To find similarity among these surveys, we first converted all data into a large matrix containing species occurrence presence/absence. A matrix of similarity was built using the Bray-Curtis distance coefficient (Krebs 1999). To compare all surveys, we used the method of complete linkage (Sneath & Sokal 1973). A dendrogram of similarities was built using the PRIMER v. 6 software, Plymouth Routines in Multivariate Ecological Research (Clarke & Gorley 2006).

Using the entire dataset, which included all wetlands, we assessed whether the richness of macrophytes in South America reaches an asymptote, or if there are many species yet to be found. The expected species accumulation curve was calculated according to a Mao Tau function. Using an accumulative curve with "studies" as units of sampling effort, an asymptote would indicate whether almost all species have already been recorded; however, the lack of an asymptote would indicate that the number of aquatic macrophyte species found until now is still far from the real total of these species. We also estimated the richness of macrophytes using a first-order Jacknife estimator (Jack1) with the objective to assess the extent to which the number of macrophyte species in South America remains underestimated. Accumulation and estimation curves were constructed using the EstimateS program (Colwell 2009).

A total of 153 species of macrophytes was recorded in the Upper Paraná River floodplain. These species were distributed in 100 genera and 37 families (Appendix 1), representing a variety of taxonomic groups (Charophyta Bryophyta, Pteridophyta, Basal Angiospermae and Angiospermae).

Sixteen of the recorded species are cryptogams and are classified as follows: two charophytes, two hepatophytes and 12 pteridophytes. Of the Angiospermae, Poales exhibited the highest number of taxa (40), followed by Alismatales (17), Myrtales and Lamiales (12 species each). The families with the highest numbers of species were Poaceae (21), Cyperaceae (17), Pontederiaceae (8), Hydrocharitaceae (7), Polygonaceae and Onagraceae (6) and Fabaceae (5). Araceae, Alismataceae, Commelinaceae, Amaranthaceae and Plantaginaceae were represented by four species each and the other families by three or fewer species.

All life forms were found in the area, and emergent and amphibious types were the most representative macrophytes, contributing 45% and 26% of the species, respectively (Fig. 2). Rooted submersed (11%) and free-floating species (9%) were also important, while the lowest numbers of species were found for rooted floating and free-submersed types (2%) and epiphytes (1%).

The second group included the Amazon basin of Ecuador, Amazon River floodplain and Amapá wetlands, with a similarity of 13% and sharing four species, Eichhornia azurea, Hymenachne amplexicaulis, Salvinia auriculata and Utricularia foliosa. Within this group, the aquatic flora of the Amazon River floodplain and the Amazon Basin of Ecuador share 13 species (2% of the total species) and exhibit 35.1% similarity. Finally, group three was formed by Peru, Ecuador and Bodoquena (Mato Grosso do Sul), with a similarity of 18% and sharing 11 species; within this group, Peru and Ecuador had the highest similarity of 33.6% and sharing 35 species (4% of all species).

Considering all of the surveys that we found for South America together with the survey we conducted in the Paraná floodplain, a total of 854 species of macrophytes was compiled. However, the species accumulation curve produced using these studies as a surrogate of sampling effort, did not reach an asymptote (Fig. 5). In fact, the number of species estimated through Jack1 was 1 388, indicating an approximate underestimation of 534 species.

According to Chambers et al. (2008), the Neotropical region has the highest number of macrophyte species in the world (984 species). The number of species of macrophytes recorded in the Upper Paraná River and its floodplain (153 species; 16% of the Neotropical region) can be considered high due to the small relative area of this ecosystem (2 500km²) compared to other aquatic areas from this region, such as the Amazon and the Pantanal. Even considering the higher species richness found in the Pantanal Matogrossensse (247 species), our survey still indicates that the Paraná floodplain is highly diverse because the Pantanal is 55 times larger in area, extending over approximately 138 183km². Thus, the conservation units contained inside this stretch of the Paraná River can be considered important for the conservation of aquatic macrophyte diversity. Again, we emphasize that the number of species we recorded does not represent the real species richness because there are a number of habitats not yet investigated in this stretch.

The comparison of the region investigated in this study with other dammed stretches suggests the importance of the Paraná floodplain as a hotspot of macrophyte species diversity in this basin. In a survey of 18 reservoirs of the Paraná River and some of its main tributaries, Martins et al. (2008) found only 39 species of macrophytes. Even in the Itaipu Reservoir, which is dendritic (and thus, favorable for macrophyte colonization), a long-term dataset showed a total of 110 species (Mormul et al. 2010). The same conclusion can be made when we compare our data with reservoirs from other basins; for example, only 23 species of macrophytes were recorded in the Guri Reservoir, Venezuela (Vilarrubia & Cova 1993). The great variety of habitats found in the Paraná floodplain, together with the natural disturbance caused by seasonal oscillations in the water level, might explain these differences in relation to different reservoirs. On the other hand, the sampling effort was not controlled in these different surveys, and thus the results should be viewed with caution. However, this pitfall might be minimized because all of the investigations that we included in this report were floristic surveys, which tend to maximize the sampling within a region. In addition, there may be differences in the definition of macrophytes used in different surveys. Considering species composition, the assemblages of the Paraná floodplain can be considered as a sample of the aquatic flora from the Pantanal, as was previously pointed out by Thomaz et al. (2009). In fact, the most representative families in number of species are largely the same for both ecosystems (Poaceae and Cyperaceae, Onagraceae, Pontederiaceae, Plantaginaceae and Fabaceae). The families with few species in both ecosystems are also the same (Typhaceae, Cucurbitaceae, Maranthaceae, Haloragaceae, Solanaceae and Orchidaceae). Our cluster analysis also indicated that these three wetlands are the most similar amongst all ecosystems in our dataset, which could be due to their geographical proximity and to hydrological similarities (all areas are subjected to seasonal variation in the water level and also include a great variety of habitats). Furthermore, both wetlands belong to the same larger Paraná basin.

However, some families differ considerably in the number of species between these two wetlands. For example, there are only two species of Nymphaeaceae in the Paraná floodplain, whereas there are eight in the Pantanal that occur mainly in rain- fed shallow ponds and seasonal standing or slow flowing water in addition to in the river floodplain, except for Victoria amazonica, which grows in oxbow lakes. Similarly, 10 species of Characeae were found in the Pantanal, which is attributed to the alkaline and brackish waters in the Southwestern Pantanal (Bueno 1993, Pott & Pott 1997), but only two were identified in the Paraná floodplain, where acid soils predominate and charophytes do not thrive. The importance of the type of habitat in determining species composition can also be observed if we compare a survey carried out on lakes, reservoirs and wetlands in the Southern Paraná State (Cervi et al. 2009), only 200km away, that shares only 23% of macrophytes with the Paraná floodplain.

The low richness of aquatic epiphytes reported in the Paraná floodplain is related to the small sampling effort that has been carried out on floating meadows. Epiphytes usually colonize advanced stages of aquatic succession (Pott & Pott 2003). For example, surveys carried out by Tur (1972) and Neiff (1982) in the Middle Paraná (Argentina) identified 70 species of epiphytes. Even though both regions are on the Paraná River, the flood pulses differ between the Middle and the Upper Paraná basins, which may influence the accumulation of organic matter and, thus, the formation of floating-substrates, as well as the displacement of these islands, and this may explain the differences in the richness of epiphytes found in these wetlands.

The species number increase over time reported for the Paraná floodplain can be mainly attributed to the refinement of the taxonomic searches and identification carried out. In addition to this effect, we also considered a higher sampling effort in the Paraná floodplain, with collections made in habitats not previously investigated (e.g., rocks in the Paraná channel) and the arrival of new species (e.g. Hydrilla verticillata). The high level of species richness that we found indicates that the Paraná floodplain is still in a good conservation state, despite the strong anthropogenic pressures in the region related to changes in hydrometric levels, nutrient cycling and suspended solid loadings (Souza-Filho 2009).

However, despite the good status of conservation with respect to the aquatic flora, we contend that there is a concern related to the presence of two invasive species, H. verticillata and Urochloa subquadripara. The first is a submersed species native to Asia and the North of Africa, that colonizes the Paraná main channel and has a high competitive ability, threatening native species due its rapid regeneration following hydrological disturbances (Sousa et al. 2009, Thomaz et al. 2009). Its success in the Paraná main channel is associated with the same effect leading to an increase in the colonization by submersed species, i.e., the increase in water transparency and propagule pressures originating in the upstream reservoirs (Thomaz et al. 2009). Hydrilla verticillata has not yet colonized lakes in either the Baía or Ivinhema river habitats (Sousa et al. 2009). The second species, U. subquadripara, belongs to the family Poaceae, which contributes with several invasive species (Petenon & Pivello 2008). Although U. subquadripara has been rarely recorded in the Paraná floodplain, it reduced significantly the diversity of macrophytes in a lake close to the Baía River, the only place where it occurs with high biomass in this floodplain (Michelan et al. 2010). Disturbances associated with the oscillation in water levels may explain why this species is so rare in most habitats in the floodplain, but in light of its severe threat to macrophyte diversity, its monitoring is a priority, especially in the best preserved areas of this region.

The results of our cluster analysis indicate that Neotropical wetlands are different regarding macrophyte composition. Thus, we infer that such differences may be due to multiple factors, such as climate, flood regime and geography.

In fact, the most similar areas (Pantanal and Paraná floodplain) share many similarities: they are both large floodplains located from 80-160 m.a.s.l., have a great variety of habitats and are subjected to seasonal water level fluctuations (Agostinho & Zalewiski 1995, Vila da Silva 1995). However, the cluster analysis also shows that South American wetlands are diverse regarding macrophyte assemblages, and even ecosystems located in the same basin may differ considerably (e.g., the upper and middle/lower Paraná floodplains). The differences observed for these two floodplains may be accounted for by differences in their nutrient regimes (the Argentinean floodplains receive high phosphorus inputs from the Andes tributaries) and also to the types of habitats investigated (e.g., the widespread occurrence of floating meadows in Argentina with a high richness of epiphytes). As previously mentioned, the groups formed by the cluster analysis suggest that, though they represent geographically distant environments, such as the Amazonian floodplains and the Argentina plain, the sampled landscapes are determinant in forming groupings.

The accumulation curve reflects the differences found in the cluster analysis. In other words, the great differences among the South American surveys included in the cluster analysis indicate a high beta-diversity, leading to a lack of an asymptote in the accumulation curve. The total number of species found in all 12 surveys represents 87% of the number found by Chambers et al. (2008) for the Neotropical region. Despite the fact that our findings are close to the total number of Neotropical macrophytes, the lack of an asymptote, together with the high underestimation of true richness suggested by the Jack1 estimator, indicates that we are still far from describing the actual richness for this region. The number of plants to be described in Brazil, what may reflect the situation of South America, is considered very high (Pimm et al. 2010). In fact, there has been a clear lack of investigations conducted in pristine habitats in South America, such as in parts of the Amazon and the Andes, which are areas of high biodiversity. Future investigations at these sites, together with the description of new species (e.g., Bove et al. 2006, Amaral & Bittrich 2008), will certainly increase the number of species of Neotropical macrophytes recorded and give a better idea of the biodiversity provided by the great variety of ecosystems of this biogeographical region.

In summary, this report highlights the flora of different wetlands of South America and indicates that the actual species richness of macrophytes of this continent is far from being well understood. Our hypothesis sustains that macrophyte records, together with existing surveys, indicate a continuous need for carrying out increasing numbers of collections in new areas in the upper Paraná river-floodplain system and in other South American wetlands, as the number of species so far reported remains far from the predicted total. The checklist generated in this study is intended to support other research in wetlands and, in particular, to assure the continuity of ongoing long-term ecological programs, and it reveals a rich flora that is practically unknown to botanists and ecologists.

This study was funded by the Brazilian Council of Research (CNPq) through a Long Term Ecological Program (site number 6). S. M. Thomaz received a CNPq Productivity Research Grant and acknowledges this agency for long-term funding. A. Pott received a CNPq and CAPES Productivity Research. Additionally, F. A. Ferreira and R. P. Mormul acknowledge CNPq and CAPES (Coordenação de Aperfeiçoamento de Pessoal de Nível Superior) for furnishing Ph.D. fellowships, respectively.