Which company had the most diverse 2019 releases?

[Halichoeres]

I know there are some among us who only collect specific taxonomic groups (Dinosauria, Theropoda, Ornithopoda), but I know there are some like me, who are interested in biodiversity through time and paleoecology. To people like us, when a company announces a figure of an oddball genus, it's the best news.

What I want to know is, which company had the broadest range of taxa in 2019? Well, we could just count up the number of different species, and declare that the company with the highest number is the winner. But let's say a company made 20 different species, but they were all different kinds hadrosaur. That's not really very diverse, and if you think it is, I'd ask you to consider whether you'd still think so if they were all asaphid trilobites, a group that lasted longer than hadrosaurs and had more species.

So what's the fair way to measure this? Based on some metrics used in community ecology, I think I've come up with a good way to do it: phylogenetic diversity. Essentially, you specify a phylogram/cladogram that unites all of the species represented in a company's lineup, and you calculate the branch length in millions of years. Branch length is just the time between when Species A existed and its last common ancestor with Species B. Add up all of those lengths across a whole tree, and you get a tree length. For example, this is how you would calculate the diversity of EoFauna's first three figures:

The last common ancestor of Mammuthus and Palaeoloxodon was probably about 25 million years before each existed. The last common ancestor of elephants and Giganotosaurus is identical to the common ancestor of any sauropsid + any synapsid (so it could be a sloth and a tuatara if you wanted) at around 312 million years ago. Add the branch lengths up and the tree is 551.1 million years long. We can do this for any set of taxa. For more detail click on the spoiler:

Spoiler

For last common ancestors of clades with living representatives, I used the conservative end of the ranges at Blair Hedges's timetree.org, which aggregates divergence times from molecular phylogenetic studies. For any divergence times between fossil taxa with no direct descendants, I looked for the temporal range of the most inclusive clade that included each but not the other, and used the larger of the two values, which I believe is the only logical way to estimate this. For time of existence, I used the midpoint of the statigraphic range of a particular taxon. Where there is controversy about the position of a group (turtles, ichthyosaurs), it is usually precisely because divergences were happening in rapid succession, so the effect on tree length of alternative positions is generally trivial.

[close]

Now, community ecology, by definition, only deals with species that are living in the same place at the same time. Obviously, in the realm of prehistoric toys, we're dealing with the entire Phanerozoic and a little bit of the Proterozoic. If we use tree length alone, a Cambrian set, even if it sampled every phylum, would look really un-diverse, since the divergence times would only extend a few tens of millions of years into the Ediacaran. Conversely, a set of Pleistocene animals that was almost entirely mammals and one bird would get a huge bonus for the long-ago split between sauropsids and synapsids. To my mind, that wouldn't really be diverse. In short, tree length penalizes the Paleozoic and disproportionately rewards the Cenozoic. So I'm adding to the basic tree length a Deep Time Correction (cite me, Halichoeres 2019), which adds a bonus for each time period represented in a company's lineup, equal to how long ago the period ended. Each period gets the following bonuses (in millions of years):

0     Quaternary      
2.6  Neogene         
23   Paleogene
66   Cretaceous
145  Jurassic      
201.3 Triassic           
251.9 Permian         
298.9 Carboniferous    
358.9 Devonian         
419.2 Silurian         
443.8 Ordovician   
485.4 Cambrian   
541    Ediacaran   

Without this correction, a company that released a Permian synapsid, a Cenozoic mammal, and a dinosaur would get credit for the long branches (hundreds of millions of years) leading from the synapsid-sauropsid split, but at most 10 or 15 million years for the Permian synapsid. I think we can all agree that this lineup is more interesting and diverse than one with a dinosaur and two Cenozoic mammals, even though the latter would likely score better by raw tree length. For large lineups (>10 figures), the Deep Time Correction has a smaller effect, but it can be important for small lineups.

Anyway, I've calculated these lengths for a bunch of different companies that released prehistoric toys in 2019. Before I show the winners, I'd love to see people's guesses about which company they thought had the most diverse lineup this year. Or, if you have a better idea about how to measure it, I'd like to hear that too.

Edit November 27: Here are the top 5 companies by this metric for 2019:

Sonokong:

Adjusted length: 4,880.5 million years

Mattel:

Adjusted length: 2,592.2 million years

Paleozoo:

Adjusted length: 2,471.2 million years

PNSO:

Adjusted length: 2,232.6 million years

Safari Ltd:

Adjusted length: 1,973.3 million years

And here's a summary of a bunch of companies' releases. The dark portion of the bar is the raw tree length, the light portion is the deep time bonus.

[Concavenator]

My bet is Safari.

[Duna]

Very interesting topic. Please go on!

[PumperKrickel]

deleted

[Faelrin]

My best bet would be CollectA, just on account of having the Edaphosaurus (Permian) and Elasmotherium (Neogene to Quaternary) in their line up, in addition to the dinosaurs (1 thyreophoran - , 3 theropods, all Cretaceous), and one pterosaur (Cretaceous as well), and maybe if we're counting the tube which featured minis of the mammals (Paleogene to Quaternary), the terror bird (Neogene), and the Estemmenosuchus (Permian) they had made previously.


Other possible contender's?:

Safari Ltd had a good variety of species across time as well, with one mammal (Quaternary), pterosaur (Cretaceous), and psuedosuchian (Triassic), with 5 theropods (Jurassic - Cretaceous), 1 thyreophoran (Jurassic), 1 ceratopsian (Cretaceous), and 1 sauropod (Jurassic).

PNSO had some good diversity, but most from the Mesozoic (the exception being the Megalodon, from the Neogene). 4 marine reptiles (2 were ichthyosaurs, spanning from the Jurassic to Cretaceous), 3 thyreophorans (Jurassic to Cretaceous), 7 theropods (Jurassic to Cretaceous), 3 sauropods (Jurassic to Cretaceous), 1 hadrosaurid (Cretaceous), 2 ceratopsians (Cretaceous).

Mattel I think had good diversity too, though most from the Mesozoic as well (exception being the Dimetrodon, if counted). If we are only counting the 15 new species (not counting the Indominus rex, which wasn't a real animal, and perhaps the Dimetrodon could be counted towards the number of new species as well, which is a Permian period animal, edit: and a synapsid, as although the Destruct-A-Saurs version was intended for 2018, it was canceled, and so the Savage Strike one is the first released figure of that genus). There were 3 new species of pterosaurs (Jurassic to Cretaceous), 1 species of plesiosauria (Jurassic), 2 sauropods (Jurassic to Cretaceous), 1 sauropodomorph non-sauropod (Triassic), 1 pachycephalosaurid (Creataceous), 1 hadrosaurid (Cretaceous), 2 ceratopsians (Cretaceous), and 4 theropods (Jurassic to Cretaceous).

[Shonisaurus]

With a difference Collecta and Safari and also PNSO although many of the PNSO sculptures are miniatures.

[Ravonium]

I'm going to predict CollectA.

[stargatedalek]

I feel like it has to be quantifiably CollectA, since no one else has done invertebrates this year meanwhile CollectA has both arthropods and Cephalopods. There simply isn't any competition.

If we want to account for outliers and make it Ornithodiran exclusive (or make it a competition for second place) I think Mattel actually has a decent shot for the largest diversity of Ornithodirans.

[Shonisaurus]

Collecta is an authentic encyclopedia of the prehistoric world apart from current, wild, marine or invertebrate fauna. It is the most complete collection of dinosaurs and other prehistoric animals, I think of the whole story. In fact its founder Anthony Beeson wanted Collecta to be an encyclopedic referent of prehistoric animal figures and I think he has succeeded. Although we must also take into account Safari and PNSO and in this case PNSO is on the heels in terms of variety of figures to Collecta but logically does not have the variety of prehistoric mammals and Paleocene animals that has the Collecta brand.

As for Collecta variety this year 2019 has been very varied but with few figures (seven figures plus miniatures).

[Halichoeres]

Interesting guesses, everyone! I think you all have a pretty good intuition for what a winning lineup looks like. @Faelrin, your hypotheses were well-reasoned in particular.

@stargatedalek, I think you're right that for 2020, CollectA will be all but impossible to beat. It's an incredibly diverse lineup featuring, as you say, both spiralians and an ecdysozoan. Unless PNSO, Favorite, or Kaiyodo does something really unexpected, CollectA has already won. As you suggest, the invertebrates are a huge part of that--they're worth a billion years all on their own, and that's not even counting the extant horseshoe crab and nautilus.

Meanwhile, here are the results for 2019! Below, above the break, I've listed the five most popular companies (based on visits to their respective New for 2019 threads), plus another that was mentioned as a contender, plus the winner, which nobody guessed. The letters after the corrected length just represent the geological periods for which the company received a bonus. Other companies are after the break.

Safari Ltd:

raw tree length: 1,332 million years (=1.332 billion)
corrected length: 1,973 million years (TJKQ)

CollectA:

raw tree length: 1,521 million years
corrected length: 1,864 million years (PKPgNQ)
I gave them credit for the miniature box of mammals. A very synapsid-heavy year as a result.

Schleich:

raw tree length: 828 million years
corrected length: 1,290.9 million years (PJK)

Mattel:

raw tree length: 1,167.5 million years
corrected length: 1,831.7 million years (PTJK)
I'm not 100% sure I got all of Mattel's 2019 offerings, as I don't follow the line very closely. I gave them credit for repaints. Dimetrodon is helping out a lot.

Papo:

raw tree length: 475 million years
corrected length: 686 million years (JK)
Even though I gave Papo credit for repaints, they have one of the shortest trees, and by far the shortest of the most popular companies. 2018 would have looked a little better for them.

PNSO, a contender that @Faelrin mentioned:

raw tree length: 2,019 million years
corrected length: 2,232.6 million years (JKN)
PNSO got a very long tree the hard way. They got a very modest deep time correction, so the length is driven by sampling of lots of different sauropsid lineages. The mosasaur, ichthyosaurs, and crocodile were pretty important. A late Cenozoic shark did a lot of heavy lifting too. I included the sawfish that came with the Spinosaurus, which adds 325 million years to the tree. Without it, PNSO still beats CollectA but comes in just behind Safari.

But neither of them is actually the winner. Nobody guessed this, understandably since the company seems to be below most people's radar, but the winner is Sonokong:

raw tree length: 3,372 million years
corrected length: 4,880 million years (ꞒDPTJKQ)
All three major dinosaur lineages? Check. Pterosaurs? Check. Crocodile-line archosaur? Check. Lepidosaur? Check. Turtle? Check. Ichthyosaur? Check. Synapsids? Check. Fish? Check. But the protostomes carry the day: Sonokong offered both an ecdysozoan (Anomalocaris) and a spiralian (the ammonite, which has a barnacle on it that I didn't count, but that would have been even better). Together, the invertebrates add 910 million years (524 for their own branches, and another 326 for extending the branch that gives rise to vertebrates). This despite the fact that I used a conservative Triassic date for the unspecified ammonite (this doesn't affect the time correction, since Shastasaurus had the Triassic covered). Also, I did not include the "frogamander," which might be intended as Gerobatrachus, although I can't be sure.

Some people might, at this point, be saying, 'Who gives a rip about Sonokong?' I mean, their dinosaurs are about as realistic as Mattel's, but probably nobody on the forum has the same emotional connection with Dino Mecard that they have with the Jurassic Park franchise. So that's fine! Of the companies we're most familiar with, I would say that PNSO won 2019, followed closely by Safari. If you don't want to give PNSO credit for the sawfish, then Safari was the winner. 

In the future, companies wishing to claim the title of most diverse lineup can just play the trump card: plants. The divergence time between plants and animals is so large that Creative Beast in 2018, despite the only animals in the line being maniraptorans, would have beaten every company except Sonokong, just because they included Araucaria trees in the environment packs.

Below the spoiler I've posted the stats of some more minor companies, in case people want to check those out. If a company you're interested in isn't included, let me know, and I can whip up a tree and branch length estimate for it.

Spoiler

Colorata:

raw tree length: 973 million years
corrected length: 998.6 million years (PgNQ)

Creative Beast:

raw tree length: 365 million years
corrected length: 431 million years (K)

Eikoh:

raw tree length: 1,354 million years
corrected length: 1,565 million years (JK)

EoFauna:

raw tree length: 587 million years
corrected length: 800.6 million years (JKN)

Favorite:

raw tree length: 431 million years
corrected length: 642 million years (JK)
This was a boring year from Favorite. 2018 would have looked much better, and 2016 would have looked very, very good.

Kaiyodo:

raw tree length: 781 million years
corrected length: 992 million years (JK)

Mojö:

raw tree length: 516 million years
corrected length: 940.9 million years (DK)

Paleozoo:

raw tree length: 834 million years
corrected length: 2,471.2 million years (Pre-ꞒꞒDP)
This is one of the few instances where phylogenetic uncertainty might have an appreciable effect on the estimate. Ornithoscelida vs. Saurischia doesn't really change tree length much, but if Fractofusus is a eumetazoan, or a stem-animal, or a bilaterian, well, that can matter a lot. Here I treat it as a stem-animal, but I gave it really short branches because who the hell knows. This is also the biggest proportional gain from the deep time correction. I think Paleozoo's offerings are rightly thought of as very diverse, but by virtue of the animals being early, their divergence times are shallow.

Rebor:

raw tree length: 245 million years
corrected length: 456 million years (JK)
Given that up to now they've mainly been a theropod factory, this short tree length should be unsurprising.

Takara Tomy:

raw tree length: 1,054 million years
corrected length: 1,256 million years (JK)
I've only counted the Ania line here, but they have some other, cruder gashapons toys that would have added a little bit of length.

[close]

[Faelrin]

I know I was forgetting some brands, especially the one that won, Sonokong. Kind of a shame I did since they did go and make another figure of one of my favorite genus, the Anomalocaris. But seeing as I haven't been able to get around to getting any of them and since they aren't mentioned as much as CollectA, Papo, Safari Ltd, etc around here, they really sadly did slip off my radar again. Very good write up though and analysis altogether. I would like to see more of this done going forward. It is fascinating to see who was the most diverse in either method for the year. I imagine these tree's would have been time consuming to create as well.

Yes there were some species missing from the Mattel line up, perhaps some important, being the new species to the line (Brachiosaurus, Concavenator, Nasutoceratops, Parasaurolophus, Quetzalcoatlus), while many others were repaints and/or retools (Allosaurus, Ankylosaurus, Baryonyx, Ceratosaurus, Dilophosaurus, Dimorphodon, Metriacanthosaurus, Pachycephalosaurus, Proceratosaurus, Spinosaurus, Stegosaurus, Suchomimus, Triceratops, and Tyrannosaurus). If we want to include repacks, then Compsognathus (included in the Battle Damage Ultimate Baryonyx Breakout set), and Gallimimus (Battle Damage) could be there too. Not really a huge deal, since it wouldn't change much anyways, but may as well worth mention for the curious I guess.

Also to add to the point about the plants from the BotM accessory packs, we got ferns and horsetails too, which are often neglected too (although plants in general are, extinct and extant). I'm very glad to have gotten them, as they certainly help create interesting scenes for other figures as well (which I incorporated into a few of my reviews, and perhaps it wouldn't hurt to use them more often).

[CityRaptor]

Well, like you said, "accesory packs". Most plants are packed with animals as assecories. General preference seems to be a lot higher on the food chain. So by popularity, it is Carnivores>Herbivores>Plants. Well, expect for horses for some reason.

[Halichoeres]

I imagine these tree's would have been time consuming to create as well.

I did phylogenetics for my PhD, so generating the trees isn't too difficult or time-consuming because I've had lots of practice. I can generate a Newick tree in just a couple of minutes as long as I have the branch lengths looked up beforehand. Looking up the divergence times and resolving conflicting sources takes a little more time, but that's also something I've practiced a lot for work.

For the curious, this is what a Newick tree looks like. This format can be read and/or displayed by most phylogenetic software, including R using the package phytools:
Eikoh<-read.tree(text="((((((((Therizinosaurus:97,Velociraptor:94):1,Tyrannosaurus:103):7,Allosaurus:25):26,Carnotaurus:129):1,Dilophosaurus:9):29,((Apatosaurus:22,Amargasaurus:48):1,Saltasaurus:97):56):2,(((Maiasaura:9,Olorotitan:17):76,(((Sinoceratops:7,Styracoceratops:6):1,Triceratops:16):79,Stygimoloch:95):1):37,Tuojiangosaurus:39):34):12,Quetzalcoatlus:178);")

Thanks for the Mattel JW list, that's so many! I feel pretty silly missing the Brachiosaurus in particular because of all the hype around it. I've been giving everyone else credit for repaints, so it's only fair to do so for Mattel as well. And retools (like adding the Battle Damage panels) I think are fair game too. I don't think repacks should count if they're the same mold with the same paint job. I'm not as sure as you are that the missing species won't make a difference. Whereas none of these new species adds a huge branch, the revised tree will include twice as many taxa as I had originally included. Short branches can really add up if there are enough of them. Probably the biggest deal is the inclusion of Ankylosaurus--the first version had no thyreophorans.

Is this now a complete list?
Albertosaurus
Allosaurus
Amargasaurus
Ankylosaurus
Baryonyx
Brachiosaurus
Ceratosaurus
Coelurus
Concavenator
Dilophosaurus
Dimetrodon
Dimorphodon
Dracorex
Herrerasaurus
Metriacanthosaurus
Monolophosaurus
Mononykus
Mussaurus
Nasutoceratops
Pachycephalosaurus
Parasaurolophus
Plesiosaurus
Proceratosaurus
Protoceratops
Pteranodon
Quetzalcoatlus
Rhamphorhynchus
Spinosaurus
Stegosaurus
Stygimoloch
Styracosaurus
Suchomimus
Tapejara
Triceratops
Tyrannosaurus
Velociraptor

And were there any new mini dinos released, or at minimum repainted? I'm not finding any at jptoys.com, but I don't know if that's complete. Thanks again for the help!

[suspsy]

This is cool stuff, @Halichoeres. You'll have to do this again once all the 2020 reveals are done!

[Faelrin]

@Halichoeres Thanks for sharing how you've gone about this. I suppose it makes sense for this to be much easier to do with your experience and knowledge.

Also yes that Mattel list looks complete to my eyes, although come to think of I was only focusing on the main line at the time. If including the budget figures (6 and 12 inch), all species there are accounted for, and in the case of the snap squad figures only the Carnotaurus and maybe the Mosasaurus would be missing (not sure if it is intended to be a 2020 release or not as some places have gotten the wave with it already, although I believe both Jurassic Outpost and Collect Jurassic had it included with their 2020 product image dumps). I'm not sure if those side lines should be taken into account here or not though. Up to you I guess. The only new mini dinos to my knowledge were repaints of the Dilophosaurus, Indominus rex, and Velociraptor in metallic paint. The rest were repacks.


Edit: Come to think of it, Stygimoloch should probably be not included as both figures (Battle Damage and Savage Strike) are technically repacked figures. Sorry Stiggy.


Edit 2: Although the 6 inch budget figure of it was new. So maybe never mind if that side line is included here. I may as well list those species here too (for both the 6 and 12 inch), as well as the snap squad figures:


6 inch budget figures (all new sculpts):
Stygimoloch
Tyrannosaurus
Velociraptor


12 inch budget figures (only new and repaints):
Dilophosaurus (repaint)
Pteranodon (new)
Proceratosaurus (new)


Snap Squad (hybrids omitted):
Wave 1:
Tyrannosaurus
Velociraptor
Wave 2:
Carnotaurus
Dilophosaurus
Tyrannosaurus
Wave 3:
Velociraptor
Wave 4 (might actually be 2020):
Mosasaurus
Spinosaurus

[Halichoeres]

This is cool stuff, @Halichoeres. You'll have to do this again once all the 2020 reveals are done!

Thanks! I might do a preliminary analysis once "reveal season" is over, but enough companies have rolling releases (PNSO, Favorite, Sonokong) that it won't be possible to do it comprehensively until around this time next year.

@Faelrin thanks for all that! I'm going to include all the sub-lines--in for a penny, in for a pound. As it happens, Stygimoloch doesn't affect anything--those three pachycephalosaurs have branch lengths of 0. Carnotaurus makes a difference, however. I'll exclude Mosasaurus because it seems more consistent with a 2020 release. I'll run the revised tree and probably post it tomorrow. Thanks again!

[Halichoeres]

Holy cow, Mattel's huge slate of releases comes through. Ankylosaurus and Stegosaurus are most consequential; together they add more than 100 million years of branch length. All told, Mattel shoots past PNSO to take second place behind Sonokong:



Mattel pulls this off in complete brute force fashion, by releasing an ungodly 37 different nominal taxa (not counting completely invented ones). Most of these branches are pretty darned short on their own, and only four geological periods are represented, but with that many species you can make up a lot of ground the hard way. If you don't want to give them credit for repaints, second place reverts to Paleozoo. If you don't want to count Paleozoo because they aren't toys, exactly, then second place reverts to PNSO. If you don't want to give PNSO credit for the sawfish that came with their Spinosaurus, then second place reverts to Safari.

You can see how inefficiently Mattel got to second place in this graph, which shows the number of taxa included in a company's lineup on the horizontal axis, and the adjusted tree length (in millions of years) on the vertical axis:

The line is just a linear regression of length on taxon number. Any company below that line is punching below its weight: its tree has a short length relative to the number of taxa. This is what you see in dinosaur-only or mostly-dinosaur lines (or Colorata's Cenozoic box, their only prehistoric release this year). A company above the line is getting a lot of bang for its buck: its tree is long relative to the number of taxa. CollectA was below the line this year, but I promise it won't be for 2020! And whereas Mojo was just above the line this year thanks to a pterosaur and placoderm, for 2020 its 100% dinosaur lineup will put it below the line.

PS: I've updated the first post with 2019's winners. Let me know if there's anything I missed!

[Newt]

This, sir, is some intense paleogeekery. Kudos!

[Halichoeres]

This, sir, is some intense paleogeekery. Kudos!

As I tell my students, it's literally my job to be a geek.

I've added a graph of most companies' rankings to the first page.