Salmon Survival: Transforming Aquaculture with Predictive Science

Corinne

Welcome to 'Salty Talks.' I'm Corinne, the communication specialist at the Aquaculture Research Institute and Today, we’re going to hear from Dr. Heather Hamlin and her graduate student, Halli Bear. Their work in the field of Atlantic salmon aquaculture, focuses on developing predictive metrics for salmon broodstock and offspring quality. We’ll take a closer look into their research and the potential impact it could have on sustainable aquaculture practices.

Heather

So I'm Heather Hamlin. I am a professor in and director of the School of Marine Sciences, and I'm also a member of the Aquaculture Research Institute. My background is years ago to put myself through college, I actually worked in a pet store and the fish and culturing the fish and understanding all of the different species. And you know, that just became sort of one of my passions. And so I managed the pet store to get myself through college. But in doing that, I met someone who came in on a regular basis, and she happened to work at the university, and she worked for one of the professors there and said, you know, you really should come in and talk to her name was Linda, my boss, and who is a professor, a professor in what is now the School of Marine Sciences. And so I went and talked with her and just became really interested in what she was doing.

And that's how I started in actually my master's. I didn't even know about aquaculture before that. I just knew fish were cool. And then after talking with Linda, I realized, oh my God, aquaculture is this field, this whole, you know, field that I could actually have a job in. And then sort of my passion became, oh my God, we could feed the world like aquaculture. Is this really cool opportunity to grow fish? Um, which I was learning more and more is just such a great protein source, more than, you know, than a lot of terrestrial animals for a bunch of reasons. And so that just sort of became my passion.

From there, I got a job at NOAA, and then from there I worked at Mote Marine Lab, and that's sort of where I had my first real job, postmasters. And that was at something called the Mote Aquaculture Park, which is part of Mote Marine Lab. In doing that, I became the production manager for a commercial sturgeon demonstration farm. With that, that just really ignited my sort of passion for growing fish to feed people. And in that case, it was a demonstration facility in which they were trying to demonstrate that you can grow fish, but in a really environmentally sustainable and ecologically responsible way. And so we did it on a culture farm where we were trying we were aiming for 5% water reuse, which is really low. And so it would be really sustainable. And it was a really exciting project. And so I worked there for. Nine years a crazy long time. And in that time I also got my PhD through the University of Florida. In that, I met someone named Lou Gillette who became a tremendous mentor to me, probably the most profound mentor I've had. And he got me really interested in combining aquaculture with something called endocrine disruption. And so this is understanding how chemicals in the environment can alter your endocrine system or your ability to produce hormones in a way that can impact the reproductive status of the fishes. That was super exciting. And it sort of opened up this whole new opportunity of, oh, I can look at toxicology, I can look at reproductive function, I can look at all these different things through all of these cool lenses.

And after I got my PhD, I did a postdoc with him, and then I went to the Medical University of South Carolina and the ob gyn department, which seems really sort of this weird and secure test route. But it was I was part of a group that looked at maternal fetal interactions but using alternative models. And so at that time, I was working at the Hollings Marine Lab and trying to use fish and alligators, strangely enough, but as models to look at questions of human health but that just aren't in humans, like people who are like dolphins, for instance, they can dive deep and then when they come back up, they can re inflate their lungs. Like if we tried to collapse our lungs and re inflate them, they would not do that. Right. And so again, this wasn't my project, but this is just the types of things where you can use something like a dolphin to ask questions about human health. And so I was asking questions about how environmental contaminants might impact impact endocrine function. Humans have a very similar system. So it was just it was a good fit.

From there I went to the University of Maine, which is where I wanted to be the whole time. But getting a job in the location you really want to be that often doesn't match up, you know, job opportunities and where you want to be. So I was really lucky when an opportunity came up through the Aquaculture Research Institute to have me end up where I am now and done. Background check.

Halli

That was awesome. I learned so much in those few moments. Really. Hi, my name is Halli Bear. I am a graduate research assistant in the Hamlin Lab, and I'm working on predictive metrics for Atlantic salmon Broodstock and offspring quality. And I guess a little bit of my background is that I did an undergrad degree in marine biology at the University of New England. So down in Biddeford, Maine, and there I was in the marine biology program. I worked in a pinnipeds ecology lab. And so I really focused on pinnipeds and fisheries interactions. So we were actually going through gray seal scat and identifying the otoliths and trying to kind of let the fisheries know what exactly the seals were consuming, because everybody was up in arms and thought it was the Cod. And that's why the fishery was going down after that.

When I graduated, I went to work as a technician at the Ohio State University called Stern Laboratory, and it was on the western basin of Lake Erie. And we were doing water quality work to identify what was going on with their harmful algal blooms. So they had a cyanobacteria bloom that was producing microcystin, which is a liver toxin. And it really got so bad that they had to shut down the public water to Toledo and a Port Clinton folks. So I worked on that for two summers, and we really just. Kind of tried to see what was going on with nitrogen and phosphorus, since it's near a lot of agricultural communities.

Once that season wrapped up in November, I ended up kind of getting more on the route towards fish and had a really kind of wonderful, strange opportunity to adjunct at a community college in Nelsonville, Ohio, and ended up teaching ichthyology. So we really just focused on stream species of southeastern Ohio and kind of learned about how the history of glaciation, of the streams and rivers local to that area served as good niches for the fish. I did that just for one semester, but it was a really awesome opportunity to learn and be taught by students, and also just kind of have like a light within me come back realizing that I like to squeeze fish.

And so Covid happened, took a few years doing some service industry work and you know, was down in Athens and just happened upon Heather and Scarlet. We ended up having a conversation about Fish and Maine and somehow, like, I don't know if things crossed paths, things lined. And we just continued that conversation through email and zoom turned out to be an opportunity to go back to school and get a graduate degree. So here I am.

Corinne

The heart of today's discussion revolves around a USDA funded grant, standing at the forefront of aquaculture research. This project has many layers, focusing on identifying biomarkers in Atlantic salmon, which could change the way we predict and manage broodstock quality and offspring development.

Heather

Yeah. So this particular grant is with the USDA and it's something called the Nifa program. And its goal is to really help identify something that we call biomarkers. So it's either parts of the fish's biology or behavior or something that we can use as a metric to predict whether or not they will be either a good performing broodstock or maybe the cohorts of the group of siblings within a certain cohort is what we call them can maybe be fast growers or some other quality metric.

So our whole goal is to identify these biomarkers in a couple of different groups. So one in what we call the broodstock, the parental fish that give rise to the remaining to their offspring. And so in those parents we're looking particularly at females. And we're trying to use a whole bunch of different metrics to understand if she will be a good breeder. And then we're trying to look at her offspring. So the embryos that she produces, we're trying to look at a whole bunch of different biomarkers in the offspring, both in the embryos. So before they hatch and then in what we call the alevins or maybe fingerlings or these later stages to see if we can identify markers that can predict cohort quality. So, will this cohort have good survival or will they grow fast. Those are the kinds of things that are really useful.

Because if we can identify biomarkers that correlate really strongly with these predictive indices, then we can say, oh, this cohort is not going to be very good. It might make sense now while they're, you know, really small, maybe even before they've hatched to cull this group or maybe later we don't put it we can use that to predict, well, we're going to keep them, but we know they don't grow well. So we're going to use that to know we need more eggs, or we'll just be able to sort of predict long term financially what that's going to mean for us.

So having that predictive capacity is really, really important. And then for the broodstock females. That is really important in identifying females that you definitely want to breed, right? Because those are really those that are just really critical for any kind of predictive capacity for a facility in their hatchery production and hatchery management.

Corinne

In the realm of salmon aquaculture, the challenge of maintaining high embryo survival rates has become increasingly pronounced. Dr. Hamlin's research stands as a timely response to these industry hurdles. Historically, the industry has prioritized growth and disease resistance, but this new approach underscores the importance of the breeding female's role. Understanding and identifying a good breeding female early on is crucial, as it sets the foundation for the entire production cycle. This insight propels the field towards a more holistic and effective strategy in salmon breeding, offering hope for addressing the decline in embryo survival and improving the future of salmon breeding.

Heather

Each facility and I say each there's really not that many salmon breeding facilities, you know, in the world. But at the facility in Franklin, um, they use a number of different metrics, but they have something called an estimated breeding value or EBV. And that has changed over time based on what they're prioritizing. But overwhelmingly their key metric is growth. Maybe don't quote me on this, but I think right now it's like 70%. Wait, is that correct? Yeah. Has a 70% wait on growth. And then I think they also take into account, um, I don't know if it's disease resistance, particularly sea lice and some other things. I think flesh color has also been, you know, because, you know, pink and, um, has also been within the breeding value, but really it's been primarily focused on growth. And so the biomarkers that we are trying to identify and focus on may not necessarily correlate with that very well because one of the issues is fish. They only have so much energy, right. Like all of us. Right. We only have so much energy. And if you're putting a lot of your energy into growth, you necessarily have less energy for something like reproduction and putting in a whole bunch of energy into your gonads, right into the ovary and the female and testes in males. And so you put all that energy in and it's hard to say. If so, right now, if we're selecting for growth, does that mean potentially that we're not necessarily selecting for the best breeding fish? Because one of the issues that we're trying to solve is that embryo survival has been steadily declining for the past 15, 20 years. And so farmers used to be able to rely on 85 and 90% embryo survival. So every female, when you spawn that female, say 90%, 85 to 90% of those fertilized eggs would result in hatched embryos, hatched fry or larvae at the time. But now it varies from year to year, but on average maybe 55%, which is so low. So that trajectory is you know, it's kind of gone up and down. But now that average is down at the 55 range. And so that's pretty it's pretty low from a historical high. And so that's one of the things that we're trying to identify is okay. What biomarkers can we look at that might be able to predict that embryo's arrival. Because that is really, really important for a couple of reasons. Because obviously it takes a lot of energy and a lot of effort to breed those fish to create those larvae. If you're only getting 55% versus a circle 90%, that's a lot more effort sort of per unit volume. The other thing is a lot of the farms will sell surplus eggs, so you can only fit so many in your facility. But if you can spawn more, you can sell those to other facilities. Right. Who may not have the breeding females. But you can, you can, you know, that can be another source of revenue. And so when you can't do that that's another loss. Right. So not having high percentages of survival is something we call IOP. And so you just say AIA because in salmon you can't see through the egg like you can and say like a zebrafish or something. That's clear. So that's the first time at which you can see, oh, I can see this dark spot, which is the I, I know it's growing, I know it's developing. And I also know that from here on, the odds of survival and hatching are pretty high. So that's a really key sort of marker in embryonic development.

Corinne

Halli research offers a complementary and equally vital perspective to their research goals. By tracking the development of salmon embryos at various stages, Halli's work zeroes in on identifying key developmental milestones and potential issues. This includes assessing embryo viability, morphometrics, essentially body shape and size, as well as responses to environmental stressors. These observations and data collection help illustrate a fuller understanding of salmon development, directly contributing to the overarching objective of improving broodstock quality and offspring survival in aquaculture.

Halli

With the reduction of the egg viability, um, and the loss of eye up numbers. Um, kind of the take that we've been doing in lab is trying to really hone in on when they are dying in order to kind of create some metrics and flesh those out. So, um, we know that the viability is reduced at IOP, but, um, finding when that's happening can really, um, determine some of the metrics. So, um, last year, uh, in the preliminary data, I will say, um, we looked at morphometrics of eyed individuals and got egg volumes. And then at hatch we got total lengths, eye diameter, yolk sac volume. And those are metrics that could be included in kind of growth and trajectory. And kind of the beauty of working with the USDA is they have those cohorts living on in their facilities, so we can follow them at a year when they are PA like. This year, we plan to get some total length, eye diameters and see if there's any correlation with, um, the growth that we see when there are embryos moving on as they grow and, you know, subsequent developmental stages into, um, older adults. And I guess I'll just talk a little bit now. So we got eggs. Last week, a week ago on Wednesday from Bingham, and we are rearing them up at the DRL. And I have been pulling eggs every other day, sometimes every day, and have been looking at them to see kind of how they are developing and keeping track of those mortalities. Um, a lot of people have shown that eggs are extremely susceptible and sensitive during gastrulation periods, which will start for our eggs tomorrow. So we'll see kind of how that goes. I think once we get a good look at that data, and if you see a lot of individuals dying, um, before gastrulation during cleavage, that can kind of guide the molecular questions you're going to ask for quality. Like are they relying on their maternal genome? In that case, it would be advantageous to look at a maternal transcript and kind of see if there's indicators of upregulated or downregulated genes that could kind of give a metric of success. I guess that would be showing you poor quality eggs, but you would have eggs surpass that. And you could see that through a transcriptome analysis this year, um, with our eggs. We're going to continue taking morphometrics of total length, eye diameter, yolk sac volume, egg volume. But um, a little bit different to get some metrics during some distinct stages of development. We found that a lot of our individuals died last year during vascularization. So we're going to try to look into when those respiratory functions are like really starting to develop as the heart develops. So the heart tube, when it turns, when the heart spontaneously starts to contract, we're going to, um, look at the metabolic rate of those individuals as well as test them for some thermal tolerance and see their heart rate as they're exposed to graded temperatures. That will kind of give some insight to physiological properties of good performing eggs versus bad performing eggs. We could see if there's variation in the metabolic rate or their thermal tolerance. So that would be the like. We'd look at it when the heart, the embryos, when the heart begins to contract, once 50% of them have hatched, and then when they have completely absorbed that yolk sac and get kind of the metabolic rates. And I think that'll be helpful to have morphometrics kind of blended together with that data, because the yolk sac volume and how they utilize it is how they feed off of energy stores. So seeing those with metabolism might be good metrics to see how they, you know, put forth energy towards survival and growth just as individuals, and continue following those cohorts and see if there are any correlating metrics. I feel like it's kind of difficult to explain metrics of embryos correlating to like their ultimate subsequent development, because, I mean, that's the whole thing about quality is you want an egg with capabilities to undergo subsequent development, ultimately into broodstock to produce good forming eggs. But that's, um, like four years down the line. But like I said, that's the best part of working with the USDA is that we always have those fish growing and can keep track of the lineages.

Corinne

To put Halli’s research into simpler terms… she's examining salmon eggs at various stages of development, focusing on identifying reasons behind their survival or failure. Key to her study is morphometrics, a method where she measures the physical attributes of eggs and embryos, like size, shape, and yolk volume. She's also tracking heart development and response to temperature changes, which are critical for embryo health. By correlating these measurements with survival rates, Halli aims to pinpoint specific characteristics that indicate a healthy development trajectory.

In both Halli’s and Dr. Hamlin's research, they are aiming to converge techniques like transcriptomics, proteomics, and metabolomics to feed a sophisticated AI model. By examining a wealth of data points and exploring diverse biological markers like mucus, plasma, ovarian fluids and hormones they can integrate these varied insights into an AI-driven model, which could sift through the complexities to predict vital factors like breeding success and fish health.

Heather

There's different things that we can measure. So something called transcriptomics basically measures the intention I guess of the cell to do something. And proteomics kind of tells you that it took steps to do that, to make the proteins, to go ahead and do the process that it planned on doing. But then metabolomics tells you that it went through and that process happened. And and now you're seeing the metabolites that are the end result of that. One of the things that we're trying to do is use so the the fish have a mucus coating which protects them, but also a lot of the bodily processes that go on within the fish. Those metabolites end up in the mucus. We're trying to understand if we can use the metabolites that are in the mucus, like if we took a mucus swab, or maybe there's metabolites in the plasma, or maybe there's metabolites in the ovarian fluid. Those are all tissues that we're looking at to see if that can itself be a biomarker, metabolomics. You end up with millions of metabolites because. That's how many are there. That becomes then a huge data analysis, you know, endeavor. And so what we're trying to do there is use all of those metabolites and combine those millions of data points with artificial intelligence. So AI and machine learning to see if it can then create a model that says, oh, if the metabolites have this pattern right, maybe this group is higher and this group is lower. And all of these different, they pattern out in this different way if that can be predictive of performance later on. And so we're working with collaborators that that sort of is their bailiwick. And that's what they do. And then we're trying to kind of gather all of those data points, which could be metabolomics. It could be Haley's other metrics like embryonic measurements and morphometrics. And so she can use all of that. So basically gather all of these data points, combine them into some way that then machine learning can say, ah, now it's predictive.

Corinne

So what exactly can that information can that give us?

Heather

One of the things we're asking is can it predict whether she will have good IOP success, whether she will have 90% IOP success or 0 or 50, or, you know, whether it can correlate with her IOP success in her offspring, will it correlate with how well her offspring perform meaning? Will they grow fast? Will they have high survival? You know, maybe disease resistance or, you know, or something along those lines. So it's hard to say what it will predict, but we're trying to understand the whole breadth of what. It can predict for us, because really, the mucus is kind of a really novel that what they call a matrix is a really kind of novel tissue to look at. We really didn't expect necessarily that we were going to find anything. So this is pretty cool that it really. So we backtrack a little bit. The USDA has fish that if they have a really high estimated breeding value, they get bred. If they have a low estimated breeding value, often they will call them. And again, remember that the estimated breeding value, the most heavily weighted component of that is growth. And so when we took mucus samples before sort of as a proof of concept, we took mucus samples and then correlated that with estimated breeding value. And those correlated really well like like 80 to 86% or something like that, 80 to 90% correlation, which is amazing. We're not necessarily finding that that correlates with eye up, but therein lies the question of maybe estimated breeding value and I up success are are sort of competing with each other because estimated breeding value is more about growth. And so we're still trying to understand how all those those pieces fit together. So now we're moving away from estimating breeding value and going toward maybe eye up or some other metric to see what we can use as predictive. So we're really actually trying to understand now what can metabolomics tell us.

Corinne

Building on their insights about the AI model and the exploration of metabolomics in salmon, we can look ahead to the future. The research in Dr Hamlins lab has potential to open a gateway to practical applications in the aquaculture industry. Envisioning a collaboration with other experts in the industry, these scientific findings could offer more efficient, cost-effective, and sustainable practices, ultimately benefiting the entire aquaculture sector, from producers to consumers.

Heather

Yeah. So that that I think is the super exciting part because then we will start to look to partner with industry technology, you know, sort of in the technology sector to find people who may want to capitalize on that and say, ooh, we can build an X, Y, or Z thing that can then sample that and then, you know, maybe somehow streamline what is the what is the word I'm looking for? But, um, yeah, you know, or use something to develop their own technology that can be, you know, patented or used or whatever, that then could benefit because we're not technological engineers, you know. But if we can identify. Whoa, this is super predictive. I guess we'll leave that up to the other experts, you know, and innovators to come up with a way to make that technology or contraption or whatever that, that farmers can use. But it would obviously have to be cost effective or it's not going to be useful to farmers. So at least it's that critical piece of information so that other people know what to build, I guess, to create that thing that can then be utilized by because at its face, farmers aren't going to go and swab their fish and send them off for metabolomic sequencing. And then, you know, and then you know, and then use this model, you know what I mean? But but our goal is to give all that information to, to, to other people who can, um, because that is public information. And then hopefully then that can, you know, again, be something farmers can use. Yes. Exactly. Yeah. Much less economic loss, because that will really help farmers identify which fish are the most valuable to invest their time and resources in. So it will really limit the costs associated with hatchery production. That obviously is tremendously beneficial because ultimately that's what it's really all about, is producing fish at a low enough cost that it makes it viable. And then eventually that could be translated to consumers, because I think that would also increase capacity in a number of a number of ways.

Heather

It would really be to identify biomarkers that are predictive of both broodstock and offspring success.

Corinne rocks. Corinne rocks. Woo!