Agnatha All Along? Changing the Evolutionary Narrative of Vertebrate Origins

Who's been messing up everything? It's been the Agnatha all along

Like Comment
Read the paper

Scientists or not, we are all suckers for a good story.  Sure, we can sense a story too good to be true –– such as when avocados are on sale, or when your dating app pings the match made in Heaven.  But we also ‘believe’ in stories that we know are not exactly true.  Don’t tell me that we don’t cast ourselves in favourite fictions.  And as a baseball fan I can tell you to never cite advanced fielding stats with someone who believes Derek Jeter was the greatest shortstop ever to call the Yankee Stadium home.  Privately and professionally, we all live in that continuum.  We learn to make judicious use of just-so stories.


Here is one of those stories that clicked in my distant memory of university education.  In my sophomore year, everyone in my major needed to take comparative vertebrate anatomy, and anyone having done so remembers three months of dissecting a dogfish and a cat in the fume of preservatives and deteriorating flesh.  Now, it generally was not hard for students to understand homologous relationships of structures between these gnathostomes, but it was a long shot, I think, for students to fully visualize evolutionary transformation of the vertebrate body plan with just a squalid, a felid, and prepared specimens of a few other modern vertebrates in between.  Living tips of the vertebrate tree are far removed from their common ancestors, and it takes a pinch of fallacy to reconstruct the transformation series solely using those examples.  One transition, however, was as clear to me as a morning icicle of a Canadian spring.  That was the vertebrate origins.


Ammocoete larva of a Pacific lamprey Entosphenus tridentatus (Photograph: Gregory Kovalchuk).
Ammocoete larva of a Pacific lamprey Entosphenus tridentatus (Photograph: Gregory Kovalchuk).

In lab 1, we all put a slide of an amphioxus –– an invertebrate chordate –– under the microscope and proceeded to the next slide that had an animal almost identical to the amphioxus.  This second slide contained an ammocoete, a larva of modern lamprey.  Both amphioxus and ammocoetes are functionally blind, burrow in sand, and feed by filtering out food particles in the water column.  But unlike amphioxus, ammocoetes do not stop there.  After spending years in that primordial form, an ammocoete makes a dramatic transformation into an eel-like ectoparasitic blood-sucking adult form we typically associate with the image of a lamprey.  This metamorphosis is so breathtaking that up until the mid-19th century ammocoetes and adult lampreys were thought to be completely different groups of animals, hence persistence of the old taxonomic name in modern times to refer specifically to the larval phase.


The Water Babies?


Now you see the story here.  First, the two animals set apart by the vertebrate origins look alike and share apparently primordial anatomy of a filter feeder.  Second, the lamprey life history proceeds from a blind filter-feeder to a specialized predatory fish almost as if to make an analogue of early vertebrate evolution, to echo distant ancestors, or (if you will) to recapitulate the half-billion-year-old evolutionary transformation.  The idea made logical and optical sense.


Ammocoete, a filter-feeding larval form of a modern lamprey, was believed as an evolutionary relic of –– and therefore a proxy for –– the common ancestor from which all vertebrate lineages descended.
Image credits: modern lamprey, adult: US Geological Survey; modern lamprey, larva: Gregory Kovalchuk; fossil lamprey (Priscomyzon): Kristen Tietjen; ostracoderm (Arandaspis): Franz Anthony

This Haeckelian view has held its status as conventional wisdom for 150 years, around for as long as Victorian novels.  Ammocoetes, and by extension lampreys as a whole, have been treated and presented as “swimming time capsules” (as quoted in a media interview with my coauthor Michael Coates) in literature and classrooms.  Sure, experts always had a healthy dose of skepticism and varied in their versions of the story.  Amphioxus are now relegated to the earliest branching chordate lineage, no longer a cherished sister group to vertebrates.  Stem vertebrate fossils from Cambrian times do not exactly appear ammocoete-like in their anatomy.  Hagfish –– the other living jawless vertebrate lineage –– develop straight from embryos without any distinct larval phase.  And we cannot forget a prophetic review of ammocoete origin hypotheses by Thomas Evans, Philippe Janvier, and Margaret Docker.

Nevertheless, we continue to depict a vertebrate ancestor that looks like nothing but an ammocoete.  This is somewhat akin to the treatment of chondrichthyans (sharks, rays, skates, and chimaeras) in vertebrate zoology.  Despite unequivocal fossil evidence to the contrary, the classical narrative persists, which views the serially patterned developmental anatomy of elasmobranchs as the paragon of vertebrate archetype.  Similarly, the Ammocoete-First (or Ammocoete-Primitive) story is just so convenient that it has infiltrated our way of thinking about vertebrate origins irreplaceably.

Once fortified with historical inertia, just-so stories are difficult to interrogate.  In this case any ammocoete-driven narrative hinges on one prediction:  ammocoetes must extend deep into the vertebrate tree.  Though seemingly straightforward, this prediction is hard to test.  Being entirely soft-bodied, lamprey fossils are exceedingly rare.  When it comes to their larval form, which in modern species can be as small as a centimeter in length and occur in fast-flowing streams, the chance of finding such fossils would be as good as you going out to the ballpark and catching a Bartolo Colón home run.  Now, the spectacular Early Cretaceous lake deposits of Liaoning, China (home of the feathered dinosaurs) preserves an ammocoete phase of a lamprey Mesomyzon mengae.  But these fossils are barely older than 100 million years, less than a fifth of the way to the vertebrate ancestry.


The Team Stem Lamprey. Left to right from the top row: Tetsuto Miyashita (Canadian Museum of Nature); Robert W. Gess (Albany Museum); Kristen Tietjen (University of Kansas); Michael I. Coates (University of Chicago.  

Sucker for Life


It takes four balls for a batter to walk to the first base, and there are as many Palaeozoic localities that yield lamprey fossils: Waterloo Farm, South Africa (Devonian); Mazon Creek/Francis Creek Shale, Illinois; Bear Gulch Limestone, Montana; and Montceau-les-Mines, France (all Pennsylvanian).  These localities are at least 180 million years older than the Cretaceous lamprey.  Of these, my collaborator Robert Gess will follow my post with a detailed look at the Waterloo Farm Lagerstätte.  This is a particularly remarkable site.  Not only is Waterloo Farm home to the oldest fossil lamprey known (approx. 360 million years old), but it also represents a rare high-latitude locality (South Africa sat well within the southern polar circle in Devonian times), and the only one of the four localities to preserve a rich vertebrate fauna leading up to the major extinction episodes in the Fammenian stage.  Rob has the sharpest pair of eyes known in a fossil collector, and he is responsible for finding all specimens of Priscomyzon riniensis we included in our study.  With this and other localities, we had a pretty good range for a study of fossil lampreys: the four taxa of stem lamprey span across 50 million years in occurrence, come from two continents in low and high latitudes, and sit chronologically at more than two thirds of the way from modern lampreys to the vertebrate ancestry.

In addition to the Devonian genus Priscomyzon, we reported larvae and juveniles from three other stem lamprey taxa: Mayomyzon pieckoensis and Pipiscius zangerli from Mazon Creek; and Hardistiella montanensis from Bear Gulch.  These include hatchlings that still carry a yolk sac.  The smallest is a specimen of Priscomyzon slightly less than 14 mm in body length.  These immature individuals have large eyes, a sucker with cusps, and left and right baskets of gill arches united at the back.  The first two are straight-forward to interpret: these are defining traits of the predatory adult phase of modern lampreys.  The branchial baskets are important because in the filter-feeding ammocoetes the baskets are long and parallel, with food and respiratory current both passing through in between.  In contrast, modern lamprey adults have the baskets that are squat and united between the left and right side, where the feeding and respiratory passages are separate after the mouth.  Without this decoupling, they would not be able to suck.  This is a necessary adaptation for their predatory lifestyle.


A hatchling of Priscomyzon riniensis already has features associated with the predatory adult phase of modern lamprey and lacks those of the blind, filter-feeding acommocoate phase.  A Canadian 25-cent coin for scale (Photograph and drawing by Tetsuto Miyashita)

The hatchlings of Palaeozoic stem lampreys had the same feeding apparatus as the adult.  In Priscomyzon, the oral sucker has an elliptical shape at the end of an elongate snout in a hatchling, and undergoes changes in proportion toward the adult.  Given the diminutive size of the hatchlings (water is much more viscous for them than for adults), and given how the oral sucker can change shapes in modern species, I consider that these hatchlings were capable of feeding more or less like their adults, and perhaps in a way analogous to leeches.  This continues to be the area of internal discussion among the authors, and you can follow this in our supplementary file.  Personally, I would not be surprised if you waded in a Devonian lagoon in sandals and pulled up your feet to find these baby lampreys latching and sucking.


A life reconstruction of a hatchling of Priscomyzon riniensis.  Art by Kristen Tietjen

The presence of adult-like traits and the absence of ammocoete-like morphology led us to our conclusion: they did not have an ammocoete phase.  Otherwise, they would have to hatch as ammocoetes and promptly metamorphose (while still carrying a yolk sac) at a size that they can sit comfortably on your little fingernail.  If we found this pattern just in Priscomyzon, we would be torn how to interpret what could be a standalone phenomenon.  But with four species representing three independent branches of the lamprey stem, we are as confident about our interpretation as for Ichiro Suzuki to catch a flyball in the right field: the absence of an ammocoete phase was likely the general condition from which modern lampreys evolved.


A hatchling of Pipiscius zangerli, a Carboniferous stem lamprey from the Mazon Creek fauna, Illinois.  This individual also carries a yolk sac, indicating its premature death (Photograph and drawing by Tetsuto Miyashita)

Two questions stand.  If not an evolutionary throwback, how do we explain the origin of ammocoetes?  And if not ammocoetes, do we have any alternative model of the vertebrate ancestry at hand?  This is not a typical double play situation –– fielding one doesn’t answer the other.

Let’s take a look at the first question with some attention to ecological background.  The Palaeozoic stem lampreys inhabited nearshore (Mayomyzon and Pipiscius), lagoon (Hardistiella), and estuarine lake (Priscomyzon).  Each of these marginal marine localities yields a diverse assemblage of larval, juvenile, and mature, gravid fish, and was likely a nursery to many species that cruised the adjacent sea.  In contrast, the vast majority of modern lampreys occur in rivers and lakes.  Although some species are anadromous, lampreys with an ammocoete larval phase invariably and exclusively start their life cycle in freshwater.  While the saltwater lineages of the Palaeozoic fossil lampreys all went extinct, those that populated freshwater survived four of the five mass extinction events.

The life cycle of a modern lamprey through the ammocoete phase to metamorphosis to the predatory adult phase (Graphics by the US Geological Survey)

As those who watched the Alien movies can appreciate, closed systems where prey comes in low abundance and punctuation (like humans on passing spaceships) are tough environments for specialized macroscopic predators to adapt to.  Occupying the habitats so ephemeral, fluctuating, and poorly connected as freshwater systems, it is not hard to imagine the advantage of biding time in a more generalist larval phase until you can count on juicy fish.  Crucially, you don’t have to be too complicated as a larva for such lifestyle, which extensively overlaps with that of an amphioxus.  There you have our speculation –– the ancestors of the lamprey crown group inserted a filter-feeding larval phase in their life history to adapt to freshwater environments.  All the primordial resemblances I saw between ammocoetes and amphioxus in my Anatomy course would be the result of convergence due to similar lifestyle.


A representative ostracoderm, Arandaspis.  Ostracoderms are a diverse and extinct assemblage of jawless fish that have an armour of bony scales and plates. Art by Franz Anthony

This brings us to the second question: What did the last common ancestor between lampreys and us (all living vertebrates) look like?  Going back to the work I led in 2019, we proposed an intriguing view on vertebrate phylogeny.  Preceding phylogenies had placed naked jawless vertebrates (hagfishes and lampreys) as primitive, and ostracoderms –– an assemblage of Palaeozoic jawless fishes that covered their body with bony scales and plates –– to form successive branches off the stem of jawed vertebrates.  We suggested an alternative that some ostracoderms are more closely related to modern hagfishes and lampreys.  Now that we removed ammocoetes to the lamprey crown group, our phylogeny predicts that living vertebrate lineages can be traced to the ostracoderm assemblage.  This generates two implications: 1) ancestors of hagfishes and lampreys lost bony dermal skeleton, thus the living descendants are more derived anatomically than ostracoderms; and 2) ostracoderms now bracket the last common ancestor of all living vertebrates, therefore presenting a clear alternative to ammocoetes for the common ancestor we all come from.

The discovery of a new larval form in fossil lampreys now indicates that an ammocoete is a more recent invention. Instead, ostracoderms bracket the vertebrate ancestor. This new view of vertebrate ancestry suggests that the first vertebrates already had bones in their skeleton, and the development of modern lampreys offers little clue to the ancestral condition of vertebrates.
Image credits: modern lamprey, adult: US Geological Survey; modern lamprey, larva: Gregory Kovalchuk; fossil lamprey (Priscomyzon): Kristen Tietjen; ostracoderm (Arandaspis): Franz Anthony


A Creature Luminous, Ghastly, and Spectral


On how thin of the proverbial ice do we stand?  We might see cracks running in all directions.  Just like the elasmobranch worship, one could follow an endless interrogation of ad hoc explanations to restore ammocoetes in their glorious ancestral status.  Any phylogeny that groups all the Palaeozoic stem lampreys in a single exclusive clade will necessarily pay a pilgrimage to the Ammocoete-First model.  Even without such phylogeny, one only needs to be less afraid of flying above parsimony to consider the possibility of ammocoetes lost independently.  Unless hard evidence turns up of ammocoetes running deep in vertebrate phylogeny, however, we would be essentially chasing a hound of the Baskervilles in footprints and growls that reinforce the imagination.  It would also be worth revisiting exactly what substance we have seen in common between ammocoetes and amphioxus, if there is anything more than a convenient analogue by circumstance. 


A hatchling of a modern sea lamprey (Petromyzon marinus).  Born blind and 'toothless', it burrows in sand and grows as a filter-feeding ammocoete larva (Photograph by Tetsuto Miyashita)

If we are right –– which still is a big if –– we essentially made it a little more difficult for those teaching vertebrate anatomy or using lampreys in their research as a model for primitive vertebrate conditions.  Even if one is looking at lamprey embryos, the development should be interpreted in the context of leading to this secondarily evolved larval phase.  On one hand it is rather awkward to dance to the new twist.  On the other hand, the amphibian field has taken a similar revelation to their own advantage and expanded the breadth of evolutionary insights to gain from frog and salamander embryos.  To do so in vertebrate origins, we must be able to sift superficial congruence and integrate the data from embryos with morphological evolution tracked by the still poorly understood fossil record –– perhaps in an approach analogous to that of the courageous, imaginative, but widely criticized work by the late Richard Jefferies (reviewed in Before the Backbones).  Thomas Holtz at the University of Maryland gave voice to my sentiment by commenting that this is “why fossils matter for studying vertebrate evolution.”


Sucker for Life
This reconstruction of the Late Devonian estuarine lake in today's South Africa captures the life history of a stem lamprey Priscomyzon riniensis. Three individuals representing different ontogenetic stages take a shelter in the meadow of charophyte algae Octochara crassa. Clockwise from right: a yolk-sac-carrying hatchling tucked in the charophyte; a juvenile attached to the substrate in the foreground; and an adult looming over the other individuals and showing its feeding apparatus. In the background, a school of the coelacanth Selenichthys kowiensis swim by.
Art by Kristen Tietjen

How do I look back on writing this paper personally?  For want of analogy, I return to my favourite subject.  The first thing that comes to my mind about Derek Jeter is none of those “Captain Clutch” moments, but Game 1 of the 2012 ALCS when he broke his ankle in five steps going for a Jhonny Peralta groundball in the top of the 12th.  With his left ankle popped, the Tigers took the lead.  As he got carried back to the dugout, the feeling of despair among the Yankees fans was palpable through a screen that they lost the game, the series, and the season right there and then.  Five years on, Jeter wrote of this experience to The Players Tribune.  In it, he talks about how Ichiro Suzuki, a Zen-like outfielder with the team, came to sit beside him in the changing room after the game.  The two shared long silence until the clubhouse completely vacated.  When Jeter picked himself up, Ichiro, still in his uniform, stood and watched as he left, as if to show his respect.

Ichiro strikes his signature pose, graced by two living cyclostomes.  A parting gift from a lab mate Stephanie Sang upon my departure from the University of Chicago, March 2020.  Art by Stephanie Sang


The ammocoete model enjoyed the long popularity.  It remains to be the simplest and optically sensible explanation for vertebrate origins.  But we also know that the hypothesis is a product of our own senses just as any other about what we would like to see in the world.  Evolution owes us nothing to seal its key moment in ontogeny as if to do us a favour.  Neither is evolution obliged to conserve it over hundreds of millions of years against the crushing wave of selection for no sensible reason.  Just like Jeter’s career continues to be scrutinized in the wake of sabermetrics, the fate has it that new evidence often betrays our fanciful expectations.  I decided to start looking for evidence of ammocoetes in Palaeozoic rocks in 2014 as a PhD student, and ended up with something quite to the contrary.  So I would like to learn from Ichiro a thing or two, and take my hat off for advances of the field made possible by the study of ammocoetes.  Our paper was not to invalidate this line of research, but to add to it by giving the animals different meanings in the study of vertebrate evolution.  There will be other (and arguably better) players taking shortstop for the Yankees in my lifetime, but none so illustrious, suave, and much of a class act as the one who wore Number 2.  It’s just so.

*Footnote: "Agnatha All Along" in the title is borrowed from Eric Scott's social media comment with #AgnathaAllAlong, which is a nod to "Agatha All Along" from the Marvel Studios miniseries WandaVision.

For additional resources, please visit the webpage summarizing this paper on my research website.

Tetsuto Miyashita

Research Scientist, Canadian Museum of Nature

I am a broadly trained zoologist with an interest in evolution of animal forms, patterns in the fossil record, and developmental mechanisms generating morphological variations at both micro- and macroevolutionary scales