Vegetated earthmounds in
tropical savannas of
Central Brazil:  A synthesis

With special reference to the Pantanal do Mato Grosso

Victor M. Ponce and Catia N. da Cunha

Online version 2015

[Original version 1993]


The vegetated earthmounds which dot the great expanse of seasonally flooded grasslands of the Pantanal do Mato Grosso and other ecosystems of Central Brazil are examined from both abiotic (hydrological/geomorphological action) and biotic perspectives (termite action). The nature of the earthmounds is explained in terms of the competition between savanna woodland (cerrado) and open grassland (campo) communities. The difference in scale evident between micromounds (murundus) and macromunds (capões) suggests the action of alternate environmental forces.


The Pantanal do Mato Grosso is one of the largest tropical wetland ecosystems in the world, encompassing an area of about 140,000 km2 in the upper basin of the Rio Paraguay, in central-western Brazil (Fig. 1). "Pantanal" is Portuguese for great swampland, implying that the region is subject to extensive and recurrent flooding. However, large portions of the Pantanal are flooded only during the annual crest of the Rio Paraguay and its tributaries, and much non-flooded terra firma is interspersed throughout the region.

The area is flat to slightly undulating, with most of it lying about 100 m above mean sea level. The mixture of permanent swamp, seasonal swamp, and terra firma, as well as the contiguity of the Pantanal to other major South American ecosystems (the Amazon rainforest to the north and northwest, the tropical savannas of Central Brazil to the east, and the dry Chaco-type forest of neighbouring Bolivia to the southwest), has contributed to the richness and variety of its vegetation. In the existing literature, the vegetation is often marked as a single unit and referred to as the "Pantanal complex." Actually, this complex is a mosaic of many different communities, with frequent abrupt changes usually correlated with topography, and with many ecotones (Prance and Schaller, 1982).

Fig. 1.  Location of the Pantanal do Mato Grosso, Brazil.

A distinct geomorphological feature of the Pantanal landscape is the vegetated earthmounds (of various sizes and shapes, and approximately circular or elliptical in plan) which dot the great expanse of seasonally flooded grasslands. Brazilians have used various terms to describe these curious earthmounds: murundu, covoal, morrote, cocuruto, moncheio, capãozinho, and ilha (Oliveira Filho and Furley, 1990). The larger earthmounds, which can reach diameters of more than 100 m, have been referred to as capoes, and distinguished from the capões de aterro or aterros de bugre, which are manmade (Cunha, 1990).

Cursory accounts of these Brazilian earthmounds have appeared in the English-language literature in the last three decades (Cole, 1960; Askew et. al., 1970; Ratter et. al., 1973; Eiten, 1975; Prance and Schaller, 1982; Furley and Ratter, 1988). These attribute the genesis and evolution of the earthmounds to abiotic and biotic factors; however, the precise role of these mechanisms remains to be elucidated. Unlike the adjacent open grasslands which do not support woody vegetative species, the Pantanal earthmounds feature woody vegetation, revealing their close association with the savanna woodlands or cerrados of Central Brazil. In some instances, the earthmounds also feature xeric vegetation such as cactus and other succulents, revealing the influence of the drier Chaco-type climate to the southwest. Rather than having an endemic flora, the Pantanal is made up of elements from mato (semideciduous forest transitional to the Amazon rainforest), campo (open grassland), cerrado (savanna woodland), and caatinga (desert scrub).

This paper reviews current knowledge on the vegetated earthmounds of the tropical savannas of Central Brazil. The nature and survivability of the earthmound habitat is explained in terms of the competition between cerrado and campo plant communities. Several earthmound classifications are discussed, and a tentative classification on the basis of scale is advanced.


The Brazilian earthmounds have been briefly mentioned in the literature dealing with the vegetation of Central Brazil. Cole (1960) noted the following: "Dotted through the grasslands are clumps of trees, usually the cerrado species ipê roxa, ipê amarelo (both Tecomia spp.), and the lixeiro, the latter generally growing near termite nests around which sediments accumulate, allowing small portions of the soil to be protected against the flood waters. Away from the rivers, in the areas above the level of the normal floods, the clumps of trees become more numerous, and the transition is made from the low to the high pantanal. The latter grades to campo cerrado, where complete freedom from flood is associated with an apparently random distribution of trees."

In a soil study of northeastern Mato Grosso, Askew et. al. (1970) noted that the campos are found in the valleys (where the soil is moister and usually saturated to the surface during the wet season), and that many areas of campo feature raised islands of cerrado containing termitaria. In a related study, Ratter et. al. (1973) described a type of "termite savanna" in which mounds up to 2-3 m high and not more than 6-7 m in diameter, each carrying a group of cerrado trees and one or more termitaria, are scattered remarkably evenly over flat treeless grassland which is flooded in the wet season but dry from May or June until December or January. Ratter et. al. (1973) studied such termite savannas a few kilometres southeast of Xavantina, and reported seeing very much larger areas lower down the valley of the Rio das Mortes, and in the Araguaia river valley near the Ilha do Bananal.

Eiten (1975), refering to the earthmounds of the Serra do Roncador (northeastern Mato Grosso), noted that in almost all valley-side campos, except those on steep slopes, there were scattered circular groves of cerrado scrub several metres in diameter. These groves always stand upon soil that is slightly raised above the general level of the campo, from about 0.3 m to a little over 1 m. On top of each of these raised soil platforms is a single termite nest 1.0-1.5 m tall. The soil of the platforms is always of the same light grey as the subsoil of the campo, as is the nest itself. Apparently, the termites have built up this platform to keep the base of their nest from touching saturated soil. Eiten claimed that he never saw a platform without a well-formed nest, nor did he ever see a degraded nest on a platform, a situation which led him to believe that the platforms do not result from degradation of the nests. In addition, he observed that the quantity of soil in the platform is many times that of a nest, and is not hard as is the nest soil. Eiten noted that a landscape of saturated grasslands with scattered cerrado groves on raised earth platforms is a form of "pantanal" in the strictest sense, reminiscent of the Florida Everglades marshes with their "hammocks." In the classical Pantanal region of southwestern Mato Grosso, as well as in a large region of the northeastern corner of Mato Grosso and western Goiás, there are large expanses where this landscape type is dominant. The campus occupy vast flat areas which are annually inundated by several meters of water which remain weeks or months before draining away. The raised earth platforms are several metres high and several tens of metres wide.

Prance and Schaller (1982) noted the strong cerrado element in the Pantanal do Mato Grosso, particularly in the eastern part. These cerrados are dominated by species such as Bowdichia virgiloides, Caryocar brasiliense, Curatella americana, and Qualea parviflora, all of these typical of the cerrados of the Planalto of Central Brazil. They observed that cerrado occurs mainly in the nonflooded upland, but also towards the eastern edge of the Pantanal where the land is inundated for only short periods at the height of the flood season. Such wet cerrado tends to consist of numerous islands of cerradão (dense forest) on slightly elevated areas that are not flooded. Numerous small cerrado islands surround termite hills, and termite activity builds up the size of these islands by increasing the elevation.

Araujo Neto et. al. (1986) reported on the small mound fields or campos de murundus of Fazenda Agua Limpa, in the Federal District (Brasilia). These murundus are close to elliptical in plan, with length dimensions averaging 7.0 x 5.5 m, height from 0.05 m to more than 2 m, volume from 0.1 m3 to 141.5 m3, and aerial density from 26 to 61 ha-1. The mound fields consist of repeated patterns of rounded earthmounds covered by woody savanna plants, and neighbouring depressions dominated by grasses, sedges, and other low vegetation. These earthmound fields achieve their most spectacular development between the upper and drier formations (campo cerrado) and the lower moist valley formations of gallery forest (mata ciliar).

It is widely believed that the murundus are termite mounds, analogous to the termite savannas of south-central Africa. In Central Brazil, this belief is not based on detailed field measurements but on the general observation that many mounds are indeed occupied by termites. Thus, it is not clear whether the termites actively construct all or part of the mounds or whether they simply occupy the mounds, where environmental conditions are more suitable. Araujo Neto et. al. (1986) concluded that many of the mounds were not occupied by termites, and that no pedological evidence to indicate past occupation could be discerned. They interpreted the mounds as residual products of differential erosion, and surmised that surface runoff, seepage effects, and groundwater were the dominant processes in the genesis of the murundus. However, they noted that most of the larger mounds did show signs of termite activity.

Furley and Ratter (1988) noted that cerrado tree species which are most resistant to waterlogging (e.g., Curatella americana and Byrsonima crassifolia), are common near the boundary cerrado/campo, and on raised islands of ground in wet campos. The distribution of these islands produces the campos de murundus, consisting of an expanse of wet campos dotted with a regular pattern of raised earthmounds bearing cerrado trees, shrubs, and often termitaria.

Cunha (1990) described the capões de cerrado in the Pantanal de Poconé, one of several such subregions constituting the Pantanal do Mato Grosso. These earthmounds are circular to elliptical in shape, of 4-7 m in diameter, and tend to be dominated by the Lixeiro (Curatella americana). Cunha (1990) has also documented the much larger capões, typically of elliptical shape, of lengths up to 300 m, widths up to 80 m, and sparsely distributed around the campos subject to seasonal inundation. The aterros de bugre are dikes built by the indigenous settlers of the region, ostensibly to provide an area of high ground for protection during the annual flood season. Shallow excavations made in these aterros de bugre invariably reveal archeological remains (ceramics fragments and other paraphernalia), confirming the geologically recent and artificial origin of these geomorphological features.


The abiotic factor:  Geomorphology

The role of the abiotic factor is hypothesized as follows: after inception, due to surface or other irregularities, the incipient earthmounds are readily colonized by cerrado trees and shrubs, particularly by those species that are more tolerant of waterlogging. The establishment of vegetation on the mounds acts to reduce the speed of seasonal flood waters around them, further contributing (by differential sedimentation) to an increase in their size. During exceedingly wet years, the topping of the mounds by floodwaters may be surmised, albeit only for short periods. As the reduction of speed of the flood waters is gradual from the perimeter towards the centre of each mound, being most marked at the centre, the process of gradual mound built-up by differential sedimentation over the course of millennia is envisioned. The abiotic factor does not preclude the activity of termites, which may choose to avail themselves of the drier environment of the earthmound. The size and shape of the larger earthmounds (capões), coupled with field observations documenting the limited presence or even absence of termites, lends support to the hypothesis of differential sedimentation.

The biotic factor:  Termitaria

Mathews (1977) has developed a theory to explain the role of termitaria in the genesis of the Brazilian earthmounds. He suggested that the smaller earthmounds (murundus) are formed through a long process involving different stages, each stage corresponding to the action of a different termite species, as follows. (I) Initially, a species more tolerant of moist soils (Armitermes cerradoensis) colonizes the campo during the dry season; (2) during the wet season, another species of termite (Anoplotermes sp.) seeks refuge around the nest of A. cerradoensis, contributing larvae and soil to the mound; (3) following the death of the colony of A. cerradoensis, its nest undergoes degradation, although it continues to be enlarged by the Anoplotermes species during the wet season, and may at this time be invaded by plants; (4) in the following stage, the larger mounds would be colonized during the drier years by a third species, Cornitermes synderi, one less tolerant of flooding and, therefore, necessitating the building of large nests to provide a drier platform; and (5) in the final stage, the nests of C. snyderi are successively attacked and destroyed by predators such as the anteater (Myrmecophaga tridactyla), and followed by reconstruction and repair of the mound by the same or newer colonies, eventually leading to increases in mound size.

In a recent study of a campo de monchões in a seasonally inundated depression in the vicinity of Cuiabá (Mato Grosso), Oliveira Filho (1988) found evidence to indicate that the soils of the earthmounds were probably formed through the gradual degradation and erosion of successive generations of termitaria. This led him to propose a theory to explain the formation of the monchões in terms of a process of primary succession in which ecological islands (i.e., the monchões) grow gradually in size and, consequently, in complexity, species diversity, biomass, and total energy.


There are three physiognomically and floristically distinct vegetation types populating the highly dissected plains of northeastern Mato Grosso: (1) cerrado (including its variant cerradão), (2) campo, and (3) gallery forest (Ratter et. al., 1973). These vegetation types occur in a characteristic pattern: gallery forest, often bordered by narrow strips of campo, follows the streams, and cerrado (and cerradão) covers all the intervening higher ground (interfluves).

Eiten (1975) has documented the sharpness of the cerrado/campo boundary. Within 1 m, or even 0.5 m, the change from the shrubs and low trees of the cerrado to the grassy layer without woody plants of the campo is complete. The reason that the cerrado suddenly stops and gives way to the campo seems to be that the cerrado plants cannot establish themselves from seed in continually wet soil. In general, the campo occupies a site with a lower and more fluctuating water table, while the cerrado occupies higher ground where the soil seldom if ever remains saturated. Therefore, the sharply defined cerrado/campo boundary is attributed to the waterlogging itself, and not to the changed chemical condition (and colour) of the soil, which is a result of the waterlogging. In almost all cases, the cerrado stops suddenly at the edge of the campo, ostensibly due to the competition between the two vegetation types as whole plant communities. As soon as the surface of the soil is elevated by the termites and that it remains permanently nonsaturated and the surface can dry during the dry season, the cerrado plants invade it in full force.

In most places the groundwater level has a very precise relation to the cerrado/campo boundary. The relation of groundwater level to the boundary can be explained in terms of the high water levels in the wet season. Cerrado species tolerant of waterlogging are able to grow in open campos in places close to the cerrado boundary where the soil level is only a few centimetres higher than elsewhere. The observation that larger islands on pantanal landscape are thickly clothed with cerrado vegetation supports the view that groundwater level exerts a precise control on the cerrado/campo boundary.

The seasonal fluctuation in the depth of the water table is of crucial importance to the vegetation in two ways: (1) by influencing the ability of plants to take up water, especially in drought years; and (2) by restricting root activity in the wet season to the comparatively well-oxygenated layers of the soil, except for the gallery forest vegetation which has root systems especially adapted to waterlogged and anaerobic conditions. Under such conditions, very small differences in topography or soil are sufficient to tilt the balance in favor of one type of vegetation or the other.

Soil moisture is highly correlated with the main trend of vegetational variation. The patent lack of trees in the wet campos is striking, particularly since a wide range of woody species successfully colonizes both the interfluves (which are drier than the campos) and the stream sides (which are wetter). The absence of tall, woody species from areas which are intermediate in their physical characteristics may be due to the fluctuating nature of the water table and consequent soil moisture. Thus, trees appear to be able to tolerate both permanently wet (gallery forest) and dry (cerrado) conditions but not an extreme alternation of saturation and desiccation (Goldsmith, 1974).


The Brazilian earthmounds vary in size, shape, orientation, plant cover, soil properties, termite activity, surface runoff and groundwater. In their study of the campos de murundus of Fazenda Agua Limpa, Araujo Neto et. al. (1986) attempted to classify the murundus on the basis of morphology, including size and shape. The results proved inconclusive; neither size nor shape could be used as an adequate criterion for classification. Araujo Neto et. al. then turned their attention to the formative processes, including surface and groundwater movements, and the role of termites. Their observations led to a tentative twofold classification of campos de murundus into: (1) those associated predominantly with groundwater and seepage, featuring wooded savanna vegetation (cerrado) on the mounds and wet grasslands (campo limpo) in the depressions, and readily colonized by termites; and (2) those associated with surface runoff only, having no contact with groundwater, and featuring herbaceous grassland with scattered low shrubs (campo sujo and campo de campestre). In a contemporary paper, Furley (1986) echoed the ideas of Araujo Neto et. al.

Oliveira Filho and Furley (1990) have recently attempted a classification of the murundus based on formative processes. They observed that the role of the termites seemed to be of about equal importance as that of the hydrological/geomorphological processes, particularly with regard to the smaller earthmounds (murundus). This led them to classify the earthmounds into: (1) those formed primarily by abiotic processes, i.e. hydrological/ geomorphological processes, including the effect of floods, surface runoff, and ground-water; (2) those formed primarily by biotic processes, i.e. termite activity, and (3) those of mixed abiotic/biotic genesis, with combined action of hydrological/ geomorphological processes and termites. Oliveira Filho and Furley (1990) observed that surface runoff, leading to differential erosion, appeared to be a dominant process in the formation and development of some of the murundus, particularly those that were not affected by groundwater. They surmised that a process of differential erosion and sedimentation could well explain the existence of the larger earthmounds or capões found in the low Pantanal of Mato Grosso. However, the smaller earthmounds, or murundus de pantanal, appeared more likely to be caused by the combined action of hydrological/geomorphological processes and termites.

The marked scale difference of the murundus and capões suggests a tentative classification of the earthmounds on this basis. According to this concept, the earthmounds are classified into: (1) micromounds, or murundus, and (2) macromounds, or capões. The micromounds may often show a remarkable spatial regularity when viewed from the air, particularly those that have a substantial biotic influence. Smith (1971) has shown an aerial view of the Araguaia floodplains, where the quasi-mathematical spatial regularity of the small earthmounds defied comprehension. Mathews (1977) has explained the spatial regularity in terms of the territorial domains of termite colonies which are hostile to each other.

Fig. 2.   Micromound (murundu) in the Pantanal do Mato Grosso.

The micromounds or murundus are typically those studied by Araujo Neto et. al. (1986) in Fazenda Agua Limpa, Brasilia, although they also occur in other regions of central Brazil, including northwestern Mato Grosso, western Goiás, and in the high Pantanal of Mato Grosso. The micromounds are usually conical in shape, have length dimensions of up to several metres, small volume and high aerial density (Fig. 2). In contrast, the macromounds or capões are typically semi-oval in shape, have length dimensions of several tens of metres, large volume, and low aerial density (Fig. 3) (Cunha, 1990).

Fig. 3.   Macromound (capão) in the Pantanal do Mato Grosso.


The Pantanal do Mato Grosso and other ecosystems of Central Brazil lie within the climatic region of the cerrado, a savanna-type biome characterized by its adaptation to a long dry season, often lasting up to seven months. Several intergrading series of cerrado stands are possible, with the depth to the water table being the critical environmental factor determining physiognomic and floristic variation.

The topographic sequence cerrado/campo/gallery forest is common to many areas of Mato Grosso. Cerrado vegetation requires a well-drained soil, with no long-term accumulation of water during the rainy season. Conversely, the trees of a gallery forest have root systems especially adapted to poor drainage and associated waterlogging.

Positioned between cerrado and gallery forest, the grassy campos are the only type of vegetation well-adapted to extreme alternations of saturation and desiccation. In the highly dissected landscape of northeastern Mato Grosso, the campos may be entirely absent or constitute only a narrow strip separating cerrado and gallery forest. Under these conditions, the presence of campos is determined by a complex combination of environmental factors, including groundwater regime, geomorphology, and possibly soils. The campos, however, tend to dominate in the floodplains and regions grading to pantanal landscape. In the latter case, a competition exists between cerrado and campo communities. Cerrado plants will usually take the higher ground, since they are unable to establish themselves from seed in continually wet soil. Surface and other irregularities, whether caused by abiotic (hydrologica/geomorphological) or biotic factors (termitaria), lead to small differences in elevation and provide an appropriate environment for the cerrado to encroach on the wet campo. Over the course of millennia, this leads to the formation of the vegetated earthmounds. Thus, earthmounds and their colonizing cerrado vegetation engage in a symbiotic relationship, which assures both mound growth and cerrado survival in spite of the hostile environment of the surrounding wet campo.

The nature of the earthmounds is explained in terms of the competition between two plant communities, in this case, the cerrado, endemic to the savannas of Central Brazil, and the campo, whose grasses are better adapted to extreme alternations of saturation and desiccation. The cerrado encroaches upon the grassy campo whenever topographic conditions are such as to enable it to remain slightly above the surrounding wet campo. This state of affairs is abetted by the action of termites, who, by building their nests above the main level of the campo, provide the appropriate environment for the establishment and survival of the clumps of cerrado vegetation. Once formed, the earthmounds tend to gradually increase in size, aided by either additional termite activity or by differential erosion/sedimentation. In the great expanse of seasonally flooded grasslands of the Pantanal do Mato Grosso, this latter process may be substantial enough to account for the development of the larger earthmounds or capões.


The writers wish to acknowledge the support of the U.S. National Science Foundation.


Araujo Neto, M. D., Furley, P. A., Haridasan, M., and Johnson, C. F.. (1986). The murundus of the cerrado region of Central Brazil. J. Trop. Ecol. 2, 17-35.

Askew, G.P., Moffatt, D. J., Montgomery, R. F. and Searl, P. L. (1970). Soil landscapes in northeastern Mato Grosso. Geogrl. J. 136,211-227.

Cole, M. M. (1960). Cerrado, caatinga and pantanal: the distribution and origin of the savanna vegetation of Brazil. Geogrl J. 126, 16,179.

Cunha, C. N. (1990). Estudos florísticos e fitofisionómicos das principais formações arbóreas do Pantanal de Poconé-MT. Master's thesis, Universidade Estadual de Campinas, São Paulo, Brazil.

Eiten. G. (1975). The vegetation of the Serra do Roncador. Biotrop. 7, 112-135.

Farley. P. A. (1986). Classification and distribution of murundus in the cerrado of Central Brazil. J. Biogeogr. 13, 265-268.

Farley, P. A. and Ratter, J. A. (1988). Soil resources and plant communities of the central Brazilian cerrado and their development. J. Biogeogr. 15, 97-108.

Goldsmith, F. B. (1974). Multivariate analysis of tropical grassland communities in Mato Grosso, Brazil. J. Biogeogr. 1, 111-122.

Mathews, A. G. A. (1977). Studies on termites from the Mato Grosso State. Brazil. Academia Brasileira de Ciências, Rio de Janeiro.

Oliveira Filho, A. T. (1988). A vegetação de um campo de monchões-microrrelevos associados a cupins--na regiã de Cuiabá (MT). Ph.D. dissertation, Universidade Fstadual de Campinas, São Paulo, Brazil.

Oliveira Filho, A. T. (1992). Floodplain 'murundus' of Central Brazil: evidence for the termite-origin hypothesis. J. trop. Ecol. 8, 1-19.

Oliveira Filho, A. T. and Farley, P. A. (1990). Monchão, cocuruto, murundu. Ciência Hoje. 11, 30-37.

Prance, G. T. and Schaller, G. B. (1982). Preliminary study of some vegetation types of the Pantanal, Mato Grosso, Brazil. Brittonia, 34, 228-251.

Ratter, J. A., Richards, P. W., Argent, G. and Gifford, D. R. (1973). Observations of the vegetation of northeastern Mato Grosso I. The woody vegetation types of the Xavantina-Cachimbo expedition area. Phil. Trans. R. Soc. 266B, 449-492.

Smith, A. (1971). Mato Grosso:  Last virgin land. Dutton. New York.


Documents in Portable Document Format (PDF) require Adobe Acrobat Reader 5.0 or higher to view; download Adobe Acrobat Reader.