Is the Coelacanth Special?

I assume everyone knows what a coelacanth is, or at least I did until WordPress suggested I had meant to write “complacent” instead, but just in case you are one of the few who missed the largest sarcopterygian1 discovery of the 20th century the story is as follows.

In the early 1930s scientists knew of a group of ancient, lobe-finned fishes2 thought to be either the ancestors of tetrapods or closely related to them, known as the coelacanths. These fishes were thought to have gone extinct prior to the beginning of the Paleogene, either slightly before or during the extinction that ended the age of dinosaurs. However, in 1938 a scientist in South Africa received a strange fish from a local fisherman that proved (after some investigation) to be a coelacanth. This species is now known to science as Latimeria chalumnae after the scientist who first found it (Marjorie Courtney-Latimer) and the river mouth it was found near (the Chalumna).

The coelacanth is one of the best examples of a Lazarus taxon around – a name applied to species that have disappeared from the fossil record only to reappear later with a substantial gap between its apparent extinction and its reappearance. Because of this it frequently gets co-opted as evidence that many other supposedly-extinct animals might still live among us, normally (it is claimed) as semi-mythical monsters. It’s hard to run across stories of living dinosaurs in the jungles of Africa, living pterosaurs in New Guinea, Gigantopithecus in North America, or plesiosaurs in Scottish lakes without someone invoking the coelacanth as evidence that pretty much any animal from the past might still be around. In fact, this particular post was inspired by watching “Expedition Mungo” on Animal Planet claim that the coelacanth was a good reason to suspect that the Triassic archosaur Postosuchus was behind the semi-legendary “ghbali” of Liberia3. So does the coelacanth herald a new world in which creatures from the past come back like sequels to movies from my childhood? Or is the coelacanth special, a once-off discovery?

A few things to note: there are actually two coelacanths. In 1999 a second species of coelacanth was discovered in Indonesia. This species received the name Latimeria menadoensis. Moreover, the total number of coelacanth specimens known to science is tiny. In 2005 Inoue et al. reported that L. chalumnae was represented by “more than 200 specimens” and that L. menadoensis was represented by only two specimens4. Moreover, there was a 14-year gap between the initial discovery of Latimeria and the discovery of a second specimen.

So how could we figure out what the discovery of the coelacanth means for other potential Lazarus taxa? Ideally we would want to know what Latimeria (and its precursors) have been doing since the K-Pg extinction event.

This is what we do know. Inoue et al. (2005) estimated the divergence time between the two Latimeria species using the mitochondrial genomes of the two species. The short take-home on the study is that they estimate that the two Latimerias split 40-30 million years ago, which is right around the time India collided with Eurasia. Indeed, Inoue et al. believe that this event is what split an African/Eurasian coastal population into two reproductively-isolated populations. This is also a remarkably long period of separation for two species in the same genus (although genera are pretty arbitrary – mammalogists tend to erect lots of genera for their special snowflakes whereas entomologists tend to pack species into genera like genera are being rationed). In fact, Inoue et al. point out that some prior studies had even suggested that the two Latimeria species might need to be synonymized because the differences between the species were so small.

So what’s going on? Another clue is in how much further back Inoue et al’s divergence time is from other studies. Instead of assuming a mutation rate from studies in other species they calculated it using the split between the Actinopterygii and Sarcopterygii as a calibration point5. The results they got suggest that Latimeria has had a relatively unchanging genome for a long, long time.

This is not entirely surprising. L. chalumnae is ovoviviparous (Smith et al., 1975) and found in water around 200 m deep (Fricke et al. 1991) where it appears to move largely by passive drifting. It appears that Latimeria is a slow-growing, slow-living, slow-reproducing species. These things tend to mitigate against rapid speciation.

So if the modern Latimeria spp. are slow-living fish of deeper (although not truly deep) water what have then been doing since the K-Pg extinction? Probably living right where they are. Modern coelacanths are nocturnal and have eyes adapted to low-light conditions (Yokayama et al. 1999) suggesting either a long period as nocturnal animals or a long period as deep-sea animals, or both. What this also means is that coelacanths are probably in a long, drawn-out process of going extinct. At one point coelacanths were a speciose group found in a variety of habitats. Now they are a clade with only two species found in a single habitat type. While it is possible that coelacanths could rebound and become a species-rich group again I think they’ve probably missed the boat on that. If in 40-30 million years coelacanths can barely split into distinguishable species it seems unlikely that in the next 10 million years they will suddenly evolve into a multiplicity of forms and take back the seas from the actinopterygians. (Mind you, the other possibility is that coelacanths are under strong stabilizing selection that keeps them from diverging significantly. This would suggest that coelacanths have inhabited a habitat that has experienced little to no change in millions of years, which again suggests that they aren’t really cut out for the rest of the world.)

A lot of this suggests that the main reason coelacanths are such an amazing Lazarus taxon is that they’ve spent a lot of the “age of mammals” in an environment of very poor fossilization where their unchanging body plans still serve them well. It is likely that if we had coelacanths from the age of coelacanth peak diversity to compare our modern Latimeria to we would find that Latimeria is a strange, deep-water offshoot of the main coelacanth branch. It just happens to be the branch that dodged the meteor 66 million years ago as well.

So are you about to find a dinosaur deep in the African jungles? Well, yes. Dinosaurs are remarkably abundant with over 10,000 extant species…oh, wait, you didn’t mean birds. Then no (probably). Theoretically a mountain or rainforest species could dodge fossilization in the same way that Latimeria did but terrestrial environments haven’t been nearly as static as the ocean. The world has warmed, cooled, dried, and, uh, wettened in ways that have had enormous impacts on terrestrial ecosystems but these same changes probably did little to the 200 m depths Latimeria prefers. The odds that a Mesozoic holdover is hanging out in a rainforest waiting to be re-discovered seems low.

But wait! It gets worse! You see, Latimeria is the sort of thing that is hard to find anyway. (Oh, and just in case you forgot, 1938 is a long time ago. Marjorie Courtney-Latimer, the coelacanth’s discoverer, died at age 97, and also died before I graduated college. We’ve yet to top the coelacanth.)

This lovely graph is a graph showing how fast we’ve been discovering fishes. Really, this is a graph using data from FishBase to attempt to determine when species were discovered. Two caveats: FishBase doesn’t include every species (although it tries) and I’m using the date associated with the description as the discovery date because that’s something I can slice out of the dataset with a few lines of code. Third bonus caveat: FishBase is sometimes wrong. However, all of this is small potatoes compared to the size of the dataset. There’s a lot of data here and so the occasional mistake or omission probably doesn’t change much.

What this graph shows is that we are still discovering fish pretty rapidly. While it might look like there’s a rapid leveling-off very recently (which would indicate that either we have collectively become terrible at finding new fish species or that we found them all) that’s an artifact of the data. Instead, we are, if anything, accelerating our rate of finding species. For comparison, the data for birds (which I had somewhere and can’t find) has pretty much flatlined. We’re basically done.

Here’s why this matters: if we are still discovering fish like crazy then we probably have a lot left to discover. The less we have already found the more likely it is that another big discovery awaits. Latimeria is in a group that we haven’t discovered much of (although in 78 years we haven’t outdone Latimeria) whereas many of the creatures people wish to support via Lazarus taxa are in groups that we know much, much more about. Someone will discover a new bacterium this year (or month, or week). Someone will discover a new beetle. Large mammals? We’re pretty much done with those, and the ones we “discover” tend to be things we knew about but didn’t realize they were different from something we’d already described.

But it gets better/worse! What if we split fish by depth? After all, Latimeria is, as I’ve noted, not very easy to get ahold of. Here I’ve split all fish species into one of three categories: shallow water fish (fish whose deepest depth range in FishBase is less than Latimeria chalumnae‘s shallow depth range of 150 m), deep water fish (L. chalumnae range or deeper), and medium depth fish who straddle these other ranges. Sure enough, there’s a real trend where deeper water fish are less well known.

Latimeria chalumnae (at least) is also a fairly large fish, close to the size of an adult human. If we split the available fish by size (smaller than L. menadoensis at 140 cm and larger) we see that larger fish are better known than smaller ones. Latimeria chalumnae was discovered right at the end of the age of describing big fish. Latimeria mendoensis was discovered ridiculously late, although since Latimeria was already known the Lazarus-ness of L. menadoensis‘s discovery is significantly less.

So what does this mean? It means, I think, that referencing the coelacanth as if it shows that some other creature from the depths of deep time is about to reappear is probably pretty wrong-headed. The coelacanth seems to be a pretty special case for the following reasons:

  1. First, the coelacanth lineage had to tail off into oblivion slowly. It’s not the only example of this by far (off the top of my head, temnospondyls closed down the party at the end of the Triassic but left a few loners at the bar drinking until the end of the Cretaceous) but there are other models as well, including catastrophic extinction over a short time. Without this the group in question is either numerous, and easily found already, or extinct.
  2. Second, the coelacanth lineage had to go somewhere where it was unlikely to fossilize. Not too hard, since fossilization is hard, but it’s a hoop to jump through. Shallow, inland waters wouldn’t have worked at all.
  3. Third, the coelacanths had to go somewhere where the world stood still for millions of years. The world just doesn’t do this. When India slammed into Eurasia’s underbelly and blasted the Himalayas skyward coelacanths noticed that there was a bit too much silt and the populations on either side of India lost contact. For animals on land, freshwater, or in the shallow coastlines this event was probably a good bit more dramatic6. If they had failed to do this one of two things would have happened: 1) they would have died 2) they would have changed and wouldn’t be the rediscovery of an ancient fish as much as the discovery of a weird fish that, eventually, we would find was related to a group we thought had died out.
  4. Fourth – and this one is really important – coelacanths had to end up somewhere where they were fundamentally hard to find. When people invoke the coelacanth to say that ancient creatures could remain undiscovered this undiscovered bit is pretty key. Coelacanths are still pretty hard to find. They may have small population sizes, they may just be hard to catch, but it really, really helps (as the graphs above show) that they are deeper-water fish. Not only are they deeper-water fish but the Indian Ocean is probably a better place to hide from scientists than, say, the North Atlantic. Coelacanths are the sort of fish that we would expect to have trouble finding regardless of their “living fossil” status.

In short, I tend to believe that the coelacanth rolled a lot of sixes. Coelacanths really are special. Almost any other path through their history would have either ended their lineage before humans could have discovered them or made them either common or known from ancient (historical) times. It’s not impossible that another species could pull off something similar, but only because adding up probabilities doesn’t generally get you to “impossible”.



Fricke, H., Hissmann, K., Schauer, J., Reinicke, O., Kasang, L., & Plante, R. (1991). Habitat and population size of the coelacanth Latimeria chalumnae at Grand Comoro. Environmental Biology of Fishes, 32(1), 287–300.

Inoue, J. G., Miya, M., Venkatesh, B., & Nishida, M. (2005). The mitochondrial genome of Indonesian coelacanth Latimeria menadoensis (Sarcopterygii: Coelacanthiformes) and divergence time estimation between the two coelacanths. Gene, 349, 227–235.

Smith, C. L., Rand, C. S., Schaeffer, B., & Atz, J. W. (1975). Latimeria, the Living Coelacanth, Is Ovoviviparous. Science, 190(4219), 1105 LP-1106. Retrieved from

Yokoyama, S., Zhang, H., Radlwimmer, F. B., & Blow, N. S. (1999). Adaptive Evolution of Color Vision of the Comoran Coelacanth (Latimeria chalumnae). Proceedings of the National Academy of Sciences of the United States of America, 96(11), 6279–6284. Retrieved from