Abstracts

ARTICLES

Natural forest regeneration in abandoned sugarcane fields in northeastern Brazil: floristic changes

Regeneração natural de mata em áreas de cana abandonadas no nordeste do Brasil: mudanças florísticas

Ladivania Medeiros do Nascimento^I,1 1 Corresponding author: Ladivania Medeiros do Nascimento, e-mail: ladivania@hotmail.com ; Everardo Valadares de Sá Barretto Sampaio^II; Maria Jesus Nogueira Rodal^III; Suzene Izídio da Silva^III; Ana Carolina Borges Lins e Silva^IV

^IJardim Botânico do Recife, Secretaria de Meio Ambiente, Prefeitura do Recife, Rod. BR 232, Km 7, Coqueiral, CEP 54240-450, Recife, PE, Brasil

^IIDepartamento de Energia Nuclear, Universidade Federal de Pernambuco – UFPE, Av. Professor Moraes Rego, 1235, Cidade Universitária, CEP 50670-901, Recife, PE, Brasil. www.ufpe.br

^IIIDepartamento de Biologia, Área de Botânica, Universidade Federal Rural de Pernambuco – UFRPE, Rua Dom Manoel de Medeiros, s/n, Dois Irmãos, CEP 52171-900, Recife, PE, Brasil. www.ufrpe.br

^IVDepartamento de Biologia, Área de Ecologia, Universidade Federal Rural de Pernambuco – UFRPE, Rua Dom Manoel de Medeiros, s/n, Dois Irmãos, CEP 52171-900, Recife, PE, Brasil. www.ufrpe.br

ABSTRACT

Surveys were undertaken to examine the floristic changes during secondary succession in three areas of 12 and three of 20-year-old secondary forests in Pernambuco State, Brazil. Two hundred and six species were identified, with 136 being found in the 12-year-old secondary forest and 161 species in the 20-year-old forest. Fabaceae and Myrtaceae were the most important families, increasing in species numbers with regeneration age. Of the 216 species, 115 were trees, 48 shrubs, 16 herbaceous plants, and 24 woody lianas, without significant differences between the two regeneration site ages. NMDS analysis revealed a formation of two floristic groups, distinguishing secondary and mature forests, with a further division within secondary forests in accordance with the time since abandonment. Similarity analysis ANOSIM confirmed the significance of the groups, which had floristic composition significant distinct (R=0.96) and 63% of dissimilarity (SIMPER). However, the sharing of 68 arboreal species between the secondary and mature forests suggests a floristic convergence. DCA analysis of the arboreal component as well as the other plant habits suggested that the separation of the subgroups is correlated with physical and chemical variables of the soils. All of these results indicate that, within the chronosequence analyzed, the velocity and direction of the floristic composition during secondary succession was influenced not only by the time of their abandonment, but also by a wide range of environmental variables.

Keywords: secondary forests, floristic, secondary succession.

RESUMO

Com objetivo de detectar mudanças florísticas ao longo da sucessão secundária e subsidiar futuros projetos de recuperação florestal foi realizado levantamento florístico de seis áreas de floresta secundária (capoeira) de 12 e 20 anos em Pernambuco. Foram registradas 206 espécies, sendo 136 nas capoeiras de 12 anos e 161 nas de 20 anos. Fabaceae e Myrtaceae foram as famílias mais importantes, aumentando no número de espécies com a idade de regeneração. Do total de espécies, 115 foram árvores, 48 arbustos, 16 herbáceas e 24 trepadeiras, sem diferença significativa por idade de regeneração. Análise de NMDS indicou a separação dos grupos florísticos das florestas maduras e das capoeiras, assim como a formação de subgrupos de capoeiras (SF1-2-3 e SF4-5-6) com idade de regeneração distinta. A análise de similaridade ANOSIM mostrou que os grupos formados apresentaram composição florística significativamente distintas (R = 0.96) e 63% de dissimilaridade (SIMPER). Entretanto, a presença nas capoeiras de 67 espécies em comum com as florestas maduras indicam uma tendência de convergência florística. A análise de DCA do componente arbóreo e dos outros hábitos sugere que a separação dos subgrupos por idade estaria correlacionada com variáveis edáficas físicas e químicas. Todos esses resultados indicam que, numa análise de cronossequência, não apenas o tempo de abandono, mas todas as variáveis ambientais influenciam a velocidade e direção de formação da composição florística durante a sucessão secundária.

Palavras-chave: capoeira, flora vascular, sucessão secundária.

Introduction

The degradation and destruction of natural habitats figure among the principal modern threats to biodiversity (Primack 2008). With the acceleration of the conversion of forests into pasture lands and agricultural fields in recent decades in Latin America (Geist & Lambin 2001) it has come to the point that mature forests altered by anthropogenic actions and regenerating secondary forests now compose approximately half of all remaining forest areas in the tropics (International... 2002). On the other hand, large areas of formerly cleared lands have also been abandoned and are evolving into secondary forests (Wright 2005).

Fragments of mature and secondary forests can be found in many landscapes otherwise dominated by agriculture or pasture lands. This situation can be seen, for example, in the Brazilian Atlantic Forest biome that has suffered negative impacts from a number of economic cycles, resulting in significant landscape modifications and ecosystem destruction (Dean 2002). The conservation of the Atlantic Forest and its biodiversity now presents a significant challenge to conservation efforts, principally due to its advanced state of forest substitution, and depends on the protection of the remaining fragments of mature forest (Tabarelli & Gascon 2005) and the correct management of regenerating forests after abandonment (Dent & Wright 2009). The Atlantic Forest now persists on only between 11% and 16% of its original land area coverage, and even these remnants are extremely fragmented (Ribeiro et al. 2009). This once extensive forest is currently composed of just a few well-preserved mature forest sites with numerous areas of varying ages with diverse use-histories undergoing secondary succession, and these are often surrounded by matrixes of cultivated areas and pasturelands (Ranta et al. 1998, Trindade et al. 2008).

A number of workers have pointed out the importance of secondary forests: as biodiversity reservoirs within fragmented landscapes (Chazdon 1998, 2003, Letcher & Chazdon 2009); as sources of sustenance for wildlife (Parry et al. 2007, Herrera-Montes & Brokaw 2010), providing sites for conservation of rare and endemic species (Liebsch et al. 2008) and wood and non-wood products (Chazdon & Coe 1999); in the accumulation of biomass (Gehring et al. 2004, Grace 2004, Feldpausch et al. 2007); for controlling carbon emissions (Feldpausch et al. 2005); and for diminishing pressure on natural habitats (Wright & Muller-Landau 2006).

The structural attributes of tropical forests such as density, biomass, richness, diversity, and composition can slowly recover after significant natural or anthropogenic disturbances (Brown & Lugo 1990, Guariguata & Ostertag 2001, Chazdon 2003, Finegan & Nasi 2004), but an important question in terms of our understanding of successional processes of secondary forests is if their floristic composition also tends to converge towards that of nearby mature forests or if they maintain their differences. According to Chazdon et al. (2009), convergence seems to be related to the level of anthropogenic disturbance, the duration of the time that the area was used, and the regional landscape context. The interrelations of these parameters make the outcomes of succession less predictable, and measurable differences in the floristic compositions of secondary forests may persist even centuries after they have been abandoned and left to recuperate (Finegan 1996, Aide et al. 2000, Denslow & Guzman 2000, Calvo et al. 2002, Ribas et al. 2003).

Previous researchers (refer to a review by Guariguata & Ostertag 2001) have demonstrated that a number of factors influence the recuperation of the floristic composition after disturbances, although many questions still remain unanswered due, in large part, to the idiosyncratic, non-directional, and largely unpredictable processes of succession in tropical forests (Letcher & Chazdon 2009).

The present work analyzed the floristic composition of 12 and 20-year-old secondary forests (known in northeastern Brazil as capoeiras) growing within a landscape dominated by sugarcane plantations and compared them with mature forest fragments of various sizes and showing varying degrees of disturbance, with the intention of addressing two basic questions: 1) Do 12 and 20-year-old secondary forest tend to floristically converge to resemble neighboring mature forests? 2) Is their time of abandonment/recuperation the principal factor determining this conversion process?

Materials and Methods

Weekly collections were made between June/2006 and July/2009 in six secondary forest fragments (capoeiras) belonging to the Usina São José (USJ) sugar refinery in Igarassu, Pernambuco State (PE), Brazil. The fragments were located between the geographical coordinates 07º 41' 04.9" and 07º 54' 41.6" S and 34º 54' 17.6" and 35º 05' 07.2" W, within a total area of approximately 280 km² (Figure 1); 88% of that area was occupied by a monoculture of sugarcane (Trindade et al. 2008). The six forest fragments had sizes that varied from 5 to 11 ha, and they were up to 300 m from mature forest fragments, the latter being considered mature in not having suffered any clear-cutting for at least 60 years.

The local climate is of the type As' (Köppen 1936), characterized as hot and humid, with an average annual rainfall rate of 1687 mm, average temperature of 24.9 ºC, and a dry season lasting more than three months (Schessl et al. 2008). The predominant geological formation in the area is the Barreiras Formation of the plio- Pleistocene (the most extensive geologic formation in this region of the coast) that comprises nonconsolidated sandy-clay sediments of continental origin (Companhia... 2003). The landscape is dominated by coastal plateaus 40 to 160 m above sea level that are cut by deep and narrow valleys whose sides have inclinations greater than 30% (Companhia... 2003). The regional vegetation is classified as Dense Ombrophilous Lowland Forest (Veloso et al. 1991).

Site selections considered the approximate ages of the secondary forest (capoeira) remnants based on the vegetational characteristics captured in aerial photographic sequences (CONDEPE/FIDEM) covering the decades of 1960, 1970, and 1980 (at a scale of 1: 30,000), as well as satellite images acquired in 2005 and disturbance histories from the 1990s obtained from interviews with long-time residents in the area.

Of the six secondary forest sites examined, three had been undergoing natural regeneration for approximately 12 years (sites 1, 2 and 6) and three had regenerated for 20 years (sites 3, 4 and 5) after suspending sugarcane cultivation during the 1980s. When they were selected in 2006, the six areas had shrub/arboreal canopy physiognomies varying from 6 to 18 m in height.

Botanical material was collected, using traditional techniques (Mori et al. 1989), crisscrossing the fragment in random walks (Filgueiras et al. 1994) and also examining each plant within one hundred and eighty 10 × 10 m plots installed in the six fragments (30 plots in each fragment). In each fragment, the whole fragment was surveyed, including edge and interior areas. Growth habits examined were arboreal (diameter at breast height, DBH > 15 cm), shrub/subshrubs and terrestrial herbs (DBH < 15 cm), epiphytic herbs and woody lianas, all with reproductive parts, whether flowers or fruits. Seedlings or saplings of arboreal species were not included.

Soil samples were collected in each plot from the first 20 centimeters below the surface and analyzed in terms of their soil texture (sand, silt, and clay), pH in water, P, K⁺, Ca²⁺, Mg²⁺, and extractable Al³⁺, according to protocols described in the Manual of Soil Analysis Methods (Embrapa 1997).

Reference collections were incorporated into the Geraldo Mariz Herbarium (UFP), with duplicates to the Dárdano de Andrade Lima Herbarium (IPA). Species identifications were made by specialists at different Brazilian institutions and by comparisons with collections deposited at the Professor Vasconcelos Sobrinho (PEUFR) and IPA herbaria.

A species list was prepared, listed according to family, with information about plant habits, localities of occurrence, and their herbarium collection numbers. The classification of the angiosperm families followed the recommendations of the APG III (Angiosperm... 2009), and those of the pteridophytes followed Smith & Wolf (2006), with modifications as presented by Smith et al. (2008). The authors' names and scientific names were confirmed using The International Plant Names Index (www.ipni.org) database.

Statistical differences in the soils were tested using one-way ANOVA (Stat Soft 2001). Values were transformed before analysis if they did not exhibit normal distribution or variance homogeneity, following Sokal & Rohlf (1995).

The G test (Sokal & Rohlf 1995) was used to identify changes in the species richness of each plant habit between the 12 and 20-year-old secondary forests.

The floristic compositions of the arboreal component of mature forest fragments in the study area were obtained from the works of Silva (2004), Rocha et al. (2008), and Silva et al. (2008), which included all trees with DBH > 15 cm, similarly to the present study. Four types of analyses were performed to examine the arboreal habit data (NMDS, ANOSIM, SIMPER and DCA-Detrended Correspondence Analysis), utilizing the PAST 2.01 (Hammer et al. 2001) and PC-ORD version 4.0 software program (McCune & Mefford 1999).

The NMDS analysis (Non-metric Multi-dimensional Scaling) was used to analyze the degree of floristic difference between the two secondary forest ages (12 and 20-year-old), and between those and nearby mature forests. To that end, a binary matrix (presence/absence) of the secondary forests (115 species) and the mature forests (171 spp.) was constructed. Infra-specific taxa or those without precise identification to the species level (listed as "sp.", "cf." or "aff."), which occurred in at least two areas, were excluded from analysis, resulting in a matrix of 136 species. In the analysis, the Jaccard (J) index coefficient was applied to generate a graph (Legendre & Legendre 1998).

The non-parametric analyzis ANOSIM (Clarke 1993) was also performed, com 10,000 permutations, aiming at confirming the significance of the groups formed by the NMDS analysis. This method generates a global R-statistic, which is a measure of the distance between groups. An R-value close to 1 indicates strongly dissimilar assemblages, while an R-value close to zero indicates that assemblages are scarcely distinguishable (Clarke 1993). These R-values were used to compare floristic assemblages between secondary vegetation ages. Where ANOSIM revealed significant differences between groups, Similarity percentages (SIMPER) analysis was used to identify those species that contributed most to the observed assemblage distinction (Clarke 1993). Cumulative contributions were cut arbitrarily at 50%. Species with the highest dissimilarity to standard deviation ratios were identified as good discriminators for each comparison (Clarke 1993).

A third analysis (DCA) was applied to determine correlations between the species distributions and environmental variables. Data concerning the presence/absence (0-1) of the arboreal species (115 spp.) and the set of other habits (shrubs, herbs, and woody lianas = 91 spp.) were used to form two primary matrices. A categorical variable matrix was created based of the average values of the chemical (pH, P, K⁺, Ca²⁺, Mg²⁺, Al³⁺) and physical (percentages of sand, clay, and silt) analyses of the soils and with the time of vegetation recuperation (12 and 20 years). The Pearson and Kendall correlation was applied to evaluate the representativeness of the variables on the axes; r values > 0.70 were considered high (Cohen 1988, Dancey & Reidy 2006).

Results

Sixty-six families, 120 genera, and 206 species were recorded in the six secondary forests at the USJ. Fifty families and 136 species were identified in the 12-year-old areas, and 57 families and 161 species were encountered in the 20-year-old areas (Table 1), which represented an approximately 15% increase in the number of species. Of the 206 species encountered, 115 were trees (56%), 48 shrubs (23%), 16 herbs (8%), 24 woody lianas (12%), and three were epiphytes (1.5%) (Table 2). In spite of the fact that there was a 20% increase in the number of arboreal species in the 20-year-old secondary forest, there were no significant overall differences in terms of the habits of the plants in the two regeneration ages (G test, p = 0.55, Table 2).

The NMDS analysis, applied for all arboreal species of secondary forests (12 and 20 years) and mature forests, revealed the formation of three groups, with stress of 0.093 (Figure 2). The first group was composed by the 12-year secondary forests (SF 1, 2 and 3), the second by the 20-year secondary forests (SF 4, 5 and 6), and the third, by mature forests (Silva 2004, Rocha et al. 2008, Silva et al. 2008).

When the ANOSIM was performed, the three groups previously formed by NMDS were found to be significant (R = 0.983, p = 0.0036), with dissimilarity (SIMPER) of 63%. The two groups composed by secondary forests (SF 1-2-3 and SF4-5-6) differed in floristic composition, with dissimilarity of 57%. Floristic differences were larger when 12-year secondary forest group was compared to mature forests (MF 1-2-3), with R = 1 and dissimilarity of 78% (Table 3). The high dissimilarity on floristic composition between mature and secondary forests was shaped by the presence of tree species exclusive to mature forests, such as: Aspidosperma discolor, Chamaecrista ensiformis, Parkia pendula, Inga blanchetiana, I. capitata, Pouteria bangii, P. peduncularis.

The physical and chemical parameters of soil varied significantly between the secondary forest (p < 0.05), regardless of age. The secondary forests differed significantly (p < 0.05) in their soil physical and chemical parameters and the differences were not related to the forest age. SF1, SF2, and SF3 (20-years) had similar values differing from SF6 (12-years), SF4, and SF5 (Table 4).

This separation of the secondary forests independent of their times of recuperation as observed in the grouping analysis (sub-groups A and B) was confirmed by the DCA analysis, indicating on axis I (eigenvalue = 0.43) that the concentrations of K⁺ (r = 0.92), Ca²⁺ (r = 0.64), Mg²⁺ (r = 0.83), and the percentages of clay (r = 0.89) and silt (r = 0.96) were responsible for the formation of the secondary forest groups 1-2-3, while pH (r = –0.74) and sand (r = –0.93) were responsible for the formation of the secondary forest groups 4-5-6 (Table 4; Figure 3a).

The DCA analysis for the other plant habits indicated the formation of two groups with distinct ages (Figure 3b). On axis 1 (eigenvalues = 0.82) there was a correlation of the first group (SF 1-3) with variables K⁺ (r = 0.57), clay (r = 0.68), and silt (r = 0.69) and of the second group (SF 2-4-5-6) with pH (r = –0.97) and sand (r = –0.69), Table 4.

Discussion

The difference in the numbers of families between the two regenerating forest ages may seem small, although the presence of families typical of mature forests (such as Aspleniaceae, Dryopteridaceae, Orchidaceae, and Pteridaceae) in the oldest sites and the increases in the numbers of species of Fabaceae, Myrtaceae, Annonaceae, and Sapotaceae indicated that there was a significant floristic enrichment over the years – even though this was not as notable when considering the plant habits.

The occurrence of a large number of species (206) in secondary forests with different times since abandonment within the same landscape, experiencing the same climate, geology and history, could be attributed to spatial heterogeneity, which allows that a high number of plant species persists due to the high resource supply (Tilman & Pacala 1993). According to Connell (1978), different successional stages can be seen as moderately disturbed scenarios, in which disturbance occurs with moderate frequency, duration and intensity, enabling that pioneer and secondary species cohabit the same area, resulting in larger species richness, when compared to less disturbed sites (Castillo-Campos et al. 2008).

The low level of floristic similarity between the secondary and mature forests indicates that they are in distinct successional stages and corroborates with the hypothesis that the recuperation of the floristic composition of secondary forests occurs only slowly (Chazdon 2003, Chazdon et al. 2009, Piotto et al. 2009, Powers et al. 2009). However, the sharing of 68 arboreal species between the secondary and mature forests suggests that the flora of the secondary forests at the USJ tends to convergence to that of the mature forests – as has been observed in other tropical forests (Guariguata et al. 1997, Peña-Claros 2003, Capers et al. 2005, Carim et al. 2007, Chazdon et al. 2007, 2009, Liebsch et al. 2007, Castillo-Campos et al. 2008, Norden et al. 2009, Lebrija-Trejos et al. 2010).

The formation of secondary forest groups composed of distinct regeneration ages, as observed in the grouping analysis and ordinations of the arboreal habit, can be justified by the similarities of the soil characteristics within the groups (which is often a consequence of the close proximity of these areas). In terms of the other plant habits, there is also a high correlation within the groups that were formed, principally due to the similarities of soil textures (with the exception of SF 2).

All of these results indicate that not just recuperation times, but many other environmental variables influence the velocity and direction of the formation of floristic composition during secondary succession. Our results corroborate those reported for secondary forests in Costa Rica (Chazdon et al. 2009), Mexico (Powers et al. 2009), and the Brazilian Amazon (Prata et al. 2010) – which also did not encounter any significant correlation between the forest ages and their floristic compositions – with each locality following a distinct and idiosyncratic path of species accumulation driven by edaphic factors (Guariguata & Ostertag 2001), colonizing species (Chazdon 2003, Junqueira et al. 2010), and the landscape matrix (Bierregaard et al. 1992, Nascimento & Laurance 2006). Together these results indicate that the preservation of conserved forest fragments near areas undergoing regeneration permits genetic flux and the continuity of successional processes.

Acknowledgements

The authors would like to thank the Usina São José/Grupo Cavalcanti Petribú for allowing us to undertake this research on their property as well as the research groups at the Laboratório de Ecologia Vegetal (LEVE) and the Laboratório de Fitossociologia (LAFIT) of the Universidade Federal Rural de Pernambuco for their assistance with the fieldwork. This study was part of the Fragments Project, in the Brazilian-Germany Cooperation, Science and Technology for the Atlantic Forest (Proc. 690147/01-5) financed by CNPq (Conselho Nacional de Desenvolvimento Científico e Tecnológico, Brasil) and BMBF (01 LB 0203 A1), with a Doctorate CNPq grant to the first author.

Received 07/06/2012

Revised 19/10/2012

Accepted 14/11/2012

Publication Dates

History