Flooding-tolerant legume symbioses from the Brazilian Pantanal


Author for correspondence: Euan K. James Tel: +44 1382344741 Fax: +44 1382345893 Email:e.k.james@dundee.ac.uk


  •  Nodulated legumes in some of the flooded and seasonally flooded areas of the Pantanal Mato-Grossense wetlands in Brazil are described here.
  •  In the permanently flooded lakes (baias) of the Caracara national park Discolobium pulchellum, Mimosa pellita, Sesbania exasperata and Vigna lasiocarpa (syn. Phaseolus pilosus) were the most abundant, whereas close to Corumbá, at the edges of the river Paraguai, Neptunia spp. were also common. Adaptations that allow these legumes to fix N2 in a flooded environment included a floating growth habit, aerenchyma and nodulated adventitious roots.
  •  By contrast, Aeschynomene spp. (A. ciliata, A. denticulata, A. fluminensis, and A. sensitiva) were the dominant nodulated legumes in the seasonally flooded pastures of Nhumirim.
  •  Stem-nodulation was commonly observed, particularly on seasonally flooded Aeschynomene and seasonal/permanently flooded Discolobium spp., but also, in a modified form, on floating stems of V. lasiocarpa. The structures of stem and/or root nodules on Aeschynomene spp., Discolobium leptophyllum and V. lasiocarpa are described in detail, and nodulation by D. leptophyllum and Neptunia pubescens is reported for the first time.


The Pantanal Mato-Grossense is located in west Brazil and to the south of Amazonia (Fig. 1), and has a total area of 138 183 km2, excluding the approximately 10% of its area that lies in neighbouring Bolivia and Paraguay (da Silva & Abdon, 1998). It is the largest pristine wetland in the world and is of extreme importance in South America, both ecologically, due to its biodiversity, and economically, as a source of tourism, fresh water, fish, wildlife, forage and timber plants (Allem & Valls, 1987; Heckman, 1998). However, as with other tropical wetlands, this little-studied and fragile region is increasingly under threat from drainage and/or dyke-building for agriculture and housing, deforestation and pollution (Allem & Valls, 1987; Heckman, 1998). In the northern part of the Pantanal, south of Cuiabá (the capital of the state of Mato Grosso), the unique flora is particularly threatened by drainage and dyke-building for cattle ranching, and also from overgrazing (Allem & Valls, 1987; Heckman, 1998).

Figure 1.

Sketch map of the Pantanal Mato-Grossense, indicating collecting sites, including the Caracara national park (Ca), the Transpantaneira highway between Poconé and Porto Jofre (Tp), Corumbá, Fazenda Nhumirim (Nh), and Pousada Arara Azul (PA) located between Fazenda Leque and the Rio Abobral (bar, 100 km).

Tropical wetlands, such as the Pantanal, the Amazon region and the Orinoco basin generally experience two types of flooding, each having a unique flora: plants in the central regions close to the water courses are more or less permanently flooded, whereas those in the peripheries are subject to seasonal flooding. In both cases, many of the plants are nodulated legumes, and recent work has shown that not only can these fix N2 whilst flooded, but there may even be positive selection pressure for them to do so (Moreira et al., 1992; Barrios & Herrera, 1993a,b; Heckman, 1998; Loureiro et al., 1998; Saur et al., 1998; Sprent, 1999). This selection pressure may be due to the inherently low N-status of the heavily leached soils brought about by seasonal flooding, and also because, under more permanently flooded conditions, there is a decrease in the mineralization of organic matter and an increase in denitrification, all of which results in a shortage of available N (Buresh et al., 1980; Martinelli et al., 1992; Barrios & Herrera, 1993b). Moreover, and specifically in the case of the Pantanal, the headwaters of the rivers that feed it (Rio Cuiabá, Rio Paraguai) can be low in organic and inorganic nutrients. This lack of external input of N further exacerbates the oligotrophication of the wetland during the flooded period, and hence increases the demand for biological N2 fixation (Heckman, 1998). In a recent exhaustive study of the ecology of the Poconé region in the north of the Pantanal, Heckman (1998) concluded that the predominant source of fixed N was almost certainly nodulated legumes. Indeed, the latter have already been shown to contribute significantly to the N-balance of other tropical wetlands and rainforests, such as the Amazon region (Salati et al., 1982; Martinelli et al., 1992), French Guiana (Roggy et al., 1999a,b; Roggy & Prévost, 1999), and the flooded forests of the Orinoco basin (Barrios & Herrera, 1993b; Sprent, 1999), and are considered to be the main contributors of fixed N in all pristine ecosystems (Cleveland et al., 1999).

With the exception of the studies by Moreira et al. (1992), Barrios & Herrera (1993a), Barrios & Herrera (1993b) and Saur et al. (1998), there has been comparatively little research on tropical wetland legumes. Not all of them have been checked for an ability to nodulate: a study of nodulation in 172 legume species in the Amazon region of Brazil found that 56% of the reports of nodulation were new, with most of the nodulated legumes being found within the seasonally flooded Varzea/Igapo areas rather than in the drier Terra firme regions (Moreira et al., 1992). The legumes of the Pantanal have been studied even less than those in the Amazon, although the Pantanal has recently been the source of a number of new discoveries of flooding-tolerant legume symbioses, including a new genus of stem-nodulating shrubs, Discolobium (Loureiro et al., 1994), and a new report of stem nodulation by a species of Aeschynomene (A. fluminensis;Loureiro et al., 1995). Initial field studies have confirmed that not only are these legumes abundant in the Pantanal (Allem & Valls, 1987; Prado et al., 1994; de Faria & Lima, 1998; Heckman, 1998; Pinder & Rosso, 1998; Schessl, 1999) but also that they are extensively nodulated and hence may have substantial rates of N2 fixation (Heckman, 1998; Loureiro et al., 1998).

In this paper we describe the nodulated legumes that were encountered in a series of expeditions during the dry season of 1996 to the north and western regions of the Pantanal Mato-Grossense. In particular, we focused upon those found in permanently flooded sites within the Caracara IBAMA reserve and national park, and those found in seasonally flooded sites at Fazenda Nhumirim (Fig. 1). This paper constitutes the first report of nodulated legumes in the central, permanently flooded, regions of the Brazilian Pantanal.

Materials and Methods

Description of the collecting areas

In the wet season, the water level of the Rio Paraguai measured at Ladario, which is close to the port of Corumbá (Fig. 1), typically rises 4 m before the river breaks its banks and floods much of the adjacent region. In an average year the water level in the flooded region rises to a maximum of 7.5 m, and at this time, which is known locally as the Cheia, the total flooded area in the Pantanal can reach 110 000 km2. In most years this area includes the regions of Fazenda Nhumirim, Poconé and the areas adjacent to the Transpantaneira highway (Allem & Valls, 1987; Prado et al., 1994; Heckman, 1998). In the dry season (Seca) the floodwater usually recedes from the peripheral areas of the Pantanal and it is only the central parts that remain flooded, particularly those close to the courses of the major rivers (Rios Paraguai, Cuiabá, Taquari, Nabileque, Negro and Miranda). The Caracara Parque Nacional do Pantanal is located at the junction of the Rio Paraguai and Rio Cuiabá (Fig. 1), and has a total area of 145 000 ha. As it is so close to the rivers, much of the national park is permanently flooded. In the Seca this type of area consists of numerous small shallow lakes or baias that are separated from each other by narrow strips of land, whereas in the Cheia it is made up of large lakes interspersed with islands.

The first expedition in our study was made to Caracara (Table 1, Fig. 1) but plants were also collected en route from permanently flooded water courses and seasonally flooded pastures along the Transpantaneira highway between Poconé and Porto Jofre. In the second expedition we examined the seasonally flooded pastures of the Fazenda Nhumirim to the south of Corumbá, as well as the banks of the Rio Paraguai close to Corumbá itself (Table 1, Fig. 1).

Table 1.  Collection areas in the Northern (Area 1) and Western (Area 2) Pantanal Mato-Grossense
LocationAltitude (m)Mean annual rainfall (mm)Mean annual temp. (°C)Comments
Collection Area 1    
Caracara (Ca)    
17 51 S, 57 24 W 90126226.8Permanently flooded baias at the confluence of the Rio Paraguai and Rio Cuiabá
Transpantaneira (Tp) (Poconé– Porto Jofre)  
16 23 S, 56 38 W – 17 38 S, 56 44 W100127525.8Main highway south of Cuiabá passing through a mosaic of seasonal savannahs and grasslands, flood-free ridges and wood islets, marshes and backswamps, and seasonal channels
Collection Area 2    
Corumbá (Co) (Rio Paraguai)  
19 01 S, 57 39 W141107025.1Port city on the Rio Paraguai
Fazenda Nhumirim (Nh)    
18 59 S, 56 39 W 98118325.5Cattle ranch. Seasonally flooded baias or permanent and seasonal ponds. Rainfed
Pousada Arara Azul (PA) (Fazenda Leque – Rio Abobral)
19 14 S, 57 01 W – 20 23 S, 57 03 W 85119826.6Ranchland. Under the influence of the Rio Negro and the false Rio Abobral (an ancient riverbed)


Nodules for microscopy were collected in the field, where they were initially fixed in 2% (v/v) glutaraldehyde in sodium bicarbonate. As soon as the samples were returned to the laboratory at UFMT, Cuiabá, they were immediately transferred to electron microscopy grade glutaraldehyde (Sigma-Aldrich Company Ltd., Poole, UK) diluted to 5% (v/v) in 50 mM phosphate buffer, pH 7.0. Further processing was performed at the University of Dundee according to James et al. (1992a). The nodules were either dehydrated and infiltrated in LR White resin for light microscopy or, for transmission electron microscopy (TEM), they were postfixed in 1% (w/v) osmium tetroxide prior to dehydration and infiltration. Sections were taken on a Reichert OM U3 ultramicrotome at 1 µm thickness for light microscopy and at 70 nm thickness for TEM. Sections for light microscopy were collected on glass slides and stained in toluidine blue, whereas TEM sections were collected on copper grids and stained in uranyl acetate and lead citrate. Light and TEM sections were viewed and photographed using an Olympus BH2 light microscope and a JEOL 1200EX TEM, respectively.

Results and Discussion

Nodulated legumes found in the north-western Pantanal Mato-Grossense


Four nodulated wetland species were found: A. ciliata, A. denticulata, A. fluminensis and A. sensitiva (Table 2). Two other N2-fixing species, A. americana and A. rudis, are also common in the Pantanal (Pott & Pott, 2000), but nodulated examples of these were not encountered in the present study. Of the four species found here, A. fluminensis is probably the best described (Allem & Valls, 1987; Prado et al., 1994; Loureiro et al., 1995; Heckman, 1998; Schessl, 1999, Fig. 2a,b). In the south-eastern Pantanal it has been recognized as a potential indicator species for ‘marsh ponds’ and ‘waterlogged basins’, that is shallow depressions that have waterlogged soils in the dry season and up to 0.8 m of water in the wet season (Pinder & Rosso, 1998). Although our data are only qualitative, they suggest that A. fluminensis is equally common in the seasonally flooded depressions in the pastures of the Fazenda Nhumirim, and in the fields adjacent to the Transpantaneira highway near Poconé (Fig. 2a,b). Loureiro et al. (1995) have suggested that A. fluminensis is well-adapted to seasonal flooding as, unlike most legumes, it forms active N2-fixing nodules on the submerged stem. After the flooding recedes these stem nodules remain functional, together with the root nodules. Because of its ability to survive and fix N2 whilst flooded, Heckman (1998) considers A. fluminensis to be amongst the most important plants contributing to the N-cycle of the Pantanal.

Table 2.  Some nodulated legumes observed in various locations in the north-western, central south and south-western Pantanal Mato-Grossense (see Table 1,Fig. 1) during the dry season (August 1996)
  • a

    Flooded = plants partially submerged in water or rooted in saturated soil.

  • b

    b Nodules on floating stems but attached vascularly to the bases of subtending adventitious roots.

  • c

    c First report of nodulation.

Aeschynomene ciliataVog.NhStemDry
Aeschynomene denticulataRuddCoStemFlooded
Aeschynomene fluminensisVell.Nh, PA, TpStem, RootDry, Flooded
Aeschynomene sensitivaSw.NhRootDry
Discolobium leptophyllumBenth.TpStemc, RootcFlooded
Discolobium pulchellumBenth.Ca, TpStem, RootFlooded
Mimosa pellitaH.B. ex Willd.CaRootFlooded
Mimosa polycarpaKunthCaRootFlooded
Neptunia plenaBenth.CoRootFlooded
Neptunia prostrata(Lam.) Baill.CoStemb, RootFlooded
Neptunia pubescensBenth.CoRootcDry
Sesbania exasperataH.B.K.CaStemb, RootFlooded
Vigna lasiocarpa(Benth.) VerdcCa, CoStembc, RootFlooded
(syn. Phaseolus pilosus)(H.B.K.)   
Figure 2.

(a) Large (approx. 1.5 m), woody Aeschynomene fluminensis plant growing within a partially flooded field adjacent to the Transpantaneira highway close to Poconé. The plant shown here is heavily laden with fruit (small arrows). Note the cattle in the background (large arrow). (b) Aeschynomene fluminensis growing in dry conditions in a field adjacent to the Transpantaneira highway close to Poconé. Note the distinctive yellow flowers (small arrows), and also that it has been grazed (large arrow). (c) Stand of Discolobium pulchellum rooted in a baia in the Caracara national park. The depth of the water is approx. 1.5 m, and the submerged parts of the stems and the roots are nodulated (Fig. 2d). (d) Nodules on the stem of Discolobium leptophyllum (arrows). As with D. pulchellum these only form on the submerged portions of the stem. Note the lenticellular material on the stems (*). Bar, 1 cm.

Much less information is available about the other Aeschynomene spp., although stem nodules have been reported on all of them (Eaglesham & Szalay, 1983; Alazard, 1985; Stegink & Vaughn, 1988; Boivin et al., 1997). In the present study, they were often found growing adjacent to A. fluminensis (Table 2) and it is possible that their similar habitats, particularly the seasonal flooding, have resulted in similar adaptations, including the presence of stem nodules. Interestingly, only root nodules were found on A. sensitiva, and this may have been due to the relatively dry conditions in which it was found. De Faria & Lima (1998) also encountered this species in the Pantanal (close to Corumbá) and confirmed its ability to form root nodules in pot experiments. However, they did not report stem nodules in either field or pot-grown plants and did not report whether the plants were found in flooded or nonflooded conditions. In Senegal, A. sensitiva is known to develop unique ‘collar’ nodules around the stem after flooding (Boivin et al., 1997), and it remains to be seen if A. sensitiva in the Pantanal also has this ability. Aeschynomene denticulata and A. sensitiva are also abundant in seasonally flooded Chaco in Paraguay, although no mention was made of their nodulation status (Hacker et al., 1996).

Nodules on A. fluminensis have been described in detail by Loureiro et al. (1995), but this is the first report of the ultrastructure of stem or root nodules on A. ciliata, A. denticulata and A. sensitiva (Figs 3a–d, 4a–d). Stem nodules on Aeschynomene ciliata and A. denticulata are similar in appearance to those on A. indica (Arora, 1954; Yatazawa & Yoshida, 1979) and A. afraspera (Alazard & Duhoux, 1987) in that they arise as bumps on the stem (e.g. A. denticulata;Fig. 3a). This contrasts with A. fluminensis (Loureiro et al., 1995) and Sesbania rostrata (Dreyfus & Dommergues, 1981; James et al., 1996; Boivin et al., 1997), where nodules are more prominent and have a distinct stalk attaching them to the subtending stem. The infected zones of both stem and root nodules on A. ciliata, A. denticulata and A. sensitiva were typically aeschynomenoid (Sprent et al., 1989), with few uninfected cells (Figs 3a–d, 4a,b). In some A. sensitiva root nodules, meristematic tissue was observed at the ends of lobes of infected tissue (Fig. 3c); as no infection threads were observed it is likely that new infected cells are formed via division of already infected ones. This is commonly reported in other aeschynomenoid types, including stem nodules on A. afraspera (Alazard & Duhoux, 1990), although occasional infection threads were observed in nodules of A. fluminensisLoureiro et al. (1995), and in aeschynomenoid Discolobium nodules (Loureiro et al., 1994; this study).

Figure 3.

Stem and root nodules on various Aeschynomene spp. The infected zones (I) of all these are typically aeschynomenoid, that is there are no uninfected cells amongst the infected ones. (a) Longitudinal section (LS) of a stem nodule on A. denticulata. The nodule is essentially a ‘bump’ on the surface of the stem (S); it is not attached by a stalk, as in stem nodules on species such as A. fluminensis. Bar, 100 µm. (b) LS of a root nodule on Aeschynomenesensitiva. Although it is typically aeschynomenoid, on one of the lobes of the nodule there is a meristematic region (M) where new infected cells are forming (Fig. 3c). R, root. Bar, 100 µm. (c) Higher magnification of a meristematic lobe of the A. sensitiva root nodule shown in Fig. 3(b). All the newly divided cells in the infected zone are infected, but there are no infection threads visible. Note that many of the bacteria in these cells are rod-shaped (arrows) and not the characteristic spherical shape of bacteroids in the mature infected cells (see Fig. 4a–c). Bar, 10 µm. (d) Light micrograph of the cortex and infected zone of a stem nodule on A. ciliata. As with A. denticulata, but even more so on this plant; these nodules appear as almost subliminal bumps on the surface of the stem. Numerous chloroplasts are visible within the cortical cells (arrows). These were also apparent in A. denticulata stem nodules (Fig. 4d). E, epidermis. Bar, 20 µm.

Figure 4.

Details of stem and root nodules on various Aeschynomene spp. (a) Light micrograph of the cortex and infected zone (I) of a root nodule on A. sensitiva. Note that many of the cortical cells contain chloroplast-like structures (small arrows), and also that the bacteroids in the infected cells are large and spherical (‘coccoid’) in shape. E, epidermis. Bar, 10 µm. (b) Transmission electron micrograph (TEM) of infected cells in a stem nodule on A. ciliata. As with A. sensitiva, these are full of large coccoid bacteroids (shown with an asterisk), most of which have electron-transparent vacuoles and/or electron-dense nuclear material within them (Fig. 4c). N, nucleus. Bar, 2 µm. (c) TEM of symbiosomes in an A. denticulata stem nodule. Vacuoles (small arrows) and nuclear material (large arrows) can be seen within the ‘coccoid’ bacteroids. An asterisk shows the peribacteroid space. Bar, 1 µm. (d) TEM of a chloroplast (arrow) within a cortical cell in an A. denticulata stem nodule. Note the stacked thylakoids (T). W, cell wall; S, starch granule. Bar, 500 nm.

With the exception of A. fluminensis (not shown, but see Loureiro et al., 1995), nodules from all the Aeschynomene spp. in the present study, had large oval or spherical bacteroids, with only one bacteroid per symbiosome (Fig. 4a–c). These ‘coccoid’ bacteroids are commonly observed within nodules on other aeschynomenoid species, such as peanut (Arachis hypogaea) root nodules (Sen et al., 1986), and in stem and root nodules on other Aeschynomene spp., for example A. indica (Yatazawa et al., 1984; Vaughn & Elmore, 1985; Evans et al., 1990). As in A. indica (Yatazawa et al., 1984; Vaughn & Elmore, 1985; Evans et al., 1990), bacteroids in A. ciliata and A. denticulata nodules contained distinctive electron-transparent vesicles and electron-dense nuclear material (Fig. 4b,c). In Aeschynomene spp., the unusual appearance of the bacteroids may be related to their ability to photosynthesize both ex planta and in planta, as many isolates from stem nodules contain bacteriochlorophyll a (Bchl a) and carotenoids (Evans et al., 1990; Hungria et al., 1992; Boivin et al., 1997; Fleischman & Kramer, 1998; Molouba et al., 1999).

Chloroplasts were observed in the uninfected cell layers of the green stem nodules on A. ciliata and A. denticulata (Figs 3d, 4d), and have been shown previously in A. fluminensis stem nodules (Loureiro et al., 1995). Interestingly, chloroplast-like structures were also seen in the root nodules of A. sensitiva (Fig. 4a). Chloroplasts are considered to be a distinguishing feature of stem nodules, and it has been suggested that their close proximity to the N2-fixing cells may be responsible for the high nitrogenase activities reported in stem-nodulated legumes (Ladha et al., 1992; Boivin et al., 1997; Fleischman & Kramer, 1998; Loureiro et al., 1998). On the other hand, chloroplasts have so far been shown to be functional only in aerial stem nodules on A. indica (Evans et al., 1990), A. scabra (Hungria et al., 1992) and S. rostrata (James et al., 1998).


Large stands of D. pulchellum (2–3 m in height) were found throughout the Caracara national park, where they were usually rooted within permanently submerged mud in the shallow (1–2 m) baias (Fig. 2c). Discolobium pulchellum and smaller numbers of Discolobiumleptophyllum were also seen growing within rivers and flooded pastures adjacent to the north–south stretch of the Transpantaneira highway between Poconé and Porto Jofre. Nodules were observed on the flooded stems and roots (Fig. 2d), and the stems were often covered in profuse aerenchyma (Fig. 2d). Compared with Aeschynomene spp. there is very little information about nodules on Discolobium spp. Nodulation (of stems and roots) has so far been reported in only two of the eight species (D. pulchellum and D. psoralaefolium;Loureiro et al., 1994). In the present study, confirmation of nodules on D. leptophyllum brings this total to three. Although they appear to share similar (i.e. flooded) habitats with A. fluminensis and the other Aeschynomene spp. (Loureiro et al., 1995; Heckman, 1998), Discolobium spp. (especially D. pulchellum) are almost exclusively hydrophytic and have not been reported from seasonally flooded basins (Pinder & Rosso, 1998; Heckman, 1998; Loureiro et al., 1998; Pott & Pott, 2000). They were, however, occasionally observed in Caracara rooted in mud on islets exposed after the floodwater had receded. Our study agrees with that of Prado et al. (1994) which described D. pulchellum among those species that are always present in the low-lying flooded areas bordering the Transpantaneira that do not usually dry up, even in the dry season. In the case of D. leptophyllum, the plant is somewhat smaller than D. pulchellum and it appears to be adapted to living in shallower waters. Indeed, the specimens that are shown here (Figs 2d, 5a–c) were not found in the relatively deep baias of Caracara but at the edges of narrow water courses adjacent to the Transpantaneira.

Figure 5.

(a) Section through a young root nodule on a submerged root of Discolobium leptophyllum. Note that the stele (S) of the root has secondary thickening (black arrow) and that the cortex contains lysigenous aerenchyma (L). The infected zone (I) of the nodule is essentially aeschynomenoid with the few uninfected cells being confined to discrete groups or files (white arrows). Note that the nodule cortex (shown by an asterisk) is thick and spongy in appearance. Bar, 100 µm. (b) Detail of part of the infected tissue of a young D. leptophyllum nodule showing an infection thread-like structure (large arrow) entering one of a small group of uninfected cells (shown with an asterisk) within the infected zone. Note that the bacteroids in the infected cells are generally rod- or oval-shaped (small arrows). Bar, 5 µm. (c) Transmission electron micrograph (TEM) showing an infection thread-like structure (shown with an asterisk) within a cell in a young D. leptophyllum nodule. Bacteria are being released from it (arrows). V, vacuole. Bar, 1 µm. (d) Longitudinal section of young root nodule on Neptunia pubescens found in wet (but not flooded) soil. This is a typical indeterminate nodule, with a distinct meristem (M), invasion (I) and N2-fixing zone (asterisk). It is also typically mimosoid, that is with a cortical phellogen (P) that can expand into a secondary aerenchyma. Bar, 50 µm. (e) Nodule attached to an adventitious root (R) on a floating stem (S) of Sesbania exasperata. The meristem (M) is visible at the tip of the nodule. There is a layer of sclereids and tannin/phenolic-containing cells (short arrows) separating the inner cortex from the outer cortex. Note the vascular traces (long arrow) entering the nodule from the adventitious root. Bar, 100 µm.

Although Discolobium nodules are basically aeschynomenoid (Fig. 5a), they differ from Aeschynomene nodules in a number of features, both structural and physiological (Loureiro et al., 1994; Martins et al., 2000). The most notable difference is that the stem nodules (but not the root nodules) on D. pulchellum senesce rapidly (within 24 h) upon exposure to air, suggesting that they are specifically adapted to permanent submergence/inundation (Loureiro et al., 1994). This is also the case with D. leptophyllum, whose stem nodules are largely identical to those on D. pulchellum (Martins et al., 2000; data not shown). The processes involved in the infection and development of stem and root nodules on Discolobium have yet to be established, but the present study has shown that infection thread-like structures are abundant in young root nodules on D. leptophyllum, being localized within discrete clusters or files of uninfected cells within the infected tissue (Fig. 5b,c). Again, this shows that Discolobium nodules are different from those on Aeschynomene, in which infection threads are rarely, if ever, seen (Sprent et al., 1989; Alazard & Duhoux, 1990). Indeed, the short wide infection thread-like structures reported here actually resemble those reported within nodules on Lupinus albus (James et al., 1997), a species not directly related to Discolobium, but which also has nodules with an infected zone that contains few, if any, uninfected cells. The bacteroids in nodules of D. leptophyllum (this study; Fig. 5b) and D. pulchellum (Loureiro et al., 1994) were always rod or oval-shaped and the coccoid bacteroids so typical of Aeschynomene nodules (see earlier) have so far not been observed in Discolobium spp. Another interesting feature of young root nodules of D. leptophyllum was their occasional formation on mature, secondarily thickened roots in which the cortex was replaced by a lysigenous aerenchyma (Fig. 5a). The latter is generally considered to be a feature of deep-rooted wetland species, such as rice (Oryza sativa) (Jackson & Armstrong, 1999), and its occurrence here supports the hypothesis that Discolobium spp. are specifically adapted to permanent flooding/inundation (Loureiro et al., 1998).


Two nodulated shrubby species were encountered at Caracara; M. pellita and M. polycarpa. Both are pan-tropical and have previously been reported as being nodulated (Barrios & Gonzalez, 1971; Allen & Allen, 1981). In the Pantanal, de Faria & de Lima (1998) found nodules on both species growing in the Corumbá region but did not comment on whether the plants were flooded or not. In Caracara, Mimosa pellita (syn. M. pigra L.) was particularly abundant, growing profusely along river banks and the edges of baias. Submerged portions of stems were covered in lenticellular material/aerenchyma and the adventitious roots arizing from these were nodulated. This species is distributed throughout tropical South America and is known to be flooding-tolerant. For example, nodulated plants were reported growing in the seasonally flooded forests of the Orinoco basin (Barrios & Herrera, 1993a), in Varzea/Igapo areas of the Amazon (Moreira et al., 1992), and in the Pantanal (Pott & Pott, 1994; Heckman, 1998; Schessl, 1999). A brief description of M. pellita nodules grown under flooded conditions was given by James et al. (1995). Mimosa polycarpa is less well described, but in Caracara it was found growing in wet (but not flooded) soil and the roots had very small nodules (< 2 mm in diameter) (data not shown).


Most members of this genus are pan-tropical, and are particularly abundant in wetland regions of Central and South America (Windler, 1966), including the Pantanal (Pott & Pott, 2000). Interestingly, in the present study no Neptunia spp. were found in the Caracara national park. There is no obvious explanation for this, but it could be attributed partly to the action of waves on the wide, open waters. These may inhibit the growth on river banks of the relatively sensitive N. plena (James et al., 1992b). In addition, the existing floating mats of vegetation (as mentioned previously) may be too dense for the pioneering, free-floating N. prostrata (syn. N. natans, N. oleracea) (Subba Rao et al., 1995; Pott & Pott, 2000). By contrast to Caracara, Neptunia spp. were abundant at the other collecting sites, such as close to Corumbá, where large stands of N. plena were encountered rooted in the flooded mud at the edges of the Rio Paraguai. These plants were profusely nodulated, and the nodules were located on submerged, spongy tap roots, as well as on adventitious roots that had arisen from flooded stems. Some nodulated N. prostrata plants were also found in this region close to Corumbá. Neptunia prostrata differs from N. plena in having a floating growth habit, with the nodules forming at the base of adventitious roots emerging from very spongy, floating stems (Schaede, 1940; Subba Rao et al., 1995). The structure of N. prostrata nodules has been described by Schaede (1940) and Subba Rao et al. (1995), and that of N. plena by James et al. (1992a). In both species, the nodules are indeterminate with a distinct meristem typical of mimosoid nodules. On flooded plants, nodules are connected to the stem via profuse aerenchyma.

A few plants of N. pubescens were found growing in wet (but not flooded) soil on seasonally flooded lake edges 20–40 km south of Corumbá, close to the Bolivian/Paraguaian Chaco regions. Extremely small nodules (1 mm diameter) were seen on the roots, and the structure of these (Fig. 5d) was typical of Neptunia (Schaede, 1940; James et al., 1992a; Subba Rao et al., 1995). However, they did not have any obvious aerenchyma, probably because they were found in nonflooded conditions (James et al., 1992a). This is the first report of nodulation by N. pubescens, although it remains to be seen if it can also be nodulated when flooded. Interestingly, Hacker et al. (1996) also found N. pubescens in seasonally flooded Chaco in nearby Paraguay, but did not comment on its nodulation status.


Although they are often reported in the Pantanal (Heckman, 1998; Schessl, 1999; Pott & Pott, 2000), Sesbania spp. were not common at any of our collecting sites. An exception was S. exasperata, which was found in Caracara, usually as a component of floating mats of vegetation in the baias, but also rooted in flooded soil on some of the islands. S. exasperata is known throughout tropical and subtropical South America. For instance, it has recently been reported in seasonally flooded Chaco in Paraguay (Hacker et al., 1996) and in the Poconé region of the Pantanal (Heckman, 1998). It has long been known that many Sesbania spp. are both hydrophytic (Arber, 1920) and nodulated (Allen & Allen, 1981; Ladha et al., 1992; Boivin et al., 1997). In the case of S. exasperata, a number of authors have alluded to its potential ability to fix N2 (e.g. Heckman, 1998; Franco et al., 1998; Pott & Pott, 2000), but to our knowledge this is the first actual description of its nodules. The nodules were usually found on floating stems at the base of adventitious roots, in a manner similar to nodules on N. prostrata (Schaede, 1940; Subba Rao et al., 1995). They were often green in colour and superficially resembled the stem nodules on S. rostrata (Dreyfus & Dommergues, 1981; James et al., 1996). However, when nodules subtended by stems were sectioned it was revealed that they were actually connected vascularly to the adventitious roots (Fig. 5e) and not to the stems, and therefore they cannot be described as ‘stem’ nodules according to the definition of James et al. (1992a). Moreover, although they appeared green the nodules did not contain the chloroplasts (data not shown) that are such a significant feature of stem nodules on S. rostrata (Dreyfus & Dommergues, 1981; Duhoux, 1984; James et al., 1996, 1998). Except for their close proximity to the stem, in all other respects S. exasperata root nodules were similar to those observed on other Sesbania species (Harris et al., 1949; Brown & Walsh, 1994; Ndoye et al., 1994; Rana & Krishnan, 1995), being essentially spherical in shape but with a meristematic region at the tip, and with a cortex containing a layer of sclereids and tannin-containing cells (Fig. 5e).

Vigna lasiocarpa

V. lasiocarpa (syn. Phaseolus pilosus) is common in wetland regions of South America (Pott & Pott, 2000), and root nodules were reported on plants in swampy savannahs in Venezuela (Barrios & Gonzalez, 1971). In the present study, V. lasiocarpa was abundant in the baias of Caracara where it was a component of floating mats of vegetation (Fig. 6a). It was also found in wet soil on islands, where it often grew alongside Sesbania exasperata and Mimosa pellita. The plants were well nodulated, with nodules being present both on adventitious roots arizing from floating stems as well as on the stems themselves (Fig. 6b). The ‘stem nodules’ were green in colour, as were the root nodules that were close to the surface and exposed to daylight (Fig. 6b). Those root nodules that were more deeply submerged and/or covered in vegetation were usually white or pinkish (not shown). The stem nodules were always associated with adventitious roots (Fig. 6b), and when they were sectioned it could be seen that the nodules were attached to the subtending stem and the adventitious root via a small stalk of tissue containing a vascular trace (Fig. 7a). This is very similar to the vascular connections recently reported for ‘stem nodules’ on the temperate wetland legume, Lotus uliginosus (James & Sprent, 1999). Those authors concluded that L. uliginosus was ‘intermediate’ between true stem-nodulated legumes, such as Aeschynomene spp. Discolobium and Sesbania rostrata (as previously discussed), and those legumes that have nodules connected vascularly to the base of the subtending adventitious root, such as Neptuniaprostrata (Schaede, 1940; Subba Rao et al., 1995), N. plena (James et al., 1992a) and Sesbania exasperata (this study). From the present study of their vascular system it would appear that stem nodules on Vigna lasiocarpa fit into the same intermediate category as those on L. uliginosus. However, in support of their categorization as ‘stem nodules’, a feature of V. lasiocarpa nodules was the presence of numerous chloroplast-like structures within the cortex surrounding the infected tissue (Fig. 7b,c). Further studies are needed to determine if these structures really are chloroplasts and if so, whether they are photosynthetic (James et al., 1998).

Figure 6.

(a) Vigna lasiocarpa growing within a mat of vegetation floating on a baia in the Caracara national park. Note the yellow flower and the climbing habit. (b) Pale green nodules on the stem (black arrows) and adventitious roots (white arrow) of V. lasiocarpa. Note that there are adventitious roots adjacent to the ‘stem’ nodules. Bar, 2 cm.

Figure 7.

Structural study of nodules on Vigna lasiocarpa. (a) Nodule (N) attached vascularly to a floating stem via a short stalk of tissue (arrow) that is also connected vascularly to a root (not shown). With the exception of being attached to a stem, this nodule is very similar in structure to mature nodules on the roots of terrestrial Vigna spp. that is it is determinate, with no apical meristem. Note the spongy pith in the stem (P); this allows it to float. Bar, 100 µm. (b) Detail of cortex (asterisk) and infected tissue (I) of a young nodule attached to a floating stem. This nodule was green (Fig. 6b) and has chloroplast-like structures within the cortical cells (arrows). E, epidermis. Bar, 10 µm. (c) Transmission electron micrograph (TEM) of chloroplast-like structure (arrow) in a cortical cell of a green nodule (Fig. 6b). The structure has starch granules within it (S). W, cell wall. Bar, 500 nm. (d) TEM of infected cell in a young green nodule attached to a floating stem (Fig. 6b). The cell is packed with bacteroids, with only 1–3 bacteroids per symbiosome (*). Note the large nucleus with nucleolus (N). Bar, 1 µm.

Nodules on terrestrial Vigna spp., for example V. unguiculata, are not usually tolerant of the sudden imposition of prolonged flooding (Minchin & Summerfield, 1976), although they will acclimatize to low pO2 if they are grown under these conditions from germination, producing profuse lenticellular tissue (Dakora & Atkins, 1990a,b). However, although V. lasiocarpa lives in a flooded habitat, its nodules were located mainly on floating stems and hence would probably not actually be subjected to particularly low pO2 values. This probably explains the absence of profuse lenticels (Fig. 7a,b). Certainly, the infected tissue of V. lasiocarpa nodules showed no obvious signs of stress and appeared to be fully effective (Fig. 7a,b,d). In this respect V. lasiocarpa is similar to N. prostrata (Schaede, 1940; Subba Rao et al., 1995) and S. exasperata (this study).

Vigna lasiocarpa is similar in morphology to the closely related species, Vigna longifolia (syn. Phaseolus trichocarpa), which is also found in the Pantanal (Pott & Pott, 1994). However, V. lasiocarpa differs from V. longifolia in having hairy pods, and also in that the latter usually grows only in seasonally flooded or dry areas (Pott & Pott, 1994). Interestingly, V. longifolia is also found in other wetland regions of South and Central America, such as the humid Chaco of Paraguay (Hacker et al., 1996) and in ‘permanently wet’ locations in Puerto Rico (Dubey et al., 1972).


Potential importance of flooding-tolerant legumes to the ecology of the Pantanal

This study has presented qualitative data in support of the hypothesis that nodulated legumes are abundant in the Pantanal (Heckman, 1998; Loureiro et al., 1998). It has also shown that the type of symbiosis may differ according to the flooding regime and that there are adaptations to two types of flooding: permanent and seasonal. The former is best exemplified by Discolobium pulchellum, whose stem nodules cannot form or function without being surrounded by water or wet soil, and the latter by various Aeschynomene spp. on which stem nodules form during the flooding period and remain functional after the flooding recedes. There are also a number of semiaquatic legumes, such as Mimosa pellita and Neptunia plena that are rooted in mud at the peripheries of the river channels and permanently flooded baias. These are not stem-nodulated, but instead have their nodules on adventitious roots that form on their flooded stems. Finally, in the permanently flooded regions of the Pantanal there is a great abundance of legumes that are not actually ‘flooding tolerant’ as such, but have a floating habit that prevents their nodules (borne on stems and/or adventitious roots) from being substantially submerged. Good examples are Neptunia prostrata, Sesbania exasperata and Vigna lasiocarpa.

Initial studies have shown that the nodulated legumes encountered in the present study are potentially of great importance to the ecology of the Pantanal. For example, in addition to their N2-fixing ability, all of them are very palatable and readily foraged by indigenous fauna and/or cattle. This is especially true of the Aeschynomene spp. (Fig. 2a,b, this study; Allem & Valls, 1987; Prado et al., 1994; Loureiro et al., 1995; Hacker et al., 1996; Heckman, 1998; Pott & Pott, 2000) and the Discolobium spp., whose submerged stems are eaten by the herbivorous fish pacu (Piaractus mesopotamicus) and by capybara (Hydrochaeris hydrochaeris). The indigenous people of the Pantanal also consume Aeschynomene and Discolobium spp. for medicinal purposes (Pott & Pott, 2000). Owing to their ability to fix N2 and to tolerate flooding, A. fluminensis, A. sensitiva and S. exasperata are showing promise outside the Pantanal as pioneer species for recovery of flooded ponds filled with residues of bauxite minings in regions such as Porto Trombetas in the Amazon (Franco et al., 1998). However, in order to determine more exactly the importance of nodulated legumes to the ecology of the Pantanal and other tropical wetlands, it will be necessary to quantify their populations in regions with different flooding regimes. In addition, 15N natural abundance studies, such as those that have been undertaken in the Amazon floodplain (Martinelli et al., 1992) and in the rainforests of French Guiana (Roggy et al., 1999a), are urgently needed in order to assess the potential contribution of nodulated legumes to the N-budget of this largely oligotrophic region.


We thank R. M. M. Crawford and S. M. de Faria for helpful discussions, and M. Gruber, M. Kierans, S. G. McInroy for technical assistance. EKJ was funded by NERC grant GR3/9037′A’, and AAF, AP, CMM, VJP by the Brazilian national research foundation (CNPq). We gratefully acknowledge IBAMA for permission to visit the Caracara national park, and Benjamin D. Silva for taking care of us during our visit.