Introduction

The transformation of conserved areas into anthropogenic areas is a growing problem in wet and dry areas of the world (Paulos 2001; Sá et al. 2004; McCray et al. 2005; Figuerôa et al. 2006; Mekuria et al. 2007; Rasingam and Parthasarathy 2009). The establishment of agriculture and livestock and the exploitation of firewood and lumber are common practices in the anthropogenic processes that occur in habitats (McLaren and McDonald 2003; Nyssen et al. 2004; Figuerôa et al. 2006; Nascimento et al. 2008). These practices led to a scarcity of resources and reduced land productive efficiency and affected the resilience of habitats (Girma 2001; Sá e Silva et al. 2008).

A reduction in the production capacity of the land prompts humans to search for other areas to maintain production. Consequently, native vegetation has become increasingly fragmented, resulting in the loss of natural habitats. Pronounced changes occur in landscapes that ultimately influence the dynamics of the lives of the local human population. In arid and semiarid regions, for example, the agropastoral activities developed over several centuries and the continued harvesting of lumber have led to soil erosion and the inability of the ecosystems to sustain life of any kind (Teckle 1999; Paulos 2001; Nyssen et al. 2004; Mekuria et al. 2007). In semiarid northeastern region from Brazil, there are some areas where the degradation processes are occurring so fast that the land has reached the level of desertification (Drumond et al. 2004).

The potential for the regeneration of anthropogenic regions has been reported for several areas (Sampaio et al. 1998; Negrero-Castilho and Hall 2000; Pereira et al. 2003; Luoga et al. 2004), but there are still too few studies to gain an understanding of the processes that enable the renewal of disrupted populations and communities. In general, studies have shown that anthropogenic plant communities show changes in the composition and richness of resident species, in the life forms present, and in the structure of its populations; the consequences of these changes can affect productive human activities (Sampaio et al. 1998; Pereira et al. 2003; Drumond et al. 2004; Albuquerque and Lucena 2005; Sá e Silva et al. 2008; Duarte et al. 2009).

One striking environmental feature in semiarid regions that influences the natural regeneration of conserved areas is the temporal variation in water availability (Knapp et al. 2002; Clary 2008). For example, in conserved dry forest habitats in Brazil, the interannual irregularity of the rainfall distribution leads to a reduced population size for many species, especially herbaceous species that represent more than 50 % of the flora in these habitats and form an extensive layer of protection for the soil (Reis et al. 2006; Araújo et al. 2007).

In anthropogenic habitats, the role that temporal variations have on the regeneration of plant populations is poorly understood. However, if the irregular distribution of rainfall affects the floristics and structure of the community in conserved areas, these same effects are expected to influence the resilience of the anthropogenic habitats that are undergoing a natural regeneration process. Anthropogenic activities are also expected to cause changes in the floristic composition and the community structure compared with conserved habitats. Therefore, this study developed the following hypotheses: (1) the richness, floristic composition, and herbaceous community structure vary year to year in response to interannual variations in total precipitation and (2) the life forms, richness, floristic composition, and herbaceous community structure of anthropogenic habitats are different from the herbaceous community in conserved habitats. Thus, the aims of the present study are the following: (1) describe and compare the interannual changes in the richness, species composition, diversity, and structure of the herbaceous component of an anthropogenic area and (2) identify similarities and differences that exist in the biological attributes of the anthropogenic community compared with the regeneration characteristics of a fragment of conserved habitat, as described by Reis et al. (2006).

Materials and methods

Description of the study area

The study was conducted at the Agronomic Institute of Pernambuco (Instituto Agronômico de Pernambuco—IPA) (8°14′ S and 35°55′ W, 537 m in altitude), located in the municipality of Caruaru, Pernambuco, Brazil. This station is within a rural zone that is 9 km from the city. The soil type for the region is classified as Eutrophic Yellow Podzolic and has some rocky outcrops. The area is drained by Olaria Creek (Pottery Creek), a tributary of the Ipojuca River (Alcoforado-Filho et al. 2003).

The regional climate is seasonal, with an average annual temperature of 22.5 °C that varies between 25 and 31 °C in the dry season and between 16 and 20 °C in the rainy season. The average annual rainfall is 694 mm, and the rainy season occurs mainly from March to August; a few months (generally May and June) have more than 100 mm of rainfall. The dry season occurs between September and February, and typical rainfall levels during this period are less than 30 mm per month. However, there may be occasional or erratic rainfall during the dry season and dry spells during the rainy season (Araújo 2005; Araújo et al. 2005a). During the study period, the annual precipitation was 686.6 mm in 2008 and 766.5 mm in 2009. During the study of Reis et al. (2006), the annual precipitation was 819.5 mm in 2002 and 448.8 mm in 2003 (Fig. 1).

Fig. 1
figure 1

Total monthly rainfall in the rainy and dry seasons in the years 2002, 2003, 2008, and 2009 in a semiarid region in northeastern Brazil. The precipitation values were provided by the Agronomic Institute of Pernambuco (IPA)

Full size image

The experimental station covers 190 ha, most of which is used for livestock and agricultural research activities. The natural vegetation of the entire area was previously occupied by a single patch of natural caatinga, which has now been reduced to approximately 20 ha (Alcoforado-Filho et al. 2003). For 50 years, this 20-ha fragment has been a conserved area, and the transit of domestic animals and the removal of vegetation are not permitted.

In 2002 and 2003, the floristics and structure of the herbaceous component of this conserved fragment were described by Reis et al. (2006). The woody and herbaceous flora of the conserved fragment is rich in Fabaceae, Malvaceae, Asteraceae, Convolvulaceae, Poaceae, and Euphorbiaceae species (Alcoforado-Filho et al. 2003; Araújo et al. 2005b). The herbaceous populations undergo changes in their structure and dynamics year to year due to the influence of climatic seasonality. Within the conserved fragment, there are different microhabitat conditions that also influence the establishment of herbs (Reis et al. 2006; Andrade et al. 2007; Santos et al. 2007; Silva et al. 2008).

Bordering this conserved fragment, there is an area measuring approximately 1 ha that was clearcut to establish a crop of giant cactus (Opuntia ficus-indica Mill.). This area was later abandoned and has been naturally regenerating for 15 years. In some parts of this anthropogenic area, there are already plants that have grown and developed canopies, providing a greater amount of shade for the ground. However, most parts do not have well-developed woody vegetation and are largely insolated during the dry season. These areas are also directly impacted by the rainfall during the rainy season. The woody plants form a discontinuous canopy that is approximately 5 m in height (Lopes et al. 2012), and the herbaceous vegetation is very dense, forming a covering on the ground during the rainy season.

Sampling of herbaceous plants

A total of 105 permanent plots were established in the anthropogenic area. The plots were distributed along seven equidistant 7-m transects and perpendicular to the conserved fragment of the caatinga vegetation. On each transect, 15 equidistant plots were placed every 9 m. The plot size was 1 × 1 m, which is the size often used in studies on the floristics and phytosociology of herbaceous plants (Behera and Misra 2006; Rasingam and Parthasarathy 2009; Kushwaha and Nandy 2012)

In two consecutive years (April to June 2008 and April to June 2009), each individual herb present within the plots was counted, and its height and stem/pseudostem diameter at the soil level were measured. Each plant with chlorophyll in its stem/pseudostem or with low levels of lignification but that were not woody seedlings was considered an herb. All plants were considered as individuals when they were not connected to another plant at ground level. The measurements of the height of the herbs and the diameter of the stems/pseudostems were taken with a ruler or tape measure and a digital caliper, respectively. For individuals with stems/pseudostems branching at ground level, individual measurements were taken for each branch. The height of the vines was estimated from the height of the supporting plant.

All of the nonflowering herbs in the plots were marked, and samples of flowering plants in the plots and/or of close individuals were collected. Herbs sampled just outside the plots were indicated as observed in the floristic list. Then, monthly visits were made to the sampled plots, and walkthroughs were conducted throughout the area to expand the collection effort and collect reproductive material of the species that were not flowering at the start of the sampling.

In the field, some seedlings of woody species were mistaken for herbaceous species due to morphological similarities at that stage. To correct this problem, individuals of these species were collected, planted in polyethylene bags containing soil from the site, transported to a greenhouse, and monitored until reproductive material was obtained. After proper identification, the seedlings of woody species were excluded from the analysis.

The species recorded in the present study were classified (therophytes, geophytes, or chamaeophytes) according to the Raunkiaer life-form classification system (Raunkiaer 1934) to determine whether the severe disruptions caused by humans influence the frequency of life forms in the herbaceous component.

In the conserved area, the floristic and phytosociology data of the herbaceous component of the study were compiled from Reis et al. (2006).

Data analysis

The plant material was preserved according to the usual techniques for preparing, drying, and mounting herbarium specimens (Mori et al. 1989). Plant identification was performed by comparing the plant to herbarium specimens deposited in the herbaria and by using taxonomic keys and descriptions in specialized literature. For species for which the identification was inconclusive, the material was sent to experts. The list of families, species, and phytosociological parameters recorded in this study and compiled in the study by Reis et al. (2006) was organized according to the Cronquist system (1988). The spelling of the species name was determined by consulting the Index Kewensis (www.ipni.org/ipni/plantnamesearchpage.do) and the Missouri Botanical Garden’s (MOBOT) VAST nomenclatural database (www.mobot.mobot.org/W3T/Search/vast.html). The abbreviations of the names of the authors of the species were performed by consulting Brumitt and Powell (1992) and MOBOT. After identification, the species were incorporated into the Professor Vasconcelos Sobrinho Herbarium collection (PEURF).

Matrices were constructed in the Excel program with data for the number of individuals, the height, and the diameter per plot, which generated a database that was imported into the FITOPAC program (Shepherd 1995) to describe the structure of the sampled community using the following parameters: density, dominance, frequency, and importance value (IV) index. The floristic similarity between years and between the anthropogenic and conserved areas (studied by Reis et al. 2006) was calculated using the Sørensen index (Krebs 1989).

In the anthropogenic areas, differences in the precipitation, richness, density, average height, and average diameter between years were determined using a Mann–Whitney variance test. Differences in the average density between the conserved and anthropogenic areas were determined using a single-sample t test. Differences in the total basal area and overall richness between areas were evaluated using a chi-squared test with Yates’ correction. The diversity of both areas during each year was determined using the Shannon diversity index (H′). The difference in diversity between areas and between years was assessed pairwise using a t test modified by Hutchinson (Zar 1996). The life form of the species is given as a percentage, considering the occurrence of each species in each area, independent of the year.

Results

Floristic composition and diversity

The herbaceous flora of the anthropogenic area was represented by 86 species that are distributed among 70 genera and 27 families, of which 17 species and four families were outside of the plots. In total, 84 species were recorded in the first year and 80 species in the second year (Table 1). There was no significant difference in the average richness between years in the anthropogenic area (Table 2). The conserved fragment described by Reis et al. (2006), which is located near the anthropogenic area, was represented by 71 species distributed into 63 genera and 35 families. In total, 55 species were found in the first year and 59 species in the second year. The total richness was significantly different between areas and between years, except when the richness of the conserved and anthropogenic areas was compared in the second year (Table 3).

Table 1 Families and species of herbaceous plants present in a natural regeneration area of caatinga in Pernambuco, Brazil, according to the value of importance of the families in 2008
Full size table
Table 2 Differences in the richness, density, height, and mean diameter (±SD) between years in an area of anthropogenic vegetation in a semiarid region in northeastern Brazil (degrees of freedom = 208)
Full size table
Table 3 Differences in the total species richness and total basal area (chi-squared test with Yates’ correction; degrees of freedom = 1), diversity (Hutchinson’s modified t test; v = degrees of freedom), and mean density (single-sample t test; degrees of freedom = 104) between conserved and anthropogenic areas (Reis et al. 2006) over 2 years in a semiarid region in northeastern Brazil
Full size table

In total, 123 species were listed, 53 of which only occurred in the anthropogenic area, 37 only in the conserved area, and 33 in both areas (Table 1). The floristic similarity between the two areas based on the Sorensen index was 60 %. The floristic similarity between years was 95 % in the anthropogenic area and 74 % in the conserved fragment.

Considering the temporal occurrence of species in the anthropogenic area, Pleurophora anomala, Spermacoce verticillata, Sida sp. 1, Sida sp. 2, Spananthe paniculata, and Momordica charantia occurred only in the first year, and Chamaesyce hyssopifolia and Convolvulaceae sp1 were present only in the second year. In the conserved area, the species exclusively encountered during the first year were Acalypha multicaulis, Aristolochia birostris, Begonia reniformis, Bomarea salsillioides, Cuphea prunellaefolia, Heliotropium angiospermum, Merremia aegyptia, Peperomia sp., Petalostelma sp., and Selaginella sulcata. Those encountered exclusively in the second year were Alternanthera brasiliana, Chaetocalyx longiflora, Cayaponia sp., Desmodium glabrum, Jacquemontia hirsuta, Malvastrum scabrum, Plumbago scandens, Polygala paniculata, Portulaca oleracea, Ruellia asperula, Sarcoglottis acaulis, Sidastrum multiflorum, Talinum paniculatum, Talinum triangulare, Tragia sp., and Tragia volubilis.

The Shannon–Wiener diversity index for the anthropogenic area was 1.91 natural digits per individuals (nats.ind−1) in 2008 and 1.94 nats.ind−1 in 2009. According to Reis et al. (2006), the diversity index of the conserved area was significantly higher in 2003 (3.01 nats.ind−1) than in 2002 (2.66 nats.ind−1). The diversity index was significantly different between years and between areas (Table 3).

In the anthropogenic area, 64 (74.5 %), 16 (18.5 %), and 6 (7 %) species were classified as therophytic, chamaephytic, and geophytic life forms, respectively. In the conserved area, 44 species were therophytes (62 %), 19 were chamaeophytes (26.5 %), and 8 were geophytes (11.5 %).

The following families had the greatest number of species in the anthropogenic area: Fabaceae (12 species); Asteraceae, Malvaceae, and Poaceae (ten species each); and Convolvulaceae and Euphorbiaceae (six species each). The other families were represented by one (the majority), two, or three species (Table 1). The following families were best represented in terms of the number of species in the conserved fragment (Reis et al. 2006): Malvaceae (eight species); Euphorbiaceae (seven species); Poaceae (six species); and Convolvulaceae, Fabaceae, and Portulacaceae (four species each).

Considering the two fragments studied, there were a total of 43 families recorded: 6 families were unique to the anthropogenic area, 14 were unique to the conserved area, and 23 were present in both areas (Table 1). The families that were recorded exclusively in the anthropogenic area were Apiaceae, Cucurbitaceae, Lamiaceae, Malpighiaceae, Rubiaceae, and Turneraceae. The families that were recorded exclusively in the conserved area were Alstroemeriaceae, Araceae, Aristolochiaceae, Asclepiadaceae, Bromeliaceae, Cucurbitaceae, Liliaceae, Loasaceae, Malpighiaceae, Moraceae, Nyctaginaceae, Piperaceae, Portulacaceae, and Selaginellaceae.

Density, basal area, and height

In 2008, there were 8,035 individuals per 105 m−2, ranging from 1 to 385 individuals per square meter (ind.m−2), with an average of 76 ind.m−2. In 2009, there were 9,284 ind.105 m−2, ranging from 1 to 451 ind.m−2, with an average of 87 ind.m−2. There was no significant difference in the mean density between years in the anthropogenic fragment (Table 2). In the conserved fragment, there was a significant reduction in the density from the first (4,039 ind.105 m−2) to the second year (826 ind.105 m−2) (Reis et al. 2006). When the densities between areas were compared, they were also significantly different (Table 3).

In the anthropogenic area, the total basal area was 1.67 m2.ha−1 in 2008 and 3.27 m2.ha−1 in 2009; the difference between years was significant (X 2 = 51,822, P < 0.05). In the conserved caatinga fragment, Reis et al. (2006) reported that the total basal area was 1.79 m2 ha−1 in the first year and 0.28 m2 ha−1 in the second year; the difference was also significant (X 2 = 110.15, P < 0.05). The total basal area was significantly different between the conserved and anthropogenic areas in the second year but not in the first year (Table 3).

The minimum, average, and maximum diameter values were 0.01, 1.3, and 2.00 cm, respectively, in 2008 and were 0.01, 1.1, and 1.56 cm, respectively, in 2009. There was no difference in the average diameters between years (Table 2). The largest diameters were recorded for a few individuals in populations of Urochloa maxima, Pappophoum pappipherum, Commelina obliqua, Stylosanthes scabra, Ruellia asperula, Ruellia bahiensis, and Heliotropium angiospermum.

The minimum, average, and maximum height values were 0.04, 13.7, and 230 cm, respectively, in 2008 and 0.5, 17.5, and 260 cm, respectively, in 2009. There was no significant difference in the average heights between years (Table 2). Poaceae had the tallest heights recorded in the area.

There was no significant difference in the average rainfall [Z(U) = 0.6351; p = 0.5254] between 2008 (57.2 mm) and 2009 (63.9 mm). Although the total precipitation has decreased 45 % from 2002 to 2003, there was no significant difference in the average rainfall [Z(U) = 1.3856; p = 0.1659] between 2002 (68.3 mm) and 2003 (37.4 mm)

Importance values of taxa

The families with the highest IVs were practically the same in the 2 years studied, only differing in the order of their importance. In 2008, the notable families in terms of IV were Asteraceae, Poaceae, Fabaceae, Malvaceae, and Commelinaceae, composing approximately 80 % of the total IV. In 2009, five families accounted for approximately 82 % of the total IV and were essentially the same families as in 2008, with the exception of Commelinaceae, which was replaced by Acanthaceae in the ranking of families of high ecological significance (Table 1).

The species of greatest importance in the first year were Delilia biflora, U. maxima, Desmodium glabrum, C. oblique, and Herissantia crispa, which together accounted for 63 % of the total IV. In the following year, there was little change, and the species with the greatest IVs were D. biflora, R. bahiensis, D. glabrum, U. maxima, and H. crispa, accounting for 66 % of the total IV.

In the 2 years analyzed, the high IV for D. biflora and D. glabrum was due to their elevated density and frequency, characterizing these species as large and well-distributed populations. The high IV for U. maxima, C obliqua, H. crispa, and R. bahiensis could be attributed to their dominance, indicating that these species denote the physiognomy of the area because they have larger stem diameters.

Of the species that did not belong to the group with the five largest IVs in 2008, Bidens bipinnata, Pappophorum pappiferum, R. bahiensis, H. angiospermum, Chaetocalyx sp., and Gomphrena vaga were also notable in the physiognomy of the community due to their high density and/or frequency. In 2009, P. pappiferum, Pylea hialina, B. bipinnata, Chaetocalyx sp., Blainvillea acmela, and Poinsettia heterophylla had a considerable density and/or frequency in the area, although they did not have high IVs.

Discussion

Studies conducted in different dry forest areas around the world have shown that variations in the total rainfall can influence the richness, floristic composition, and density of herbaceous populations (Ferguson et al. 2003; Price and Morgan 2007; Clary 2008). In Brazil, these findings were also reported by Reis et al. (2006), Silva et al. (2008), Feitoza et al. (2008), and Silva et al. (2009) in studies of the herbaceous caatinga communities.

In the anthropogenic area, the total rainfall was not significantly different between years, and consequently the richness, floristic composition, and species diversity were similar (Table 2), supporting the hypothesis presented herein. The diversity and richness values in the anthropogenic area are similar to the values reported for other semiarid environments (Behera and Misra 2006; Maracajá and Benevides 2006; Reis et al. 2006; Feitoza et al. 2008; Silva et al. 2009; Jia et al. 2011); the diversity was higher than in some conserved areas that have less total rainfall than that recorded in this study area (Maracajá and Benevides 2006; Feitoza et al. 2008).

Undoubtedly, in dry environments, climatic seasonality is considered a factor that shapes the ecosystem and population dynamics, given that there is a tendency of increasing densities of herbaceous plants as the total precipitation increases (Belsky 1990; Salo 2004; Volis et al. 2004; Araújo 2005; Reis et al. 2006; Lima et al. 2007; Price and Morgan 2007; Clary 2008; Feitoza et al. 2008; Silva et al. 2008). Notably, there are studies that did not find a significant positive correlation between the number of herbaceous individuals and annual precipitation values (Nippert et al. 2006). The height, diameter, and total basal area of the herbaceous community also tend to reflect changes in the total rainfall in conserved communities, which have lower values during drier years (Reis et al. 2006; Andrade et al. 2007; Lima et al. 2007; Santos et al. 2007; Feitoza et al. 2008; Silva et al. 2008, 2009). In the present study, there was not a significant increase in the precipitation between years, and consequently, the average density, height, and diameter of the plants were similar between years, supporting the hypothesis of the same ones.

Although no interannual differences were recorded for species diversity in the anthropogenic area, there were significant differences in the diversity and density of herbaceous species between the anthropogenic and conserved areas (Table 2). The lower diversity recorded in the anthropogenic area may be explained by the large number of individuals representing only a few species, which is a reflection of the low equability compared with the conserved fragment. The density of the anthropogenic community was approximately four times that of the conserved community due to the prominent density of some species (Table 1) and may be hindering the recolonization by native species because they have a greater competitive capability (Pivello et al. 1999; Yayneshet et al. 2009), which should be evaluated further in future studies. These findings indicate that the anthropogenic fragment has not yet developed the specific microclimatic conditions that support the same level of diversity as in the protected areas (Clark and McLachlan 2003).

Notably, the family Poaceae had a remarkable density in the anthropogenic area, contributing largely to the significant difference between the areas. In the plots in which species of this family dominated, individuals of other species were rarely observed. This same behavior was also recorded in dry environments that are in the early stages of regeneration because Poaceae are common in most anthropogenic areas and its establishment tends to inhibit the germination of other species until tree species develop and provide shade over them (Ferguson 2001; Francis and Parrotta 2006; Yayneshet et al. 2009). Francis and Parrotta (2006) also reported that the biomass of other herb species was inversely proportional to the presence of grass species, indicating a strong competition between graminoid and broadleaved herbs. These results also indicate that the established competition by grasses may be possibly the primary barrier to the regeneration of native species in dry regions.

In both the regenerating and conserved fragments, the greater part of the herbaceous assembly was formed by therophytes (Table 1), indicating that the seed-based survival strategy is similar between areas, independent of the conservation level of the habitat (Reis et al. 2006; Silva et al. 2009). In unfavorable seasons (dry season), herbaceous individuals die, but the species persist in the area in the form of seeds that germinate when abiotic conditions are favorable (Volis et al. 2004; Costa et al. 2007; Santos et al. 2007; Silva et al. 2008, 2009; Feitoza et al. 2008). The present study also found that there is a tendency to reduce the number of therophyte species and increase the number of chamaephyte species as the season progresses.

In the anthropogenic area, the IVs of the populations were similar between years, and the position of only some species changed, suggesting that there is little change in the efficiency of the exploitation of resources provided by the habitat (Parthasarathy and Karthikeyan 1997). D. biflora was the species with the highest IV in both years, but this species was not among the ten primary species with high IVs in the conserved area in any of the years studied by Reis et al. (2006).

According to Behera and Misra (2006), in dry environments, the composition of herbaceous species of anthropogenic habitat is influenced by the regeneration time (age of the young forest) However, those authors compared areas very distant from each other such that it was unclear whether the floristic composition of these areas it was similar prior to human intervention.

In the present study, the sites that were compared were less than 3 m from each other and had a single floristic assembly prior to human intervention. After 15 years of regeneration, only 43 % of the floristic assembly returned and now occupies the anthropogenic area. Thus, 15 years is likely not sufficient to establish the conditions required for the 37 herbaceous species recorded by Reis et al. (2006) in the conserved area to become reestablished (Table 1).

Species that arrive to the regeneration area are possibly more tolerant to the extreme insolation conditions, initially fulfilling the role of soil cover and protecting the soil from high temperatures. Moreover, these species that are tolerant to the insolation conditions also protect the soil against the direct impact of rainfall, which can disaggregate particles, strip nutrients, promote erosion, and impair the germination of less tolerant species (Forbis et al. 2004; Noel et al. 2006; Nippert et al. 2006; Mekuria et al. 2007).

Interestingly, many of the herbaceous species of the caatinga in conserved areas are transient; these species may be present during 1 year and absent during others and may have cyclic birth rates (Araújo and Ferraz 2003; Araújo et al. 2005b; Reis et al. 2006). In the anthropogenic area, only six species (S. verticillata, S. paniculata, M. charantia, C. hyssopifolia, Sida sp. 1, and Sida sp. 2) showed an irregular occurrence between the years. Therefore, following the regeneration process of anthropogenic areas, a longer period of time describing the average trends in the cyclic demographic variations of these species is important because the regeneration time may not be the only significant factor for the restoration of the herbaceous component. The annual abiotic characteristics (particularly climate characteristics) that influence the germination, survival, and productivity of herbaceous species in semiarid environments may also play a role.

The conserved fragment had 14 families that did not occur in the anthropogenic fragment, although most of these families have been reported in the herbaceous component of other conserved areas (Feitoza et al. 2008; Silva et al. 2009). This finding suggests that some families of the herbaceous component can be used as indicators of the state of the conserved area or as susceptible to the disappearance of woody vegetation.

The heterogeneity of the microhabitat conditions may also induce variation in the species composition of an area. For example, the families exclusively found in the conserved fragment, with the exception of Piperaceae and Portulacaceae, were specific to the shaded areas or to areas with a smaller temperature range (Reis et al. 2006). The absence of these families in anthropogenic areas may be a strong indicator that this area has not yet developed the microhabitat conditions similar to the caatinga conditions in the conserved area. Therefore, this area still needs more time under protection to restore its entire herbaceous community.

Finally, the results of the present study and information found in the literature led us to propose the following hypothetical model of the natural regeneration of herbaceous communities in semiarid regions of northeastern Brazil. In recently abandoned agricultural areas, species richness is zero or very near to zero. During the early years, the colonizing/facilitating herbaceous species functional group, which comprises species that are unique to areas with recently abandoned agricultural enterprises and that have the ability to inhabit extreme environments (e.g., direct sunlight, high temperatures, low humidity), occupies the space. This occupation is important for the following reasons: (1) it promotes a marked increase in the richness and density of herbaceous plants, (2) it confers some degree of coverage for the soil and consequently prevents erosion due to the direct impact of rain, (3) it traps woody plant seeds present in the top soil or deposited by seed rain, (4) it keeps the soil warm and moist and may function as a germination chamber for woody species, and (5) it can facilitate the establishment and survival of woody species seedlings.

Subsequently, with the development of shrub and tree vegetation, an increase in the basal area promotes a greater occupation of horizontal space and the development of a canopy that provides more shade. This progression promotes local microclimatic changes and consequently reduces the density of the herbaceous community, leading to the conditions needed to establish a functional group of early secondary herbaceous species. This group would comprise species present in the climax community or very close to the climax community, which also have the ability to colonize areas with at least 15 years of natural regeneration. With the evolution of this succession, the herbaceous community will again be represented by a functional group of late secondary herbaceous species that are sensitive to the microclimatic changes caused by human actions and are, therefore, exclusively found in the climax community or very close to the climax community.

Conclusion

In consecutive years with similar amounts of total rainfall, there were no differences recorded in the species richness, floristic composition, and structure of the herbaceous community. Despite the elapsed regeneration time (15 years), there were still few floristic and structural similarities between the herbaceous community in the conserved fragment and the community in the anthropogenic area, indicating that the latter has not yet developed the specific microclimate conditions that support the same level of diversity observed in the protected areas. Anthropogenic areas require longer protection times to achieve the same level of diversity, abundance, and composition of the species and families in the herbaceous strata as conserved fragments of caatinga.