depression, mainly regarding climatic conditions and the floristic composition of the vegetation cover. In this sense, in the lower portion of the topographic profile, species such as Poincianella pyramidalis, Piptadenia stipulacea, Jatropha molissima, among others typical of the caatinga biome, have been collected (PRADO, 2003).

The establishment of cerrado patches inserted in the semi-arid domain finds an explanation in the Refuge Theory (HAFFER, 1969). In this sense, Troppmair (2006) highlights that refuges are areas that can be forest or not, where a biota is established in relatively restricted environments, surrounded by different geo-environmental conditions, causing the impediment of its expansion.

According to the Refuge Theory, due to a series of glaciations that occurred in the Quaternary, the vegetative dynamics of Brazil went through phases of forest retraction and expansion, changing the position of the caatingas and cerrados (AB'SABER, 2005). Due to the alternation between dry and humid climates in the tropical regions, Conti and Furlan (2003) cite that the Brazilian geobotanical set is the result of the expansion and retraction of forests, cerrados and caatingas, triggered during periods of climate change.

In the transition between the Pleistocene and Holocene, 18 to 12 thousand years B.P., the climatic conditions made the caatingas occupy a much larger area than they do today, extending into the domains of the cerrado and Amazonian biomes. During this drier period of expansion of the caatinga, the areas where the cerrado vegetation is found became refuges; in some places, the caatinga has remained as strongholds (SILVA, 2011). Thus, the Refuge Theory explains the cerrado found at the top of the Brejo de Saloá.

To explain the occurrence of a typical vegetation from another phytogeographic province, Ab'Sáber (2005) used the principle of the enclave, or residual relict. In this context, Brejo de Saloá can be considered as one enclave, since the vegetation cover, considered as cerrado, is surrounded by another domain of a completely different nature and composition. This author also adds that for the establishment of these species in these locations, possible corridors would have existed at some time, probably at the geological moment when the climate was favorable for the expansion of such vegetation.

On the dynamics of expansion and retraction of the cerrado and caatinga during the Quaternary, Silva (2011) comments that, after the period of retraction of the caatinga, due to climate fluctuations and with a more humid climate, these conditions led the cerrado to expand. Thus, the author adds that the enclaves of cerrado found outside its domain (Central Brazil), are refuges, interpreted as testimonies of paleoenvironments.

Moreover, it is also noteworthy the occurrence of lichens from the Cladoniaceae family, which are common in coastal sandy areas, in the Amazonian cerrado (senso lato) and in the central Brazilian cerrado (senso stricto). In a study of the lichen species Cladonia verticillaris, Buril et al. (2010) made collections of this species in two areas of cerrado, in the coastal tablelands of Alhandra (Paraíba) and in the cerrado that occurs in the Brejo de Saloá (Pernambuco). The authors corroborated that this species occurs in phytophysiognomically similar areas, but climatically and geographically distinct ones, including by tracing its genetic profile. This fact corroborates our indications of the establishment of the cerrado patch in Saloá, and that in those climatic conditions, lichens can be a bioindicator of cerrado environments. In other regions of the planet, especially boreal ones, Cladoniaceae occur in acid soils of Pinus spp. forests (PIZNAK; BAKOR, 2019).

This study becomes more comprehensive and enlightening as the proposed spatial transect is analyzed by drawing a physiographic overview of the landscape transect. In this sense, it cover aspects of the relief units, soils, and the different vegetation coverages involved in this process of vegetation differentiation along the topographic profile. These features can be seen in Figure 1, which shows the different environmental aspects of the area studied. The georeferenced coordinates, used for the elaboration of the Landscape Profile of Figure 1, are shown in Table 1. The spatial distribution of the points (plots) along the municipalities is represented in Map 1.

Thus, through groupings of environmental characteristics, such as vegetation types, aridity index, soil types and relief units, it was possible to establish a schematic profile of how the different plant formations are distributed along the topographic gradient.

Figure 2 shows a mosaic with photographs of the plots along the whole landscape profile (plots 1 to 10).

Such vegetational formations vary from the typical caatinga of dry environments, to environments where physical conditions favor the development of montane forest, and culminating in the cerrado enclave at the top of the Brejo de Saloá.