Much of our world is in hiding. It lies below the surface of the oceans and is largely invisible to us, a Maria Incognito. It is a hostile world: we are creatures of the solid ground, who long left the seas behind and cannot return. But more of this forbidden world is being uncovered, by sonar and the occasional human submarine visitor. Gravity maps have revealed the ocean floor in places where ship-based surveys have been sparse. These gravity maps use the height of the ocean surface to determine how much mass there is below. Obviously this does not work on land, and therefore these maps show empty space where the land is, a true Terra Incognito. In contrast, the oceans come to life, showing mountains, plains and chains of (ex-)volcanoes – all of it deep under water, a true Titanic world. (In Greek mythology, the Titans were the children of Earth and Sky – Gaia and Uranus – who, like their namesake the Titanic, became banished to the abyss.) Spreading ridges and transform faults are visible. Many parts of this new world remain next to impossible to explore directly, too deep for all but a few very carefully designed submarines – as we were disastrously reminded by the demise of the Titan submarine. But other parts of this world are adjacent to known islands and clearly are submerged extensions to those, larger structures of which the visible island is just the very top: a submerged part of our world of which we were unaware.
When exploring those maps, one such large underwater plateau shows up in the far southern Indian Ocean. It carries an island on top that looks like an inverted version of Iceland. A quick search shows that the island is already discovered (by Kerguelen – twice), named (by Cook – twice), owned (once, by France), and occupied (by a colony of itinerant scientists). The main island is surrounded by an archipelago of perhaps 300 much smaller islands. I am not sure whether anyone has actually counted them! The archipelago is called the Kerguelen Islands (Cook’s second name was Desolation Island, a name still adopted in French). Kerguelen is supposed to be the most isolated piece of occupied real estate in the world. There is no airport; it can be reached only by a week-long boat journey across a storm-swept ocean. Questions arise – why is it there? Why the underwater plateau? And why does it look like an Iceland of the South?
A direct comparison shows that Kerguelen island is rather smaller than Iceland. Adding the surrounding submerged plateau makes it more comparable but still smaller. However, there is much more to the underworld area. Kerguelen forms one end of a much longer, but almost entirely submarine, volcanic plateau, stretching southeastward. The full underwater plateau is shown below (image credit: Australian government). It is over 2000 km long, covers an area of about 1.3 million square kilometres (four times the size of the British Isles, one sixth of Australia) — and it was created by volcanic eruptions. This is a super-Iceland! It consists of three parts, the northern plateau with Kerguelen itself, a central plateau with two small islands (Heard island and McDonald island) and a southern plateau which is fully submerged. Elan Bank, to the side, may also be part of the large plateau. Further south, a trough separates it from Antarctica.
All together, the volcanic plateau is the largest one in the Indian Ocean, and larger even than the Deccan traps! The size qualifies it as a large igneous province (LIP): by definition, this is an eruption covering an area more than 100,000 square kilometres. Eruptions this size are often attributed to hot plumes melting the lithosphere below the eruption site. The head of such a plume can extend over 1000 km. The ‘Kerguelen mantle plume’ is a plume of legend. Kerguelen island is a harbinger of past doom. For all its remote inconspicuousness, this frozen island with its submarine plateau once changed the world. Australia is still living with the consequences.
Kerguelen Island and the surrounding northern plateau are volcanically dormant or dead. So is the southern plateau. But the central area still has life in it. Both Heard Island and McDonald Island are active. They are in fact the only active volcanoes owned by Australia (a gift from the UK).
Heard island is 2800 meters tall (it has grown a bit since wikipedia) making it the tallest Australian mountain outside of Antarctica. You would be lucky to see it though: cloud cover can seem near perpetual. On a rare good day, the mountain shows a shape which is a bit like El Teide, with a pyramid cone sitting on a much larger summit. Rather more common is that it is covered in cloud. The island is called Heard Island, the volcano is strangely named ‘Big Ben’ (occupying pretty much all of the island), and the erupting cone is Mawson Peak. It is among the most frequently active volcanoes in the world, probably erupting a lava flow every few months. The most recent one was 25 May 2023, noticed by no one but captured by the Sentinel satellite.
The island experienced devastating seal hunting in the 19th century. As one document describes: “The plain (Atlas Cove) was strewn with bones of the Sea Elephant and Sea Leopard, those of the former being most abundant. There were remains of thousands of skeletons, and I gathered a good many tusks of old-males. The bones lay in curved lines, looking like tide lines, on either side of the plain above the beaches, marking the rookeries of old times and tracks of slaughter of the sealers.” That was the past. Nowadays Heard Island seems little touched by humanity, being uninhabitable and inaccessible. Even the ubiquitous rats are absent, leaving the sea birds to nest in peace (apart from the regular volcanic eruptions, or course). There are also no introduced plants. Apparently an attempt was made to introduce sheep during the Australian National Antarctic Research Expedition from 1947 to 1954 (Australians don’t feel at home without them, apparently) but they did not survive. Sledging dogs were also brought in and did well but served no useful purpose so they were taken to Antarctica. This leaves a recovering but authentic subantarctic ecosystem. The eastern spit is apparently very popular with elephant seals. Leopard seals are still found here in winter (beware) and the 2 million macaroni penguins are as smelly as you can imagine.McDonald Island is nearby (30 km) but very different. While Heard island is covered in glaciers, McDonald is just rock. And while Heard erupts frequent and effusive, McDonald erupts rarely but explosively. After 75,000 years of dormancy (call it a mid-winter nap) it became active in 1992 and exploded in 1996. A green island was turned into a volcanic wasteland. Since that time there have been several further eruptions, which build different cones. The island has grown so much (doubling in size) that
nearby Flat island ( a descriptive name if ever there was one) has become incorporated in it.
Kerguelen Island is very different from these two volcanic islands. It is much larger, at 150 km across. A description of the island can be found here. The picture at the top of this post shows a scene from the island. The coast is deeply indented with bays; the land is mountainous, rising to almost 2 km. The island shows multiple horizontal layers of basaltic lava, forming a plateau which covers much of the interior of the island. It is deeply eroded and covered by a tundra vegetation. A bonus to people is the presence of edible plants, in particular a local cabbage which provides vitamin-C. The climate is like Iceland’s north coast. Around 4000 reindeer are present, showing the influence of people. So do the wild cats and rabbits! In spite of its isolation, this land is a certified part of the human ecosphere. The reindeer were originally brought to a nearby island by Norwegian whalers (Norwegians don’t feel at home without them, apparently), but they (the reindeer, not the Norwegians) swam their way to Kerguelen freedom. They are the only remaining southern-hemisphere reindeer, after the population at South Georgia was removed. The rabbits came from South Africa in the 19th century and were meant to provide food to the whalers. They are now a major problem especially in the eastern part of Kerguelen. Sheep were a 20th century addition. The survival of feral cats in this tundra world came as a surprise. They descended from cats kept on ships who managed to get to land (probably with human help as they are not noted for swimming). The cats managed to live on rabbits and sea birds.
After the whalers and the sealers had left, an itineracy of scientists came. There are now some 100 people here, living mainly in one town and each staying for a limited time. Tourism is next to non-existent.
In addition to the central basaltic plateau, close to the coast are several younger volcanic complexes. In the southwest is a series of ring dykes, emplaced underground but visible now because of erosion. They are as young as 5 million years. In the same region are remnants of explosive volcanism: a large caldera and pyroclastic deposits. These may be only 1 million years old and include the last major explosion on Kerguelen. Ross volcano in the south similarly is 1-2 million years old. All these volcanic complexes are extinct. But the inaccessible southwestern peninsula still shows active fumaroles as well as hot water basins reaching 80C. It would be a popular place to visit, had it not been so unapproachable that helicopters are needed for a visit. The suggestion has been made that a hydrothermal system is located underneath the icecap. The heat cannot easily escape from there, and so instead the hot water below the ice feeds the activity beyond the edges of the glacier.
The hydrothermal activity may provide a clue as to why Kerguelen Island is an island. Why is it the only part of the submerged plateau where a large block has been pushed up to above sea level? Other than this, the only islands here are active volcanoes, build up by young lava. To get the old crust to rise up requires a buoyant force below. There must be a heat source below the crust which has lowered the density and thus pushed up the land. The fact that the rest of the plateau is submerged suggests there is less or no heat elsewhere: the warmth is located only underneath Kerguelen Island and nowhere else. But it is currently too weak to produce eruptible magma. The explosive activity a million years ago may have been the last straw of a cooling magma.
At one time, coal mining was attempted at Kerguelen. The very attempt shows there is much more to Kerguelen than volcanics. Coal takes a long time to form: for Kerguelen to have coal, it must be old, far older than a volcanic island should be. This is no Hawai’i, not even a Hawai’i in decline: it has a much more ancient history. (Update: The fossil wood is found sandwiched below the upper basaltic layer and the basement, and was noted already in the early 1800’s during a visit to the northern shores of Kerguelen Island. See below for a description from Geology Today, 2012, Falcon-Lang)
That long history is true for the rest of the underwater plateau as well. Drilling has discovered fossil wood and charcoal in the central and southern plateaus. It shows that at times, part of the plateau, now 2 km deep, was above water and was covered in forests. At that time the plateau was an island the size of Great Britain.
Clearly this is no ordinary ocean floor! Indeed, the Kerguelen plateau is considered a micro-continent, but one that has fallen on hard times and is keeping its head above water only with great difficulty and some help from below. But you can never stop being a continent, even when submerged by 2 km. Continents are the world’s great gift: they are forever.
If this is a micro-continent, where did it come from? They are normally offspring of a bigger continent. We can see the process in Africa: a rift forms, which splits a continent in two quite unequal parts. The African Rift is currently working on dislodging Somalia and sending it into the Indian Ocean. That would create a microcontinent. The Seychelles are a microcontinent which split from India. Micro-continents are geological cast-offs caused by off-centre rifting. But Kerguelen is surrounded by four different continents: Africa, Antarctica, Asia and Australia, in their battle for alphabetical supremacy. So which one is the parent?
And is all of Kerguelen a micro-continent? We have previously seen that the Seychelles are a combination of a micro-continent and a volcanic addition: part has continental crust underneath, but another part is build on oceanic crust. It turns out, this is true for Kerguelen as well. It was not easy to determine. The whole area is covered in kilometers of basalt which gives little clue to what lies beneath. Seismic studies have shown that in the crust underlying the southern plateau, earthquake waves travel at seismic velocities which are typical for continental crust. But in the north, underneath Kerguelen itself, that layer is not seen and here the crust appears to be oceanic. It therefore appears that the microcontinent is the southern part and that the north is a later addition without any continental affinity. The main area that sticks out from the sea is therefore oceanic, and the deepest submerged parts are continental!How old are the different parts? The basaltic lavas drilled up across the plateau have been dated using radioactive decay. This has shown a very large range of dates. The oldest dates are found in the southern plateau, and are all around 115 million years ago. Towards the north the ages rapidly become younger, reaching 35 million year around Kerguelen itself. The volcanism on this plateau has been remarkably long lived. One could even argue that it continues to this day, at Heard and McDonald Islands, and until a million years ago on Kerguelen Island itself.
The continental part of Kerguelen had volcanism in two phases: between 110 and 120 million years ago in the southern part and between 90 and 100 million years ago in the central part. The oceanic extension in the north is much younger. Kerguelen island itself is 25-30 million years old, with a minor volcanic resurgence 10 million years ago which over time migrated south to Heard Island. The gaps in time are so large that the various phases of volcanism probably have different origins. This does not look like a very long-lived hot spot which miraculously stayed underneath Kerguelen, moving with the ocean floor the whole time. Were there several mantle plumes impacting Kerguelen at different times? Or was it something else?
The answers to these are found elsewhere. We need to look around the Indian Ocean.
The Indian Ocean
The map below shows the location of Kerguelen in the Indian Ocean (click to see full detail). It is not the only feature in the ocean. Spreading ridges separate the main plates. Faintly visible is the SouthEast Indian Ridge, the ocean spreading centre between Kerguelen and Australia (the transform faults are better visible than the spreading ridge). Much clearer is the spreading ridge towards Africa. A straight line extending almost to Myanmar is called the Ninety East Ridge and looks very much like a hotspot trail. It points directly at central Kerguelen. To the right of its southern end is the Broken Ridge. Broken Ridge is a missing sliver of the central Kerguelen Plateau, separated from it by virtue of becoming stuck on the wrong plate. A spreading ridge is located midway between the two – microcontinents can apparently split and produce nanocontinents. To quote Jonathan Swift,
Naturalists observe, a flea
Has smaller fleas that on him prey,
And these have smaller yet to bite ’em,
And so proceed, ad infinitum
And so supercontinents form continents which shed microcontinents and those split off nano-continents. The smallest, too small to survive, perhaps become the in-continents.
The ages confirm that this all this oceanic debris is linked. The Broken Ridge is dated to 95 million years ago, whilst the Ninety East Ridge ranges from 83 million years at its northern end to 37 million years in the south. Further afield, fossil dykes and lava flows in India, at the Bay of Bengal, are 115 million years old. So are fossil dykes in Antarctica, on the far side of Kerguelen.
Let’s first match up the younger volcanics. The tip of the Ninety East Ridge nearest to Kerguelen Island has the same age as the Island. The spreading ridge is midway between them. Clearly, Kerguelen Island belongs to Ninety East Ridge. It is the youngest part of this ridge, and came from a lava flood which occured on or near the spreading ridge. Kerguelen Island really did form like Iceland. The rest of Ninety East Ridge formed earlier, over the 50 million period before this event. It came from the same hot spot and was carried away by the moving Indian ocean plate. The Ninety East Ridge is one-sided, showing that the hot spot was not on the spreading ridge for that time, but was located on the Indian Ocean plate. It ended up on the spreading centre only during the final lava flood when it also jumped to the other plate. The spreading centre that had been further south shifted northward, to the hot spot, and separated it from its tail.
The hot spot did not end its existence with the formation of Kerguelen Island. It had lost much of its vigour, but it remained on the spreading ridge. There are two other volcanic islands at its current position. These are the islands of Amsterdam and St Paul, with a series of nearby submarine volcanoes which are strangely named Chains of the Dead Poets. (Indeed, they carry the names of dead poets.) These two islands are at the very centre of the Indian Ocean, at the central point between Antarctica, Australia and Madagascar. It is a telling location.
We need to go back further in time to even deader poets. The older Kerguelen volcanics comes in two groups, at 95 and at 115 million years old. The group at around 95 million years old shows up only in central Kerguelen and in Broken Ridge. These were clearly a single entity at the time. The separation from the spreading ridge is similar to or just a tad more than that of the Ninety East Ridge. This shows that the separation occurred at about the time of the younger volcanics (37 million years ago) or perhaps a few million years before, a long time after the eruptions here had ended. This was the time the spreading ridge jumped northward, and it cut right along the edge of the (probably already submerged) central Kerguelen plateau. A small bit of the plateau found itself on the wrong side, started to be carried away and became the Broken Ridge – an appropriate name.
The spreading centre itself was much older. Nowadays it is located between Australia and Antarctica. The spreading began in earnest 85 million years ago, as indicated by the oldest ages seen at the furthest side of the Ninety East Ridge. What had happened between 85 and 37 million years ago? The long line of the Ninety East Ridge shows that the ocean floor was moving north. This was the same time that India was crossing the ocean and eventually slammed into Asia; the ocean floor around Ninety East Ridge was moving with India. From the length of the Ninety East Ridge and its age spread, the speed of the plate was around 10 cm/yr. The formation of the Ninety East Ridge continued for all this time, until it culminated 37 million years ago with the formation of the northern Kerguelen plateau. This was the time when India had collided with Asia, eradicating the subduction zone between them. This stopped the pull on the oceanic plate, and the spreading ridge slowed down markedly. Perhaps this is why the northern plateau grew so large: for a while, the plate remained near stationary above the magma producing area.
Is the hot spot that formed the central Kerguelen plateau 95 million years ago the same one that formed the Ninety East Ridge? It seems so. The plateau is attached to Antarctica and this continent is very reluctant to move. The plateau has perhaps drifted a bit south-southeast but did not go far. The spreading centre was located such that during much of this time the Kerguelen central plateau was on one side (Antarctica) but the nearby hot spot trail was on the other side.
Interestingly, there were other changes happening 85 million years ago. This is the time when Australia began to separate from Antarctica. Did this separation start with the arrival on the surface of the hot spot that formed the central Kerguelen plateau? Or are we too far from Australia for this?
We need to make another step back in time. The earlier burst of volcanic eruptions occurred 115 million years ago. This formed the southern Kerguelen plateau. The same event is also seen in dykes in nearby Antarctica which formed at the same time. The Rajmahal traps in north east India and Bangladesh also formed at the same time. All these areas were closely connected 115 million years ago. India, of course, at this time was part of eastern Gondwana, sitting between Africa and Antarctica. The current long-distance relations came from the rifting of this part of Gondwana.
But this was still not the beginning. 150 million years ago, this was the heart of eastern Gondwana, when Australia, Antarctica and India were all connected and Africa was beginning to separate. To the north of eastern Gondwana was the Tethys ocean; beyond that, Asia was beginning to take shape. Of course India, Bangladesh and the Himalayan countries weren’t there yet. They were still part of Gondwana, migrants still waiting for the chance to make the crossing.
They waited 15 million more years, while the Jurassic gave away to the Cretaceous and far away the Atlantic Ocean began to form. A minor extinction event, called the Tithonian extinction, heralded the incoming Cretaceous. It is named after a prince of Troy, another lost world. The cause of this extinction is not known. But the world was being redrawn and a new age dawned. 135 million years ago, a flood basalt burst out in what is now southwestern Australia. The remnants are seen in Bunbury, but also very far away in southeastern Tibet. Obviously, these two places were in the same location at the time, in the heart of eastern Gondwana and far from any ocean. India was larger than it is now: some of its area would later be lost in the collision with Asia. The Bunbury basalt erupted where southwest Australia joined with northern Greater India. It was the beginning of the end: eastern Gondwana was beginning to break up in its very heartland.
The eruption subsided but it was too late for the continent: Gondwana had broken. But 20 million years lava flows restarted in what is now the southern Kerguelen platform, at the time south or west (500 or 1000 km away?) of Bunbury. And 20 million years after that, the eruption deja vu’d again in what is now the central Kerguelen platform. All three events are often considered part of the Kerguelen flood basalt, but they can also be seen as three separate, unconnected eruptions. A comparison in our own world is provided by the African rift, which started with a flood basalt 30 million years ago and has migrated over that time thousands of kilometers southward. There is now a hot region below eastern Africa, but is this the same hot spot of 30 million years ago? And is elevated Zimbabwe the same hot spot or a different one? Do hot spots perhaps come in families and generations?
The three Gondwana eruptions first caused a rift between Australia and Greater India, and later between India and Antarctica and between Australia and Antarctica. It started with a triple point where the three different branches cut the links between three continents. A slight jump of the rifts separated the southern and central Kerguelen platform from their continents. The question which continent they came from is difficult to answer as the three were one. Based on age and location, one can argue that Kerguelen is a lost piece of Antarctica, but India could also put in a reasonable claim. The Broken Ridge still forms a line that points towards Bunbury, showing that although Kerguelen/Broken Ridge were not part of current Australia, they were on the rift that extended from it.
After the end, the hot spot remained active but it was now under a fast moving plate which caused it to form a long, straight trail of volcanoes: the Ninety East Ridge. India was moving north, crossing the Tethys ocean and carrying parts of the first two flood basalts with it. One ended up near Lhasa and one near Bangladesh – their separation shows those two events occurred at least 500 km apart – possibly much more because of the compression of India. Over 20 million years, that amounts to only a few centimeters per year, so it may be that it was not the volcanics that was moving, but the newly created plates.
Finally, 37 million years ago, when India had collided with Asia, the Kerguelen region again erupted in flames. Was this a new flood or was it just because the plate had stopped moving? At the same time, Australia accelerated in its northward motion – there was a re-organisation off the subduction zones in the Indian and Pacific Oceans (still connected at the time). The fire created a new extension to Kerguelen, but this time build on oceanic crust. After that, Kerguelen found itself moving away from the speading ridge and the volcanics subsided. Did the collision of India affect the magma supply here, far in the Indian Ocean? Perhaps the hot spot had long gone, and the volcanics came from a crustal weakness together with the spreading, allowing mantle material to reach for the surface. Or perhaps there still is residual heat here dating back to Gondwana. Either way, some volcanics continued at the spreading ridge.
There was one more chapter to the story: why did Kerguelen island remain above, or re-emerge from, the sea? Once it left the hot spot behind, it would have sunk below the waves. There was another volcanic episode 10 million years ago, at a time when it was already far from the spreading ridge. Weak activity continued afterwards. It is not clear where that re-activation came from – a sideways convection current from the spreading ridge, perhaps, stopped by the thick crust of the Kerguelen plateau? Or was it related to the beginning of the collision between Australia and Southeast Asia, east of the Wallace line? That seems more likely: the subduction pattern again changed, the pull on the Indian Ocean changed and this reactivated the weakness still present near the edge of the Kerguelen plateau. The map shows a trough. Stress does not like edges.
And that is the story of Kerguelen. It holds memories of a very different world when the southern hemisphere was united. Eastern Gondwana broke down because of massive eruptions which happened over tens of millions of years. The rifting gave us the Indian ocean – but it also left us debris all over the ocean. Kerguelen has drifted away from the original centre of the action – that is now at Amsterdam/St Paul. A little volcanic activity is still continuing but it is no longer attached to the hot spot to which it gave its name. The world has moved on. But we should look at Kerguelen, and remember. It is an elegy written in lava. Gondwana, in memoriam.
Albert, August 2023
Paul J. Wallace et al: Origin and Evolution of the Kerguelen Plateau, Broken Ridge and Kerguelen Archipelago: Editorial, Journal of Petrology, Volume 43, Issue 7, July 2002, Pages 1105–1108
Regis Ballestracci, Jacques Nougier:
Detection by infrared thermography and modelling of an icecapped geothermal system in Kerguelen archipelago, Journal of Volcanology and Geothermal Research, Volume 20, Issues 1–2, 1984, Pages 85-99
1. Leg 183 Summary: Kerguelen Plateau-Broken Ridge—A Large Igneous Province: http://www-odp.tamu.edu/publications/183_IR/chap_01/c1_1.htm
Sheng-Sheng Chen and others: The Tethyan Himalaya Igneous Province: Early Melting Products of the Kerguelen Mantle Plume, Journal of Petrology, Volume 62, Issue 11, November 2021
Francine Bénard, Jean-Paul Callot, Roland Vially, Julien Schmitz, Walter Roest, Martin Patriat, Benoît Loubrieu:
The Kerguelen plateau: Records from a long-living/composite microcontinent,
Marine and Petroleum Geology, Volume 27, 2010, Pages 633-649
Rajesh K. Srivastava: Early Cretaceous alkaline/ultra-alkaline silicate and carbonatite magmatism in the Indian Shield – a review: implications for a possible remnant of the Greater Kerguelen Large Igneous Province, Episodes, Volume 43 (2020), Pages 300-311 (https://doi.org/10.18814/epiiugs/2020/020017)