Fish shapes and scales

I thought I would take this talk to explain a little bit about the research that we're doing with these fishes. And the fish that we're measuring right now will go into a database, a second generation of what we're calling the fish shapes database. And so I thought I would give an introduction to the original database, what we've used it for, what we have learned by measuring all these different fishes, and then move on to what we hope to learn in the future with this new data set.

We were inspired by the absolutely incredible diversity of teleosts. They account for about half of all vertebrate diversity in terms of species number and what really interests us are their morphology and particularly their body shape. And I'm just showing some illustrations here in the different silhouettes. Fishes have incredible diversity of body shapes. We have very deep bodied, very elongated, very flat, very wide, things that are almost globular and then you have seahorses, which really just don't look like fish at all. And so, if we want to try and understand the evolutionary processes that generate this diversity, we first have to be able to quantify the diversity and then analyse it in a phylogenetic context and that's why I have this big phylogeny up here. And so this is a phylogeny actually including all of those 6,100 or so species that we collected in our first data set. And the kind of data that we're taking are very sort of simple morphological measures.

So we've got generally sort of lengths, depths, and widths of the fish, sort of overall standard length and depth of the caudal peduncle, which is important for locomotion of fishes. We then also have length and width of the lower jaw and the mouth, again important for feeding and we've got overall maximum width, maximum depth, for example. We also have information on the longest and the first spine in each of the fins, but I'm not going to be talking about that today. I'm going to be focusing on these nine different traits. We also take photographs as well, so we can then go on and do geometric morphometrics on those photographs. In fact, we've just started a collaboration with a group of doing that automatically through machine learning.

We measure these different traits, and as Amanda said, we spent six, seven months at the Smithsonian. We took a lot of undergraduate researchers. We were there for three different summers. You get exhausted if you’re there for more than a month, so we rotated out and we had so many people involved but it was all worth it because we managed to collect a huge amount of data. So as Amanda said, we collected over 16,000 specimens. We have currently data of 6,151 species, and we have a very broad sampling: 90% of extant teleost families and 96% of living orders and definitely while we’ve been here, we’ve definitely been able to increase some of those family levels. Though we had some here that we weren’t able to get to the Smithsonian, so that’s pretty exciting.

And of course, it involves a huge amount of work by a lot of different people. So this is everybody that came to the museum for at least a month. These are all co-authors on the paper, and the data is now available for everybody. So we’ve published it, everybody has access to all of those 16,000 different specimens and our measurements. And most of these were undergraduate researchers, there were only six of us that are not. My Co-PI, Peter Wainwright, three graduate students, two of Peter’s, Sarah Friedman and Katherine Corn, my master’s student Kat, and my Postdoc, Olivier Larouche. All of these people were really integral to generating this data. And feel free to go and download the data and use it for all your exciting fish projects.

Though we had measured all of these specimens, how do we analyse it? Well, first of all, we have to take into account size. We just can’t use absolute measures. These aren’t even pictures of the largest and the smallest fish in our data. But you can see that if we just use absolute measures, this fish is, we’re going to find out this fish is smaller than this fish. Not very exciting. But if we looked at the relative size, maybe their mouth is relatively large for the size of the fish relative to this fish. And you can remove size in various ways and we've actually used multiple different ways, but I'm going to start off today showing you data where we've used log shape ratios. And that's how you calculate the trait divided by the geometric mean of size and how we're measuring size? We're using standard length as well as maximum body depth and maximum body width. So, we have an overall size dimension. Commonly in fishes, we'll often use standard length or total length but that is part of the shape, that's really just part of the elongation of that fish. So, we've used these, a combination of these three traits as our size measure. So, once we've done that, we can actually start to look at the patterns that we see in these 6,000 or so species.

So the first thing, obviously, we want to do is build a morpho space, look at the variation of that shape, and see what the primary axes are. And so this is a principal components analysis taking into account the phylogeny. And these are the first two axes, and you can see they explain similar amounts of variance, 22% and 21% so we've got two very important axes. This first axis is a width axis, so we're describing sort of these laterally compressed bodies all the way through these very wide dorsal eventually flattened bodies. The second axis is elongation; so down here, we have the very short and fat and here very elongate and shallow. And these colours hark back to the phylogeny so they show somewhat of the relationship. So you can see these blue colours here, these are all the eels. And if you look here, this is showing a density plot of this and so you can see most fishes reside in a very tight area of this morpho space but because we have so many species, it looks like they're pretty well spread out. But this really shows that they're all in this sort of very tight area here.

And just to illustrate what some of these extremes look like, we have, of course, down here the very deep and very flattened flatfishes. We've got our very elongate eels and our pipefishes. We've got our lophiiformes and the very wide fishes over here. And then these widows up here, which are both wide and elongated, are the banjo catfishes. So we now have quantified the diversity of shape across 6,000 teleosts. But then it's interesting to know, well, why? Why has this evolved? I'm really an evolutionary biologist with a very strong heart of a natural historian. And so, if we think from an evolutionary biology point of view, any shape that enhances fitness, that is, it positively increases survival or reproduction, can be acted upon by natural selection and thus a box. And if we think about shape, shape can influence a variety of different sort of performances that influence survival and reproduction. Locomotor performance, that can help fishes both feed and capture their prey as well as prevent predation. Obviously, being able to feed, you need food to be able to survive, you need to be able to avoid predators to survive as well. And then, of course, you’ve got mate attraction and mating success that can increase your reproductive output. And so here, these provide the link of body shape to fitness. But the precise body shape that is going to enhance the performance is going to depend a lot on the ecology and the environment of the fish itself. Things like habitat complexity, the water temperature, the predation pressure and type, what they eat, how they eat, how they swim, all of these different things are going to influence the performance that maximizes fitness and thus the body shape that will be selected for. And so what we have done since collecting this data set is spend a lot of time looking at a variety of these different sort of ecological variables. And I thought I’d walk you through some of the findings that we have managed to generate so far with our current data set.

So we’re going to start off with habitat complexity. And the result is that fish shape is influenced by habitat complexity. And this makes a lot of sense because we know from sort of biomechanical and hydrodynamic models of fish swimming that fish turn in curvature, so their ability to manoeuvre around obstacles is related to their body depth. The deeper the body and the deeper the caudal peduncle, the more manoeuvrable they are. Whereas if they have these wonderful fusiform shapes of the tuna here, absolutely great at fast endurance straight line swimming, but not good for manoeuvring around obstacles. If you have a highly complex habitat, you’re coming into contact with obstacles. So, we would expect that fishes living in highly complex habitats are going to have deeper bodies and less fusiform. And how do we do that? Obviously, you can measure habitat complexity through rugosity, but you can’t get that data for all these species very well. And so, we just took a very simple approach of just looking within marine fishes and comparing those that live in reefs, which are undeniably the most complex marine habitat, to all the others. And when we did that, and we looked at a variety of different traits but I’m just summarizing here using fineness ratio. So fineness ratio shows how different, how streamlined the fish is. So lower values indicate a deeper body. Intermediate values are going to be this tuna shape, and then very high values are going to be the very elongated forms. And so, what we find is that our fineness ratios are average higher in the open habitat, just as we would predict and we've got deeper bodies in the complex habitats. And this was actually a study that was done by our undergraduate researchers, led by my Postdoc, Olivier. And all of the work I'm really presenting has been done by the graduate students and Postdocs. They have done such an amazing job with all their work on this project. So within reef fishes, we have deeper bodies relative to other marine habitats.

Interestingly enough, when we look just within reef fish, the shape is also related to their colour pattern. And this is work by my master student, Katherina Zapfe. And here we're showing fineness ratio just as we saw before. But we're comparing fishes that have these vertical bars to those that have horizontal stripes and these fishes with these deeper bodies have more likely to have these vertical bars. So barred fishes have deeper body than those of stripes. And this is consistent with sort of expectations of potentially how these lines are breaking up the body and making it more difficult for predators to see them. And if you're particularly interested in this topic and you're going to the International Vertebrate Congress up in Cairns next week, I'll be talking more on this there.

Not only can we look at the overall shape and how they differ, but we can look at the rates of evolution of shape. And this is a project led by Katherine Corn. And she found that the rate of reef fish evolution in terms of all those different body proportions that we measured is influenced by feeding mode. And so this is a very complex diagram over here. Focus mainly on the orange here. This is the biting feeding mode. So these are fishes that are pulling off attached benthic prey. So they're using their head and their body to rip attached benthic prey. And we're going to compare that to suction feeders. This is the blue here. And the majority of fishes, this is what we generally think is the way fishes feed. But you can see that there are a variety of things and biting has evolved many times across the phylogeny. And when you look at the rates of morphological evolution in these different feeding modes, Katherine was able to show that there's about a 1.7 times higher rate of evolution within those attached feeding fishes, those that bite, relative to the suction feeders. And that's particularly interesting because biting only appears to have evolved in extant groups since the KPG. So we've got these sort of changes in rates associated with changes in ecology.

And moving beyond reef fishes and now looking across all of marine fishes, we can go back and look at differences in shape again rather than rates. And here we're looking at the very deep fishes. And this is led by Chris Martinez, he was a Postdoc in the Wainwright Lab, he now has his own lab down in Southern California at UC Irvine. And he looked at this and found that water depth has a huge impact on fish shape. And to a lot of ichthyologists, this is not a surprise. We have all these real weirdos down in the deep sea. But he was really able to show that you get a lot of variation in body shape in the deep sea but it all sort of involves adaptations to being able to eat whatever you come across because there's not a lot of prey down there, so you have these large mouths. And often you get these sort of very elongated forms with tapering tails. And that is likely to do to not needing to have such strong sort of locomotion in the deep sea. And it's very hard to become a deep-sea fish. You get the number of transitions into that deep sea are very low. So not a surprise, but we've been able to provide quantitative evidence of the general observations that have been in ichthyology textbooks for a long time.

Not only have we looked at sort of absolute depth, but we've also looked at how the position that the fish takes in the water column, whether they're on the bottom, whether they're demersal or whether they're pelagic right up in the water column, can influence body shape as well. And this was led by Sarah Friedman, grad student now working, now has her PhD and working at NOAA up in Seattle. And so what she found is that it's the benthic fishes that really contribute to the diversity of body shapes across fishes. So you can see here in the orange, all of our extremes are sort of represented by these benthic fishes. So we have a lot of variation in our body shape here. And this is just showing marine fishes, but we see exactly the same thing in freshwater fishes as well. But very interestingly, Sarah did a follow up paper to this and showed that the evolution of this sort of benthic pelagic adaptations actually are driven by two different evolutionary processes. So in the marine realm, the morphology that allows them to adapt to the benthic environment is very phylogenetically patterned. It means that these lineages sort of radiated into a particular niche. And so benthic adaptations sort of happened on a per lineage basis. But in the freshwater, you get a lot of repeated evolutions of various different benthic forms. So they didn't move into that and then radiate. There’re multiple different lineages evolving into the multiple benthic forms. And that makes sense when you think about the sort of dynamics of the environment. Freshwater has a lot more vicariants events, much more likely to undergo speciation. So you're probably getting repeated evolutions of that.

So that was looking at both freshwater and marine, and then also we have looked across all of our data as well, looking at how fish shape is influenced by swimming. So, there are multiple different modes, but you can kind of break them up into two major ones. The first is sort of a body caudal so they're using the body and the rear tail fin to be able to create most of their thrust. And others are made medium paired fin swimmers using the dorsal and anal, the pectoral or a combination of those fins to swim. And perhaps again, not surprising is that we see very different shapes between these two different forms of swimming. Those that swim with body caudal locomotion are sort of much more streamlined and elongated. And those that swim with medium paired fins are much deeper bodied.

So, we've been able to do all of this work only by working at the very largest of scales. And we can only do that by using museum collections. I know I am preaching to the choir, but museum collections can really help us do fascinating sort of macro evolutionary studies. So it's not possible without access to the museum collections. And I just want to really acknowledge the amazing people at the Smithsonian, Kris Murphy, Sandra Raredon and Diane Pitassy and Jeff Williams, who made us incredibly welcome, put up with us for three years, filling their corridor with all sorts of fishes and measuring. And so we really appreciate that. So now what is next? Why are we now here at the Australian Museum collecting more? And we have like 20 to 25% of extant teleosts. How are we going to add to this? And so this grant finished up in March 2020. That's probably a very familiar date to everybody because something else happened then, right? This was when the pandemic happened. And that stopped me having any access to collections for the next couple of years, and as we know, there's publish and perish in science. It's also really important to get funding, particularly I'm going up for tenure this year. I needed to be applying for grants, but I didn't have access to the collections to do anything new. So I had to get creative, and think about this dataset that I already had, how I could use that for preliminary data. But as you all know, when you’re trying to get data out of governmental research, granting agencies, they’re not just going to give you money to continue working on something. You've got to have a very new question and a really good reason for collecting more fish body shape data. So how could I do that but have a very new question?

And the one thing that I hit upon was looking at scale and not as you might have thought in terms of scale. Fishes have scales. They do indeed. But I'm actually talking about differences in the units of measurement. And so if we think about it, ecologists are really good at thinking about spatial scale. We can think about sort of communities or regions or bioregions, and they spent a lot of time looking at how scale can influence their results and the sort of processes that they identify. For example, in community ecology, at very small scales, people often find it's biotic interactions that drive community composition, whereas at much larger scales, it tends to be abiotic. So, we sort of have identified sort of scale dependency of our results in terms of spatial scale. But there's another aspect to scale, and that is the phylogenetic scale. So what taxa are you working on? Are you working on a particular family? Are you working on a particular order? Are you working on teleosts? Are you working on vertebrates? And how does that scale influence our results? And so I was able to persuade the National Science Foundation that we could build upon this fish shapes dataset and use that to then be able to start answering these questions about the impact of phylogenetic scale on our analysis. So that's what we're doing here, is that we're now going back, we're building on our fish shapes dataset that had all teleosts and we're now going back in and collecting a lot more information on sort of individual genera and families. So my specific questions really are, are the patterns of quantitative trait evolution, that just means continuous characters (all the things that I've shown you like body size, depth, length, head depth, all those sorts of things) are they identically random or do they vary systematically across phylogenetic scales? So do we see the same thing in every genus and then when we scale up into families and into orders and beyond?

And so we're targeting families and genera with lots of species in the collection. That's what we're doing here. This is my team. You may have seen us wandering around the collections or the corridors. And so far, in the last sort of three and a half weeks, we have collected and measured 1,164 specimens across 651 species and 81 families. And we've still got a few more days left and we're hoping to increase that. And it's been really wonderful because we've been able to see some families and get data on that we wouldn't have been able to get anywhere else like the Galiixidae and other things, so it's been really wonderful to be here in those collections. And the idea is that we will then be able to do the kinds of analyses I presented those results on. So all the results I presented on have an extent of all teleosts. So we have a very large scale extent and our grain size with species, those were the sort of individual units that we were looking at. But we have many other ways of looking at this data set. We could instead of looking at species, we can increase the grain size to look at genera. Why would you want to do that? A lot of palaeontologists can only identify species down to a genus level. And so it's interesting to see whether if we use genera instead of species, do we get the same results? And similarly, we can do that in families and orders for grain sizes and then also the extents. And so that's what we were going to be doing. We're going to be analysing all this data, looking at the influence of scale and not only just based on taxonomy. I know this is a museum and taxonomy here is very dear to everyone and it's such an important part of biology. But some people take issue with that and find that, you know, what is a species? What is a genus? And they sort of think, well, maybe we should take a more phylogenetic approach to that and take particular clade ages or sizes or tree depths. And so we're actually going to be able to compare taxonomy to phylogeny as well. And so that is what we are hoping to do with this data. So it just remains for me to thank you all for listening!