NATIVE CAATINGA SPECIES FOR THE RECOVERY OF DEGRADED AREAS IN THE BRAZILIAN SEMIARID REGION

This study aimed to prospect, among the species that grow spontaneously in compacted landfi lls, native Caatinga plants with potential for ground cover in extremely impacted areas, with exposed soil. Initially, a general fl oristic survey was carried out in the four study areas. To prospect the species, data referring to the richness, coverage, and densifi cation of plants in the herbaceous stratum were collected in each study area, using the method of plots. The selection was based on species-specifi c characteristics: origin, plant habit, life cycle, propagation, dispersion syndrome, coverage, densifi cation, and allelopathic eff ect. The general fl oristic inventory revealed the presence of 73 species belonging to 63 genera and 26 botanical families. In the survey of the coverage and densifi cation of the herbaceous stratum, 33 species belonging to 32 genera and 16 families were found, being the most representative: Fabaceae (5), Malvaceae (5), and Poaceae (5). As for the origin, 26 are native, one is naturalized, and six are exotic, of which 66.6% are in the Poaceae family. Moreover, most of these species are herbs, with an annual life cycle, dissemination through seeds, and present autochoric dispersion. The coverage and densifi cation of these species ranged from 0.44% to 9.5% of the area and 1 to 4.44 of individuals/m, respectively. The species Senna unifl ora, Rhaphiodon echinus, Sida galheirensis, Mesosphaerum suaveolens, Hexasepalum teres, Waltheria rotundifolia, Trianthema portulacastrum, and Herissantia crispa showed potential for use in recovery plans for degraded areas based on the results presented by each of them in the parameters analyzed, especially in coverage, densifi cation, dispersion syndrome, and life cycle.


INTRODUCTION
Caatinga corresponds to a phytogeographic domain known as CPD -Caatinga Phytogeographic Domain (Moro et al. 2014), which harbors the most vulnerable ecological systems to climate change, is also poorly studied and protected . It is the largest, most continuous, and most diverse semiarid vegetation in the New World (Fernandes and Queiroz, 2018). Occupying 912,529 km 2 in Northeastern Brazil, the Caatinga is best described as a rich and complex socio-ecological system that houses a unique natural and cultural heritage of global importance Fernandes and Queiroz, 2018).
This biome presents fl oristic, physiognomic, and ecologically distinct plant formations, with the steppe savanna being the predominant vegetation (Moro et al., 2014). This signifi cant fl oristic heterogeneity refl ects fl ora adaptations to local climate and soil conditions in the semiarid region. Plants have developed skills to survive adverse conditions that include irregular rainfall and recurrent droughts, with the most striking feature being deciduousness (Fernandes and Queiroz, 2018).
However, the Caatinga has been undergoing phytophysiognomic and structural changes related to anthropogenic processes that date back to the time of colonization in Brazil (Bezerra et al., 2014). The result is that at least 63.6% of the Caatinga was modifi ed by human activities . The systematic elimination of vegetation cover and the inadequate and unsustainable use of the soil has caused severe environmental problems for the Caatinga, located almost entirely in the Brazilian semiarid region, reducing biodiversity, soil degradation, and desertifi cation stand out (Bezerra et al., 2014). Around 94% of the Caatinga are at moderate to high risk of desertifi cation, which can be further intensifi ed due to the biome's climate vulnerability .
In the last few decades, the Northeast region of Brazil has received and has been receiving investments in infrastructure that have been improving the quality of life of rural people. Irrigation systems revolutionized agriculture in the lower middle São Francisco, generating employment and income in important urban centers, such as Petrolina, in the state of Pernambuco, and Juazeiro, in the state of Bahia. Obviously, pressure on remnants of Caatinga has also increased, and areas formerly with extensive native vegetation have given way to large grape and mango farms. More recently, PISF, the São Francisco Integration Project, has been taking water to Brazil's aridest areas through the country's largest infrastructure project, the transposition of the São Francisco River.
The present work was carried out in these areas directly aff ected by the PISF, which correspond to severe anthropic interventions (with heavy machinery). In other words, after the machines pass, the landscape is no longer the same. Therefore, the main objective is to prospect species that are capable of colonizing exposed, poor, and overturned soil, sometimes even areas of waste disposal. The objective of the action is to move from a scenario of exposed soil (without vegetation) to one of soil with vegetation cover, being, therefore, a recovery action and not a restoration action.
Thus, studies and recovery plans for these areas are essential for mitigating environmental impacts. The selection of species for use as vegetation cover in anthropic areas should be based on edaphic, climatic, and environmental factors and the plants' characteristics (Gris et al., 2012). Native species are the most suitable for covering areas without vegetation since, in addition to ensuring the preservation of the autochthonous gene bank, they make the environment closer to the original one and more ecologically balanced (Sousa and Sutili, 2017).
Then, this study aimed to prospect, among the species that grow spontaneously in compacted landfi lls, native plants from the Caatinga with potential for covering the soil in highly impacted areas and/or exposed soil in the Brazilian semiarid region.

Study area
The study was carried out on slopes of compacted landfi lls distributed along the channels of the north and east axes, in the states of PE and CE, in the PISF work area (Integration Project of the São Francisco River). Four areas with older slopes were selected. Area I is located on the east axis and corresponds to the dark line on the right in the image, and areas II, III, IV are located on the north axis, the dark line on the left. (Figure 1).
Information about the four sampled areas is contained in table 1.
It is important to emphasize that the species surveyed were found on construction structures composed of compacted landfi lls with diff erent physical and chemical structures from the soils present in the surrounding matrix.

Floristic survey
Initially, a fl oristic survey was carried out on all slopes of the study areas to verify their general fl oristic composition. The survey took place during November (dry season) of 2014 and February and March (rainy season) of 2015, with the identifi cations based on prior knowledge of the taxa, collecting botanical material only from plants found in the phenological stage reproductive system (Fidalgo and Bononi, 1989). The exsiccates of the collected material were deposited in the PISF botanical collection. Taxonomic identifi cation was carried out through consultations with specialists and the literature. The classifi cation of species followed APG IV (2016).

Coverage survey and densifi cation of the herbaceous layer
Weeds (including creeping herbs) and sub-shrubs were considered as part of the herbaceous strata (Oliveira et al., 2017). The plots method was used to collect data regarding the coverage and densifi cation of plants in the herbaceous stratum. Sampling was carried out in the four areas (I, II, III, and IV), in a stretch of 5 km contiguous, with slopes totaling 20 km in length. At the beginning of each kilometer, a point was georeferenced at the base of the slope, and from this point, another fi ve points 10 m apart from each other were also georeferenced, forming a line of 50 m. At each point of the line, a transect was established perpendicularly to the top of the slope. In each transect, plots of 1 m ² (1 × 1 m) were placed, spaced 1 m apart from each other, and placed alternately on the right and left of the transect. Thus, the number of plots in each area varied depending on the height of the slopes. The number of plots in areas I, II, III, and IV were 94, 102, 134, and 51, respectively, totaling a sampled area of 381 m ² .
In each 1 m ² plot, all individuals belonging to the herbaceous stratum were sampled, and the following data were collected: densifi cation and percentage of coverage of each species. Densifi cation was measured by counting the number of individuals of each species that occur in 1 m ² , considering only the plots where the species was present. Some species have underground stems that allow an extensive vegetative propagation, causing the aerial part to appear cespitose, making it diffi cult to measure and defi ne what an individual is. Thus, any plant that did not show a connection with another at ground level was considered as a single individual, justifying the use of the term densifi cation and not density.
The coverage percentage was performed through the visual estimation, supported by a 1 × 1 m plot, divided with nylon threads into 100 smaller squares, each representing 1% of the area. In order to make the visual estimate of the percentage of species coverage more accurate and due to the diffi culties of sampling some of them because of their morphology and/or form of growth, we established the minimum coverage percentage of 1 percent for the herbaceous individuals. This approach also supports the adoption of the term densifi cation instead of density.

Attributes for prospecting species
In this study, the following species-specifi c attributes were considered: origin, plant habit, life cycle, propagation, dispersion syndrome, coverage, densifi cation, and allelopathic eff ect.
Species sampled were classifi ed according to origin (native or exotic) and habit (grass or subshrub) (Andrade et al., 2020). The classifi cation regarding life cycle (annual or perennial) and propagation (seed or seed/stolons) was based on Lorenzi (2014) and Silva et al. (2012). As regards the dispersion syndrome, the classifi cation was based on Van der Pijl (1982), which is grouped into three basic groups: zoochoric (dispersed by animals), anemochoric (dispersed by the wind), and autochoric (self-dispersed).
The coverage and densifi cation values were calculated from fi eld survey data, obtaining the average of the coverage percentage and the number of individuals of each species in 1 m ² , respectively. For the calculations of both coverage and densifi cation of a species, only the plots in which the species was present were considered.
The allelopathic eff ect was evaluated according to Grisi et al. (2013), and extracts from the root, stem, and leaves were used to evaluate the germination of Lettuce (Lactuca sativa L.), a species commonly used as a test plant in allelopathy studies. The species whose extracts interfered with the germination of L. sativa were considered as low, medium, high, and totally allelopathic, resulting in germination percentages greater than 80%, between 80 and 60%, less than 60% and 0%, respectively. The fi nal score for each species was obtained from the average of leaf, stem, and root extracts.
The score for each native species, since exotic species are not the focus of this study, resulted from the sum of the products (weight × score) of the attributes of the plant habit, propagation, life cycle, dispersion syndrome, coverage, and densifi cation. The allelopathic eff ect was evaluated for species with the highest score concerning the attributes mentioned above, and these allelopathic scores were added to the previously obtained (Table 2).
Therefore, a list of species with potential for land cover in extremely impacted areas was obtained.

Floristic survey
The general fl oristic survey, including all forms of life and origin, carried out in the four slope areas (I, II, III, and IV), indicated 73 species belonging to 63 genera and 26 botanical families. The most representative families, that is, with the highest number of species were Fabaceae (13 species), Poaceae (9 species), Malvaceae (8 species), Euphorbiaceae (6 species), and Apocynaceae (4 species), corresponding to 54.79% of the total species surveyed in this study (Table 3).
In addition to species from the herbaceous stratum (grass and sub-shrubs), shrub and creeping phanerogams were found, as well as woody species in the form of seedlings, probably arising from natural regeneration and colonization of the slopes.

Coverage survey and densifi cation of the herbaceous layer
In the survey of densifi cation and coverage of plants belonging to the herbaceous stratum, 33 species were found, distributed in 32 genera and 16 families. The families with the highest species richness were Fabaceae (5), Malvaceae (5), and Poaceae (5) ( Table 4).
Of the 33 species sampled, 26 were considered natives, one naturalized, and six exotics, with 66.6% belonging to the Poaceae family. Thus, even though the Poaceae family is among the most representative of the herbaceous stratum, its species were of little importance because they are mostly exotic, thus opposing the focus of the study, which are native species. These species are African in origin but already well-established in the Caatinga, with a few already being classifi ed as naturalized by some authors. Such exotic species have a high power of infestation since they reproduce by caryopsis-type fruits (epizoochory or anemochory) and have rhizomes or other underground stems that help vegetative propagation. Therefore, they present a high degree of biological invasion. However, the diversity (richness) found, apparently induces competition and establishment, ending up biologically controlling the infestation.
In terms of plant habit, 21 species were classifi ed as grasses (the creepers included here), nine as subshrubs, and three as grass/sub-shrubs. Regarding the life cycle, 17 species are annual, 13 are perennial, and three can be perennial/annual depending on the environment since herbaceous plants in the Caatinga can present variable growth patterns depending on water availability (Table 4).
Concerning propagation, all species are disseminated through seeds (Table 4). However, two of these species (Commelina erecta and Dactyloctenium aegyptium), in addition to propagating through seeds, can also reproduce through stolons (thin aerial stems that have horizontal growth, originating new plants). In addition, most species (63.6%) have autochoric dispersion, that is, through mechanisms of the plant itself, and 36.4% need external agents to disperse, whether wind (anemochoric) or animals ( zoochoric) ( Table 4).
The coverage area ratio per individual ranged between 0.44% and 9.57% (Table 4). The species that presented the lowest cover percentages were Tridax procumbens, Spigelia anthelmia, Oxalis glaucescens, and Hexasepalum teres, while the highest percentages of cover were Trianthema portulacastrum (9.57%) and Senna unifl ora (8.18%). Regarding densifi cation,  of endemic species (Córdula et al., 2014). The predominance of Fabaceae may be associated with the fact that these species have diff erent survival strategies in xeric environments, such as root nodules, compound pinnate leaves, and pulvinus, in addition to extrafl oral nectaries that help in the transpiration and the protection of the plant in general. (Córdula et al., 2014). Probably species belonging to this family are more resistant to the stressful conditions of the environment and, therefore, more adapted, a fact that is justifi ed by the signifi cant irradiation of these families in the Quaternary period (Klein, 1975). In addition, Senna unifl ora (Fabaceae) proved to be the most important species in this study, with the best values in the evaluated attributes, especially in coverage, densifi cation, and dispersion syndrome. This species has as its main characteristic the ability to fi x atmospheric nitrogen through a symbiotic relationship with diazotrophic (nitrogen-fi xing) bacteria (Neves et al., 2017). Furthermore, they present rapid growth and regrowth capacity, making them suitable for recovering degraded areas (Neves et al., 2017). Therefore, Pereira and Rodrigues (2012) highlight that Fabaceae play an important role in the dynamics of ecosystems, presenting enormous potential for recovery and, therefore, they are systematically used in environmental projects.
Considering the objective of the work, we have a universe of 33 species, that is, those found on the slopes. It is imperative to highlight the high percentage of native species (78.9%) compared to exotic (18.9%). As previously mentioned, most of these exotics belong to the Poaceae (Graminae) family and are originally from Africa, being quite aggressive in the colonization processes of exposed soil. According to Carpanezzi (1998), the rehabilitation of local biodiversity is an expected consequence of actions to recover degraded areas, so the selected species should preferably be native to the recovering environment, though some exotic species, such as the grasses mentioned above, also can be benefi cial in primary succession and colonization of exposed soil.
The high fl oristic richness of native species adds value to the results since little is known about their potential in succession processes in the Caatinga (Moro et al., 2014). They attract fauna with their fl owers, fruits, and extrafl oral nectaries, in addition to being a shelter for the rest of the fauna that does the species that have more individuals inhabiting the same m ² , therefore greater aggregation capacity, are: Dactyloctenium aegyptium (4.44), Mesosphaerum suaveolens (4.41), Tridax procumbens (4.17), and Hexasepalum teres (4.00). The lowest densifi cation values (1.00) were from Commelina erecta, Astraea surinamensis, Euphorbia adenoptera, and Zornia sericea. Except for Tridax procumbens, all the others are native, showing the importance of intercropping between plants of diff erent origins in the colonization of the slopes.
Assessing the allelopathic eff ect of the leaf, stem, and root of the selected species (Table 4), we found that only Senna unifl ora had low allelopathy in the leaf, stem, and root, with 85%, 85%, and 94% of germination of the tested plant (Lettuce), respectively. On the other hand, Trianthema portulacastrum proved to be a completely allelopathic species for leaf and stem, both with 0% germination and low allelopathy for root (83% germination).

DISCUSSION
Considering fl oristic studies with smaller life forms (habits), 73 species is a relevant richness value. Nevertheless, 33 species in slope areas with high human-modifi ed impacts, poor, shallow, and stony soil are also signifi cant. In a study compiling fl oristic and phytosociological literature on plant diversity in the Caatinga Phytogeographic Domain (CPD), Moro et al. (2014) found that the families Fabaceae, Euphorbiaceae, Malvaceae, Asteraceae, Convolvulaceae, Poaceae, Bignoniaceae, Cyperaceae, Rubiaceae, and Apocynaceae are the ten families with the most outstanding richness in the domain.
The Fabaceae family is the best represented in the Caatinga and has the most signifi cant number not use them for food (Araújo and Silva, 2017). The sum of these attributes accelerates the biological interactions and, with that, there is the advancement of succession (Rodrigues et al., 2009). Furthermore, in the natural succession process, pioneer species are generally ruderal herbaceous plants, which enrich the soil, allowing the establishment of late species (Guariguata and Ostertag, 2001).
Additionally, among the 33 species, eight were selected mainly based on coverage and densifi cation values. Namely, they are: Senna unifl ora, Rhaphiodon echinus, Sida galheirensis, Mesosphaerum suaveolens, Hexasepalum teres, Waltheria rotundifolia, Trianthema portulacastrum, and Herissantia crispa. Sida galheirensis and Waltheria rotundifolia are two Malvaceae species that are very common shrubs in the Caatinga, presenting a high degree of cespitose, therefore, with signifi cant coverage and densifi cation in communities where they are dominant. Rhaphiodon echinus (Lamiaceae) and Herissantia crispa (Malvaceae) are creeping plants, that is, herbaceous vines devoid of gripping structures. Consequently, they have a high coverage value and are also widely distributed in the CPD. On the other hand, Mesosphaerum suaveolens (Lamiceae) and Hexasepalum teres (Rubiaceae) are shrubby-herbaceous plants with low tillering with ample dispersive capacity, since their seeds are small and have autochoric dispersion. Moreover, the dispersion syndrome is the attribute that comes next in the sequence of relevance concerning the prospection objective of this study.
Species with diff erent forms of dispersal syndrome are important for recovery and degraded areas. The advantage of autochory is that there is no need for dispersing agents (Van der Pijl, 1982). Anemochory is considered a specialization, that is, adaptation to unfavorable environments, in addition to being important in open environments (Fenner and Thompson, 2005). On the other hand, zoochory has the advantages of distancing the seeds from the surroundings of the mother plant and the colonization of gaps in degraded areas (Araújo and Silva, 2017). However, in areas with severe levels of degradation, such as those commonly found in PISF areas, the presence of fauna is modest. Therefore, most of the 33 species (63.6%) found to have autochoric dispersion is a result that cannot be neglected. Indeed, the non-dependence of biological actors in seed dispersal is vital in degraded places and distant from islands of diversity, such as the present case.
All eight species selected in this study are considered spontaneous and pioneer plants presenting large seed production as a survival mechanism, with effi cient dispersion and longevity (Lorenzi, 2014). According to Rodrigues et al. (2009), the use of fast-growing pioneer species is fundamental in recovery programs, as it facilitates the beginning of the successional dynamic, mainly due to shading and nutrient incorporation, primarily covering the soil, incorporating organic matter, and inhibiting the development of exotic species invasive (Alves et al., 2018). In addition, ruderal herbaceous-shrub plants are more aggressive and pioneer, making them suitable for the base of the secondary succession process (Holanda et al., 2015).
The eight species selected are eudicots, which have axial or pivoting roots, which are more profound, and lateral ramifi cations that can be more or less developed depending on the species (Judd et al., 2015). This type of root system can extract nutrients found in deeper soil layers, which will be made available after decomposition and incorporation of the plant into the soil (Favero et al., 2000).
The greater the coverage and densifi cation, the greater the biomass production. In this sense, the highest percentages of coverage were of Trianthema portulacastrum (9.57%) and Senna unifl ora (8.18%). Concerning densifi cation, the species that have more individuals inhabiting the same square meter, therefore greater aggregation capacity, are: Dactyloctenium aegyptium (4.44), Mesosphaerum suaveolens (4.41), Tridax procumbens (4.17), and Hexasepalum teres (4.00). The only exotic one is Dactyloctenium aegyptium, and the others are all common and widely distributed species within the CPD, that is, with easy propagation management and construction of germplasm banks. Even so, only Senna unifl ora was already studied from the ecological perspective of conservation. Favero et al. (2000) and Ferreira et al. (2013) concluded that Senna unifl ora could promote the same eff ects of land cover, biomass production, and nutrient cycling as species introduced or cultivated for green manure, in addition to having forage potential. Carrying out herbaceous vegetation cover in exposed soil with potentially suitable species tends to accelerate the process of natural regeneration and establishment of secondary and late species. This occurs because, despite the occurrence of natural succession processes in degraded environments, where herbaceous, shrub and arboreal species are gradually added to the community, in space and in time (Coe and Souza, 2014), the effi ciency of this process depends on several factors, such as the availability of propagules in the soil, the coverage capacity of pioneer species and the level of impact on the soil (Coe and Souza, 2014).
In this perspective, two species potentially suitable for vegetation cover in degraded areas, Senna unifl ora and Rhaphiodon echinus, prospected in this study, are being tested in soil vegetation cover methodologies in the PISF Degraded Area Recovery Plans, the fundamental scope of the Basic Environmental Plan (PBA) 09. PBA is a condition required by the Brazilian Institute of Environment and Renewable Natural Resources (IBAMA) to the Ministry of Regional Development for the environmental licensing of the PISF's operating license. Rhaphiodon echinus became the target of this action for being easily accessible concerning seed collection and vegetative propagation compared to the others.
The exotic species found in this study can be classifi ed, considering the family, in Poaceae and other families. Only two non-Poaceae species were recorded, Amaranthus viridis (Amaranthaceae) and Boerhavia diff usa (Nyctaginaceae), both with low to moderate degree of invasion, because they do not have underground structures for vegetative propagation and present autochory and epizoochory dispersal, respectively. However, for the Poaceae family, the scenario is diff erent. They are all species with a high degree of invasion due to their aforementioned characteristics. Nonetheless, in the study area, they apparently do not threaten the community, since the richness found was signifi cant (33 species).
Although studies of prospecting species for recovery, mainly in the Caatinga, are pretty scarce, what is evidenced in this work is that there are native species suitable for use in methodologies of Degraded-Area Recovery Projects (PRADs).

CONCLUSION
The species Senna unifl ora, Rhaphiodon echinus, Sida galheirensis, Mesosphaerum suaveolens, Hexasepalum teres, Waltheria rotundifolia, Trianthema portulacastrum and Herissantia crispa have potential for use as vegetation cover for exposed soil in degraded areas. Through their speciesspecifi c attributes, such species showed that they have interesting characteristics of high propagation, coverage and densifi cation. Planting these species can also initiate the natural regeneration processes of the impacted area, either through facilitated dispersion, nitrogen fi xation and/or soil protection.

AUTHOR CONTRIBUTIONS
The manuscript is part of the master's thesis of the fi rst author. R.G.R, M.Z.B.C and D.S.P contributed to the study conception and design. The data collection was performed by J.N.C and A.P.F and the analysis by J.N.C, R.G.R, A.P.F and D.S.P. The fi rst draft of the manuscript was written by J.N.C and all authors commented on previous versions of the manuscript and gave fi nal approval for publication. R.G.R and D.S.P secured funding and logistic support.