GoToMeeting Auto Voice >> This conference will now be recorded.

Heather Tabisola >> It's good when I remember to hit record.

My name is Heather Tabisola, I'm the co-lead of the seminar series with Jens Nielsen.

And this seminar is part of NOAA's EcoFOCI bi-annual seminar series focused on the ecosystems of the North Pacific Ocean, Bering Sea, and US Arctic to improve our understanding of ecosystem dynamics and applications of that understanding to the management of living marine resources.

Today's talk is the last of our spring seminar series. We sincerely thank you for joining us this season.

Our fall speaker lineup will be found via the One NOAA Science seminar series and on the NOAA PMEL calendar of events. We anticipate the series to start in November. We have not confirmed the date as of yet, so just please stay tuned and watch those two avenues just to stay up to date.

Right now, please just double check that your microphones are muted and you are not using video.

During the talk, please feel free to type your questions into the chat, and we'll address those at the end of the talk.

So, today's speaker is Calder Atta. He joins us to discuss how he is using a comprehensive,
genome-wide dataset to address a complex history of disagreement between flatfish studies that have spanned more than a century. He is a graduate student at the University of Washington in the Tornabene Lab and a research assistant in the U Dub Ichthyology Collection. His primary work in the Collection involves facilitating the transferral and archival of larvae and egg specimens from NOAA's annual Alaskan ichthyoplankton surveys. And this talk is actually part of his first chapter of his Master's thesis.

I was also reading some behind, what Calder had put on Tornabene's website for his Master's as part of the group. And happy Earth Day, everybody. And since it's Earth Day, Calder had put this quote, and in his bio, and I just thought it really resonated for today, so Calder, I'm sorry for calling out on this, but it was really great. So, Calder wrote, "I believe that every organism has a story to tell that teaches us to appreciate the beauty of this finite planet and about our role and understanding and protecting its invaluable resources."

And with that, I turn it over to Calder.

Calder Atta >> Thank you, Heather, for that wonderful introduction.

So, as Heather mentioned, today, I'm going to be presenting on some of the work that I've been doing in Luke's Lab.

A lot of what we do focuses on biodiversity and biogenetics, taxonomy. And so I kinda want to get us started by just asking a very basic question, What is a flatfish?

Now, many of you are probably familiar with flatfishes. But when I think about the morphology of what is a flatfish, these three main characteristics come to mind. They have asymmetrical eyes. So they have eyes on one side of their head. They have an extremely compressed body. And they lie sideways, usually near on the sea floor.

And they also have these very long dorsal and anal fins that run the entire length of the side of their body. And that really helps them get more surface area to help them propel themselves through the water.

The flatfishes make up an order called Pleuronectiformes. They are very, very diverse. 816 species, and, and it's still growing.

Um, so with such a unique body plan and lifestyle, we would expect that all existing species would share a common ancestor that also had these traits. But there's actually been a lot of debate to whether that's actually true.

So this is a phylogeny that was published fairly recently by Shi et al., 2018. Um, and this is one example where I've just
set up several studies that have actually suggested that there are two origins of the flatfish body plan.

Um.

So, is this true, or is this an artifact of our methodology? Um, this, this debate has been going on for
awhile.

It's been suggested almost a century ago, and then it really picked up steam about eight years ago. And to this day, it's still not a well-resolved question. There's not a whole lot of consensus on it. Um, so...

And this, this problem in the flatfish phylogeny is only one issue and the flatfish phylogeny actually has a large history of many problems that have often not gotten talked about.

Um, and so, how did we get here? So, um.

Like we briefly touched upon, the flatfish, the history of the flatfish classification and phylogeny has gone on for well more than a century. And it be, to understand that we kind of have to go back to the beginning.

So early in the 1800s, when people are sort of just putting together the flatfish classification. You know, we didn't have as many species of flatfishes, so all of them were sort of put into one family Pleuronectidae. But then, over the years, many species got discovered. And then it's kind of hard to, people started realizing it was hard to place all of this diversity into one family, so they created more families. So they restricted Pleuronectidae to only flatfishes that people typically called flounders and made a new family for the soles, and then a new family for what are they called the Spiny turbots.

Um, but this wasn't the end, of course. New species keep getting discovered. Um, they're like, well, OK, we have to do
this again.

And so with flounders, they decided to split them up by flounders that were right-eyed. So had eyes on the right side of their head or flounders that were left-eyed, which were in the family Bothidae. So that's all well and good until you get
more species.

And then, over the years, they kind of decided to throw that whole model out the window.

You know, having eyes on the right side or the left side of your head is very easy to see for a human, but maybe it's not the most evolutionary relevant characteristic.

So now, they broke up the Pleuronectidae into all of these families and now we have these families for the Bothidae and then the soles got split up. And now we have about 16 families now.

Um, so most of that was using morphological characters, looking at things that look the same as other things and grouping together based on morphology. But eventually, people started using more quantitative tools and using Feiler genetics. And to try and untangle this phylogeny, but even with phylogenetics you run into a lot of problems.

The reason this group keeps getting broken up into smaller and smaller pieces, that with every new phylogeny that gets published, there seems to be conflicting results with past phylogenies. Um, so to understand that we kind of have
to think about how phylogenies are made.

So this is sort of an example of how historically phylogenies have been made.

So on the left you have a matrix of usually presence/absence characters that are morphological data, and then you use a computer to generate a tree that that is the most likely to produce the results of your data set. Um, so ideally, from the day so that you get the computer, the collective information that we would get from these characters would result in one, well-supported phylogeny.

But in reality, this is rarely what happens in, especially in the flatfishes, there is an extremely high number of trees that are very likely to fit the data. Um, so one way to decrease biases from any one character trait is to use more traits. So instead of using the morphological traits in this table, we could instead use genetic sequences. So that's what a lot of people that's what a lot of people have been doing for the last several decades.

So, this is just an example of a few of the most comprehensive data sets that people have done in the last few years. And basically, what I want you to take away from this is that there's a big disparity in how much dynamic material is used and how many genes are used. And this is somewhat problematic, because, if different, if different studies are giving you different results, that could stem from it them using different genes.

Genes evolved at different rates and are under different selective pressures, so your results are going to depend on what gene you select. So, one solution to this problem is, once again, just to get more genetic data. By sampling from more genes, the ultimate solution to that would be, to get up, get full genomes is for, for all of your species, for all of your species, but that's not very feasible. Doing this for many species would be very expensive, very time consuming. And you would probably break the computer trying to make a tree with all of that information.

So instead, what we've been doing or what we did for this was use a method called exon-capture. So exon-capture is a method that was developed by our collaborators in Shanghai, China. And the idea is that we have, we've already mapped the genes and a lot of model organisms, like zebrafish and tilapia. So we can use these these data sets to try and find the ones that are also present in the organisms that you want the study.

Um, so the benefits of this, they've already tested this to, or used this to work out relatively recently evolved species, and also deeper relationships. So it's kind of in that scope of evolution that we're work, that we're working at.

Um, so the set of genes that we're using comes out of this publication, Jiang et al., 2018. And in this study, they provide a data set of 4434 exons that that should work for any ray-finned fish. Um, so these are all single copy nuclear genes. The size range is pretty wide. Typically, they're around 250 basepairs that the largest is, as you can see, has up to
5000 basepairs. And the total genetic material that's in this is over a million basepairs.

So, if we looked at, if we, if we add this to the, to the previous graph. Um, we can see that, in comparison to the
previous studies, this has a lot more, a lot more data that we're going to, that we're dealing with. It's about a hundred times more genetic material. However, our, this method is fairly time consuming, so we're limited in the amount of species
that we could include.

So, this is by species, so, not the highest, but, we did manage to get 80, 89 species added in 816. And we tried to get maybe we made sure to cover most of the major groups that we got 11 families out of 16.

Um, our samples came from several fish collections so the University of Washington, University of Kansas, Louisiana State University, CSIRO in Australia. And these samples also came from from here, AFSC.

Um, so once we have the sequence data, there is pretty extensive computational work pipeline that, that, that we did to prepare the data before making trees. Um, and because of the uncertainty, sort of sort of hunting the flatfishes, we wanted
to try several different tree construction methods. Um, so, these were the two main methods. We did the concatenation method.

So, in this method, all the genes are sort of stitched together into one long sequence per sample. Then the gene is constructed, or sorry, the tree is constructed from this master gene sequence. Then, on the right, with the gene trees method, we created one tree per, per individual gene. And then we combined all of those gene trees into one species tree. Um, then after this, the last thing we did was we did tests to find genes that were most clock-like. Meaning they appear to have mutated in a in a linear fashion over time. So in theory, this should give you more accurate results. I'll show you how that came out.

And then with our filter data set of clock-like genes, we then applied both methods once again. So that gives us four trees in total. And here are the results.

So one of the aims going into this was that, by inferring from more data, we would see some improvement and clarity on what the true phylogeny looks like. But that didn't really happen.

While the trees sort of superficially look similar, they actually differ quite a lot in terms of how the families are related to one another, which has also been a recurring problem in previous studies.

So what this is showing that even with is that even with the same data, the result can differ based on the method that you use to construct a tree. Like in previous studies, we kind of found that many of the relationships are still still unclear.

So, where does that leave us? Where do we go from here?

So, I went back and reviewed previous studies again, and it turns out this is some, it turns out there's some consistency about what is unclear. So I decided to write down some guidelines on how we should be tackling this problem in the future.

So first of all, we've been trying to tackle the issue of limited data by increasing the amount of genetic material being used, but even at this scale, different genes give us different results. And if we think about it, if we think about
it, that might not actually be a big problem. The data is just data and it's just, and in the case of the phylogeny, some, in the case of phylogeny, sometimes there are rapid radiations that creates VC's very quickly and in these cases, in these
cases I think we should be asking ourselves how important is it to parse out the details of every speciation event? How likely are we to be able to find meaningful results and how much would it cost?

Um, so, in phylogenetics if we can't figure out who's more closely related to who, we can collapse branches into polytomies
and I'll show you what that looks like on the next slide. Besides using more sequence data, one way to gain information is to, to gain more information is to sample from a greater number of species. But, as said before, it's kind of a fea..not
feasible to sample from a large amount of data, and especially from 816 species.

So to workaround this, we can infer relationships at a very broad level, which has sort of what everyone's been doing so far.
But we can also focus in on specific groups and do more comprehensive studies where and focus on where there are knowledge gaps. In order to find the knowledge gaps and polytomies that are sort of created from these radiation events and are unlikely to be results soon. I used our trees and corroborated our data with previous phylogenies.

This was a pretty, this was a pretty intense challenge since the literature on a flatfish Feiler genetics and classification goes back so far into the 1800s.

Um.

And so, basically, this is the result of of combining our results, the results of many, many papers, with phylogenies of the flatfishes.

So this is a tree showing the how all the families are related within the order Pleuronectiformes. So there are two medium, main radiation events that I point, I point out in this tree. So the first one, the one on the left. Um, this is, this one is at the very base of the tree, and this is the event that's making us wonder if the flatfish body planner rose multiple times.

Then the second one, the polytomy in the middle of the tree, has five different lineages coming out. Looking at the phylogenies across publications, there's very little consensus on what's really happening in this part of the tree. So both of these areas, I think, are going to not be very likely to be resolved anytime soon.

And then if we sort of zoom in to the family levels, I also have family level trees. So these are the, the, the phylogenies, some of them to genus, some of them to species, they're well-studied, to the best of our knowledge so far.

The Pleuronectidae on the right is a very well established group at this point, and that probably doesn't need much more analysis in the way of species relationships, but there, I did find a lot of families that still needed comprehensive work.

So.

These eight families, I, these eight families really are data gaps in the flatfish phylogeny. There needs to be a comprehensive phylogenetic analysis with all species, or nearly all species sort of result or report on those relationships.

I've also compiled a list of where...

[Unknown speaker interrupting] >> Cautiously take it.

Calder Atta >> I've also compiled a list of genera, where one or more studies have suggested that they were that they could be invalid. So I would say all of these genera need to be re-examined at some point. And then the last list I have are genera that have never been included in any molecular phylogeny. So we don't know where they go.

We know, we know, we know they're within their respective families, but we have no idea where they go on the tree within those families. Um, so, and and my hope going forward is that we can use this information to guide us on where we need to focus our attention, and then you don't have to tackle each of these groups individually.

If you did a comprehensive analysis of the Bothidae, you could include all of the genera for the Bothidae and sort of knock those out at the same time.

Um, so, for me, personally, I, for, for my future plans, I'm working with our collaborators, and we are compiling a database of exon-capture data. So, it's not just flatfishes but with all fishes. So, our data is going to be part of that that
data set for future studies.

Uh, the other potential thing we could, we could do is GGI, or gene genealogy interrogation. And this is a technique to sort of look at individual genes to see how much each one is contributing to the final tree and if there's potential bias. So that can sort of tell us where, what's, what's, what's causing some of this specific patterns that are conflicting with each other.

Then, with the trees that I produced, I was going to do time calibration. So time calibration is basically mapping the
phylogeny onto a geological timeline, and we do this typically with fossils. And so you use the fossils to put on a date
on certain points on the tree, sort of scale it to to real geological time.

And then after once the tree is time calibrated, we can ask more interesting questions about how flatfish has evolved and moves spatially through time. And so this is sort of hinting at chapter two of my thesis, which is looking at Alaskan flatfishes. How certain Alaskan flatfish species have evolved in the North Pacific.

So think with that, I'll take start taking questions.

Heather Tabisola >> Thank you so much, Calder.

That was really good I clap for everybody because I know it's weird probably giving a talk and not hearing that.

All right, questions, please type them into the chat and Jens and I will monitor that, and then facilitate that with Calder.
And, again, just a reminder too, this is the last of the EcoFOCI talks this season. So, we will see everybody in the fall, and dates for that will come out later in the summer.

Calder, you...oh! Here we go. Ok, I was gonna say you stymied everyone.

So, here's a question from DS: Any way to filter the data set to improve the signal to noise ratio?

Calder Atta >> So, filtering noise. Um, so, so, one way we could do that would be the clock-like filtering.

Um.

Basically, as, uh...Works by selecting genes that, um...By selecting genes that are that are sort of, I guess, that seems to be accumulating mutations more evenly.

Uh. That's the, that's the goal, is to reduce the noise, and remove those, remove genes that are sort of behaving more radically.

I'm not sure if that completely answered your question, but, yeah.

Heather Tabisola >> Thank you, Calder.

People are slow to type in questions today. So I know it's a rainiest day in Seattle, so maybe everybody's just still asleep.

Calder Atta >> Maybe.

Heather Tabisola >> I had no idea how many times, like the phy...like, how many different classification groups. Like, how many times that has been moved through. That doesn't surprise me, but it was fun to learn about that.

Calder Atta >> Yeah. It was fun for me to dig through, dig through all of that.

[Laughs]
Heather Tabisola >> Yeah? Was it? Was it really fun?

[Laughs]
I liked how you laid it out on the slides.

OK, Cause men, do species from the same family encounter similar environmental variables?

Calder Atta >> Oooh!

So, on a family level, it differs widely. Let me go back to the...This picture, so, these are the 16 families.

So just by looking at how many species are in it, you can sort of get a grasp. More species are going to have a bigger, you
know, overall family-wide geographic range. Some species...So the Pleuronectoidei which is probably what most of, most of, what we are familiar with. Those are permanent in the northern, temperate and polar, northern hemisphere.

Um, and so typically with that you're going to see there's still going to be a variety of substrates that they're going to be on, whether it's like, like more silt versus more rocky. But a lot of these are so that so there are there are temperate and polar species.

The Achiropsettidae with the four species, are only known from Antarctica. There are a lot of species that are tropical
too. So the, the Bothidae left-eye flounders and the Rhomboso...oh, well Rhombosoleidae are a little more temperate Australia. But when you get into the tropics, you know, and you're spanning the global, the global tropics, that's a lot of area. Um, and you're, you're, in the Pleuronectoidei, you're dealing with species that are going down as deep as a thousand meters, so there's, there's a huge variety in terms of its environmental variables as well.

Um, yeah.

Heather Tabisola >> Thanks, Calder. Colleen says: Thank you for the great talk, Calder! I'm excited to see the step with the fossils. Where do you plan to get the fossils from U Dub Collection or beyond?

Calder Atta >> So, the fossil...there aren't very many fossils right now. Right now, I have a list of 24 fossils. So I wouldn't, I wouldn't physically be handling the fossils. But basically, I guess this would be another, dig through literature, dig through what other people have used for fossils.

One thing you can do for time calibration is also use other people's trees. So basically, you just have to put a timestamp
on a node and then how you want that node to be restricted to that time, that timestamp. So, basically, the process would be going going through the literature, going to finding finding publication that describes a fossil and then what what time period it was from. And then placing that on somewhere on the tree. And by using multiple timestamps, you can sort of calibrate it. It's not the easiest process. There's definitely a lot of error associated with it.

One, one challenge is that you need to have and I'm doing air quotes, "accurate tree", so something I have to decide is what what I think is the best tree sort of to use would be for that.

Yeah. I wish I were handling the fossils. That would be really cool.

Heather Tabisola >> Thank you, Calder. Thank you for the question, Colleen. CJ, I see your question. I'm actually going to save that one for last. I'm going to jump back to it.

A question from Melanie: With the families that need further analyses, among those are they particular speciose relative...

Wait, I don't know if I read this right. OK, wait.

With the families that need further analysis. Among those families, are they particular speciose relative to the other better resolved families?

Calder Atta >> Um...It's a bit of a mix.

So, the three biggest families just looking at, again, looking at the numbers, the Bothidae, the Soleidae, and the Cynoglossidae, definitely all need thorough re-examination. There doesn't seem to be too much of a pattern in terms of which ones were studied.

The Pleuronectoidei has been studied a lot, just because it was that first, it was that first family that included all
of the flatfishes, and as it sort of accumulated all these things based on arbitrary, somewhat arbitrary "this thing looks like a flounder", it accumulated a lot of like, the classification accumulated a lot of structure that was misleading based on the morphologies that people were using.

So, there, so, once genetic, genetic techniques picked up, really, people are sort of hammering out trying to figure out the Pleuronectoidei. I think some of this, some of the groups that are sort of hard to come by, the Achiropsettidae, the Antarctic ones. No one's ever looked at the interrelationships of those species, or just those four species.

Um.

But, I mean, it's a lot of work. If it, if I, if I go back to this slide, on the bottom, I'm showing you sort of the numbers of, so, so, the families that need comprehensive analysis are eight out of 16, so, so half of the group and, of course, three of those are our, have over 100 species.

Um, and then looking at the, the, the, the two right columns. So, just general with, with problems that need to be addressed, we're looking at 45 out of 127, so it's a, it's a lot. A lot of work needs to be done.

I wish it were easier to collaborate with people, because this is such a large group, and they're global, right? So, we have to look at different collections. Um, especially groups and that are harder to find and in remote areas are going to be harder to get it...uh, get ahold of.

Yeah.

Heather Tabisola >> Thanks, Calder. Thanks for the question, Melanie.

Morgan just had a statement and he said: I really like your comment that there are some consistency about what is unclear.

That is the fundamental reason for the study of ichthyology. I would probably add the fundamental reason
study for most science.

And so I'm going to jump back to CJs question here, if nobody else has other questions. And CJ's question is: How could you use what you've learned from a systematics perspective to help inform fisheries management?

Calder Atta >> Mmm.

So. This is sort of the taxonomy response to to fisheries management. A lot of these, a lot of these species, especially
in the Pleuronectoidei, and so, if we want to talk about Alaska, I think there are 23 species of Pleuronectidae there and there there is also, uh, Paralichthyidae, I think, as well.

So for species and a lot of those species are commercially harvested, um, and relating it to how this affects fisheries, I would say that taxonomy, you need to have taxonomy as the foundation for any sort of getting any sort of ecological information.

So, for example, if you had a population that you thought was one species, and you thought it was doing well. But then you find out by doing population genetics, that they're two populations. And then, even more so, if there are two species.
You know, your numbers are going to completely change if they're, if they're not...If it's not one continuous population, then how you interpret population size is going to differ dramatically. And so so getting all this figured out is
is it provides, like the foundation for all other ecological questions.

I hope that answers.

I hope that's a sufficient answer.

Heather Tabisola >> CJ does say: Good answer and good talk. Thanks, Calder.

Heather Tabisola >> All right, if there are no other questions.

Calder, is there anything last minute that you want to add?

Calder Atta >> This was the first time I've given a remote presentation, and it was an interesting experience.

Heather Tabisola >> Yeah, I think you did a great job. I know it's weird not seeing people, but,
um...

Calder Atta >> Yeah.

Heather Tabisola >> People do tend to be, they are engaged.

All right, so Calder, again, thank you so much for joining us.

Thank you for closing out our season and to everybody else, we'll see you in the fall.

Thanks again!