Earthen mounds (heuweltjies) of South Africa and their termite occupants: applicability of concepts of the extended phenotype, ecosystem engineering and niche construction

Abstract

Heuweltjies are earthen mounds found throughout the Succulent Karoo of South Africa and are inhabited by the termite Microhodotermes viator. Many have assumed that heuweltjies are constructed by the occupying termites. Consequently, heuweltjies have been used as an example of several important concepts in ecology and evolution: the extended phenotype, ecosystem engineering and niche construction. However, recent findings demonstrate that M. viator does not directly construct heuweltjies. Rather, termite colonies enrich the soil around their nests with plant nutrients, which promotes development of widely separated patches of denser vegetation. Eventual formation of heuweltjies represents a response of the physical environment to the windbreak effect of the denser vegetation patches (localized reduction of wind velocity and resultant deposition and accumulation of airborne sediment). Other structures constructed by the termites are justifiably regarded as extended phenotypes. Identification and investigation of a complex cascade of processes are required to more precisely assess the manner in which this termite species functions as an ecosystem engineer or niche constructor, thereby significantly influencing the availability of resources within local ecosystems. Environmental alterations that are either directly or indirectly generated by social animals that construct large, communal nests represent ecological processes that contribute significantly to local biodiversity.

This article is part of the theme issue ‘The evolutionary ecology of nests: a cross-taxon approach’.

1. Introduction

Large earthen mounds constructed by termites are among the most remarkable animal-built structures in terrestrial environments. Termite mounds and the nests they house are fitting examples of extended phenotypes [1], which represent behaviours and physical manifestations of those behaviours that are under genetic control, and are therefore subject to natural selection and evolution. An extended phenotype can include physical structures or environmental modifications that are constructed by the organism but are not a physical part of that organism. Mounds constructed by different termite species exhibit considerable diversity in external shape and internal structure, but in general, the mounds contribute to the success of colonies by providing protection from predators and greatly modifying atmospheric conditions of the internal environment [2,3]. For example, the massive, indurated exteriors of mounds of fungus-cultivating Macrotermes species provide protection against large, burrowing predators, such as aardvarks (Orycerotopus afer), and the internal arrangement of small passages and access tunnels enables defense against small predators, such as ants [4].

The external form and internal architecture of those mounds also provide ventilation systems for the close regulation of humidity and CO₂ concentrations required by the cultivated fungi [5,6].

The influences of termite mounds extend far beyond benefits accrued by the colonies that construct and inhabit them. Throughout tropical and subtropical savannah ecosystems of Africa, termite activity significantly alters the physical and chemical characteristics of soils. In doing so, termites act as ecosystem engineers by modulating the availability of resources available to many other kinds of organisms [7,8]. As a consequence, termites greatly influence vegetation (species composition, spatial pattern and productivity), faunal distribution and abundance, ecosystem processes and ultimately, overall biodiversity [9–14].

The purpose of this paper is to review and critically examine information about a particular type of earthen mound associated with colonies and nests of the southern harvester termite (Microhodotermes viator), found in western South Africa. These mounds are called heuweltjies (pronounced ‘hué-vil-kees’), which means little hills in Afrikaans. Despite considerable controversy regarding the origin of these mounds, they have been widely regarded to be constructed by the termites and accordingly, have been cited as an example of an extended phenotype or ‘extended organism’ of M. viator [15], and also an example of ecosystem engineering by termites [16,17] and of niche construction [18]. These concepts in evolution and ecology are linked in their history of development: the extended phenotype focuses on the evolution of behaviours involved in altering environmental conditions; ecosystem engineering subsequently considers, in part, the way in which extended phenotypes (e.g. beaver dams) influence ecosystem function and structure of communities, and the most recent conceptual contribution, niche construction, goes further to include the evolutionary consequences of environmental changes created by organisms. This paper critically examines the manner in which heuweltjies and their termite occupants have been used as examples in each of these conceptual areas. Recent advances in knowledge about multiple, interacting processes responsible for formation of these mounds provide a more detailed evaluation of the degree to which the formation of heuweltjies can be directly attributed to the termites, and also a more complete understanding of how colonies of M. viator actually function as ecosystem engineers and niche constructors.

2. Mounds of controversy—heuweltjies of western South Africa

Heuweltjies are found throughout approximately 67 000 km² of western South Africa, principally in the Northern Cape and Western Cape provinces [19–21]. Their principal range is throughout the Succulent Karoo Biome, in a swath 100–200 km wide, inland from the western (Atlantic) coast, and then eastward in the Little Karoo region of the inland basins of the Cape Fold Mountains. Their northern distribution extends into southwestern Namibia, approximately 150 km northwest of the Orange River, the border with South Africa (JR McAuliffe 2022, personal observation). They also occur in some adjacent portions of the Fynbos Biome in the southwestern corner of South Africa [22], and Lovegrove & Siegfried [19] provide a map of their distribution within South Africa. Heuweltjies commonly range in diameter from 10–30 m, with heights of 0.5–2.0 m, and in some regions, the mounds cover more than a quarter of the land surface [23,24]. Heuweltjies exhibit remarkable spatial overdispersion, with the average distance between centres of nearest neighbouring mounds typically ranging from 40–50 m (figure 1a). Soil of the mounds differs from that of the surrounding area and supports distinct vegetation. The regular spacing of the mounds, typically covered with vegetation that is different in species composition from the surrounding area, gives much of the Succulent Karoo a singularly unusual appearance (figure 1). They occur in diverse geological settings, including hillslopes and pediments with very shallow soils (less than 15 cm thick) over many types of bedrock (sedimentary, igneous and metamorphic), ancient aeolian sand deposits, and alluvial deposits on valley floors [25].

Figure 1.

(a) Google Earth Pro vertical aerial view of hillslopes covered with regularly spaced heuweltjies that are more densely vegetated than their surroundings. The site is located 15 km southeast of Klawer, Western Cape Province, South africa. The yellow ‘X’ and arrow indicate the camera position and view direction, respectively. Coordinates of the camera location are 31.88105°S, 18.70527°E. The yellow ellipse indicates the approximate area shown in the centre of (b). (b) Ground-level photograph in the direction indicated by the yellow arrow in (a). The darker swath through the centre of view is a northwest-facing hillslope indicated by the yellow ellipse in (a). Vegetation of heuweltjies in this general area ranges from 50–150 cm in height and plant canopies cover three-quarters or more of the soil surface. Vegetation off of the mounds has a maximum height of approximately 30 cm and covers less than half of the soil surface. Diameters of heuweltjies are approximately 30 m, with heights of approximately 1 m. The rocky hillslopes consist of quartzitic sandstone bedrock with thin soils; the silt-rich soils of heuweltjies are positioned on top of those materials [25]. Photo by JR McAuliffe, 16 October 2016.

Heuweltjies differ from large epigeal mounds constructed by other termite species in several ways. First, the large size of many heuweltjies (e.g. a 30 m diameter heuweltjie has a basal area greater than 700 m² and an earthen volume greater than 400 m³ [23]) is an order of magnitude greater than that of typical, large mounds of Macrotermes spp. [8,26–29]. Second, upper layers of the low, domed surfaces of heuweltjies typically consist of non-cemented and unconsolidated, silt-rich soil, rather than materials that are directly arranged and cemented in place by mound-builders such as Macrotermes spp. [30]. Another significant contrast is the great age of some heuweltjies. Radiocarbon dating of cemented accumulations of calcium carbonate that commonly form in association with long-term termite occupancy indicates that the ages of some mounds exceed 30 000 years [31,32] (published radiocarbon ages compiled with calibrated calendar dates in [25]). By contrast, the oldest reported ages of large Macrotermes mounds, also based on radiocarbon ages of soil carbonates, is only a tenth of that duration [33].

Processes responsible for formation of these unusual, long-lived landscape features have been debated for more than 80 years, and explanations have included various physical and biological mechanisms. Lovegrove [34] catergorized and listed various proposed mechanisms and McAuliffe [25] presented an updated list that includes additional, more recently proposed hypotheses. A widespread view has been that the termites create the mounds [35,36], potentially over long periods of time, possibly in combination with earth-moving activities of burrowing mammals [19,37–39]. Consequently, termitaria—the name used for earthen mounds directly constructed by termites like the fungus-growing Macrotermes spp.—has been applied to heuweltjies [40,41].

More recently, Cramer et al. [42] proposed that termites have nothing to do with heuweltjie formation, but rather, the mounds represent relicts of an ancient land surface, protected from erosion by regularly spaced vegetation patches. Original development of those patches was attributed entirely to interactions among plants in generating vegetation structure. This hypothetical scenario is based on theoretical models of spatial self-organization in ecological systems arising through the interplay of long-range negative feedback (i.e. competition) and short-range facilitation [43–45]. The occurrence of termites on the mounds was viewed as a secondary response to the more favourable, pre-existing soil conditions found there. However, the operation of spatial self-organization in this system was inferred entirely from the pattern of regular spacing of the mounds, not any direct investigation of process [42]. Furthermore, the spacing between heuweltjies (40–50 m) and their density are remarkably invariant across diverse environments, ranging from extremely arid (mean annual precipitation (MAP) less than 100 mm) to considerably more mesic (MAP greater than 400 mm), and with soil conditions varying from deep soils of ancient aeolian sand sheets to steep hillslopes with extremely thin (less than 15 cm) rocky soils underlain by different kinds of bedrock. The proposition that the relatively consistent spatial patterning of heuweltjies across such diverse physical environments (precipitation, soil characteristics, substrate-limited rooting depth, etc.) is generated solely through processes of vegetation self-organization, requiring plant-to-plant competitive interactions over the scale of tens of metres, is implausible. Indeed, the authors who originally proposed the operation of this process [42] recognized this difficulty, conceding in a subsequent publication that ‘Since the vegetation of all sites where mounds are found in this region is generally less than 1 m tall, it is unlikely that vegetation patches could experience direct root competition over long distances' [20]. No published papers have subsequently provided direct evidence for vegetation self-regulation as a process that is ultimately responsible for the formation of heuweltjies.

Nevertheless, controversy regarding how heuweltjies are formed has continued, with members of the same group of investigators in [42] recently proposing that erosion of landscape surfaces by water initiates the process of mound development and their spacing due to the existence of drainage channels in topographic lows between mounds [46]. However, the spatial pattern of drainages is more plausibly explained as the normal hydrological response to the positioning of mounds, rather than a process that triggered the mounds' initial development and regular spacing. Furthermore, the latter does not provide an adequate explanation because the mounds occur in highly contrasting environments that fundamentally differ in hydrological behaviour, the generation of runoff and hence, erodability (e.g. nearly level, extensive aeolian sand sheets that generate little or no runoff and completely lack surface drainage channels and stream networks, to hillslopes with thin soils underlain by bedrock, which are capable of generating runoff). Yet the spatial patterning and density of heuweltjies remain relatively invariant across the range of contrasting landscapes [25].

Despite the controversy within the community of researchers directly involved in research on heuweltjies, there are multiple cases where other ecologists and evolutionary biologists have uncritically embraced the view that colonies of M. viator are responsible for constructing the mounds. For example, a paper focused on concepts related to the extended phenotype concept in evolutionary biology tacitly expressed that view: ‘they [colonies of M. viator] produce curious landforms known as heuweltjies…’ [15]. Hastings et al. [16] listed M. viator as an ecosystem engineer responsible for creating heuweltjies. A paper on the related, conceptual framework of niche construction theory, which includes the evolutionary consequences of environmental alterations created by organisms, repeats the claim that heuweltjies are ‘produced’ by M. viator [18]. That paper states that because individual heuweltjies persist for much longer than the longevity of an individual colony of the termites, dispersing reproductive pairs that alight on unoccupied heuweltjies encounter soil conditions more suitable to their survival and colony establishment because ‘previous generations of termites constructed them [heuweltjies] to be that way’. Nearly three pages were devoted to the same thesis in a final chapter of a book focused on questions of evolutionary biology published by Harvard University Press [47, pp. 222–224]. These perspectives merit closer examination and evaluation.

3. Recent advances in understanding

McAuliffe [25] provides the most recent and detailed review and evaluation of the many explanations regarding the formation of heuweltjies. The long history of controversy and lack of progress in achieving substantial understanding stem in part from a tendency to divide potential mechanisms into mutually exclusive sets of hypotheses. For example, Lovegrove [34] presented a dichotomous scheme of physical versus biological mechanisms and concluded that physical processes are incapable of generating the highly regular spacing, therefore a biological mechanism, i.e. the spacing of territorial animals (termite colonies or burrowing mammals), must be the cause. This tendency extends to explanations focused on biological influences, where ‘faunal and floral’ processes were cast as mutually exclusive hypotheses [20]. Despite the appeal of its seeming simplicity and generality, proposals based on theoretical models of vegetation self-organization [42,45] are yet another manifestation of this tendency, where only the potential influence of plant-based interactions are considered. These kinds of mutually exclusive perspectives have conceptually restricted development of broader views of ways in which multiple and very different realms of processes can potentially interact to generate observed phenomena. Broader conceptual approaches, including theoretical ones (e.g. [13,48]), that explicitly consider the combined effects of territorial behaviours of social insects and interactions involving plants, hold greater promise in explaining regular patterning like that exhibited by heuweltjies, and resolving long-running, often acrimonious debates.

Within the past decade, evidence has emerged that indicates that heuweltjies are not constructed through the earth-moving activities of M. viator. Rather, long-term occupancy of a patch by the termites creates soil conditions that promote localized development of vegetation that is taller and denser than the immediately surrounding area. The denser vegetation functions as an aeolian sediment trap that enhances localized deposition and accumulation of fine, wind-transported sediments, thereby forming the large mounds. Multiple lines of empirical evidence provide keys to the understanding the broad constellation of processes that are involved. Intraspecific, agonistic interactions, documented for multiple taxa of termites, including M. viator, among members of different colonies [49,50] provide a mechanism for the overdispersion of termite colonies. Field investigations demonstrate that M. viator is directly responsible for the accumulation of plant nutrients at the focal location of their colonies [51]. Detailed examination of soil characteristics and soil stratigraphy of heuweltjies indicates that the soil materials in the mounds consist largely of accumulated aeolian (wind-deposited) sediments [23,52], and those materials are superimposed on top of pre-existing land surfaces, without evidence of erosion by water having initiated mound formation [25]. McAuliffe et al. [17,23] proposed that the conditions responsible for those aeolian accumulations are generated by a set of biological processes (termite territoriality and alterations of soil and vegetation), coupled with responses of the physical environment to those initial biogenic changes. The following list summarizes this developmental sequence:

• — Successful colonies of M. viator tend to be highly overdispersed (regularly spaced) due to territorial interactions between neighbouring colonies.
• — Within a colony of M. viator, members of the worker caste gather plant food materials from the area surrounding their single, centrally located nest. Over time, the transport of those materials towards the colony's central focus generates a unidirectional flow and concentration of materials, creating a fertile island enriched with nutrients.
• — The regularly spaced fertile islands foster development of patches of denser vegetation embedded within the surrounding, more sparsely vegetated matrix.
• — The dense vegetation patches function as aeolian sediment traps in the semi-arid environment of the Succulent Karoo, accumulating greater amounts of sediments compared to their more sparsely vegetated surroundings, forming a mound composed largely of those sediments.
• — Positive feedback loops involving both biological and physical processes contribute to continuity of occupation by termites and persistence of processes responsible for creation and maintenance of the mound.

A detailed, evidence-based system dynamics model that includes the above is presented in [17,25]. Information from another investigation [53] independently substantiates the conclusion that deposition and accumulation of aeolian sediments generate the large volume of earthen materials in heuweltjies. That latter study investigated the geochemical characteristics of soil from heuweltjies, soil from adjacent (off-mound) areas, and the underlyling bedrock at a site near Clanwilliam, Western Cape, South Africa. The bedrock consists of quartzitic sandstone (quartz arenite) that contains 93.6% SiO₂. Thin, off-mound soils that mantle the bedrock also contain primarily SiO₂ (91.4%), reflecting the dominant contribution of sandy materials derived from weathered bedrock. By contrast, soil from heuweltjies averaged 71.3% SiO₂, but contained approximately double the Al₂O₃ content of bedrock and off-mound soil (6.1% versus 2.5% and 3.1%). The much higher Al₂O₃ content of heuweltjie soils reflects products derived from the weathering of aluminosilicate and phyllosilicate minerals (feldspars, micas and clays) that are common in many kinds of rocks far removed from the site (e.g. granites, palaeozoic shales and mudstones), but are present in negligible amounts in the local quartz arenite bedrock. Additionally, concentration of a variety of terrigenous trace elements, including rubidium, zirconium and hafnium, all representing weathering products of a variety of rock types other than the local quartz arenite, were 2.5 to 4 times greater in soils from heuweltjies than in the quartz arenite bedrock of the site. The distinct geochemistry of soils of heuweltjies unequivocally demonstrates that they are not principally derived from weathering of the local bedrock or localized redistribution of soils from adjacent off-mound areas. At the same location near Clanwilliam, Cramer et al. [42] reported that silt content in the soil of heuweltjies (11%) was approximately double that in adjacent off-mound areas (5%). The hillslope location of the study area (∼ 32.21015°S, 18.87364°E), positioned well above the floodplain of the Olifants River and tributaries, rules out the possibility that the silt-rich soils of heuweltjies with distinct geochemical signatures are alluvial deposits. At this site, aeolian deposition of fine-grained sediments derived from a wide geographical area is the only means by which those materials could have been delivered to generate the observed soil volumes within heuweltjies.

The central role of aeolian sediment accumulation in the formation of heuweltjies is substantiated by marked geographical variation in heuweltjie size as a function of regional differences in aeolian sediment supply. Within the Atlantic coastal region, a low-relief landscape largely covered by aeolian sand deposits that extend 50 km or more inland, heuweltjies commonly reach diameters exceeding 30 m and heights of 2 m, and frequently cover a quarter of the land surface [17,23,54]. By contrast, in more interior regions characterized by very thin soils over bedrock surfaces and lacking a plentiful supply of potential aeolian sediments, heuweltjies are considerably smaller. In those areas, diameters and heights of the mounds are typically less than half that of those in the Atlantic coastal plain region, and they cover a much smaller proportion of the land surface, despite similar spacing between the mounds (40–50 m) and densities (4–5/ha). In some interior locales, heuweltjies do not reflect the literal meaning of their Afrikaans name, as they have no vertical relief, but instead consist of flat disc-like areas of contrasting soil characteristics and vegetation, centred on colonies of M. viator [25,51].

Quantitative analysis of aerial imagery of 7117 randomly chosen locations across the entire range of occurrence of heuweltjies in South Africa further demonstrates that the size of heuweltjies varies in response to regional contrasts in aeolian sediment supply [20]. That study showed that heuweltjies achieved the greatest diameters and covered a larger fraction of the landscape in the western coastal region, where sparsely vegetated areas with sandy soils, relatively strong winds and predictable summer drought generate, mobilize and deliver an ample supply of aeolian sediments.

A recently proposed argument against the role of aeolian sediment accumulation as a principal process involved in formation of the mounds is that that heuweltjies differ in shape and spatial dispersion from another kind of aeolian landform called nabkhas, which are accumulations of wind-blown sediment beneath plant canopies in arid lands [46]. However, that conclusion limited the comparison to a specific type of nabkha that forms in response to strong, overwhelmingly unidirectional wind regimes. Wind regimes of that sort do not occur in the region occupied by heuweltjies, as demonstrated by monthly wind rose diagrams for locations throughout western South Africa (https://mesonet.agron.iastate.edu/sites/locate.php?network=ZA__ASOS). Furthermore, a previous comparison of heuweltjies to nabkhas portrays the process of accumulation of aeolian sediments in the two as similar, but with an important exception – ‘the result of territorial interactions between members of neighboring colonies [of termites]’ [17, p. 1907].

4. Nest structures directly built by M. viator

Although the recent advances in understanding outlined above indicate that heuweltjies are not formed by earth-moving activities of M. viator, colonies actively construct other structures as part of their nesting behaviour. Active, mature colonies of M. viator are not restricted exclusively to heuweltjies, and in some settings, colonies have other physical manifestations. In substrates where deep burrowing is possible, such as fine-grained alluvial deposits or deeply weathered, friable bedrock, a colony constructs and inhabits a single, spheroidal, subterranean hive within a cavity excavated at a depth of a metre or slightly more beneath the surface. In such settings, no type of epigeal mound appears on the surface above the hive (figure 2a,b). The hive contains sub-horizontal layers of comb-like sheeting composed of dark papery carton [50,55] and serves as the central focus of the colony and a protected location for the care and rearing of eggs and developing nymphal stages. The single reproductive female of a colony (the queen) is not confined to a specially constructed cell or royal chamber within the hive, as are for example, queens of Macrotermes spp., but can move around freely and if necessary, can exit the central hive via tunnel-like passageways [55]. Microhodotermes viator is a member of an older, basal group of termites [56], and does not cultivate fungi in special subterranean chambers within the mound structure as do some members of more evolutionary advanced groups, such as Macrotermes. Instead, M. viator consumes plant material directly, with digestion aided by intestinal protozoans and bacteria. Stores of harvested food items, consisting of small sticks and other dry plant materials, are stashed in small chambers, typically a few centimeters in diameter, that can be reached via passageways that radiate from the hive [55]. Colonies of M. viator that inhabit heuweltjies construct hives with the same shape, composition, structure and size, and also create the same kind of food storage chambers within the soil mass of heuweltjies ([25,35], JR McAuliffe 2012, 2016, personal observations).

Figure 2.

(a) Vertical exposure through an alluvial terrace along an ephemeral stream located west of Prince Albert, Western Cape, South Africa. Oval structure is the remains of a subterranean hive of Microhodotermes viator. The structure is 38 cm wide and its upper edge is located 80 cm below the soil surface. (b) Close-up view of cavity that originally contained the nest. The sub-horizontal layers of carton shelving have nearly completely disintegrated. Pictured is M. Timm Hoffman; 8 October 2016. (c) Small conical mound located 10 km southeast of Nuwerus, Western Cape, South Africa, on thin soil (10 cm thick) underlain by gneiss bedrock. The triangular patch of darker material consists of soil material recently expelled by termites. Inset, lower right granular structure of expelled soil materials; granules are 1–2 mm in diameter. (d) Close-up view of the same mound with expelled soil materials removed near the apex, revealing the portal (white arrow) through which the soil materials are expelled by termites. Massive solidification of some of the previously expelled granular materials has occurred, indicated by dark shadow above fingers, October, 2022. (e) Cross-section through a conical mound of M. viator revealing the basal positioning of the spheroidal nest. (f). Close-up photograph of spheroidal nest showing the structure of hive, composed of dark, organic carton. The mound had been previously attacked by an aardvark, which hollowed out and destroyed a portion of the hive, and apparently the entire colony. 25 km northeast of Vanrhynsdorp, Western Cape Province, South Africa, October 2016. Photographs by JR McAuliffe.

In places where shallow bedrock or other impenetrable barriers prevent deep excavation and positioning of a subterranean hive, colonies of M. viator construct small, conical mounds that typically reach maximum heights of 1 m (figure 2c). Coaton [57, p. 324] commented that such mounds ‘…may be expected to occur only where erosion has proceeded for so long that there is virtually no topsoil left at all’. In contrast to the unconsolidated surface soil of heuweltjies, conical mounds are very solidly indurated and extremely difficult to penetrate with a shovel, especially on larger, apparently older structures ([57], JR McAuliffe 2012, 2016, 2022, personal observations). Conical mounds are formed as the termites expel loose, slightly moistened, granular soil materials from temporary tunnel openings in the mound surface (figure 2c,d). However, unlike the manner in which Macrotermes construct their elaborate mounds with the deliberate placement of soil material mixed with salivary secretions [30], workers of M. viator do nothing to position or cement the granular expelled materials ([50], JR McAuliffe 2012, 2016, 2022, personal observations). Instead, the conical form develops through the gravitational sliding of freshly expelled materials down the inclined sides of the growing mound. Moistening and drying eventually harden the expelled soil materials into a solidly indurated mass. The spheroidal hive composed of sheetings of dark carton is constructed within a centrally located cavity at or slightly below the base of the mound, with small passageways leading from the hive to the surface of the mound; surface openings are fully closed when not in use (figure 2e,f).

5. Factors responsible for formation of heuweltjies in some environments, but not others

The physical environment, including the nature of the geological substrate, climate and aeolian sediment supply, all apparently contribute to whether or not heuweltjies can form in a particular location, and the kind of nest structures that are constructed and occupied by colonies of M. viator [17,25]. The capacity of vegetation to respond to nutrient-enriched patches of soil generated by the presence of a termite colony depends strongly on climate, particularly precipitation. Addition of nutrients to the soil environment has little impact on vegetation productivity in extremely arid environments due to the overwhelming limitation of plant-available moisture [58,59]. In more arid portions of the Succulent Karoo, where heuweltjies are absent, colonies of M. viator nevertheless generate patches of nutrient-enriched soil, yet denser vegetation required to create aeolian sediment traps does not develop [17]. The absence of heuweltjies in those locations, despite the presence of regularly spaced colonies of M. viator, is explained by this climate-dependent lack of a marked vegetation response.

At the other end of the climate continuum where ample moisture sustains closed-canopy vegetation of tall, dense shrubs (e.g. fynbos), colonies of M. viator also locally alter soil conditions. Although those alterations significantly affect plant species composition, they do not produce sufficient changes in the physical structure of vegetation (canopy cover, density and height) to contrast greatly with the surrounding vegetation. Consequently, distinct, island-like aeolian sediment traps do not develop in more mesic environments [17]. Given a sufficient aeolian sediment supply, heuweltjies apparently develop only within a restricted range of semi-arid environments—neither too dry nor too moist—where localized soil enrichment is capable of generating vegetation patches that are significantly denser than the sparser, surrounding vegetation (figure 1). In environments where heuweltjies do not form (or where old heuweltjies have eroded), depending on substrate conditions, the termites either occupy nests positioned deep within the soil or at the base of the protective conical mounds where soils lack sufficient depth due to impenetrable barriers such as solid bedrock [17,25].

Large heuweltjies and the small conical mounds can occur together in the same area, with the conical mounds typically located on thin soils in spaces between existing heuweltjies. In these settings, the conical mounds are typically small, representing the activity of relatively new, recently established colonies. However, survival and longevity of such nascent colonies are likely limited, due to territorial interactions with larger, well-established colonies that occupy neighbouring heuweltjies, and also to predation, particularly by aardvarks that readily access the colony by digging through the walls of the small mounds (see §6). At sites where much of the fine soil mass of heuweltjies has been removed through erosion by water, colonies of M. viator remain, but reside in the small, cemented conical mounds of their own construction, often within the eroded footprints of former heuweltjies [17]. These variable manifestations demonstrate the capacity for flexibility in nesting behaviour as a function of localized physical constraints.

6. Functions of nesting structures built by M. viator

(a) . Predation

Construction of the spheroidal hive (figure 2a,b,e,f) is universal in M. viator, whether or not a colony occupies a heuweltjie. However, the hive, as the central colony focus, is particularly vulnerable to attack by predators, especially aardvarks, as is also the case for other termite species in Africa [27]. Aardvarks are capable of complete destruction of colonies of M. viator. In a monograph on the harvester termites (Hodotermitidae) of southern Africa, which includes M. viator, W.G.H. Coaton¹ remarked: ‘Primary place amongst these predators must be given to the ant-bear, Orycerotopus afer, the only natural enemy which digs down and destroys the harvester when they are concentrated in their hives, and therefore can and does destroy the functioning reproductives of the colony’ [50, p. 48].

Colonies of M. viator exhibit what appears to be one of two general strategies to lessen this risk of destruction by aardvarks: (1) locating the hive at considerable depth in substrates where excavation by the termites is possible (in either the fine-grained soils of heuweltjies or other deep soil environments; figure 2a,b) or (2) construction of the protective, indurated conical mound directly above the hive in substrates where deep excavation is impossible (e.g. thin soils underlain by impenetrable bedrock; figure 2c,d).

The importance of predation by aardvarks as a major selective pressure that has shaped the evolution of nesting structures built by M. viator is borne out by the proportion of mounds that are penetrated by excavations made by this mammal. Whether nests are located in heuweltjies, beneath small conical mounds, or beneath the surface with no overlying epigeal mound, foraging excavations by aardvarks typically occur in more than half of all nest locations, with the exception of reports from one locality (table 1). However, the three different manifestations of nesting behaviour likely vary considerably in the risk of complete destruction of a colony. The confined placement of the spherical hive near the base of a conical mound is likely accompanied by a much greater risk of complete destruction. Aardvarks attack conical mounds by excavating at the base of the structure to gain direct access to the central hive (figure 3). In such a position, the termites, including the colony's reproductive pair, are at particular risk due to the lack of subterranean areas to which they can retreat.

Table 1.

Occurrence of foraging excavations by aardvarks (Orycerotopus afer) on colonies of the southern harvester termites (Microhodotermes viator) located in heuweltjies, conical mounds and sites without surface mounds.

location of internal nests	percent with aardvark excavationsa	total examined	location & date of data collection	latitude & longitude	reference and date of data collection
within heuweltjies	75	(visual estimation without data)	near Clanwilliam	—	[34]
					1980s
	53	63	near Clanwilliam	32.2090°S	[35]
			Dam Lake	18.8794°E	1980s
	53	15	Inverdoorn	32.1620°S	b
				19.7440°E	October 2012
	65	20	Soebatsfontein	30.1770°S	b
				17.5540°E	October 2012
	70	10	Namaqua	30.0182°S	b
			National Park	17.4529°E	October 2012
within small, conical mounds	71	28	Kapelsfontein	32.6750°S	c
				19.7164°E	October 2012
	86	21	East of Nuwerus	31.1367°S	c
				18.7531°E	October 2016
	35	143	near Beaufort	—	[61]
			West		Sept. 1962
	12	992	near Beaufort	—	[61]
			West		May 1963
	10	1154	near Beaufort	—	[61]
			West		April 1965
subterranean, lacking any surface mound	58	29	Tankwa Karoo	32.2206°S	b
			National Park	19.7050°E	October 2012

^aPublished values rounded to nearest whole percentage.

^bUnpublished data collected by MT Hoffman, MP King and the author; site descriptions in [23,51].

^cUnpublished data collected by JR McAuliffe, October 2016; site description in [51].

Figure 3.

(a) Foraging excavation by an aardvark at the base of a conical mound. The eroded surface of the mound indicates a lack of termite activity, likely due to complete destruction of the colony by predation. Soils at the site are extremely thin (less than 25 cm) over bedrock; consequently, internal spherical hives within the mounds are located at the approximate level of the mound base. In such settings, the indurated conical mound provides the only protection from predation by aardvarks. Pictured is M. Timm Hoffman, October 2012, at the Kapelsfontein site location (table 1). (b) Close-up view of recent excavation by an aardvark into the base of a conical mound. Subparallel, linear cuts through the exposed surface below the pocketknife were made by the stout digging claws of an aardvark. The red portion of the knife is 9 cm in length. Northeast of Vanrynsdorp, Western Cape, South Africa, October 2016. Photographs by JR McAuliffe.

By contrast, for colonies inhabiting subterranean nests within either the large earthen volumes of heuweltjies or in deeper soils where epigeal conical mounds are not constructed, typical foraging excavations by aardvarks likely, on average, inflict far less damage to a colony than to one inhabiting a conical mound. Aardvarks locate termites principally by olfaction [62], and the presence of termites on or immediately beneath the soil surface at foraging ports distributed about the large surface of a heuweltjie likely elicits burrowing by aardvarks in places that can be far removed from the location of a deeper, more distantly located hive. This is indicated by the frequent occurrence of aardvark foraging excavations at multiple locations across the surfaces of individual heuweltjies (JR McAuliffe 2012, 2016, personal observations). Additionally, when the hive of a colony is positioned at greater depth and within the much larger soil volume of a heuweltjie, there is probably a greater possibility for survival of a significant fraction of the colony, including the reproductive individuals, by retreat from the central hive via the many small access tunnels leading away from the hive. As mentioned previously, the reproductive individuals in a colony of M. viator are not confined to an enclosed royal chamber and are free to move about and even exit the hive, including when the termites apparently detect the mound being penetrated [50]. However, the escape and persistence of a critical fraction of a colony are likely greater for those positioned within large soil volumes of either heuweltjies or in soils where deep placement of the hive is possible.

The death of the primary reproductives (the paired female and male responsible for founding of the original colony) due to predation does not necessarily cause complete destruction of a colony of M. viator, as long as a sufficient portion of the colony remains. As with termites in general, with the of loss of the primary reproductives, substitute reproductive females and males can develop from mature, pre-dispersal individuals of the alate class (winged individuals that are capable of dispersal and establishment of new colonies), replacing the original primary reproductives, allowing the colony to persist. Additionally, if fully developed alates are not present to assume this function, immature alate nymphs become sexually mature, developing into what are known as secondary reproductives and allowing the colony to survive [50]. Despite this possibility, the average longevity of any particular colony of M. viator located within the protective volume of heuweltjies or deep within other soil environments is likely much greater than colonies inhabiting conical mounds, which provide a predator like the aardvark with a much better-defined target to attack. Given that conical mounds are constructed only in environments where deep excavation and positioning of the hive are impossible, the conical mounds represent a behavioural manifestation that affords at least some measure of defense against predation by aardvarks.

(b) . Other possible functions of conical mounds

The massive composition of conical mounds contains small tunnels that reach the surface, but lack large internal open spaces or prominent external openings. Consequently, it is unlikely that the mounds serve a primary function of ventilation as accomplished by the complex mound structures of the fungus cultivating termites Macrotermes spp. [5,6]. Close regulation of temperature, humidity and CO₂ concentrations within the mounds of Macrotermes is essential due to the narrow tolerances of the Termitomyces fungi cultivated by the termites [3]. Comparable maintenance of such a narrow range of atmospheric conditions within the nest environment would be unnecessary for colonies of M. viator, as they do not rely on fungal symbionts. However, there is limited evidence that the internal positioning of the spherical hive within the conical mound may influence temperatures experienced by the colony, particularly during cooler nights. Van Ark [61] found that the internal spheroidal hives within conical mounds were positioned slightly closer to the northern surfaces of mounds (i.e. facing the equator), suggesting a thermoregulatory function of such positioning.

7. Benefits afforded to termites by heuweltjies

In addition to how positioning of the central hive of M. viator deep within the soil mass of a heuweltjie probably provides greater protection from burrowing predators, the mounds likely provide other benefits. Turner [15] claimed that more favourable soil moisture conditions within heuweltjies are the main factor leading to continued occupancy of the structures, and those more favourable conditions are due to soil modifications created primarily by the excavating activities of termites. Evidence does exist for the more equable soil moisture conditions of heuweltjies: soils of heuweltjies more readily absorb precipitation than those of surrounding areas [63], and soil water contents in heuweltjies are generally higher and persist longer than in surrounding soils [64,65]. However, the more favourable soil moisture environment cannot be attributed exclusively to the direct actions of termites. Rather, those conditions are a consequence of a chain of biotic and abiotic processes that are responsible for formation of the large mounds (e.g. development of denser vegetation cover and the resultant trapping of wind-blown, silt-rich sediments that have higher moisture holding capacity). Furthermore, the development of heuweltjies is highly dependent on the physical environmental context, and not solely the direct actions of termites. In the more arid environments where water availability strongly limits vegetation development, trapping of aeolian sediments does not occur, despite localized alteration of soil conditions by the termites (nutrient levels, permeability, etc.; [17]).

8. Are heuweltjies extended phenotypes of M. viator?

No. Evidence indicates that heuweltjies are not ‘built structures’, created through the earth-moving activities of M. viator. Rather, the formation of heuweltjies represents a response of the physical environment to the windbreak effect of a denser vegetation patch (localized reduction of wind velocity and resultant deposition and accumulation of airborne sediment). Although the termites are directly responsible for localized soil enrichment that promotes development of patches of denser vegetation, subsequent aeolian deposition and formation of a heuweltjie represent an incidental environmental response, one step removed from the direct impacts of the termites on soil nutrient content, rather than an extended phenotype of M. viator.

The only structures associated with nests of M. viator that can justifiably be regarded as extended phenotypes are the spheroidal hives and indurated conical mounds (figure 2). Both are phenotypic characters expressed outside the collective body of the termite colony through direct construction of artefacts by the termites. The spheroidal hives represent a universally expressed extended phenotype of M. viator, regardless of whether a colony resides in a heuweltjie, beneath the base of a small, conical mound or deeply positioned within the soil. However, creation of the conical mounds is a behavioural manifestation that depends on the environmental context. Environmental context-dependent expressions of nest construction behaviour also occur in other termite species [2,3,66], ants [67], fish [68] and birds [69–71]. Perhaps the best known example consists of structures built (or not built) by the American beaver (Castor canadensis). Dams and accompanying lodges constructed by beavers are referred to repeatedly by Richard Dawkins in The extended phenotype [1], and also commented [72, p. 379] that beaver dams ‘are true extended phenotypes insofar as they are adaptations for the benefit of replicators (presumably alleles but conceivably something else) that statistically have a causal influence on their construction’. However, expression of the beaver's extended phenotype in the form of these constructed artefacts is entirely dependent on the environmental context—beavers do not always construct dams and accompanying lodges. In large lakes with rocky shores, dams serve no function and beavers simply construct lodges in shallow water but separated from the shoreline. In small streams with earthen banks, dams are built, but instead of constructing stick lodges within the resulting pond, beavers excavate burrows with underwater entries for their dens in earthen banks along sides of the impounded stream. In large, deep and swift rivers where construction of dams or lodges is neither feasible nor functional, the animals den exclusively in excavated burrows with underwater entries within earthen riverbanks ([73]; JR McAuliffe 1974–2009, personal observations).

9. Does M. viator function as an ecosystem engineer and niche constructor?

Microhodotermes viator clearly functions as an ecosystem engineer, but the types of changes induced by the termites depend on the physical environmental context. Colonies act as allogenic engineers [7], in that they modify environments by changing materials (i.e. soil) from one state to another. The termites initiate environmental change by accumulating plant nutrients in patches centred on colony locations. That primary change has the potential to generate a series of subsequent transitions involving plant responses to nutrient-enriched soils and eventual aeolian sediment deposition beneath patches of denser vegetation. The termites are primary ecosystem engineers, but subsequent vegetation responses to soil enrichment serve as the proximal factor responsible for trapping and retaining aeolian sediments. In this sense, termites and plants both operate as ecosystem engineers (i.e. multiple engineers [74]) in the formation of heuweltjies through their collective alteration of the environment.

Other than the impacts of human beings and their domestic livestock during the past two centuries, there is probably no other animal other than M. viator that has so profoundly influenced ecosystems of the Succulent Karoo Biome of western South Africa. Colonies of this termite, through localized alteration of soil conditions and the accompanying cascade of further biological and physical changes, strongly influence the region's biodiversity and the appearance of entire landscapes (figure 1). Environmental conditions offered by heuweltjies, including unique soil conditions and vegetation cover, provide essential habitat and food resources to a wide variety of animals that would likely be rare or absent were it not for the existence of heuweltjies [38]. Other termite species exert comparable influences [9–14]. Similarly, other communal nesting organisms, through environmental alterations they create directly or set in motion, have the capacity to profoundly influence local biodiversity. For example, sociable weavers (Philtairus socius) of arid- to semiarid environments of southern Africa construct large communal nests up to several metres in width and depth that resemble irregular haystacks perched within trees. The nests can contain hundreds of individual nest chambers occupied by the birds and the structures may in some cases persist for a century or more. These large, persistent structures modify the environment in many ways through the formation of a fertile soil island beneath the nests and by ameliorating local environmental conditions. These modifications benefit a wide variety of invertebrate and vertebrate animal populations, increasing biodiversity, particularly in more the more arid regions where sociable weavers are found [75,76].

Niche construction considers the evolutionary implications of environmental changes created by organisms and is defined as the ‘The process whereby organisms, through their metabolism, their activities, and their choices, modify their own and/or each other's niches’ [77. p. 8]. Aspects of this body of thought overlap broadly with that of ecosystem engineering, but additionally it considers the role of organism-induced environmental change in modifying selective pressures, thereby influencing the evolution of the species generating environmental change, as well as others that occupy the same ecosystem. Given its extremely broad definition, colonies of M. viator clearly are ‘niche constructors’, just as they are ‘ecosystem engineers’ in that their occupation of a patch within the landscape fundamentally alters soil conditions, which sets further environmental change in motion.

However, a controversial aspect of niche construction theory is the concept of ecological inheritance, which refers to legacies of environmental change bequeathed by niche-constructing organisms to subsequent populations, thereby modifying selection pressures on descendent organisms (see criticism by R. Dawkins in [72] and a balanced appraisal of conflicting views in [78]). Heuweltjies have been cited as an example of ecological inheritance with the suggestion that ‘Adaptation of termites to the semi-arid environment of the Karoo involves more the evolution of heuweltjies rather than of the termites themselves' [15, p. 346]. However, this view, based on claims that M. viator ‘produces’ heuweltjies [15,18], greatly oversimplifies and misrepresents a complex system involving a combination of biotic and abiotic processes. In reference to research required in investigations of phenomena involved in niche construction, Odling-Smee et al. [77] commented ‘The key to progress is to break down complicated pathways into tractable component pieces'. To do so first requires a careful and detailed identification of those pathways. To date, the manner in which several publications on the theme of niche construction and ecological inheritance have used heuweltjies as a key example contributes to neither a substantive foundation for that body of theory, nor to the advancement of knowledge about this fascinating ecological system within the Succulent Karoo.

10. Practical considerations

The labels extended phenotype, ecosystem engineer and niche construction, when thoughtfully and appropriately applied, provide a useful shorthand for communicating general categories of evolutionary and ecological processes that can help focus further inquiry. Additionally, detailed and accurate depiction of the nature of these phenomena and processes has practical relevance that extends beyond mere academic debate. The Succulent Karoo Biome is recognized as an unusual biodiversity hotspot among all the arid and semi-arid regions of the world [79], and the unique vegetation composition of heuweltjies and accompanying habitat for all types of animals contributes significantly to this biodiversity [39,80–82]. An accurate and detailed understanding of the complex, linked set of biological and physical processes responsible for the formation of heuweltjies consequently can provide valuable information to guide conservation practice. Take, for example, two contrasting representations of how heuweltjies form: (1) the oft-claimed, simple conclusion that termites create the mounds versus (2) termites initiate a chain of biotic and abiotic transitions, culminating in deposition of aeolian sediments and mound formation. Viewpoint (1) could conceivably lead to the conclusion that conservation or protection of the termite populations in and of itself would be sufficient to the maintain the mounds and biodiversity associated with them. On the other hand, with viewpoint (2), the importance of the linked set of biotic and abiotic interactions comes to the forefront, and this can potentially lead to different conclusions regarding needs of conservation and land management practices, e.g. any use of landscapes that diminishes sediment trapping effects of vegetation on heuweltjies impairs or eliminates the very processes by which the mounds are formed and maintained [17,25]. In a system such as this, detailed knowledge of roles played by multiple processes matters greatly to understanding potential environmental trajectories of the future.

Data accessibility

All of the new data presented in the paper are included in table 1.

Authors' contributions

J.R.M.: conceptualization, data curation, formal analysis, funding acquisition, investigation, methodology, project administration, resources, writing—original draft.

Conflict of interest declaration

I declare I have no competing interests.

Funding

I received no funding for this study.