Sunday, August 17, 2014

New publication: phylogenetic relationships of fur seals and sea lions (Otariidae)

Last week saw publication of a new study by Morgan Churchill, Mark Clementz, and myself, which presents a new cladistic analysis of the pinniped family Otariidae (Morphobank account/matrix available HERE) - known informally as fur seals and sea lions. I've been fascinated with fossil otariids since I started research on a specimen of the Pliocene dwarf fur seal Callorhinus gilmorei from the Rio Dell Formation in Northern California. This collaboration has been about two years in the making, and began shortly after we started work on the Pelagiarctos study published last year in PLOS One.

There are about 13-16 species of modern otariids in 7 (or 8) genera - 9 of which are fur seals in the genera Arctocephalus (some may be Arctophoca) and Callorhinus. Fur seals are generally smaller-bodied than sea lions, and are primitively characterized by retaining dense underfur - whereas sea lions have a thicker blubber layer, only possess hair (as opposed to underfur) and are generally much larger. Modern otariids are externally very similar, and it can be very difficult to tell them apart. Fur seals (with the exception of the Northern Fur Seal Callorhinus ursinus which has a distinctive snout shape) in particular are nearly impossible to tell apart externally, and identifications of osteological remains are often made based upon the geographic location as few geographically overlap with one another. Different species of Arctocephalus are nearly identical in skull morphology - however, sea lions are more morphologically distinct from one another. For example, California sea lions (Zalophus californianus) have a huge sagittal crest but an otherwise Arctocephalus-like skull, and the Steller's sea lion (Eumetopias jubatus) and New Zealand sea lion (Phocarctos hookeri) have a "domed" forehead. The Australian sea lion (Neophoca cinerea) has an unusually robust intertemporal region, while the South American sea lion (Otaria byronia) has a very elongate palate, procumbent upper third incisors and canines, and rugose tubercles on the side of the braincase (an elongate palate is also present in Phocarctos).

Relative to other pinnipeds (walruses, Odobenidae, and true seals, Phocidae), otariids primitively possess external ear pinnae (the type genus Otaria comes from the greek Otarion, meaning little ear, the most adorable name possible for such an intimidating monster as the south american sea lion), and can rotate their hindflippers forward and walk when on land - walruses are capable of this type of terrestrial locomotion as well, but true seals have an ankle that is more extremely adapted for swimming and can no longer rotate forward. The result is that their feet are permanently extended posteriorly - the same motion your ankle makes when standing on on your tiptoes. The long wing-like forelimbs of otariids permits underwater flying, in a manner reminiscent of bird flight.

Pinniped phylogeny of Barnes et al. (1985). Note that true seals are notably lacking from this hypothesis: most studies of pinniped evolution prior to the late 1980's were done under the now-outdated paradigm of pinniped diphyly.

Modern hypotheses for fur seal and sea lion relationships began under the assumption of pinniped diphyly, and this idea is pervasive in practically all papers on pinniped evolution published between 1960 and 1987 - while the idea has mostly faded away, a few paleontologists adhere to the idea. Effectively, pinniped diphyly maintains that otariids, odobenids, and the extinct desmatophocid seals form a natural group (the Otarioidea) that evolved from bear-like ancestors (Ursidae) and that true seals independently evolved from the weasel family (Mustelidae). This hypothesis was first challenged in 1987 when Andre Wyss identified a number of basicranial features linking walruses with true seals - suggesting not only that pinnipeds were diphyletic, but also that Otarioidea may not be monophyletic. Pinniped monophyly has been corroborated by many subsequent morphological studies and virtually every single molecular analysis of carnivoran phylogeny ever published, whereas pinniped diphyly has never been robustly supported by a single cladistic analysis of morphological data (in other words, proponents of pinniped diphyly have relied upon hand drawn cladograms).

Interrelationships of otariids were generally not investigated much prior to the advent of cladistics, with the exception that otariids were assumed to contain two natural groups: the fur seals (Arctocephalinae: Arctocephalus/Arctophoca + Callorhinus) and the sea lions (Otariinae: Eumetopias, Neophoca, Otaria, Phocarctos, Zalophus). The first cladistic analysis of otariid relationships conducted by Berta and Demere (1986) initially supported the monophyly of these two subfamilies, but has been subsequently challenged by most molecular analyses which indicate that both subfamilies are paraphyletic, with large body size and dense underfur being lost and gained several times.

Some molecular cladistic results from prior studies.

Aside from the first study by Berta and Demere (1986), only two following studies published morphological analyses of the otariidae: a paper by Demere and Berta (2005) included a cladogram from a separate unpublished study, and a fully published cladistic analysis published by Barnes et al. (2006) for the newly described fossil sea lion Proterozetes. Neither of these studies used more than 45 characters, which is more than some other papers I've read, but Morgan and I felt like more could be done. So, we surveyed the literature for additional characters, including some from published studies on phocid phylogeny which were just as useful to look at morphological variation within otariids, in addition to brainstorming totally new characters nobody else had thought of yet.

One of the problems facing us is that for some features, there is quite a lot of variation within a single species - for example, in most mammals there is a healthy amount of variation within a species. Think about how different everyone looks on your next bus or subway ride - the same applies to other mammals. Dental variation is remarkably common, with many individuals having double rooted teeth versus single rooted in others, or have more strongly developed cusps or ridges called cingula. When you code characters for a cladistic analysis, you have different character states summarized by numbers. An example would be character #X - labial cingulum on upper postcanine teeth (in English, a small ridge around the base of the crown on the cheek side of the tooth): 0- present. 1- absent. It's present in primitive pinnipeds like Enaliarctos, but absent in most modern pinnipeds - except some sea lions, where it has secondarily evolved (called a reversal). The problem is that for some species, some individuals have lingual cingula and others do not - this sort of variation is called a polymorphism, and it means that species is coded for both character states (01 in the cladistic matrix). When a 01 is coded, the program effectively treats it as a big "?" and the character is uninformative for that taxon. A way to get around this and still harness the usefulness of that polymorphic condition (because in many cases it represents the incipient development of the derived condition) is to introduce an intermediate "polymorphic" character state (0=present, 1=polymorphic, 2=absent) and run the character as an ordered (or additive) character (in other words, the program treats the character states as progressing from one to another, rather than unordered, the typical manner in which vertebrate morphologists treat cladistic characters).

We were also careful to avoid ambiguous character definitions, which are prevalent in many older studies. Whenever possible, we used ratios to define different states (e.g. long, >80% of the length of some other measurement). Vaguely defined characters basically prevent future researchers from replicating your analysis if they have no damn clue what you were talking about. Another problem prevalent in pinnipeds is sexual dimorphism: males are much larger than females, a life history trait that carries over into their skeletons. The skulls and mandibles of males look quite a bit different than females - female skulls are more generalized and gracile and often lack some of the structural peculiarities that permit identification of skulls, to the point where most female otariids look extremely similar. To circumvent this problem, whenever possible we only coded from adult male specimens. I have no idea how to run molecular analyses, but Morgan's done plenty of it, so he additionally ran some molecular analyses and a combined (morphology + molecules) analysis.

Morphology-based phylogenetic results from our new paper.

As for our results? Well, we found that some differences in sampling and search method produced a few minor changes in topology. Our morphology-only analyses tended to support sea lion (Otariinae) monophyly, and also supported fur seal (Arctocephalinae) paraphyly. Our molecular and combined analyses, however, showed that both groups are likely paraphyletic. One of the only results that was universal among these different analyses was a Eumetopias + Zalophus clade, which is supported by most previous analyses of molecular data (we also confirmed this for morphology, which is a good step forward). Similarly, we also consistently recovered a sister taxon relationship between the Pleistocene sea lion Proterozetes and the extant Steller's sea lion Eumetopias, to which Zalophus was sister, a clade we gave the unimaginative name "Northern sea lion clade". I have for a long while thought that Proterozetes was an unnecessary name, and that it should be recombined as Eumetopias ulysses, although Proterozetes does have a couple of Zalophus-like features, such as a higher sagittal crest and a more gracile rostrum and a skull that is slightly proportionally shorter than Eumetopias, suggesting that Eumetopias evolved as a sort of "mega" California sea lion.

Most North Pacific fossil fur seals (Thalassoleon mexicanus, macnallyae, and inouei) were consistently placed on the otariid stem, but without good support. The extant Northern fur seal, Callorhinus ursinus, was typically the earliest diverging extant otariid; no sister group relationship with Callorhinus gilmorei was supported, although some of the key features of Callorhinus ursinus (dorsally inflated maxilla, higher facial angle) are not yet known for published specimens of Callorhinus gilmorei and we were unable to code for these characters. Discovery of skulls of Callorhinus gilmorei can probably alleviate this problem - stay tuned. The fossil fur seal Hydrarctos lomasiensis from Peru, on the other hand, consistently plotted out in the sea lion-Arctocephalus/Arctophoca clade. The last significant fossil otariid, Neophoca palatina from the Pleistocene of New Zealand, kinda came out all over the place: it's a nice skull (what's left of it - more on that in the future, so again, stay tuned) but it critically lacks the tip of the rostrum and the dentition, and has no other elements.

Why were our results for fossil otariids so... disappointing? It's largely due to the fact that there isn't much variation between even modern otariid species, and if a fossil has anything other than a nearly complete skull, some of the more obscure peculiarities necessary for good phylogenetic resolution are going to be absent and not codable. Furthermore, fossil otariids are relatively derived and we don't have any examples of otariids that have plesimorphic morphology intermediate between something like a fur seal and an early enaliarctine pinnipedimorph. We know they're pinnipeds, and we know they evolved from an enaliarctine-like ancestor, but the transitional fossils just don't exist (yet... stay tuned, again!). The otariid fossil record is also temporally shallow - the oldest known otariid is a Pithanotaria-like mandible from the late Miocene Aoki Formation of Japan (12.5-11.8 Ma; Kohno et al. 2007). In contrast, fossil phocids go back to 19 Ma (Afrophoca), fossil odobenids are known from about 16-17.5 Ma (Prototaria, Proneotherium, Pelagiarctos), and desmatophocids go back to about 19 Ma as well (Desmatophoca brachycephala). Difficulty for modern otariids results from a general lack of ecological specialization, a relatively recent diversification (see below) and frequent hybridization. Furthermore, some species - particularly Artcophoca phillippii, A. tropicalis, and A. gazella, are currently rare and distributed only on remote islands, and large accessible collections of these species were not really available for our study.

Phylogenetic relationships can be interesting in and of themselves, but after hundreds of hours staring at a big matrix of 0s and 1s you may be attempted to slit your wrists just to see some color - cladistics is a great tool, and while a lot of people enjoy doing nothing else, it's healthy to mix it up a little bit and use those cladistic results for addressing something more interesting. One question we addressed was biogeography - the fossil record of otariids is pretty craptacular, and various biogeographic hypotheses have been advanced purely on molecular results. That's fine, and often necessary, but even a crappy fossil record is a great opportunity to constrain or spot check hypotheses - since fossils invariably reflect the presence (or in some cases, absence) of a particular taxon in a particular region (ocean basins in our case). Otariids have generally thought to have had a North Pacific origin, thanks due in part to the fossil record (Pithanotaria, Thalassoleon) and the earliest diverging extant otariid - Callorhinus - still lives here (and has never left). However, only about 1/3 of extant otariid species live here today, and most inhabit the southern hemisphere. Previous studies suggested a Plio-Pleistocene dispersal to the southern hemisphere. Our results - an ancestral character state analysis of biogeography - show a North Pacific origin for otariids (unsurprisingly), with the Australian sea lion, Neophoca, and the monophyletic southern otariid clade (Phocarctos + Otaria + Arctocephalus/Artcophoca) independently dispersing to the southern hemisphere. The fossil Hydrarctos, which is apparently as old as 6.6 Ma or so (and as young as 3.9 Ma), sets a minimum and maximum date for this dispersal. A single femur identified as Arctocephalus from South Africa about 3-5 Ma also sets some constraints for the dispersal.

Further constraint may be included when sea surface temperatures tolerated by extant otariids are plotted onto the phylogeny. We reconstructed the ancestral water temperature toleration for the southern otariid clade as 22-20 degrees celsius at the most: for the uninitiated, warm water is an effective barrier towards marine mammal dispersal, and is generally more important than cold water for restricting marine mammal migration/dispersal (sea and pack ice can be another). Many modern species cannot cross the equator, largely owing to the absence of abundant food. In cold, nutrient-rich cold temperate waters of the eastern Pacific, the ocean teems with life and marine mammals are locally abundant. To put in a plug for another paper, I discussed this quite a bit in my recent paper in Geodiversitas and proposed that the equatorial warm water barrier (in concert with the unopened Bering strait and the recently closed Central American seaway) permitted a highly distinct, provincial marine mammal fauna to evolve in the eastern North Pacific (Boessenecker, 2013). Two periods in the late Neogene saw equatorial waters become cool enough to permit temperate/cold temperate otariids cross from North to South: late Pliocene global cooling beginning about 3 Ma, and an older period of cooler water and increased upwelling took place about 6-7 Ma. From 3-5 Ma the equatorial Pacific was relatively warm and characterized by permanent El Nino conditions, and warmer than the reconstructed temperature tolerance of the southern otariid clade. Prior to 7 Ma, the equatorial waters were relatively warm and would have been an effective barrier to otariid dispersal. Coincidentally, a separate study (Yonezawa et al. 2009 - to date, the most convincing molecular study of otariids in my opinion) found a ~7 Ma molecular date for the diversification of the southern otariid clade. The fossil record indicates that otariids were already present in the southern hemisphere before the 3 Ma initiation of glacial-related global sea surface cooling, and the combination of the fossil record, molecular date, and temperature data strongly suggest a southern dispersal around 6-7 Ma. Certainly, the densely sampled Pisco Formation of Peru has yielded no otariids older than about 6.6 Ma (and has yielded an otherwise well-sampled and rich marine mammal fauna from multiple stratigraphic horizons), corroborating this estimate.

What future work remains to be done? More work on some of the more obscure southern hemisphere fur seals is necessary (and the Guadalupe fur seal, for that matter). Better molecular sampling of otariids, and a larger osteological sample size of some species is necessary. We of course need more fossil otariids - more digging in non-Sharktooth Hill middle Miocene deposits! Otariids apparently just weren't in the Temblor Sea, and early otariids may yet be hiding in middle Miocene rocks from the California and Oregon coast, and Japan.

For our morphobank account, CLICK HERE.


Barnes LG, Domning DP, Ray CE. 1985. Status of studies on fossil marine mammals. Marine Mammal Science 1:15-53.

Barnes LG, Ray CE, Koretsky IA. 2006. A new Pliocene sea lion Proterozetes ulysses (Mammalia: Otariidae) from Oregon, U.S.A. In: Csiki Z, ed. Mesozoic and Cenozoic vertebrates and paleoenvironments: tributes to the career of Prof. Dan Grigorescu. Bucharest: Ars Docendi, 57–77.

Berta A, Deméré TA. 1986. Callorhinus gilmorei n. sp., (Carnivora: Otariidae) from the San Diego Formation (Blancan) and its implications for otariid phylogeny. Transactions of the San Diego Society of Natural History 21: 111–126.

Boessenecker RW. 2011. New records of the fur seal Callorhinus (Carnivora: Otariidae) from the Plio-Pleistocene Rio Dell Formation of Northern California and comments on otariid dental evolution. Journal of Vertebrate Paleontology 31: 454–467.

Boessenecker RW. 2013. A new marine vertebrate assemblage from the Late Neogene Purisima Formation in Central California, part II: pinnipeds and cetaceans. Geodiversitas 35: 815-940.

Churchill, M., Boessenecker, R.W., and Clementz, M.T. 2014. Colonization of the Southern Hemisphere by fur seals and sea lions (Carnivora: Otariidae) revealed by combined evidence phylogenetic and Bayesian biogeographical analysis. Zoological Journal of the Linnean Society. In press, online early:
Deméré TA, Berta A. 2005. New skeletal material of Thalassoleon (Otariidae: Pinnipedia) from the Late Miocene-Early Pliocene (Hemphillian) of California. Bulletin of the Florida Museum of Natural History 45: 379–411.

Kohno, N., Koike, H. and Narita, K., 2007: Outline of fossil marine mammals from the Middle Miocene Bessho and Aoki Formations, Nagano Prefecture, Japan. Research Report of the Shinshushinmachi Fossil Museum 10: 145.

Yonezawa T, Kohno N, Hasegawa M. 2009. The monophyletic origin of sea lions and fur seals (Carnivora: Otariidae) in the Southern Hemisphere. Gene 441: 89–99.


Darren Naish said...

Congrats on such a fundamentally useful paper!

Robert Boessenecker said...

Thanks Darren! Let us know if you end up having any hate mail for us.