Architecture and morphogenesis in the mound of Macrotermes … pdfs/Cimbebasia 16_00.pdf ·...

Cimbebasia 16: 143-175, 2000 143

Architecture and morphogenesis in the mound ofMacrotermes michaelseni (Sjöstedt) (Isoptera: Termitidae,

Macrotermitinae) in northern Namibia

J Scott TurnerDepartment of Environmental & Forest Biology, State University of New York, College of Environmental

Science & Forestry, Syracuse, New York 13210, U.S.A.; e-mail: [email protected]

The mounds of Macrotermes michaelseni (Sjöstedt) in northern Namibia have acharacteristic architecture, being divided into three structurally distinct components: acentral cone-shaped mound, topped by a tall, thin spire which tilts northward at an anglesimilar to the sun's average zenith angle, and a broad outwash pediment that results fromerosion off the mound. Internally, the mound is permeated by a complex network oftunnels that can be divided into three broad types: a capacious central chimney thatextends upward through the center of the mound from the colony up to the apex of thespire; a vertically-biased network of surface conduits that lies roughly one to two cmbelow the surface, and an interweaving network of lateral tunnels that connect thechimney and surface conduits. A model of mound morphogenesis is proposed whichaccounts for most of these structural features, and which points toward a general modelfor the relationship between mound architecture and social homeostasis among themacrotermitine termites.

INTRODUCTION

The mounds of the macrotermitine termites ofthe genus Macrotermes Holmgren, 1909, areone of southern Africa's most remarkablelandscape features. These structures arecommon throughout a variety of habitat types,occurring at densities of roughly one to four perhectare (Abe & Darlington 1985; Grassé &Noirot 1961; Pomeroy 1977; Ruelle 1964,1985). In some regions, these mounds canattain heights of several metres. In Namibia,the genus is represented by four speciesnamely: Macrotermes natalensis (Haviland,1898), M. subhyalinus (Rambur, 1842), Mvitrialatus (Sjöstedt, 1899) and M.michaelseni (Sjöstedt, 1914), with theircombined ranges occupying most of thecountry to the north and east of the 200 mmmean annual rainfall isohyet, which runsroughly from Mata Mata in the southeast to theRuacana Falls in the northwest (Coaton &Sheasby 1972; Ruelle et al. 1975). This rangeencompasses most of the arid Savanna Biomein Namibia, as well as some marginal areas ofthe Nama Karoo Biome.

In the case of M. michaelseni, the mounditself is not a habitation for the colony. Thecolony proper, i.e. the locations where thequeen, workers and fungus gardens reside, issituated below the mound, occupying aspherical volume roughly 1.5-2 m in diameter(Darlington 1987). Given the prodigious task ofconstructing the mound, it is worth asking whythe termites should build such a structure if notto live within? Currently, most entomologistsview the mound as playing some rôle in theregulation of the colony environment, such asits temperature, humidity or the concentrationof respiratory gases (Darlington 1987; Lüscher1956, 1961; Wilson 1971). The mound does thissupposedly by promoting ventilation of thecolony, and the various mound architecturesamong the macrotermitines seem well suited toparticular ventilatory mechanisms (Darlington1987; Lüscher 1956, 1961; Ruelle 1964). Forexample, completely enclosed mounds, such asthose constructed by Macrotermes natalensisor M. michaelseni, are permeated with anetwork of passageways that are seeminglyorganized to capture energy dissipated from

Cimbebasia 16,2000 144

colony metabolism, and to use it to drive acirculatory flow of air within the mound(Darlington 1985, 1987; Lüscher 1956, 1961;Wilson 1971). Other mounds, such as thosecommonly built by M. subhyalinus, have largediameter openings, sometimes at the tops ofvertical chimneys. These presumably capturewind energy to drive air through the colony in aone-way flow (Darlington 1984, 1987; Ruelle1964; Weir 1973), by a mechanism known asinduced flow (Vogel 1978; Vogel & Bretz1972).

This neat correspondence between mound structureand function implies that termites construct moundsaccording to a certain plan, and that naturalselection has refined this plan so that mounds havethe ‘proper’ architecture for the particularventilatory mechanism they employ. This approachis fraught with difficulty, though, in part because itimplies the mound to be the superorganismalanalogue of phenotype, with all the attendantdifficulties and ambiguities that are inherent in thesuperorganism concept (Wilson 1971). Thechallenge for students of social insects is to explain,in a reasonably reductionist way, how moundarchitecture and function arise, and in the context ofthe rather limited repertoire of behaviours thatworkers can bring to the construction of the mound.

As part of a broader exploration of the respiratoryfunction of termite mounds, I have undertaken adetailed examination of the internal structure of themounds of Macrotermes michaelseni (formerlymossambicus: vide Ruelle 1975) in northern Namibia.Although the structure of the M. michaelseni moundhas been described elsewhere (Darlington 1986),descriptions of mound architecture among themacrotermitines have, in the interests of descriptiveclarity, tended to omit a great deal of the structuralcomplexity of these mounds (Darlington 1984; Inoueet al. 1997; Lüscher 1956, 1961; Ruelle 1964). Themound is a complex structure, however, and I havetaken the approach in this paper that this complexitymust be part of a full understanding of basicquestions of mound morphogenesis and of the linkbetween mound structure and function.

C39

D2780

C38

Outjo

Namatubis

Namibia

AtlanticOcean

Windhoek

Out jo

Figure 1. Location of the study site. Thelocation of the farm, Namatubis, is indicated bya filled star.

MATERIALS AND METHODS

The study areaThe study area was located in northern Namibia, ona private farm roughly 22 km north of the town ofOutjo, Outjo District, Kunene region (Figure 1). Thefarm, Namatubis, consists of several fenced camps,each roughly 300 ha, used for rotational grazing ofcattle and domesticated game. The study area waslocated on one of these camps, 289 ha in extent,centered roughly at the coordinates 16°4.50’E,19°59.05’S. The habitat is characterized by Coaton& Sheasby (1972) as mopane savanna, dominatedby Colophospermum mopane (Kirk ex Benth.) Kirk exJ. Leonard (Fabaceae) intermixed with other trees,including Acacia tortillis (Forssk.) Hayne, Acaciaerioloba E. May., Acacia mellifera (Vahl) Benth.,Mundulea sericea (Willd.) A. Chev. (Fabaceae),Croton gratissimus Burch. (Euphorbiaceae), Petalidiumengleranum, (Schinz). (Acanthaceae), and Ximeniaamericana L. (Olacaceae). The terrain is flat, withsandy red soil overlying a shallow calcareous basewhich rises occasionally to the surface as chalky

Turner - Architecture and Morphogenesis of Macrotermes mounds 145

outcrops. The habitat supports a rich grasspopulation, mostly Stipagrostis Nees, and this in turnsupports a rich termite fauna, dominated byMacrotermes michaelseni, but with evidence ofabundant harvester termite (Hodotermesmossambicus) activity as well. A rich geophyte florais also present, but these are evident only after thespring rains in January. There was no evidence ofovergrazing in the habitat, such as encroachment byAcacia mellifera, and it appeared to be typical of thenatural vegetation of this region.

Survey methods and general measurement of moundshape and sizeA systematic survey of mounds on this camp wasconducted which fixed each mound’s location andrecorded simple measures of mound size and shape.The survey was carried out by dividing the campinto 80 north-south corridors, 0.05’ of longitudewide (roughly 92 m), and 0.2’ of latitude deep(roughly 368 m). Demarcation points for eachcorridor were located using a hand-held globalpositioning system (GPS) receiver (Garmin GPS II).

The survey was carried by a team of four to sixvolunteers, who swept locating each mound andfixed each mound’s location and measured its size.Size measurements included the mound’s basalwidth along its north-south axis, db, and themound’s height, zM. Both were measured indirectly,according to the following procedure (Figure 3).One volunteer, the instrument operator, wasstationed at some convenient distance to the east orwest of the mound (A), whichever afforded theeasier access and greater visibility. Anothervolunteer, the staff holder, held a vertical staff ateither the north or south base of the mound (B orC). The staff was equipped with a sound reflectingtarget which was aimed at the instrument operator.Distance from the instrument operator to the staff(segment AB or AC) was then measured using anultrasonic range finder. The compass heading from

the instrument operator to the staff holder (angles αor β) was also measured using an electronicsighting compass. The procedure was then repeatedfor the opposite side of the mound. Once distances

19 59.8’So

19 59.3’So

19 59.8’So

16

3.8

’Eo

16

4.3

’Eo

16

4.8

’Eo

16

5.3

’Eo

Latitude

LongitudeFigure 2. Detailed map of the camp where the studies were carried out, and the locations of themounds of Macrotermes michaelseni residing therein. The heavy dotted lines indicate fencelines. The fine solid lines indicate the quadrats used in the censusing sweeps. Locations ofindividual mounds are indicated by open diamonds. Water points are indicated by well towers. Thesighting disk symbols indicate base points for the GPS receiver.


(segments AC and AB) and directions (α or β \) tothe north and south bases of the mound wasmeasured, the instrument operator then measured

the angle (angle δ) to the mound’s apex, E, using asighting angle meter. From these measurements,basal diameter (segment BC) and height of themound (segment A’B) could be calculated

trigonometrically. The angle γ was calculated as thedifference in compass directions to points C and B:

γ = (α - β) [1]

The basal diameter, BC, was calculated using theLaw of Cosines:

BC = √(AC2 + AB2 - 2 AC AB cosγ) [2]

The height is estimated from the angle δ, thedistance DF, and the distance from the ground tothe eyes of the observer, AD. The distance DF is

estimated from the angle ε , which is calculated fromthe law of sines:

sin ε = AC sin γ / BC [3]

The length of segment DF is equivalent to thedistance A’A, and is estimated from the law of sinesas the bisector of the basal diameter, segment CB :

DF = CB sin ε / 2 sin (δ/2) [4]

The length of segment EF is then estimated usingthe trigonometric identity:

EF = DF tan δ [5]

Height, A’E is then the sum:

A’E = AD + EF [6]

The mounds of Macrotermes michaelseni exhibit apronounced northward orientation, both in the tiltof the spire and in the skew of the conical base. Toquantify this, the azimuth and zenith of the moundswere measured. Azimuth is the heading on thehorizon to which the mound points. Zenith is theangle of tilt from the vertical. To estimate azimuth,an observer stood roughly 10 m to the north of themound and moved to the east or west until themound profile seemed to the observer to be themost symmetrical. The azimuth was then estimatedas the complement of the compass heading from theobserver to the apex of the mound. Two methodswere used to estimate zenith. In method A (Figure4), the mound’s basal diameter, db, was estimated

db

lN

lS

β φ γ

α

θ

N

Figure 4. Method A for estimating zenith ofthe mound spire. Details in text.

A

B

αβ

a.C

A’ γ

ε A

D

E

δ

b.

F

A’

Figure 3. Measurement protocols for dimensions of mounds measured during the surveysweeps through the study area. Details in text.


from the circumference of the base of the mound, cb,measured using a steel tape and the formula:

db = cb / π. [7]

Measurements were then made of the distancesfrom the base of the mound to its apex at the northand south sides, lN and lS, respectively. The threesides, db, lN and lS, form a triangle that correspondsto the profile of the mound when viewed from the

east or west. The angles α, β and γ, can beestimated from the law of cosines:

cos α = lN2 + lS

2 - db2 / 2 lN lS [8a]

cos β = lN2 + db

2 - lS2 / 2 lN db [8b]cos γ = lS

2 + db2 - lN

2 / 2 lS db [8c]

The zenith is indicated by the angle φ formedby the bisector of angle α and db, which formsa new triangle with angles β , α/2 and φ . Thezenith, θ is readily calculated as:

θ = 90 -φ = 90 - (180 - β - α/2) [9]

Method B for estimating the zenith used a pivotingsighting protractor, consisting of a 2 m long staff ofwood with a wooden strip mounted on it so that thestrip could pivot about a known point (Figure 5). Anobserver was stationed roughly 20 m from the eastor west face of the mound, whichever offered theclearest view. Another pair of volunteers wasstationed about midway between the observer andmound. One of these volunteers held the staffvertically, maintaining its position with the aid of aspirit level. The other volunteer adjusted the pivotuntil the observer agreed that it paralleled the spire.The zenith was then estimated using measurements

from the protractor. Two lengths, designated lV andlA, were marked lengths on the staff and pivot,respectively, and with the pivot point as theircommon origin. The distance between the endpoints of lV and lA was then measured using a steel

tape measure (lT). The zenith θ is then estimatedfrom the law of cosines:

cos θ = lA2 + lV

2 - lT2 / 2 lA lV [10]

Mound sectioning methodsA mound’s internal structure was assessed bysectioning it in a series of flat sections, similar inprinciple to the sectioning of a tissue block by amicrotome. Three types of sectioning wereundertaken. In one, designated horizontal section,the mound was sectioned from top to the base in 10cm horizontal slices. In the second, designatedvertical section, the mound was sectioned in 10 cmvertical slices from one side to the other. The third,designated surface section, peeled away the surfacein 1 cm slices at four sites located at the mound’scardinal points. In all types of section, the exposedsection face was photographed with ASA 200 slideand print film. Outlines of the mound section and of

db

lN

lS

β φ γ

α

θ

N

Figure 4. Method A for estimating zenith ofthe mound spire. Details in text.

N

lV

lA

lT

a.

b.

Figure 5. Method B for estimating zenith ofthe mound spire. Details in text.


the exposed walls of the tunnels were traced usingthe digitized images of the section faces, and theimages analyzed as described below. To ensureproper alignment and scaling of the photographs inthe sequence, fiduciary points were set as indicatedbelow.

Horizontal sectionOne mound, about 170 cm tall, was sectionedhorizontally. No fiduciary points were set in thiscase, but alignment of the section sequence wasestimated using landmarks visible in thephotographs. Each section proceeded as follows. Arough cut was first made using a trenching tool andtrowel. Once the overburden of the mound abovethe section was removed, the exposed face wasflattened and smoothed by a progressively finerseries of cuts, first with the flat metal edge of aplastering trowel, and finally with a stiff brush. Theexposed face was then photographed and theprocedure repeated.

In the lower sections, the exposed face had to bereconstructed from several photographsreassembled into a photomontage. This, along withuncertainties in the positions of the alignmentlandmarks made it impossible to trace withconfidence the passage of particular tunnels fromone section to the next. This mound was thereforeused only for qualitative description and formeasurements of the tunnels and exposed faces.

Vertical sectionsThree mounds were sectioned vertically. In one ofthe three, serious errors arose in the alignment ofthe fiduciary points and the information from thismound was therefore not used. In the remainingtwo, one (designated M03) was sectioned from eastto west, while the other (designated M43) wassectioned from north to south. In both mounds,fiduciary points were established by setting up leveland parallel reference lines on either side of themound. The lines were leveled using a spirit linelevel. Parallelism of the lines was ensured bymeasuring the perpendicular distance between themat several points. The reference lines weresupported on stakes held in place by guy lines.Errors due to stretching of the reference lines werecontrolled for by marking reference points on eachline with brightly coloured nail varnish, which couldbe aligned with the stakes. Fiduciary points weremarked by a small bow of flagging tape tied looselyto each reference line so that the bow could slidealong it and be aligned with the cut vertical face of

the mound. Parallax errors were minimized by alwaysplacing the camera at a set distance from the moundface, i.e. the position of the camera advanced alongwith the sections through the mound. Each moundface was photographed in a single frame, i.e. noreconstructions of the image throughphotomontage were necessary for these sections.However, several close-up photographs of the facewere made for verification of the features in theimage of the entire face.

The procedure for vertical sectioning was similar tothat used in the horizontal sections, that is a roughcut, followed by progressively finer cuts until asmooth vertical face was ready for photographing.The first, or rough cut was laid out by tracing a line10 cm behind or below the previously cut face,judged by direct measurement from the previouslycut face, or with reference to the fiduciary points onthe reference lines. The rough cut was made withthe flat blade of a trenching tool, followed by a finerset of cuts with the flat blade of a shovel.Recalcitrant portions of the mound face wereremoved with a broad-bladed cold chisel andhammer. The mound face was then smoothed withthe flat edge of a trowel blade, and finally ‘polished’using a stiff brush.

Surface sectionsSurface sections were made by ‘peeling’ the moundsurface away in 1 cm layers. These dissections wereintended to reveal the structure of the surfaceconduits that underlie the mound’s surface, and toassess their proximity to the surface, theirrelationship to areas of surface building, and theirextent.

Surface sections were made at the cardinal pointson the mound surface, roughly 1.5 metres aboveground level. The cardinal points were fixed bysighting with an electronic compass. At eachcardinal point, a 40 cm by 40 cm square was laid outon the mound surface using a steel tape and chalkline. Fiduciary points were set at points just outsidealternate corners of the square, and marked with apiece of flagging tape held in place by nails driveninto the mound. A photograph was then taken ofeach mound surface. Soil was then scraped off to adepth of 1 cm, using a narrow bladed trowel. Theexposed face was then smoothed as much aspossible, excess dust removed with a brush, andthen photographed. This procedure was repeatedtwice more down to a depth of 3 cm. By this depth,the sectioning had usually penetrated well into thenetwork of tunnels in the mound.


Analysis of imagesA flat bed scanner was used to digitize the imagesat a resolution of 300 dpi. The digitized image wasthen imported into a graphics program (CorelDrawv8), and tracings made of the outlines of themound’s exposed faces or of the exposed surfaceconduits. The digitized images were also used forconstruction of photomontages in the horizontalsections, and for making any scaling and alignmentadjustments necessary. The tracings were thenexported to IntelliCAD, in which the areas andperimeter lengths of the outlines of the tunnels andsection faces could be measured. Thesemeasurements were: the length of the mound’souter surface outline, lm ; the section area of theexposed face, Am; the combined circumferences ofthe tunnels in the exposed face, lt; and the combinedareas of tunnels in the exposed face, At. From thesemeasurements, several subsidiary measurementswere calculated. Mound surface area, AM, wascalculated as the summed outlines of the moundmultiplied by the section thickness, lsx:

AM = lsx Σ lm [11]

Mound volume, VM, is similarly calculated from themound section areas:

VM = lsx Σ Am [12]

Tunnel surface area, AT, was estimated from thetunnel outline lengths:

AT = lsx Σ lt [13]

and tunnel volume, VT was estimated from thesummed tunnel areas in the sections:

VT = lsx Σ At [14]

Rotation of the spire and subsequent effectsTo assess the effect of tilt angle on the distributionof temperature and insolation in the spire, spires oftwo mounds were detached and rotated by 180o.The procedure for rotating the spire was as follows.A mound was chosen, and the cardinal pointsmarked on the spire’s surface using a sightingcompass. The cardinal points were marked usingnails driven into the spire and tagged with flaggingtape. Once the cardinal points were marked, thespire was wrapped in nylon webbing to keep it fromcrumbling during the operation. A horizontal cutwas then made through the spire with a bow saw,detaching it from the rest of the mound. Thedetached spire was then rotated 180o around itsbase, as determined by the formerly marked

southern cardinal point coming into alignment witha sighting compass stationed to the mound’s north.The spire was then unwrapped and left to stand fortwenty-four hours. By this time, the mound had‘healed’, with damage to the spire and the cutthrough it sealed by the workers.

Three mounds were treated in this manner. One, CV,served as a control, and was located in a sunny andopen situation. A second, T05, had its spire rotated,and was located in a sunny and open situationsimilar to mound CV’s. A third, T06, also had itsspire rotated, but was located in a shady shelteredsituation. During the course of a sunny day, surfacetemperatures and incident radiation were measuredusing a surface thermocouple and light meter,respectively. The three mounds’ temperatures andinsolation at each cardinal point were measuredevery thirty minutes from about 10:30 AM to 4:30PM, with measurements taken sequentially from thethree mounds, i.e. from CV first, then from T05, thenfrom T06. The mounds were close enough to oneanother so that all the measurements could be madeover a span of 10 minutes. Data were analyzed usinga repeated measures analysis of variance, with

spire

conical base

outwash pediment

Figure 6. A typical mound of Macrotermesmichaelseni at the Namatubis study site. Themound has an Acacia tree embedded within it.


mound face (N, S, E and W) and mound (CV, T05,T06) being the class variables.

Each cardinal point of the mound wasphotographed at the time. To assess the effect ofspire rotation on building activity, the mound faceswere photographed again after six months forcomparison.

RESULTS

External structure of the moundThe M. michaelseni mound typically consists of threestructurally distinct parts (Figure 6): a cone-shapedbase, from which a broad outwash pediment fans,and which is topped by a more cylindrical spire thatcommonly rises to about twice the height of thebase. Within this type, though, there isconsiderable diversity of shape. Most commonly,the variation comes in how distinct the spire iscompared to the conical base. In many mounds, thespire is nearly cylindrical, and contrasts stronglywith the sloping walls of the conical base. In othermounds, the base merges more or less imperceptiblyinto the spire, so that the mound’s overall shape isthat of a skewed cone.

The texture of the mound’s surface can usually bedifferentiated into two distinct types: a relativelysmooth and hard surface which has been exposedto smoothing by wind and water erosion, and ahighly friable and rough surface that is the site of

recent building activity. In roughly 10% of themounds, however, the surface was more uniform intexture, comprised of very loose soil that flakedwhen dry. This type of mound was usually darker incolour than the more common reddish sands thatcomprised most mounds, perhaps as a result of adifferent soil type being brought up from a different

soil horizon at the mound’s location.

The mound is covered to varying degrees by areasof new building (Figure 7). New building isevidenced by an increased surface roughness of themound, or if it is fresh enough, by a noticeabledampness. There seem to be two types of newbuilding. The spire usually shows signs of beingrecently worked, most intensely at the spire’s apex.Sites of new building also frequently occur at thesloping sides of the mound’s conical base. Thesesites of new building are dome-shaped, seeming tobillow out from the side of the mound. These sitesof new building, designated ‘domes’, are oftenassociated with deep tunnels that extend downwardand centripetally to the colony, and with expansivegalleries located just below the surface (see below).Less frequently, one finds ‘moundlets’ of newbuilding that occur at some distance from theconical base (Figure 8).

The mounds are frequently associated with trees orother vegetation, even to the extent that trees arefound embedded within them. Of a sample of 27mounds, nine were adjacent to a tree or surroundedby woody vegetation that could afford the moundsome significant shading, and a further eleven had

new buildingon spire

domes

Figure 7. A mound of Macrotermesmichaelseni showing a typical distribution ofnew building activity at the top of the spire,and two ‘domes’ emerging at the conicalbase. This particular mound was M43, one ofthe vertically sectioned mounds. Some of thereference stakes are visible in the photograph.

Table 1. Distribution of heights and estimatedvolumes for 303 censused mounds ofMacrotermes michaelseni in the Namatubisstudy site.

Height (m) Estimated ,volume (m3)

Mean 2.16 3.41

Standard deviation 0.53 2.72

Maximum 3.75 19.08

3rd Quartile 2.51 4.36

Median 2.13 2.65

1st Quartile 1.82 1.70

Minimum 0.68 0.13


trees embedded within them. Despite the closeassociation of mounds with trees, however, thecolonies do not appear to use the trees forstructural support: seven mounds in the samplestood on their own, with no association at all withwoody vegetation. Furthermore, the trees did notappear to serve as significant sources of food.Although bark was frequently stripped from theportions of the trunk embedded within the mound,the embedded trees appeared to be in robust health,suggesting the cambium was left unharmed (videFigure 6). Indeed, the trees seem to derive somebenefit from the association - in the dry season,trees embedded within mounds frequentlysupported the only green vegetation on site, withmost of the other trees having shed their leaves (S.Turner pers. obs.). The association of trees andmounds probably stems from greater survivorshipof natal colonies in the shade provided by thesetrees or bushes.

Population density and mound sizesThe study camp supports a total of 303 moundsover an extent of 289 ha, yielding a populationdensity of 1.045 mounds ha-1. The modal distance to

mounds’ nearest neighbours is roughly 70 m,spanning a range of zero (two mounds immediatelyadjacent to one another) to 148 m. The average

number of mounds in a 0.1’×0.1’ quadrat (roughly3.43 ha) was 3.49, with a variance between quadratsof 3.35. The distribution of mounds per quadrat(Figure 9) was not significantly different from a

Poisson distribution (χ2 = 5.23, P = 0.823), and thesimilarity of mean and variance of mounds perquadrat indicates a random distribution of mounds,with no evidence of clumping or repulsion of onemound with respect to another. On average,mounds in this sample were 2.16 m high, spanning arange of 0.7 m to 3.75 m (Table 1). Volumes wereestimated from the measurements of basal diameterand height, assuming the mound approximated acone. Estimated mound volumes averaged 3.41 m3,spanning a range from 0.13 to 19.08 m3 (Table 1).

Mound orientationMeasurements estimating azimuth and zenith byMethod A were carried out on a subsample of 34mounds on site. Among these mounds, the conicalbase averaged 7.5 m in circumference, whichcorresponds to a basal diameter of 2.4 m, assuming

a. b.

mound

moundlet

moundlet

Figure 8. Two examples of accessory moundlets. a. A mound with a very narrow conical baseand a moundlet emerging in the outwash pediment. b. Close-up of a moundlet adjacent to themound.


a circular base (Table 2). At half-height, thecircumference is 42% that of the base. There is amarked asymmetry between the lengths of the northand south faces of the mound, lN and lS. The southface was approximately 11.5% longer than the southface (Table 2), reflecting the mounds’ northward tilt.A paired t-test shows the difference to besignificant (one-tailed p = 0.0417). The mound’saverage zenith is estimated by this method to be19.6oN (Table 3). Method B, which estimates zenithmore directly, gives an average value of 18.2oN(Table 3). By either method, a Student’s t test showsthe mound’s zenith to not significantly differentfrom the average zenith angle of the sun through

the year, which is 19.98o at this latitude (For methodA, p = 0.926, and for method B, p = 0.161) . Themounds’ average azimuth is biased about 16o to thewest (Table 3): this bias may arise from the inclusionin the sample of one young mound which wasoriented to the south-west (Table 3). The medianazimuth is 3.7° W. Both zenith and azimuth exhibitconsiderable variation (Table 3).

Horizontal sectionsThe mound chosen for horizontal section wasroughly 170 cm tall, and was sectioned down toabout 20 cm above ground level, that is, just to thebottom of the conical base. The horizontal crosssection ranges from elliptical in the spire to circularat the base (Figure 10). In the spire (Figure 10,sections +140 to +90), the tunnels are few innumber, and branch and merge repeatedly to form aramifying network of vertically-oriented tunnels. Atthe conical base (Figure 10, sections +80 to +20), thetunnels of the spire begin to branch laterally. In thecentripetal direction, these branches merge into alarge elongate open space in the center of themound, designated the chimney (‘ch’ in Figure 10).Centrifugally, the branches ramify into a series ofradial channels (‘rc’ in Figure 10), which eventuallyproject downward and centripetally. Laterexcavation showed the colony’s fungus gardens tobe located about 40 cm below the lowest horizontalsection.

Mounds per quadrat

Qua

dra

ts

25

20

15

10

5

0

Quadrats (Observed)

Quadrats (Expected)

0 1 2 3 4 5 6 7 8 9

Figure 9. Distribution of mound density in theNamatubis study camp. Solid bars representthe observed numbers of quadrats of theindicated density. Open bars indicate theexpected distributions of mound densities.

Table 2. Linear measurements of 34 mounds of Macrotermes michaelseni at the Namatubissite. Basal circumference is measured at the boundary between the outwash pediment and theconical base of the mound. South and north face lengths are measured from the base of theconical mound to the apex at the respective cardinal points. Half-height circumference ismeasured at the point halfway between the apex and base of the conical portion of the mound.

Basalcircumference(m)

Half heightcircumference (m)

South facelength (m)

North facelength (m)

Mean 7.54 3.19 3.19 2.86

Standard Deviation 1.75 1.25 1.16 0.80

Minimum 4.90 1.30 1.25 1.60

1st quartile 5.91 2.29 2.16 2.20

Median 7.70 3.15 3.08 2.80

3rd quartile 8.38 3.78 4.10 3.40

Maximum 11.80 6.85 5.60 4.90


The sectioned portion of this mound occupiedabout 0.9 m3, of which 18% on average wasoccupied by tunnel (Table 4). This figure can onlybe approximate, given the incomplete coverage ofthe photographs. The ratio of the mound’s surfacearea (AM) to its volume (VM) is roughly 4.3 m-1, whilethe ratio of tunnel surface area (AT) to tunnel volume(VT) was roughly 10 times higher, 44.1 m-1. Thisexpanded tunnel area presumably indicates thehighly folded structure of the mound’s internal airspaces.

The sizes of the tunnels and mound show a two-phase distribution with height, correspondingroughly to the spire and conical base (Figure 11). Inthe spire, about 25% of the volume of the mound isoccupied by tunnel, while in the base, only about16% of the mound volume is so occupied. This isreflected in the distribution of areas and lineardimensions in the various horizontal sections. Inboth the spire and conical base, the section areas ofthe mound, Am, and tunnels, At, converge slightlywith respect to height: in other words, tunnels cometo occupy a larger proportional volume of themound in the higher sections of the mound (Figure11a). Interestingly, the circumferences of moundsections, lm, and tunnel sections, lt, convergesteeply in the spire, but not in the conical base

(Figure 11b). This may indicate the morecomplicated shapes of the tunnels in the portion ofthe mound experiencing the most recent and intensebuilding.

Vertical sectionsThe two vertically-sectioned mounds differedslightly in size, with M43 being about 4% taller thanM03, although M43 was about 50% morevoluminous (Table 5). As in the horizontal sections,the volume of mound occupied by the tunnels wassmall, roughly 28% for M03 and 16% for M43. Theratio of surface area to volume, AM/VM, was similarlysmall for the mound (2.59 m-1 for M03, 2.30 m-1 forM43), and remarkably large for the tunnels, AT/VT

(36.6 m-1 for M03, 42.5 m-1 for M43).

As in the horizontal sections, the vertical sectionsrevealed a complicated reticulated network oftunnels within the mound, although the verticalsections reveal them with more structural detail.Outlines of the tunnels, exposed colony and moundare given in Figures 12-14 for M03 and Figures 15-17for M43.

Perhaps the most striking feature of both mounds isthe large vertical chimney (‘ch’ in Figures 12-17),which corresponds to the elongate tunnel thatappeared in the lower levels of the horizontally-

Table 3. Distributions of zenith and azimuth angles of mounds of Macrotermes michaelsenimounds at the Namatubis site.

Zenith angle (degrees),(Method A)

Zenith angle (degrees),(Method B)

Azimuth angle(degrees)

Mean 19.6 N 18.2 N 16.8 W

N 32 37 32

Standard Deviation 14.7 7.3 36.2

Minimum 5.9 S 4.6 S 153.7 W

1st quartile 10.0 N 13.6 N 33.7 W

Median 14.1 N 18.8 N 3.7 W

3rd quartile 23.7 N 23.0 N 6.3 E

Maximum 64.2 N 33.1 N 16.3 E

Expected 19.98 N(latitude)

19.98 N(latitude)

0.0

Z 0.450 0.926 0.996


+20 cm

+30 cm

+40 cm

+50 cm

+60 cm

+70 cm

+80 cm

+90 cm

+100 cm

+110 cm

+120 cm

+130 cm

+140 cm

ch

rc

rc

rc

rc

rc

rc

rc rc

ch

ch

scsc

ch

sc

ch

sc

ch

Figure 10. Tracings of a horizontally sectioned mound of Macrotermes michaelseni. The solidoutlines indicate mound surface and tunnel outlines that were visible in the photomontage. Dottedoutlines indicate hypothetical mound surface that could not be seen in the photomontage. Thescale bar represents 30 cm. Labels refer to height of the sectioned face above ground level. Label‘rc’ refers to a radial tunnel, ‘sc’ refers to a surface conduit, and ‘ch’ refers to the chimney.


sectioned mound (‘ch’ in Figure 10). In the verticalsections, the chimney opens just above the colony,and extends upward into the apex of the mound(Figure 18). Extensive lateral tunnels connect the airspace of the chimney to air spaces throughout themound. In some instances, clear continuous

pathways can be seen between the chimney and thesurface tunnels (Figure 12, 130 cm; Figure 13, 150cm, 160 cm; Figure 16, 220 cm). Near the top of thecolony, the fungus gardens seem to be serviced bynumerous tunnels that appear to wrap around themargins of the colony, merging eventually to formthe chimney (Figure 12, 120 cm to 130 cm; Figure 13,140 cm, 180 cm, 190 cm; Figure 14, 210 cm to 250 cm;Figure 17, 230 cm to 250 cm).

Another common feature of these two mounds is anopen chamber that forms at the top of the chimney(Figure 12, 120 cm to 130 cm; Figure 16, 170 cm to190 cm). This chamber also connects extensivelywith the surface conduits that run under the outersurface of the mound (Figure 12, 130 cm; Figure 13,140 cm to 150 cm; Figure 16, 130 cm to 140 cm, 180cm to 210 cm). The cap of the mound and itsassociated chamber varies in degree of opennessand shape from mound to mound (Figure 19). Inmany mounds, the cap is rounded, and is coveredby a thin friable layer of soil (Figure 19a). In others,

1.0 1.6 2.2

1.5

2.5

3.5

log height (cm)lo

g s

ect

ion c

ircu

mfe

rence

(cm

)

tunneltunnel

moundmound

m = -0.62

m = -0.68

m = -1.10

1.0 1.6 2.2

2.0

3.0

4.0

log height (cm)

log s

ect

ion a

rea (

cm)

2

m = -1.96m = -1.07

m = -1.16

m -2.13

m = -2.22

a. b.

Figure 11. Relationship between height and section area and circumference of tunnels andmounds in the horizontally sectioned mound. Section area refers to the plan area of mound andtunnel sections. a. Mound section area, Am, includes the area enclosed by the outer boundary ofthe mound. Tunnel section area, At, is the summed plan areas of the tunnel outlines in that section.b. Mound circumference, lm, refers to the outline length of the sectioned face. Tunnelcircumference, lt, refers to the summed outline lengths of the tunnels in the section. Slopes of thevarious component data are adjacent to each line.

Table 4. Estimates of volumes and surfaceareas of mound and tunnels in the horizontallysectioned Macrotermes michaelseni mound.

Property Quantity

Mound height (m) 1.5

Mound surface area (m2) 4.066

Mound volume (m3) 0.939

Tunnel surface area (m2) 7.504

Tunnel volume (m3) 0.170


the cap is strongly peaked, and the wall separatingthe upper chamber from the air is thicker, althoughseemingly still very porous (Figure 19b). Amongother mounds, the chamber space is moredisorganized and difficult to trace, with numerousfinely divided subchambers and tunnels (Figure19c).

conduit forms on the lower part of the mound’swest face (Figure 16, 180 cm), breaking up intonumerous small tunnels at 190 cm, and coalescingagain at 200 cm through 220 cm.

Both sectioned mounds had extensive networks oflarge calibre surface tunnels, designated surfaceconduits, which underlie the entire surface of themound. For example, an extensive surface conduitunderlies the upper two thirds of the north face ofM03 (Figure 12, 120 cm), and it can be followed insubsequent sections as it extends downward and tothe west (Figure 12, 130 cm; Figure 13, 140 cm to 150cm). Other surface tunnels appear on this mound’ssouth face, and again, one can trace the intertwiningtunnels through several sections of the mound(Figure 13: 150 cm to 180 cm). Similar features areseen in M43. In this mound, an extensive surface

The surface conduits are separated from the outsideair by a thin friable wall, similar to the thin wallseparating the upper chamber from the air at the capof the mound (Figure 20). Often, the interior surfaceof the conduits are pitted with small holes, roughly2 mm in diameter. These usually underlie areas ofsurface building, and these probably serve asegress holes for worker termites to the surface.

A particularly interesting feature of moundconstruction are surface ‘boils’, which are narrowlydelimited areas of apparently intense buildingactivity. The sections through mound M43sectioned through two surface ‘boils’ (Figures 22-23), one more extensively built than the other. Theintense building activity that accompanies these‘boils’ is evidenced by the numerous smallchambers underlying the ‘boil’ (Figure 22). There isobviously transport of soil to the mound surfaceduring the construction of a ‘boil’, because thetexture and colour of the soil sometimes differs fromthe mound base it is built upon (Figure 21). A moredeveloped ‘boil’ will show extensive workinggalleries, and the highly convoluted internalsurfaces that are characteristic of intense buildingactivity (Figure 22). A particularly interestingfeature of these surface ‘boils’ are the centripetalradial channels, which appear to penetrate directlyto the colony. Follow the tracings for the tunnelsunderlying the ‘boil’ from 150 cm to 180 cm (Figure16) - the working gallery pictured in Figure 22 is partof this series. The development of the centripetaltunnel is clearly seen in these sections. Thedirectness of the connection is also seen clearly inFigure 23: the tunnel underlying the ‘boil’ runsstraight to the colony.

Surface ‘boils’ appear to be the precursor to thedome-shaped extrusions of soil from the surface,like those pictured in Figure 7. The vertical sectionsreveal four series through ‘domes’ (‘d’ in Figures12-17), two on M03, from 60 cm through 90 cm(Figure 12) and again from 190 cm (Figure 13)through 260 cm, and two on M43, from 120 cmthrough 140 cm (Figure 15) and again from 220 cm(Figure 16) through 320 cm (Figure 17). The lastmerges into a surface ‘boil’ at 330 cm to 340 cm(Figure 17). The ‘domes’ are clearly areas of intensebuilding activity, particularly toward the surface.Note particularly the gradation from large diametertunnels in the center of the ‘dome’ to numeroussmall tunnels toward the surface in the series 230 cmthrough 270 in M43 (Figure 17).

colony

fungus gardens

chimney

Figure 18. Photograph through one of thecenter sections of M03, showing relationship ofchimney, fungus gardens and colony. Thisphotograph corresponds to the outline at +160cm in Figure 13.

Cimbebasia 16,2000

+10 cm

+20 cm

+30 cm

+40 cm

+50 cm

+60 cm

+70 cm

+80 cm

+90 cm

+100 cm

+110 cm

+120 cm

+130 cm

d

d

d

d

rc

rc

sc

sc

sc

sc

sc

ch

ch

sc

Figure 12. Outlines of the vertically sectioned mound M03. Numbers adjacent to the outlines refer to thedepth of the section face with respect to the eastern edge of the outwash pediment. Areas filled with darkgray represent the colony, visible in these sections mostly as fungus gardens. Areas filled with light grayrepresent mound section face. Areas filled with intermediate gray are outlines of tunnels exposed at the face.Labels as in Figure 10. The label ‘d’ refers to a ‘dome’.

Turner - Architecture and Morphogenesis in Macrotermes mounds

+140 cm

+150 cm

+160 cm

+170 cm

+180 cm

+190 cm

sc

sc ch

ch

sc

sc

ch

ch

scsc

sc

sc

d

rc

sc

sc

sc

sc

sc

sc

Figure 13. Continuation of series of vertical sections from vertically sectioned mound M03.Conventions are as in Figure 12.

Cimbebasia 16,2000

+200 cm

+210 cm

+220 cm

+230 cm

+240 cm

+250 cm

+260 cm

+270 cm

+280 cm

+290 cm

+300 cm

+310 cm

+320 cm

d

d

d

d

d

d

d

d

rc

b

sc

sc

scsc

sc

sc

sc

sc

rc

b

b

sc



+10 cm

+20 cm

+30 cm

+40 cm

+50 cm

+60 cm

+70 cm

+80 cm

+90 cm

+100 cm

+110 cm

+120 cm

+130 cm

+140 cm

d

s c

Figure 15. Outlines of the vertically sectioned mound M43 of Macrotermes michaelseni.Conventions are as in Figure 12.

Cimbebasia 16,2000

+150 cm

+160 cm

+170 cm

+180 cm

+190 cm

+200 cm

+210 cm

+220 cm

sc

d

b

rc

sc

rc

b

rc

rc

bsc

ch

ch

sc

sc

sc

rc

scch

sc



+230 cm

+240 cm

+250 cm

+260 cm

+270 cm

+280 cm

+290 cm

+300 cm

+310 cm

+320 cm

+330 cm

+340 cm

+350 cm

d

ch

sc

d

rcrcsc

sc

sc

d

d

d

d

sc

rc

d

d

rc

d

b

b


Turner - Architecture and Morphogenesis in Macrotermes mounds 157

Despite the similarity of heights (within 3%), thetwo mounds differ substantially in volume andsurface area (Table 5). Nevertheless, the ratios ofmound surface area and volume (AM/VM) are verysimilar, between 2.3 and 2.6 m-1 (Table 5). This issubstantially smaller than the same ratio in thehorizontally-sectioned mound (Table 4): thedifference is probably due to the horizontallysectioned mound not being sectioned down toground level, leaving out the substantial volume ofoutwash pediment that is included in themeasurements for both the vertically sectionedmounds.

The size of tunnels in the mounds varies markedlyin sections through the mound. In sections throughthe outwash pediment, tunnels occupied only asmall proportion of both mounds (Figure 24), whilein sections through the base and center of themound show a much greater proportion of thesection occupied by tunnels, reaching as high as40% of the section volume, Vm. On average, thetunnels occupied 28% of the volume of M03, andabout 16% of the volume of M43 (Figure 24).Comparisons of the tunnel surface area with respectto the mound surface area show a similar pattern(Figure 25). On average, the tunnel surface area, AT,is roughly 3.5 to 2.8 times greater than the outsidesurface area of the mounds, AM (Table 5, Figure 25).Again, the absence of tunnels in the outwashpediment and their proliferation in the center areevident from the distribution of areas across thesections (Figure 25). In the sections through the

outwash pediment, the proportion of the sectionface occupied by tunnel is visibly small. As thesections proceed through the center parts of thebase and spire, the ratio of tunnel area to moundface area increases dramatically. The mound alsoexpands the surface area of ground surfaceinteracting with air by, on average 40% (Figure 26;Table 5). Clearly, the major expansion of groundsurface occurs in the central parts of the mound. Inthe central portion of the mound, the expansion ofground surface is substantially above average, witha maximum ground surface expansion of about 2.3(Figure 26).

Surface sectionsThe surface conduits form an extensive networkthat underlies the entire surface of the mound whichis illustrated by ‘peeling’ away layers of moundsurface. Eight series of surface sections were madeusing the methods described above. Arepresentative sample of three of these series arereproduced in Figures 27-29. A common feature ofthese sections is the strong vertical orientation ofthe surface conduits. The conduits are also quiteshallowly placed, with nearly all the sectionsbreaking into conduit at depths of only threecentimetres, sometimes shallower. Surface conduitsfrequently are directly under regions of surfacebuilding, for example, the south-face series in Figure27, and the west face series in Figure 28. As often,however, surface conduits are adjacent to regionsof surface building, for example in the west-faceseries in Figure 27, or the east face series south-face

a. b. c.

Figure 19. Three different apices of the mound of Macrotermes michaelseni. a. A thin friablemound cap. b. A thick and porous mound cap showing extensive new building. c. A complexmound cap showing evidence of reworking of the interior of the upper chamber of the chimney.


series in Figure 28. In one section (south face ofFigure 28), in which most of the surface wascovered by new building, the surface is penetratedby numerous tiny holes, presumably egress holesand galleries of the newly built surface.

The extent of new building on the surface sectionsaveraged about 21%, ranging from 100% to nil(Table 6). On average, the extent of the underlyingtunnels doubled for each centimetre of depth,starting at 4% of the sample area at 1 cm depth andincreasing to about 20% of the sample area at 3 cmdepth. The range of each measure was large,however, ranging from a maximum of about 45% ofthe underlying area to nil, even in sections as deepas three centimetres (Table 6). The expansion of thetunnels with depth was significant (F2,95 = 7.73, p =0.0008). However, there was no discernible

difference between the samples at the cardinalpoints, either in the extent of new building on thesurface (F3,28 = 1.91, p = 0.151), or in the extent of thetunnels beneath (F3,95 = 0.500, p = 0.683). There waslikewise no significant correlation between theextent of the tunnels and the extent of buildingactivity on the overlying surface (Figure 30).

Effect of rotating the spireRotating the spire affected the distribution ofinsolation, temperature and building activity on themound surface (Figure 31). Overall, mound surfacesfacing the north were both warmer (F3,143 = 20.26, F*

= 2.67, p<0.0001) and more intensely insolated (F3,143

= 86.05, F* = 2.67, p<0.0001) than were the otherfaces (Figure 31). The western faces tended to beboth warmer and more insolated than the easternfaces were, primarily because the majority of themeasurements were made before solar noon, whichbiased the distribution of heat within the mound tothe west. Also, the western faces of both CV andT05 were relatively free of vegetation that couldshade the mound. Rotating the spire 180o alsosignificantly affected both surface temperature(F2,143 = 21.60, F* = 3.06, p < 0.0001) and insolation(F2,143 = 39.11, F* = 3.06, p<0.0001), with the rotatedspires being warmer and more intensely insolatedthan the unrotated spires (Figure 31). There is also asignificant interaction effect for both temperature(F6,143 = 3.08, F* = 2.17, p = 0.007) and insolation(F6,143 = 16.96, F* = 2.17, p < 0.0001), with the northface warmed and insolated more in the rotated spirethan in the unrotated or shaded spire.

Figure 20. The interior surface of a surfaceconduit, showing egress holes indicated bywhite arrows. One egress hole opens directlyto the outside, as indicated by the light shiningthrough. The black dotted line indicates theouter boundary of the mound. Some of theouter wall of the surface conduit had flakedaway during preparation of the mound face.

mound

boil

Figure 21. A newly formed ‘boil’ indicatingthe different texture of the soil in the moundcompared to the ‘boil’. Note also the smallworking tunnels permeating the ‘boil’.


Subsequent observation of the mounds with rotatedspires indicated a greater extent of building activityon the north face of the spire than on other faces.Specifically, on the north face, marker pegs that hadbeen used to indicate the cardinal points on thespire were nearly buried under new building. On thesouth faces, however, soil around the marker pegshad been eroded away, in one case sufficiently tocause the peg to fall out.

DISCUSSION

Much of the classical work on termite mounds hastreated these structures as analogous to phenotype,with particular species-specific architectures, orwith structural features ‘designed’ to serve aparticular function, such as colony homeostasis(Darlington 1984, 1985, 1986; Lüscher 1956; 1961;Weir 1973). Whether purposefully or not, thisanalogy draws its inspiration from the conception of

the insect colony as a ‘superorganism’, with traitsand behaviours reminiscent of those observed inconventional organisms, albeit at a higher level oforganization (Wilson 1971).

Conventional phenotype can be operationallydefined as a structural feature of an organism that ischaracteristic of the species in which it is found. Inturn, this implies that between-species variation inthe structure is demonstrably larger than the within-species variation. One striking feature ofmacrotermitine mounds is their extraordinarystructural variation, and how to partition thisvariation between and among species has long beenuncertain. At one time, between-species variationwas thought to be the most significant, with certainspecies thought to construct mounds withdistinctive architectures. For example, Macrotermessubhyalinus is often described as building relativelylow dome-shaped mounds, with large openings thatcan serve as wind scoops or Venturi vents (Weir1973). M. natalensis, for its part, is usually describedas building cone-shaped mounds that are

boil>moundlet

fungusgarden

radialchannel

>surface conduit

Figure 23. Another example of a ‘boil’ and aradial channel (indicated by arrows) extendingdown to the colony.

boil > moundlet

radial channel

workinggallery

new building

Figure 22. A more extensively worked ‘boil’,showing the large working gallery, and themarked gradation in tunnel size from gallery tomound surface. The radial channel, presumablythe original egress hole for alates, runs directlyto the colony.


completely enclosed (Lüscher 1961; Ruelle 1964).Odontotermes transvaalensis locates its colony deepunderground, and builds vertical open chimneysroughly a metre tall (Ruelle 1985; Turner 1994).Finally, the mound of M. mossambicus (now M.michaelseni) is described as having a cone-shapedbase topped with a thin vertical spire, or steeple(Coaton & Sheasby 1972; Ruelle et al. 1975).

To reliably partition this variation among speciesrequires both a sound understanding of thetaxonomic relations among the termites, as well as agood understanding of the variation within thosespecies. Unfortunately, the taxonomy of termites,particularly of the macrotermitines, has long beenconfused, and many of the classical works ontermite mound architecture (e.g. Grassé & Noirot1961) were done prior to recent revisions ofmacrotermitine taxonomy (e.g. Ruelle 1970, 1975). Asthe confused taxonomy of termites has becomeclearer, and as more descriptions of termite moundarchitecture have accumulated, it has become

obvious that the within-species variation ofarchitecture among the macrotermitines issubstantial, perhaps greater than the purportedbetween-species variations that drew the attentionof earlier investigators. For example, Ruelle (1964)describes M. natalensis mounds in Uganda ascharacteristically open at the top, while the samespecies in the Côte d’Ivoire builds mounds that arecommonly closed. Grassé & Noirot (1961) describe avast variation of architecture among mounds ofBellicositermes natalensis (= Macrotermes natalensis)ranging from low conical mounds to moundsdecorated with numerous accessory moundlets toenormous ‘dome’ mounds that span diameters aslarge as 30 m. Similarly, mounds of M. subhyalinus inKenya exhibit two common morphs (Darlington1984), with one (the ‘Bissel type’ mound) being lowand hemispherical in shape, and penetrated bynumerous vent holes, and another (the ‘Marigattype’) being topped by a very tall and thin chimney.Similar degrees of within-species variation have alsobeen described for mounds of the more distantly

Sec

tion

area

(m)

2

Sec

tion

area

(m)

2

Depth of section (cm)



Depth of section (cm)10 1050 50100 100150 150200 200250 250300 300 350

1

0

2

3

4

10 50 100 150 200 250 300

mound

tunnel

1

0

2

3

4

10 50 100 150 200 250 300 350

mound

tunnel

tunnel/mound

average

tunnel/mound

average

Tunn

el s

ectio

n ar

ea/M

ound

sec

tion

area

Tunn

el s

ectio

n ar

ea/M

ound

sec

tion

area

0.4 0.4

0.3 0.3

0.2 0.2

0.1 0.1

0 0

a.

b.

c.

d.

Figure 24. Distribution of section area for mound (Am) and tunnels (At) and the ratio of tunnel section areato mound section area through the series of vertical sections. a. and b. Mound M03. c. and d. Mound M43.The horizontal lines in b. and d. indicate the average value over the series of sections.


related Brazilian nasute termite Cornitermes bequaerti(Cancello 1991). Less distinct within-speciesvariation in mound architecture has also beendescribed, correlated with climatic gradients. Forexample, Pomeroy (1977) observed that ‘turrets’ onthe mounds of M. bellicosus are common in hotterparts of its range, while mounds in cooler partsfrequently lack these structures. Harris (1956) hasnoted that M. natalensis mounds tend to have lowerprofiles at higher elevations than they do at lowerelevations. Coaton & Sheasby (1972) have notedsimilar variations in the mounds of M. natalensis,with mounds in the more open and windy areas ofthe Nama Karoo being low in profile, while moundsin more heavily vegetated arid savanna tend to betaller. Overall, the picture that seems to be emergingis the mound as a dynamic structure which has arelatively minor genetic component, overlain withsubstantial, perhaps dominant, epigenetic factorsthat correlate to broad patterns of climate, soil type,vegetation and so forth.

Any attempt to explain this variation of moundarchitecture must be reconciled with the fairlylimited behavioural repertoire that termites bring to

mound construction (Emerson 1956; Stuart 1972;von Frisch & von Frisch 1974). Macrotermitinemounds also have function, however, and to somedegree, this function is dependent upon themound’s structure. Thus, any explanation for themorphogenesis of termite mounds must alsoaccount for how the structure is attuned to itsfunction. In the discussion to follow, I would like topropose such a model. Most of my attention will bedevoted to morphogenetic models which explain themajor architectural features of the M. michaelsenimounds here described. The integration of structureand function is based

upon the results of physiological experiments yet tobe published, so I shall only touch upon its mainfeatures here. Finally, I shall offer a speculativemodel for the between-species variation of moundarchitectures amongst the macrotermitine termites.

Mound construction results from termites engagingin a simple activity: translocating grains of sand orsoil from one spot to another. A worker termite thatis disturbed will pick up a grain of soil, either from aspoil pile or from a tunnel wall, carry it for a time,

2.0

1.0

0

10

1010

10 50

5050

50 100

100100

100 150

150150

150 200

200200

200 250

250250

250 300

300300

300 350

350

Sur

face

are

a (m

)2

Sur

face

are

a (m

)2


Depth of section (cm)Depth of section (cm)


tunnelsmound

tunnelsmound

0 0

2 2

4 4

6 6

8 8

Tun

nel s

urfa

ce a

rea/

Mo

und

su

rfac

e a

rea

Tun

nel s

urfa

ce a

rea/

Mo

und

su

rfac

e a

rea

tunnel area/surface area

average

tunnel area/surface area

average

0

1

2

3

4

5a.

b.

c.

d.

Figure 25. Distributions of surface area of mound (10* lm) and tunnel (10* lt) and the ratio of tunnel surfacearea to mound surface area through the series of vertical sections. Conventions are as in Figure 24.


and then deposit it onto a surface, daubing it intoplace with adhesive secretions (Stuart 1972; vonFrisch and von Frisch 1974). When takencollectively in a colony of 1-2 million inhabitants,this results in termites being prodigious movers ofsoil, particularly in arid tropical ecosystems(Dangerfield et al. 1998; Lobry de Bruyn &Conacher 1990). For example, the Ugandan termiteMacrotermes bellicosus transports soil upward intothe mound at rates varying from 0.4 to 0.9 m3 ha -1 y -1.This upward movement of soil is offset by similarlevels of downward movement through erosion, butthe building activities are sufficient to support a‘standing crop’ of soil in termite mounds of 4-8 m3

ha -1 (Pomeroy 1976).

Termite mounds are, therefore, dynamic structures,and their architecture is a result of how thesemovements of soil are directed (Harris 1956). Ingeneral, the shape of such a dynamic structurefollows from two principal movements of soil. Onthe one hand, any activity in which workundertaken to deposit soil particles at places theyotherwise would not be will result in building, whichwe might describe using the general term of blastis.Blastic processes in mounds are nearly always

biological in nature, because the translocation ofsoil against gravity or weathering processesrequires active work being done on the soil by theworkers. On the other hand are processes whichremove soil particles, perhaps to return them topositions closer to equilibrium, which we will termclastis. Clastic processes include abiotic processes,such as erosion, but may also be biological. Forexample, termites often open holes in the surfaces oftermite mounds, either as a normal part of extensionof the mound surface (Figure 20), or to open egressholes for the emergence of alates: both involveclastis. Also, the simple act of translocating soilfrom one part of the mound to another necessarilyinvolves a clastic process, when a grain of soil isdislodged from the mound, coupled to a blasticprocess, when it is redeposited.

10

1010

50

5050

100

100100

150

150150

200

200200

250

250250

300

300300 350

0.2 0.2

0.4 0.4

0.6 0.6

0.8 0.8


Depth of section (cm)Depth of section (cm)

Su

rfac

e a

rea (

m)

2

50 100 150 200 250 300 350Depth of section (cm)

Su

rfac

e a

rea (

m)

2

m o u n d m o u n db a s e b a s e

1.5 1.5

2.0 2.0

2.5 2.5

1.0 1.0

Mou

nd s

urface a

rea

/base

surf

ace a

rea

Mou

nd s

urface a

rea

/base

surf

ace a

rea

s u r f a c e / b a s a l s u r f a c e / b a s a la v e r a g e a v e r a g e

a.

b.

c.

d.

Figure 26. Distributions of surface area of the mound and base through the series of vertical sections.Conventions are as in Figure 24. Base area refers to the product of section thickness and the horizontaldistance from the left to the right of the mound base.


Figure 27. A representative series of surfacesections. Columns refer to series at thecardinal points indicated. Rows are the depthsof the section in centimeters. Light grayindicates the plot marked on the mound forsectioning, roughly 40 cm by 40 cm. Whiteareas at 0 cm depth refer to areas of newbuilding. Dark gray in the other sectionsrepresent outlines of tunnels.

Figure 28. A representative series of surfacesections. Conventions are as in Figure 27.

Figure 29. A representative series of surfacesections. Conventions are as in Figure 27.

Understanding the architecture of termitemounds thus requires an understanding of howthese blastic and clastic processes interact. As asimple example, consider two of the commonoutward features of the M. michaelseni mound:the broad outwash pediment and northward-tilting spire (Figure 6). The origins of theoutwash pediment and spire are easily explainedby a straightforward interaction between ablastic process that translocates soil directlyupward, and a clastic process, like erosion, thattransports soil downward again (Figure 32). Inthe absence of clastis, the upwardly-directedvector of blastis should result in the constructionof a cylindrical column (Figure 32a). If a clasticprocess, like erosion, is now added, thereemerges a mound with the central columntapered at an angle steeper than the soil’s angleof repose, and a broad outwash pediment, withthe slope of the pediment being similar to orshallower than the soil’s angle of repose (Figure32b). The northward tilt of the spire results fromadding a third vector of blastis (Figure 32c).Imagine that the rate at which a termite builds isdependent upon the termite’s body temperature.Summed across all the workers, the rate atwhich blastic work is done should varythroughout the mound in parallel with thedistribution of temperature within the mound. Ifthe mound is warmed by the sun more on itsnorth side than on its south side (Figure 31), thespire’s north side will grow at a faster rate thanwill its south, east or west sides. This additionalvector of blastis will be oriented in the direction

East North South West

0

-1

-2

-3

E a s t N o r t h S o u t h W e s t

0

-1

-2

-3

East North South West

0

-1

-2

-3


of greatest warming by insolation, namely to thesun’s average zenith.

Thus, the mound will grow in the direction thatequalizes the distributions of temperature aroundthe spire, which will occur when the spire pointstoward the sun’s average zenith.

The correspondence of the mounds’ tilt with thesun’s average zenith angle at Namatubis istantalizing, albeit not conclusive, evidence in favorof this model. For example, one would predict fromthis a systematic variation in the mounds’northward tilts with latitude, and I am not aware ofany evidence that supports or refutes thisprediction. However, the patterns of building on themounds whose spires were rotated does suggestthat the rotated spire was slowly growingnorthward. Furthermore, variations of temperatureare known to affect mound architecture ofMacrotermes bellicosus in ways consistent with thismodel (Korb & Linsenmair 1998).

Another striking architectural feature of the M.michaelseni mound is the size differentiation of thetunnels, with more interior tunnels being larger andmore capacious than tunnels closer to the surface.This is evidenced in a number of ways. In the twosets of vertical sections, for example (Figures 12-17),the tunnels in the sections through the outwashpediments and the peripheral portions of the coneappear smaller and fewer in number than the tunnelsin the more interior sections of the mound,culminating in the opening of the chimney in thecenter of the mound (‘ch’ in Figures 12-17). Theseapparent patterns are substantiated by the variousmeasures of tunnel volume and surface areathrough the sections (Figures 27-29). Areas in themound which exhibit intense building activity, suchas at the top of the spire (Figures 19b-c), or at thesite of a new ‘boil’ (Figure 22), also show a

differentiation between small tunnels at the surfacegrading into larger open spaces toward the interior.Finally, in the surface sections, volume occupied bytunnels roughly doubles for every centimetre ofdepth (Figures 27-30).

This distribution of tunnel sizes can be explained bya type of conservation mechanism, in which thetranslocation of a grain of soil from one place toanother involves the simultaneous appearance of a‘heap’ - a grain of soil glued to something, either atunnel wall, or a previously deposited grain of soil -and a ‘hole’, a space previously occupied by thatgrain of soil (Figure 33). In the absence of anybiasing of the movements of the termites, soiltranslocation will be random, as will the spatialdistribution of heaps and holes. If the movement oftermites carrying soil is biased in a particulardirection, however, the distribution of heaps andholes no longer will be random, and will reflect thebias in the locomotory activities of the termites. Oneconsequence of such a bias will be distribution ofhole sizes similar to that seen in the mound.Suppose, for example, the termites tend totranslocate soil centrifugally, from the center of themound to the surface.

Table 6. Proportion of surface section areas occupied by new building or tunnels. Numbers arereported as mean (standard deviation) [ minimum - maximum].

Proportion of area occupied by new building

Surface 0.21 (0.23) [1.000 - 0.000]

Depth of section (cm) Proportion of area occupied by tunnel

-1 cm 0.04 (0.07) [0.275 - 0.000]

-2 cm 0.10 (0.10) [0.309 - 0.001]

-3 cm 0.20 (0.13) [0.444 - 0.000]


As soil is deposited at the surface, i.e. as new heapsare added there, the mound surface grows outward.However, the material to supply these new heapsmust result in the appearance of holes, and becausetermites only carry soil for a limited distance, thenew hole either appears at the site of a nearby heap,leveling it, or from a previously created hole,enlarging it (Figure 33). Thus, mound buildinginvolves a sort of ‘bucket-brigade’, with outwardextension of the mound surface being supported byincreasing amounts of excavation from the interiorparts of the mound. This results in a directcorrelation between the age of a tunnel and its size.When a tunnel is newly formed, it has not served asa source of soil for new heaps for very long, and it’ssize - reflective of the number of ‘holes’ openedthere - will therefore be small. Whereas oldertunnels will have served as a source for longerperiod, and it will therefore be large. Consequently,the capacious chimney probably results simply fromthis region of the mound having served for thelongest time as a source of material for extendingthe mound surface outwards.

Probably the most common biasing mechanisms inmound construction involve gradients ofconcentrations in respiratory gases (including watervapor) and temperature (Emerson 1956; Stuart 1972).There is much anecdotal evidence that these alterbuilding behaviour of termites and other socialinsects (Nicholas & Sillans 1989; Stuart 1972). For

example, macrotermitine termites will begin toexcavate openings in their nests when nest CO2

concentrations increase from 1% to 2% (Nicholas &Sillans 1989). Termites kept in glass cases will tendto concentrate their building near the case’s ventholes (Stuart 1972). Behaviours of ants and termitescan be altered by exposure to small alterations ofconcentrations of oxygen, carbon dioxide, humidityor gentle movements of air (Stuart 1972).

It follows that the structure of a mound could beshaped in part by gradients within the mound of theconcentrations of respiratory gases (i.e. pO2, pCO2,pH2O) or temperature (Korb & Linsenmair 1998).Consider, for example, the sequence of behavioursthat follow the opening of a breach in the moundwall of M. michaelseni (S. Turner pers. obs.). Whenthe mound wall is first breached, worker termites are

Sur

face

te

mpe

ratu

re (

C)

oS

urf

ace

inso

latio

n (

kLux

)

Mound face

Mound face

West

West

North

North

East

East

South

South

120

80

40

0

25

30

35

40 Control

Rotated spire (sunny)

Rotated spire (shady)

Control

Rotated spire (sunny)

Rotated spire (shady)

a.

b.

Figure 31. Surface temperature and insolation of themounds in the spire rotation experiment. Symbolsindicate mean values over the course of theexperiment. Error bars represent the least significantrange of Tukey. Non-overlap of the bars indicates astatistically significant difference at the 95%confidence level. a. Surface temperature. b. Surfaceinsolation.

0

0.1

0.2

0.3

0.4

0.5

0 0.2 0.4 0.6 0.8 1.0

Proportion of surface with building activity

Pro

po

rtio

n o

f su

rfa

ce

occ

up

ied

by

tun

ne

l

d e p t h = - 3 c m

d e p t h = - 2 c m

d e p t h = - 1 c m

Figure 30. Relationship of tunnel area to surfacebuilding activity. Results from different depths areindicated by stars (-1 cm), circles (-2 cm) andtriangles (-3 cm). Lines are the least squaresregression lines for the depths indicated.Proportions calculated are the ratio of tunnel area ornew building area to the outline area at each depth.


rare or absent. After about five minutes, workersbegin to appear near the breach and to depositgrains of sand on nearby tunnel walls. Initially, thedeposition is haphazard, but as the initial repairwork proceeds, there is continued recruitment ofworkers to the task of repair, and increased numbersof deposited grains of sand. Eventually, workertermites seeking to deposit grains of sand are likelyto encounter previously deposited grains, and thisbiases deposition of new grains toward thosepreviously deposited, a process known as‘stigmergy’. The result is the erection of the pillars,walls and roofed galleries that grow to eventuallyseal off the tunnels leading to the breach. Once

sealed, the workers recruited to the repair work driftaway to other tasks.

The curious thing about this behaviour, particularlyin response to a large breach in the mound wall, isthat it occurs mostly at some distance inward fromthe breach. As a consequence, the repair does notreplace the damaged wall: rather it seals the tunnelsleading into the breach. This is consistent with theworkers’ activities being concentrated at a particularthreshold of pO2, pCO2, or pH2O, or withtranslocation being directed down gradients in theconcentrations of these gases. When the moundwall is breached, colony air with high pCO2 and

Buildinggravitaxis

BuildinggravitaxisErosion

BuildinggravitaxisErosion

Buildingthermotaxis

a. b.

c.

Figure 32. Hypothetical scheme for the development of the outwash pediment and spire, and the northwardtilt of the spire. a. Growth with vertical blastis only. b. Tapered spire and fanned outwash pediment resultingfrom vertical blastis and downward clastis. c. Growth of a tilted spire resulting from vertical blastis,temperature-directed blastis and downward clastis. Details are discussed in text.


pH2O, and low pO2 is brought into contact withoutside air with low pCO2 and pH2O, and high pO2.Gradients in these concentrations will inevitablyfollow, but the mixing of these two air masses, andhence the location and steepness of the gradients,will be influenced by the interaction of the twoenergy sources that drive the mixing. From thecolony interior, metabolic energy in the form of heatand humidity will drive colony air out toward thebreach, while wind energy will drive outside air in. Ifthe effect of the winds is great, as they could beexpected to be in the case of a large breach, thegradients will be pushed inward, concentratingbuilding activity there. However, a different patternof repair results when the opening in the moundwall is small (S. Turner pers. obs.). Part of myphysiological work with these mounds has requiredme to drill small holes (25 mm diameter) into thesurface conduits. In those circumstances, repairwork is always concentrated on the mound’s outersurface. For small holes such as these, theadmission of wind energy might be smaller, themetabolic energy driving colony air outward mayprevail, and gradients in pO2, pCO2, and pH2O mayactually extend outwards from the hole,concentrating building activity there.

If building activity is generally biased by fields ofpO2, pCO2 and pH2O, it may help explain the mound-building habit in general: mound-building resultsfrom termites translocating soils along fields ofconcentration in respiratory gases, temperature andperhaps momentum in air (proportional to velocity).

These fields will radiate outward from the colony,and soil translocation will also be outward.Furthermore, termites can only extend the moundsurface outward by opening up small egress holesin the mound, such as those depicted in Figure 20. Ifthese act similarly to the small holes I have drilled inmounds, this may concentrate building on themound’s outer surface, causing it to grow outward.

In this conception, architecture emerges from theways a mound interacts with all the factors thatinfluence its concentration fields for the respiratorygases. These could include colony metabolism,temperature distribution or kinetic energy in wind.The crucial thing is that mound structure is bothresult and cause of the concentration fields, in otherwords, mound architecture is part of a closed-loopsystem in which feedback interactions are likely toplay an important rôle.

How might such feedback loops operate? Thequantitative measurements of tunnel and mounddimensions suggest that the mound consists of twostructurally distinct components: the spire, and theconical base (e.g. Figures 11, 24-26). This mightimply two such closed loops commonly operate togenerate the building of the mound. I suggest thatboth loops operate as positive feedback loops, withmound growth being driven by a kind ofautocatalytic process whereby an action taken tocorrect a problem (i.e. structural alteration of themound to correct an anomalous concentration ofrespiratory gases) serves only to aggravate thestimulus that initiated the action.

Let us first turn to the upward growth of the spire(Figure 34). Colony metabolism changes not onlythe composition of the colony air but also itsdensity, by warming and humidifying it.Consequently, colony metabolism imparts buoyantforces to the colony air that will tend to loft itupwards. In the initial stages of mound growth,these forces will be small, because the colonymetabolism will be small, and the mound shouldgrow uniformly outward with no particular bias. Asthe mound grows, however, it will begin to excavatelarge internal spaces, according to the ‘bucketbrigade’ model outlined in Figure 33, and itsmetabolic rate will increase, which magnifies thebuoyant forces on colony air. Thus, colony air willlikely be more forcefully lofted upwards through thenewly-opened large tunnels in the center of themound. Consequently, gradients of the respiratorygases and water vapor will steepen in the upperparts of the chimney, and blastis should be most

outwardgrowth

holeheap

Figure 33. The ‘bucket brigade’ model of tunnelgrowth. Details in text.


intense there, causing the mound to grow upwards.As the mound grows upwards, however, it alsogrows into the gradient of wind speeds thatcomprise the surface boundary layer. Because thesehigher winds will bring more wind energy to themixing of colony air, these should also steepen theconcentration gradients at the uppermost parts ofthe spire, stimulating blastis there, with the stimulusintensifying as the spire grows upwards into evermore energetic winds. Eventually, this aggravatingeffect of increasing wind speeds will taper off as themound grows high enough to get into the freestream (where momentum no longer varies withheight), or where the mound height equilibratesbetween rates of upwardly directed blastis, anderosion-mediated clastis (Figure 34).

The mound’s conical base may also grow outwardby a similar autocatalytic process, although onediffering in some detail from that proposed for thespire (Figure 35). The similarity of ‘boils’ (Figures21, 22), ‘moundlets ’ (Figure 8) and ‘domes’ (Figure7) suggests a developmental sequence whereby the‘boil’ is precursor to the ‘moundlet’, which isprecursor to the ‘dome’. The process probablybegins with the appearance of a diagonal tunnelthat emerges from the colony (‘rc’ in Figures 12-17,Figures 22-23), which Darlington (1984) hasidentified as egress holes constructed foremergence of the alates. However, their location andsize indicate that they may also be formed by theemergence of mushrooms of the symbiotic funguscultivated by the termites. Irrespective of how theholes are formed, though, immediately followingtheir opening, the termites work to seal the hole.Because of the small calibre of these holes, the

gradients in respiratory gas concentrations mightextend outward from the mound’s outer surface,thus concentrating the repair work there, formingthe ‘boil’ (Figure 21). As the ‘boil’ grows, however,its outside surface area also expands, whichenhances the capacity of the growing ‘boil’ for gasexchange and for interaction with momentum inwinds (Figure 35). The direct connection of the‘boil’ to the colony through the radial channel alsoprovides a direct conduit for mixing energy fromcolony metabolism. The ‘boil’ may therefore growautocatalytically, with the attempt to seal the holeintensifying the stimulus that initiated the buildingin the first place, which worsens the more thetermites work to fix it (Figure 35).

Thus, several of the architectural features of the M.michaelseni mound can be explained through suchautocatalytic feedback loops. This suggests ageneral model of mound morphogenesis which canbe summarized roughly as follows (Figure 36). In theyoung mound, which is small and in the shape of asimple heap, partial pressure gradients are uniform,reflecting the diffuse and relatively disorganizedarrangement of tunnels in the mound (Figure 36a).As the mound grows, and as ‘bucket-brigade’ soiltransport continues, large spaces begin to open inthe mound center, and which are the precursor forthe chimney. These serve to channel theincreasingly buoyant colony air up to the top of themound, steepening the concentration gradients ofthe respiratory gases there, and consequentlyfocusing blastis there (Figure 36b). The result is theupward elongation of the mound to form the spire(Figure 36c). Once the colony has attained sufficientsize, either to produce alates or a crop of

wind energy

ventilation ofupper air spaces

of mound

local↓↑

↓

CO O

H O

2

2

2

blastis attop of mound

upward extensionof mound

higher windspeeds

+

+

+

+

+

+b.

a.

wind speed

height

Figure 34. Hypothetical scheme for the autocatalytic upward growth of the spire. Details in text. a. As themound grows, as indicated by the darkening shades of gray, it projects into energetically more powerfulwinds, indicated by the velocity profile to the left. b. Series of positive feedback steps which result inautocatalytic upward growth.


mushrooms, the resulting diagonal tunnels willprovide a direct channel for the movement of colonyair to the surface, with gradients of carbon dioxide,oxygen and water vapor being steepened there asthey are in the upper parts of the chimney (Figure36c). These steepened gradients will elicit blastisthere, which will develop eventually into theautocatalytic growth of the ‘dome’. As theautocatalytic growth proceeds, working gallerieswill develop beneath the developing ‘dome’ (as inFigure 22), which will eventually grow and ramify toform the surface conduits. Add to this the subtlebiasing effects of solar heating and erosion (Figure32), and a reasonably simple model formorphogenesis in the M. michaelseni moundemerges.

One of the tantalizing aspects of this model ofmound morphogenesis is that it may help explainhow social homeostasis emerges from theconstruction of the mound. Social homeostasis, inwhich the collective activities of a colony’sinhabitants results in the regulation of the colonyenvironment, has been a long-standing problem inthe biology of social insects, in part because itdraws its inspiration from the problematic conceptof the insect colony as a superorganism (Wilson1971). For example, how are the activities of thecolony’s multitudinous inhabitants coordinated sothat colony homeostasis results? With respect toMacrotermes in particular, nest architecture is widelyheld to be the key to social homeostasis in theseinsects (Lüscher 1961; von Frisch and von Frisch

1974; Wilson 1971). Yet, how do the inhabitants ofthe colony know when their nest has attained the‘correct’ structure, and how are corrections todeviant structures carried out?

A solution to these quandaries can be found byconsidering homeostasis not as a regulatoryphenomenon, but as a process that brings it about.Homeostasis results when the fluxes of matter andenergy into and out of the colony are matched. Forexample, when the rate of oxygen consumption bythe colony is matched to the rate of oxygen fluxacross the boundary separating the colony from itsenvironment, a homeostasis of oxygenconcentration in the colony is the result. If soiltranslocation and the mound structure that resultsaffects the concentration gradients for oxygen inthe mound, this may, in turn result in the matchingof oxygen flux to oxygen consumption. Socialhomeostasis, then, is not the product of aregulatory process, as it is in the regulation of, say,body temperature. Rather, it is a simple by-productof soil transport by termites, insofar as the soiltransport both affects, and is affected by thegradients in oxygen concentration within themound.

This simple model also might point the way tounderstanding morphogenesis of mound structureamong other macrotermitines. The model of moundmorphogenesis I have outlined is based upon fiveprincipal assertions:

ventilation ofsurface air

spaces

blastis atside of mound

outward expansionof dome

+

+

++

+

b.

opening ofegress hole

increased surfacearea for

exchangelocalCOO

H O

2

2

2

↓↑

↓

a.

Figure 35. Hypothetical scheme for the autocatalytic growth of a ‘dome’. Details in text. a. The repair of asmall egress hole results in the continued outward growth of the mound, culminating in the development ofa ‘dome’. b. Series of positive feedback steps which result in autocatalytic outward growth.


1. Soil translocation occurs along gradients inconcentrations of oxygen, carbon dioxide,water vapor and momentum

2. The intensity of soil translocation, i.e. the rateof translocation work, or more succinctly, thetranslocation power, is proportional to thesteepness of these gradients.

3. Any factor which ‘verticalizes’ thesegradients, i.e. steepens them in the verticaldimension, will tend termite colonies towardthe construction of mounds. Such factors caninclude metabolic rate of the colony, porosityof the soil compared to tunnels excavated bytermites, and concentration of metabolic powerinto compact colonies.

4. Gradients in the respiratory gases can also beinfluenced by external environmental factors,such as winds, insolation or rainfall. Theinteraction of soil translocation and colonymetabolism with these external factors caninvolve worker termites into autocatalyticloops of building, which cause the mound togrow uncontrollably.

5. The tendency of mounds to open throughchimneys or large vents is dependent uponhow intensely verticalized the gradients in themound are. The steeper these gradients are inthe vertical dimension, the more likely it will be

that the mound will open through a chimney.

This model is admittedly speculative, but it doesoffer an explanation for some of the major variationin mound architectures to be found amongst themacrotermitines. Consider, for example, the dramaticvariations described by Grassé & Noirot (1961) forBellicositermes (= Macrotermes) natalensis, the moundas a broad, low hillock, or as a strongly verticallyoriented mound. The hillock type of mound iscorrelated with a broadly dispersed colony, withfungus gardens, and the metabolic power thatemanates from them, scattered over a broad extent.The vertical mound, in contrast, has its fungusgardens concentrated in a ‘cap’, roughly 1.5 to 2 min diameter, that sits atop the colony. In the hillockmound, even if the metabolic production of thecolony was prodigious, the metabolic heat andhumidification of the air can be blunted byconduction into the surrounding soil, reducing thebuoyant forces that could produce strong verticalgradients in respiratory gases. Blastis in the colonywill be broadly distributed, but locally weak, andmore likely to be leveled by erosion, with the resultbeing a low hillock mound. Concentrating thecolony, on the other hand, will also concentrate therelease of metabolic energy that heats andhumidifies the colony air, which will magnify theresulting buoyant forces, and hence the verticalsteepness of the gradients in the mound. The localconcentration of blastis will result in a columnarmound.

Properties of the soil also are likely to influence howstrongly verticalized the gradients in respiratorygases are. A very porous soil, for example, willenable horizontal movements of gases by diffusion,reducing the vertical dimensions of gas movement,and the vertical translocation of soilcommensurably. A very compact soil, on the otherhand, will bottle up respiratory gases, and theirmovements will be strongly biased by the pathwaystermites themselves open through the soil. This mayhelp explain the tendency of mounds to produce, insome circumstance, many ‘moundlets ’ or ‘boils’,while mounds in other circumstances do not. On theone hand, in a mound that is built from verycompact soil, any egress hole, whether it be for theexit of alates, or for mushrooms, or for workers, asseen in Figure 20, will result in the formation of a‘boil’, and potentially into a ‘dome’. If the mound isvery porous, on the other hand, the opening of anyegress hole may have less effect on theconcentration gradients in the tunnels, and henceon the stimulation of ‘boils’ and ‘domes’.

a.

b.

c.

d.

Figure 36. A model of morphogenesis in the moundof Macrotermes michaelseni. Details in text.


Similarly, the vertical position of the colony mightoperate through this mechanism to alter whether themound opens through a chimney or not. In general,mounds that are located deep underground are morelikely to have open chimneys than those that arenot. Two widespread species with open chimneys,Macrotermes subhyalinus and Odontotermestransvaalensis , both locate their colonies moredeeply than closed-mound types such as M.michaelseni or M. natalensis. Again, how verticallyoriented the gradients can be will depend upon theextent of horizontal movements of the gases. In adeeply buried colony, the possibilities for horizontalmovements may be limited more than they would bein a colony that extends upward into the mound asdoes the M. michaelseni colony. A severe enoughrestriction on the horizontal movements of gasesmay eventually force the gradients to ‘spill out’over the tops of the mound, with the result being acontinual translocation of soil out and over the top,producing a chimney.

ACKNOWLEDGMENTS

This research was supported by a grant from the Earthwatch Institute.I am grateful to Margaret Voss and Robbin Turner for theirinvaluable help as field assistants on this project. Eugène Marais wasvery helpful in pointing me to sites where Macrotermes colonieswere abundant. I am also grateful to Mr. Freddie Pretorius, whokindly allowed us to work on his farm, Namatubis, near Outjo, andto Jaco and Riette Liebenburg, who provided us with comfortableaccommodation and warm hospitality through the three years of thisproject. This work could not have been undertaken without thegenerous contributions of time, effort and money on the part of 69Earthcorps volunteers who helped on this project: Arthur Anderson,Barbara Anderson, Penny Austin, Dorothy Bailey, Margaret Baisley,Kay Bandell, Callan Bentley, Ronald Bentley, Mari Blanchard,Valeris Brissett, Mary Coroneos, William Coyle, Margaret Dales,Rita Dent, Greg Fendrich, Sue Fielden, Cheryl Foster, ClareFreeman, Gary Freeman, Marilyn Gardner, David Green, EileenGroth, Robert Habedanck, Mirjam Hamunghete, Liv Hansen,Margaret Hillenbrand, Yoshiki Hosojima, Paul Husted, ToshiakiIchimura, Rupert Ingram, Olaf Jansen, Michael Johnson, ErikoKanayo, Sam Kaufmann, Matthew Law, Nancy Law, Mark Lewis,Phil Loveday, Noelle Lusane, Shaw Lynds, Oliver Manaka, PeterMcCartney, Mari McIntosh, Sue McNeill, Darla Meier, ChristineMergenthaler, Cheryl Mikuly, Patrine Mwetulundila, MichaelNovikoff, Thomas O'Donnell, David Pennack, Gilda Peress, FrankPerkins, Laurie Salisbury, Dorothee Saltmarsh, James Sentman,Josiane Schutz, Elaine Stambaugh, John Swanwick, Kit Swanwick,Juan Carlos Vicente, Stan Walerczyk, Steven Wolf, and LyndaZartarian. The work was carried out under a permit from the NamibiaMinistry of Environment and Tourism, Republic of Namibia.

REFERENCES

ABE, T. & DARLINGTON, J. P. E. C. 1985. Distribution and abundanceof a mound-building termite, Macrotermes michaelseni, with specialreference to its subterranean colonies and ant predators. Physiologyand Ecology Japan 22: 59-74.

CANCELLO, E. M. 1991. Two different mounds of Cornitermesbequaerti (Termitidae, Nasutitermitinae): An example of theplasticity in termite nest architecture in the neotropics. RevistaBrasileria de Entomologia 35(3): 603-606.

COATON, W. G. H. & SHEASBY, J. L. 1972. Preliminary report on asurvey of the termites (Isoptera) of South West Africa. CimbebasiaMemoir 2: 1-129.

DANGERFIELD, J. M., MCCARTHY, T. S. & ELLERY, W. N. 1998. Themound-building termite Macrotermes michaelseni as an ecosystemengineer. Journal of Tropical Ecology 14: 507-520.

DARLINGTON, J. P. E. C. 1984. Two types of mounds built by thetermite Macrotermes subhyalinus in Kenya. Insect Science and itsApplication 5(6): 481-492.

DARLINGTON, J. P. E. C. 1985. The structure of mature mounds of thetermite Macrotermes michaelseni in Kenya. Insect Science and itsApplication 6(2): 149-156.

DARLINGTON, J. P. E. C. 1987. How termites keep their cool. TheEntomological Society of Queensland News Bulletin 15: 45-46.

EMERSON, A. E. 1956. Regenerative behaviour and social homeostasisin termites. Ecology 37: 248-258.

GRASSÉ, P. P. & NOIROT, C. 1961. Nouvelles recherches sur lasystématique et l'éthologie des termites champignonnistes du genreBellicositermes Emerson. Insectes Sociaux 8: 311-359.

HARRIS, W. V. 1956. Termite mound building. Insectes Sociaux3(2): 261-268.

INOUE, T., VIJARNSORN, P. & ABE, T. 1997. Mound structure of thefungus-growing termite Macrotermes gilvus in Thailand. Journal ofTropical Ecology 13: 115-124.

KORB, J. & LINSENMAIR , K. E. 1998. The effects of temperature on thearchitecture and distribution of Macrotermes bellicosus (Isoptera,Macrotermitinae) mounds in different habitats of a west AfricanGuinea savanna. Insectes Sociaux 45: 51-65.

LOBRY DE BRUYN, L. A. & CONACHER , A. J. 1990. The role of termitesand ants in soil modification: A review. Australian Journal of SoilResearch 28: 55-93.

LÜSCHER , M. 1956. Die Lufterneuerung im nest der termiteMacrotermes natalensis (Haviland). Insectes Sociaux 3: 273-276.

LÜSCHER , M. 1961. Air conditioned termite nests. ScientificAmerican 238(1): 138-145.

NICOLAS, G. & SILLANS, D. 1989. Immediate and latent effects ofcarbon dioxide on insects. Annual Review of Entomology 34: 97-116.

POMEROY, D. E. 1976. Studies on a population of large termitemounds in Uganda. Ecological Entomology 1: 49-61.

POMEROY, D. E. 1977. The distribution and abundance of largetermite mounds in Uganda. Journal of Applied Ecology 14: 465-475.

RUELLE, J. E. 1964. L'architecture du nid de Macrotermes natalensiset son sens fonctionnel (pp. 327-362). In. A. Bouillon (ed.) Etudessur les termites Africains. . Paris, Maisson et Cie.

RUELLE, J. E. 1970. A revision of the termites of the genusMacrotermes from the Ethiopian region (Isoptera: Termitidae).Bulletin of the British Museum of Natural History (Entomology)24(9): 366-444.

RUELLE, J. E. 1975. Macrotermes michaelseni (Sjöstedt). A newname for Macrotermes mossambicus (Hagen) (Isoptera: Termitidae).Journal of the Entomological Society of Southern Africa 40: 119.

RUELLE, J. E. 1985. 6. Order Isoptera (termites) (pp. 53-61). InSCHOLTZ, C. H. & HOLM, E. (eds). Insects of southern Africa,Butterworths, Durban, 1-502 pp.


RUELLE, J. E., COATON, W. & SHEASBY, J. 1975. National survey ofthe Isoptera of southern Africa 8. The genus Macrotermes Holmgren(Termitidae: Macrotermitinae). Cimbebasia 3(A): 73-94.

STUART, A. M. 1972. Behavioural regulatory mechanisms in thesocial homeostasis of termites (Isoptera). American Zoologist 12:589-594.

TURNER , J. S. 1994. Ventilation and thermal constancy of a colony ofa southern African termite (Odontotermes transvaalensis:Macrotermitinae). Journal of Arid Environments 28: 231-248.

TURNER , J. S. in press. The extended organism: the functionalbiology of animal-built structures. Harvard University Press,Cambridge, Massachusetts.

VOGEL, S. 1978. Organisms that capture currents. Scientific American239: 128-139.

VOGEL, S. A. & BRETZ, W. L. 1972. Interfacial organisms: passiveventilation in the velocity gradients near surfaces. Science 175: 210-211.

VON FRISCH, K. & VON FRISCH, O. 1974. Animal architecture. NewYork, Harcourt Brace Jovanovich, 306 pp.

WEIR , J. S. 197). Air flow, evaporation and mineral accumulation inmounds of Macrotermes subhyalinus. Journal of Animal Ecology42: 509-520.

WILSON, E. O. 1971. The insect societies . Cambridge, Massachusetts,Belknap/Harvard University Press, 548 pp.

Manuscript received November 1999, accepted February 2000

Added note: This is a facsimile reproduction of thepaper that appeared in Cimbebasia 16: 147-175.Consequently, it differs from the print version informatting and layout, and hence, pagination. It alsoincludes two figures inadvertently left out of theprint version. Otherwise, the content is the same asin the print version.

Architecture and morphogenesis in the mound of Macrotermes … pdfs/Cimbebasia 16_00.pdf ·...

Documents

Transcript of Architecture and morphogenesis in the mound of Macrotermes … pdfs/Cimbebasia 16_00.pdf ·...