2.5 Statistical procedures
Biomass of individuals was calculated based on the DGL using the equation: Biomass = 0.173 DGLcm2.295 (Amorim et al., 2005). We tested the normality of the data using the Shapiro Wilk test and homogeneity using the Bartlett’s test (Crawley, 2013). Plot values for richness, abundance, height, diameter, biomass and each physical and chemical soil property were averaged for each area. The areas were compared using Student’s T-tests and Pearson’s chi-squared test for parametric data and non-parametric data, respectively.
We performed Pearson’s correlation analyses to elucidate the relationships between vegetation and soil properties. Simple linear regressions were performed between the variables with the highest correlation, establishing the model which best described the results. The adult and regenerative strata were also correlated, as the adult vegetation acts as a vegetation cover and influences the regenerating stratum. We subsequently generated graphs only for the significantly correlated variables, using the plot function of the R statistical software program (R Core Team, 2018). Only the soil properties in the surface horizons were used in correlation analyses for the regenerating stratum. As a result, all horizons below the surface horizon were disregarded in the following analyses.
We performed a non-metric multidimensional scaling (NMDS) ranking to observe possible ordering in the species composition of regenerating stratum between the different areas using the metaMDS function of the vegan software package. In addition, the similarity between environments was analyzed using the ANOSIM test (Oksanen et al., 2018). All analyzes were performed using the R Studio 3.5.0. statistical software program (R Core Team, 2018).
Results
A total of 3,436 individuals were sampled, 2,519 of them in the adult stratum, distributed in 41 species and 17 families (Appendix 1), which represent the vegetation cover of our study. A total of 917 individuals were sampled in the regenerating stratum, with 29 species identified distributed in 15 families. The species Aspidosperma pyrifoliumMart (Apocynaceae) was the most frequent. The most frequent family was Euphorbiaceae (six species). Area I presented 526 individuals from the regenerating stratum, which were distributed in 11 species and six families. Area II presented 391 individuals, 27 species and 15 families. Two species were exclusive to Area I, while 18 species and nine families were exclusive to Area II.
The species were ecologically classified based on their distribution among the areas, with four categories being established: Aridity indicators, being dominant or exclusive species from the driest area (Area I); Sensitive species, being those found only in the most humid environments (Area II); Tolerant species, mainly distributed in the most humid environments, however, some individuals were also found in the drier area, confirming that the species tolerates such conditions; and finally, Common species, being those that are uniformly distributed between both areas (Appendix 1).
The NMDS ranking showed a low-stress value (stress = 0.06), and the similarity analysis showed significant differences between the areas regarding the composition and abundance of species present in the regenerating stratum (ANOSIM, r = 0.20; P=0.03) (Figure 2).
Note : Here figure 2 will be inserted
The mean comparison tests between Area I and II showed significant differences in the seedling diameter and diversity. Area I presented the largest average diameter of seedlings , and Area II had the highest average diversity of orders 1 (RegD1) and 2 (RegD2). The Herbivory Pressure Index in the plots ranged from 0.08 to 13.37, however there was no significant difference between the areas (Table 1).
Note : Here Table 1 will be inserted
Cambisol was the dominant soil group identified in the study area with six soil profiles. We also found three Luvisols, one Gleysol, one Planosol, and one Regosol. All soil properties have high standard deviation values in Area I and Area II, indicating high soil variability (Table 2). All A horizons were classified as ochric horizons. They varied between 2 and 10 cm in thickness. The A horizon in Area I and Area II is dominantly sandy clay loam and has base saturation above 80%. The Ca2+, Mg2+, K+ and Na+, and P contents were not significantly different between Areas I and II. Only the average C and N levels showed significant differences, with a higher average in Area II (Table 2).
Note : Here Table 2 will be inserted
The herbivory pressure index showed a weakly positive relationship with the diameter at ground level of saplings (DGL) (Figure 3b) and a negative relationship with the diversity of order 0 of the regenerating stratum (RegD0) (Figure 3c). RegD0 is also showed a positive relationship with biomass (Figure 3d), diversity (D0) (Figure 3e) and abundance of the adult stratum (Figure 3f). The DGL of the regenerating stratum showed a negative relationship with the abundance (Figure 3a) and biomass of the adult stratum. The abundance of seedlings also showed a negative relationship with the height of the adult stratum.
The diversity (RegD0) and biomass of the regenerating stratum also showed positive relationships with soil carbon and nitrogen levels (Figure 3g and Figure 3h). Despite showing relatively weak relationships, these linear regressions are all significant (Figure 3).
Note : Here figure 3 will be inserted
Discussion
Our results indicate significant differences in the floristic composition of regenerating stratum, diversity, DGL, and C and N levels in the soil between different vegetation cover areas. Species more resistant to water and anthropic stress and common in degraded areas, such as A. pyrifolium, C. pyramidales, and T. palmadora(Meiado, 2012; Souza et al., 2015a, 2015b), are dominant in Area I. On the other hand, Area II presents more than twice the number of species in the first area, and also shows 18 exclusive species. The Hill numbers D1 and D2 also showed higher mean richness in the area of denser vegetable coverage (Area II), indicating a higher number of common and dominant species in it, and consequently a more uniform distribution in the abundances of each species.
The abundance of species and individuals of Euphorbiaceae family can be attributed to their adaptive capacity, palatability, and potential for use by humans (Ribeiro et al., 2015, Rito et al., 2017). The dominance of A. pyrifolium , especially in Area I, is attributed to its high resistance to drought and because it is unpalatable to some herbivores, mainly goats and sheep. Such properties are conferred by its epicuticular wax, which in addition to ensuring greater efficiency in the use of water, has a composition rich in toxic compounds, thus presenting a competitive advantage over species more sensitive to semi-arid climates (Medeiros et al., 2017). Moreover, the anthropic pressure on this species is low since its wood has little utility value.T. palmadora , the second most abundant species in our study, is morphophysiologically adapted and resistant to water stress, unpalatable to animals due to its large amount of spines, and has a high reproductive rate. It can spread asexually through its branches and fruits even under limiting conditions (Meiado, 2012). C. pyramidale , the third most abundant specie in the study, is resistant to water deficits, high salt concentrations, and high temperatures, being widely found in degraded areas (Matias et al., 2018). Finally, species of genus Jatropha also have low wood density, making it of little use for human use. It is considered unpalatable to animals, and it has a high water reserve capacity, guaranteeing its survival and reproductive success (Ribeiro et al., 2015; Silva et al., 2004).
The absence of statistical differences in the content of exchangeable bases and texture between soils of Areas I and II can be attributed to: a) affinity of the soils with the parent material, and; b) relative homogeneity of sampled lithologies. The semi-arid climate in Northeast Brazil favors low chemical weathering of minerals and incipient pedogenesis (Araújo et al., 2017). These soils consequently express characteristics associated with their parent material, especially Cambisols and Neossols (FAO, 2015). Since gneiss is the main lithology of Cariri Paraibano (Lages et al., 2018), it is expected that the soils have similar characteristics. On the other hand, the statistical difference in the C and N levels between Areas I and II indicate environments with different inputs of plant residues to the soil and/or different preservation levels.
The environmental conditions provided by the greater abundance, biomass and diversity of adults observed in Area II favor the establishment of a more diverse regenerating stratum by: a) guaranteeing a larger seed bank in the community (Chazdon, 2014); b) attracting dispersing and pollinating agents (Guevara et al., 1986); c) facilitating seedling germination and establishment; and d) and providing greater protection to seedlings. Adult stratum areas with higher biomass protect the regenerating stratum against predation and trampling by herbivores (Derroire et al., 2016b; Marinho et al., 2016). Large herbivores prefer to forage in more open environments due to greater accessibility and to the palatability of the established herbaceous layer (Skarpe et al., 2007). Herbivore pressure negatively influenced species diversity, which was lower in more open areas (Area I). We suggest that the pressure of herbivores in Area I is favored by the ease of access and greater abundance of herbaceous plants present in the area, where their pressure can reduce species diversity.
In addition, microclimate conditions are milder under higher vegetation levels. The shade provided in environments with a higher number of trees reduces the transpiration rates of seedlings by decreasing their exposure to high light and temperature levels, as well as retaining greater moisture in the soil and providing a higher amount of organic waste to the soil (Derroire et al., 2016a; Lebrija-Trejos et al., 2011).
The positive correlations between the diversity of regenerative stratum (RegD0) with C and N levels in the soil indicate the importance of soil organic matter content in the ecological succession of areas. The supply of plant residues and their incorporation into the soil ensures nutrient cycling (Menezes et al., 2012; Sousa et al., 2012). As semi-arid soils are mostly eutrophic, the availability of nutrients derived from mineral weathering, such as Ca, Mg, and K, is not limiting for plant growth. On the other hand, C and N are primary macronutrients derived from the decomposition of soil organic matter (Six et al., 2004), therefore they are highly associated with soil organic matter content and quality (Ostrowska and Porębska, 2015). In addition, higher soil organic matter levels favor aggregate formation, reducing soil density and increasing porosity and water infiltration (Ferreira et al., 2018; Silva e Mendonça, 2007).
The negative correlation between the abundance of seedlings and the adult stratum height can be attributed to establishing a denser canopy in forest conditions than open areas. Areas with taller trees guarantee more favorable environmental conditions for plant growth (Lebrija-Trejos et al., 2011), providing greater balance in the community than the proliferation of more resistant species observed in more disturbed environments such as Area I (Rito et al., 2017). The higher number of exclusive species sensitive to degradation in Area II reinforces this hypothesis (Barbosa et al., 2007, Souza et al., 2015a).
The higher mean DGL of the regenerating stratum in the area of lower plant density (Area I) can be attributed to the decrease in competition for light (Mclaren and Mcdonald, 2003). However, despite the greater shading faced by seedlings in Area II, it is known that light is not one of the most limiting resources in the Caatinga (Sampaio, 2003). Thus, this factor alone does not fully explain our results. The DGL also showed a positive correlation with pressure by herbivory, and we speculate that these effects are indirect, and the greater secondary growth in Area I is related to the plant’s physiological adaptations to water stress to which it is subjected.
The growth of some plant species in arid and semi-arid environments is limited to the season of water availability, passing through growth pulses during the rainy season and dormancy during the driest months (García-Cervigón et al., 2017). Therefore, plants respond quickly to precipitation events because of this limitation, investing in wood and leaf production while water is available in soil. The irregular distribution of rainfall observed in the Brazilian semi-arid region with sporadic precipitation events in the dry months can stimulate several growth pulses during the year (Aragão et al., 2019). Furthermore, the distribution of photoassimilates by the plant varies according to the availability of resources, and root growth and secondary growth become the priority over primary growth when under limiting conditions (Mattos, 1999). Thus, the largest mean diameter observed in Area I can be attributed to such factors. However, complementary physiological studies are necessary to confirm this idea since these responses vary between different species (Aragão et al., 2019).
As expected, RegD0 was negatively related to pressure by herbivory. Herbivores generally cause adverse effects on plant communities, causing habitat degradation, limiting the growth of individuals, and affecting the taxonomic composition (Sfair et al., 2018; Souza et al., 2015a). The effects on the regenerative stratum are even more robust, since the animals have a preference for younger individuals due to their palatability, and they are even more susceptible to trampling mortality, compromising the diversity, structure, and maintenance of the plant community in the future (Marinho et al., 2016).
Our results indicate a more advanced secondary succession stage in Area II. In addition to the low diversity, Area I has many individuals identified as A. pyrilifolium , which suggests biotic homogenization due to anthropic disturbance. Diversity in degraded environments is reduced due to a decrease in species which are more sensitive to environmental changes and an increase in the abundance of the most resistant species (Ribeiro-Neto et al., 2016). The lower C and N levels in the soils of Area I also seem to be directly related to the anthropic disturbance. This scenario is the first step towards establishing the desertification process. These results are a consequence of the degradation regime imposed on the Caatinga since its colonization, including agricultural activities (Travassos and Souza, 2014). Nowadays, deforestation and inappropriate land use in agricultural areas compromise productivity, hampering natural regeneration in these areas (Leal et al., 2005; Marinho et al., 2016; Sousa et al., 2012; Sfair et al., 2018).
Considering the colonization and occupation processes of Cariri Paraibano, the impacts could mainly be related to livestock activities, especially goat farming, since it is a prevalent practice in the area, in which animals are raised freely to feed on native vegetation (Leal et al., 2005). This was evidenced in our study through the high biomass of fecal pellets found. This fact is pointed out as one of the main reasons for the vegetation cover loss in drylands and a strong driver of the desertification process (Marinho et al., 2016).
These results highlight the importance of adjusting the grazing practice of these areas to guarantee maintenance of plant communities and biological diversity. One of the most efficient land management practices is fallow, which isolates the land and allows it to recover through its resilience, promoting recovery of soil fertility, water storage capacity, and plant communities (Ferreira, et al., 2018). In the case of the Brazilian semi-arid region, this practice is not yet realistic for social and economic reasons. However, the development of public policies to raise awareness and incentives for more sustainable development is indicated. For example, better livestock management along with the creation of trapped herders avoids continuous removal of native vegetation andallows its restoration through natural regeneration (Leal et al., 2005). Another measure of fundamental importance would be creating Conservation Units of Integral Protection, which are still scarce in the Caatinga (Antongiovanni et al., 2020; MMA, 2011).
Conclusions
Our results confirm the effects of vegetation cover on the natural regeneration of plant communities in the Brazilian semi-arid region. The higher biomass and diversity of the adult stratum added to a higher C and N content in Area II soil led to the establishment of a higher number of species, with even greater equity.
The results indicate overgrazing as one of the main threats and drivers of the desertification process in the region, highlighting the importance of implementing management techniques in the use of these lands. Creating Conservation Units of Integral Protection and encouraging more sustainable development through environmental education are essential measures to be implemented.