Submillimetre mechanistic designs of termite-built structures

Abstract

Termites inhabit complex underground mounds of intricate stigmergic labyrinthine designs with multiple functions as nursery, food storage and refuge, while maintaining a homeostatic microclimate. Past research studied termite building activities rather than the actual material structure. Yet, prior to understanding how multi-functionality shaped termite building, a thorough grasp of submillimetre mechanistic architecture of mounds is required. Here, we identify for Nasutitermes exitiosus via granulometry and Fourier transform infrared spectroscopy analysis, preferential particle sizes related to coarse silts and unknown mixtures of organic/inorganic components. High-resolution micro-computed X-ray tomography and microindentation tests reveal wall patterns of filigree laminated layers and sub-millimetre porosity wrapped around a coarse-grained inner scaffold. The scaffold geometry, which is designed of a lignin-based composite and densely biocementitious stercoral mortar, resembles that of trabecula cancellous bones. Fractal dimension estimates indicate multi-scaled porosity, important for enhanced evaporative cooling and structural stability. The indentation moduli increase from the outer to the inner wall parts to values higher than those found in loose clays and which exceed locally the properties of anthropogenic cementitious materials. Termites engineer intricately layered biocementitious composites of high elasticity. The multiple-scales and porosity of the structure indicate a potential to pioneer bio-architected lightweight and high-strength materials.

1. Background

The morphology of termite mounds (termitaria) and other build structures (e.g. foraging galleries) is diverse and can be linked to different trophic categories, such as soil and soil/wood interface feeding, wood and litter foraging or specialized and incidental feeding, evolved over more than 181 Ma as a consequence of changed habitat, predation and tightly linked to sociality and brood care [1–4]. Termitaria are classified into hypogeal (subterranean), epigeal (above ground, usually called ‘mounds') and arboreal (within or attached to tree) nests, for details cf. [1]. Even though living in complete darkness, termites have evolved the ability to allocate their tasks efficiently among up to several million nest-mates [5,6]. Staying cryptic permits termites to live a life without the dangers of the outside world, with little predation or being exposed to desiccation [6–8]. Apart from being protective, termite mound construction must provide various functions including stability, ventilation, serving as food storage and nursery and to transmit colony information [1,10–12]. In addition to using biotremology, termites use pheromones and tactile information as a direct communication modality as well as stigmergy as indirect communication, with much research of termite building being dedicated to this aspect of self-organization [1,12].

The general complexity within the mound and its multifunctionality, the self-organization of the colony and related stigmergic effects of termite-built structures continue to fascinate scientists and engineers [1,5,7,12]. The nest, especially the inner carton material that can be found in some genera of termites, functions as food storage [10,13,14] and as intricate ventilation system that thermoregulates the climate in the mound (homeostasis) [7,8]. The homeostasis aspect has already motivated architects to conceptualize eco-friendly evaporative cooling systems for human dwellings [15,16]. Especially, the macro and microporosity of termite manipulated soils is assumed to take an important role in storing moisture and releasing humidity essential for the evaporative cooling functionality of the whole nest [1,7].¹ Soil porosity levels reportedly vary between 20 and 40% [1,9,12,17–19], yet no reported studies has ever been conducted in determining microporosity [1]. Some termites, such as Coptotermes acinaciformis, have shown the ability to use their clay-built structures as a foraging tool to support load and to unlock otherwise inaccessible food resources [10,20]. The mound and the constructions built on their foraging sites have been shown to be macroscopically structurally different: Odontotermes obesus mounds to take up to 1.5 MPa compressive loads [21–23]; the clay walls designed to keep loads at a foraging site in C. acinaciformis had a pressure load capacity of 0.22 MPa, which maximize foraging efficiency by maintaining compressive loads from tree weight [20]. Perna et al. [24] introduced computed tomography (CT) to visualize the architecture of Cubitermes sp. nests to study them using graph theory, while Himmi et al. [25] used X-ray tomography to study the tunnel system of Incisitermes minor in wood, highlighting the termite nest as the morphological expression of the sum of termite behavioural patterns. However, as pointed out by Oberst et al. [1] only medical CT has been applied to the study of termite structures so far.

Apart from those macroscopic mechanical properties, which have received some attention in recent years contemporary research has been mainly concerned with the topology of tunnel systems, aspects of self-organization or swarm behaviour, often only computationally [1,8,18,19,24] and no study yet is focussed on the microscale mechanical or micro-geometrical properties of real-life termite nests [1]. While Zachariah et al. and Oberst et al. studied the moisture content, stability and also weather resistance to moisture and conducted compression tests on soil samples [10,21,22], including termite-contacted timbers, there are few detailed analyses of other influential factors for many termite species, especially particle size and soil organic matter have been conducted. Zachariah et al. [23] studied the building blocks (boluses) of O. obesus, which are ca 1 mm—sized to fit into termite worker jaws—and described them as key elements of spherical unitary shape made of granular hydrophilic, osmotically inactive, non-hydroscopic materials with surface roughness, rigidity and organic matter. While this describes some physical properties, it can be assumed that the micro-structure is different and characteristics such as elasticity or building aspects on the local level, in parts of the termitarium, associated with certain corridors and chambers, or even within the structure of an internal construction, will show variations important for the understanding of termite functional, hierarchical constructions.

This study is the first attempt to understand the submillimetre mechanistic designs of a mound building termite, including geometry, materials used and material properties. Submillimetre mechanistic details are important to understand the termite mound's functionality including subfunctions such as stability, geochemical characterization, self-ventilation and climatization, microporosity and microfluidics, gas exchange or wave propagation properties, all of which also relate to termite behaviour and the colony as a superorganism [1,12]. Nasutitermes exitiosus (Termitidae) [13,14], which is known for its phylogenic diversity and various constructions, serves as study species with its paraecie nest fractions examined. N. exitiosus takes a special role in the phylogenetic tree as it bridges the evolution of mound builders to other Nasutitermes spp. which are more often arboreal [27,28]. Nasutitermes exitiosus provides a good opportunity to consider the different areas of the mound and the properties of the materials used in each area. Nasutitermes species build their nests i.e. the construction containing the reproductive, brood and moulting areas, thin parallel walls/floors, either primarily or entirely made from with pale coloured faecal materials. Surrounding the nest is the ‘mound material', with its thick irregular walls, made from a mixture of soil and dark faecal matter. Surrounding the mound material is the mound wall or soil capping, made mostly from soil with some faecal material.

Granulometry and Fourier transform infrared spectroscopy (FTIR) are applied to parts of the termitarium to gather information about which particle sizes termites use to build and to what content the wall structure is composed of organic or inorganic constituents. High-resolution X-ray micro-CT (μCT) scans are then conducted to study the geometric properties on the submillimetre detail level. We then conduct microindentation tests to study local mechanical properties.

2. Methods

2.1. Collection of specimens

We collected internal structures of two termitaria of N. exitiosus near the Australian Defence Force Academy in Canberra (−35°17'22.80″ S 149°09'30.00″ E). The termitaria were mounds of comparable size; heights of about 300 mm and diameters of 650 mm, cf. electronic supplementary material, S1a. First, the outer soil capping of the selected mound section was removed. We cut two samples of inner endoecie from two mounds of about 100 mm below the outer soil capping (conceptually depicted, see figure 1), with a size of 78.6 mm × 64.5 mm × 94.6 mm using a spatula. We could perceive manually a difference in hardness and brittleness between the outer and inner endoecie, which guided the location of the sample.

Figure 1.

Schematic of a Nasutitermes exitiosus termite mound with its different parts highlighted. The sample location is indicated as the ‘inner endoecie' which we found to be a harder than the outer endoecie. For the inner endoecie, we studied the wall structure (magnified in magenta coloured circle), which we assumed from visual inspection (electronic supplementary material, figure S1) to have two parts, an outer wall (OW) and an inner wall structure (IW).

We discarded loose parts and debris, as well as termites, and placed the samples into self-sealing bags then into plastic food containers for transportation (electronic supplementary material, figure S1); we later vacuum sealed the bags. We also took soil samples from six locations around the mound at 1 m distance to the centre of the mound and 60° apart by impacting a 1.5 m metal cylinder with 14 mm diameter (1 mm thick wall) to about 50 cm depth and extracting the inner soil cylinder for analysis. Details of this method are provided within electronic supplementary material, S1.

2.2. Granulometry

We cleaned and sieved the samples from the termitaria and the surrounding soil to remove lose debris, and organic matter such as leave litter and twigs (cf. electronic supplementary material, S2a). We oven dried samples at about 50°C for about 8 h to remove water. The samples from the termitaria are typically assumed to consist of clay particles (selected by the termites), in chewed boluses mixed with faeces, whereas the surrounding soils should consist of a wider range of particle size distributions [29,30] (standard protocol, in general, specific for termite soils—details can be found in electronic supplementary material, S2). We removed after visual inspection parts of the outer layers of wall segments to be ground and analysed separately from the whole wall segments. The inner wall structure was visibly brighter in colour under bright illumination (2000 lumens) due to seemingly coarser grained boluses being used, so separation of the outer wall (OW) from the inner segment was realized with a spatula and scalpel.

For the particle analysis, ground soil samples (as previously collected from the ground 1 m away from the termitaria) were mixed with water until the laser obscuration was between 10 and 20%. To achieve that 1.54 g of ground termite wall structure were dissolved in 1 l warm water and 20 ml of SHMP solution (sodium hexametaphosphate surfactant 20.14 g SHMP in 0.1 l H₂O at 50°C), cf. electronic supplementary material, S2b. We used a particle analyser (Mastersize 2000, Malvern Panalytical, Malvern, UK, electronic supplementary material, figure S2), with a 3R HeNe gas laser (4 mW, 633 nm) to measure the scattering pattern of laser diffraction, assuming spherical or non-uniform particles, cf. electronic supplementary material, S2c. The detector array takes 2000 (1 ms duration) snapshots of the particle scattering pattern passing through the analyser beam (n = 3 averages). First, the Malvern particle analyser was calibrated for fly ash. Then the clay materials were studied, and the refractive index of the particle analyser was then set to 1.55 with an absorption coefficient of 0.1. The emulsion of water with SHMP and particles was constantly stirred at 1250 r.p.m. for 5 min and after each measurement further dispersed using ultrasonic excitation probe set at 20 µm tip displacement for 2 min to disperse large agglomerates of particles and to avoid re-agglomeration.

2.3. Fourier transform infrared spectroscopy

Using FTIR would enable us to determine inorganic and organic contents within the sample structure compared to the surrounding soil (details electronic supplementary material, S3), which has not been done before (details electronic supplementary material, S3). We ground a small sample piece (2 g) of termitaria from the endoecie sample using a Retsch PM 100 (less than 1 µm) planetary ball mill, cf. electronic supplementary material, figure S3.1. The samples were first ground using a metal pestle and mortar, then oven dried (105°C, 5 h). The balls as well as the vial of the ball mill were cleaned with ethanol before and after each use, then the ground sample was milled for about 180 s at 290 r.p.m. We then used an FTIR scanner (Nicolet 6700, Thermo Fisher Scientific, Waltham, MA, USA, electronic supplementary material, figure S3.1) to obtain the ‘fingerprint' of the material's molecular absorption/transmission, by measuring absorption peaks of the vibration frequencies between atomic bonds [31]. After measuring the background noise level, we took 32 averages at a spatial resolution of approximately 4 µm with the spectral resolution set to 0.482 cm⁻¹ (wavenumbers ranging from 4000 cm⁻¹ to 450 cm⁻¹). The average density of the endoecie's OW structure was generally estimated by using Archimedes' principle, details within electronic supplementary material, S5.

2.4. Micro-computed X-ray tomography

We used high resolution X-ray helical μCT to analyse the fine grain geometry of termite-built structures. The images and data were analysed at the Australian National University's (ANU) μCT facility (CTLab) using a transmission-type X-ray source, with beam voltage and current of 80 kV and 110 µA, respectively (details within electronic supplementary material, S4). CT allows for high resolution, high signal-to-noise ratio, high cone beam geometry and large field of view imaging [32] to identify the detail structures of termite-built inner walls for the first time. We will call the parts we identify in the following tunnels, nodes (chambers) [1], inner wall (IW) and OW structures. We scanned the termite samples and converted these into 2D image stacks using ANU's Multi-modal Australian ScienceS Imaging and Visualization Environment (MASSIVE) imaging and visualization cluster. Tomography renderings and porosity percentages were produced using ANU's Medial Axis and Network Generation software (Mango) [32] and Drishti volume rendering software [33]. We estimated the pore volume, the micro- and macroporosity as well as the fractal dimension from the images, using Taud's method and according to Minkowski–Bouligand as outlined in electronic supplementary material, S4 and we compared our calculations with the volume fractal dimension [34,35]. The commercially available Avizo^® (FEI™) software was used to binarize, threshold, segment, smooth and output volumes for each individual pore within a given dataset. Each particle was given an equivalent spherical diameter based on its volume to determine the median sizes of particles. The scan had a voxel size of about 42.3 µm.

2.5. Microindentation tests

Microintendation tests were conducted on thinly sliced termite nest samples. Due to small particle size and potentially fractal structure, we used microindentation tests to measure hardness on the microscale [36] (details within electronic supplementary material, S6). Microindentation (indentation depth greater than 0.2 µm) was developed to determine the mechanical properties of materials at small scales, where a tip, generally made from diamond, is pressed into a sample at a known load to estimate indentation hardness and indentation modulus by curve fitting. The average hardness over the area of the sample was then calculated. To conduct the microindentation tests, a diamond saw was used to cut thin (3 mm) slices of a termite wall structure sample, electronic supplementary material, figure S6a. Here, we used a tribo-indenter TI 900 (Hysitron) with a maximum load of 10 mN, with an optical microscope, modulus mapping and a Berkovich diamond tip, electronic supplementary material, figure S6c (details within electronic supplementary material, S6).

3. Results

3.1. Granulometry and Fourier transform infrared spectroscopy

Soil particle sizes are classified into clay (fine, less than 4 µm), silt (less than 62 µm) and sand (coarse, greater than 62 µm) [29,35], with termites usually preferring clay-like material for the endoecie [1,5,7,12,29]. Using granulometry, the particle sizes of soil samples taken in the surrounding of two termite mounds (colony A and B) show peaks at about 49.1 µm and higher sand content (greater than 100 µm), figure 2a.1.

Figure 2.

Granulometry and Fourier transform infrared spectroscopy (FTIR) study. (a.1) Surrounding soil with particles also greater than 10² μm; and (a.2) termite mound structure which has mostly particles sizes of 4 µm to 47.5 µm; fly ash provided for comparison; Fourier transform infrared spectroscopy (b.1) surrounding soil with strong inorganic components, (b.2) outer wall structure, indicating organic components and (b.3) outer and inner wall structures (whole wall segment, ground) with paraffin-like components within the inner wall segments.

The termite structures show peaks at about 18.3 µm and 47.95 µm for colony A; and peaks at about 4.01 µm, 18.5 µm and 47.5 µm for colony B (figure 2a.2), i.e. mostly silt, with no sand. FTIR analysis (figure 2b) indicates inorganics (minerals, most common quartz, second most common kaolinite) for wavelengths below 1000 cm⁻¹. Figure 2b.2 highlights that the outer layers of termite wall structures contain more organic groups than the surrounding soil between 1600 cm⁻¹ and 1100 cm⁻¹ (C–O deformations, lignin, cellulose, C=O). By grinding the termite wall structure, the IW segments (figure 2b.3) are implicitly measured as well which results in reduced transmittance at lower wavelengths, yet more pronounced in the range of organic molecules; with a new trough appearing at around 2920 cm⁻¹ and 2840 cm⁻¹—wavelengths of CH3 and CH2 molecules' asymmetrical and symmetrical stretching vibrations commonly found for waxes (e.g. paraffins) and terpenes [37,38]. Nasutitermes exitiosus, while foraging, takes in lignocellulosic components of the cell walls, which contains embedded resins, and which contribute to the formation of sesquiterpene hydrocarbons in the fontanelle glands of the solders which they squirt as a sticky secretion in defence [39–42]. Sticky, terpene hydrocarbon and nitroalkene-rich defence secretions effectively repel predators including mammals (e.g. Tamadua mexicana (Mymecophagidae)) and are presumably included into the termite nest wall for exactly the same reason [43].

3.2. High-resolution micro-computed tomography

By conducting a high-resolution μCT scan of termite nest samples, the existence of an OW and an inner scaffold is confirmed (figure 3). From figure 3b,c, as well as from the histogram taken from the μCT scan (see electronic supplementary material, figure S4.2), the OW appears homogeneous with respect to density as opposed to the inner wall (IW) structure (cf. figure 3c). The latter is highly dense, biocemented and coarse grain with many cavities resembling in its visual appearance to a bone structure, similar to trabecula cancellous bones [44].

Figure 3.

Results of high-resolution μCT study. Termite wall structure geometry and its composition. (a) X–Z cut termite structure (with sample holder in background); (b) X–Z plane of (a) and in (c) its magnification visualizing coarse grain, bone-like scaffolding with tunnel diameters of about 2.8 mm between larger nodes (chambers) [8], inner wall (IW) and outer wall (OW) structures; and (d) magnification of rectangle in (c) of highly dense, layered particles within coarse-grain inner scaffold. The colours in intensity counts (proportional to density, values within the 0.1 and the 99.9 percentile of the distribution) indicate that the centre of the dense scaffold particles is about double as dense as the outer wall layer taking 99.9 per cent of the distribution into account.

Particles of the IW structure are generally up to twice the density (2079.6 kg m⁻³ for Mound A) and up to five times denser (5199 kg m⁻³ for Mound B) than the OW structure (estimation described in electronic supplementary material, S5)—the average macro-porosity for parts of mound A and mound B is estimated to be about 32.19% and 31.71%—while the porosity of the IW structure and the microporosity of the layered OW segments are estimated to be about 22.1% and 32.8% (details of method used see electronic supplementary material, S4), respectively [17]. The topological dimension of the two mound samples is estimated to be 2.87 (Mound A) and 2.71 (Mound B) (details of calculation method see electronic supplementary material, S4) based on the correlation dimension D₂; non-integer numbers indicate fractality [45] which is a known feature in many scientific areas including geology and soil science [34,35,46]. Fractality indicates the presence of multiple scales [46,47], a unique property of many structures in nature and also characteristic of termite nest walls, indicative of increasing porosity on the submillimetre and microscale.

3.3. Microindentation tests

The thin slices of termitaria samples shown in figure 4a highlight the different wall structure component, magnified in figure 4b, which show the coarse grain inner scaffolding (figure 4b.1) as well as a laminated OW structure (figure 4b.2). These samples are representative of the general structure observed to be present in the mound. Figure 3b.2 has been modified using various filters in iPhoto to visualize the complex layering (electronic supplementary material, figure S7). For selected points (figures 4c.1 to 4c.3), microindentation tests were conducted showing hysteretic behaviour (figure 3d), increasing from the outside to the inside, indentation modulus (ranging from 4 to 12.4 GP, 2σ-value) and hardness (ranging from 0.21 GPa to 0.62 GPa, 2σ-value), and which were on average 0.45 GPa and 7.2 GPa, respectively, (figure 4e). Using Hertzian theory and the reduced modulus of elasticity as combination of sample material and indenter elastic deformations [34], we estimated Young's modulus to be on average 24.7 GPa (ranging between 14.2 and 45.1 GPa, details see electronic supplementary material, S6). Such values are orders of magnitude higher than those found for compacted soils which are commonly below 200 MPa [35].

Figure 4.

Microindentation tests. (a) The termite wall structure sample shows a section of (b.1) fine grain, layered outer wall material and coarse-grain, porous inner wall structure as well as a clear (b.2) multi-layered outer wall structure. (c.1) to (c.3) show typical structures magnified during microindentation measurements; with (d) examples for those regions and (e) the distribution of microindentation hardness and modulus.

4. Discussion

Our results for N. exitiosus (Isoptera) are the first to exemplify univocally submillimetre details related to the finer mechanistic structures of termite mound components. Knowing these finer structural details is required for a better … materials, a topic of significant interest in general, as also … required for the understanding of multifunctionality in biogenic materials as also previously raised by Perna & Theraulaz [12], Singh et al. [7] or more recently by Oberst et al. [1].

4.1. Role of porosity

The porosity of termite nests, especially that of the shell and the inner structures, is essential for gas exchange processes, climatization and ventilation properties [7,17–19]. Micro- and macroporosity have been estimated previously to range between 24% to 35% and 19% to 47%, respectively, using photogrammetry or medical CT scans; and the fractal dimension of the outer mound surface was estimated to be about 1.9 [17,18]. Our measurements were based on high-resolution μCT imaging (which is about one order of magnitude more accurate than commonly employed medical CT). We used Taud's method to calculate porosity and we provided an estimate of the fractal dimension using volumetric data, instead of images photogrammetric surface models and 2D cross-sectional images which allow us to quantify the complexity of the internal nest structure more accurately on a smaller scale [1,17,44,45,49]. Our estimates of macro- and microporosity for a sample of two colonies of N. exitiosus ranged between 32.19% to 31.71%, and 34.1% to 32.8%, with an estimated total porosity of 66.29% and 64.51% when compared with about 58% to 59% [34] (for Microceratermes nervosus, Macrognathotermes sunteri and Tumulitermes pastinator). We estimated a volume filling fractal dimension of about 2.87 for colony A (and 2.83, colony B); this converted into two dimensions is about 1.87 (1.83), which is similar to the 1.9 estimate extracted from images of soil samples [17,34].² A pore area (volume) fractal dimension can be related to porosity [48–50]: a 2D fractal dimension of about 1.87 (1.83) therefore should result in porosity values between 60% and 75% (depending on the ratio of maximum to minimum pore diameter)—which is consistent with our estimated total porosity of 66.29% and 64.51% [46,49]. Mapping the pore area fractal dimension with the total porosity [45] provided consistent results in the current scenario neglecting that since the structure is fractal, actual porosity values should also be fractal. Even though we used a relatively fine resolution, the porosity within the smooth plaster wall structure could not be measured. To visualize the microporosity of connected pores down to 4.5 µm voxels, a new method is needed; perhaps a smaller fragment of the outer plaster structure positioned within a vacuumed, then Xenon-filled cylinder [1].

In contrast to the OW structure, the coarse grain inner scaffolding (IW) has large pores found in both samples. Apart from potentially providing stability, the pores may have a potential to store larger amounts of water to support the evaporative cooling function within the narrow channel system of the termite mound [7]. Water applied by termites on the wall surfaces gets absorbed via microporosity and is then used in evaporative cooling and for thermoregulation [1,8]. Alternatively, or additionally, the spongy structure of the inner scaffolding may absorb excess liquids also. Here, simulating the airflow and the thermodynamics via a CFD simulation and conducting tests using 3D printed materials within a chamber with controlled airflow and temperature conditions could provide insights into understanding the function of permeability and water retention capability of termite nest walls and compare this with real-life samples.

4.2. Particle Size

Termitaria, especially the nursery are known to be composed of clays or fine silts [5,12,29] (less than 1 µm [19], cf. figure 2), with walls plastered with layers of faeces [1,5–9,12–14,51]. However, here the inner endoecie of N. exitiosus just below the outer soil capping wall and surrounding the central nursery was composed of coarse silts (figure 2). Clay, due to its finer particle size, contains more water, which means clays have higher shrinkage compared with silts. Clay, especially if exposed to both temperature and humidity extremes, has a higher risk of crack formations due to tension [35,51–54], resulting in lower stability. Cracks in the outer shell, however, may be ‘designed' for the purpose of supporting porous peripheries and permeability for cooling or for water retention as recently evidenced in nests of the fungus growing termite O. obesus [22]; or are simply an effect of weathering. Often, mounds have a rough surface which gets smoothed by rain and wind and which may also lead to different chemical reactions and segregation of particles.

The conditions within the mound also may not be homogeneous, with concentrically, layered temperature and humidity gradients from the nursery to the outer shell. However, the inner endoecie within the mound should not expand or contract too much. Using coarse silts as the main construction material for the inner structure provides the colony with an additional safety margin, to reduce non-essential, energy consuming maintenance activities [10,20]. Particle size is just one of many factors that affect material properties in nest construction. Others include the access to food and water, favourable climatic conditions, or mineralization of soils used [1,8,12,22,29]—termites are likely to be indifferent to some building materials and use just what they find [1,12]. More research is required to better understand the material selection of termites, especially in relation to geological conditions, geography or colony maturity [1,8,12,19]. In this context, it would be of interest to design an instrumented mound in the laboratory, with and without termites, to monitor and possibly control building activities.

4.3. Different densities

Details of properties of the different layers of the OW structure (shown in figure 4b.2 and visualized in electronic supplementary material, figure S7) e.g. densities, indentation hardness or indentation modulus or their multiple scale property have never been documented—novel findings for this paper. We show that the OW plaster is a complex homogeneously laminar layered composite presumably built from liquid mortar (from faeces) and which is known to provide an antibacterial and smooth surface finish ideal to fight pathogens and to avoid unnecessary vortices for enhanced ventilation [1,7–10,48]. Its density is about 1052 kg m⁻³ (similar to colloidal clay) while that of the inner scaffold varies between 2079.6 kg m⁻³ (90 percentile of intensity distribution, figure 3) which is similar to uniform silt with 1738 kg m⁻³ (averaged from the minimum and maximum values which are 1297 kg m⁻³ and 2179 kg m⁻³ [49]), and 5199 kg m⁻³ (99 percentile) which is similar to pyrite with about 4800 to 5000 kg m⁻³ [35,51,54]. Since we know from the granulometry study that the layered plaster wall is mostly coarse silt with some clay components, its microporosity can be roughly estimated using the estimated average densities of uniform silt as reference which produces a relative low density of the OW. This low density is at least indicative for some micro- or even sub-microporosity, estimated to be around 39.1%—which is consistent with our estimate provided via pore volume fractal dimension.

Different densities of various layers, as well as the coarse grain highly dense inner scaffolding, affect the stability of the wall structure and the corridors, acting like a skeleton on top of the soft carton material which houses the colony's nursery as well as the royal cell [1,12,50]. The inner scaffolding resembles cancellous bones in its appearance [55,56], cf. electronic supplementary material, figure S5. It consists of inorganic and organic materials admixed with endogenous secretions made of lignin and cellulose, originally from the plant cell walls [1,10,13,20,22] and which show up at wavelengths of paraffins—distinctly related to terpenes and resins (cf. figure 2b). These compounds are used by termite workers to augment the biocementation of stercoral mortar [8,9,12,14,21,22]. This stercoral mortar of the inner scaffolding is the densest within the centre of a sphere-shaped structure surrounded by layers of lower density. High density is also found in cancellous bones where heavily loaded areas show the highest densities (in humans though only up to ca 1900 kg m⁻³) [56] and in termites it may support the load bearing function during foraging [20]. Also, the material and the particle size itself, the high silt/clay ratio of the inner endoecie, supports stability of a structure, which is less prone to cracks as a result of alternating desiccation and humidification (swelling) when compared with clays, while having similar or higher density, permeability, stiffness and cohesive strength [35,51,52].

4.4. Fine grain termite bricks

The boluses [21–24], the individual ball-shaped ‘termite bricks' [1] (soil-pellets [57]) formed from soil particles and accumulated by individual termites with their mouth parts, are commonly reported as mechanistic building components [9,57,58]. Here, we show that termites do appear to build the inner scaffolding from those termite bricks, while the layered OW structure, similar to layers of plaster are made from liquid faeces of increasingly lower density and hardness (figure 4). Prohamitermes mirabilis has been reported [58] to use prefabricated plugs made from a small foreign particle (sand grain) wrapped with nest cement. Here, the building blocks of the inner scaffolding seem to be made the same way, with highly dense particles in the centre, wrapped with layers of less dense material. The local Young modulus (modulus of elasticity) estimated from the microindentation tests (for more details see electronic supplementary material) was astonishingly high, ranging locally from 14.2 to 45.1 GPa. For comparison, the upper values exceed man-made concrete (40 GPa) and are orders of magnitude higher than compacted soils (200 MPa) [35,51]—by taking only about 32% of its volume owing to its porosity. However, it needs to be noted that these are estimates for local properties and not directly comparable to global values as found in engineering homogeneous materials. For comparison, the global compressive strength of load supporting walls at foraging sites of Coptotermes acinaciformis were estimated to be 0.22 MPa, much lower than concrete, indicating that there might be a difference between the nest where highest safety is required and the foraging site where the termites invest just as much energy as necessary [10,20]. However, the strength of soil depends on mineral composition, grain size distribution and the ratio of clay/silt and other particles/materials, degree of saturation (water in interparticle junctions and related water surface tension), stress history [21,22]. Zachariah et al. [23] showed for O. obesus that construction particle choice being close to its liquid limit attained high compressive strength similar to that of mounds found in the field when dry, by forming a monolithic, densely packed structure. Endogenous secretions were assumed to be mainly responsible for resistance against premature weathering. However, more tests are required, especially for N. exitiosus, including the validation on different scales to deduce the macroscopic properties such as compressive strength, flexural strength and tensile strength from the mechanistic, submillimetre characteristics and multiscale topology as studied here; and to understand how local scales in this heterogeneous material leads to multi-functionality in biogenic, termite-built structures.

While it is without doubt challenging to adapt these multiscale structures for human applications, it can be foreseen that a technology transfer from material properties as found in these biogenic structures to artificial materials will have potential impact on novel concepts of meta-materials and multi-functional materials. The curvature of corridors and structures built by termites as recently studied by Calovi et al. [59], Ocko et al. [60] and Facchini et al. [61] has been shown to play a key role in construction and other properties of termite nests. Similar to the inner scaffolding, curvature may play a central role with regards to stability, wave propagation and other functions [1]. While we did not study curvature in this present study, it seems plausible that species' functional characteristics are directly moulded into geometric features of different termite nests [1,12,22] and self-organized biotectonics should be studied using network quantification, graph theory, multi-agent systems and complex nonlinear times series/dynamics approaches [1,12,22,62,63].

How the formation of the smooth low-density exterior with its coarse-grain, high density interior within a meandering macroscopic structure can be modelled mathematically is another aspect of interest. Granular convection as observed in the Brazil nut effect and fluid flow convection (coffee-ring effect) [64–66] paired with multi-scale reaction and diffusion processes (for the macroscopic structure) [46,47] may provide a model to approximate the observed geometry. If so, while the physical processes in fluid and granular material are similar, their connection to termite building is not obvious but may relate to a higher-arching self-organization/self-segregation principle connecting physics of solids and fluids with that of behaviour in eusocial insects.

Data accessibility

The raw data of the CT scan and the microindentation as well as the density calculations are available from Dryad Digital Repository: https://doi.org/10.5061/dryad.kwh70rz3g. All other information is contained in the electronic supplementary material.

Authors' contributions

Author contributions have been evaluated according to CRediT. S.O. conceptualization, methodology, software, validation, formal analysis, investigation, resources, data curation, writing the original draft, revising and editing the manuscript, visualization, project administration, funding acquisition; R.M. software, revising and editing, visualization; J.C.S.L. methodology, revising and editing; B.J.H. revising and editing; resources; T.A.E. methodology, revising and editing; M.S. formal analysis, investigation, data curation, revising and editing, visualization.

Competing interests

We declare we have no competing interests.

Funding

This research was supported under Australian Research Councils (ARC) Discovery Projects funding scheme (S.O., J.C.S.L., T.A.E.: project no. DP200100358); and it was partially supported through the ARC Training Centre for M3D Innovation (M.S. IC180100008). Furthermore, the authors acknowledge a license for scientific activities under the Nature Conservation Act 2014 within the ACT Tidbinbilla Nature Reserve (S.O. license TS20188).