Floristic and phytosociological analysis of palm swamps in the central part of the Brazilian savanna

We analyzed the fl oristics and phytosociology of three palm swamps in the municipality of Bela Vista de Goiás, located in the state of Goiás, Brazil, in the central part of the Brazilian savanna (Cerrado). Th e fl oristic surveys were conducted monthly from May 2008 to April 2009, and 310 species were recorded (seven bryophytes, 15 ferns and 288 angiosperms). Bryophytes belonged to fi ve genera and fi ve families; ferns belonged to nine genera and nine families; and angiosperms belonged to 134 genera and 45 families. Th e angiosperm families with the highest species richness were Poaceae, Cyperaceae, Asteraceae, Eriocaulaceae, Xyridaceae, Lentibulariaceae, Melastomataceae, Rubiaceae and Fabaceae. Th e palm swamps were divided into three zones of increasing humidity: edge, middle and core. Th e number of species was higher in the middle than at the edge and the core. Th e families with the highest cover values were Cyperaceae, Melastomataceae, Arecaceae and Poaceae. Although the palm swamps had been disturbed to varying degrees, those disturbances did not aff ect the fl ora in the middle or the core. Floristic similarity was high between these two zones within a given palm swamp and low between the edges of diff erent palm swamps.


Introduction
Wetlands are environments in which water is the determining factor of abiotic and biotic conditions (Keddy 2000, Steinke & Saito 2008).Th ese areas support several typical animal and plant species and foster a large number of ecological processes that sustain those species (Maltchik et al. 2003), making wetlands some of the most productive environments on the planet (Steinke & Saito 2008).
In the Brazilian savanna (Cerrado), despite the dominance of phytophysiognomies on well-drained soils, wetlands such as palm swamps also occur (Meireles et al. 2002).Palm swamps are communities composed of a continuous herb-grass layer, which occupies most of their area, and a tree-shrub layer dominated by individuals of the palm Mauritia fl exuosa L. f. (local name, buriti), with a canopy cover of between 5% and 10% (Ribeiro & Walter 2008).
In addition to the diversity of plant species (Araújo et al. 2002, Guimarães et al. 2002), palm swamps present some distinct micro-ecosystems in which there are some plant-animal interactions that are still poorly understood (Oliveira et al. 2009).Th ese environments serve as refuges, food sources and reproduction sites for the aquatic and terrestrial fauna of the adjacent phytophysiognomies (Guimarães et al. 2002).Th ey are also sites of great scenic beauty with economic and sustainable potential for ecotourism and for extractivism involving species such as M. fl exuosa and Syngonanthus nitens (Bong.)Ruhland (capim-dourado, which means "golden grass") by small rural communities (Schmidt et al. 2008).
Palm swamps are categorized as areas of permanent preservation, for ensuring the maintenance of springs and the quality of watercourses (Araújo et al. 2002).Although legally protected, palm swamps have suff ered human disturbances that occasionally become irreversible, mainly because of their low regenerative capacity (Carvalho 1991, Araújo et al. 2002, Guimarães et al. 2002).Considering the importance and the condition of degradation of palm swamps, it is necessary to expand studies on the physical, biotic, social and anthropic aspects of this physiognomy, in a dynamic and multidisciplinary manner.Th us, it will be possible to understand the aspects that characterize palm swamps and to support measures to mitigate possible environmental impacts.Th ere have been a few studies on palm swamp vegetation (Araújo et al. 2002;Guimarães et al. 2002;Ramos 2004;Oliveira et al. 2009).However, there have been no studies on the plant diversity of this physiognomy in the central part of the Cerrado.Th erefore, the aim of this study was to contribute to the knowledge of the fl oristic composition of palm swamps in the central part of the Cerrado and to conduct a phytosociological survey in those swamps.

Material and methods
Th e study sites were three disturbed palm swamps, which were selected based on previous studies involving satellite images and fi eld trips.All three sites are located on private properties in the city of Bela Vista de Goiás, in the state of Goiás, Brazil, and are part of the riparian vegetation of the tributaries of the Peixe River, which is in turn a tributary of the Corumbá River, pertaining to the Paranaíba River watershed (Fig. 1).
Th e palm swamp with the lowest degree of disturbance was in an area of transition to cerradão and was located upstream of dams for the provisioning of cattle.Th e M. fl exuosa individuals at that study site displayed signs of burning on the stems and of trampling by cattle.Th e other two palm swamps did not display signs of burning, and, at the time of sampling, one was surrounded by cultivated pasture dominated by Urochloa spp., whereas the other was surrounded by crops (primarily corn and soybean).Th e palm swamp surrounded by crops was the only one that was not at least partially fenced off .Th is site suff ered interference by cattle, which were released by cattle farmers in order to graze on crop residues (mainly of corn) during the postharvest period.We also observed signs of swine, which were allowed to feed at the site on a regular basis.
According to the Köppen classification system, the climate in the region is type Aw, which is defi ned as megathermal, with a rainy summer and a dry winter.Mean monthly temperatures range from 19°C to 28°C and mean annual rainfall is below 2000 mm (INMET 2009).
Th e fl oristic survey was conducted between May 2008 and April 2009, with monthly collection of fertile botanical material observed in asystematic walks throughout the three sites.Each palm swamp was subdivided into three zones: edge, middle and core (Araújo et al. 2002, Guimarães et al. 2002).Th e edge is closest to the transitional vegetation and exhibits soil that is lighter in color and more well-drained.Th e middle exhibits soil that is darker and periodically saturated with water.Th e core is permanently fl ooded.For all sampled material, we recorded data on the growth habit, zone of occurrence and morphological characteristics.
Th e botanical material was mounted following the usual procedures, and the vouchers were deposited at the Herbarium of the Federal University of Goiás.Species were iden-tifi ed with the aid of specialists in various botanical families and by comparison with identifi ed vouchers in herbaria.We adopted classifi cation systems based on the following works: Goffi net & Buck (2004) for Bryophyta; He-Nigrén et al. (2006) for Marchantiophyta; Tryon & Tryon (1982) for Pteridophyta; and the Angiosperm Phylogeny Group 2009 update (APG III 2009) for Magnoliophyta (angiosperms).
Th e phytosociological survey was conducted with the line intercept method, adapted from Munhoz & Felfi li (2006), to determine the composition and linear cover of species.At each site, we selected four transects, 10 m apart, comprising two transects on each bank of the drainage canal.Th e transects were positioned perpendicular to the drainage line, and, for each transect, we extended a line from the edge to the core of the palm swamp.Th e line was divided in 1-m sections, or sample units.We analyzed 129 sample units in the palm swamp surrounded by cerradão, 246 in that surrounded by pasture and 115 in that surrounded by crops.For each 1-m section, we recorded the occurrence of each species and its horizontal projection.Th e sum of the horizontal projections of a given species in all sample units was taken as the absolute cover value for that species in each palm swamp.Th e relative cover value was obtained by dividing the absolute cover value of each species by the sum of the absolute cover values of all species, and multiplying the result by 100.Th e record of occurrence of each species in the sample units was used in order to calculate the frequency of species in each palm swamp.To investigate the dynamics of the vegetation, the fi rst phytosociological survey was performed in October 2009 (at the end of the dry season) and the second was performed in March 2010 (at the end of the rainy season), both surveys being performed in the same transects.
To evaluate the diversity of the community, we used an adaptation of the Shannon diversity index, using the cover values of both sampling periods instead of the number of individuals, in accordance with the method suggested by Munhoz & Felfi li (2006).Th e fl oristic similarity between the palm swamps and between the zones (edge, middle and core) of each site was evaluated by the quantitative Sørensen similarity index (Mueller-Dombois & Ellenberg 1974).Th e data of fl oristic similarity (I) were transformed to dissimilarity indices (D = 1 -I) and evaluated with a cluster analysis using the unweighted pair group method with arithmetic mean, with the soft ware GENES (Cruz, 2006).
Th e number of species recorded was equal or comparable among the three sites: 187 in the palm swamp surrounded by pasture; 187 in the palm swamp surrounded by crops; and 186 in the palm swamp surrounded by cerradão.Floristic similarity between the sites was high: 58.2% between the palm swamp surrounded by pasture and that surrounded by crops; 61.9% between the palm swamp surrounded by pasture and that surrounded by cerradão; and 55.1% between the palm swamp surrounded by cerradão and that surrounded by crops.According to Kent & Coker (1992), values ≥ 50% indicate high similarity.Although the sites had been disturbed to varying degrees, those disturbances did not aff ect the fl ora in the middle or core zones of these palm swamps, only the species at their edges.
Floristic similarity was higher between the middle and core zones and lower between the edge zones (Tab.2).Th e dendrogram resulting from a cluster analysis shows the fl oristic similarity between the middle and the core within each palm swamp, as well as the dissimilarity between the sites (Fig. 2).Th e coeffi cient of cophenetic correlation was 0.70, indicating good representativeness of the original dissimilarity indices in the dendrogram.Th ere was fl oristic dissimilarity between the edges of the three palm swamps and between the edge and the other two zones within each palm swamp.Despite belonging to diff erent sites, the three edge zones were more similar to each other than to their respective middle and core zones.Th is might be attributable to the fact that an edge is a zone of transition to the adjacent vegetation.In the palm swamp surrounded by pasture and the palm swamp surrounded by crops, the colonization by diaspores of exotic or invasive species, originating from the human activities in the vicinity, might have contributed to the diff erences between edge zones.
Th e Sørensen similarity index between the zones (edge, middle and core) within each palm swamp was higher between the middle and the core and lower between the edge and the core (Tab.2).Th e highest fl oristic similarity (76.4%) was between the middle and the core of the site surrounded Conyza bonariensis (L.) Cronquist neotropical* X

Elephantopus mollis Kunth pantropical -neotropical X
Emilia fosbergii Nicolson pantropical* -Bahamas?; old world?X X X Emilia sonchifolia (L.) DC. old world* -African?X Erechtites hieraciifolius (Fisch.ex Spreng.)Griseb.South American* Rhynchospora marisculus Lindl.ex Nees Central and South American* X Rhynchospora riparia (Nees) Boeckeler South American* X X Rhynchospora tenerrima Nees ex Sprengel Central and South American* X X     *Species that were observed at the study sites but have a wide geographical distribution (neotropical, pantropical, cosmopolitan, etc.), for which the origin, or initial distribution, cannot be indicated in the Cerrado, including ruderal, invasive and exotic species, according to Mendonça et al. (2008).**New record for the state of Goiás.***New record for the Cerrado, according to Mendonça et al. (2008).
2 Vernonia polyanthes -although not cited by Mendonça et al. (2008), this species occurs in the states of Goiás, Mato Grosso, Paraná, Bahia, São Paulo and Minas Gerais, and is considered a common species in the Cerrado (Almeida 2008).
3 although not cited by Mendonça et al. (2008), this species has been sampled in the municipality of Jalapão, state of Tocantins (Rezende 2007). 4species previously sampled in palm swamps in the Triângulo Mineiro area, state of Minas Gerais (Oliveira et al. 2009). 5species previously sampled in the Federal District and in the state of Goiás (Delprete 2007). 6species previously sampled in the state of Goiás (Souza 2009).
7 species with a wide distribution in South America (Delprete 2007).
Table 1.Continuation.by cerradão.Th e higher similarity between the middle and the core can be explained by the increase in humidity from the edge toward the core within a palm swamp, and by the fact that the middle and the core are less disturbed (Guimarães et al. 2002;Araújo et al. 2002).In contrast, edges are transitional zones to the adjacent phytophysiognomy and are more vulnerable to disturbances (such as cattle grazing and trampling, in palm swamps surrounded by pasture; Guimarães et al. 2002;Araújo et al. 2002) and invasion by exotic species (Oliveira et al. 2009).Sousa (2009) studied the soil texture in the three palm swamps evaluated in the present study and reported high sand content in the core zone, whereas the soil in the middle and at the edge had a high clay content, without signifi cant textural diff erences.Th erefore, we might infer that the topography is more important than is the soil texture in determining the establishment of humidity zones and the occurrence and fl oristic distribution of palm swamps, as Ramos et al. (2006) previously demonstrated in a study on the characterization of soils of palm swamps on diff erent geomorphological surfaces in the Triângulo Mineiro region (state of Minas Gerais).From the dry edge toward the wet core within a palm swamp, there is a gradient of soil declivity and depth of the water table (Ramos et al. 2006;Guimarães et al. 2002), which alters the fl oristic composition of the three zones (Araújo et al. 2002).
(Malvaceae); Lysimachia pumila (Primulaceae); Digitaria sp.; four of the fi ve Eragrostis species; and Urochloa humidicola (Poaceae).Th ese were located mainly at the edge and might be considered invasive species.Th e diaspores of these plants originated from the surrounding crops.Species exclusive to the site surrounded by pasture included fi ve of the seven bryophytes recorded in this study, as well as Geissomeria pubescens (Acanthaceae); Cardiopetalum calophyllum (Annonaceae); Conyza bonariensis, Emilia sonchifolia and Vernonia polyanthes (Asteraceae); Bulbostylis capillaris, Calyptrocarya luzuliformis and Fimbristylis complanata (Cyperaceae); both Leiothrix species and Syngonanthus anthemidifl orus (Eriocaulaceae); Camptosema coriaceum (Fabaceae); Tetrapollinia caerulescens (Gentianaceae); Hyptis subrotunda and H. velutina (Lamiaceae); Utricularia huntii (Lentibulariaceae); Ctenium sp., Paspalum imbricatum and Schizachyrium tenerum (Poaceae); and Xyris tenella (Xyridaceae).Despite the presence of invasive species, usually located at the edge, the species exclusive to this site were more characteristic of palm swamps and wetlands, such as bryophytes.Th is might have resulted from the fact that this site is fenced, which avoiding grazing and trampling by the cattle from the surrounding pasture.
Th e fern Ophioglossum crotalophoroides (Ophioglossaceae), observed only in the palm swamp surrounded by cerrado, was not recorded in Goiás in the recent literature, and is considered a critically endangered species in Minas Gerais (COPAM 2008).Until now, it was sampled only in the Pantanal of the state of Mato Grosso do Sul (Cardoso et al. 2000), in the Federal District (Pereira et al. 2004;Felfi li et al. 2007) and in the Quadrilátero Ferrífero area of Minas Gerais (Salino & Almeida 2008).Regarding the bryophytes, this study is the fi rst record of Leucoloma serrulatum (Dicranaceae) and the second record of Sphag-num cyclophyllum (Sphagnaceae) in Goiás (see Sousa et al. 2010;Yano 2011).
For all three study sites, the number of species was higher in the middle than in the other zones.However, for the palm swamp surrounded by cerradão, the diff erence between the middle and the core was small in comparison with that observed for the other two sites (Tab.1).At the three sites, only Praxelis grandifl ora (Asteraceae) and Setaria parvifl ora (Poaceae) were exclusive to the edge, and Cyathea atrovirens (Cyatheaceae) were exclusive to the core.In the middle and core of the three palm swamps, we identifi ed the species Drosera communis (Droseraceae); Eriocaulon humboldtii, Syngonanthus caulescens, S. densifl orus and S. nitens (Eriocaulaceae); Utricularia nigrescens, U. praelonga and U. trichophylla (Lentibulariaceae); Miconia chamissois (Melastomataceae); Otachyrium seminudum (Poaceae); Cecropia pachystachya (Urticaceae); and Xyris jupicai, X. laxifolia, X. savanensis, and Xyris sp.(Xyridaceae).In the edge and middle of the three sites, the species Desmodium barbatum (Fabaceae) and Sipanea hispida (Rubiaceae) occurred.Th e distribution patterns of the remaining species diff ered among the three sites.
Th e Shannon diversity index was 2.8 for the palm swamp surrounded by pasture, 3.0 for the palm swamp surrounded by crops and 3.4 for the palm swamp surrounded by cerradão.In the few studies on the vegetation of palm swamps, this index has not been calculated.Th erefore, we cannot yet determine whether our results are representative of these physiognomies, i.e., whether they can be classifi ed as conserved or disturbed.
In the phytosociological surveys, we sampled 130 species belonging to 80 genera and 37 families (Tab.3).In the two sampling periods, we recorded 86 species at the site surrounded by pasture, 81 at the site surrounded by crops and 96 at the site surrounded by cerradão.Th e families with the highest cover values in October 2009 and March 2010, respectively, were as follows: Cyperaceae (37.9% and 21.4%), Melastomataceae (21.0% and 20.6%), Arecaceae (16.8% and 18.0%), and Poaceae (11.1% and 6.2%) in the palm swamp surrounded by pasture; Melastomataceae (25.6% and 37.9%), Cyperaceae (20.7% and 11.7%), Poaceae (17.5% and 19.7%), and Arecaceae (10.3% and 16.3%) in the palm swamp surrounded by crops; and Poaceae (29.3% and 26.3%), Cyperaceae (25.8% and 25.8%), Melastomataceae (18.0% and 11.8%), and Arecaceae (3.3% and 11.0%) in the palm swamp surrounded by cerradão (Fig. 3).For most of the families, relative frequency was higher than was relative cover, at all three study sites (Fig. 3).Th is might be attributable to the fact that most palm swamp species are herbs or subshrubs, small in height and diameter, and are widely distributed through the area (Munhoz & Felfi li 2006).
Th e families for which the relative cover was higher than was the relative frequency included Melastomataceae, Arecaceae, Cyperaceae, Poaceae and Urticaceae.Th e species with the highest relative cover were Rhynchospora consanguinea, R. tenuis, R. globosa, R. robusta, Mauritia fl exuosa, Miconia chamissois, Andropogon virgatus, A. bicornis, Rhynchanthera grandifl ora and Saccharum angustifolium.Th ese species were among the ten main species in terms of the cover values, in both sampling periods; there was only one change in order from one period to the next.In addition to these species, the following species were also among the ten species with the highest relative cover: Urospatha sagittifolia, Xyris sp., Psychotria paracatuensis, Syngonanthus densifl orus, S. caulescens and Ilex affi nis in the palm swamp surrounded by pasture; Cecropia pachystachya, Hyptis lantanifolia, Xyris laxifolia and Piper aduncum in the palm swamp surrounded by crops; and Ludwigia nervosa, Xanthosoma striatipes, Andropogon sp. and Syngonanthus densifl orus in the palm swamp surrounded by cerradão.Species of Poaceae and Cyperaceae usually form dense clumps or have individuals that possess rhizomes or stolons and are widely distributed in the vegetation.
In contrast, the Melastomataceae species are shrubs and Arecaceae species are trees, as are Urticaceae species.Th is explains why relative cover was higher than was the relative frequency for those species.
Th ere was a small diff erence between the two sampling periods in terms of the number of species.Th e number of species recorded in October 2009 and in March 2010 was, respectively, 60 and 74 in the palm swamp surrounded by cerradão, 65 and 72 in the palm swamp surrounded by pasture, and 61 and 57 in the palm swamp surrounded by crops.As can be seen, at the fi rst two sites, there was a slight increase from October to March.Th is small variation in species number might have occurred because palm swamps have a seasonal area and a perennial area, the soil in the latter remaining water-saturated for several months of the year (Ramos 2004, Oliveira 2005).
At the study sites, the families with the highest species richness were Poaceae, Cyperaceae, Asteraceae, Eriocaulaceae, Xyridaceae, Lentibulariaceae, Melastomataceae, Ru-   Bold indicates the ten main species with the highest RC per period and per area.biaceae and Fabaceae.We found that the fl oristic similarity was greatest between the middle and core zones within a palm swamp and lowest between the edge zones of diff erent palm swamps.However, additional studies comparing diff erent areas for larger periods of time are needed in order to determine whether there are patterns in the spatial and temporal distribution of plant species in palm swamps, as well as to quantify the eff ect that human disturbance has on these environments.
Although the herb-subshrub layer is dominant in several phytophysiognomies in the Cerrado and exhibits high species richness, as in the case of palm swamps, its vegetation has been poorly studied.Th is lack of data impedes the enforcement of public policies for the conservation and sustainable management of palm swamps.Th ese environments support signifi cant biodiversity and a rich fl ora.Because palm swamps occur in areas where there are natural springs, protecting these swamps also protects water resources.

Figure 1 .
Figure 1.Location of the three palm swamp study sites in the city of Bela Vista de Goiás, state of Goiás, Brazil.

Figure 2 .
Figure 2. Dendrogram generated by the complement of the Sørensen similarity index ( (D = 1 -I) between the nine zones of the palm swamps studied in the city of Bela Vista de Goiás, state of Goiás, Brazil.Prefi xes: E -edge; M -middle; C -core.Suffi xes: C -palm swamp surrounded by cerradão; P -palm swamp surrounded by pasture; Cr -palm swamp surrounded by crops.Cophenetic correlation=0.6982.

Figure 3 .
Figure 3. Distribution of cover values and relative frequencies (means of the two sampling periods), by family, for the fl ora of three palm swamps in the city of Bela Vista de Goiás, state of Goiás, Brazil.PP -palm swamp surrounded by pasture; PCr -palm swamp surrounded by crops; PC -palm swamp surrounded by cerradão.

Table 1 .
Species, ordered by category and family, sampled in the edge, middle and core zones (E, M and C, respectively) of three palm swamps-one surrounded by pasture (PP); one surrounded by crops (PCr); and one surrounded by cerradão (PC)-in the city of Bela Vista de Goiás, state of Goiás, Brazil.

Table 2 .
Matrix of fl oristic similarity (Sørensen similarity index) between the nine zones of the palm swamps studied in the city of Bela Vista de Goiás, state of Goiás, Brazil.Prefi xes: E -edge; M -middle; C -core.Suffi xes: C -palm swamp surrounded by cerrado; P -palm swamp surrounded by pasture; Cr -palm swamp surrounded by crops.

Table 3 .
Relative frequency (RF) and relative cover (RC) of the species sampled in three palm swamps-one surrounded by pasture (PP); one surrounded by crops (PCr); and one surrounded by cerradão (PC)-in the city of Bela Vista de Goiás, state ofGoiás, Brazil, in October 2009 and March 2010.