 Okay, so I'm going to talk about the work of my team here. So Rosie Ware is the postdoc on the project was around for the first three years. Unfortunately, she can't make it here today. She's moved on to new pastures. And Becky and Rosie here are two PhD students who've worked over the past few years on some of the stuff I'm going to talk about here. Incredibly hard at times. So for the uninitiated sedimentary DNA is ancient DNA that occurs freely in sediments. And it comes from a number of different types of sediments. And as an area, it's about a decade and a half just over an age. So we typically get stuff from permafrosts. There has been some river systems and lake systems, ice cores. There's fantastic work that's coming out on cave sediments and hominin in evolution currently. And obviously marine cores, which we're going to talk about today. So you're taking these back to the beginning of the Lost Frontiers project. And I've stolen here my favorite rendering from Martin of the landscape as we understand it to look just to visualize what's going on. We have this fairly sort of simple idea that we're going to use DNA to populate the landscape. So we'll bung things like trees on it. But like I say, sedimentary DNA is that it's still a developing field. And particularly at this time, there were a number of challenges to overcome. Not least the most basic, is it really there? Is it real? Can we actually determine what types of organisms are there? Which is not a trivial problem, as you'll see in a moment. Kind of connected to establishing that it's real in the first place. And then can we say things about the frequency of organisms in the landscape? Something about the biomass, which I won't talk about so much, but it did come up when we were analyzing the tsunami stuff a year or so ago. And then we move on to questions about how the DNA that we see that's there and we think we know what it is and we're pretty sure it's genuine. How did it get there? And we'll move on to that later in the talk as well. So starting off with the, is it there? There we go, excuse me. Is it there? We have various aspects here. So we need to understand how DNA breakdown occurs in the environment. I shall begin on by talking about that. And we'll also talk about these other aspects of authentication, phylogenetic assignation, and some of the issues of topology that we're coming to really just now that all the data is coming together. So of all the different ways that DNA can potentially break down, we were thinking about this a lot in the 90s. It turns out that it mostly only breaks down in two ways that we see in ancient DNA. And both of those are hydrolytically driven. So we get a depurination process. I get my laser pointer up here. Okay, so depurination, which basically knocks out a purine base, one of the bases in the DNA chain which exposes the backbone. And that leads to fragmentation. So when we look at DNA reads, DNA sequences, we tend to see quite short sequences. We will see a distribution of DNA sequences that looks something like this. In reality, the distribution of DNA molecules actually follows this sort of exponential distribution. And we tend not to see these when we're doing our analyses, partly because they're not being extracted in the DNA process and partly because we can't process them bioinformatically. There's not enough information. They're not long enough. But the important point here to make is that it's not really the mode here. That's the important part of the maths. It's the parameter that described this exponential slope which is called Lambda. And that'll pop up in a different guys a little bit later in the talk. And the second type of breakdown is this process down here of deamination where cytosine, one of the four bases of DNA will lose an amine group where it actually becomes a uracil which is the RNA equivalent of the thymine. So we actually get a change in the DNA sequence. And that shows up on the ends of molecules and that's become over the past decade or so the standard way of identifying whether your DNA is really ancient or not. So this graph here demonstrates the length of the DNA molecule. And then towards the end it's showing numbers of mismatches or proportions of mismatch. And this red line ticking up here is showing that it's finding thymines when it expects to see cytosines when it compares DNA sequences to a database. Okay, so those are our two basic sorts of parameters. Now, a few years back, our group started looking quite intensely at environmental correlations to DNA breakdown. Partly in the wake of our first work with Vince actually around 2014, 2015, looking at the Moldner cliff sites. And in this particular study here, we looked at basically all the polygenomic data that was around at the time, about 183 studies and correlated with environmental variables to understand how you can explain that fragmentation. So this is that lambda variable. And the important thing that I need to say from the start, and again, this is quite important in understanding the topology of the DNA in the Stogoland landscape is fragmentation doesn't increase with age in a noticeable kind of a way when we take stuff out. That's because fragmentation happens very quickly. And after that, it's very difficult to tell the difference between say 10,000 years and 500,000 year old DNA. On the other hand, deamolation that change from a cytosine to a thymine does increase over time. Now, what that tells us is that DNA in the environment isn't continually breaking up over time. It's basically disappearing in signal because of bulk diffusion processes. So it all depends about how closed your system is. So things like bones, like Petros bones in the skull, which are very, very dense, are great closed systems. So they're just as damaged as any other DNA of the same age. But because they're being held in a small box, there's more of them. So it's basically a frequency game that we're playing here. So we expect to see DNA damage correlate with environment time. And a first sort of glimpse of the type of thing that we're seeing from the Dogoland data is that pretty much seems to be the case. We say we've got people which are more acidic environments. You get a lot more in the way of hydrolysis going on there. We see more damage, so greater lambda values here. Whereas if you go to more salty environments, so there are good reasons why I don't really have time to go into why salt will slow down the process, you see less damage. So we've got a nice sort of correlation between environment and the level of DNA damage. So we needed to do a number of innovations in this project to be able to study the sedimentary DNA. And the first of which was establishing an approach to authentication. So this reflects a lot of the work that Ashley Rosie Everett has been doing during her PhD. But what we really need to see is the ends of the DNA molecules. Now, although authentication has been used for the past decade or so as a method of identifying whether your ancient DNA is real or not, the conventional approaches require a huge number of DNA sequences from a single species to be mapped against the genome. And our problem is that we've basically got a metagenomic situation here where we'll have maybe just a few reads from each different species. So we've developed a tool called Metadamage which will shortly be available for download through the unconventionly named Github. Beloved of computer geek types to download the program. It should be immensely useful to the sedimentary DNA community. And basically this gives that sort of typical output showing, as you can see in our embryo fighter, we've got damage on the ends of the molecules and we've got a confidence interval because we can deal with quite low numbers of reads. This has actually got a fair number here, 3000, but we can be really quite sensitive. We can go down into just a few hundred reads and still see a damaged signal which is immensely powerful relative to what's available out there at the moment. Okay, so we've got a process of identifying whether our DNA is real or not. And then we've also been developing this phylogenetic as a nation approach. So so-called PIA or phylogenetic intersection analysis which is published and available for download. And it's a very stringent pipeline because once you are using metagenomic DNA in order to get that authentication signal, you're basically looking anywhere in the genome. So it becomes actually quite difficult to accurately say what organism you've got. So quite briefly what this algorithm will do will give you a phylogenetic range in which your organism can occur. And it also, from the way the algorithm's set up, starts off with the starting assumption that the DNA you're looking for may not be in the database and quite often it isn't. So it takes into account database density of occupation. You're gonna go into that too much, but the long and the short of it is we're about 90% plus confident of our phylogenetic as a nations, sometimes more, sometimes less. Okay, so the third innovation to deal with sedimentary DNA is to worry about DNA movement post deposition. So whether there's leaching up or down. And this has been a big problem in terrestrial sediments in particular. So what we've done here, and you're not gonna see too much of this, is but it's a statistical framework to test whether samples are different from each other using things like beta distributions to check to see whether there's any homogenization process going on. Okay, and then the last thing we did, again, I'm not gonna talk too much about this, but it's in the tsunami paper in geosciences, is to get a proxy of biomass. So then the amount of DNA that's left behind by an organism is a function of the organism's size and the size of its genome. So that takes into account those two factors. So things with very large genomes are gonna be more present, especially if they're large. And in that way we could show during the tsunami episode that we get this increased presence of biomass of trees in core one A. Okay, so let's give you the background. What I'm gonna talk about in our data here, and again, like all the other talks previously, we are literally just analyzing this at the moment in very early stages of integrating, particularly with the other environmental sources. And I'll talk about this line, of course, going up and down the river system. So this is a lot of work that was generated by Becky more recently and then Rosie Ware in the first few years of the project and a little mention of some cause up here. So particularly core 27, right. So I've shown you this already. We are generally quite happy that our data are genuine ancient DNA. As you can see, all our PIA authenticated data is over four million reads and we can see that in our extraction banks, we don't see damaged DNA coming through. So we're pretty confident in the data that we've produced. So that's what the data look like. The problem from here on in is we've got about 600 plus taxa, which would make quite awful sort of paralogs of pollen diagrams and diatom diagrams. So what we've done is to analyze the correlation of occurrence of organisms together. So basically we've got a correlation coefficient matrix which tells us which taxa occur with which other taxa and helps get rid of some of the degeneracy. So sometimes we've got family level assignations, sometimes their genus level and often those occur in together because they're the same organism. We can invert those into a distance matrix from which we can make a principal component analysis, more accurately principal coordinate analysis. And from there, we can skip up into dimensional space making hyper cubes. So that's a fancy term. You can just think of it in three dimensions if you've got a three dimensional graph and just broke it up into squares into cubes. We're looking for similar occupancy of the various different taxa. And from that we build what we call plant guilds. So those represent basically different depositional categories of plant taxa. And I'm gonna talk to you mostly about plant taxa in this because there's a lot more of plant data than anything else. Okay, so we've got about 56, no, not about exactly 56 plant guilds and these are they and some are more interesting than others. And some, even if we skip up to 10 dimensions remain quite large. So this one in particular here, we've still got 60 taxa or so in here that do not want to stop holding hands with each other. So they always co-occur together. And then we've got other groups. So up here we've got our Zostra, so seagrass groups. So this is our representation of marine environment. And we have other groups that represent some things like reeds. So I'm gonna pick out some color codes here. So green, remember represents this sort of marine inundation. We've got a reedy group here in green. We've got a grassy group here in brown. And very much like Ben was talking about earlier we've got this a lot of salics, a lot of willow occurring, willow and friends. And then this big group here you'll see it's got a lot of taxa in it but it doesn't feature very much. It's very restricted geographically. And then I will come to this towards the end of the talk. We've got some interesting groups. I noticed quite early on this occurrence of jugland ACA and you'll see next door juglands. So Walmart, which both surprised me, worried me and excited me in turn. So I was driving my group up the wall I think over the next couple of years trying to establish whether this was a real signal or not. We'll come back to that a bit later. So we've got this guild structure. And what we can do is go through each of our samples within the cores and there are 322 in all. And we can ask of that sample what the guild structure is, what percentage of the guilds represented. And we end up with this much simpler sort of thing to look at. So this is a typical three types of guild that we see a lot of. So we've got this marine kind of situation going on here. Zostra dominated, we've got a Willow dominated and Grassland dominated often with that reed group with it as well. Okay, so if we look at those against sediments, now sorry because we're going up to birds eye view it's very difficult to pick out here the key hasn't perfectly rendered but the general point I wanted to make here was that we see a lot of changes in sediment types. So a change in depositional regime going on without much in the way of a change in the plant guild structure. It doesn't look like the incoming sediments are changing the plant make-ups that we see. And similarly, we have plants changing over time without the co-change with the sediments. So my first impression of this data is that the sediments coming in are not really driving the plant presence which sort of takes us to the first step of thinking about the DNA depositional model here. So in cases like Zostra and Reeds we might expect them to be actually in situ. And then things like Phaelix and the other trees we have a sort of more proximal situation going on where to position is anybody's guess, leaves coming in. And then we've got our sediment sources coming in here. And I've sort of formalized this a little bit. So we've got our influx from sediments, but, and we're coming back to the importance of the lambda and the half-life of DNA. You've got DNA disappearing from the water or decaying. Disappearing is probably a better word for D there. So our deposition of DNA is basically a sum of these various different factors and it's got to be the influx minus the decay. And obviously we've got efflux going on here. And what it seems to me is that S and H for the most part are larger than I minus T. And here we have just building up. So what's an important fact here is the half-life of water, of DNA and water. And we do know from sedimentary, I'm sorry, from modern DNA studies that that lambda, and this is the same lambda that I showed you at the beginning is greater than one. That means that the half-life of DNA in water systems is typically less than an hour. So this stuff is getting stripped out very, very quickly. So I expect under normal conditions that DNA is not really coming in on sediments influxing. But when you have a sudden influx such as the tsunami, and we do have examples of sudden gravel deposits, then this value of I minus DT can suddenly increase and have more of an influence. So we can have influential aspects going on there. So I'm under the impression at the moment that actually what we're seeing in the DNA is mostly a local deposition of DNA. Okay, and this sort of summarizes what I'm saying. I'm gonna have to speed up. So these other points you can ask me about afterwards. So here's the river system and we've got a couple of transects going up the river system and then across. And that is going to be displayed in the next slide. We also have ELF 27, which is around up here somewhere that we're gonna look at now. That's just to remind you what the guild structures tend to look like. And then there's this other guild structure which turns up, which we'll talk about in a moment, but this has that warmer signal that I was talking about previously. Okay, again, we've gone up to bird's eye view. So this is the Southern River system with a mouth here moving up to the top. And then we've got the second transect across the top there. And then we've got seven and 20 that Ben and Tom were just talking about as we're further away and then 27 further away again and 22 is right up on Dogger Island. And what we've got here is the sample stacked against all the various different radiocarbon and OSL dates. And this is sitting with Tim and Kevin at the moment to map against sort of age models. So what we have here is some absolute dates and samples tend to be stacked around you. So we've got an OSL date here and samples stacked up around. So we've got a sort of a first impression of the age distribution going on here. And broadly speaking, you can see lots of that green structure of Zostra going up and around the top here. And in the center here, you've got lots of salix structures going on. Over time, we've sort of got this broad inundation impression across cause going on. We can, I can take you through the, I'm not sure I've got enough time actually to take you through in too much detail. What we've got in early stages, we've got Willow down in the bottom here. So this is deep in the Pleistocene and the sort of Rivermouth cause. And we've got Willow, we've got cold taxa like the Dryas and Aracaceae. And then as we move into sort of 11 and a half to 13 and a half thousand years ago, we're sort of spanning that sort of Allerod and younger Dryas periods where we're seeing again, some Dryas going on. But what I shall actually start showing you is some of the distribution of some of the interesting taxa going on here. So we've got Dryas popping up around about the younger Dryas. It also pops up later on. And we can also, just to demonstrate the usefulness of the guilds here, we fill that information out a little bit more if we've got the guild that occupies that has Dryas as a member. So it's a sort of a cold guild, interestingly popping up in this period around here also. And if we look at something like Betulaceae, so this is Birch, Birch and Older actually. We see that it pops up all over the place fairly early on, but uniform throughout the cause. Something like Coralus, we've got it turning up. We've got it turning up very early. So we've got what might be presence of Coralus in Allerod period. And then quite similar to, I think what Ben was showing later, we've got this sudden appearance of it around about that sort of 10 and a half year, 10 and a half thousand year mark over time. And again, then pretty much pops up everywhere. And then we've also got Oaks following a similar sort of distribution that occurring fairly early here. Okay, more interestingly, I think in some ways is Lyme. So we've got Lyme, and I know this is a complex deposit over here off shore, quite old and in Elf 27, but occurring a couple of a thousand or two years earlier than the UK mainland. So it looks like it's generally a bit warmer earlier. Again, pending the development of age models as we sort of integrate everything together. And there, there's an example of where we've looked at specifically these Tillia reeds. And these are the ones that have been through extreme I'm sorting and we see a damage signal. We're pretty sure that's real line. Okay, so just sort of getting towards the end, we've got the presence of this, this juggling signal. So this Walnut signal. Now Walnut, we don't expect to be in Northwestern Europe. As I'm sure you appreciate, doesn't produce a lot of pollen either. Recent studies do place it in sort of South French Gaucho refugia, but in the archeological record, actually the oldest Walnut remains are about 10 and a half thousand years ago from Spain. And there's quite a long tradition of using Walnut or moving Walnut around. And what we're seeing here is what's apparently is about from 11.3 thousand years up to about eight and a half thousand years of presence of Walnut. And then much later in much lower quantities further up the valley. Now the Walnut signal is real. So it's definitely Walnut, they're 100% matches and there's not really anything we can mistake it for. And it's genuinely ancient stuff. So it also occurs in its guilds, as I remind you here, with some typical disturbance indicators like nettles and wheat. Not, sorry, not wheat, that's just the wheat tribe, but we actual wheat is not there in the system. So we've just got wild cereals going on. So I'm currently thinking we're developing the opinion that this is a possible anthropogenic signal. So if we would put that on back onto that landscape, this is where it occurs and the size of the trees here are relative to frequency. So that's your 11.3 to eight and a half window. And then much later on, we get small instances that it's further upstream and then eventually up around core seven. So I've burned up quite a lot of time here, but broadly speaking, again, it's a rapid overview of everything that's going on. The sedimentary DNA appears to be mostly local deposition. This is something we're digging into and trying to understand at the moment. And we're developing depositional models as we speak. So we're in the process of integrating with the rest of the team. We basically see a willow tundra to start off with and like the others, we move into a birch woodland and we can broadly see the inundation process going on. And I put it to you that there's possibly an anthropogenic signal in the data. Okay, so this has been, again, a lot of teamwork. So thanks to the broader team and particularly the dating guys at the moment. And I also say an extra thanks to a guy called Logan Kester, who was a fellow in my group a few years back who we still interact with. It's helped us out with some of the analysis here. And that I shall stop and I shall stop share. Thank you very much, Robin. We're right on the hour, but I think we should have time for a few questions as we're going into a one hour lunch break here. So I'm going to come up the obvious one of what animal DNA do you have or any? Right, so it would have been a two hour talk if we were going to cover that as well. So like I was saying, there is a lot more of the plant DNA and without a lot more interpretation to be done with the plant DNA. Animals do turn up and they're very low frequency. So it's very hard to put sort of statistical surety on anything that we see there, but we see broad trends. I mean, the usual sorts of suspects, there are going to be boar and bovids and that sort of thing, canids in the system. Interestingly, Becky found European turtles in the data which she traced through. So for a time, and we might still be thinking this actually that there's a mild aspect to the landscape at some time points. But there's not a huge amount to say really on the animal front. Okay, so Mark Bateman's asking, is there any reason to believe changes in DNA not seen in sediment changes could be due to post-depositional biotubation? So right, yes, so this is one of the things that we worried about from the get go and that's what we did our statistical framework. So what I should have pointed out actually when I showed you the broad data in its raw form is that for the most part across cores we see stratigraphical integrity. So they are statistically different to each other as you move up and down the cores which tells us that DNA hasn't been moving. So we don't, I mean, there's always gonna be exceptions to the rule obviously, but as a rule it doesn't look like the DNA has moved much since deposition. And Christopher Brooke is asking about the difference in the half-life of DNA between fresh and saline water. Okay, so then, so okay, there are two aspects there. In free water it's all gonna be about diffusion. So it's gonna be similarly quick to get rid of. In terms of actual breakdown, so there's two different aspects to half-life here. If you're in a high ionic environment, what you do is you interfere with the hydrolysis process. It's basically works on the polarity of molecules. And we have done some work on this and we see for the sorts of temperatures that you expect in sea water about a 16-fold reduction in the rate of DNA breakdown in terms of accumulation of deamination signals. But in terms of the actual signal disappearing, that's the DNA sort of diffusing out. So there wouldn't be such a great difference. So it's all about the DNA getting fixed in situ. Once it's fixed, then it'll last a lot longer in a saline environment. Okay, I mean, there's few around the macrophosal traces of juglands and the set of DNA, clarify how certain this is late glacial holocene and not derived samples, given that the macrophosals such as azalea are present in your cause and commonly found in earlier Pleistocene deposits. So my thinking on, so this was suggested as well, we're trying, I mean, I'm presenting this to the group and they're trying to understand, I wouldn't say I persuaded everybody for sure. But my thinking is when sediments are redistributed, and the move, so you got complexity going on, you come into the situation where you've got this half-life of DNA coming into play. So where you've got rearrangement of sediments, I think you lose that old DNA. So I don't think this is an old interglacial deposition. Now the exception to that rule, I would say, would be when you get a certain influx, a violent event that tears up a load of sediment, such that it remains cloudy. And again, we're outside of my area of expertise. It'd be a question for Tim, but I would imagine that's the sort of situation that you retain your OSL signal. So the sediments remain looking old. So this was my point here that the DNA as it's going down, I think should reflect the age that the sediment appears to be by the proxy's measurements, even though that might actually be older sediment, but the OSL signal, I think it probably shouldn't actually. So my current thinking, and again, we're at early stages, is that I don't think it's some super old DNA from a previous interglacial. Whether it's later, like modern walnut going in, I'm not sure how it can be, but it does show the damage that's consistent with everything else of that age. It seems to match. Yeah, and I think we'll be hearing more about that consistency in ages later on this afternoon, probably on some of the OSL profiling and turnabout. So we can leave that again for commentary then. A question here from Melanie Munt, mentioned that it's particularly problematic in terrestrial environments for the post-depositional movement of the set of DNA. But in terms of submerged landscapes, is there likely to be any possibility of the existence of set of DNA where preserved materials are largely absent or where the environment is particularly turbulent? So that question seems to be, can you get DNA when you don't have macrofossils broadly? And yes, there's a fairly solid track record of that in the literature now. And we're starting to understand better in terms of DNA decay processes that's within the realms of possibility. What I would say the area generally in ancient DNA and sedimentary DNA in particular is we don't understand yet in detail the interactions with surfaces of the DNA molecule. So it's not just, it's same with proteins as well that the preservation in the environment that you're talking about here is largely about the exclusion of water to sites of reaction. So if you get tight binding, so anecdotally everybody knows in the field that if you've got clay, you've got silica, DNA binds to silica, you tend to see better preservation going on. So these things can potentially be held in certain substrates for long after there's any larger macrofossil remain. So there's something about it being turbulent, what's? I think the environment of deposition I presume in that case. Yeah, and I don't think DNA doesn't do terribly well in a turbulent situation. And this is how we break DNA up in the laboratory by rattling it through sort of physical stress. So I would not expect, and in a turbulent situation I would have thought the other process diffusion would probably kick in. But that said, I mean, depends what you mean by turbulent. So when we see the Tsunami episode, it's very short period of time and probably this stuff is being brought in actually on biological physical material being dumped trees, the situation in the DNA is there a long time after the trees have rotted.