 And thank you to those of you who are online being here as well. For those who are online, almost everybody here knows who I am. My name is Tuba and I am the co-chair of this Decadal survey for the ocean sciences committee. My other co-chair is Jim Yoder who is online. I'm here. Good morning, Jim. Yesterday we had some amazing sessions and I wanted once more to thank Leila and Mona for facilitating them. That was really excellent and it was wonderful for me to sort of just be in a listening mode. So I really appreciated the work that you all did. Today, this morning, we do have another set of very exciting panels. I'm really looking forward to hearing more about artificial intelligence and what we can do with that. And also talk about ocean life, biodiversity issues above response to climate change. So we're going to be some very exciting panels in the course of more and more this morning. And before we jump right into the panels, again, I want to thank our hosts, Leila and the University of Southern Mississippi. This has been an amazing room to be in. Really worked well in terms of just the sound quality and the fact that we have windows and the lovely bulk co-chair. Thank you. And you said maybe you want to say a few words. Very quickly. That was a perfect segue. I really wanted to thank these two gentlemen, Chris Kirby and Ricky Slaughter. They did all. I think they're really keen and hard to make sure that we have the setting and the environment to be very communicative and they're probably dying of embarrassment. So thank you. I just wanted to welcome everyone again to the University of Southern Mississippi. I'm really happy we were able to provide the setting for discussion. A couple of reminders if you need bathrooms, like once again, they're in either direction. I didn't say this yesterday and I should have there are emergency exits on either ends of the hall. If it happens, go either direction and you will be fine. And if anyone needs to step out and take the meeting or call throughout the day, the conference room immediately across the hall is open. So just pop in and use it as much as you like. The thing I wanted just to say this morning, you know, just as kind of a welcome was to tell just a really quick story about Gilbert R Mason. We've been talking about ships a lot in the last couple of days or the last day. We've also been talking about building a diverse and equitable future for ocean science and I just wanted to take a minute to talk about that name. He is the namesake for the future NSF research vessel that is being built at this moment about an hour or two hours from here. That will operate out of Gulf War Mississippi. And while you may have heard the name you may not know anything about him. I think that might be true for most of the books in the room. But you know it's the name of a naming the ship. Mason. It's not just him. It's his family is that is wrapped in that name and it is an aspirational story but it's also an accountability story, and it's one that, you know, we are hoping to to be accountable to so Gilbert Mason was neither a ocean scientist, he was also not an astronaut so different name he actually was a physician, and he had his practice in Biloxi throughout the entirety of his life it's just the next city over. He's known for a number of things throughout his life. He provided medical service to the mariner community in South Mississippi, as well as in the New Orleans area. But the reason he has placards throughout the, the gold South is that in his earlier years in the 1950s, he organized waitings and the beaches of Biloxi to segregate those beaches and those waitings often turned very violent. They were they were dangerous things to do but he was committed to doing this, he and his community so that the beaches could be open to all. And he's also known for larger civil rights activism throughout the state of Mississippi and his work to desegregate the schools in Mississippi, which ultimately happened in 1967 and there's a fun side story there about the challenge points that we had at the QA for, for the vessel several years ago that I can tell you, tell you online about that that particular day. But he did these things for his family so that his children could have access to education, and one of his children came, but went on to be a physician as well. His civil rights activities that he did, he carried them out with his wife Natalie as well, who was a social work professor at the University of Southern Mississippi. He did this all for his community so that all could have equal access to the beaches of Mississippi. And this is the reason why we have named the ship after a civil rights leader who was a physician in Mississippi, not an astronaut. And it's why the ship carries the model motto of equal access to the sea. That's where that all comes from it's his legacy. It's his family's legacy. And it's a legacy that we're holding up for future scientists and the scientific enterprise that that we will be able to support. And it's also a legacy of accountability. We know that, you know, we put up a name like that, and we have a responsibility to be accountable. You know, to the work that we want to do, we being the scientific community, and I will say that the Mason family is watching very closely and we want them involved in this because we want that kind of accountability. So it's a, it's a fun aspirational story. Our plan is to tell everybody this story every time they come on the ship, but we don't have the ship so I'm just figuring out. Thanks. That was wonderful. It's really great to hear what the background about it. He passed away. He passed away in 1997, and his son, Gilver Mason, Jr. passed away last year. Thank you. Thank you. It's a great story, good motivation and a nice opener for today. With that, I'm going to hand it over to you. Right. So today, we're going to have a panel for opportunities for AI amounted advanced social sciences. And so what we're looking for today is to learn a little bit more about opportunities for artificial intelligence and machine learning to advance social sciences. So we have three panelists today. Warren Wood from the U.S. Naval Research Laboratory, Peter Gerstoff from Scripps Institute of Oceanography, and Heidi Sosik from the National Oceanographic Institute. We've asked each of the panelists to start with a short presentation on their own work. And then we have a few questions that we've come up with to talk about this topic and then we can open it up for questions on the slide. So with that, I'm going to turn it over to Warren. Thank you very much. And is it possible for you to have let me talk some up there. Yeah, I can. Did you email me your site? I'm not familiar with Zoom, maybe in other cities. Okay, so I'll just do a brief introduction here. My name is Warren Wood. I work for the U.S. Naval Research Laboratory. I work with a bunch of people and we have been working in the field learning now for probably several years. It actually kind of sneaks up on it. I just wanted to start off with a few things. You know, a lot of people who are unfamiliar with the topic say, well, you know, can AI or machines, are they going to take over and do science? I don't know, but they can certainly help. And the way we've used machine learning and artificial intelligence is as a tool to help us do more with what we have. So, you know, in the past, we've got, you know, people can program machines to do simple things very, very fast. Right now we are teaching machines to do more sophisticated things, recognize the various patterns and so on, and then use those patterns to make predictions. And in the future, you know, self teaching or self learning will, you know, AI take over the world. I'm not sure how good humans are at self teaching sometimes. So it's a very powerful software tool. I'm not sure it's much more than that at this point. So in the past, we've had tremendous, you know, changes in our society from things like photography or programmable computer mass vaccinations. I think of AI as just kind of one, one more step and not anything, you know, horrific or dangerous. So with our motivation here at the year, at the Naval Research Laboratory and I work just not too far from here from the Gulf Board at the Spanish Space Center. Our motivation is understanding the environment in a comprehensive way. And we have for the atmosphere, you know, the atmosphere in the oceans have necessarily been sort of dealt with on a holistic global scale for, you know, an entire existence. Because clearly the air is connected everywhere and the seas connected everywhere. Geology has a different history, one of resource extraction. And so we ended up doing a lot of sort of postage stamp type of high super high resolution studies. Very lots of data, but it's very local, very local geographic spots. And so happy with a more holistic picture and prediction of the sea bed. So the idea is our in our, this is an example of machine learning. So we have all these sparse data points and what we'd love to do is come up with a comprehensive map. And so we've developed this tool that we call geospatial machine learning. Essentially, it's a it's a very fancy way of interpolating and mapping. And I'm not going to go through all of this but the fundamental steps here are data curation. This turns out to be way harder than everybody would be. Data, especially geological logic data is in the variety of forms of formats getting those all. So machine can and can understand them is is no small task. We also need predictors. So we, the way this works is we have things that we know everywhere and things that we know in certain places and we want to know we want to predict those data values. But we can only predict where we have features. So the features are things like the symmetry and spatial statistics of the symmetry and I'll show a couple of examples. Let me use our machine learning. We have a training phase where we train where we have the observation. So is machine learning can really only predict in the area. So we put that to train it and then we can predict. And then we use a informal uncertainty to estimate how sure we are at that. So the data curation, again, tons and tons of data out there, getting into a usable form is a is a is a task that I bought me again. So we have predictors. We have things like elevation, global elevation, which is a roughness over radius of 100 meters. So a very, very small scale we go all the way from, you know, 100 to 500 to 100 meters to a lot of scale and spatial statistics. We calculate things like positioning of that. The peakiness essentially of what's down. Also, I wanted to point out, you know, this is oceans oriented. The meeting, but we also can do things online. So we have a fraction of outside of geological maps. And again, we calculate the opposite decision. And again, I'm not going to go through too much of the detail here in my in my cognitive intro. But essentially what we're doing is we think about this map up here as a map of what we want to know. And these would be our features or predictors. What we're going to do is say, okay, well, here's where we want to predict. And if we look down here at all these predictors, the values of each of these predictors is there's might be closest in terms of in terms of geology space, not geographic space, geographic, but it's on the other side of the world. But geologically might be very similar to the given observed value. So what we do is say, okay, well, but if this is the observed value that's closest in geology space to this, then we assume that the value of what we want to know is the same as that. And this is, you know, it's essentially a statistical engine, and it comes up with a essentially a prediction or an outcast, we get an uncertainty associated with that. And then what's kind of nice is that we also get a guide to the next where to make the next observation, right, so we can statistically point to a place where another observation would be valued. And then our conformal uncertainty, this is, this is much harder to explain, but essentially what we're one way to a certain uncertainty is just to assume a single uncertainty everywhere. So let's say we're going to predict them with them and you could say, well, the symmetry uncertainty is just a given value. And that represents this thing here, which sometimes overestimates the uncertainty and sometimes underestimates it. A little bit better technique is to use this informal uncertainty technique, where we can have a data, the data itself drives the uncertainty prediction and drive the uncertainty estimate. So here's a, here's a prediction of our symmetry and here's a prediction of the uncertainty. And you can see that the different features are so different features have different uncertainties. Bloom is low here, so where we have actually bathymetric surveys. So, I just want to finish up with a couple of published examples. Here's one for total organic carbon at the sea floor that we published a few years ago out of our lab. That's been getting a lot of, a lot of sudden, but it's very useful to, to take an inventory, taking inventories of performance, so see that properties is very useful. We have also looked at categorical predictions, one of which is these C4 or C4 toward expulsion anomalies. Predicting anomalies, it turns out, is kind of challenging because like I said, the prediction engine really can only predict things that it has seen before. So if it's anomalous, that's a little bit tricky, but read the paper. Then better give John a hand. Another, another example is mass accumulation rate, mass accumulation rate of seven essential accumulation rates, published a couple of years ago, and I think this represented a major upgrade to the estimates they've done previously. Also, I mentioned you can do things on land. I've been looking at a little bit of geochemistry. Most of the data actually just comes from Australia in the United States, but fortunately, Australia in the United States and also Brazil and Chile cover a wide variety of groundwater situations. And from those, I think you can get a pretty good estimate of this, we can span the space if you want to do chemistry, groundwater chemistry space. Then we come up with predictions of things like global bicarbonate concentration. So that's, that's my intro talk, and obviously happy to entertain you. Thanks, Lauren. Right now I don't see any questions with Slido, but I'm sure that people are thinking about what to ask people have a bunch of time for questions at the panel discussion. So next we're going to hear from Mati Sosa. Good morning everyone. Good morning. Okay, I hope you see my title slide. Does everything look okay there? Okay, great. Thank you. Thanks for the opportunity to share some ideas about the value of AI and ML and research on plant and ecosystems. Marine ecosystems in general. I'm going to focus. I talk mostly on imaging in the ocean. Our ability to image has really exploded in recent years. Imaging flow cytobotyl highlight is one example. This is an automated submersible microscope. It takes images like you see here at rates of tens of thousands per hour. And because these images have about one micron resolution, we can identify many imaged organisms to genus or species with with this approach. The technology is automated and it's designed for extended deployments. So we now have high resolution multi decade time series of images at places like the Martha's Vineyard coastal observatory, as well as now growing spatial from approaches like seasonal process studies that extend from the near shore observatory across the continental shell. And now a decade of even broader spatial surveys along the eastern seaboard of the US from North Carolina to to Canada. And each of the points on these maps is a place where an ICB sample and thousands of images have been collected. These already really large data sets are growing rapidly that an ICB is imaging it at NBCO right now while I'm speaking to you. I got back from a winter process Trent that cruise just two days ago, and a broad scale survey on a no worship is underway right now. With these data streams we can characterize absolutely amazing species specific dynamics. Here I'm just going to highlight for you an important chain forming diatom in this ecosystem, this daily resolved time series from NBCO helps us see in an unprecedented the very strong seasonality high inter annual variation in this organism. And then with the, with the broad scale surveys. We can generate things like these composite seasonal maps that come from decade observations and show that this organism that we know is important in the near shore is also blooming in specific patterns with seasons across the entire region. This is only one of many, many species in the in the ecosystem, and images of this particular organism comprise way less than 1% of the nearly one and a half billion images that are in these data sets. So, ML is very critical for analysis of this data set there's no way we can as humans we can sit through billions of images. We're currently using a 155 category convolutional neural network to classify these images this is a supervised algorithm so highly high quality training data is really essential to get this kind of good performance. And I realize that we're not only using ML for this kind of quantitative abundance patterns through time and space that I showed on the previous slide, but we're also able to study important processes that are happening at microscopic scales this classifier performs well for the diatom species I've been talking about both when it's healthy, but also when it's infected by these tiny lethal parasites that show up very clearly in the images. This means that we can learn that we have been able to learn that the parasite infections in this ecosystem are recurrent. This is at the near shore observatory this is that diatom time series again where the darker points indicate very high levels of parasite infection prevalence this is a lethal parasite parasite I should say so it's very important to the dynamics of this organism. It's important that the near shore observatory and also across the broad scale. This revelation has implications not only for the dynamics of this particular species but also for food webs carbon and nutrient transfer through the ecosystem, etc. and it's leading us to new hypotheses linked to the ecosystem that will not change in this region. I want to tell you that these days that Martha's Vineyard Observatory is just one of many time series. These are being deployed all along the US coastline by a wide variety of users now. And in addition to basic research this network of sites is helping to provide early warning, and better understanding of many things like harmful algal bloom species that produce toxins that can threaten food safety for humans, and also cause other ecological problems like fish kills. You can imagine ML is essential for timely and effective use of these observations, this in the same way that we're doing with the basic research. And this is, it doesn't really stop here because I've seen the users around the world are starting new, new time series locations and conducting large scale spatial surveys with high throughput imaging. And just, you know, to beat a dead horse strategic use of AI is the key to making the most of these huge and growing data sets. I want to just close my last minute or so by bringing up another imaging approach, we're complimenting ICB with centimeter scale imaging to capture patterns higher up the food web so zooplankton, and also the dynamics of the wind snow, the video is not going the dynamics of marine snow distributions. Here you see data collected with this sensor platform that's profiling up and down as it's towed behind a ship and collecting images at 15 Hertz 15 images a second and we're giving our ability to use AI and ML models to locate, classify and characterize targets in these really complex images is a big, a really big challenge that we're tackling now. But I just, just got back from a winter process cruise, we were doing that kind of toad operation on on the ship earlier this week, but we had to end the cruise early and run into sheltered near shore waters because as I think most of you know a major Gail blew up through through the entire Eastern US, and we could not safely operate from the research ship. It was really exciting because as we were coming in running away from the storm, we passed this long range autonomous vehicle heading out directly into the winter storm. And I'm excited to tell you that I'm collaborating with a talented engineering team and was whole and also a camera system developers at a small company, Belomere to get this prototype imaging system similar to the one on the It's now on the nose of this autonomous vehicle it's on its first science mission. As of this morning the vehicle had had completed the entire cross shell transfect and is headed turned around at a far point it's on its way home, and it's collecting about a terabyte of images a day. And now of course it's not practical to transmit that kind of raw data back home from an autonomous vehicle out in the open waters. So the demand now is for effective AI on board to say these, so that we can send home only a lower volume processed information. And this will also open up the door for real time interpretation and adaptive sampling based on the imaging while it's happening so that going forward we can do the smartest observing we can across time and space, while also resolving the kind of biological detail that we really need to understand how ecosystems function and how they're changing. I will leave you with that. Thank you. Thank you so much Heidi. That was fascinating. I think we'll go ahead and go to Peter and when we can take a few questions off slide of before we start the panel discussions. Peter. Thanks for joining us so early on the West Coast. I think you're on mute. Okay, so now you can see everything. We're good. Okay, so. So this is I'm Peter Gustav. Thank you for inviting me and I'm at the UCSD on the West Coast here and I have my website here you can see more of what I do I do machine learning in acoustics and seismics. So machine learning and artificial intelligences. I think about the same thing. Why. Okay, so machine learning they combine the strong statistics and computer science in machine learning. To obtain the model so it's not just magic when we combine them machine learning is based on the theory for model identification control estimation. A neural network approximation. I saw very fast with modern architecture leading to very excitement. An example of machine learning and how it influences that I did a class on on data assimilation. I had about 10 students. And then I decided to change it to machine learning for physical applications. Then the only thing I did was making the title, then I had 65 students. And it grow to 245 students graduate students in 2020. So I think to attract students we have to have some machine learning. So I like to have the basic principle of machine learning here this is a function approximation problem. We approximate this function here we have our outputs here and we have features that we observe. And from that we want to extract this unknown function relationship here. We can use supervised learning. The outputs are labeled and we need lots of data to train it. Or the more interesting part at least scientific is unsupervised learning when there's no label. The goal is to find the interesting properties. And a good example is what we call an auto encoder where we learn to represent the data with itself. And in order to learn this shows it here we have weights of the neural network we have to learn so we have to train on model data. And to verify the model we keep a part of the data separately and we test on these to measure how well the method generalized. And this is very relevant to use this approach when we can't be explained with existing models. So if we have a good physical model we should of course use this. I would stress that it's much more than just neural network and actually I work a lot with all these methods here. There's many more methods. So in order to do machine learning we need some physics in the system. If we just use a basic machine learning approach I illustrated with this pendulum here. We have these points here. We just measure them here. And then we can make the curve fit to here but it doesn't generalize. Remember the important thing is that our methods generalize well. And here we are just measuring the least square. If we knew it was some kind of physics system we know the PDE, the partial differential equation. We can put that into the physical system. This is what's done here. We have the PDE here encoded into the system. And that way we went back. It's a much simpler system. And we can now generalize well. You can see I can predict everything well. This applies to data simulation as well. Same thing. There we have equations into the... The unique thing is that these derivatives we can actually do analytics with. We have back propagation because back propagation automatically gives us these derivatives. I think an important thing in physics is that we have some uncertainty quantifications in computer science. It's not so important to have much uncertainty. But if we want to predict the value, uncertainty is just as important as the value itself. So uncertainty qualification is a little bit underrepresented in machine learning at the moment. I illustrated here with an example for Gaussian processes. And here we have a function with a prior here, uniform prior. And this is possible realizations I have here. And as we get more points, you can see we have more knowledge of the environment because we have measurements. And then we can reduce the uncertainty in neighboring areas. And this seems to adapt as we move along as certain things here. There's many ways to do machine learning. We do Monte Carlo Gaussian process. Some kind of uncertainty qualification where we just have typically a number representing the uncertainty. And with conformal prediction, we can guarantee it arrives within a certain uncertainty interval. And this influence also may be propagation of uncertainty. We can put in uncertainty in the latent variables in a neural network. If we have explainable measures that would also give a confidence in the metal and that's also related to uncertainty. I think pattern detection, all these are good. We do also a machine learning for ocean and climate science when we do data assimilation. This is important. We can learn emulators and surrogators by modeling. These will be a higher resolution. We don't have a solution problem with these. We can do better physics because we learn the surrogate models for the unresolved processes and faster to compute because we can use a neural network. And so the data assimilation with machine learning we can adjust more observational data. We can learn observational operators, but we need more uncertainty qualification in this process. So similar to things we already do, but faster and more precise. This is my take on machine learning in the next decade. People of course have AI assistance for the scientists to perform tasks that can be specified, literature survey, gather data, write code, implement the analysis, create the draft papers, review papers and these things. So with all these here, we still have human learning profiles, for example improved forecasts. I think some of the cool topics is when we can do multimodal sensing with AI. I could see combining a video and sound in observation in a single ocean observatory. If we have a camera, can we put sound on it also? So we can combine sound and in video. There's no, I don't know of a physical process that can do this. So we have to use machine learning to teach how they hang together. We could say a satellite could. By just looking at a satellite, we could have predict ocean background noise. We observe rain waves, chlorophyll, and from that we can extract the noise. We can predict turbulence, hyperspectral sensing, climate change, ocean model. So in all this humans will work closely with AI to better the outputs that we learn. I was thinking other things we could reduce the whole academic fleet and just have robots sailing out in the ocean and drop the acoustic voice, the seismic voice, and take some water samples all around. And that would maybe be very efficient. Risk of AI is of course we will get fake papers, we will get too many papers. They might even replace the apprenticeships, graduate students, replace scientists. And we all see an example here of fake people coming up with the GAN, generalized adversary networks. I had an introduction of machine learning and ocean acoustics, where you have more details here. Thank you. Thank you so much Peter, that was great. We have a number of really good questions on Slido, so I'm just going to ask one question from my questions for the panel, and then we can move to Slido. So the one question that I want to start with is, what do you think the future needs for the cyber infrastructure will be just AI and now, and we'll just go in the order that we have a panel. So, Lauren. Yeah, I would say the short answer is data curation and data communication, essentially what AI is, is the information of the science. You know, we're all most of us here are probably trained in the earth sciences. And I think, well, the biggest hurdle in getting to use AI effectively is to become enough of a data scientist that we understand how data are stored and how to avoid it. Heidi. Yeah, I think I would like to emphasize that we really need data to information full pipelines that are documented reproducible, and that include full provenance tracking not just traceability from end products back to raw data but also all the systems models and parameters along that full path from one to the from one end to the other. And this is a, this is a really big challenge to build and standardize enough so that we can catalyze an active community of of data generators and the somewhat akin to the big data centers operated by, you know, agencies like NASA, where, you know, the critical data streams from earth, earth observing satellites are supported with products that can be easily reprocessed and there's no full and standard pipelines, but the problem is bigger for our community because our data sources are so heterogeneous. And the data generators are very diverse and heterogeneous. So it's not, there's kind of point source data, data all over the place, instead of a centralized small number of highly curated satellite pay lies. So it's a big challenge. So yes, I think we need a data center that could collect all the data together. And of course that will be very diverse but I think we just start by putting it in and maybe machine learning could put system in them one day. But, but some kind of collecting of data, maybe an ocean chat tqp might make, make this easier to access. I think accessibility is going to be a big issue. So, I'm actually going to move to Slido, because there's some really great questions on there, and Lava, you have a question on Slido. Yeah. So that the zeitgeist is a particular mood or spirit of a particular point in history and I, all of the speakers touch on a sense of the zeitgeist around machine learning. So I wanted to see if you all would lean into a little bit of words of inspiration for caution, as we try to capture this moment in this zeitgeist in our report, and how it can, can or cannot address grand challenges. Okay, Sean. So, as I said, I think AI is a step change in the, you know, in the power of illness. So, as such, it's a tool and any, I think any tool has, you know, can be used, you know, good work for it. I think that some of the risks, you know, there are risks with every new technology. We have new weapons, you know, atomic weapons that created a lot of risk. And, you know, we have any kind of new technology represents a way that people can help society or hurt society. And so I think there's always a little bit of a wild west anytime you get a new technology and I think we're, we're in that right now. And so I don't know if the, there are no societal norms really I think established to prevent somebody from submitting a fake paper. And then we have to ask ourselves, well, if the fake paper is really good, maybe that's not a bad thing. So let's just not be so fearful. And let's, let's think about maybe the fake papers better than some of the other papers, right. So let's think about how to use this problem, right, and how we can, how we can best use it to advance science for society. I think there are always going to be people out there who are going to want to want to use it to do. And I think that's true in every, every possible tool you can do that. This information. This information, this information, right. But, you know, the first time somebody, you know, some caveman or something picked up a stick and killed his neighbor. I think it was a cave woman who killed her husband. I think it was a cave woman who killed her husband. I'm not judging. You know, the one analogy that I shared that makes sense to me is the calculator calculator first team on board people were like what our kids must know how to add and subtract and multiply and, but now the calculator is an integral tool in us teaching math to students. And so, but it was definitely perceived as a threat to mass. Yeah, people will always feel threatened. Some people will always be threatened by new technology. I'm a little threatened by soon as we can see it. If I could jump in, I'd like to add another slightly different perspective. I think, you know, in terms of inspiration, we're, we can do things with AI ML that we just can't do any other way. So that I mean hands down I feel strongly about that. So instead of talking about risks in terms of bad actors though I think we should also be very mindful about, you know, good intentioned scientific or research uses that might go awry. And if we don't, if we're not careful enough about doing the hard work to verify and the performance of any kind of algorithm that we're going to use to either interpret data or discover patterns. And there's no way to get around that we can't just take it ML AI and ML as a, as a black box, and it's hard work. Yeah, I mean that box on inspiration. I'm here I don't have questions at the moment, comments at the moment. I think Kristen you have a question. Yeah, this is for all of the panelists. Hi, and I guess also machine learning training be incorporated into our teaching at undergraduate and graduate levels for those who are going into ocean sciences. I can see. Maybe we're lagging the need at the moment and ramping was up with going to probably take a lot of strategies for doing that with you. So you're not talking about teaching AI amount you're talking about using AI ML know the whole thing like if we get the work. This is now a workforce need that we're on the edge of work. Then, how do we need that workforce development need in higher education. Peter says you have 245 students. And when I first started, actually it was just image processing we did, because there was no physical incorporation of physics into the system so it was sort of like cat and dogs and whatever you could see in the picture. Now, now there's more advanced methods where they have have physics into it so I think that's cool. In terms of curriculum. At SIO, there is no. Of putting machine learning directly into it yet, but they're free to take the class I teach is on upper campus in electrical engineering, but I have students from from scripts, institution, both inography. I think they take classes that way they're very interested in learning it so it comes in here but I think eventually it will come in as a mass similar to a mass class I would say right but data focused. So obligatory class. Heidi, do you have any thoughts on this. I think this can be handled similar to the way we, we have our students learn the appropriate statistical methods for data and science scientific applications. I would agree I think as Heidi mentioned earlier I think this is a, you know, this is this is something there are things that machine learning does that other other techniques can't accomplish right and so decades ago as a geologist I went into this, you know, the first people to start using computers and geology, most geologists drew their maps by hand. You know, we learned over years and decades that there was just more effective ways to do this. And I think that that's, there will be a natural progression of teaching the students as they need to learn these tools to do the science. But they can't do it without machine learning. I can see professional development being needed very quickly for the faculty to be able to teach what's needed. I think there's training workforce training at that level to be able to train the students. So that's the follow up question. So, this question maybe the data but when you train students to so teach students about ethical and responsible use of the out and out methods or is it just about data analysis and assimilation and making sense of large. In other words, how much of ethical and responsible use of AI is included in the education and training. So, so when I teach it I don't have any responsible use of it yet. I mean, we're just trying to make the basic methods work and and see how much we can use of data so it's not. We're talking about the message just like my slides was more focused on the message here I'm not focused on responsible use. Yeah, I would, I would chime in and say that yeah everything should be used that whether it's machine learning for any other kind of. I think just like ethical use of any, any other kind of sort of a factory procedure or any, any tool that's used to do anything in society needs to be used everybody. Yeah, those are that comes right down to what is ethical and what is that behavior. And, you know, it's, there are many stories of people using AI to try to do something good like determine, you know, who we should hire to leave our company. And it turns out that we're putting the data and the PI said we need to hire white males that graduating from our students. Well, it wasn't the answer you're looking right, but it was, it was dated with the bed was biased. And so, you know, you recognize the biases that you, that you put into AI. These are all these are almost impossible questions to answer satisfactorily. You know, what is, what is ethical behavior but I think we always need to keep a check. So, on everything to ensure that it's, it's just ethical and AI is. One more question about ethical, ethical use of AI and being able to trust AI, I think AI trust goes into that. And how does that fall into uncertainty quantifications. Yep. Yeah, you know, we thought about this a lot. It turns out the statistics is very hard. We, you know, we come up with an uncertainty estimate how certain is the uncertainty. Well, you can have uncertainty on the answer. And ultimately, you know, I think for in terms of, in terms of ethics, there's never going to be a, there's never going to be a single easy solution. Constantly going to have to check and question like we do in science and science we question authority all the time. If you're not questioning the scientific authority, probably not doing good science, you need to always be pushing the envelope and always be checking to make sure that the answer is better. In fact, they don't want to get ahead of you. And there's no, there's no easy way out of this. This is what makes science, but you know, challenging and interesting is that it's, it's a constant day after day push against the unknown. And we constantly push. Nice to follow to that. So what I'm interested in is one of the concepts of science is reproducibility. Is there a family have reproducibility in terms of say a publication that uses AI. So ours. Yes, our reproducibility the validation steps in the data analysis. We're very concerned about this. We sometimes will do a prediction one of our special predictions. We'll do it several times and comparing the answers. Yeah, I'm thinking about in terms of like a scientific review process. Could you have in someone from, say a reviewer of a paper, then use the same protocol to see if they can reduce the results. That's, that's a good question. I'm sort of training and testing. So if you look at the data. Yeah, I mean, there are ways to test the science if you're if you're testing whether somebody is trying to pull one over on the scientific community by putting out fake results. Maybe you could train an AI to to to sit out, you know, the bad actors, the misinformation, the disinformation. I don't know. Yeah, but I mean, I'm thinking about it, not just people who are deliberately misleading, but maybe they made a mistake on how they, how they went about it. Yeah, yeah. So certainly you can misprogram something we can make mistakes have been made throughout history. People will make mistakes. I'm sure we've made mistakes in some of our publications. They exist. And again, it's just this constant sort of checking and questioning. Is this the right thing? Didn't we make a mistake? How can we check to make sure this is all part of the scientific method. And the risks of AI, AI allows you to do things bigger or faster. So your mistakes are going to be bigger and faster. So it's probably requires more effort to make sure that they're not. Mona, you had a question. Sure. So my question is one of the statement of tasks that we have to do with the research infrastructure. So our AI and those who's exacerbating existing infrastructure challenges because of the amount of data that researchers have to download to train and our models. So I can see how this would easily do inequities between our one and resource rich universities and those that do not have that kind of infrastructure in place. Sure. No, that makes total sense. Yeah, so I think, as I said, data, this is a data science and data, you know, big computers, lots of disk storage space. And for communication, all these are expensive infrastructure components. And these are, these are going to be used more effectively by wealthier entities. You could use UNOLS type approach where the big centers have fair access through some sort of, you know, evaluation criteria based on funded proposals that type of thing. Yeah, I mean, for sure. The whole cloud can storage and compute infrastructure could support something like that very readily. Yeah, no roadblocks there besides money. Some things about Marsha. You know, one of the things that we should also not forget about are how, how advanced the large language models have become. There will come a time when we're, we won't be sharing time series with people anymore that the terabytes of data will sit somewhere and will query it, you know, with a large language model, like, you know, and that large language model won't necessarily need to go back and, you know, dig into the data, it will, it will have learned what the data is telling us. It's sort of like how do you think the large language models do it now when you just pose a question. It does need to trust and explain ability to them. And I'm sorry. I have a question from the, from the audience. I think it's very related to the last question, but the question was sort of what kind of underlying data are needed to supervise the machine learning and training and how can we enable the development of such data sets globally, especially in hyperbiotivist, hyperbiotiverse brain ecosystems, development agents. Yeah, well, I guess, I mean, in terms of what kind of data are needed for supervised training, if you're, if you're thinking about bio biological diversity and approach, and I'll approach is to try to characterize distributions of that diversity. But there's no two ways around you need someone who knows the taxonomy of those organisms, and is willing to put the time in to annotate images or whatever, whether it's, you know, acoustic spectra or whatever kind of data you have associated with those organisms, and to, you know, match them up and with annotations. How can we enable that in development of global data sets I think this kind of comes back to shared resources and infrastructure to support standardized and centralized and community building around that hard work of annotating image data. With internet access I don't think there's a big bar to step over, but we do need the infrastructure to support sharing that information, and that work. Maybe someone else wants to help clarify. Oh yeah, and I would chime in and say, you know, once you think you haven't figured out and you develop a system like that then something is going to change. You're going to learn something else that you need to do. Some other kind of data to answer some other kind of. So I don't see that this is a, there's no destiny, it's not a destination, it's a direction. Definitely not a one and done task it will be part of that ongoing verification validation continuing to build out in, you know, we've been annotating images to build our deep learning models for more than a decade and we're, we're still, you know, adding to the training sets when we're studying the same region because new things emerge or we find new problems we want to explore and we need more facets in the training data. It's definitely part of the research. Yeah, my question. My question has to do with the 3D AI and 3D in the water column where your X and Y sampling scales are very different from Z. And is that, is that pose a fundamental problem or is it simply a computation issue and if computation only is there a rule that you could figure out how much more time it would take them to do it to D AI. We're actually, you know, I'm all deep in that question right now so we were talking, we're thinking not so much about the water column although that's part of it but also the 70. So in our cases, the 70 was very much more heterogeneous than the water column. And so we're, we're looking into that right now and I might our hunch has been that it is simply a computational problem but there are there are some vast differences. When we think of the horizontal homogeneity of the system might be orders of magnitude different from the vertical. So, I, our values that it's, it's simply going to be a more of a computational thing but I do think that there's going to be some kind of change you know that that would be required to deal with these. It's basically a vastly catastrophic problem. Yeah, thanks. I'm. Thank you all the panelists this is really a fascinating discussion that I'm kind of curious and whether the thoughts are in the world and essentially guiding the supporting the skills we need for everything from support from computational capacity to interdisciplinary research, developing ethical best practices I'm just curious on what you think NSF should be doing. Let's, let's start with Peter. That I don't think I have much opinion on this with supervised learning with data sets I think we could be you could use unsupervised learning and that would reduce the demand for having all these label data sets so we can label things automatically in the future. Supervised unsupervised learning similar supervised learning or so there comes many new methods on this different quick answer. Yes. I want to say I want to kind of come back to my point earlier about the need for data to information pipelines that are documented reproducible and include provenance tracking and add that that are also linked to ready computational data sets that is available for a wide diversity of users and it becomes inclusive and isn't, you know, cost limited for anyone to participate in the research. That's a, you know, that's a tall order, and I think there's an opportunity for NSF to lead an interagency charge to kind of tackle these problems that cross objectives from NSF. To NASA to know to EPA. And kind of all get together to join forces and tackle this big challenge that's cyber infrastructure and cost that could be shared across agencies because there's so much you know we all need to get together to be able to do what needs to be done. Instead of dividing up the pie. I'm just going to, since, since this is really important to our study, I want to make sure that we capture what you said by the. So what I got was develop information pipelines, whichever reproducible with provenance tracking have data that are computationally ready and accessible. And then maybe lead an interagency charge to make data accessible across agencies. Yeah, I think that's right. And, you know, we could, we could build on, we don't have to start from scratch we build on existing success stories. You know, we have important infrastructure and repositories like the Beco demo program that have the potential with the infusion of investment to be linked to the kinds of needs we have for everything we've been talking about today. They're not ready for that right now because demo is. Yeah, it could it needs. It needs infusion to be able to move forward in that right direction in the future. But that's just one example that's an NSF supported within geo. There are other efforts within NASA, within NOAA within other agencies that could all be the kind of take lessons learned and build on the strengths of the combined effort. Maybe I'm being a little Polly Anna but interagency might get us where we need to be in the next five to 10 years. I would echo that and I've been pushing on that and one of the one of the hurdles that we that we've come across is that program managers and agencies in general are reticent to fund and put funding into something that's all data. So these data going to a repository, but those repositories. I think they might be resemble data dots. And again, the data are different forms of formats. So I think one of the things that NSF and other agencies can do is, as Heidi suggested, get on board with, you know, standardization. What is this is the first bar to get across to use AI is getting the data to some kind of standard. That's been a huge hurdle for us without you spend a lot more time than I would like dealing with dirty data cleaning the data. Program managers with ties, people to keep their keep their data clean and usable by others in the future and encourage submissions that use legacy data instead, essentially instead of firing from you. This is a question on that. Yeah, I'm sort of curious to hear from all of the panelists, how much publicly accessible data they use now for training their various models. And in this particular case, I'm really getting back to this question that if we had public, if we had databases, it's going to help us right there are databases right now, you spend hundreds of millions of dollars and setting up databases. How much of that is actually being used today. I would say that we've set up data repositories that are, you know, slumps of data, you know, people just download all your data and whatever format. They have a rolling deck to repository, which has been phenomenal for making sure that the data are at least preserved, but that's only the first step. The next step is actually making sure that they're usually right and accessible. And so this is this is where I'm referring to. And to answer your first question, we use almost exclusively publicly. So we can be there. And Gia and your fans. What there's lots of lots of data repositories out there that we acquire data from. And, and then we. Peter and Heidi, do you have any thoughts on using publicly accessible data. Yeah, I guess I have a couple of thoughts. I mean, in everything I showed today with the IFCB data. All even the raw data as well as all the products we produce from it are publicly accessible. But I would love to hear from a G or others who who think there are national places where that there that that type of data can be deposited appropriately. We built our own system to make the IFCB data accessible. The kinds of repositories I'm aware of are very good at accepting. Scalar data, you know, you measure temperature and salinity, there's a lot of places to put it. If you measure billions of images, there's like nowhere. Thanks, I think that's actually that gets to it. What I was thinking about what I was trying to scratch away at which is that when I was talking about publicly accessible as thinking about where you're using somebody else's data and how we can really help with that because with, you know, like what we have done is phenomenal. I mean, with that work in Alaska in terms of equity and this is astounding, but how I see it being huge. But for me, this idea that if you set up a database, it will help it will solve a problem, or if all folks submitted data to solve a problem is something that I want to sort of just understand better perspective. I think it's not just setting up a database. It's setting up a system of software tools and accessible accessibility tools that make it easier for everyone, the data generators and the data users. You know, the example you may be familiar with a Jeep that I'm, I use is the ICB dashboard system which is supported by web services and API's that was built within my research group. It's now used by almost almost every ICB user around the world, because it makes their research and their work easier to do. And those are the kind of tools we need to build. The ICB dashboard is it's a research tool that we you know we built we're not software systems engineers we can't centralize it. And it's a bit challenging for users all around the world to be able to set up that software and infrastructure and that's the kind of thing that could be hardened and centralized. I think with actually not too big a lift. Yeah, but Heidi, thank you for articulating so nicely how, how much work it is to do it right so that people use your data. And this comes back to what you were saying you know how come there's already data out there more and you're saying when the data dumps. I was trying to resist, but one of the reasons I think we're by people just dump their data instead of really curating it is because our promotion and tenure system doesn't value it. Right. And so, as long as that's the case, you know you're going to write your next paper or you're going to curate your data, you're going to write your next paper. And we have to change our old fashioned evaluation system and actually align it with our stated values in order to make progress on this. If I'm not saying something if we could somehow equally make equivalent data curation with citations right so if somebody uses your data, you're essentially getting a citation and I think we've taken steps to that with the DIY data sets. And I think it's that as valuable as writing a paper to get citations and if your promotion board is then looking at not just your publications, but also how many people, very good, use the data that you have acquired. Then there is someone else in the center. Very good. And in fact, some institutions are trying to move in that direction but what I'm often seeing happen and maybe things as do they have all the papers that I'm going to cover many per year that. Oh, and then they also have to say great, but not, you know, they don't tell people what you were suggesting. That's a cultural. Yeah. I was curious also about some of the existing data required for positive or is the ability of some of these development systems to accept kinds of non traditional data, like different coming from, I guess, not traditional sources. I'm asking that also because it kind of went back to the training set issue. I noticed that for one of the training sets shown that like the Indian Ocean at zero. This is a running joke in our lab that basically all of our data comes from around the US and Europe and Japan. The southern hemisphere is very, very important. I think this actually one of the advantages of using this, like the larger stage of gene learning is that we can now these make visions. Assuming that the geology, some of the geology is similar, but, and also I think it highlights how much of going to going to these non traditional places to acquire. Yeah, there are a lot of stretches of the global. That is an undersea. Extremely important example. If you look at, you will get, for example, each of it. It's it's definitely centered around. So I want to be mindful of the time. I really want to thank our families. Thank you so much. Thank you. Thank you really learned a lot. Very, very thoughtful. Thank you. Thanks for the invitation to join you. Much appreciated. Much appreciated all of y'all's input. So here we are right on time. Marcia amazing. Thank you. Thank you so much. And I think we have, it looks like we have no rights. Not yet. Not yet. Okay. I think maybe a little bit of a bio break and then start the ocean life panel at 725. Is that okay? 95. My computer is still on that. I'm really excited about this next panel and the basin agreed to moderate this one. Thank you everyone. This morning we're not going to segue into biology. So we've got a session here going through the rest of the morning. We have seven speakers, five, five talks. We'll have a panel discussion that you can. We kind of chunked it up into three talks in the session. The primary things we'll do. We'll have two talks or so for each speaker. And then we'll have some time to meet to a little chunks to interact with. The first one is by Gabrielle. I'm going to share my screen and hopefully it'll work. And since it's built, they have 15 minutes. But we like to focus this first little session and session on how diversity issues. And that's followed up by Joe. But without me rambling too much further. Thank you, Jason. I'm going to share my screen and hopefully it'll work. And next, the question is. Can I go to presenter view? Can I ask, do people see the current slide or do they see my notes as well? We're seeing your notes. Okay. All right. Never mind then. We'll just. Stop that. And go straight. To here. Okay, are we good? No, we cannot see your position. You can't see. Oh, I'm sorry. One moment, please. You'd think I would have learned this by now. Here we are. Well, thank you again for this opportunity and honor to speak to the, to this committee on, on the ocean decadal survey. I'm speaking with, with my colleague Gabrielle here and we'll tag team this a little bit. I'd like to start with the punch line, which is that we believe that the coming decade should be a decade of ocean biodiversity. And there are, the question is why, and the answer is, because the ocean is alive living organisms. Make the earth and ocean ecosystem work. They capture in organic carbon converted into biomass that we use for food pump atmospheric carbon into the deep ocean and so on. Biodiversity is also important. And at the center of essentially everything that we care about as humans, food security, coastal protection. A favorable climate, even cultural identity. So without biodiversity, the ocean is just water. This audience, of course, gets that, but not everyone does. If ever there was a grand frontier for science, it is understanding what that biodiversity is doing. So what do we need to cross that frontier? We propose that we need a major initiative with 2 components. The 1st. Is a dynamic map of ocean biodiversity. The 1st step in understanding how biodiversity is is influencing ecosystems is knowing what we have and where it is. This is a map of the current long term biological observations of the ocean of ocean biology that are known. They span about 7% of the global ocean surface. And of course, much less of the deep ocean. Secondly, in addition to that dynamic map, we need some way of systematizing the functions of the species and how they are working in ecosystem functions. In ecosystems, something akin to a periodic table of niches. So this is not a new frontier. The biodiversity in 1992, the nations of the world got together and agreed that biodiversity loss is an existential challenge to humanity along with climate change and desertification. That was a generation and a half ago. National Science Foundation also has identified the decline of biodiversity and what that is doing in ocean ecosystems as a key challenge. As reflected in the previous decadal survey from 2015, what is the role of biodiversity in the resilience of marine ecosystems and how will it be affected by natural and anthropogenic changes. We have made a lot of progress and answering that question, as I believe Dr. Bernhardt will mention in the next talk, but it is still with us. So one might say that this is a long standing old question, but what's new is that we think that in this coming decade, we have the opportunity really for the first time to be able to answer the question. And also what's new is that we are increasingly being pressed from multiple quarters for answers to these questions that require data on biodiversity. For example, we now have pressing national demands for biodiversity data that are emerging and that we are frankly struggling to meet society in the government of the US get the importance of biodiversity and this is illustrated by a flood of policy documents over the last several years, including the president's America, the beautiful initiative, the national nature assessment, which is just getting underway now, the ocean justice strategy, which is tied to the quality of natural habitats and resources. And finally, the transforming transformations for sustainable blue economy, which is very largely based on living resources in the ocean. And we need to know how those species are functioning in order to do that in order to really bring the blue economy forward. All of these aspirations require data on what species we have and what they're doing and we are missing a lot of that data. So again, the first step we think is a dynamic map of ocean biodiversity. I say dynamic because the ocean is alive and even more so than on land organisms are constantly moving and changing. So this is this is a long term sustained issue to manage those resources. We need to know where they are on land. The federal government has invested a lot over the long term in understanding where biodiversity is through, for example, the USGS gap analysis program, which also addresses the state of protection of biodiversity and how it may be threatened by climate change. So this figure shows the map that was published in the New York Times a couple of years ago that shows the imperiled where biodiversity on land is most imperiled. This map shows where biodiversity is most at risk in America, but it doesn't show North Atlantic right whales. It doesn't show polar bears. It doesn't show the endangered staghorn and alcorn corals in the Florida Keys. So there there is a key gap in public appreciation and understanding, I think this is the kind of thing that we need for the ocean, which in this map is simply a blank white space. So the second part of a decade for ocean biodiversity is some way of systematizing the functions or niches of the species in the system and what they're doing. So on the left, we have something like a the kind of models that we usually use for understanding how species interact with some of their traits and how those may influence functions that emerge as fisheries yield atmospheric composition, export flux and so on. How do we organize those and how do we get the data to organize those into systematic ways of thinking about how body size and stoichiometry and many other traits of species influence functions something akin to a periodic table of niches. This is a topic that has been dabbled with, I might say, in ecology for a number of years, and it's a very difficult task, but we may we are on the cusp of being able to do it with, for example, publications of major new gene catalogs of ocean organisms and so on. So at this point, I would like to turn this over to my colleague Gabrielle to continue with where we are going in the future. So I think if you would advance the slides and it to the next slide please. So we can do this now. Underpinning a periodic table a dynamic map is the explosion of new technologies that are going to be critical in meeting any new demands for biodiversity data. You as a committee talked at length about AI image processing machine learning big data requirements this morning. You'll be discussing EDNA later today and hopefully there will be further opportunities to talk about some of these technologies in your in your future conversations. We're advancing coordinated genomics approaches high resolution remote sensing active and passive acoustics telemetry, other other approaches but for most of these groups are also considering standardized data management and data flow. Also data storage and archival to ensure wide availability of the information and importantly that it's available to support action on the ground for management and decision making. So that's an important gap to consider and one that we hope the committee will consider technologies promise to enable observation over the temporal and spatial skills for which we're being asked for biodiversity information. And also in difficult to reach areas and in some cases, importantly at lower costs than some traditional methods. Innovations in data systems in cloud computing are happening in tandem with in water and remote sensing technologies and will further enable all of this and we need it to really expand the availability of actionable information for sustainable and equitable development and also responsible use of the resources that we're tapping. So next slide please. We do have a strong foundation now for the science and data that are required to establish and maintain that dynamic map and to understand species functions. And this is an evolving space and this positions us differently than in 2015. It includes decades of exploration and data assembly work by many organizations and individuals, including some of you who are participating today. We have the networks on the picture here and this is not a complete picture but many of these are focused on co development of approaches with users and communities to ensure relevance of the science and data and that those science and data output outputs are benefiting society and openly accessible. Some of these are insufficiently resourced but offer great promise to advance our national needs for ocean biodiversity information and, in fact, we think this represents a significant body of work and existing partnership and investment that NSF could really effectively leverage. The census of marine life produced the first nationwide overview of US marine biodiversity. So we have that overview from the census but it was really qualitative. We can now get quantitative. So to the next slide please. So we have the tools to get quantitative but we can't measure everything everywhere all the time. So how do we prioritize what to measure where to measure and what data to be collecting. The global community has identified frameworks for core biology and ecosystem measurements that we have, you know, as a nation, we have yet to adopt in a systematic way. Building on that a recent global expert task force developed a framework to assess the abundance and distribution of marine biodiversity and the link to that work is here on the slide. But it prioritizes species and habitats that are ecologically important, societally important and feasible to measure and it helps to inform priorities for science and stewardship and an investment in that. We applied the framework in a case study for US marine protected areas and determined that we likely overestimate biodiversity protection because of data gaps, including sparse information from outside of MPAs sparse information from ocean versus coastal waters and for invertebrate species. So there's a lot of opportunity there. Next slide please. So we have tools, technologies, networks and opportunity. We also now have a mandate from above from the highest levels of US government for the specific focus on this work. So this is a, you know, a great window of opportunity to come full circle back to the purpose of the session the grand vision that we propose will really require a major scientific investment that is directly aligned with NSF's mission. And in fact, NSF is well positioned and I think we were hearing reflections on that this morning to lead a US wide initiative across disciplines and methods to enable biodiversity understanding. The US is developing a national strategy for ocean biodiversity that will provide high level vision for actions to support ocean biodiversity coordination communication conservation. And the scoping of that is underway and this turtle represents a QR code that you could click on and give us input to the development of that strategy. But implementing the strategy will depend on scientific advances and really critically effective leveraging and partnering and NSF and the community that it supports. Well NSF is engaged in the strategy development and will continue to be central to this. It's going to require vigorous wide ranging research to better understand how living nature figures in our lives in a changing world. And it's important to note too though that this strategy will be complemented by a national aquatic DNA strategy that will also be released this summer, focusing on one key tool for biodiversity understanding. So I think maybe the next slide Emmett was a closing slide. Well, it's, it is very much a closing slide. Most if not all of the challenges that we're facing as a society in the coming decade are intimately tied with the network of interactions with living species and habitats around us. And responding to climate change is inextricably linked to the changes in biodiversity and I think we'll hear more about that later. And fundamental challenges social equity is also inextricably linked to the challenges we face with biodiversity loss and change. So effectively addressing the challenges of climate of social equity will fail without incorporating and really implementing a big vision for biodiversity. So thank you so much for the opportunity to sort of set the stage for this conversation today and I know we're looking forward to some good discussion. Thank you, Gabrielle. I appreciate that. What we'll do is hold up on comments and questions you're talking about. We'll go to Joey's talk and then when she's done we'll open up to you. Joey, can you all see my slides here? Okay, we're on the wrong screen here for me, but I'll just turn my laptop this way. Okay, yeah, thanks. Thanks to Jason for the invitation to speak here today. I'm really excited to give you a little bit of sense of my perspective on biodiversity change and how I approach studying it and what the implications of those changes are for human well-being. And I'm going to do my very best to stick to my time, but I may have slightly, you know, so you might have to poke me with a stick here. But before I tell you about the work that I do in this realm, I think it's important to tell you how I come to this work. This is a photo, not far from where I grew up on the West Coast of Vancouver Island in the Pacific Northwest. This is on the traditional ancestral and unceded territory of the Coast Salish peoples, including the Squamish, Slewa-Tooth and Musqueam nations. And for me as an ecologist and as a biodiversity scientist, it's important to recognize that the biodiversity that exists within these traditional territories exists because of thousands of years of effective stewardship on the part of Coast Salish people in the past, in the present, and ideally into the future. And what you're looking at in this photo is not any old stretch of shoreline. This is actually a clam garden, which is an ancient form of aquaculture that Coast Salish people use to maintain high levels of shellfish productivity and biodiversity in the intertidal zone. And the abundance of these shorelines is reflected in one of my favorite traditional Coast Salish sayings, when the tide goes out, the table is set. And what I really like about this saying is that, you know, in just seven or eight short words, it evokes a living system in which human well-being and ecosystem health are intricately linked, not only through the food benefits provided by the shellfish, but also the cultural and spiritual benefits. The other component here, when the tide goes out, evokes a changing dynamic living system. And coastal systems like this one and others around the world support livelihoods, but these systems are not only, you know, experiencing daily tidal cycles, but they're also now experiencing rapid planetary scale change. And this rapid planetary scale change is effectively threatening the ability of these systems to support livelihoods as captured in the United Nations Sustainable Development Goals. And so I would argue that understanding how living systems across scales respond to environmental change is the most important scientific challenge that we face over the next century. And so in my work, I aim to develop a mechanistic understanding of biodiversity change in aquatic systems, and then connect that understanding to consequences of those changes in biodiversity for human health and human well-being. In particular, I'm going to dive into an example today that focuses on seafood and human nutrition. But when it comes to generating a mechanistic understanding of biodiversity change and ecology, this often requires us to connect our understanding of processes operating at another scale. So for example, if we want to predict how a community will respond to a change in temperature, we might need to know something about how the populations within that community are responding to a change in temperature. And in order to do that, we might need to know something about how the individuals within those populations are themselves responding to a change in temperature. And so kind of one of the central themes of my research program is to develop and test theoretical frameworks that allow us to connect understanding of processes operating at one scale to dynamics and outcomes at another. And I do this across scales from processes operating within cells all the way up to human well-being. So Jason told me I should give you my bottom line up front here today. So my bottom line up front is that theoretical and empirical evidence demonstrates that biodiversity at multiple scales is critical to enhancing resilience. And where we define resilience as the capacity of a system to maintain functioning structure and feedbacks in the face of environmental change. And then I'll give you an example of how biodiversity and maintaining biodiversity is critical to support human well-being. So I'll start just by giving a quick overview of the evidence that we have that biodiversity is critical to enhancing resilience across scales of biological organization. So over short time scales, biodiversity in the form of genetic diversity and phenotypic diversity and species diversity enhance resistance to change by increasing the range of responses to the environment and the likelihood that species can functionally compensate for one another following a disturbance or an environmental change. And then both over short and long time scales responses to changing environments include a combination of phenotypic plasticity and rapid evolution of traits better suited to new conditions. And in the case of evolutionary adaptation, we know that high genetic diversity, for example, facilitates rapid evolution to environmental change. And so together, these processes enable population persistence in changing environments. At higher levels of ecological and biological organization, we know that connections among species, populations and ecosystems contribute to self-organization and stabilize ecosystems in the face of fluctuating environments. And they can enhance recovery following severe disturbances. And so for example, receding of individuals from other sites. So across communities can prevent local extinction following disturbance. And so overall, we have evidence across scales that biodiversity plays an important role in maintaining community stability over time by increasing the chance that species will be resistant over the short term. Also by allowing species to functionally compensate for one another and by facilitating processes such as recruitment, which ultimately enhance recovery over longer time scales. So given the evidence that biodiversity enhances resilience, the challenges that we're now living in a biodiversity crisis where biodiversity itself is declining in many ecosystems and there's growing concerns that these changes could directly impact human health. And so my research here in this area attempts to answer this question of what do these changes in biodiversity mean for human health and human well-being using biodiversity theory. So biodiversity theory predicts that for species to coexist, they must differ in their resource niches. And this leads to complementarity in resource use across species, which ultimately leads to a positive and saturating effect of biodiversity on ecosystem functioning. This is an example from grasslands. This relationship has been observed in dozens, if not hundreds of different studies. And we can actually estimate the strength of this relationship by fitting a power function to this relationship and this B scaling exponent here quantifies the biodiversity effect. Oops, sorry, skip the slide there. And within this framework, we've connected our understanding of changes in biodiversity to ecosystem services. But the vast majority of cases in this context have been focused on services that are based directly on yield, total ecosystem production, for example. But there are actually very few robust links in this framework between biodiversity and human health and human well-being. And we know that in the case of human health and human well-being, total biomass yields are not predictive of health benefits. And we have really nice example of this from Bangladesh, where we've observed that over a 20-year period, people switched their diet from consuming a high biodiversity diet comprised of many species to a low biodiversity diet comprised mostly of farmed carp. And over this 20-year period, even though people consumed more seafood biomass overall, so total yields from the system increased, micronutrient intake actually decreased, leading to a problem called hidden hunger, in which people have access to sufficient protein and calories, but insufficient micronutrients. And the reason for that is that seafood is valuable in the human diet because it provides a range of critical nutrients essential for human health, including protein and fat but also omega-3 fatty acids and micronutrients like calcium, iron and zinc. And so what you can see is that adequate human health is really like a multi-dimensional optimization problem. It's about not just reaching one of these nutrient targets alone, but reaching them all simultaneously. So here we use biodiversity theory to test the hypothesis that seafood species richness enhances nutritional benefits and that it's not just the number of species in a system that matters, but it's also the diversity of functional traits like Emmett was talking about that matters. And so we tested the hypothesis that ecological functional diversity is positively associated with nutrient diversity and therefore increased nutritional benefits. So I'll just give a quick overview of what this looks like. We envision a scenario of high biodiversity in the diet and if species differ from one another in their nutrient profile such that species are complementary, one species is high in calcium while another is high in zinc, then biodiversity theory leads us to predict that we should see enhanced nutritional benefits. And in this way we can then connect biodiversity and changes in biodiversity within a system directly to human health and human well-being because we would predict then that in a case of high biodiversity that a greater proportion of the human population would meet their nutritional needs. So just a quick overview of results. We found that there's absolutely no benefit of increasing biodiversity in the diet or maintaining biodiversity in natural systems when it comes to protein provisioning, which is this yellow line here. But when it comes to calcium and zinc and the micronutrients, we actually see strong effects of biodiversity where as biodiversity increases, the minimum portion size required to reach a given nutritional target decreases, which means that we can reach our nutritional needs more efficiently with lower biomass consumption overall. And finally, we found that nutritional benefits are not just associated with species richness but also ecological functional diversity. And so communities that are comprised of a higher or a broader diversity of ecological functional traits allow us to reach a wider range of nutritional targets simultaneously. And so to summarize that, that link between ecological functional diversity and nutritional benefits is important because it allows us to link our understanding of the processes that maintain biodiversity species interactions, like competition, importation to the benefits that these systems provide to human health. And so to summarize what I've demonstrated to you is that biodiversity in this case we've demonstrated is critical for micronutrients but not for protein and that biodiversity is essentially essential to meeting nutritional requirements efficiently. And so this is important implications when we think about balancing sustainable sustainability goals for human health and human nutrition with also maintaining biodiversity. So what we're showing is that if we can maintain or restore biodiversity we can actually reach nutritional goals more efficiently. So I'll conclude by returning to this challenge that I alluded to at the beginning, which is that kind of a major challenge is to understand causes and consequences of biodiversity change and the consequences of those changes for human well-being. And I demonstrated that an approach to addressing this problem is to develop and test theoretical frameworks that allow us to relate processes at one scale to outcomes at another. And I would argue that this is a powerful approach because it allows us to generate a general understanding of change in living systems and what those changes mean for human well-being. And I'll end there and say thanks and I'm very happy to answer any questions and contribute to discussion over the next for the rest of the session. Thank you Joey, appreciate your talk. We have a few questions already in Slido. The one I want to flag and Allison, I'll ask the four-year-old. Joey, sit tight for a second, but Emmett and Gabrielle, you also helped the ocean biodiversity summit a few weeks ago. I'm wondering if you could give a real brief summary of the outcomes of that, just for the communities benefit and to harmonize where that is going with some of this on the table. And particularly the Allison, you asked me if there was an NSF presence there? There was an NSF presence at the summit. And we've been engaged with some NSF colleagues sort of in the development of that in talking about priorities, for example, for tribal engagement. And also, NSF is an active member of the National Ocean Biodiversity Strategy writing team. So there's a lot of nice linkage there. I can start with the summary of the summit and then Emmett can pick up whatever I miss. But we convened the summit last month. We had about 100 participants from across federal agencies, tribes and territories, industry, investment, exploration, really dynamic NGOs, a really dynamic group. We tried very hard to get broad participation because we wanted to bring together leaders across sectors and communities to give perspectives about why the focus on ocean biodiversity right now is so critical. And why leadership engagement and deep support is really needed to bolster some of those kind of grassroots efforts that I pointed to in one of our slides this morning and transform and advance our capabilities with regard to ocean biodiversity understanding and maintaining that understanding over time and developing and understanding I think speaking to Joey's key points about how biodiversity is also intertwined with some really fundamental human needs. So we invited conversations during the summit. Well, first we had an kind of a fireside chat with Jane Lubchenko from the White House and Andrew Steer from Bezos Earth Fund, talking about the context including looking back to Rio for the ocean biodiversity dialogue and the need to really develop our collective narrative about the importance of a focus on ocean biodiversity in this time in this decade. And so they carried really strong messages there and we can put the link to the summit in the live stream in the chat or the recording. And then we we hosted a series of conversations to really invite points of view on the needs for ocean biodiversity information among frontline communities who are really experiencing deeply this change in terms of biodiversity loss or shifts. We invited a conversation about different perspectives on valuing biodiversity so you know how we think about biodiversity from the point of view of how to invest in protections or conservation, but also the value of biodiversity from a cultural perspective. Many groups to find value differently and we really need to embrace that and understand it. And then a third conversation around exploration and innovation, taking a broad view of that and thinking about innovation from a technology perspective but also innovations that are needed in terms of policy and societal engagement and that sort of thing. And so we had some really robust panel discussions. There is a flurry of activity coming out of the summit we heard a lot of really positive feedback that those leaders that we invited were really welcome to the opportunity to interact and engage and they've been actively following up with us about opportunities to pursue actions following the summit and that that was our biggest hope for outcome was that we would start to see a different level of action and awareness and I would say that's come that's that's become real. So it's a it's a nice way to sort of underpin and bolster the work that that we've been doing and the engagement we've been trying to invite around the National Ocean biodiversity strategy, which we will release this summer. And as, you know, as we mentioned, the first step in that strategy will be to really focus on implementation and what actions are needed, and what groups can engage because it's not just a federal enterprise that needs to move that forward. And then what would you add. That was a great summary just a couple of sort of subjective impressions to add on top of that, you know that there was tremendous energy which was really exciting and of course gratifying for us, you know, having worked in this field for a while, you know, I think there was a real sense that we are at a tipping point in terms of really, first of all, understanding the importance of biodiversity broadly speaking, you know, in any in the earth system and in human society. And secondly, in the, you know, the, the motivation to really do what needs to be done and we heard this from private sector and investment firms and so on. So it's very exciting to see the broader, the broader community getting involved here. So just one thing as well is about the communication and how we really present this to the larger society, those of you who who follow Jane Lubchenko's work, you know, have heard her talk about the ocean being too big to ignore. And what she and Andrew steer talked about was that we need a new narrative for biodiversity. And I think it's really striking that that Jane does not use the word biodiversity. She talks about nature and we are the United States is starting this national nature assessment which for all practical purposes is a national biodiversity assessment under a different name nature resonates more with people. And so the idea of, of working on a new narrative, I think is has a compelling aspect to it. So, yeah, I'll stop there. We have a few other questions, but some of these housing. Do you want to jump in and ask you a further question. I can't. So, both Peter, and who else wasn't asking about the role of NSF and supporting biology or CM or any of the panelists have any sense of how well we're going to support biology and whether or not that's changing and touched on a little bit in your response to the last question, but I think that would be important for us to have a sense of that. So, that was a question for us or for the committee. I'm sorry, thank for us. It's interesting that you asked that question because we did check in actually with our NSF member of the strategy writing team on this very question and, and I have some specifics. With regard to awards or grants NSF has funded 130 awards, starting in 2014 with a total of approximately $62 million. And so that's, and then there's a number of publications cited as a funder in 135 publications since 2014. And these are things coming out of the dimensions on on biodiversity initiative, for example, which some of you may be familiar with and that's probably a good initiative to look back at. I'll just add that it's a little bit difficult to answer that question because of course there's a lot of research funded by NSF that might not have biodiversity in the abstract but but of course it is really important in that way. You know NSF has really been a major supporter of this kind of work. Not just in ocean sciences, but through the systematics and, you know, programs in in D. E. B. and the L. T. E. R. is that, you know, have to understand the biodiversity where they're working. So, especially I think for the fundamental science the basic science Heidi mentioned in the earlier session about it's so critical that you actually have taxonomous who have that knowledge that are able to calibrate the DNA and the remote sensing and all of these other things. And so much of that is also supported by some of the other federal agencies. Noah has systematics lab at the Natural History Museum in Washington USDA as well. But NSF is certainly a major funding funder for for ocean ocean biodiversity, both both that basic level and also, you know, the ecology of what those pieces are doing. Thank you. She, I'd like to invite me to ask you a question. Yeah. Thank you. This question is for Jolie. What are some key partnerships to achieving the projects like you mentioned. You know, you talk about bringing ecological health for human well-being and I know you probably have time to get, you know, really into how you might want to talk, you know, qualitatively or quantitatively measured well-being, but you must have some partnerships. And you want to get your thoughts about partnerships. But as part of our charge, you know, we're supposed to identify strategies and opportunities for partnerships such as this. Yeah, great question. I think you can hear me. I've lost my mouse. So the question was about partnerships. And what sorts of partnerships we should be investing in developing in order to understand connections between ecosystem health and human well-being. Is that what it is? Is that I might understand your question correctly? Yeah. So I guess I would say, yeah, I would all kinds of partnerships. We currently, where I'm working now, we have established partnerships with government, you know, relevant government agencies, provincial, federal, you know, I'm in Canada now, so fisheries and oceans Canada as well as provincial governments. And then also developing and establishing trusting relationships with indigenous communities is something that we're investing a lot in right now. And those are partnerships that take a long time to develop establishing trust and so on. But I think the key thing about understanding biodiversity and its connections to human well-being is that those relationships are going to differ in every single community. So for example, the argument that I gave today that protecting or restoring biodiversity is critical for human health and human nutrition is a argument that might work for some communities. But for example, for some of the First Nations communities that I'm working with now, that argument doesn't make any sense to them. It just doesn't resonate to them. All forms of life are important and it's not because of their micronutrient content or their calcium content. It's because nature is kin, nature is family. And it doesn't make sense to value nature as we would when you look at the side of a cereal box and you look at, you know what I mean, like the, our recommended dietary allowance that way that argument just doesn't resonate with them and it's just not a compelling conversation around protecting biodiversity. So I guess my point there is that I would, I'm a big, I would just underscore the importance of understanding connections in multiple contexts in multiple, multiple communities, multiple cultural contexts because those relationships are really differ depending on the community that you're interested in working with or understanding. Thank you. We have one more question to wrap it up. I'm going to call on Rick to ask your question to essentially channeling Alison's question. Alison, you want to go, you want me to. Okay. Hi. So this is sort of a combination question Alison framed it as what do we still need to understand about the importance of biodiversity that we don't know now. And I followed up with a little more specificity in the sense of, and I say that my question is truly coming from a friendly place. What is different now about the need for further research on biodiversity in the context of what are one or two big drivers that are not more of the same. I'm asking the question, because I need help when I'm asked that question, what my answer should be so thank you. Joey, I'm waiting to see if you're going to take that first. Great. Can we so the question is, can we, what is different now about understanding the importance of biodiversity is that I'm getting that. Essentially, because as you pointed out it wasn't see change and there's a ton of other. I mean we all get about biodiversity but really what are the specific one or two big drivers now that are not just needing more of the same more research and what's unique about this moment in time. Yeah, okay. Well, I can take a stab at it and then maybe Emmett or Gabrielle can follow up but I think the key thing is that now we are experiencing rapid scale change in biodiversity in a way that we haven't that's unprecedented. So the rate and magnitude of change in itself warrants attention now relative to other times. And the other component is that I think we're still in early days of discovering and understanding connections between biodiversity and human well being. So the link that I demonstrated today between biodiversity and human health is one of just a few like a handful of actual robust links that we have directly to human well being. And the challenge there is that human well being is multi dimensional and it's difficult to quantify but we still, there's still so much that we don't understand there so links between biodiversity and disease prevalence and transmission. There's many, many examples that that's still a mechanistic understanding of how biodiversity per se is changing and what that means for human well being we there's still so much that we don't know so I would say there's just. There's there's a there's a lot still to understand and discover there. That would be the argument for me for why we should be studying it. Yeah, I would add a couple of points to that I completely agree with with Joey there I mean, you know, perhaps not 10 years ago but shortly before that people were still arguing about whether biodiversity actually has a significant effect on ecosystem function. And in the specific context of the, you know, the number of species, which is, of course, only one, you know, axis of biodiversity. That's changed. I mean, it's essentially accepted now that that's the case, but it still remains true that the vast majority of research that's been done connecting biodiversity to ecosystem function has been with with plants and primarily terrestrial plants. You know, ecosystems are much more complex than that. You know, Joey's work is a really nice exception to that in in going beyond terrestrial plants but also in the really critical aspect of connecting that to people that that is still a frontier. And because it's not necessarily a, you know, a straight pipeline that goes from biodiversity to function to people people have values. As Joey mentioned, different communities consider the same kind of service in different ways. So I guess if I had to summarize it, I would say that the challenge now is really increasing the resolution. So instead of using, you know, box models with NZP, we now we now should be able to do box models with with, you know, dozens or even hundreds of different boxes representing different, you know, biological components as as as the AI and computational capacity and knowledge are able to catch up with that. Okay, thank you both for that. That was very helpful. I know we're running a few minutes late. I see Lisa from NSF has a hand up Lisa, please wait. Thank you Jason. I just want to put one other spin on and come back to something that Joey's been saying to the ocean climate action plan that Emmett listed as one of his documents. I'm not going to lie, it's the ocean manipulation plan. And so there's a big urgency around how are we going to impact all of this amazing living ecosystem, as we necessarily manipulate the ocean to help on on climate action. So, and, and, and I do think that it's pretty different now in our framing to of the. I like to listen and on cold production stuff to write non human relatives. We have a different framing of what ecosystems actually are and what they mean to us so I think both of those are significant changes over the last decade. Thank you. We're going to shift gears. Before we do, I want to thank the three panelists. So far, I think he gave us an excellent overview of not only bio diversity in the marine system for staff so that wherever you are. And now we're going to switch gears to how biodiversity and other things might be responding to climate change. Thank you. Hi folks, thanks. My name is Peter. This it's my creep. Let me be the moderator for this panel on responses to climate change. I want to start by thanking our panelists for being here today. I know it's, it's quite a bit of work and I want to really appreciate your efforts and putting together your slides. Our panelists are going to be talking about our organizational and ecosystem responses to Earth, changing climate. And I've asked each panel is to keep their presentation to 10 minutes. And the hope is that we get to the presentation today just for the half hour for discussion. So committee members and colleagues listening in while the panelists are presenting please please type your questions and slide out. And that will give us a chance to try and address as many of them as possible. We are a bit behind, so we'll probably be wrapping up close to 1115 or 1116. So without any further ado, let's go ahead and get started. Our first panel is today is mainland, since he from University of California, the Santa Cruz so mainland floors yours. Thanks so much. I'm just there is the share screen. That work out. Can you see my screen? That's great. Thank you. Perfect. Yeah, thanks so much. The work this committee doing is vitally important and I really appreciate the invitation to talk today. I'm male and pink associate professor at the University of California Santa Cruz. And today I was asked to talk about what I see as key research opportunities related to species distributions and climate. Hopefully can complement what my colleagues contributed already being many, many overlapping themes. So just to put them up front, my main messages are that species are rapidly shifting to new locations in the ocean, and that this has dramatic impact on marine ecosystems and society. Second, there's a massive hole in our knowledge related to the biological processes that drive these ships and the ecological and social mechanisms of adaptation to these ships as well. Third, there's a strong need for technology data and computational infrastructure to observe species distributions, especially in the tropics and southern hemisphere, where few observations now exist. And finally, I'd argue that we need stronger workforce development in data literacy and the data science skills to enable the interpretation forecast and analysis of adaptation options to these climate impacts. So on my first point, we have many observations like the following where this map shows in orange and red. The American lobster used to be found all the way south to Virginia and the Northeast US. But now has populations across southern New England have largely collapsed and the population has contracted hundreds of miles north as temperatures in the northeast rapidly warmed. More broadly, marine species are shifting nearly 60 kilometers per decade toward the poles on average. It's about it's nearly six times faster than has been observed on land. So put another way, marine species are the canaries in the coal mine when it comes to climate impacts. So these ships also reshape entire ecosystems. One example is when urchins shifted south into Australian kelp forests and in effect, hit them up really dramatic changes to those ecosystems. These ocean changes also reshape human society as well. Just to show you one example, this map shows with a red square where large fishing trawlers from Beaufort, North Carolina were fishing in the mid 1990s. Boats now travel 500 kilometers north, following summer flounder that shifted north as well. Even more common in fishing operations are changes in the species that fisheries target. Seafood is the most widely traded food commodity. So this has massive economic and political implications as well. The Environmental Protection Agency estimated that future distribution shifts of commercial fishery species in the US could cause the loss of at least a billion dollars every year by the end of the century. Distribution shifts also cause political conflict. We've seen legal conflict between states within the US and trade wars and failures to operate on sustainable fishing, even between countries that are close allies internationally. So, and yet despite the importance of species distribution shifts, we don't understand why they're happening and why there's such a wide variation from one species to the next. We know climate and changing ocean conditions are the ultimate driver of many of these shifts, but it's the biological processes in the middle that remain a mystery. Current approaches to this problem often use what are called species distribution models that in effect assume species directly followed the environmental conditions with which they were historically associated. Even though we know this is much too simplistic, these models get almost all the biology that we know is important. Just to unpack this slightly, shifting species distributions consist of two processes. There's leading edge expansions and their trailing edge contractions, shown on this slide as persistence, which is the lack of contraction. So trailing edges contract only when a whole host of biological processes fail to enable persistence that includes physiological tolerance, behavior, phenotypic plasticity, phenological shifts and adaptive evolution. And yet we don't understand yet why some species are able to persist while others contract rapidly. At the leading edge, it's the processes of dispersal and population growth, as well as acclimation and evolution that enable that leading edge expansion into new geographic areas. But again, we don't understand why some species expand quickly, others slowly or not at all. It's in effect this missing biological middle in the species distribution puzzle. One of the part of the problem is that we historically have studied this process within disciplinary silos, rather than integrating across biological scales and across oceanographic disciplines. A related problem is that most of the research has focused on the physical environment, and yet we know that changing species interactions can have even larger impacts. So these are the indirect effects of climate change, and we're not yet able to generalize beyond individual case studies. So to address these questions going forward, we have a wealth of observations on where species are and where they have been, including from opportunistic observations and from scientific surveys. So these are really an underutilized resource to understand changes in species distributions, especially when combined with data on historical ocean condition and biotic environments. But what's needed is to more effectively integrate these observations with mathematical theory, and especially process-based mathematical theory for these range shifts, and with the genomics and the experiments to measure rapid evolution, species interactions, and physiological tolerance. I propose that what we need is a more focused oceanographic research program focused on integrating across oceanographic and biological scales to address these issues. The second key research need I want to highlight are the social ecological feedbacks around species distribution shifts. Coastal marine ecosystems are not only heavily impacted by human behavior, but changes in these ecosystems also strongly influence human behavior, including how and where to pursue fishing, recreational opportunities, biodiversity conservation, even shipping and offshore development questions. And many of the key questions are around how social ecological systems adapt to species distribution shifts, how feedbacks flow through these systems in both directions, and also which structures or adaptation approaches achieve societal goals. So I think this represents a key opportunity for collaboration between ocean sciences and the director for social behavioral economic sciences on a funding program to address climate impacts on social ecological systems in the ocean. There are also infrastructure issues to consider as well. So this map shows with distortion those areas with more observations of distribution shifts, and you can see by how small they are that the tropics and the southern hemisphere are poorly observed. This is an equity and justice issue in addition to a scientific impediment. So in many cases, data on species distributions in the tropics and the southern hemisphere do actually exist from historical surveys, but they're not available for research. But changing that requires long term engagement to build trust to develop culturally appropriate methods for sharing data, and to develop new programs for observing where species are and how this changes through time. This is a case where single sites aren't enough, especially to understand changes in species distribution. So instead we're talking about surveys across regions, 1000 kilometers or larger, repeated at regular intervals and sort of a regional long term ecological research program. I think it's especially focused probably on the on the tropics and some of the southern hemisphere oceans new methods like we'll hear about around environmental DNA, but also machine learning from images can be can be useful here though. I also think we don't want to over overlook low tech observations that work really well in low income context. This would also help create the dynamic map of ocean life that Emmett and Gabriel talked about as well. Understanding species distribution responses to climate requires making sense of massive quantities of data, linking mathematical models with data and the computer science skills to operational operationalize the pipeline from real time biological observations to forecast. So these are highly transferable skills across fields and skills that are in high demand by industry so this argues for workforce training programs that prioritize data literacy, as well as data science skills at all education levels. So the benefits of this research infrastructure and training would extend far beyond NSF to benefit national security, especially around questions of conflict and international relations over shifting species, but also benefit living marine resource management including the climate ecosystem and fisheries initiative and more broadly development of public climate services for the biological impacts of climate change, but everyone has the information to adapt to climate change, not just the wealthy and the corporations that can, they can afford it. So with that, thanks so much for your time and I really look forward to discussing this more on the panel. Thank you. That was fantastic and right on time. So really appreciate that. Next up, we'll be hearing from Steven Marazzi from the University of South Florida. Yes, good morning everybody and it's good to see so many familiar faces. So I want to follow up on a number of points that both mail in and I think Emmett made about the importance of not only understanding the distributional shifts of animals and plant systems due to climate change, but also their functioning in the system and how these species interact. So basically like to do three things. Number one, talk a little bit about the co evolution of species and how species adaptation can occur under climate change and what the implications of change variability are. I'd like to cite three quick examples of evolving communities that that we have indexed and the potential changes not only in the in the distributions of animal but but as importantly and perhaps more importantly, the ecosystem dynamics underpinning them. And then last I'd like to follow up on some recommendations some of which you've you've heard in a different context already. So if we look at the sort of potential biological implications of, of changing climate that there's a whole host of things that we would anticipate happening under, under warming seas under sea level rise ocean acidification and other symptoms of a claim changing climate. And they include metabolic rates. Mostly, you know, the some of these are temperature dependent processes that we would assume that metabolic rates would increase, although not not you necessarily uniformly obviously implications for productivity and growth, which can go either way. Primary productivity, trophic interactions. These are difficult to predict because of the spatial processes and dislocations that occur and can occur under climate change. Obviously competition and predation are are really important elements of this, particularly when under climate change you may have predators and prey changing distributions differentially. One of the particular important aspects of understanding this dynamic is looking at so called Depensatory Mortality and Ali effects at low population sizes. And in particular, as we try to recover many ecosystems that have been degraded. We also have the compounding effects of climate change and so this represents a really major issue. If we have things moving around in space and that that basically implies that we're going to have invasions and potentially local localized extinctions as well. And so understanding, you know, how a species moving into a new area can be accommodated in the system that's there. And of course, this spins off issues of host parasite relationships, disease implications, etc. One of the things that really confounds our ability to deconvolve the climate signal is that we have so many other simultaneous drivers and systems including fishing, huge vacation and other human derived. Thrust the system rate paper by Scott don't at all just published on some of this that is worth reading. So when one looks at a system, you have a series of co evolved species that distribute themselves on various resource continuum. These are just two dimensions that we often look at in the ocean temperature and depth and you can see that, you know, these are overlapping ovals in the sense that there are tales of these distributions where species overlap. The idea, primarily is that the oval, the major part of the oval doesn't overlap so as to minimize competition between co evolved species. Under climate change, we can we can actually anticipate these oval shifting around a bit. And so, for example, you could see, you know, the two species on the right hand side, moving off as a couplet. And so they become disassociated with the other species in the system. You can also see two ovals where actually, you know, they don't overlap or that they simply move in in depth space to a different place and potentially interact with other species. So you can imagine how many dynamical processes are involved in these kinds of things. So I'd like to shift gears and actually put some meat on these bones a little bit with a couple of work examples. Here's a very recent study by Kathy Mills at the Gulf of Maine Research Institute and her colleagues looking at. These are the NIMS trawl survey data from the Northeast region and you can see that over the 70s to the, to the early 2000s versus the 2010s, you can see the center of mass of many of these species that shifted northward, which is on process that Melon talked about. And when you look at these species relative to how they're, sorry, they're, how their temperature preferences and the underlying temperatures have changed. You can classify these animals into what we would call ineffective trackers. That is, they're tracking along, but they're not changing as fast as a temperature effective trackers, which are moving along. And some that have no response at all, nor did they experience warming and then the so-called effective trackers, right? The animals that might be most sensitive to change. And interestingly, if you start looking at the ecological processes involved, these are long-term changes in the growth rates of many of these species. So for example, you see long-term negative changes in the growth rates of haddock on Georgia's bank and other things. And so, so understanding that these movement patterns of various species come with implications for the population dynamics of animals and communities is important. The second example I want to cite is a very well-known situation where we've got the so-called march of the mangroves. It's occurring not only in the United States and the Gulf Coast and the Atlantic Seaboard, but all over the world. The mangroves are limited by lower lethal temperature, and as the temperature warms, what we're seeing is expansion into higher latitudes. And so what's happening here is they're displacing existing types of ecosystems, and in particular salt marshes. So what are the implications if you're just substituting one habitat for another? It's a really interesting study where they paired up so-called ecotones where you've got salt marsh invaded by mangroves and the underlying salt marsh. What you're seeing here is a large difference between mangroves versus the Spartan or the salt marsh grass in terms of their rates of decomposition. And so as the rates of decomposition of an evolving ecosystem change, that means that things like the structural complexity, nutrient cycling, and other things are going to change as well. The third example I want to cite is of northward range expansion of inshore species called fiddler crab. And this is an example relevant to one of the LTER nodes up on the north part of the state of Massachusetts. Over time, historically, they were never occurred above Cape Cod and what we see in the range expansion in the northern areas. And so this particular study looked at three areas in the Plum Island Sound, LTER. And what they were finding is that in exclusion experiments in these marshes with the crabs, you had a reduction in the above ground biomass of Spartina as opposed to out the excluded areas. And the finding here, the speculation was that by adding the crabs, they actually are undermining the root systems of Spartina. And so clearly there's implications of these things. So I wanted to shift to a number of what I would call findings or recommendations for addressing these and other problems that are contingent in terms of the implications of this. And first of all, although the phenotypic distribution, a phenomenon of distribution shifts to global warming has been studied extensively, there's proportionally very little on the dynamical system responses, including nutrient cycling, metabolic rates, etc. And so I think really do need a program of ecosystem based research on systems undergoing rapid change and a number of the examples are certainly those that are associated with rapid change. Secondly, no any other mission oriented agencies do heavily rely on their ocean going time series and much of what we know about what's going on in the continental shelves are these long term trawl surveys and other surveys that that the mission agencies have, but there's a real gap in particular in coastal regions. So when you think about it, there are a lot of coastal infrastructure that already exists. For example, the NAML laboratories in the air sites, the LTDRs, the neon sites. So actually combining them in some kind of phenological network would actually make a lot of sense. A good point I want to make is that these kinds of large scale ecological programs are not foreign to NSF. In fact, the USA GLOBEC program, which occurred from the 1990s to the 2010s was exactly emphasizing some of these problems of both the top down and bottom up effects on things. So coming up with a similar type of program that might be a joint academic and government initiatives to address climate change impacts seems like something that NSF has already done and the agencies are familiar with. And just to make the point, these are the distributions of some of these nodes already. So combining them in some way to establish in this case a coastal phenology network would be appropriate. So, and of course we have statewide surveys that are, you know, haven't necessarily been used for these kinds of things. These are trawl surveys in the northern Gulf of Mexico. It's really interesting if you look at the Reyes and Detra of US GLOBEC, its goal is to identify how a changing global climate will affect the abundance and dynamics of marine animal populations. That's a really good goal and it sounds like exactly what you all are talking about today. And so this was the GLOBEC timeline and you can see that, you know, field programs were implemented in three places of very rapid climate change as it turns out. The California Currant System, the Northeast United States, the Alaskan Field Program, and then also in the Southern Ocean. And it's interesting because many people have said that GLOBEC was the right program, but it was just a little bit too early. So in terms of a couple more recommendations, number one, much of what we know about climate change impacts comes from the global north. This is a point that both Emmett and Malin made. But the impacts may be in fact as severe or more severe in the global south. Interestingly, the country of Norway has tried to address this by stationing some of its infrastructure in Africa to get the 30 plus countries there to collaborate on this. And so we might want to consider something that might be funded by a number of US agencies to do a similar kinds of things. So we're trying things, for example, in South America or Oceana to build capacity and to address this fundamental research gap that we've got. And just to emphasize that point, here are fishery landings as they're traded around the world. So with that, thank you very much. That was great, Steve. Thank you so much. And then rounding out our panelists presentations today is Sarah Davies from Boston University. Sarah, floor is yours. Okay. Hi, everyone. Thanks so much for this invite. I'm really, I guess sort of excited, also nervous to tell my thoughts about coral reef changes over the next 100 years. So I'm an assistant professor here at BU and in our lab we're really fascinated by coral reefs as are many people. So these are foundational organisms that are building blocks of entire ecosystems. And when you think about where coral reefs live, they live in these kind of oligotrophic waters where there's not a lot of nutrients. So they really rely on this symbiotic relationship. So I've zoomed in here on a single coral polyp. And this is an Idaean host. So it's an animal host. And inside each of those brown dots is an algal cell. And these are algae in the family, symbiotic DNA. And the symbiotic relationship is the algae photosynthesized. They provide carbon sugars to the coral host and the host provides them a home. But there's also a complex microbiome that associates with both the host and the algae. And together we call this the coral holobion. So I'm really focusing on this coral holobion. So taking a single ecosystem approach from what Steven was talking about. And it's really these complex interactions between these whole about members that service the building blocks for these entire reef ecosystems. And these ecosystems are kind of like tropical rainforests. So they provide tons of services ranging from human well-being to jobs in the economy, coastline protection. So coral reefs are the primary defense against tropical storms, for example, and marine biodiversity. So coral reefs are really the foundation of an entire food web. So they provide nurseries for tiny fishes. They obviously here we have a fisherman fishing for tuna. So larval tuna, for example, are part of the coral reef system. And scuba divers right providing jobs and economics to local communities. And this marine biodiversity when it's lost, it has direct human impacts. So when we think about how much coral reefs or how important coral reefs are. So we've got their global economic value being about 10 trillion US dollars per year. And this is because of social cultural benefits. So 13% of the human population lives within 100 kilometers of reefs. And remember that coral reefs only live 25 North to 25 South ish, although some are expanding. 94% of small island nation populations live within 100 kilometers. So really, when we think about a healthy coral reef system, it really leads to healthy people and healthy populations. Unfortunately, the future socioeconomic viability of coral reefs is uncertain. And this is because of a variety of stressors. So there's coral bleaching. So this is the healthy coral here that I showed you at the beginning. It's beautiful, lovely. And you might argue that this picture on the left is also lovely, but it's actually really devastating. This is a bleached coral. So what you're seeing is the Translucent Nidarian host with its calcium carbonate skeleton showing and it has no algae. So it's done a process called coral bleaching. And if this state persists, it can lead to mass coral mortality. And there's also rampant coral diseases. And there's a variety of reasons leading to increased frequency of bleaching and disease. So when we think about threats to coral reefs, there's a variety of stressors. So they range from this kind of direct exploitation where you have organisms being removed from the reef for food. So top predators like sharks, herbivores being removed from reefs and change in the entire trophic structure of that ecosystem. You also have invasive species. So things like the lionfish invading and eating tons of small fish that locally that wasn't a predator from before pollution. So you can have things like microplastics and eutrophication from fertilizers coming in to the water and increasing eutrophication of the system changes in land use. So removal of mangroves for things like shrimp farms or hotels. But really the primary challenge facing coral reefs is the increase in greenhouse gas emissions. So this increase in CO2 and the atmosphere is warming it up and the ocean really does the lion share of taking in those temperatures. So as the oceans warm, it really creates a fundamental threat to coral reefs that increases the frequency of bleaching and also increases the frequency of disease. And together this means that coral reef cover is predicted to decline by up to 99% if global warming reaches two degrees above pre-industrial levels. And this was based on the IPCC report in 2018. I want to bring us here to this quote from Nolten et al in 2021. The coming year and decade likely offer the last chance for international, regional, national and local entities to change the trajectory of coral reefs from heading towards worldwide collapse to heading towards slow but steady recovery. So I think this is really kind of optimistic, right? We're thinking about the future, thinking about saving reefs. However, that was in 2021. And unfortunately, what the last two years has shown us is that climate change is happening at a much more dramatic scale for coral reefs than we were hoping. So this is some dramatic imagery from that was taken in 2023 and 2022. This is a healthy alchorn coral. So you can see by that brown color, the coral is healthy. This coral is highly endangered. This was taken in August 2022. And this is the coral bleaching in the summer bleaching event this last summer. So this last summer was devastating for the Caribbean heatwaves lasting months. So really quite a wild summer we have last year. And so, but really we remain very not great at predicting coral responses to climate change. And the question is, why is it so challenging? I hope to argue today that there's three main reasons. The first is that there's unrecognized coral genetic diversity. There's a normal, enormous algal genetic diversity. And also there's these complex interactions amongst whole about members that are really challenging to understand. So I'm going to take you to a story about why some corals win and others lose. So it's thinking about this unrecognized coral diversity. So I'm going to take you to Palau. This archipelago. It's beautiful. We sampled at six sites and we collected this coral perides lobata. So these sites we think of these offshore sites as being more classic sites and these intro sites as being more extreme. And we went in and sequenced the genomes of these corals. We found strong evidence for three lineages within a species. So this is looking at the sites along here in the different colors. And these bars are the ancestry of individuals. So you can see that each individual mostly belongs to one color. And we see strong evidence for three kind of like blocks of individuals. And when we took those individuals and put them under heat stress experiment, we found variation among these lineages in how they respond to thermal challenge. So we see here the purple lineage, for example, even under really extreme heat had 100% survival. So there's really functional variation within a species that we might be missing. This is maybe a glass half full story. So when we went into the literature, this was paper just published this week, looking at how prevalent these cryptic lineages are among corals. We found evidence in 24 genera across every ocean basin. So the recommendation here is thinking about broad strategies to increase international sampling efforts across species ranges and I'm using quotes here that include leadership by local and indigenous communities. So thinking about kind of like multinational collaborative opportunities. Now I'm going to focus on the algal genetic diversity. So the genetic diversity of the algae is vast, although it's quite messy and quite contentious in the community. So this led us to do an NSF workshop. So myself, Adrian Carrera and John Parkinson run a workshop to try to build consensus about how we assess and interpret symbiotic diversity. So diversity of that algal symbiont. And we came up with the idea that there's these low complexity communities that are maybe a little easier to understand where you see that they host in their cells, the same color of symbiont. But there's also these high complexity communities that are maybe more challenging to understand because your every sample of coral is a community of organisms present. And so I'm just going to riff off of the recommendations that all of these authors came up with. So just because they're already there, we wrote a white paper if you're interested. But the first one is that we really need to leverage technological advancements for assessing diversity in symbiotic ECA. And again, these international collaborations to better link diversity with function are really needed. And then we need to expand culture collections and taxonomy. Now I want to talk about the whole about interactions and incorporate some of that microbiome. So a lot of the work that's been done is on hosts or symbionts or microbiome, but there's very little kind of incorporating all three members. So we need more work thinking about how Nigerian phenotypes will differ depending on what type of microbiome you have, what type of symbiont community, what genotype of coral you are. And together, all of these interactions amongst whole about members, this idea that you can mix and match hosts and symbionts and microbiome combinations together might yield novel phenotypes that might do well under climate change. So the recommendation here is to think about focusing research on mechanisms underlying how these interactions are shaped by changing oceans so that we might come up with ideas about how we might use these different combinations to try to be more adaptive strategies. So when we're thinking about coral reef solutions, so how can we keep reefs looking like this what we see on the left? Obviously, the main thing we need to do is slow climate change and improve local conditions. So efforts to mitigate warming and improve local conditions are absolutely paramount. However, these are major policy challenges, corals and currents that take coral larvae and CO2 emissions. They don't respect political boundaries. So I think there's a huge need in the community for incentives and mechanisms for multi-national collaborations and I know that's been touched on a few times today. And I think this will lead to broader engagement with multi-national stakeholders that will bolster both science and conservation. So now I'm going to talk about specifically actively restoring coral populations, which is becoming a major focus of coral reef research. So when we're thinking about restoring corals, so this is there's a whole coral restoration consortium. I spoke before this talk kind of thinking about what some of the recommendations might be. And these are the ones we came up with. So thinking about national repositories or biobanks to preserve host and cement genotypes. So for example, for the endangered acropored corals and the Florida Keys, a lot of those had already been biobanked before the devastation of summer 2023. So there's still those genotypes in nurseries. So things like this where you have an entire population wiped out from a bleaching event. That genetic material and that individual genotype is still available and efforts to create novel genetic combinations. So the idea that maybe you mix and match individuals from the south to the north and try to think about hybrid vigor to try to increase resiliency of different genotypes and research on the potential risks associated with novel interventions. So whether those interventions are across countries or whether those interventions are quite provocative like moving Pacific corals into the Caribbean. We need research to understand those risks. And we also lastly need investment in host and cement taxonomy policies are based on species and in coral and some of us were really bad at knowing what species are. And so last I want to end here looking back at this reef. The next decade of research will be critical. We must reduce CO2 this is absolutely critical and local stressors will also prioritizing understanding diversity and how this diversity can be leveraged for restoration and adaptive management. And I want to leave you with my final thought which is kind of the theme is we cannot conserve and restore what is out there until we actually know what is out there. So thank you. Wonderful sir. Thank you. That was a great way to wrap up this panel and thank you again for all three of you for great presentations and great timely presentations of that. We have a few questions here on Slido. And so I'm going to go ahead and I guess take it off by asking a question to all of you as well as the committee and attendees. So mainly made a point earlier about, you know, how we're seeing these species sort of changes in species distribution with a changing climate and we can kind of infer or even correlate the two. And maintenance of the details, if you will, in the middle are okay. So, I'll post this question, you know, our models at the point where we can predict either the outcome. In other words, where will species they end up after, you know, time period X right without necessarily knowing what happens in the middle, or our models at the point where we can really better understand some of the mechanisms. So I'll open it up to the three panelists to answer that. Anyone want to kick it off? Well, I'll try. I think the, the, the field of modeling species distributions has rapidly expanded right. And so there's a lot of covariates that people are including in these models, including interestingly the density of the populations themselves. I mean, there's a number of hypotheses called basin hypotheses that if populations get to a low level, they shrink their geographic distributions into the core. And so this represents an interesting problem right because you've got changing ecological conditions and changing population sizes right. So, you know, understanding how distribution shifts relate to both of those factors is important. I do think there's been a rapid progress in this where the progress hasn't been so great is actually understanding the implications of species interactions as opposed to individual species models. I can follow up on that. I tried to make this point very, very briefly, though, just that, you know, the, our simple approaches right now just relate climate to either species occurrence or biomass or abundance. And yet we know there's a lot of biology in there and the climate is actually affecting biological rates and demographic rates. And there are those process based species distribution models out there that exist. But we haven't yet brought them to bear on the incredible range of observations that exist and tied them carefully to and constrain them with experiments and the other including genomics on rapid adaptation. So I think it's about using models to link between observations and the processes we know and we haven't yet taken advantage of the opportunities that are there. There's some very simplistic approaches that are easy to apply and have been applied really widely, but they skip the biology that that matters and that includes the species interactions that Steve was just talking about. And the kinds of basin basins of attraction and dynamics like that that he was talking about as well. So, we have the models but we're not using them. Excellent. Thank you. Sarah, did you want to add anything? I just wanted to say that, like, I think also it's hard when you're thinking about species occurrence because my argument would be that like, we often don't know what species we're working with. So for example, in Japan, there's been a range expansion of acropored corals along the Japan coast but when we went in and sequenced those they're actually they're different lineages. So the individuals that are at the northern range front are different lineage than the ones that are present in the core population so it's like they look the same but they're different genetic material. So it's hard to know who's actually moving because there's no they they're like, they look the same. And, you know, my colleague she asked a question about do we have enough empirical data. You addressed that to some extent she I don't know if you wanted to add anything to that question. Yeah, I think we all might have started addressing it already but, you know, having there's there's still no empirical data on actual experiments and, you know, targeting specific questions and can we use technological advances and AI or any knowledge learning for those that you just mentioned. Well, I think that there's like AI I think can be really useful in thinking about how more there might be more logical variation that human eyes like missing that might be able to identify without sequencing so the whole idea that we need to go in and sequence everyone is obviously not a feasible idea, but developing kind of tools and techniques leveraging AI and like neural networks and things to try to. Maybe we're missing something like we think it looks the same but it's not so I think there's a lot of potential avenue for research thinking about more logical variation that we're missing to try to identify who species are and this goes with others colleagues doing this in mosquitoes for example. So it's like it's not just corals like people are trying to think about more logical variation among different lineages within kind of like a broader taxonomic group that are cryptic so to speak. So I think there's a lot of potential there. I also see opportunities around making better use of the of the images and the video that's out there using AI and machine learning to turn those into species observations. Obviously that takes quite a bit of work and careful curation and understanding what species are out there. And I also saw in the chat, you know, acoustic information and other sources of more novel information streams can be useful there too. Moving on to some other questions here Jason may ask a question. Jason you want to try me on this shirt. I missed spoke on the question. Originally asked what's the role of a couple of ecological biology model to explore all these topics. I think what I really want to ask is, what is the state of the set of models ecosystem models that they're coming to business. It isn't sufficient to be able to address not only this esteem of these distributions, but all with other impacts and see what's going to summarize and stop. What's the state of the model. And was there opportunity. You know, honestly, I think Jason is probably the one in the room most adequate to answer that question as well as asking right. You know, I would say, you know, if you look at the state of ecological models that are in use around the world, you know, the Atlantis system is in terms of in particular of larger animals is sort of the state of the art and has a number of modules, but mostly, you know, it's it's it's it's key to species interactions, more or less assuming a constant ecosystem or an ecosystem not under the rapid change. So, so having a spatially explicit component that can map these species distribution but also take into account the distributions that's probably right at the bleeding edge of modeling that that would be really useful for this right now. Yeah, and I that add on that. You know, it's, I think one of the ongoing challenges is to. And this is this is also met in climate models it's this, this balance between model complexity, you can put as much detail into an ecological model as you want. But understanding what details important and whether which detail actually is actually helpful for understanding both historical and future changes is what we haven't done. So it's really this merge of modeling and data and study on the underlying biological process that I think is so important we can't have just models. And that's part of why I was trying to emphasize this are these data science skills and the literacy to merge these these pieces together. You know, as you know, Jason and others who've worked with them like Atlantic model Atlantis models are incredibly complex but it can be hard to diagnose what's actually going on. And that's part of it's a computational channel as well as something that needs more focus. Very good. Very well. We have a couple of three more questions that at least start with that Josie. Josie want to ask your question. Sure. So, inside from the most question I was just curious about the basis species and how it can be thinking bio diversity, and what we should be thinking about the future. Yeah, I mean, one of the things I think is really important from a research perspective is is trying to generalize the impacts of novel species interactions and that's a theme that cross cuts both invasive species but also the shifting species interactions that Steve talked about. But I brought up a little bit, but I think, you know, it's often a case where we see that some species invasive species have a really big impact. Others are not invasive and have very little impact. And how do we understanding the generalizable rules for ecological interactions, building from traits and from taxonomy and from environment is really fascinating and important research question that has both practical and both applied and basic science implications. So, Josie will appreciate this answer. You know, the federal crab example is an interesting one. Is that an invasive species into the Gulf of Maine. Well, I guess it is right because it really wasn't found at any other place. Well, it's it's taking advantage of a habitat constraint that's no longer there right in terms of minimum, you know, maximum minimum temperatures right. And so, and it's going to create ecological change. In general, when we think about this, you know, highly packed niches, and you know, species moving, you know, out of there, those those relationships, it leaves open niches right that can be invaded and so I think, you know, understanding, you know, how a species would fit into a system and potentially exploit it is a really important aspect of this and so, you know, in in the Gulf of Mexico, for example, invasive lionfish have been, you know, you would have thought that that was a well packed set of niches but certainly those that species was able to take advantage of something that wasn't there which is, you know, predator of small animals right so I do think invasives. I think we need another definition for what an invasive species is if it's on a continuum, you know, of distribution, you know constrained by by by climate right so it's not like you know something coming out of the Indo Pacific right. I think Steve's right that there's like, if we've learned anything in biology there's nothing that's binary, right so it's like, you know, for example we think of coral reefs is being really great but in Japan where they've like expanded their range they're overtaking kelp community so they've changed the entire ecosystem, right that used to be dominated by kelp, and the fishes associated with kelp. And, you know, that's like, and from a coral perspective we're like yay crawls are expanding their range right but those kelp researchers like oh my gosh crawls are taking over our ecosystem. So I think invasive is also just like depends on your context to, and then thinking about the lionfish invasion into the Caribbean I think there has been some like really interesting socioeconomic impacts of lionfish so it's become like an entire fishery. You know it's created a huge economy, you know I was in Curacao and there's entire businesses based off of like lionfish jewelry and lionfish tacos and lionfish ceviche. So it's, and it has all this so it's like really people associate you know this fishing pressure is being like really positive and they feel like they're, you know, being positively impacting the environment by eating this lionfish so I think it's got this really interesting kind of socioeconomic interaction that's fun. And they're really crazy. Let's go to Leila and then Dipajana and then we'll move on to the next panel. Leila you want to close the question? Yeah, thank you very much with the same line of thought that we're in right now. I'm wondering if the models and species of study or the studies of species change and expansion are considering the built environment. Fixed structures, vessels, also ballast, you know, those are questions earlier, the things that microbes and coral larvae and other small things have a tendency to stick to and then fling off of and maybe expand their range that way. You know Leila you ask a really cogent question and I think it's from where you actually sit. So if we look for example at the built infrastructure in the Gulf of Mexico, you've got a massive number of oil platforms out there and they've actually fundamentally changed the dynamics of a number of the species and they've become stepping stones in a sense, you know, for a range expansion and moving to the next stepping stone. They've changed the nature of pelagic species that they tend to congregate around these things as opposed to, you know, being more, you know, sort of peripatetic and so, so yeah, I mean, I don't think models have necessarily caught up with a lot of this because there's a big observational gap. But nevertheless, I think, you know, humans are embedded in this environment at this point and, you know, if we want to understand how these ecological systems are going to change. I think it gets back to the point that Malin made about these are natural and economic systems at the same time, you know, they're coupled systems and so embedding the human element in the models is going to be critically important. This is, and as I'm sure folks are aware, we're at a inflection point in many ways, at least for our continental shelf ecosystems in terms of some massive plans for more built structures and offshore development, partly to address climate change but understanding the biological impacts of that is still a really important open research question. Excellent. All right, well then, we'll wrap it up with Dipanjana's question. You're on and you could ask about the factors influencing those thoughts. Yes, Dipanjana here. Gratitude for again, excellent insights and presentations. I actually the non human intelligence and resilience are also very fascinating subject. So considering the factors which you have presented in your presentation. I would like to ask which factors are influencing this responses most and can better management or conservation of such response factors those muscles, which are providing the resilience can decelerate the rate of biodiversity loss. I'm gonna try that melon. Yeah, just just started thinking through. I mean, I think on one level, you know, part of what's so important is response diversity and functional diversity. And that get back gets back to some of the cryptic diversity that Sarah was bringing up. I think some of the functional diversity that Joey brought up as well, but that diversity across species provides insurance against environmental change so we can ecosystems can continue to maintain function. Despite environmental change, I think that's incredibly important during this, during this moment in our history and your history. I don't think I think there's also the interaction between stressors to that's really important and challenging to understand. So this idea that there's like warming oceans that are interacting with like, you know, more direct influences like human, like being close to like an urban system. For example, and all the stressors that are associated with urbanization. And so I think there's like really big challenges to understanding. I'm obsessed with cryptic. So I'm going to make another cryptic thing. But, for example, when we were looking along an urban gradient in curibus in the Pacific, we found that there were three cryptic lineages. And one of the lineages was more commonly found in the anthropogenically impacted areas. And then when there was a massive heat wave in 2018 where if you didn't know the genetic data, you would have said, Oh, those corals that were living closer to urban environments all died. So they were like less resilient because of urbanization, but actually they were kind of like a specialist lineage that was like more commonly found. And when we actually genotype them, they were all the same lineage. So you saw the extinction of a single lineage in a local area. But if you interpreted that without knowing the underlying diversity of those interacting stressors you would have said, Oh, it's, it's, it's the urbanization climate change interaction. I think there is just so nuanced to understand. Yeah, so I think there's just like a lot more to learn. I'll just follow up on it and say, you know, it's interesting, you know, in trying to sort out multiple simultaneous drivers. There's only an end of one of the earth, right? And so, you know, we need a sort of a multi-passed approach to this, which includes modeling, you know, experiments on land, experiments in water, etc. For us to sort of deconvolve all of this. And, you know, I get back to, you know, what was the driver for global one was looking at top down and bottom up processes at the same time. Right. And so, so I actually think, you know, this is becoming even more important now as we we get into sort of uncharted territory with climate change. Thank you all very much. There was fantastic presentation. Great question. So I'm going to turn it over to a judge. You can try to lead the next panel. Yeah, thanks, Peter. So, I may take my most not greatest choice here and give us two minute break. But I'm going to ask, I really want the panelists to stay on. We're going to switch topics right now. We've had this great discussion on science. And now, we're going to look at and come back from the great sort of tools and technologies. And I'm calling this part one, because we're only going to be looking at DNA today and we specifically talked about this a little bit last chat going on about other technologies. And we talked about whether we could address all of the day we decided to just focus on DNA. But stay tuned because there will be a cartoon that we will pick up. But let's just take a really quick to the break and come back and panelists from the morning, please stay out because we have a full discussion. Can you hear us. Can you try talking. I can, I can hear people now. I couldn't. Okay, so we're, we're ready to go. Okay, welcome back everyone and thank you so we're now going to talk about DNA and to talk about it for this session. We've got Paul and Zach and they're going to do a joint presentation. So I think it's all you're going to go first. And so he's doing your slides. Yep. Okay, so thank you for the invitation for Zach and I to speak with you today and thank you to the other speakers for for setting us up so well in your talks. Ocean ecosystems are facing unprecedented challenges. Climate change urbanization, eutrophication over harvesting of resources and critical questions that we're facing in this coming decade are things like who are the winners and losers of climate change. How are these changes impacting ecosystem function. How can we support ecosystem resilience to anthropogenic stressors. And there's a tremendous amount of ocean observation data from drifters and gliders and satellites and all manner of things. But these observation systems can't tell us about C star wasting disease or, you know, black band disease and corals that can't tell you that Garibaldi fish, you know, now live in Monterey Bay it can't tell you about non analog communities were temperate and tropical fishes are now mixing in the oceans where they never did before. And if we think about biodiversity observations in ocean ecosystems. This is people and, you know, a lot of these traditional methods are people intensive putting divers in the water, having people go through the the catch of a troll. These methods are time and labor intensive. They're expensive. They require taxonomic expertise. You know, because of this, it focuses just on conspicuous species. And anytime you're putting people in the water, you know, you're limited to the conditions of the ocean at the time. So, if we think about the future of ocean biodiversity observation. You know, ideally, observation systems would capture entire communities. It would happen at a global scale. It would happen in near real time. And it would happen. Importantly, autonomously. So, we all know about barcodes. We've gone to the grocery store picked up a jar of peanut butter. We scan it. It comes up with a price. And DNA barcoding uses the same approach, but using species diagnostic DNA sequences to identify a particular taxa. DNA barcoding has morphed into what is known as DNA meta barcoding where using next generation sequencers, we are getting thousands, tens of thousands, millions of sequences. And this makes it possible to identify entire communities through these diagnostic DNA sequences. And, you know, we can identify who's there and, you know, create a profile of what these communities look like. And compare differences between communities, either the same community at different time points, different communities in different locations and things like that. Now, there's some common approaches to meta barcoding. And what really differs is the input into the system. So, the first thing that really counts as meta barcoding that we didn't call meta barcoding at the time is microbial community profiling. You take a sample, you amplify 16S, you put it into a chime and it spits out a distribution of microbial taxa in your community. A method that has gained a lot of prominence over the last decade is environmental DNA or EDNA. EDNA is based on the fact that all organisms are leaving behind small biological traces of themselves in the environment. And that if you go out and you collect water or sediment from that environment, either using scuba divers or a niskin bottle, you can filter, isolate, amplify and meta barcode that DNA and reconstruct entire communities. Another meta barcoding approach that is gaining prominence is autonomous reef monitoring structures. These are essentially stacks of settlement plates that get encrusted with benthic marine organisms. You take them apart, they make very pretty photos. And then you can scrape all of that biodiversity into a blender, extract the DNA and put it into a meta barcoding pipeline that allows you to reconstruct the communities that were present. So some of the points that other speakers have made that we want to emphasize is that it's really critical to think holistically about biodiversity monitoring and move beyond this very narrow focus that we have on megafauna commercial species, foundational species. So, for example, in the Channel Islands National Marine Park, they monitor 56 priority species at 37 sites once a year. And that is very coarse observation of the biodiversity in these systems. Essentially, they're flying blind. They really don't understand what is going on in these systems on a finer time scale. And if we can pair results from EDNA to the visual surveys from the National Park Service, we capture a lot of the same things, but EDNA gets a lot more. And importantly, 25% of Zach's EDNA samples had a giant black sea bass in it. In 30 plus years, National Park Service divers have seen one ever. So EDNA is capturing diversity that other methods is missing. And so if you look at the sensitivity of EDNA compared to other methods such as baited remote underwater videos compared to beach sains, again, EDNA is outperforming these other traditional survey methods. So the sensitivity of EDNA is higher as well. So in this study, Zach compared fish biodiversity inside and outside of the marine protected area. And what you can see on the right hand side is that EDNA does a better job of differentiating those two communities, allowing us to see the impact of the marine protected area. Now, we talk about meta bar coding and different approaches to meta bar coding, specifically because they capture very, very different communities. So this is data from Indonesia comparing arms and EDNA from the exact same sites. And what you see is a shocking little overlap between the tremendous amount of biodiversity that each of these meta bar coding methods captured. And one of the really cool things about these meta bar coding approaches is because you're looking at entire communities, you can do a comparison of sites across a marine pollution gradient, which you can see very clearly in those images, non photoshopped images on the right of the arms. And if you plot the bio or if you plot the abundance of specific taxa versus the pollution, what you see is that there's some taxa that increase in diversity. There are other taxa that have significant decreases in diversity. And so from a monitoring perspective, if we want to really know what the canary in the coal mine is, you know, these methods allow us to focus or identify and focus on the taxa that are most sensitive to change so that we might be able to respond to environmental changes proactively, rather than, you know, once all the corals have died. So, it's important to think holistically about biodiversity monitoring, because right now, what we're monitoring may not be the best indicator species of environmental stress. We need to be moving towards whole communities, vertebrates invertebrates microbes. And this is going to allow us to focus monitoring efforts on the things that are most sensitive. The other point that several speakers have made is that we need to be thinking about thinking globally about biodiversity monitoring and move beyond US borders territories and our exclusive economic zones. So, many years ago, Peter sale wrote a paper about connectivity being a critical gap in the development of effective marine protected areas. And this is because if you want to protect to marine ecosystems here, those ecosystems are dependent upon larval input from upstream populations. And so, if we're not protecting these upstream populations. The MPAs may not be stable in the long term. And this is important because, you know, larvae don't respect national borders. The other thing we need to be thinking about is that most of global biodiversity is not around North and South America. It's centered in Southeast Asia. And most, you know, 40% of the world's economy and 80% of the needs of the poor are derived from biological resources. This is very true in Southeast Asia, where, you know, it's home to nearly 400 million people, about a third of which have fish as a primary daily protein source. It's a big part of the GDP and supporting the livelihoods of, you know, 120 million people. And it's incredibly threatened by anthropogenic stressors. The red and yellow here highlighting reefs in at risk of local collapse. And it's important to note, as Joey did, biodiversity is food security, food security is political stability, and political stability is US national security. And so, us focusing only on US waters, US territories, US interests, we do that to our own peril. Thanks Paul. Hey everyone, Zach Gold. I just want to first thank everyone because I'm a product of NSF and NOAA. That's the only way I got here today is because of all of the undergraduate, you know, graduate funding that got me to this position today. And I'm really honored to be here. And I'm going to highlight here, you know, talking about the additional needs that we have. And this is about developing those resources that we need to be able to pull off doing EDNA, you know, at scale. And so, as this is a surprise to no one, there's a lot of studies that have been happening in the US and the global north and not a lot of applications of EDNA and meta-marketing, you know, in the most biodiverse hotspots. But the few that have been done have been really interesting in that when you compare how many species you get in the same leader of water, we get the same number in Southern California as we do in Rajampot. Which is insane because there's about two orders of magnitude more species of fish in Rajampot in the coral triangle than there are in Southern California. And so our techniques and technologies and our sample methodologies that we might be using in our biodiversity, poor, you know, comparative regions might not immediately translate to, you know, how we do these same kind of surveys, how we're going to evaluate these hyper biodiverse systems. Right. And so what that means, though, is you can get away with taking 12 liters of water at a site in Southern California, and you did a pretty good job of saturating biodiversity. You know, estimates are we need 300 liters ballpark to get the same saturation on a coral reef, right. And now, all of a sudden, that's not cheaper, better, faster. That's really difficult to do. And so we're going to have to think about how do we design these tools to be better to work in places where there are far more species. And one of those tools as highlighted and you saw in some of the slides that Paul is presenting, those are phylum level or family or order level tax and other assignments and part of the problem is our reference databases for these species and has been hyped up by previous talks today. We need those taxonomous and we have fewer taxonomous now and maybe we've ever had in the last 150 years. And here's just an example of what happens when you put in the legwork to have some of that taxonomy. You know, all of a sudden you're turning light switch on we can actually put names to faces for these species. I spent in the second chapter of my PhD, specifically doing that on frankly one of the most well studied group of organisms in the world which is fishes in the California current. But even then, we had so few sequences and reference barcodes that one by doing that effort we're able to find, you know, almost double the number of species by actually putting in the legwork, doing the museum tissue specimen collections identifying those species down the painstaking frankly boring science that is the fundamentals right and you can't win championships and can't make free throws right so we have to be able to do those fundamentals. And here's just a further example of you know when you further curate those and think really intentionally about how do we design these reference databases, and this was just one really small subset in Southern California, we can improve our ability to assign taxonomy pretty dramatically. And we were actually by just having this regional curated database that took a lot of effort. It took a lot of effort to just get a species list of who are all the fishes that live in Southern California involved interviewing 30 different you know sort of fish, biologists and pulling out their tons and tons of these lists, but we improved and got species level resolution here. And you need this data, because, ultimately, we need to have defensible taxonomy practices for when we do, you know, take a DNA do DNA meta bargaining we're putting names to basis as DNA sequences. The only way that we can do that and even test which parameters matter and what does that shape look like is by having those databases and this is just one example we did. And we found that there's actually this direct trade off between confidence and resolution. This is true in all fields of science, not just, you know, bioinformatics. But the reason why this matters is if you have $100 million detection of this invasive species showing up in the Great Lakes, you know, on the other side of this dam. You better, you know, you know this is going to the Supreme Court like you better know that that is the right identification. The whole is just to what is the optimal, you know, number of species that we can identify doing this in Indonesia we don't actually, there's not a lot on the line this is more of a research question, you will choose a completely different parameter in this instance right and so having that information developing those best practices is really fundamental, but we can't do that without having these reference databases and understanding that sequence diversity for these key markers. In addition to investing in these reference databases, we have to invest in people. And we can't just let EDNA be the science, you know, of the most well funded institutions. It has to be democratized and provided. And you know, these are the future scientists that we have to have and we need more and more of these skill sets. To my knowledge, I was one of the first three students that had graduated PhD doing environmental DNA in the US. And that was, I graduated in 2020 that was not that long ago like we need workforce development if this is going to be the future of how we're scaling our monitoring. And really to scale the monitoring we need to overcome some key technical challenges. And so, you know, it's awesome. I love spending my PhD in the Channel Islands diving and taking water samples with some of my friends pictured here, but that's not going to cut it if we need to have the kind of observations that, you know, Paul highlighted before and that all the previous speakers were talking about right, we need to have automated sampling, we need to develop it and there are some freaking cool technologies coming at the end of the day, but I'm a Bari and Woods Hole, but what about the rest of us, you know, right, like, we can't all be affording Lamborghinis, like we need to be able to buy our smart cars, right and how do we get those off the streets so that it can be democratized everybody can use these technologies. So we can actually scale this, you know, across ocean basins. And another element and this is something I've spent the last few years really working on is, you know, at the moment metabarking is really good at telling us who was present, but not how many and quantifying abundance right and we know that abundance matters, you would not manage these two marine ecosystems the same way, but they have all the same species, right and so we don't have abundance we only have presence, we're going to be really limited understanding how marine ecosystems are changing, you know, in response to climate change other anthropogenic stressors. And so part of it is developing mechanistic frameworks this is work led by Oli Sheldon, and North Pacific Science Center and Ryan Kelly at UW, we're just some simple math, this is the math the first equation is pulling marbles out of a jar. The second is the PCR equation that we've known about since at least the 80s. And it turns out that just correcting for amplification efficiency bias, doing some standard validation improves our ability to derive quantitative estimates pretty dramatically. And you might even be able to notice that on the zero there. There's a bunch of points that are not explained by the model and that was why we added the, you know, pulling marbles out of a hat equation right and this is just really simple work that's come out in the last year and a half but there's so much more. Frankly, people that are much better at math and statistics to be able to keep pushing and making better tools and models where we can get better abundance data from these. And this isn't even, you know, this is really going from, you know, the DNA tube all the way to through sequencing and more work needs to be done on understanding, you know, from a water sample from the environment how that actually relates to biomass but I think we can get there. There's been a lot of really promising results in just the last year and a half that show that we are approaching these, the ability to get there. And the other really exciting work and I think is really critical and obviously something NSF has been funding for a long time, but is thinking about long term observations because if we're really going to ask how are marine ecosystems changing over time. We didn't know where we came from. So, one, some work that I did in an NSF, you know, grip internship was working with Noah taking these ethanol jars. They're actually older than all of my siblings. And it turns out you can just pet the ethanol at the top of here and reconstruct which organisms were living, or were captured in these samples, right and so you can sample process sequence and reconstruct what's there. We've found thousands of species of zooplankton we've never had in the California current, a long term zooplankton spatially explicit time series right that's a fundamental piece of ecosystems that we've just never been able to have. And if we really want to know how marine ecosystems are changing we need to know how the food sources are changing right. And so, this is just some pilot data that we're really excited about but we were able to look see that temperature is obviously a main driver which is surprised to know that this is only 100 samples right there's 7,000 of those samples. This is one LTR, you know, so there's project associated with an LTR. There are museum samples collections of these jars sitting everywhere around the United States around the world. These can be leveraged to start answering these questions of, well, what did our marine ecosystems look like we've never had enough time in the universe to count copepods and identify the species now we do, you know, using EDNA approaches. And I just want to highlight this paper that came out in the last week that I think is something that certainly we are thinking about really strongly and Noah, moving towards which was, it took the, you know, the Australian Government CSIRO they're sort of equivalent to know spent a lot of time integrating the last decade of all EDNA samples that were taken off of the East coast of Australia, and they were able to identify, you know, in the red and the blue these sort of heat wave associated communities. And the only way that those possible was investing, you know, in the data integration, investing in the people that be able to smartly put these together and combining the physical oceanography along with, you know, the biologists to answer these questions. There's no reason, you know, my ballpark estimate is that in the US over the last 23 years since Craig mentioned the first EDNA sample, there are over 100,000 EDNA samples. There is no repository. There are three, you know, on OBS and GBIF that have ever been submitted from Noah. But we could be doing this kind of data and so it's how do we mobilize data sets that already exist. They're in supplemental data files. They're the raw data sitting on NCBI somewhere. You know, but how do we actually integrate these at scale to answer these big questions and, you know, basically provide this data to the people like the folks on the previous call who can analyze it and move that to actionable science. And so why is all this possible? Well, when I was 10 years old, I made sequence the first human genome. It cost about $5 billion in today's dollars. Now it's two bucks. There's nothing that has declined costs, you know, with inflation this fast. And so that makes all this work possible. It's also incredibly complementary to lots of tools we heard today and moving in the future, whether that's AI and machine learning to integrate, you know, what do you do with a data set with 30,000 species, you know, and 20 environmental variables? How do you find those features that are important and not right? Autonomy, putting, you know, and Barry just had a couple of papers come out recently and putting these ENA samples on sail drones on long range of UVs. You know, how do we deploy these at scale so we can actually have these ocean basin scale, you know, visualizations combining this with imagery and acoustics. You know, the thing that ENA is really good at is putting names to faces. What ENA is not great at is counting, but imagery and acoustics are very good at counting, right. And so it's how do you build those joint models, those joint statistical approaches to leverage the strength and weaknesses of both of them. And lastly, it's really scalable. The one day in Seattle, we process more PCR samples than all of NOAA has ever combined during the COVID pandemic. We have the technology to do it. This is possible. You know, just do we have the investment and resources to actually commit to doing this? And so I'll just throw these up, you know, ultimately, summarizing that we need to think holistically about biodiversity monitoring. We need to think globally about our biodiversity monitoring can't just be, you know, in the US, the species distributions are going across borders. We have to develop these resources and capacity both human and the databases. And we need to overcome some of these key technical challenges and leverage our long term observations. With that, acknowledge many, many people that have made all of this work that Paul and I have done possible. And happy to take any questions. Well, thank you, Paul, and that was really. I, my computer has just decided to die. Slowly charge that what's going on. So bear with me. If you see questions coming up in the chat. Let me know, but we have a anyways. You will have a couple of questions. Yeah, I'll meander a little bit and then you can figure out where the answer lies in this and it's really pointing to the fact that it's a it's a new measurement technique and every new measurement techniques that has calibration understanding issues and it's really, I suppose, asking the question, do we really understand where this DNA that's being measured in the water has come from what its trajectory is and whether the tween species and within environment there are real differences in persistence or abundance and availability and it's really kind of understanding of whether the kind of does it really calibrate and do you understand the implications of the presence, could an organism eat another organism and through a fecal output release DNA and so move DNA DNA around, you know, without the organism ever being present. And so we really know what's going on. Yeah, absolutely. I can answer that question and Paul at you chime in. So I think, I think the answer is we're learning more and more every single day, you know, do we have a perfect answer there. No. But there's a lot of work that is compared to variability in DNA signatures versus traditional methods. When nobody likes to admit, certainly not know, is that there's a lot of variability between Charles, but we don't quantify it because we, it's too expensive to go back and do a second trial at the same site the same location. So my DNA is actually sort of guilty of we go in and we're holding ourselves very highly accountable we take tons of replicates we know exactly what our variants is right and so we can quantify that we can look at it and I would say that is a huge advantage, knowing what your variants is being able to quantify it now your questions about space and time resolution there's a ton of really cool projects that are going on actively and there's in the last year there's been 10 DNA papers that have been published per day. I'm not sitting on top of literature I don't think anyone in the world has. And so we're learning so much. And, but to answer a question briefly, like from what we've looked at in both coastal and tragic systems. It looks like eating a really is a snapshot of here and now. And so you're something on the scale of a, you know, a couple hours to a day in terms of time resolution and, you know, spatial scale of something on the order of we've seen differences as small as 50 meters apart as different as 10 meters apart in the water column. You know, while people were actively serving so there's at least like, you know, four foot waves. And, but other places and other systems, we just, you know, I did some work up in Kenai River. You know, 40 foot plus title exchanges and it's like in those systems there's an active title board, like yeah, you can move 10 kilometers but that's not surprising because it's going moving the title board is moving faster than a car. So I think there's still a little bit unknown about exactly pinning down those questions of how long and how far and have distant but everything that we have seen so far is that it has resolution, certainly, you know, on the scale of 10s of meters to hundreds of meters and hours and so I think, and it's going to differ in different environmental systems and so what works in Antarctica is not going to work in an estuary in the Mississippi is not going to work, you know, in the twilight zone, you know, off of Cape Cod and so I think really understanding transport dynamics and, and thinking about integrating which I don't think is done at scale enough like getting those physical oceanographers to talk to the DNA biologists there's been, you know, a handful of folks that have been really doing that to nail that down. But everything that we have seen so far is that it is relatively spatially controlled, and the results, you know, make sense, you know, with a little bit of grain of the salt, and especially having those reference databases in place, they're able to answer a lot of questions. I mean, I would just, you know, say what Ryan Kelly says very frequently is that every observation method we use has biases, and, you know, whether we admit it or not, you know, visual surveys is an incredibly biased method. And we can show that, you know, 70 or 25% of our samples pick up black sea bass, they've seen one in 30 years. So clearly there's biases. I think a lot of established methods, we've sort of for we don't talk about how they're biased and so, you know, it's not, it's not about, you know, meta bar coding. Certainly, you know, arms doesn't have the same transport issues that EDNA does, you know, it's a different meta bar coding approach. It's telling you about what's in this one, you know, cubic foot of simulated reef. So, I think a lot of the methods that we've been using for decades, we kind of forget, you know, that these have biases as well. And it's not about, you know, replacing one thing with the other. It's about, you know, integrating complimentary approaches so that we can actually improve our understanding of how these ecosystems are changing and functioning over time. Susan. I have a question. And I'm just fascinated by your ability to get the EDNA out of the reserve samples to the biological collections. But I'm wondering, is there going to be a potential bias in what actually gets preserved in those collections? And how would we now understand what that bias is? Yeah, absolutely. That's something that we looked into. And there was nothing, at least in our pilot study, we had, we found no evidence there, but it's definitely something that needs to be done to calibrate these. And I think there's a really cool work that I've heard of coming from Mumbari that's going to get published in the next couple of months where they were actually looking at form length preserve samples over the last 30 years from sediment traps. And they actually did a whole bunch of intercalibration, you know, where they were able to pull recent ones, expose them to form length, go up to a year, see if they can see, detect any changes and impact of that. So I think there's things that we can do in the lab to know if there are those biases or not. And at least from what we saw, we didn't find really any evidence that that was happening. But it's super important to obviously calibrate and validate as in with any science. And then the second question I had was sort of the problem of contamination if you're looking at low abundance species, and how do we detect that? Yeah, I mean, I think it's true for the same as the way we were doing COVID tests, right? I mean, it's just like implementing controls at every step of the way I have field blanks, you know, this is this the standard of microbiology like applied that to and so there's a really great wonderful standards that people have thought about for a long time. And I think one thing that hasn't really happened yet is, you know, a lot of people came into the ED&M at a bar cutting world as fish biologists. And the microbiologists were like, this was the third paper ever published, you know, after the human genome was ED&M sequencing. It's been around since 2001, at least if not earlier in some form or another. And so, you know, having better cross communication between those research fields and leveraging what has been learned and optimized in the medical fields where they have just the same problem of contamination. And they've built out systems to control for those and it's very possible. Well, I can give you an example. I think it was SARS-CoV-2 being detected in England by the public health England and it was, you know, before they could detect it in the population and it was a contamination issue. But it wasn't detected until I think several years later that they have the contamination. So that's why I mean it's just a it's a cautionary, you know, sort of example of how it's easy to get contamination and hard to detect sometimes. Yeah, one thing that we've done around that is just to implement multiple replication at multiple scales. And then you can use that to really understand, you know, what are real detections, what are not real detections by having that kind of built into the site to sign it. So, and luckily one of the advantages of EDNA is because it is relatively cost effective, like it's, you can implement these replication levels at multiple levels in a way that you'd never be able to afford to do that, you know, for a SCUBA survey or visual survey. Okay. I just was curious to get people's thoughts on the issue of data sovereignty in terms of how we do this. Now clearly work is done collaboratively, but on the other hand, we're also seeing a lot of regulation of framework of regulation on data collection, especially generally data collection in areas where national distribution of infection is, but it's curious about how you're thinking about that. And you as an anybody, all of us. Is that addressed to the general. Well, I was hoping a panelist who I'm debating for. That is a lovely, why don't we use this as a segue. We are going to find time and now I want to open it up to sort of the more general discussion. I know this was just going to discuss all ocean life in the next 10 minutes. The transition and I think it gets into what we need to the previous panel was talking about the need to look globally. So, if you're all still online, please chime in. Hi, I'm still charging cover so you see somebody on the room that I was going to go ahead. Just to respond to that I had some connection around this issue of with the global networks are go folks and writers about the mentoring other national territory and water. The key issue that's always come up is any inclination or indication that they're making biological measurements. As long as they're measuring temperature of salinity. Most nation states have shrugged their shoulders a better for both our goal and even glad that they're not allowed to fly. There's a perception that they're making biological. No, you don't want to there and it's a big story. And so there is a lot of national sensitivity. Once you get into the biological. So I think it is a big issue. And try to pull up that information. Yeah, I just like to add to that. You know, I think. You know, the convention on biological diversity has been really, really important and highlighting. You know, just how dependent humanity is on biodiversity and getting. People to think about protecting it. I think one of the unintended consequences of this is that. Because. You know, one of the selling points to developing countries was that. You know, the cure to cancer, the cure to AIDS, the, you know, what the cure to whatever you could be on your reef in your rainforest. It is basically equated. Any sort of. Biodiversity work, particularly anything involving genetics to bio prospecting and. You know, over, over my career, like I've basically seen the evolution of biodiversity science be, you know, it. It's the one use of biodiversity. That is almost like, you know, it's approaching illegal now where it's like even, you know, even in the US, it's really hard. You know, if I want to go. You know, collect some animals, you know, to eat, that's totally fine. If I want to collect those same animals and sequence their DNA, it's like, well, hold on a second, you know, this is, you know, we got to talk about this. And so I think one of the things that we need to think about as we are talking about, you know, looking at biodiversity, both in the US and, you know, globally is, you know, to sort of push back on this on this. You know, what is essentially the criminalization of biodiversity research and I don't use that word lightly in Indonesia. They actually have made, there's a national law that any violation of a foreign researcher related to biodiversity research is punishable by multimillion dollar fines in 10 years in an Indonesian prison. So it is being. It is being criminalized and it's criminalizing people who actually are doing permitted research without actually doing anything to decrease people doing illegal research. I guess the question was also about data sovereignty, and I think that's also important sort of even beyond genetic information, but you know, sharing data internationally, especially with, you know, countries with fewer economic resources really requires a long process of building trust and understanding how benefits can go in both ways and identifying what those benefits can be. I think there is a danger if it's just making data openly available in some ways that becomes a different form of colonialism in some ways with countries that have resources to take advantage of those data using them. So I think it's that process of building trust and thinking about how benefits go both ways is an important part of that international relationship and international collaborations. Sarah, are you chiming in on this topic? Sorry, I missed that. Did you want me to speak now? I was wondering if you were adding on to that topic or if you were changing topics. Yeah, adding on. Well, I think this really is like a push for thinking about those multinational funding opportunities that I think should be creative because then you can actually give leadership opportunities for people in these places. So, you know, for example, if you want to understand the topic was brought up by Paul, like if you want to understand connectivity of a species across its range, you need to work with like many countries. And you also don't want the situation where people are prepping libraries in different ways because we know there's like drama, like, it doesn't work. So you need like, we need ways to like move samples between places, and we need to provide funding opportunities for people to collaborate with us like as equal leaders. I think the multinational funding opportunities is something that like came up as a common theme to all of my colleagues when I was chatting about like research needs. Everyone was just like, uh, working across international lines, like CITES permits, this that it's like, and they want to work with people and people in these countries want to work with us. Like, it's like, but there's no mechanism in place. So I think it's a real need. Yeah, I got, I want so badly to add what Sarah is saying. Set that aside for now and pick up on what Paul is saying, because it's more directly relates right now to this. How we're using the technology. So one, I think Paul really hit the nail on the head. It's many in our community have spent the last four years working with the United Nations on high seas. And I can tell you that there's a lot that is going into that regarding marine genetic resources. And the next few years that we're working to advise policy makers on the implementation plan, there is a lot of room for error. And I would also just want to add on a different note, there's a lot of room for opportunity. And I think that one thing I would like to see us do as a community is really keep up the engagement with the delegates who are who are really working on the implementation plan. So just leave that there. But it does connect in a way where Zachary is presenting and I have a question. I suspect there aren't enough resources being made available to, to, for lack of a better word, cross calibrate what you're seeing with EDNA and other analytical methods, both existing ones or how EDNA relates to, you know, you know, how we use EDNA approaches to look at the historical archives and the like. Do you have a recommendation to us on one thing you think you'd like to see NSF do to help us better establish the relationship between what we see in EDNA samples and related technologies to existing, you know, imagery data sets or other kind of existing approaches and data. Yeah, I think the thing that I would love to see there is, I mean, I think ultimately, in sort of the same way, Heidi was talking about the IFCB is right, it's like, in that regional local area, how many observations were needed to basically validate the IFCB. I think we need that, frankly, for every real large scale application of environmental DNA, or at least as many of those as possible, right. And so if we're going to pick one, I mean, I think there is just a real emphasis on doing those kind of pilot ground tree thing, validation, very favorite studies. And I think, and I think the components to that that I think are really critical are having those reference barcodes for the species of interests, because I mean, we're doing a separate across just species that NOAA monitors. Our best estimate is it's like 15% have genetic resources. These are the like lowest hanging fruit like these are the things that like we actively care about I mean some of these are whales for God's sakes like, you know, let alone the policy worms that like nobody cares about except for except for except for the people that you know, like are out here. And so I think like, like those need to be there, but we also need to do much more of that and I would you know I'd point to one of, you know, the coolest cooler papers that's come out, but only Sheldon's work and lots of others at Northwest history Science Center where they did side by side acoustics of hate trolls in EDNA. And it's just like, boom, on the money those maps are there like, we feel really good that we could, you know, if we just had EDNA signatures, we could interpolate them over space and time alongside those acoustics and so it's that kind of ground tree thing and work that I think is really critical and it doesn't take a lot to add in some of those components and I think we can leverage, you know, existing LTRs that are already happening existing surveys that are already being funded, you know, by, you know, indirectly or directly NSF and other sources to you know basically add in right all sort of mentioned this and you know, it's never going to replace boots on the ground, it's never going to replace white ships on the water, right, but we're never going to have a fleet of 108 global class white ships. I mean, we can dream, I don't think it's going to happen. But what you could envision is having 100 gliders complement, you know, our core set of white ships and provide us the icing on the cake to extend to spaces depth times that we could not ever get to you know with those ships and so that's really where I think we need to have those that validation so we can scale appropriately because, you know, as that slide showed like our ocean biodiversity observation, or it has not scaled with our ability to measure temperature and salinity and other biogeochemistry, but we need to get there if we want to answer these big scale questions. Yeah, I have a comment on that but also a question that might shift it away so I don't know if other people have comments on this topic. Okay. So, um, regarding the DNA, you know, we're, we're in Hawaii, thank you, thank you all for your panelists. We're working with, you know, communities and for the brain, you know, a lot of the resource managers are interested in a new technology and, you know, what like speaking of data sovereignty or try to provide communities and I know Ryan Kelly, you know, talks about this a lot to providing communities with an agency to conduct the methods themselves and then, you know, the scientists actually providing the guidance for curation so that the ownership and decision making stays within the appropriate parties and I think, you know, I would like to hear the party, you know, the final thoughts on that. So that's, that's one, sorry. And then the second thing is kind of related in that, you know, we're also I'm part of the nurse system and, you know, we're working on DNA technologies and estuaries and, you know, as we shift the thinking of oceans and humans like holistically, you know, the coastal regions estuaries don't, you know, et cetera, are very crucial areas, you know, so speaking of that, what are some strategies for an ocean, both to explore these important kind of questions about biodiversity and technological challenges that can be contained within the regions and estuary or climate effects maybe both extreme like so. So, sorry, long question, but I would love to hear any other panelists. Thanks. I would, I would just add one little bit but I've been talking a lot so I call this. The one thing I would say is like, I think what we would want to answer that question of like being able to enable anyone do the science is like, think about NCBI and NIH's support of blast. Anybody with a DNA sequence can copy paste that it's not hard to teach, you know, anyone with any computer literacy, how to annotate a single DNA sequence. And it tells you the answer and you don't even need that good of internet connection we've done this and Paul's done this a million times in Indonesia. And how can you enable that same kind of word for EDNA data, right, you could imagine that actually the Department of Energy already does this but for microbiome data, but it's only on microbes and they're not allowed to do this work in the ocean right. So how do you get the interagency calibration between there are, you know, the nationals, you know, NCBI, DOE, NOAA to be able to have sort of a data portal that implemented best practices and did all the behind the scenes work. And just, we're not keeping the data we just give it to you as an output right and so that enables anybody who gets a DNA sequence, even if you know you're a tribe and serving our quest, you take a water sample you send it off to a sequencing company they email you the raw sequence data that you know they're sitting on and they have the sovereignty, they can put names to species and analyze it and we don't need to keep any of that information right like that kind of open source and providing the tools and resources and to be able to do that scale I think would help help that and I think, you know, we think about this all the time. Lots of collaborators in Indonesia and former lab mates of mine, PhD students of halls. You know how do we enable them to be able to analyze this data in a country where they don't have a supercomputer the supercomputer that UCLA, you know it's awesome that we can do this for the people that we know and work with. But how do we, you know, extend that globally. And so I know certainly a lot of my career is building a lot of these software tools to, you know, diversify and allow anybody to do this work but I think we could also build an NSF could think about integrating use that scale to provide that kind of tool that enables anyone to do this kind of. So, reflecting back on this entire session, we heard a lot. And then Steve said we need to go back to looking at what is the moon shop. So here's my question and I would propose that we go back in the order of presentations and just step through each one of the speakers. And here's my question for you. There's going to be go back to. One sentence mission state. What are we doing. What is the emphasis of go back to whatever we call it in this context, the biodiversity and climate change impacts. What's that one bullet point was that one statement that you can use to sell this and energize global ocean community for the next few years. I think you should go in reverse order of speakers. Well I'm not shy about this. I mean, you know, I clearly see the utility and an at scale project that's large enough to actually encompass the range of issues at various taxonomic levels that allow us to understand the implications of a rapidly changing planet on the biodiversity of the earth. That's my elevator speech. Yeah, I come up with something fairly similar. I mean the future of ocean life in a changing climate. I think that that really is a cross biological scales and across oceanographic processes really requires massive integration. I think my pitch about the, we can't know what's what we're losing until we know what's there, and that will require multinational collaborations across species ranges. And I think I agree with everything that's been said, I think, you know, converging on some standard approaches and some priorities for what we're trying to document is going to be very important if we're thinking we can be, you know, far reaching but also fairly efficient, even, you know, even with significant resources. Oh, you know, the more we use zoom the worse we get. I think it boils down to integrative ocean observations and and really sort of adding this more fine scale biodiversity model monitoring into these other ocean observation platforms we have. Because, you know, it, it will allow us to see what's there it allows to see how it's changing but by integrating it with all of these other more physical observations that we're taking. We can actually start to understand some of these relationships between these physical processes and the biological communities. Yeah, the only thing that I would add to that just to echo agree with what everybody said but this, the, the connections to human well being and particularly for whom the impacts on which communities, I think is is critical if we want to understand the impacts of these changes. I know you're not understanding you. I think everyone covered it. Nicely done. I think Jason will help us up well. Can I add one component to see I'm sorry. So I think sort of underpinning all of this is where it's possible that we're co developing these kinds of approaches and research and observations and data provision with groups on the ground that you know, really require help or information or have their own knowledge streams even to contribute. So that co production co development pieces is pretty critical now. So we need a biological origin circulation experiment. We are almost out of time and I'm going to have the last support. So make it a good one. Yeah, well, I'll try my best. I'm asking this question from a perspective naive Italian lack of knowledge, because as folks know I'm not a biologist any stretch away. But I am influenced is a workforce question. I am influenced by a lot of what I'm observing in the medical genetic world road Institute and other places like that in the Boston Cambridge area. They really have a terrific model of having a lot of hardcore biologists, microbiologists partnering up with mathematicians or five mathematicians might not be trained in biology at all. And I'm wondering if there's opportunity, or if that is a direction that the marine world of genomics is going. And again, I'm asking this from perspective of lack of knowledge on the part maybe this is happening. Maybe not, but I wondered if there was a desire for that or if there's anything we can do to support in our work translational research being created and manifest by expanding our fields of expertise that we have doing this type of things. I don't want to comment, but I wait for the family. Sure, I just a couple points. I mean, one is a, I think a huge need for bringing mathematical and statistical approaches to understanding biodiversity data and biodiversity change. So that's a lot of what we've been talking about on this panel and I think there are a lot of unexplored opportunities there on the genomic side. And in the context of changing environments. There's a lot we haven't done in understanding genomic functional genomic variation, not just species and cryptic species but, you know, what are the functional genetic differences between different species between different locally adapted populations or individuals. And what role can that play in ecosystem restoration and understanding which species are likely to survive changing ocean conditions and climate conditions. So, yeah, I think there is a lot of opportunities that has been unexplored. Yeah, I just wanted to say that actually, a lot of the bioinformatics is, you know, there's a lot of crosstalk. Yeah, a lot of them, right. The crazy thing is that a lot of it came from the marine ground. I mean, folks went into human genome and things like that with training in ocean micro, aquatic microbiology. The problem is the number of zeros in front of the just, you know, the human health angle just has so much more when you show up, but you actually have a green drain going in that direction that way. Yeah, that's interesting. All right, thanks folks. I guess I would just add one little element to that which is, I think there's also a ton that they can learn from us. Like we, like one of the things when I started and, you know, Paul and I were first trying to do ED&A, we can't get to family level resolution and talk to a marine manager. If we say we found fish in the ocean, no really like cool, you know, like, but if you do that for bacteria for microbes, it's like, we've never seen this thing. So it's fine. It's good enough. And so like a lot of work came from that marine world where it's like, we need to put better names on these things. And those actually ended up leading to improved developments in the microbiome space. So I think it's, there's a lot of opportunity in that cross space and also, you know, thinking about the leveraging the physical oceanography and AI and machine learning all of these places. Like, I think there's a real need for these, these gross data. They deserve definite translational is in both directions. And a lot of times those, those form mathematical field type people, they're attracted to what we do because it's so messy compared to a relatively cleaner world. So they actually really appreciate this challenge too, so you're right. I really want to thank the panel for the whole panel for this morning. We covered a lot of topics rather quickly. The panelists, we were superb. He gave us a lot of good information. As I said, we will be continuing some discussions about acoustics and in addition, not just acoustics, but other technologies sensing marine life. And I just want to thank you all, I want to thank you all for hanging in there for a long session. And this has given us a lot of information about so appreciation and I think that wraps it up for you. Indeed, and I also want to thank all three of you moderators Jason, Peter and Joey for just doing a phenomenal job. And of course, our panelists, I don't have a biological background so I learned a lot here over the last few hours, and I have more thinking into do to make sure that I understand everything with the help of the rest of the course. Thank you.