The ecological significance of forested mounds of the termite Macrotermes in the Caprivi Strip

@tonyrebelo @jeremygilmore @hamishrobertson @joshua_tx @s_k_johnsgard @botswanabugs @bartwursten @troos

In May 2006, a colleague and I visited the eastern Caprivi Strip ( and to do the fieldwork for and

On the early morning of 19 May 2006, I studied an example of a large mound built - and still occupied - by Macrotermes ( We had slept at this location the previous night.

The height of this mound was >3.5 m above general ground level, and probably as much as 4.5 m. Its width (diameter, including the broad apron of the mound) was 25 m, measured from the general level of the floodplain that formed the matrix for the scattered mounds.

The scene in is indicative. However, the sheer area of the mound - nearly 500 square metres - makes the situation difficult to photograph.

The species of Macrotermes could not be identified at the time, and - surprisingly - this seems still to be true today. The most likely candidates are Macrotermes natalensis and Macrotermes falciger, both of which are greatly underrepresented in iNaturalist.


The plant community on the mound contained a total of about 20 spp., visible at the time of my visit.

Eight individual trees occurred on this mound, consisting of

Peripheral plants of M. mochisia, 1.5 m high, had been so heavily browsed by large folivores that their growth-form was bonsai/topiary-like ( and

This was reminiscent of Gardenia (please see, but with the crown less rounded.

The understorey on the mound consisted mainly of Euclea divinorum ( and Ximenia americana (, the latter superficially resembling Gymnosporia senegalensis (which may itself have been present as a scarce species).

Carissa spinarum ( was common on this mound, but Diospyros lycioides ( seemed absent.

Salvadora persica ( was present on this mound. Dichrostachys cinerea ( occurred only where the mound joined the surrounding ground level.

The main liane on the mound was Capparis tomentosa (, festooning not only the periphery of the mound, but also the trees. There were also two lines of Cucurbitaceae, one with small fleshy fruits (possibly, and the other bearing large, knobbly melons, lying on the ground and ripening at the time (possibly Asparagus africanus? ( was present only as small individuals.

Setaria verticillata ( was sparsely present on the mound, brown and collapsed at the time, with the burr-like seed-heads evident ( A lawn-type grass occurred immediately adjacent to the mound, but did not encroach even on to the apron of the mound.

?Kalanchoe lanceolata? (a fragile-looking ?annual succulent) was present, and in flower at the time.

I estimated the contributions to the total vertically-projected cover of the vegetation on this mound, as follows:

Diospyros mespiliformis (tall trees): 55%
Manilkara mochisia (shorter trees): 25%
Capparis tomentosa (robust liane): 10%
Ximenia americana (common spinescent shrub): 5%
Euclea divinorum: 2%
Salvadora persica (partly liane-like): 1%
Cucurbitaceous liane (large fruits): trace
Cucurbitaceous liane (small fruits): trace
Unidentified shrub with dentate leaves: trace
Setaria verticillata: trace
?Phyllanthus reticulatus: trace
Acanthaceae, unidentified perennial: trace
Asparagus africanus?: trace
Malvaceae, unidentified: trace
Crassulaceae, unidentified: trace
Grewia sp. with no apparent fruits (different from the sp. in nearby woodland of T. sericea): trace
Sundry unidentified (five spp.): trace each


In this general environment, the matrix (the surrounding plain) was a floodplain, bearing open savanna with a main stratum of short, coarse (plastic-like) grass (dry and brown at the time).

On the clayey flats (with a grey sandy surface over most of the area) of the matrix, there was a sparse stratum of trees and shrubs.

This consisted of Combretum imberbe and congeners, Senegalia nigrescens, Piliostigma thonningii (, Dichrostachys cinerea, and Ziziphus mucronata (, with a few individuals of Gardenia.

In the case of C. imberbe, most of the individuals on the floodplain were about 1-1.5 m high.

All of these spp. were absent from the mound itself.

Patches of cracking clay in the matrix supported only sapling-size/stunted Combretum imberbe (, and shrubs of Dichrostachys cinerea (

However, the large mounds were not the only situation in which trees were concentrated. Terminalia sericea (including large trees) dominated on a sandy patch about 75 m away, slightly elevated above the level of the matrix of short grassland. This sand was fine-grained and pale grey, and at most 0.5 m deep over a hard layer, as we discovered when our van became temporarily stuck.

Accompanying T. sericea on the sandy patch were Grewia sp. (in ripe fruit, the edible layer of which was naturally completely dry), Gymnosporia senegalensis (, Philenoptera violacea (, and Vachellia erioloba (uncommon) - all of which were absent from the mound described above.


The area was well-grazed, albeit only by Bos taurus X indicus. Despite this substitution of domestic livestock for the original wild grazers, the ecosystem looked natural and healthy. The nearest village or kraal was at least several km distant.

There was abundant evidence of current foraging by termites throughout this area. The faeces of B. taurus X indicus were being promptly consumed by Macrotermes.

There was much evidence (in the form of mud-runnels) that Macrotermes was foraging for litter on the mound itself. This included grass (also covered by mud-runnels) being consumed by Macrotermes on the apron of the mound. However, my impression was that only trampled grass was suitable for Macrotermes.

The mud-runnels made for foraging by Macrotermes were present on the surface in the matrix, even where the surface was cracking clay. This was often on the faeces of B. taurus X indicus.

Hodotermes mossambicus ( was abundant hereabouts, particularly on the floodplain. However, I did not see it at this location, perhaps because of the time of day.

The soil-heaps of mole-rats were absent from this whole area, despite the occurrence elsewhere in the Caprivi Strip of Fukomys damarensis (

Birds heard on/near this mound:

Vanellus coronatus (, common on the short grassland of the matrix
Ptilopsis granti (, heard repeatedly the previous night
Corythaixoides concolor (
Streptopelia semitorquata (
Streptopelia capicola (
Streptopelia decipiens (
Oriolus auratus (
Pycnonotus barbatus (


In Africa more broadly, it is well-known that mounds of Macrotermes tend to support plant spp. absent from the matrix.

However, what is remarkable in this case is that the plant communities on and off this particularly large mound (comparable in area to a typical suburban residential plot) shared virtually no species of plants.

The mound supported a considerable patch of forest, which was

  • exempt from wildfire,
  • composed of plants dispersed and sown by vertebrates, mostly via fleshy fruits and endozoochory, and
  • surrounded by open savanna, itself probably exempt from wildfire when heavily grazed.

The presence of S. persica indicates that the soils on this mound, even at its apex, were rich in cations, particularly sodium (,often%20mucronate%20at%20the%20apex.).

The complete lack of caesalpinioid legumes, and indeed legumes in general, on the mound is noteworthy.

All of the trees, and most of the shrubs and lianes, on this mound possess fleshy fruits - many of them edible to humans ( and and

The protection of the whole mound from wildfire was owing to

  • the lack of a flammable lower stratum on the mound itself, and
  • the shortness of the grass in the surrounding matrix, partly owing to grazing.

The presence of faeces indicated that the vegetation on the mound was probably attractive to B. taurus X indicus, despite the scarcity of grass on the mound.

My impression was that three main consumers of grass interact in an important way in the complex of vegetation. In the matrix and probably also on the mound, grass is eaten mainly by ungulates and Hodotermes; the faeces (derived mainly from grass) of the ungulates are consumed by Macrotermes, which also consumes woody material of various plants; and the faeces of Hodotermes fertilise the whole ecosystem, particularly in the matrix.

What makes this situation, in the Caprivi Strip, remarkable is the combination of Macrotermes and Hodotermes. Extensive, forested mounds of Macrotermes are mainly a feature of tropical African woodlands, beyond the distribution of Hodotermes. The Caprivi Strip is at, or near, the northern limit of the distribution of Hodotermes. This unusual overlap in distribution seems to have produced a regime of thorough consumption of grass and litter, tending to exclude wildfire, and promoting palatable plants - including fruits edible for humans.

Posted on 23 May, 2023 08:33 by milewski milewski


At the time of my visit, Loxodonta africana had become scarce as a resident in the Caprivi Strip (, explaining why I saw no feces of this megaherbivore on or near the mound described.

See section 7.48 in

At that time, L. africana was mainly restricted, as a resident, to:

a) the easternmost part, near the Kwando River, of the former Caprivi Game Reserve (now Bwabwata National Park,, and
b) marshes adjacent to Botswana in Nkasa Rupara (formerly Mamili) National Park (

However, iNaturalist currently contains nearly 500 observations of L. africana in the Caprivi Strip ( This suggests that the species has increased over the past two decades.

Perhaps, were I to revisit this mound today, I would find the faeces of L. africana.

Posted by milewski about 1 year ago

According to the Atlas of Namibia (, the densities of populations of wild ungulates in the eastern Caprivi Strip are - owing to excessive illegal hunting - less than those in the pro-Namib,

Although the vegetation looks pristine in the eastern Caprivi Strip, its wild ungulates have been virtually exterminated outside Mudumu ( and the other National Parks.

Posted by milewski about 1 year ago

Van Wyk and van Wyk mapped the vegetation of the Caprivi Strip, at a general scale, as Baikiaea plurijuga-Pterocarpus angolensis woodland, with woodland of Colophospermum mopane in the southern part of the eastern Caprivi Strip.

According to my own observations, in May 2006:

Eastern Caprivi Strip:

Based on the evidence in the form of piles of soil ( and at the side of the road, Hodotermes seemed to penetrate woodland dominated by Baikiaea plurijuga (

Mounds built by Macrotermes seemed to penetrate Baikiaea woodland, but not woodland of Burkea-Guibourtia-Pterocarpus.

Central Caprivi Strip:

I observed Baikiaea plurijuga on brown sand. Heaps of the damara mole-rat (Fukomys damarensis) were present. Hodotermes was common here, foraging on the surface. There was no evidence of Macrotermes or Odontotermes. I observed ?Microtermes under a pile of dead leaves of grasses and other plants.

I infer that Hodotermes penetrates woodlands on sand - at least near the road - in the Caprivi Strip, where Macrotermes and Odontotermes do not.

I also ask:
Are mole-rats (Fukomys) ubiquitous in caesalpinioid woodlands on sand in south-central Africa?

Western Caprivi Strip:

Just east of the western boundary of Bwabwata National Park (, I observed patches of 'knobthorn-marula' savanna, similar to that in Kruger National Park. The main trees in these limited areas are Senegalia nigrescens and Sclerocarya caffra.

This incidence of S. nigrescens is well-reflected by observations in iNaturalist (e.g.

Posted by milewski about 1 year ago

NOTES ON ANOTHER, EVEN LARGER MOUND IN THE SAME AREA, examined during the same visit, in May 2006:

Mound height >3.5 m, diameter 25-35 m.

Vegetation is a patch of forest, a 'jungle-tangle' around one 'matriarch' tree, forming a fire-free patch of about 875 square metres.

This consists of Diospyros mespiliformis (one large specimen 13 m high, plus a second individual about 5 m high), Capparis tomentosa, Grewia sp., Carissa spinarum, Hyphaene sp. (sapling), Ziziphus mucronata, Manilkara mochisia (hedged juvenile of shrub-size), Euclea divinorum,?Phyllanthus reticulatus, Piliostigma thonningii, Achyranthes aspera, ?Dombeya rotundifolia, Asparagus liane (deciduous, with leaves falling at the time), Ximenia americana, ?Gymnosporia senegalensis, and sundry small lianes.

There were at least 10 genera, including five genera of lianes, that I could not identify.

The edge of this patch of forest tended to consist of Diospyros lycioides (regenerating vegetatively from the roots, i.e. 'suckering'), about 3 m high. This constituted, in part, a peripheral thicket.

Of the approximately 25 genera of plants present on this extremely large mound, Diospyros was the most important, because it contributed a) more than one species, b) by far the largest individual plant, and c) the greatest proportion of cover throughout the year.

Posted by milewski about 1 year ago
Posted by milewski about 1 year ago

One of the surprises of this fieldwork in the eastern Caprivi Strip was the discovery that the large mounds of Macrotermes, although devoid of mole-rats, contained Hodotermes under the surface of the apron, and above the depth at which the fungus cultures of Macrotermes were located.

Posted by milewski about 1 year ago

Cole M M (1986, and and and

Although poorly proofread, this is a good basic reference to large mounds and their vegetation, including a) large mounds generally occur on seasonally wet landforms, b) alleviation of both wildfire and waterlogging, conducive to forest patches, c) the mounds consist of deep material from weathered bedrock, and d) the vegetation on the mounds varies greatly, according to climate and soil.

Posted by milewski about 1 year ago

Add a Comment

Sign In or Sign Up to add comments