Vibrant ecosystems: Of rumbling elephants and seismic wildlife monitoring

Our planet is at unrest. From butterfly wings to rock gigs, typhoons and megathrust earthquakes, mechanical wave disturbances permanently penetrate the Earth system across many orders of magnitude. Such material-bound wavefields carry valuable information over great distances and through otherwise (e.g. optically) often impenetrable substances, perhaps more efficiently than any other physical information-transfer process. This information transfer can help observers to constrain origin and path effects (read: conventional seismology), or it can relate to active broadcasting, eavesdropping, or as a two-way conversation between source and receiver (read: communication). Given the pristine sensitivity of seismic instruments and our understanding of these convolved effects, seismologists have been making great strides in applying their techniques to the Sun’s interior, foul apples, calving glaciers, nuclear bombs, human singing, and urban vibrations such as traffic jams, trains and pandemic mobility.

Direct solution attempt at the inverse problem of subsurface imaging. Or: levelling seismometers. Copyright: Paula Koelemeijer & Tarje Nissen-Meyer

Our journey rattles along with an imposing, intelligent and versatile creature that has probably shaken the planet more than any other species since the last mass extinction: the elephant. More recently (2015), it was an out-of-the-blue email by my zoology collaborator Beth Mortimer entitled “Elephant Seismology” which set our stage for an ongoing collaboration across many disciplines. Awwww, sweet elephants. Easy appeal, right? Good fun, but bad for funding. Compelling on the eye but compounded as prey. Nice for screen caps, but scraping crops. Good for conservation and conversation, not for conventional co-funding. Notwithstomping, it has set us on an undulating journey through the Kenyan savanna, proposals subducted between stiff panel domains, science radio interviews, agonising misinterpretations, accusations and misunderstandings, fun, sticking sharp spades into the savanna sands in the scorching sun, learning from Kenyan colleagues about the pitfalls and benefits of conservation, the atrocious and ongoing legacy of colonisation including parachute science, and a TEDWomen talk in the Californian desert on the back of a local earthquake.

Patience with hard/software in the scorching sun surrounded by the big five. Copyright: Paula Koelemeijer & Tarje Nissen-Meyer

Beyond an astonishingly reliable emotional appeal (e.g., the average “awww-elephant”-reaction versus the public’s stance on (IMHO similarly fascinating) tardigrades or axisymmetric spectral-element methods) lurk fascinating fundamental questions at the intersections of zoology, collective animal behaviour, mammal physiology, biodiversity loss, data fusion, sensor technology, inverse problems, near-surface structure, computational modelling, ambient-noise interferometry, machine learning, and pressing societal topics in the face of the climate crisis and its drastic attributed events (such as the recent drought in Kenya), human-wildlife conflict, sustainable living and land use in the savanna, as well as the again-increasing poaching problem.

Spot the cat! Copyright: Paula Koelemeijer & Tarje Nissen-Meyer

Instead of replicating greater details from papers, I’ll defer the inclined reader to the references regarding more technical aspects. After our 2019 SeisSavanna (downloadable dataset) deployment at Mpala Research Centre in Kenya with 20 seismometers, 8 acoustic sensors, 30 camera traps, and complementary data (GPS tracking, weather), here’s roughly where we (read: our team. More experienced experts, e.g. in indigenous communities likely know a lot more about elephant behaviour) stand after learning just a glimpse over the past few years (that’s seven orders of magnitude less time to train compared to elephants’ collective 24/7 field data gathered over 50 million years):

• There’s potential for elephant-generated seismic signals to propagate over vast distances (kilometres) above the noise level, but highly dependent on local geology, type of emitted signal (running versus walking, rumbling), and the noise environment. Often, their gentle walk remains aseismic, and we don’t know yet to what extent this is being used actively.
• Seismic signals are often discernible regarding their source characteristics (location, type of signal), even over great distances.
• It’s likely (in the sense that they can use this as complementary information, so why shouldn’t they?) that elephants use seismic vibrations to broadcast and gather information, but complementary to airborne conversations. Elephant ears ain’t for nothin’. Much like their sensitive feet.
• We can seismically localise elephant rumbles with great accuracy (meters) from hundreds of meters in distance with conventional triangulation. Strikingly, for our dataset, seismic localisation works largely better than acoustic localisation. Is this a hint as to how elephants balance their use of different senses?
• Seismic waveforms are sufficient for deep neural networks, which have been trained using labels from camera-trap images, to identify species (elephant versus giraffe) and characterise their signal (walking versus running, rumbling). Corresponding datasets and ML scripts published with the paper.
• Our analysis suggests a tangible pathway towards seismic monitoring, with advantages over conventional approaches such as: fewer line-of-sight problems, azimuthally isotropic capture of signals (no cone as for cameras/drones), (largely) independent of vegetation, topography and weather, cheap, relatively small but information-rich data. Do elephants seek to exploit signal-to-noise ratios in their balance of communication mode, depending on local conditions?

Many questions remain and appear. Over what distances are signals in our dataset traceable? Can we decipher a two-way communication via any one mode, or cross-mode (acoustic question, seismic answer)? Can trained networks be applied to different environments? Can we causally link external signals (natural environment, weather, cattle herds) to elephants’ reactions? What is the elephant feet’s seismic efficiency? Can a dense network of lightweight instruments become a prototype for continuous, real-time monitoring at scale? Can we distinguish biological and other constituents within the noise environment? What is the influence of local geology and geophysics on vibrational communication and behaviour? Can we link seasonal, climatic factors to the vigour of the ecosystem’s vibroscape?

Local solution for instrument safety. Copyright: Paula Koelemeijer & Tarje Nissen-Meyer

So, what’s next for us to address these? In all honesty…. the elephant in the (panel) room has been the (rolling, not water) buck. Beneath that looms a seemingly light-hearted perception of such studies that transcend academia’s self-imposed disciplinary boundaries — boundaries that unfortunately segregate our planet’s interconnected systems along the confines of tertiary education programmes. Nevertheless, after many previous attempts and months spent on submerging our heads in overhead figures, we are grateful to have received support to deploy an autonomous dense, longer-term array in 2023. This will facilitate a proto-monitoring system in proto-realtime for proto-conservation – hopefully pronto. Overcoming some not insignificant obstacles (pandemic, drought, local elections and geopolitical stability, further funding), do stay tuned or get in touch if interested in collaborative deployments and support over the coming years.

We suggest monitoring techniques should strive to maximise the following factors for instrumentation: non-invasive, implementable, high-grade, analysable, (culturally) contextual. A tall order, but seismic node instrumentation is promising on those accounts, potentially opening a path (if multi-scattered and diffusive) to learn, alongside local communities, about the interplay between megafauna, other species and everchanging environmental conditions.

Not only akin to Copernicus’ universal revelations (pun EGUrly intended), our perception as to how central we place our human existence and communication skills within the vast natural, complex environment is, and this may warm seismologists’ core organs even more, somewhat inversely proportional to the amount of informative data and theory available: The more data-informed evidence, the less central and special we find ourselves to be in our world. If we recorded the entire 4D multi-scale vibroscape, elephant only knows how much information exchange and communication there might be bopping about the underworld. As our blip (direct P-arrival, coda?) of human history and knowledge has shown, expect the unexpected in changing perspectives. What is the role of vibrational senses in active and subconscious information gathering, communication, intra- and inter-species? How deeply learned and evolved is elephants’ AI (animal intelligence) over 50 million years to sense hitherto unknown signals for natural hazards over multi-scale distances? There’s been a few million eight-tons of training data! Can they triangulate and characterise vibrational events, or detect the water table’s hyper-surface by rumbling – seismic 3D imaging for mere survival by (reverse-time) migration? Are we still entirely poking around in the dark regarding a stealth mechanical mode of interspecies communication, or a well-developed solid-medium languages within the animal world? Elephants, dik-diks, pythons and African bees seismically arranging their time slots at the water dam? Cattle herds jointly triangulating a roaming lion in real-time? Disparate elephant family members exchanging advances over a seismic dating platform? Interspecies group psychoseismolinguistics? Without more data and insight, keeping an open mind, ear to and our feet on the ground, the answer to all these questions is not: No. We know that we don’t know – no matter how densely populated our prior.

Who gets the best slot on the shore? It’s not a holiday in the Sun. Copyright: Paula Koelemeijer & Tarje Nissen-Meyer

For an inquisitive species (ourselves as much as other animals), there is much scope to better understand the vibroscape of our complex life-support systems in a non-invasive, sustainable manner. As our human world encroaches upon every last glimpse of rock, flora and fauna, the planet gets ever smaller – we have to listen to all voices and cracks in our environment to protect it, to undo anthropogenic devastation as much as human starvation, especially in the dire face of the greatest crisis facing life on Earth ever since the Jurassic: potentially tumbling over the tipping points of biodiversity loss, climate heating, and large-scale extinction. The elephant is calling, and we must go (non-invasively, to listen and act to protect them and ourselves).

Teamwork and collaboration! No seismology book contains advice on machetes, protecting instruments with cactus, understanding elephant mock charges, effectively digging into savanna sands, and avoiding buffalos and their teasing tricks if hiding in a tree. Copyright: Paula Koelemeijer & Tarje Nissen-Meyer

Further Reading:

Van Wijk, 2017. Apple Seismology, Physics Today 70, 10, 94; https://doi.org/10.1063/PT.3.3740

Scherbaum, F, W Loos, D Vollmer, F Kane. 2015. Body vibrations as source of information for the analysis of polyphonic vocal music, Proceedings of the 5th International Workshop on Folk Music Analysis, June 10-12, 5, 89-93.

Szenicer, Moseley, Reinwald, Nissen-Meyer, Mortimer, 2021. Seismic savanna: Machine learning for classifying wildlife and behaviours using ground-based vibration field recordings, Remote Sensing for Ecology and Conservation, DOI: 10.1002/rse2.242.

Reinwald, Moseley, Szenicer, Nissen-Meyer, Oduor, Markham, Vollrath, Mortimer, 2021. Seismic localisation of elephant rumbles as a monitoring approach. Journal of the Royal Society Interface 18 (180), 20210264

Mortimer, Rees, Koelemeijer, T Nissen-Meyer, 2018. Classifying elephant behaviour through seismic vibrations, Current Biology, Vol 28, 9, PR547-R548.

L Pinzon‐Rincon, F Lavoué, A Mordret, P Boué, F Brenguier, P Dales, Y Ben‐Zion, F Vernon, C J. Bean, D Hollis, 2021. Humming Trains in Seismology: An Opportune Source for Probing the Shallow Crust. Seismological Research Letters 2021;; 92 (2A): 623–635. doi: https://doi.org/10.1785/0220200248

Ceballos, G., Ehrlich, P. R. & Raven, P. H. (2020). Vertebrates on the brink as indicators of biological annihilation and the sixth mass extinction. Proceedings of the National Academy of Sciences of the United States of America 117(24), 13596– 13602.

Gibb, R., Browning, E., Glover‐Kapfer, P. and Jones, K.E., 2019. Emerging opportunities and challenges for passive acoustics in ecological assessment and monitoring. Methods in Ecology and Evolution, 10(2), pp.169-185.

Lecocq et al, 2020. Global quieting of high-frequency seismic noise due to COVID-19 pandemic lockdown measures, Science, 369, 1338-1343. 10.1126/science.abd2438.

Malhi, Y, T Lander, E le Roux, N Stevens, M Macias-Fauria, L Wedding, C Girardin, et al. 2022. The Role of Large Wild Animals in Climate Change Mitigation and Adaptation. Current Biology 32 (4): R181–R196.

O’Connell-Rodwell, 2007. Keeping an “Ear” to the Ground: Seismic Communication in Elephants, Physiology, 22,4, 287-294

Odada, E.O., Olago, D.O. and Olaka, L.A., 2020. An East African perspective of the anthropocene. Scientific African, 10, p.e00553.

Parihar, D.S., Ghosh, R., Akula, A., Kumar, S. and Sardana, H.K., 2021. Seismic signal analysis for the characterisation of elephant movements in a forest environment. Ecological Informatics, 64, p.101329

Prosekov, A., Kuznetsov, A., Rada, A. and Ivanova, S., 2020. Methods for monitoring large terrestrial animals in the wild. Forests, 11(8), p.808.

Šturm, R, J. J. López Díez, J. Polajnar, J. Sueur, M. Virant-Doberlet, 2022. Is It Time for Ecotremology? DOI:10.3389/fevo.2022.828503

Featured Image (Top): There’s more than one elephant, and we’re in their room. A seismic near-source of the menacing character. Copyright: Paula Koelemeijer & Tarje Nissen-Meyer