 Ah, hello. Yeah, it's the landscape near where I live here. So I'm really sorry not to be able to be present physically in Trieste. I wanted to join you. And then one of my flights was canceled. And it was too complicated to organize and other travel. But anyway, I enjoy the talks. And I will now switch to my share of my screen. I have a problem with sharing my screen. You know, these new computers are too safe now. And it becomes really complicated. OK, we see your slides. Can you see my slides now? Yes, yes. OK, great, great, great. However, it is in presenter mode. So we see both slides at the next slide. That's not what I had on my screen. Wait, wait, wait. This is really, yeah, maybe. Sorry, I'm going to change my video. Yeah, you can try to stop to not stop the sharing screen, but remove the presenter mode. Yeah, now try to do. Yeah, he's using Zoom. OK. Oh, yeah, now it's OK. And now it's OK? OK, good. Sorry about that. OK, so my presentation is about the long-term dynamics and sustainability of human nature interactions. So it's a new research project that I've developed during the last 10 years or so. You must know that I'm an ecologist, a theoretical ecologist. So I mostly studied so far changes in biodiversity. It's the left graph. Changes in biodiversity, how they affect this ecosystem functioning, and in term ecosystem functioning, possibly even human well-being. But actually when we made a synthesis of that research area, what we showed in that particular paper that I'm showing this figure now, is that there was a loop, actually, that needs to be closed between what humans are doing, which is in yellow at the bottom. And then the ecological dynamics by diversity ecosystems that are in blue. And then in between, we have the anthropogenic drivers and ecosystem services. There's also a scale issue, but I'm not going to address it in my presentation here. So I became interested, or before that paper already, in trying to close this loop. And so how can you do that? Well, there are different ways of doing that. The way I personally addressed it is that by trying to extend an approach that was developed by two economists, Brander and Taylor, some time ago, who were interested in why the Easter Island civilization collapsed. And so they developed a simple economic model of Easter Island. And so the interesting of that approach is that it's very simple. So basically, what they did is to build a simple ecological economic model in which you have basically two variables. One is just a natural stock, which was just assumed to have logistic growth for simplicity. Then you have a whole machinery in terms of economies, but which is vastly simplified by assuming that you have a kind of economic equilibrium and maximization of utility, et cetera. So that reduces the system quite a bit. And then you have labor, which is essentially the human population, which grows when harvest grows. And they ended up with a very simple system, which actually turns out to be a Lodz-Kavaltura that's a prey model, where labor, that is the human population, is the predator. And the harvested natural stock is the resource. OK. And so they spent a lot of time. They were really excited about these results. And they showed nice pictures of face planes and time trajectories for these kinds of systems, which incidentally are very well known in ecology already, and they claim that actually since you have a kind of damp oscillation, then this could possibly explain the collapse of the historic civilization. Well, I don't want to debate about that conclusion, but anyway, I felt that the general approach was quite interesting. And so we decided to build other kinds of models with the same philosophy. And so since my interest was on biodiversity changes, and as you know, at least in terrestrial systems, the biggest threat to biodiversity is land use change. So we started building models of a human population converting lands. So you have lands that can be in two states, either natural or converted, and then mostly used for agriculture, which of course has a negative impact on biodiversity. But then we also built in our model a whole layer of economics behind it, which makes it very complicated at first sight. But using the kind of tricks that Prander and Taylor developed, we were able to remove them. They are, in a way, built in the parameters of the model. And so the model actually reduces to three variables. This time, the human population, technology, because it turns out to play a big role, and then biodiversity dynamics. And our main interest with that first model was to look at extinction debt. And I don't know if you know that concept of extinction debt. It's basically the fact that when you destroy natural land currently, you are actually committing species to extinctions in the future. That is, species don't disappear immediately as you convert land. By the way, this is the common mistakes that people do on the radio, et cetera. Even scientists, when they talk about extinctions, they provide estimates of rates of extinctions that are much higher than in reality. Actually, we commit species to extinction, but they are not yet extinct. And so that's the extinction debt. And so the idea is that if you have delayed biodiversity dynamics, which we know happens, we have a lot of evidence for that, then you basically get the first three graphs there, panels, show what would happen for the amount of natural land. That's a gray line. And then for the human population, that's the solid line. And the three panels, say, DC, have increasing extinction debt. And you see that at first, when you don't have extinction debt, then that particular model leads to a stable equilibrium. Then if you increase a bit extinction debt, you start seeing oscillations. And eventually, if you have too large an extinction debt, then you start having a collapse of the human population. So basically, this very simple model already shows that delayed biodiversity dynamics can generate environmental crisis to the point of even leading to the collapse of the human population. We also use the same model to show that it can also make other predictions. For instance, if you start protecting land, well, the fraction of land protected has to be large enough. But then you can stabilize the system. That's with the same parameter values. That's the right panel. So clearly, protection of a large enough fraction of land can ensure sustainability. So that's a good thing because that's predicted by a lot of other approaches. We also check that actually an approach based on taxation of land conversion can work too. So we build a more elaborate economic model showing that a land tax policy that maximizes the discounted sum of benefits in the long run can also ensure sustainability. And moreover, if you see, for instance, well, the graphs when you compare the green and black lines, you see that actually this kind of policy is beneficial for both biodiversity and the other human population. So there's no trade-off here. It's beneficial to both. So something that is obviously missing in these first models is the fact that the human population is not homogeneous and also there are delays in behavioral changes in the human population. So our next step was actually to use the same kind of model but dividing the human population into two groups. It's of course a big simplification, but we define two groups, defectors and conformers. So it means that defectors are, it's a classical evolutionary game. So it's inspired by that. So basically conformers conform to a rule that benefits collective well-being and defectors play their own according to their own interests and so don't obey the rule. What we could show then is that the outcome depends a lot on what we call here the sanction level, but it's not truly a sanction. It's really the relative benefit that the conformers gain by playing the game relative to the defectors. So it can be a positive thing and not just a negative intervention. So what you see here is the human population as a function of the proportion of conformers which can vary from zero to one. And so everything depends on the sanction level. When the sanction level is too weak, what happens is that whatever the initial proportion of conformers, the system moves to the rad equilibrium which is the unsustainable equilibrium. When you start having a larger level of sanction, in between you have the middle graph shows that you have two alternative equilibria, one which is made up of only conformers and the other one with a mixture of the two strategies and only with a very strong level of sanction or relative benefit of conformers, can you get fixation of the conformer strategy? But it also depends on the initial proportion of the two strategies because you have alternative stable states. So basically, this shows that when you have several strategies in the population, it's much more difficult to reach sustainability, which makes a lot of sense. This is compounded by the effect of the extinction depth, of course, because these are the endpoints of the time trajectories. But actually, if you have a large enough extinction depth, then you get a lot of transient dynamics that can be very complicated with a lot of ups and downs. Right. So to move a bit further, we try to build a model where we no longer have, sorry, there you go. So we no longer have economics because the model is a bit constrained by the economic assumptions. But where we could have a bit more details about the ecological processes going on. And so with one of my postdocs, Kirsten Handerstone, we started building some more detailed models. There are just more details in the fact that they have three kind of land uses, A, agriculture, and it's natural. And then D is degraded. But degraded can include urban as well. So it's neither agriculture or natural. And this can include a lot of other processes like restoration, natural regeneration, natural degradation, et cetera, et cetera. So when we parametrized our model with the data on the global trends in land use and human population size. So we basically parametrized the model with data from the period 1960 to 2014. And the right graph panel shows what the model predicts for the trends in the human population and the natural land, agricultural land, and degraded land. And as you see from the blue curve, I mean the predictions match relatively well the demographic predictions for the human population. So it's not completely unrealistic. We don't claim that our model is the best representation of what will happen. But it's at least something that we can play with with a certain confidence. So basically, we use that model to actually play with the various kinds of measures that we could implement. So we define two types of measures. One is proactive and the other reactive. A proactive measure is a measure where you have an intervention before changes happen. So that could be, for instance, avoiding or increasing land degradation. That's the one that we use here. And then a reactive measure can be, OK, given that the natural degrades, then you can restore the land towards the natural state. So that's what we call reactive. And when you vary these two measures, actually the four panels show that you can change a lot because all the green and gray areas are what we can change by changing our behaviors. And you see that it's possible to change human population size and the amount of natural land that is left in the future to a very large extent. But when you look at the last graph, the bottom right graph, that's where you have a negative proactive measure that is fixed. And then you look at when you vary your reactive measures. So you restore a lot of natural land, for instance. But as you can see from that panel, it's really hard to reverse a trend that was negative in the first place. So basically the conclusion that we grow from this exercise is that changes in human behavior obviously can make a difference. But reactive measures are not enough to ensure sustainability. And so this is a really sobering message because it means that what people are doing today will make a big difference for the future. And even in the future, we might not be able to reverse the processes going on now. Well, we know the same with climate change, that it would be very difficult to go back in time once it's started. It's even worse than that because with the same approach, then we build a more complex model where we divided the world into several regions. We're here on two regions. And so it's a model that includes human migration and unequal access to resources. For instance, we divided the world in this particular case into two regions. One is a low income region. You can view this as a kind of developing world. And then a high income region, which you can see. You could see as a developed world. And of course, in time, the situation in the two regions changed and people start moving based on opportunities for, for instance, agricultural production, presence of natural land, et cetera. So the model is a bit more complicated. But the assumptions are relatively simple. So when people can gain more by moving, eventually they start moving. And as you can see, these simple facts that changes the dynamics completely. At the bottom, you have the same model with the same parameters, but you take just an average. So you have a single region where everyone behaves in the same way. And you see that based on current trends. So this is based on the trends from the previous model with a single region. You can see that you move from 2070 to 2050 to a world where you have a lot of urban or degraded land, very little natural land. But overall, the number of people is reduced. And well-being, which is indicated by the color, is quite high. So you have a few happy people in a way. Despite the fact that natural land is destroyed. But with the same parameters, if you have two interacting regions, look at what happens in 2750. In the low-income region, basically, you have overexploitation of the land with a moderate amount of people, but the black color indicates that well-being is extremely low. So it's basically famine. In the high-income region, you basically have the same situation with a bit more people, a bit less famine, but still it's red. So it means well-being very low. So this is a simple exercise to show that an equal access to resources coupled with human migration can make a huge difference. And it can be a big factor undermining global sustainability. Right. So in the second part of my talk, I would like to show you a few results from our more recent work, which is intended to represent something at a smaller scale. Let's say a regional scale, something like that, where we focus on agricultural land use. Because well, the use of land for agriculture is one of the big problems that has been debated a lot in the literature. For instance, there has been a lot of debate about two strategies to exploit the land, land sharing, and land sparing. So in land sharing, the idea is that you exploit a lot of the land, but with a low intensity. And land sparing, you exploit the land on small surfaces, but very intensely to try to save as much of the rest of the land for natural land. So we build a similar kind of model as the previous one, with three states for the land, natural, agricultural, degraded. Then this produces resources which feed the human population, and in turn, the humans decide to convert or not, to intensify or not. And so there's a parameter beta that determines whether you do land sharing or land sparing. Basically, when beta is small on the left, you have low intensity. And when it's high on the right, then you have a high intensity on smaller areas. So that's model, which is built a bit with somewhat different assumptions, actually shows the graph shows the time trajectories of the system. You see that both extremes, land sharing and land sparing, can lead to a stable equilibrium. But in between, you get a complete collapse of the system in red, the degraded land that dominates the whole system. So we looked a bit in more details about what is driving this, but I'm not going to look into the details here. What we can see is that the four panels here show what happens with the varying amount of rates of recovery from the degraded states. So there's a natural regeneration of degraded land. And what you see basically is that at the bottom, when the natural rate of regeneration is very fast, then you avoid the collapse in between. But when it becomes lower, then as you move up, you see that there's a whole range of possibilities where you have a collapse of the system. So I must make sure that you understand that there's a kind of trade-off in this particular scenario between agricultural intensity and land conversion. And that's what creates these different outcomes. So basically, actually, in this particular model, in this particular results, what we assume is a linear trade-off between the land conversion rates and agricultural intensity. In fact, if you look at where there's a collapse in between, you see that there's a whole region that is shown in orange, where the collapse is inevitable. And then there's a whole blue region where there's a viable equilibrium. And so it really depends on how you build your trade-off and where it falls, et cetera. And so our main conclusion here is that naive land use planning can easily drive the social ecological system to a reversible collapse again. It really depends on how the two kinds of behaviors trade-off with each other. And without knowing that in detail, then you can easily go to collapse in the long run. We went a bit further than that using a spatially explicit model. Also, it's a kind of a grid model where, again, you have the same principle with natural land, agricultural land, two types of agricultural lands, actually, where they are low intensity or high intensity, then degraded land, then you have human population. But this time, there's a spatial structure in the system such that the humans can decide to aggregate agriculture, to cluster it. And also, there is a spatially explicit provision of ecosystem services coming from surrounding natural areas. So when you have large natural areas in your vicinity, in the vicinity of a crop, then you get a lot of services. That's basically the general ID. The interesting thing is that by looking at the spatial dynamics of this system, then you get what is called a percolation transition, which is known in my physicists, usually. So what happens is that, and you see that, for instance, in the upper left graph, as you decrease the production of natural land, so there's a population that starts using natural land, converts it. And once you process threshold, which is shown by the vertical gray line, then you start losing your large fragments. This is shown in the upper left graph very suddenly. So that's what is called the percolation. Suddenly, you lose your large fragments. And you can see that at the bottom left, you see that suddenly the number of fragments increase while, at the same time, the fragments become smaller. The problem is that that's what you can see in the middle graph on the top. The mean ecosystem service provision depends a lot on the largest fragment size. And so the end result is that the resource production per unit time, which is shown on the upper right panel, starts to decrease, to drop sharply once you cross the percolation threshold. And what humans do in this situation, well, because they have to feed the human population, then what they do is actually what is shown in the lower right graph. They increase their propensity to expand agriculture. Just after the percolation threshold, you see that suddenly boom, the propensity to expand agriculture increases a lot. So that shows that actually spatial dynamics is quite important. And agricultural expansion can cause a percolation transition that leads to abrupt habitat fragmentation, that leads humans to further aggravate landscape degradation. So that's why you become easily trapped into a kind of collapse situation. And the model shows that actually neither land sharing nor land sparing, as they are traditionally defined, are the best strategies. Actually, the most favorable strategies are the white ones. As it's a darker red, actually, you need a lot of natural land to avoid irreversible collapse. When it's white, it means that you are less threatened. Because if you lose more natural land, you can still be safe. And you see that the best strategy is in the lower right corner. So the best strategy is clearly to prevent severe habitat fragmentation and foster sustainability, is a combination of high agricultural pressuring, which is on the right, and low intensification, which is along the vertical axis. So it provides some new perspective on this land sharing. OK, I still have a bit of time, so I wanted to show you still the same kind of work on agricultural landscapes. But it's less about the feedback between humans and biodiversity or land conversion. And by diversity or land conversion, it's here more, it's at an even lower scale, trying to reconcile biodiversity conservation and food production in these landscapes. So we are really at the landscape level this time. And our idea with this particular approach was to look at pollination as the main service provided by natural fragments. So basically, you have a system, which is quite classical, where you have crops. That depends or not, the amount of dependence can change on pollination for its production. So the first graph panel A shows that as you increase the number of natural, semi-natural habitats, then you increase biodiversity. So that's normal. That's what we expect from what we know on ecology. Then if you look at pollination, pollination is actually the amount of crop that is pollinated. So it's a kind of service that humans receive from pollination. And so you see that actually it has a home shape with a bell shaped curve with the amount of semi-natural habitat, which makes sense. Because when you have no semi-natural habitat, you have no pollination. And when you have a lot, then there's a lot of pollinators. But actually, there's not a lot of crops. And so obviously, your production goes down. So that's basically what the basic model shows, which makes a lot of sense. And then of course, you can also have an independent crop yield, which doesn't depend on pollination. And that one, of course, will gradually decrease with the amount of semi-natural habitat. So everything depends in the system how much of your crops are pollinated and how much are not pollinated. And so we build a spatially explicit model again to look at the effect of fragmentation. I'm not going to go into the details of this, but we were able to derive a kind of measure, which we call landscape pollination potential that actually captures all the spatial aspect of the system. And what we could show here is we basically get the same patterns as before. So the main trends are identical. Simply, when you have more efficient pollinators or lower level of fragmentation, because both increase this aggregated measure for the maximum amount of pollination possible over the landscape, basically, you move up along the curves that are shown here. And so basically, you increase the magnitude, which is the panel A, or the stability also of animal-dependent production, so based on pollination, and also your crop yield area. So that's what we can show with that kind of model. The interesting thing then is to show the biodiversity effects more explicitly in this kind of model. And so in this graph, we have exactly the same kind of shape, of course, or animal-dependent crop production, then the stability of this production yield per area. And you see that biodiversity effects are turned off or on. And you clearly see that when you add effects of biodiversity on pollinator carrying capacity and a kind of insurance effect on pollination, then you have a clear benefit from biodiversity in this kind of system. So the effect of biodiversity increases the magnitude and stability of all these variables. OK, yeah. Maybe I'll skip this because I want to save a bit of time. OK, before going to the last two slides, a few conclusions that emerged from all this work, I think, relatively clearly, is that linking biodiversity ecosystems and people is critical to understand and predict the long-term dynamics of the global or regional social and ecological system. Feedbacks between humans and biodiversity play a key role in sustainability. So I think this is generally underestimated when we talk about global change. We mostly talk about climate change. But I think that the dynamics of biodiversity in the long run is key. Collapse of the global social-ecological system is likely in the absence of proactive measures that are deployed by biodiversity laws. So what I mean by that is that it's a common outcome, whatever the kind of model that we build. So it doesn't depale on fine-tune details of the model. Delayed responses and unequal access to resources are two major factors that undermine sustainability. So we have to take that into account. We've just seen that native land use planning can easily drive these systems to collapse also. I didn't show that, but reconciliation of biodiversity conservation and food production is possible, but requires a big shift in agricultural management. So all that is, some will say, pretty pessimistic, because it shows that all these kind of difficulties and collapse are quite likely. But I think it's, unfortunately, quite realistic. That's the situation we are facing currently. And so my last two slides are about what we can do about this. So actually, we are convinced that changing our worldview is necessary if we want to avoid these long-term problems. And if we really want to build sustainability, then we have to change our worldview. And basically, we compare here. It's a very simplistic picture, of course. Basically, on the left, you have the current situation when we have a disconnection between humans and nature that leads to unsustainable norms and values, behaviors and policies, which leads to extinction of experience. So we no longer are in touch with nature, so it continues even further. And so it's a kind of vicious circle. And we should shift towards a more sustainable vision where we have contact with nature. We have sustainable norms and values, behaviors, and policies, and that can lead to a verters cycle. And so in this particular study, what we wanted to see is whether simply changing human nature connectedness can it make a difference? And actually, it does. So it's a recent paper that was led by a psychologist where we made a meta-analysis of many studies, more than 200 studies, either experimental or correlational. And these studies have looked at how human nature connectedness is increased or decreased and what is its possible impact. And as you see on the top, the experimental studies show that exposure to nature, that's on the left, and mindfulness practices improve human nature connectedness. So mindfulness is, for instance, on the right panel. And you see that clearly it increases the human nature connectedness. In these experimental studies, these were controlled experiments, so we know the causality. So we know that these two kinds of manipulations increase human nature connectedness, and there's some quantitative way to measure it. At the bottom of the graph are correlational studies. And you see here causality is unknown, of course. But you see pretty much the same kind of patterns. So these correlational studies confirm the experimental results. If you change your contact with real nature and mindfulness, that's on the bottom left, you see that you have a clearly positive outcome on human nature connectedness or the reverse. We don't know causality. But also human nature connectedness is positively linked to nature conservation and human welfare. So they are not in a position to each other. But it's negatively correlated to non-environmental values. That's what is seen in Brown at the bottom. You see that the effects are systematically negative, which means that as you decrease human nature connectedness, then you basically have a less favorable behavior towards the environment. And that's it. I will stop here, except mentioning my main collaborators on these projects. Ansofiles Fouix, Kirsten Henderson, Diego Vengucio Paz, Daniel Montoya. And the last work is by Gladys Baraggan-Jazzon, who comes from Experimental Psychology. Thank you very much. OK, thank you. I have two questions. So what is the, so what I understand that you have presented are essentially theoretical models. And one question is, is there a empirical support for this? I mean, how do we stand in terms of validating these models or these insights from models in real data? That's the traditional question about modeling exercises. I'm not sure that you want to reproduce your system and manipulate the system to have an experimental setup, but that's not really possible. But simply we can maybe have some small scale data that we could possibly fit to. But currently, the only thing that we can do is based on the current trends. We know the trends. We know the reasons. And our models are basically there to show what is the potential futures. It doesn't tell that it will happen, and hopefully it won't happen. That's our hope. But we need to know what is the potential futures and what we can do about it. But it's very difficult to have a full validation of any of these models unless you perform some experiments on relatively small scales. So some of these things can be done on small scales. For instance, some of my friends in France are manipulating, are doing experiments with some farmers and trying to change their behaviors and they accept to play the game. And so they see the outcome after many years. And so this is something that can be done, but it will never solve the problem of validation at the planetary scale, which is impossible. I mean, the only way to validate it is to see what will happen in the future. But that's not a very good way to do it because it might be very negative. Thanks. We have another question. Yeah, very interesting. It's not my field, so I might ask a very naive stupid question, but here we go. At the start, you showed this, let's say, few happy people scenario that have found very confusing and counterintuitive. Because you start, can you hear me, sorry? Yes, yes. We start showing that you've got, let's say, seven little people in a box. They live in a world which is mixed between concrete, forests, a bit of everything. And then you go towards the end of the seventh century, I think, on the scale. And all of a sudden you end up with one, let's say, one little man out of seven. And then it's still full of concrete and it's still happy for some reason. I found this very counterintuitive because is there a feedback between how you change the human population and how you maintain the amount of land use or concrete? It's basically the Chernobyl example. If you remove humans from Chernobyl, it takes 30, 40 years for the city to become a forest. It's back to a forested area, more or less. So I don't know. If you can comment on that, this few happy people scenario sounds very counterintuitive to me. Because if you wipe out, let's say, one seventh of the population, there would not be enough humans to maintain the amount of concrete in the cities and things would collapse and there would be more wildlife taking place. That sort of thing. Something I found, sorry, counterintuitive. I might be wrong. I might have missed something. OK, yeah, that's a good question. It all depends on the long-term dynamics of the system and remember that this is transient dynamics. Even in 2750, it's still transient dynamics because if you let the system continue for millennia, then you will see another pattern and you will probably see what you expect. But the problem is that, well, the model includes a lot of different things, in particular, the effect of technology. And so you can still imagine a society where you have a lot of technology, few little natural land left, and still people who are relatively well off. I mean, they don't suffer from famine and things like that. So that's a counterintuitive outcome, but it's quite possible within the framework of the model. And of course, in the long run, it will refer to something that has more nature, probably. But not on that time scale. Thanks. So we have joined. You want to ask a question? Yes. May I ask a question, please? Yes, go ahead. In an early slide, you showed an equation for the growth of technology. And if I saw it correctly as it went by, in that differential equation, there was a self-limiting factor, like the logistic factor for population growth. I believe there was a maximum, and as technology increased, the rate of increase of technology decreased. I don't know if I saw that. OK, I did see that right. Well, no, you are absolutely right. I didn't have the time to. Please let me finish my question. I have two questions. One is, what happens if you remove that constraint and whether that constraint is justified in the historical record? And second, does it make sense to talk about technology as a single quantity when some technologies have adverse environmental impacts and other technologies have very favorable environmental impacts, for example, the use of information to guide agricultural activities to plow in the right places. It's all the benefits that come from informational technology. So what happens if there's not that limitation, and two, what happens if you differentiate technology according to its impacts? Thank you. OK, so yeah, thanks for the question. It's a complicated story. So in that particular model, which was our first model, we indeed implemented a kind of logistic growth for technology with the idea that, well, through time, technology don't necessarily increase until it's not infinite, let's say, the growth. And so it will stabilize somewhere. Why? That's a debatable assumption, of course. You can also have other scenarios where it never stops. So in other publications, we have removed that assumption. And it turns out that actually the dynamics of technology plays a really big role. And regarding your second question, that also plays a big role, what kind of technology, where it goes, for what, et cetera. So we've played with that in another paper that I didn't present in my presentation. And it can change a lot of things. So the model predictions are very sensitive to two things, actually. The first one is how technology is included and what are its impacts. And the other thing that is not in this particular model, it's not explicit at least, is the highly nonlinear relationship between human fertility and technology and resource access. And so we've explored that a bit more in other papers. And that can lead to very different outcomes. The problem is that many of these outcomes are not necessarily positive, contrary to what you could believe. So it can go all over the place, basically. So our basic predictions are valid for these specific assumptions, but it's possible to have other futures as well. Thank you. Michel, I have a very simple question for you. You mentioned that some naive land planning could lead to collapse. Could you elaborate what you mean by naive or give examples? Well, the example in this particular model, the model about agricultural land use was actually inspired by some real data, especially from South America, where some people have formulated the hypothesis that actually we could be near a percolation transition in South America because of land degradation. And so that's actually, that's where our work started. And so the naive land planning issue arises from the fact that there is no simple relationship between intensification or despairing versus sharing perspectives. And so if you are stuck in between in such a way that you don't have the right trade-off, then you suddenly go to collapse. And that's what we meant. It doesn't mean that necessarily we will go there, but simply if we ignore this complexity and the shape of this trade-off, well, we just don't know. That's the main idea. It's not that necessarily it will necessarily go there. OK, thank you. So I think there are no more questions. So we thank you again. And then we reconvene at a quarter past four. Thank you.