Research relationships between Indigenous Australians and others involves complex issues of principle and ethics. The intense and continuous impacts of British colonisation and dispossession continue for Indigenous people today, contributing to extreme rates of mortality, morbidity and incarceration. Colonisation and post-colonial impacts affect almost every part of Indigenous people’s lives today. Cross-cultural research in Australia demands engagement to ‘do not harm’ and counter these effects. We work within a positivist paradigm building on the strengths of people.
Authors of this article includes individuals and a collective of Martu elders and traditional knowledge experts. Being the senior authors, they are listed last as is consistent with academic convention. The ‘elders’ entity is because of the unique and complex nature of Australian Indigenous knowledge traditions which accumulate from multiple oral, visual and experiential sources. The ‘expert’ entity recognises the contributing Martu individuals who are not elders but who chose to specialise in knowing and sharing their cultural knowledge.
The non-Indigenous authors are an interdisciplinary team who span natural and social sciences. Their disciplines encompass ethnoecology, arid zone ecology, botany, zoology, entomology, fine art, environmental engineering, hydrology and natural resource management. Collectively, the non-Indigenous authors hold over 130 years of field experience in Australian arid systems.
In the 1980s, knowledge shared by Martu elders was recorded by F.W. It is foundational to this article and provided its starting point. In 1987, Martu knowledge and practice was recorded at the initial request of Martu leaders (especially Ned Gibbs Milangka and Lucy Gibbs Purungu). Twenty-three Martu elders and experts spanning three generations provided specific information related to termites. Nineteen of those people have since passed away.
After Indigenous consultation then literature review and discussion, we challenge the prevailing Western scientific paradigm of authorship by individuals or groups of individuals. We considered co-authorship precedents in Australian and international literature. Previous work has found that less than 14% of Australian publications on Indigenous biocultural knowledge were co-authored46. Precedents in ecological publications include co-authorship with individual Indigenous authors47, Indigenous communities48, Indigenous corporations49, and Indigenous countries50,51. None of these precedents are appropriate to our context so we use an authorship that reflects desert Indigenous multiple and cross-generational custodianship of knowledge. For future research, we are now developing partnerships with Indigenous organisations.
We appraised publications in the scientific literature about ‘fairy circles’ for information on First People’s knowledge. In the publications, we searched for details in their background information, introductions, activities and processes in the methods, and people and organisations in the acknowledgements. Of the 27 papers considered, none included Indigenous authors, none included any activity to gain First People’s knowledge, and only two papers (7%) thanked First People for help (administration and field work). Three papers (11%) include one or sentences on a local myth attributed to Himba and/or Damara peoples about fairy circles52-54
; one of these papers cited a television show as the ultimate source of this information. This is despite San rock art interpreted to be flying termites and termitaria55.
The 1988 – 1993 ethnographic work amongst Martu (F.W.) was funded by The University of Western Australia, Australian Institute of Aboriginal and Torres Strait Islander Studies (AIATSIS) and the Aboriginal controlled organisation, Western Desert Puntukurnuparna Aboriginal Corporation. Formal and informal ethics permissions were provided by these organisations. Information from Martu recorded by Walsh contributed to the successful Martu Native Title Determination.
Data in this paper predominantly derives from this past work and desk top research. We are guided by the core principles of the Martu organisation56 and national ethical guidelines44.
We used multiple methods, sources and media to gather and analyse Indigenous People’s knowledge related to harvester termites and their pavements. This information is fragmented and widely dispersed. Communication with archaeologists, anthropologists, linguists, historians, art specialists and others, including five recorded interviews, has yielded access to primary and secondary sources. We also searched Aboriginal language dictionaries, unpublished and published narratives, photo and film archives and other repositories. Tertiary sources in ethnographies, biographies and more have been scanned. Aboriginal artworks and associated documentation have been particularly informative. Triangulation and cross-checking helped corroborate and clarify details. Further searches would yield more ethnographic information. We mapped the exact or approximate locations of ethnographic records to show the span of termite-related knowledge across language regions (Figure 2a, Figure S2).
We spoke with six Martu people and two Warlpiri – Pintupi people about termites in 2021, with further interviews with Indigenous and other people planned for the future.
From interviews, dictionaries and linguists we accumulated a list of 114 words that span 15 Aboriginal dialects of arid Australia. This list will expand and is available on request. These terms relate to termites and their human ecology, including specifically to harvester termites and pavements. Challenges in identifying records associated with harvester termites or their pavements include the bias of Euro-Australians against termites21, and misidentifications and mis-transcriptions by primary recorders of both termites and pavements57.
Videos and photographs
Photographs and especially videos provide further evidence of Aboriginal uses of pavements. Three videos from the late 1980s show seed threshing directly or in pits excavated into pavements. One is narrated by a Pitjantjatjara woman. Links to these videos can be found at Figure S3. Audio-visual content is a compelling medium for Aboriginal audiences and future project partners, around which knowledge, trust and dialogue can grow.
Compilation and analysis
Original source materials from the ethnography, photos and videos were collated into a sortable spreadsheet. To determine the suite of uses and values attributed to harvester termites, these data were analysed thematically by uses, locations, people and artists associated with pavements, flying termites and more. The locations are shown in Figure 2.
Desert Aboriginal artworks encode deep, rich, layered knowledge, some intertwined with ceremonies, sites and songlines53. Some meanings may be hidden from viewers. In the past, this art has rarely been explored for its ecological content. The Kaapa Tjampijinpa painting (Figure 1b) was the first desert artwork of termites and pavements we found. Art historian John Kean provided documentation for this work in 1976, recorded from Kaapa by historian Dick Kimber. Through such peer-to-peer contacts and a snowballing technique, we interviewed, searched, and found more and more artworks on the topic. One important step was art historian Vivien Johnson directing us to the works of Michael Nelson Jagamara58 who painted flying termites and their pavements as eight of his primary Dreamings (Figure S5).
These paintings and their documentation were compiled into an Excel spreadsheet. There are more artworks to be found. We have not yet investigated flying termite-related icons in rock art19,55.
We repeatedly cross-checked the artworks to our word list and ethnographic records. There are ethical and methodological challenges in interpreting Aboriginal art. For example, Kaapa Tjampijinpa, Michael Jagamara and other men’s paintings were titled ‘Watanuma’, as were paintings by the women Wintjiya Napaltjari and Yuyuya Nampitjinpa. Were they all painting the same place? Why would women be painting a men’s place and vice versa? As many small pieces of information accumulated, we concluded that there are two sites known as Watanuma or synonyms, located approximately 130 km apart and associated with different features of flying termites and with termite pavements. The abundance and mysteries of Aboriginal art related to termites may continue to reveal themselves.
Termite nests can be difficult to find, as 90% of species make subterranean nests, whereas mound-nests which are easier to see being above the ground. In Australia, most harvester termite species have subterranean nests. Some harvester termites construct ‘pavement nests’, so-called because the upper surface of the nest abuts the soil surface and is flat and hard resembling a concrete pavement. Pavement nests are not uniform; they often have varied sizes, shapes and features10,59. Some have small bumps or low mounds positioned anywhere on the pavement. Pavements may lie hidden under centimetres of windblown sand. One colony may build several pavement nests (i.e., they are polycalic), and move between them (i.e., not all pavement nests are occupied simultaneously). This movement between pavement nests is thought to be related to harvesting of grasses around each nest thus foraging holes (leading from the underground tunnels to the soil surface) are not maintained permanently. Termite nests may be colonised by other termite species, ants or other insects, fungi, with or without the original inhabitants. Pavements may persist for decades or more after the termites die or relocate, although termite structures can degrade in a few years after fire or flood. We have observed unoccupied pavements which have been burnt and damaged by hot wildfires and then erosive rain.
In July 14-21, 2021, in the East Pilbara of Western Australia, we surveyed plots on Nyiyaparli country east of Newman airport and near the Jigalong Road turnoff (Figure 2c). Four plot areas were selected at the same locations surveyed by Getzin et al.2,9. Following the first, subsequent subject pavements were selected by nearest neighbour proximity. We recorded latitude and longitude and the north-south and east-west diameter of each pavement. In total, we excavated into 25 pavements in the surrounding spinifex grassland, similar excavations adjacent to 11 of these pavements. The first pavement was used to develop our methods, with results presented from the next 24 pavements (Figure 4). In 16 pavements, three trenches were dug and in the remainder one trench was dug at 50 cm long, 15 cm wide, 15 cm deep. The first trench was dug in the centre of the pavement, and the others toward the edge and on opposite sides of the pavement (Figure S7). In total, 29 m of trench were dug on the pavements, plus four metres in the trial pavement.
We improved on previous methods2,14,3,9 by sampling longer trenches, using better excavation tools including an air blower. Longer trenches revealed more of the pavement substructure, and we used tools that were less likely to shatter the termite structures. These tools were a mattock or crowbar, or an 18V electric power tool with an 8 cm shank blade by contrast to a jack hammer used by Getzin et al.3,9. After excavation, we used an 18V air blower to remove excavation debris and dust which made the termite structures easy to see.
In the initial test pavement, we dug deeper, longer, and more holes (Figure S7), trialling different tools including a hand chisel and a teaspoon. We did not include quantitative results from this pavement in the analysis. From this first pavement we concluded that with the tools and time available it was not possible to dig to the bottom of the consolidated soils on the Newman pavements. Heavier machinery and alternative methods are needed to examine the deep structure of the pavements.
Adjacent to the test pavement in the spinifex grassland, we trialled trenches and pits. Adjacent to 11 pavements we cleared vegetation then excavated three trenches dug at 50 cm long, 15 cm wide, 15 cm deep (as for pavements). We cleaned these with the air blower.
Observations included presence/absence of termite chambers, termite frass chambers, termite chaff and/or termite workers or soldiers. Other observations were recorded as required, and each pavement and trench were photographed. Samples of termites, termite chaff, termite chambers and consolidated soils were collected. A single observer reported their observations to the data recorder for the first 16 pavements. One person did all tasks on the other eight pavements.
All data was written into printed data sheets. Quantitative records were presence/absence of termite chambers, termite frass chambers, termite chaff and/or termite workers or soldiers. Qualitative observations were also recorded for each trench. Each pavement and each trench in each pavement were photographed. Samples of termites and termite chaff were collected and labelled. Samples of termite chambers and consolidated soils were taken from the methodological pavement.
At the end of surveying each plot, all trenches were backfilled. The surfaces were flattened and raked flat, and spinifex returned over the sandplain trenches.
All photographs were labelled with the plot and pavement numbers. Quantitative and qualitative data from the printed data sheets were transcribed to an Excel spreadsheet.
Distribution maps60 indicated both Drepanotermes perniger and D. rubriceps occur near our sites in the east Pilbara and also on Ngalia Warlpiri country, at Australian Wildlife Conservancy’s Newhaven Wildlife Sanctuary (Newhaven). Termites from excavations were collected (Figure S7b). All termites from the Newman pavements were keyed and identified as D. perniger. One other species, probably Shedorhinotermes derosus, was found in shallow chambers under spinifex adjacent to pavement FC2-7 near Newman. Termites from Newhaven pavements were also keyed as D. perniger; again, two Drepanotermes species were possible.
Newhaven pavement reconnaissance excavation
During 1 – 2 May 2021, at Newhaven, we explored features of one pavement. This pavement (1.4 m by 1.2 m diameter) was sprayed with 50 L of water to observe water behaviour on the pavement. One side of the pavement was vertically cut through to examine its internal structure and to collect termites. The surrounding sandy soils were soft enough to dig by hand. Two north – south trenches were dug beneath the pavement then connected by a tunnel under the full pavement. A scale drawing of the excavation was prepared on site (Figure S8c, d). While D. perniger occupied both the Newhaven and the Newman pavements, pavements from these two localities showed very different structures (the Newhaven sample size is small).
Termite pavements are obvious within many unformed vehicle tracks which traverse suitable landforms. Pavements are resistant to road graders, vehicle traffic and water erosion, will stand up to a few centimetres above the track surface. We have incidental observations of pavement occurrences from tracks (Figure 2a). Track-based surveys could be used to determine densities of termite pavements.
Figure 2 records our aerial observations of termite pavement spot patterns using drones, helicopter and Google Earth images across the arid regions of Northern Territory, Western Australia, and South Australia. These states and territory are where we have on-ground experience and familiarity with landforms and ecosystems. Drepanotermes perniger and D. rubriceps each have much wider continental distributions including in Queensland and New South Wales60.
Drones and helicopters were used on four occasions for incidental pavement observations. We also gathered ground and aerial records provided by colleagues familiar with pavements. Systematic methods for aerial survey are required.
Google Earth was used to identify locations with clearly visible pavement spot patterns in spinifex vegetation. Pavement spots, if present and not obscured by sand, are visible at approximate one kilometre altitude. We can distinguish pavement spots from spinifex rings or other circular formations and did not confuse them10. Factors that obviate the clarity and confidence of identification of spots include lower resolution satellite images and the sparseness or absence of spinifex vegetation cover (typically reduced by wildfires). To detect pavement spot patterns, we looked within the prospective land units (spinifex grasslands) for darker areas where denser vegetation made spot patterns more visible (Figure 2). The ‘historical imagery’ function on Google Earth can allow observation prior to fire events which can obscure pavement spot patterns.