The largest eruption of the 21st century happened at sea. And it wasn’t Hunga Tonga.
The Comoro islands form a forgotten archipelago, somewhere between Africa and Madagascar with about 200 kilometer of open ocean on either side. On the African side the nearest land is the northern end of Mozambique, close to the border with Tanzania. On the other side it is the northern tip of Madagascar. The closest I have been myself is on the border of Malawi and Mozambique. The scars of the civil war were clearly visible on the Mozambique side, with houses pockmarked with bullet holes, whilst Malawi was in the midst of the AIDS epidemic with fresh graves an all too common sight. Sometimes this brightest and most beautiful of continents seemed a dark place. Madagascar is a different world from Africa, settled from the far side of the Indian ocean, but it too is among the poorest countries in the world. The Comoro Islands are in between, neither Africa nor Madagascar. They were settled by a diverse mixture of people from Southeast Asia (like Madagascar), Africa (perhaps with a slavery heritage) and Arabia. The islands are now densely populated but there are few resources and poverty is widespread. It is a struggle to grow enough food for everyone. Life can be hard even in beautiful places.
There are four main islands. Three are part of the Union of the Comoros, an independent nation consisting of Grande Comore, Mohéli and Anjouan. Together they form one of the smallest countries on Earth. There are local names for the islands: Grande Comore is called Ngazidja, Mohéli is Mwali and Anjouan is Ndzwani. The fourth island is politically separated: Mayotte voted against independence and remains administered by France. Perhaps wisely so: the Comoros are known for the large number of coups (two of which were in the first 6 months of its independence) and for political instability. At one point two of the islands tried to rejoin France but this was refused by France. Since 2010 the political situation has become more stable. The name ‘Comoros’ is sometimes used for the four islands together, and sometimes for the Union of the Comoros.
The islands are volcanic in nature. We will come back to this after the evolutionary interlude.
The region has suffered its share of geological upheaval. Madagascar originally came from Africa. (It actually had a separate existence in an earlier life – we may come back to that.) It split off from Africa during the break-up of Gondwana. The split-off was far larger than Madagascar is now: the entirety of India was attached to the west side – and India itself was larger than it is now. The rift developed along what is now the coast of Tanzania. Other rifts were active elsewhere. Gondwana was breaking up, but the break was not a clean one. Subduction zones around the edges of Gondwana were pulling the supercontinent apart, looking for weaknesses. Many rifts tried to form, some successfully, some quickly superseded by others. Some regions drifted off only to be stopped in their tracks as the rift behind them failed. Others kept moving: Antarctica severed its connection to India at this time.
After the rifting, the Madagascar-India fragment (pretty much a continent in its own right) drifted southward and away from Africa. But the rifting failed to progress and Madagascar was left stranded. A new rift developed on the east side which split off India from Madagascar. The process was very similar to the opening of the North Atlantic, much later, where at first Greenland split from Canada but was left behind when a new rift opened on the other side. The tip of Greenland and the tip of India still show the same angled shape, a meeting point of rifts. Madagascar was left with a straight edge on the eastern side, formed by this new rift.
India now went its own way; the learner-driver would soon smash into Asia. Madagascar was left behind, powerless but safe. India in fact suffered another accident as well. It hit a hot spot which carved a straight line along the western side, removing a bit of the continent. The cut-off moved away to become the Seychelles, a microcontinent stranded in the middle of the Indian ocean. The Seychelles have grown since: it joined forced with the trail of volcanic islands left by the hot spot, stretching as far as Reunion. It is now known as the Mascarene plateau, the second largest oceanic plateau in the Indian ocean after Kerguelen. But only the northern part of this is continental crust. The rest is oceanic.
While the Deccan hot spot carved its line into the Indian ocean, the spreading rift jumped east again.
The African plate immediately reclaimed ownership of the western side of the new rift. Madagascar had already rejoined the African plate – now the Seychelles did too. The African plate became the proud owner of the oceanic crust of the western Indian Ocean, and of the two continental fragments of Gondwana it contains.
India, Madagascar and Mozambique have an ancient continental crust. All of these continents and microcontinents are remnants of Gondwana. Of course they are more than that: they existed before Gondwana and they will still exists when the Indian ocean is long gone. Continental crust outlasts the continents they were part of.
Geology wasn’t finished yet with Africa. 30 million years ago, the Afar hot spot separated Arabia from Africa. Since that time, some of the rifts left by the Gondwana fragmentation have re-activated. This has formed the East Africa Rift, which is in fact a combination of different rift fragments. One branch has extended off shore where it formed the Kerimbas graben. Further south lies the Davies Ridge, a remnant of the transform fault along which Madagascar and India moved south. In the future the newly activated Kerimbas graben may link up with this system. There are sea mounts here: the appropriately named St Lazarus, as well Paisly and the Sakalaves. The off-shore rift is becoming volcanic.
Unlike the trio of Madagascar, the Seychelles and Eastern Africa, the Comoros are not related to Gondwana. They are solidly part of the oceanic kingdom. The islands are sitting in a 3-km deep ocean, and in common with all deep-sea islands, are volcanic in nature. If you imagine the sea away, what is left are four steep, tall, isolated mountains sitting on a deep plain. There could be no doubt that these are volcanoes. There is no continental crust here: the Comoros archipelago is a true outlier in a sea of continents.
The Comoro Islands form a rough east-west line. This direction already shows it is not related to the off-shore section of the African rift which is nearby but runs north-south. The figure shows a cross section (rather bend) which crosses the main islands. It shows two additional peaks, Vailheu Bank just below sea level and Jumelles at a larger depth. Vailheu Bank is a well-known diving location, good for hammerhead sharks. The deep ocean here is known as one of the few locations where coelacanths still exists, albeit far too deep for divers!
The age of the Comoro Islands is not too well known. Volcanism probably dates back to 15-20 million years ago at Mayotte, and perhaps 10 million years for the other islands. The activity has continued until close to the present day, at all islands. Clearly, this is not a Hawai’i-type hot spot trail.
The islands were build up by the usual processes for volcanic islands. These are four-fold.
- Inflation. The ocean crust begins to rise because the heat and melt reduces the density. A bulge forms
- Eruption. Lava erupts and builds up new surface layers. This combines with the continuing inflation to build up the cone, which now breaks the surface.
- Land slides. The steep cone suffers frequent collapses which spread out debris over the slopes and surrounding area.
- Wave erosion. This creates a plateau, some 10 to 50 meter below sea level. The plateau can be at different depths (or heights) if the sea level has changed over time.
The third one is often overlooked, but may be responsible for the majority of the volcanic islands. We have seen examples in recent years, for instance with the complete sudden collapse of Anak Krakatau. Hunga Tonga’s ejecta also seem to have ended up as submarine flows: the volume of those may even include material that came from outward collapse of the caldera, and was not part of the original explosion. Seamounts are built out of debris.
The plateau phase is clearly visible for Mayotte.
The island of Grande Comore is 65 km long and 20 km wide. It is built around three volcanos. In the north is a dormant volcano, La Grille which reaches just over 1 km. It has had numerous fissure eruptions, each of which build a separate scoria cone, not unlike La Palma’s eruptions. The most recent eruption may have been within the past few hundred years. One eruption is dated to 700-1000 years ago. Pottery found in the fall-out shows that this was after people had first arrived here, which also places it within the past 1000 years. However, there is not a single historical report of an eruption from La Grille.
The southeastern tip of the island contains the M’Badjini volcano, 740 meters tall and extinct. One of its lava flows has been dated to 130,000 years ago. There is also a submarine volcanic ridge to the southeast of Grande Comore but very little is known about it.
The island is dominated by a central, highly active volcano, Karthala. Karthala is over 2km high and regularly erupts – most recently in 2005-2007. The volcano has a shape somewhat reminiscent of Mauna Loa: a long mountain with two rift zones. Eruptions tend to be effusive and can take place in the caldera, along the rift, or sometimes from locations low on the flanks. Phreatomagmatic explosions are also relatively common. The caldera is about 3 km across, and consists of several separate collapse features which surround a central main crater. Each collapse forms a crater which is typically 1-1.5 km diameter. The central caldera often changes in appearance. Whenever it is more than 300 meters deep, a green crater lake forms. This was present briefly in the 1940’s and 1950’s and longer lasting in the period 1991-2005.
In 1918, an explosion formed a pit crater in the northern part of the caldera. Over the years the depth decreased, and in 2007 lava erupted and filled in the crater perfectly, as shown in this picture.
Karthala is one of the two most active volcanoes of the southwest Indian Ocean. The other one is far famous: Piton La Fournaise. Karthala, in contrast, may be the most overlooked volcano on Earth. Inaccessibility is part of this: there are no roads up the mountain, and the slopes are covered in dense rain forest. The upper range is barren, due to the frequent eruptions. These eruptions take place in a 45-km long and 8-km wide zone along the ridge of the island. Currently, most activity is in the northern rift but over longer times both the northern and southern rifts are approximately equal.
The magma production rate is estimated at 10-3 km3/yr which makes it a moderately active system. But it is not harmless. Much of the island is in reach of its lava flows, and these have in the past reached the coast. Eruptions in the 19th century often involved the rift zones, including five eruptions in the period 1857-1862. After 1903, eruptions mostly remained within the caldera with as exception only the 1977 eruption. The eastern side of the island is covered in lava flows of a variety of ages, some fresh looking and others old grey. The western side shows much less evidence but this is because of the climate. The eastern side is much drier while the western side gets the rain and is densely vegetated.
In 1859, a caldera eruption produced a lava flow which traveled 13 kilometers down the northwest slopes. It reached the coast just north of the capital, Moroni. In 2003, gas emission from the southwest rift zone killed 13 people. The 1977 eruption started on April 5, low on the western flank, just north of a village. A new scoria cone grew and a’a lava travelled through two villages before reaching the coast. Almost 300 houses were destroyed. In 2005, explosions covered the island in ash and pumice, causing lahars especially on the west side of the island. Karthala is a hazardous volcano! And with close to half a million people living on Grande Comore, there are a lot of people at potential risk.
There is another hazard. The east side of Karthala is unusually steep. This seems to be the scar of a large landslide. This happened a long time ago and the slope has been stable since. But there is evidence of other collapses: the submarine slopes to the southwest of Karthala are covered in debris flows. Flank collapses are very infrequent, but they can be devastating when they happen. And unlike eruptions, flank collapses may not give much warning. This volcano is not to be trusted.
This is the smallest of the Comoros islands covering no more than 300 km2. The island is longer than it is wide. There is a central ridge some 800 meters high which is the remnant of the original large shield volcano, now deeply eroded. The western side is rugged, while the eastern tip of the islands is lower lying. Dating of the flows has indicated that there were two volcanoes, a younger western one and an older eastern one. The volcanoes grew between 5 and 2 million years ago. Little is left of the eastern volcano. The western one had a flank collapse on the eastern side, perhaps 2 million years ago.
Mohéli is sitting on a plateau, which (as discussed above) shows its age. On the southern side there are a number of small islands which form ridges pointing at Moheli itself. Interesting, these originally were eroded valleys on the flank of the volcano. Later lava flows filled in the valleys, and now erosion attacked the region around it but left the harder lava in place. The outcome was a series of radiating lava ridges. As the island sank back, the ridges became islets. These lava flows are among the last eruptions here. They are dated to between half a million and one million years ago. This also formed some strombolian cones and maars.
Everything points at Mohéli having a deep volcanic history, but it lacks a volcanic present. Its volcanoes have been extinct for half a million years.
The island of Anjouan has a different shape but a similar history. It is a triangular volcano which like Mohéli started out with one volcano, after which a rift volcano (Hawai’i like) developed. The central volcano (or its remnant) is called N’Tingui. It is up to 4 million years old although the basement is likely much older. The rifts are dated to around 1.5 million years ago. After that, a series of pyroclastic cones and maars formed, some of which are clearly visible in the map. The youngest volcanism date is 370 thousand years ago.
Anjouan contains a geological mystery. There is an outcrop of white quartzite on the island which is completely out of place on a basaltic volcanic island. The ocean floor here is 150 million years old. The quartzite must be much older.
Quartzite is compressed sandstone, from a sandy sediment that formed in a river valley. The compression requires high temperature and pressure, and normally happens as part of mountain building. It did not form in the rifting of Africa. If mountain building was required, then this compression happened during the continental collision that created Gondwana. The sand itself must have come from even older mountains.
When and where could this have happened? Supercontinents tend to break up along the approximate original sutures. That is true for the Atlantic ocean: it separated Europe from America along the approximate line where the two had first come together, now traced by the mountains chain running from Norway via Scotland to the Appalachians. So it was in Gondwana. One part of Gondwana had formed when India and Madagascar collided with the old cratons of Arabia and Africa (mainly the Congo). This was the East African Orogeny, and happened over a period of time around 600 to 550 million years ago. It build a north-south mountain chain covering part of Africa and Madagascar.
Shortly after (520 million years ago) the combination collided with East Antarctica and the Kalahari craton in the so-called Malagasy orogony, which affected Madagascar and southern India. It build an east-west mountain chain from Malawi across my war-like border with Mozambique into what is now Sri Lanka. The exact order in which the fragments came together is disputed, but the effect was two mountain chains in a ‘T’ shape, with the join near Madagascar. These mountains could have been the source of the quartzite. Erosion from the older mountains could have provided the sand, collecting in river deltas, and the second collision could have compressed it into mountains. Much later, India and Madagascar broke off, as described above, near the same suture. And somehow a layer of the compressed quartzite ended up underneath what is now Anjouan.
The details of this scenario of the origin of the quartzite are speculative! And the ‘somehow’ in the previous sentence leaves a lot unexplained. It is clear, however, that the quartzite must come from a continental remnant and is not related to the eruptions that build the island. It is a memory from Gondwana, a lost fragment which ended up mid-ocean.
The sandstone lay buried below the ocean floor. When the Comoros inflation which build the island began, it ended up above sea level. The quartzite shows indications that it interact with magma, suggesting the magma had gone through the quartzite layer. But in that case, how did it end up on top of the island?
We can try to fill in the mysterious ‘somehow’ using a bit of physics. To end up on top of a growing mountain, you need vertical motion. Quartzite is a silicate, and therefore relatively low density: 2615 kg/m3, while solid basalt is typically 2950 kg/m3. Liquid basalt (also known as magma) is a bit less dense, but at typically 2700 kg/m3 still a bit more than the quartzite. So if there is a magma chamber directly above the quartzite, the quartzite has buoyancy and will attempt to float. For this to happen, the magma chamber should obviously be larger than the block of quartzite – it has to fit in the magma bath in order to float.
The Comoros magma came from 50 km deep, from the mantle. As it moved up, a magma chamber formed perhaps 15 km deep – right where the quartzite layer had ended up. A quartzite block dislodged, and began to float up in the magma. As the magma chamber moved up, the quartzite came bobbing with. The float ended a few kilometers below the surface from where the magma began to rise in narrow conduits. Now two things happened. The top of the mountain suffered erosion and collapses. While this happened, the mountain lost a bit of weight, and hydrostatic pressure from below pushed the mountain a bit back up – just like an iceberg with the top removed. So the quartzite block ended up higher, with less mountain above it, and came up a bit higher. Now that Anjouan has been extinct for a while, the quartzite has ended up on the surface, halfway up the remaining peak.
This block has an interesting story to tell, which I hope I got right. The indications for interaction with magma wasn’t because magma rose through it -instead it rose through the magma. It is a bit of Gondwana in a location where it should not be. This outcrop on Anjouan is a third fragment of Gondwana in the Indian ocean.
Mayotte is famous for its nature. If you like wild orchids, Mayotte is not a bad place to explore. (You will need to decide whether to go for the rainy season though.) The densely populated island is surrounded by a wide coral reef forming a double barrier reef. The reef surrounds one of the largest lagoons in the world, compared to a natural aquarium with reef fish, turtles and dolphins. It is up to 100 meters deep, covers an area (not including the island) of more than 1000 km2 and contains a hundred places where the coral has formed small islands (at least at low tide). The lagoon is in fact a large submarine plateau. It is not always submerged: at places there are gaps in the barrier reef which were formed by rivers at a time when sea levels were lower. The plateau already indicates that this island is older than the others.
The island itself is not high (just over 600 meter) but hilly in places. A topographical map shows four or five separate volcanic remnants, which together make up to complicated outline of the island. Mount Choungi in the south is a good example: a steep cone, just under 600 meter high, with a reasonable track (dry season only, and some scrambling required) and great views. There are other remnants: Dziani Dzaha is a crater lake on the Petite Terre island where the airport is. The ferry to this island only takes about 30 minutes, making it a worthwhile diversion during a visit. There are three or four other apparent craters and calderas along the east side of Petite Terre, all of which are missing their eastern half in a possible flank collapse.
The shield volcano that made up the island is 10-15 million years old. The oldest lava found on the island is 8 million years. The lava flows continued until 1.5 million years ago. Like the other islands, there were multiple volcanoes, each active at a different time. The older volcanics formed the southern side of the island and the younger activity was in the north. The eruptions largely ceased a million years ago. There have been a few minor eruptions since. An ash layer found in the coral has been dated to 7000 years ago. It is not clear where this ash is from, but it may be the formation of Dziana Dhana. (Alternatively, the prevailing winds mean that Grande Comore is a possible origin; this is also the only island anywhere near with active volcanoes.) GVP lists two more recent pumice layers. Gas emissions have been registered on Petite Terre. It seems plausible that there is some volcanic activity on the eastern side of Mayotte, causing rare phreatic explosions but not producing any magma, and not clearly related to the older volcanoes here. Perhaps this was not the best place to build the airport.
The normal future here would be that Mayotte would begin to sink, while at the same time the coral barrier reef grows. Eventually, an atoll forms; It would be much like Mayotte is now but without the island and with a slightly higher reef. But the longevity of all the Comoros islands shows that volcanism here has a tendency to return. Mayotte may not be finished yet. Perhaps a new volcano will eventually grow up and rejuvenate the island. It is worth keeping an eye on the stirrings at Petite Terre.
By why are the Comoros here? There is no obvious cause for this volcanism in the Mozambique channel. It is well away from the sea mounts closer to Africa. A mantle plume has been proposed, of course, but it lacks any evidence. A hotspot trail has been suggested to lead from the Seychelles to the Comoros, but this does not agree with the direction of plate motion, and the Seychelles are already linked to Reunion. There is also little evidence for a tectonic fault system underlying the Comoros.
However, this last point is not as clear as the case against a hot spot. There may indeed well be a fault system here. The east-west alignment of the Comoros links the tip of Madagascar to the Davies Ridge, the remnant of the old transform fault along the coast of Africa. The Davies Ridge is an old fault, but further north it is being reactivated by the East African rift as it is extending off-shore. Obviously the time scales are extraordinary slow – tens of millions of years. Africa is never in a hurry. The net effect of the re-activation seems to be that Somalia is beginning to separate and drift off into the Indian Ocean.
For that to happen, the would-be Somalia plate also needs to cut loose in the south. The suggestion has been made that this loosening is the cause of the Comoros. A transform fault is forming which will complete the separation of the Somali plate. There isn’t a clearly developed fault yet (as shown by the lack of seismicity) but it there is now a diffuse region of weakness. 50 kilometers below, a partly molten mantle was waiting, and as the weakness developed 10 to 20 million years ago, magma started to trickle upward. Call it trickle-up magmanomics, if you wish. This model of early initiation of a new transform fault explains the location and extend of the Comoros, the fact that the islands do not line up well, and the fact that volcanism here seems to be intermittent and is moving randomly between the islands.
And then everything changed. On 10 May 2018, people on Mayotte experienced a M5.9 earthquake. This was unheard of: the Comoros are (or were) seismically quiet. What followed was an eruption 6 times larger than the Leilani eruption of Kilauea that had started a week before. But that is for another post.
Albert, June 2023