 Thank you. I've actually just moved in the last two weeks to University of Glasgow, so, but anyway, yeah. Okay. Good. Hello everyone, good morning. Like to first of all just say thanks for the invitation to be here I really appreciate this. This, this diagram or nice illustration you see here from Johnson and Henfries in the mid 1800s provides a really nice starting point for what I want to talk about today. And what they're illustrating here is if you look at different mountain ranges from the Andes to Himalayas or Alps or Pyrenees. What they're illustrating is how as you change latitude but also elevation, you get different ecosystems different types of vegetation. What I'm going to investigate today is kind of the fast forward and then ask the question how do these different types of vegetation actually influence mountain topography. In the spirit of this, this conference we've been exploring this in my group from looking at extreme events to seasonal variations to millennial timescale variations and what I'm going to focus on today is just the millennial timescale variations and vegetation and erosion that occur. And I'm going to minimize the number of equations I show and focus more on the model coupling that we've had to do. In addition, we've been doing a lot of observations to kind of test these models but I won't have time to go into these today and would be happy to address some of these questions. So if we look at a beautiful mountain range like one of my favorite ones here in Patagonia. We commonly think about like how these, all these landscapes evolve and the conventional wisdom now is that we have this interaction between tectonics climate paleo climate and surface processes. In the last decades, we made a lot of progress and trying to understand this and understanding the complexity of this. But what's a little bit different and oftentimes forgotten is especially if you come from a tectonics background such as me is that you have all this vegetation that's acting on top of the surface here as well. And these are just incredible little beasts here they are biogeochemical reactors that are weathering rock, get the nutrients they need. And in addition to that there's also tremendous interactions with the Earth's surface in terms of their stems obstructing flow increasing surface roughness, and those types of things. So that's what we want to look into here. Now, I'm of course not the first person to be interested in this problem. Others is very seminal work here by Longbine and show them in 1950s. And what they found is that if you look at different rivers around the world and the effective precipitation and the sediment shield or you can think of it as the erosion rate of these mountain ranges. What you see is this kind of bi-directional nonlinear response and they attributed this to different ecosystems whether you're in deserts grasslands or forests here. Pretty clear makes sense more vegetation less erosion. In this following this, a lot of other people have tried to reproduce these measurements. And there was a nice summary of this by Cliff Rebe. And what you're looking at again is very similar plot average precipitation versus Hetschman average erosion rates. And each one of these lines is a different study, and you get a bowl of spaghetti basically. Now, I have an answer for why this looks this way unfortunately I can't get to it but what it raises is the question that we may not actually understand how this system works. And it raises the question then of do we actually consider biota in this kind of process coupling here as being significant if we're thinking about the long term evolution of mountain and upland landscapes here. So what I'm going to do in this talk today is test this hypothesis here that which is basically the long bind and show hypothesis that vegetation influences erosion by having a bi-directional effect depending on the climate or ecological zone that you're looking at. So the remainder of this talk I'm going to focus on these two points here I'm going to spend a lot of time kind of illustrating to you what took us the most time, which is a coupled modeling approach to address this problem. And then I'll conclude with a short little example from some work we have in progress on how this varies over millennial timescales. Okay, so this is the system we want to try and understand. So what we basically done is set up a series of coupled models that reflect each of these different boxes here so I'm going to walk you through these. The goal then is to understand how landscapes will evolve like this. Now these models I'm going to be showing you we've tuned them to South America to the coastal Chile area, but everything I'm doing is very generic landscapes trying to understand the process interactions that are occurring there. So let's start with the paleoclimate model. What we're doing is running general circulation models and we've done two different approaches here. If you're not familiar with this so we're running global scale models, these are discretized in space and in elevation. And then you have boundary conditions that reflect things such as the land surface covers the surface temperatures greenhouse gases. Okay, so the two approaches we've done here are to look at to use first of all, trace 21 K experiment. This is a beautiful experiment it's a continuous simulation from the last glacial maximum to present. The downside is that it's very coarse resolution. I've augmented that with high resolution simulations we've done with the outcome five model by high resolution I mean 80 by 80 kilometers. And we've done this at discrete time steps that you see here with the idea of checking how how valid the results are then from from these courses simulations and in fact they work quite well. To give you an example of what's going on here so this is what you're looking at the difference in mean monthly precipitation, so differencing between the LGM and pre industrial times. And what you see going on here these flashing blue and red colors actually represent precipitation differences on the order of 100 to 500% that are occurring between individual months. Now a company with this you of course also have a large temperature changes and also large co2 changes, all of which make a difference for how vegetation will evolve over this, this time period. The next step then is vegetation. And the way we've dealt with this, and similar to to paleo climate, we can't really rely on proxies because they're too disparate in space and in time to get a continuous understanding of how vegetation or climate changes. So again we've resorted to numerical modeling for this. What we've done is coupled a dynamic vegetation model with the climate and landscape model. And the model we're using is called LPJ guests this is kind of a state of the art dynamic vegetation model. It functions. You can see the climate and co2 levels, and it calculates basically plant physiological responses to those changes, and additionally looks at competition between different plant functional types. So what you see on the right hand side here everything is is parameterized in terms of a plant functional type. So, if you're not familiar with what that is this would be like grasses tree shrubs, and you typically define about 20 different plant functional types. What's shown here in the upper plot is that you have some environmental space where different species would thrive. So this is the performance within an individual niche. And then when you add in competition and how these things interact what you get as a realized niche, which is that under certain environmental conditions you would get one plant functional type more dominant than the other. So we get kind of a stand a patch here that gives you a different composition or relative abundances of things like plants. Sorry, grass shrubs and trees. So the way this kind of links with the climate model is that we start with course resolution trace 21 case simulation here with different time steps. Okay. So we have a topographic surface which I'll get to in just a minute here, but we downscale that high resolution or sorry low resolution paleo climate simulation to higher resolution over this landscape here. And then we classify this landscape in terms of different kind of ecosystem zones where different plant behaviors might occur so for example valleys, he'll slopes and versus ridges here. What's then is that after a suite of coupled simulations like that you get in in LPJ guests you get a prediction of these relative abundances of these things. So I want to give you an example of how this the results from this and how this works. I'm not looking at the topography yet and to do this, we're going to go to postal courtier of Chile. This is the first shape study areas which I'm the co director for. And this is a large German Chilean research initiative that's just coming to a close here. So we have our series of observatories that I set up, and these go from the Atacama Desert in the north, which you see over here, all the way down to temperate rainforest so you're looking at an extreme climate and vegetation gradient. And the cool thing you might be wondering why look at the Atacama. Well this is actually the control case this is a world with very very little vegetation to compare to. As an example, what you see shown here is the vegetation total vegetation cover that we predicted for Chile. The numbers refer to the study areas I just mentioned to. And this is a model prediction that was produced from the 21 K or trace 21 K climate history from LGM to present. And we've compared this with present day vegetation cover from, for example, modus data, and they agree very well. And furthermore, in addition, if you look at individual plant functional types like grasses shrubs trees. It's capturing the relative abundances of those with latitude so we have a fair amount of confidence in how that works. Now at any different point on on this type of diagram here. And look at it in more detail. And I'll show you just the southern most one because this has the largest amount of change compared to the Atacama. So this would be an example here what's on the x axis is time before present from 21,000 years ago to today. The lower most panel here is the fraction of plant cover. And so please ignore the black line and something else. So this is the vegetation cover for different types of vegetation. And what you see here is that the total amount of vegetation cover doesn't change that much. You get some oscillations down here but it's on the order of 1020% change in vegetation cover. What does change a lot is what you see in the top plot here. So PFT is plant functional type and what you're looking at is the leaf area index of different plant functional types up here. And I just want to draw your attention to as you as you move towards the Holocene here you start to get a lot more of this dark green showing up here. And that corresponds to temperate broadleaf evergreen trees becoming more dominant as you move towards the present. So the key point here is total vegetation cover doesn't change so much, but actually what's growing there is changing a lot, which means that if we're interested in how vegetation might influence landscape processes we need to be thinking about individual plant functional types like that. Okay, this then brings me to the last part of this where I want to give you an application and looking at millennial timescale vegetation effects. So what we have here is, I mean this audience is fairly familiar with landscape evolution models, but the idea is we're simulating kind of generic plots of land here. We have different study areas we have a relatively uniform rock uplift rate between all of them. We take the climate history from this trace 21 K data, but and that feeds in and gives us our vegetation history through LPJ gas. And then what we're doing is calculating fluvial and hill slope erosion as a function of different amounts of vegetation there. And then that is we're using the land lab space model. And the point here is that this, if you essentially were parameterizing things like the entrainment of sediment or the entrainment of rock as a function of different abundances of the different plant functional types here. So let's get into this just a little bit more. You know the fundamental equation we're solving here is conservation of mass so we're calculating an erosion rate versus time at different points on the landscape. That's a function of the tectonic rock uplift. And then we have the hill slope erosion, removing this and so what we're doing here is we're essentially following the approach of this down below and brass, where the hill slope diffusivity is a function of the vegetation cover. The fluvial or overland flow component here. The erosion and sedimentation are controlled by demanding roughness number of variations between grass shrubs and trees, and we wait those numbers based on the amount of those different plant functional types on any individual cell there. Okay. So the coupling of this is was not trivial. You're doing synthetic experiments like this you always need to make sure your initial condition is an influencing your solution that you're looking at so what we do is a spin up phase. And in this case it takes about 40 million years to get an equilibrium landscape at the slow tectonic rates. So we run that a simulation with that with a fixed LGM climate. Okay, then the landscape is essentially an equilibrium after that point. And then we go into our last 20,000 years where we're imposing transient climate climate, excuse me. Climate and vegetation conditions in the models here so LPJ guests gets a climate for that time step, calculates vegetation cover variations across the landscape that feeds in the land lab. Get erosion changes our landscape that changed landscape and change soil thickness feeds back into LPJ guests and we just keep going here. Okay, so what do you get from this well. I'll give you two examples here, starting with the arid at a comma region. Okay, and then the temperate south location here. Again time on the horizontal axis from LGM to present. The red line is mean annual temperature which doesn't matter so much for the erosion but matters a lot for the plants. And then what shown here on the left side is the vegetation cover or grasses shrubs and trees after amalgamating different plant functional types together. Now, what you see is okay there's variability in the vegetation cover but there's not a lot of it there, which is in a surprise. We've actually at a comma region has been been similar to today for over 19 million years. However, if you go to the south. What you see is the effect I illustrated before in the previous plot this dark green line is the trees here. And you see as you move towards present here you get a higher abundance of trees, lower abundance of shrubs and grass, and shrubs stays quite low as well. Okay, from this then after you go through the kind of the coupling with with land lab. What I show on the right hand side then are the average erosion rates over the entire model domain for that time history. Again in the arid north here. I'll point out this dash line that runs through the middle here this is mean erosion rate. And this is actually defined by tectonics. Okay. And what we're seeing in the north then is you get this high frequency variation in catchment erosion rates. It's basically oscillating around that mean value there. In the south it's a little bit different it's not oscillating so clearly around that mean here and you see some shifts either higher or lower from that. What this implies then is that these these landscapes are actually constantly in a state of transients from this forcing here and the magnitude of these changes in these transients is on the order of 10 to 10 to 25% here. That's important because a lot of the proxy records that we look at. For example from cosmogenic radio nuclides or sediment inventories over these time scales. We may not actually be capturing what we think is the rate of tectonic uplift or rock rock uplift of the landscape through these. The natural question though is well with this variability you see an erosion rates how much of this is actually driven by the vegetation change versus the climate change. For example in the south here you notice that the precipitation history has a similar shape to that as the as the erosion rates here so what's driving what. Well we've done a lot of different sensitivity tests on this with isolating fixing different parameters and I don't have time to show all that so what I'll show you is just a very simplified perspective of this where we look at the over those time series we calculate the Pearson correlation between erosion rate, which is in the left hand column here and different parameters so MAP is mean annual precipitation. And then down here we have the correlation with the abundance of these three main plant functional types and veg here represents the total vegetation cover the landscape. So what you see for the air at a common region is that you have a moderate correlation between precipitation and and the erosion rates, and you have a very weak to know correlation between vegetation and and erosion rates. So this would be very similar to the left hand side of the long bind shown curve, we have very little vegetation and the more it rains the more erosion you got. The situation is different in the temperate south and I'm sorry I can't show you the whole sequence of areas to illustrate how this changes as you move from north to south. But what I want to illustrate is that again if you're looking there, okay you have a high positive correlation with erosion and precipitation in the temperate south. The more it rains, the more it erodes, okay make sense. But if we look at the vegetation covers here. And in particular the total vegetation cover, you have a very strong inverse correlation so more vegetation, less erosion which is also what we see in the long bind shown curve that I showed at the beginning. Okay, so very different behavior. So let's come back to our hypothesis here then does vegetation in induce this kind of bidirectional effect on erosion. And the short answer is yes it does observations which I haven't been able to show you, but and modeling work that we've done shows that different amounts of vegetation can have a positive negative, or no correlation with erosion and can kind of cause this bidirectional to be predicted, or was suggested by long bond and shown. The key thing here is that arid regions are more sensitive to precipitation and temperate regions are sensitive to both precipitation and vegetation, but there's a stronger back vegetation in those types of settings there. The last thing I want to highlight is that you always have a state of transient within catchments on millennial time scales when you're looking at vegetation and climate change. These changes in erosion rates around the mean will be on the order of 10 to 25%. So you have this has a lot of implications for observations as I mentioned, but it's important to realize that tectonics is still what is defining the mean rate of erosion in these catchments. So, that is all I have. Thank you. I need a drink. Time for a question. This is, this is probably a really naive question but how much tectonic uplift do you get over 20,000 years. Over 20,000. Well, time depends where you are but well in that system in that system, 20,000 times 0.05 millimeters per year. So, yeah, thank you. Yeah. But it's so what I'm not sure we're getting out with that, but it's, oh. I mean, over, over, you know, 40 million years, the relief in these landscapes that kind of equilibrate set around 800 meters to a kilometer topographic relief. So, when you take into account tectonic uplift and the erosion, spot it down. So one more question or I should hold away there. Tamara in the back there. Faster mark faster. That was super cool to see after reading the paper. I was curious about them. When you show the plot between the company in the area and the south and the south has this like a step that matched the shape of the precipitation changes, but they already had these like wiggling around the average. Did that blood also had a precipitation or what do you think are the reasons of those changes in the erosion rate in the cell in the region in the air region. The, it's a good point. So it's mostly coming from precipitation. There's a very weak, weak correlation with vegetation there. So, and it turns out. If you move a little bit south from there and start adding a little bit more shrubs, the response starts to change a lot. So it's really is just kind of an end member. It kind of makes sense. I mean, it's at a comma you're going to be most sensitive to stochastic rainfall events coming on the landscape. That's, that's causing most of the variability. The model that you couple for the reconstruction of the paleoclimate. How just like, I don't know how that works, but like how, what information do they use to recreate the paleoclimate. The paleoclimate. Okay. That's an entire field on its own, but you need the mean orbital parameter as it's changed through time. Then you need the greenhouse gas concentrations, which we have good understanding of. You need land surface change. And so you need some sort of soil. And that's usually assumed to be similar to today's soil. That's very important because the soil moisture is a strong effect on the climate. And then in those particular simulations, they also had a dynamic very course dynamic vegetation model to influence the land cover. And then the final really important thing is the sea surface temperatures. And those are those are calculated. That's a couple of ocean atmosphere model. And there are proxy data that help constrain it, but then it's dynamically updating what the sea surface temperatures are as you move towards present. Thank you.