 from the NASA Goddard Institute for Space Studies in New York. And this session today is going to be on agricultural impacts and the climate information that we need. And we have three speakers today. I'm going to get started with an overview and bring in some of our approaches from IPCC and AgMIP. And then we're going to hear from Benjamin Sultan and Edmund Toten later in the hour and a half we have, or hour and 45 minutes that we have here. So I think I'm just going to get going on this. And we'll probably have questions depending on the time frame. We're hoping to have questions at the very end. And then we'll see how we're doing in the transitions between if there are quick clarifications. All right, so I wanted to start by talking about what I call the quadruple challenge for agriculture. And this is that the world is asking the agricultural sector to sustainably increase production, to provide healthy food for growing and developing populations. It's not just about feeding who we have today. It's about feeding the growing populations and recognizing that as countries develop, they're often asking for different types and quantities of food. There's also new kind of efforts for nutrition that are changing the types of food that are recommended even in the richest places in the world. We also are asking the agricultural sector to adapt to climate change and ongoing climate extremes, which are already upon us, while also mitigating emissions from agricultural lands, most importantly, from livestock and from rice methane, as well as the overuse of fertilizer in some places. And then also, none of this will be possible if we can't maintain some kind of financial incentive for agriculture that supports so many regional economies. So any one of these challenges is substantial. And when you put all four together at the same time, you understand the challenge that faces the food sector. So what I'm going to talk about today is an inventory of agricultural responses to climatic impact drivers, as well as that term, climatic impact drivers, CIDs being a core element of approaches that we take within the agricultural model intercomparison and improvement project, or AGWIP, which I'll introduce. Then I'm going to talk about how we build scenarios of future agricultural systems and wrap it up with some key priorities for agricultural risk information development. So this is a table that you've seen already several times this week, where we have taken the inventory of climatic impact drivers that we developed within the IPCC working group one. You can see this in chapter 12, table 12.2. And just as a quick reminder, these are the major types of climate conditions that we know drive responses in the things that we care about. And there's a whole table here with many rows. But I pulled out the crop system row, because this is where I'm going to focus much of my attention today. And I'll also note later in the table, there is a row for agricultural lands, which is a slightly different row. So this one is really about when there are crops in the field, what affects them. Whereas agricultural lands are questions around large scale, where do we grow agriculture? And is it viable? So for example, you'll notice the most obvious example of this is right here, where plants that are in the field can get affected by coastal flooding. But usually, if the relative sea level has already risen and taken over your field, you're not likely to plant. So that is more of a lands issue than a specific crop that might be in the field. But the idea behind this table is that where you see darker colors are places where we saw in the literature that people are using this climate information to determine some kind of a response. Usually, it's because they've observed it. And they have some ability to think about the projections or how we go forward. So you'll notice that there are many, many different climatic impact drivers that affect agriculture. And almost every single one has a unique pathway by which the biophysical processes are affected and unique types of metrics and indices that we have to provide as climate scientists to enable that. So I wanted to go through some of these and talk about how we do it. But before I do that, I'm going to give some context of where I'm coming from. So I am the science coordinator and the climate team leader for the Agricultural Model Intercom Parasite Improvement Project. This is AgMip. I'll stop saying the full name and just start saying AgMip from now on. But AgMip is designed very specifically and intentionally to be a lot like CMIP, the coupled Model Intercom Parasite Project that we use for climate models. We want to do the same thing for agriculture. So we saw that the agricultural modeling field was decades behind the climate modeling field when it came to systematic, inner comparison, community, engagement, and collaborations across modeling groups so that we could have direct ways of figuring out what we can and can't model and how we can prioritize and achieve model improvements. We call this our flower diagram that shows major areas in which we have model intercomparison projects. And every single petal of this is a major kind of category. And then each row is a specific set of protocol-based activities using multiple models and multiple institutions in multiple countries. And I could talk all day about what this looks like. But what I do want to say is, since we launched in 2010, this diagram has gotten more and more complete. We're now actually going off the bottom of the screen here. Vietnam is only half on here, unfortunately. But we're adding more regions. We're also working on a bunch of different scales. So I'm just going to call out the largest categories here. Let's see if I can get the pointer to work. So we work, first of all, on the global scale. So we try to take everything together and build assessments of global economics and food prices and food trade. But on the opposite side, we go all the way down to the experiment model interface, where we have entire projects based on the crop water evapotranspiration, so the fundamental physical processes in one field, trying to get that right on the flux towers, et cetera. We work not only on the global scale, but we work in the regional areas, where we try to put all of our elements together and make the best support for adaptation and risk management planning. We also have coordinated efforts to make large-scale gridded models of crops. And while we're doing all of this, of course, we're building up data and tools, looking at interactions with things like water resources and livestock. And we have all kinds of cross-cutting themes, and I'll talk about them in a little while. But maybe what Agrib is most famous for is these crop model intercomparisons, where we have teams that are fundamentally focused on each species. So we have a wheat team that has compared more than 40 different wheat models. And you can see the other ones here. I don't think anybody stands up to the wheat team, but maize has more than 20 models, rice has about 18, and you can see the diversity of what we do. All right, yeah, and then overall Agnep is this community. We've got more than 1,200 people around the world, and it really is designed to be interdisciplinary. So I have to acknowledge first that we are super proud in the Agnep community that one of the Agnep founders, Cynthia Rosenzweig, was named the 2022 World Food Prize winner. For those who don't know, this is like the Nobel Prize for food systems. And the reason I especially wanted to thank Cynthia for this is that Cynthia has actually donated a portion of this prize to this workshop to help some of you come here. This is the type of thing Cynthia, rather than taking this money and going on a boat trip somewhere, she has funneled it back into Agnep and is now funding conferences and workshops, even in small pieces, to the best that she can. All right, so the way that we think about Agnep is really that we can't understand and prepare for the big food system challenges I mentioned unless we recognize the complexity of the food system. So here is one representation of the food system. You can see there's lots of details, but the main thing to recognize is that it is not just what happens in climate. We have to understand the fundamental biology and how it interacts with climate, but we also have to recognize that food interacts with a very dynamic economic system and a political system that can interfere or enable and that's something that we have to recognize. So there are all kinds of things that we've done in Agnep to kind of build up our understanding of each element and the way they interact. But if we kind of step a big pace backwards, we can talk about how the agricultural models fundamentally are trying to understand how the world responds when there are ships in the genotype of seeds. So that's like the genetics and the seed selection. The environment itself, that's where climate change comes in. Management of what farmers do, that includes many adaptation types. And then value change because it's not just what happens on the field after you harvest and move past the farm gate, there is a whole chain of production. Could be just local markets and it could also be going through processing plants and shipped around the world. So we talk a lot about G by E by M, that's a little bit insider speak, but this is the way that the genetics and the environment and the management interact. And then I've for the first time in this presentation added the value chains on top of that. So one way we think about agricultural modeling is that of course we're fundamentally trying to represent hazards and disasters. This means that it's not just a climate change, it's all time scales. But we're gonna build up a figure here. And the first thing to note is that we fundamentally have to understand how crops and the agricultural system respond to things like extremes in temperature both on the hot and the cold side. Rainfall extremes from droughts to floods, air pollution, pests and diseases, cyclones and extreme storms. And then this plus sign in the here is doing a lot of work that covers the rest of that CID table. We know there's a lot more. But of course, if we can fundamentally model these responses, we can follow this time scale at the bottom here. We can monitor during a season up to present day represented by this line. And we can forecast with our climate models and our crop models or our seasonal forecast weather models. And then of course, we're very fundamentally interested in the lead time because that tells us how we can react and what types of interventions we can be reactive as we see disasters unfolding. Of course, we can take these same models and go backwards in time. And this gives us understanding as we fundamentally look at things like detection and attribution and counterfactual management, which is another way of saying, if we had known that that drought was coming, could we have done something differently? And we can model that because models allow us to go beyond the observed experience and test out different interventions. So between these two things, we can develop understanding and you'll notice, of course, that it also points to certain types of climate information. So we might wanna look at historical observations, retrospective analysis, data assimilation type systems. Of course, if we go far enough on this timeline here, we're really not forecasting anymore. There's this wiggly area. And then we're really talking about projections. This is because of course, the future depends not just on the initial conditions of what we see today, but on decisions that we make as a general society around climate change, the way that the markets are shifting policies, socioeconomic change and larger questions of environmental sustainability. So once we're out here, we need scenarios so that we can project and understand. And this of course is super valuable because we have to get to this proactive set of interventions that recognize non-stationarity and emerging challenges. It's not just the same as the past. We need to get to the future. So one of our more famous set of projections are these global yield projections that we've made. These were done in association with the EasyMip project, the Intersectoral Impacts Model Intercomparison project where AgNep runs the agricultural sector. And what you see here in the bottom left are maize yield projections. And in the upper right are wheat yield projections. And you'll see that they're going largely in different directions. The maize yield is getting orange. That means lowered yield. And the wheat is in some places getting green. So if we want to understand what's happening here, we need to figure out the fundamental responses. And if I can pause it. Oh, no back. It looks like I'm not gonna be able to, hold on a second. What I have to do is I have to play it again. So I'm gonna talk through just a couple things that's happening. So while you're watching this, you'll notice that the maize yield is particularly detrimentally affected in the tropical region. So places that are already near the hot thresholds are being most strongly affected. Wheat yield you'll see in some places, especially where rainfall is increasing and where it's currently cool, a little bit more temperature can actually be helpful. And then of course we have the carbon dioxide effect, which is generally going up and helping many agricultural regions. But notice that there are exceptions. Parts of Southern Canada, Pakistan, parts of India, Bangladesh, these places are also being negatively affected as well as Mexico and Southern US. So again, there's a little bit more danger in the tropical regions. But as a cool season, cool climate zone crop, wheat is faring better than maize. So this is also very strong message coming out of the agricultural sector, which is that you can't talk about crops in general. You have to talk about systems and regions. So this figure or this study was actually featured in the synthesis report figure SPM3C, which is the first time that we've ever had anything like an impact map in the synthesis report, summary for policy makers. So this is actually a really big step that we had this map. You've already seen some of the other ones on ecosystems and human health. There's a figure below on fisheries, which I won't get to. But this was a big battle in the synthesis report approval session to get these figures through. But for the first time we do have these types of projections that we have, not just the projections, but you'll see the hatching. That's the uncertainty across the models, both climate and crop. And there's a lot we can do with this. Yes. The health one? No, so there was a different battle. The original figures that we had had both maize and wheat. And we were trying to present just like I did before. There's some positive, some negative, and the system is complicated. But in the end, there was just too many figures. So they said we can only have one. And when we put the hatch marks, when you put the uncertainty on the wheat map, it was much more uncertain, which reflects the wheat team's understanding as well. So we decided it would be easier to have one row here. And this was approved pretty quickly, actually. The health was a whole other battle. It took much, much longer. All right, so coming back to this table of CIDs, I wanted to focus on kind of how we understand and respond to these. So the first thing I wanted to say is that we are at risk as a community of what's called the lamp post problem, all right? And there's a figure, a cartoon here that will help you understand it. But the idea behind this is some drunk person walks out of a bar and they can't find their car keys. And they're looking for their car keys. And they're looking for them underneath this lamp post. And the guy says, oh, did you lose your keys? You know, somewhere around here. And he said, no, no, I probably lost them over here, but there's light here, I can see. So I'm looking for my keys over here where there's light. And of course, you're never gonna find your keys over here if you lost them over there. But it's just so appealing where there's light. And I think we sometimes have the same issue with some of our impact modeling, which is that we know how to do certain things well so we focus on the things that we know well. And there's this whole other dark space over here that we have to at least recognize. We have to avoid being this person. So when we look at this table, we can start to directly assess if we are doing a good job of representing different pieces of it. So here are a couple that I'm calling out. The mean air temperature, which we might use metrics like growing degree days. That would be a CID index. For those who don't know, growing degree days basically counts every degree above some baseline temperature. So if your baseline temperature is 10 degrees Celsius, and you have a day that is 13 degrees Celsius, you have three growing degree days. It's three points above that limit. And crops actually grow according to a calculation and accumulation of growing degree days. It's a very strong metric for crops. That's what determines their growth stages and it determines whether we get to that harvest and have had enough time in the field absorbing sunlight and making the carbohydrates that we need for our grains. So mean air temperature, we do a very good job with. So I've put a green star. That's something crop models do well. Mean precipitation, we also do quite well. That's a fundamental part of all crop models. Oridity, we do a little bit better, but a lot of our crop models don't capture the long term trend and drying out of certain regions of the world. Sometimes we assume that crops are planted in a saturated field when that's not always the case. So those types of things we can improve. And you'll also notice over here that I have atmospheric CO2 at the surface. And this was a CID that we really had to fight for in chapter 12 because a lot of people didn't realize carbon dioxide, they think of as the cause of climate change, but it is itself a direct impact on agriculture. And you can't really model the future of agriculture unless you have CO2. So it's definitely an element of our human influence on the climate system that is affecting something that we care about. So it deserves to be here. And I'll show some examples of how we're understanding this. This is a picture over here of a field trial in Arizona where we had a whole field of wheat and we were able to use heat lamps in one part of the field to synthesize warmer conditions. So we had the exact same weather, the exact same precipitation, the winds, you know, all of that stuff is the same, but we have extra heat in this one part of the field. And when we actually did those experiments, we got observations here in red and as we increase seasonal mean temperature, we get a drop in the grain yield. And you can see this pattern. And the gray and green here are our crop models that have done a pretty good job overall of representing that drop off. And there are things to still figure out. There's some error bars, the line is not a perfect one-to-one match, but this is the type of thing we're doing in AgMep. We're trying to find these field experiments and make the models accurate. Yes. That field, I would say that's, I mean, so there were multiple, you can see there's multiple stations, there's other ones in the background like over here, but each one is maybe 10 meters wide, something like that, I don't know. No, I mean, there were many of these. So they were running in different fields kind of all around in that area. But yeah, there's some really fascinating stuff. I'm not even gonna show the free air carbon enrichment, but they have basically a big ring and when the wind blows from this direction, they release carbon dioxide into that wind stream so it blows over the field. And then the wind shifts to over here and now they release the carbon dioxide from here and they're trying to maintain a higher CO2 in the field. All right, so I already mentioned the kind of mean air temperature and CO2 as impacts, but of course as the climate is changing, both of these things are changing. So this is new, not yet published results, but it actually calls a little bit of question on the global warming level approach which has taken hold in working group one and working group three. It's very convenient from a large scale policy perspective to say that in general a two degree world is the same, whether we get there quickly, whether we get there late and even the climate models, the ones that are most sensitive to climate and the ones that are least sensitive to climate still basically say the two degree world is about the same and here is one reason that it's not, which is that if you actually take different climate models, represented as the different colors here and the different shapes being different scenario pathways, when you get to this two degree world, you have a very large range of carbon dioxide concentrations that are associated with that two degree world, more than 150 ppm difference. All right, by the time you get out to a four degree world, the range that you see here is about 200 and then even there you can be deceived because not all of the climate models are sensitive enough to get to four degrees. So the true distribution is even larger. The last level that all of the climate models get to is here, the three degree world and you can see it goes from 550 up to about 750 or 800. So there's very large ranges in CO2, which we know will have an impact on crops. So we actually can run these through our crop models and unsurprisingly the model, the UKESM, which is the most sensitive to climate change, that is the model that reaches higher global warming levels fastest. So when it gets there, the CO2 is not as high as other models which need more CO2 concentration to reach those higher levels. So that UKESM model gets to the high global warming levels earlier with a lower CO2 and that is a very bad combination for crops because now it's the same temperature but the CO2 is lower and that means unsurprisingly down here, that is the most pessimistic model. So the yield losses, this being a maize yield loss, for the models with the highest equilibrium climate sensitivity, so the strongest response to climate change, those models are the most pessimistic which means I can draw a direct line from pessimistic crop projections to the equilibrium climate sensitivity of the climate models. So this is work that's hopefully coming out soon but something that we really need to grapple with. All right, so when we're looking at extreme heat, the other thing that we are trying to do is we have to recognize that there are, it's not one category, there are many thresholds within extreme heat as a category. So we had this figure in chapter 12 in our FAQ section that is just a simple way of thinking about, there is a set of temperatures, represented on the x-axis here, where growth is not really limited by temperature. It basically is growing happily but as you get across a critical temperature threshold, you often see this kind of step change where now you're getting reduced growth. So it's a little bit too warm, the plant's not quite happy and growth goes down but then you can reach a second threshold, a limiting temperature threshold after which the crop can rapidly drop off and fail. So we need to spend time in the agricultural community to identify these thresholds and make sure we understand how close to those limits we are and what types of adaptations we might be able to do. For example, a genetic adaptation could literally move this threshold to the right. Give us more space before we start to lose our yields. Another thing that I just can't resist but showing is a result that we got back in 2016, where we compared, I'm gonna have to talk through this. This is many different models that were run in the AgMIP wheat team and on the bottom axis, we have looked at the year by year variation of that crop model's response to climate and we have made a correlation between the yield and the average temperature. So as you'd expect, most of these models have a negative correlation. When it's hot, the yield is low. When it's a little bit cooler, the yield does better, right? So that negative correlation makes sense. But then we asked the question of are the models that are more sensitive to that seasonal variation also more sensitive to climate change? So when you increase the mean temperature, do the models that are reactive to climate or to temperature, do they drop more? And the general pattern is pretty consistent. This diagonal line here means this was a very warm sensitivity test but very easy to see. The models that responded most to seasonal temperature also responded the most to climate. So that general pattern makes sense but you'll also notice that the models do not cross the zero line at this middle point and what that means is that even the models that did not respond to the seasonal temperature variation show a climate change response. And the reason for that is there is a fundamental difference between a warm season and an overall warmer climate. So you may have remembered that warm year that happened in your country. It may have been a two week, very hot spell that came through and made the temperature for the whole season warm. Plants respond very differently to a two week heat wave than they do to every single day being a little bit warmer, every single day being a little bit faster in your growth stages. So this gap between the origin and all of these crossing lines for each of these sensitivity experiments, the blue one being a cooling experiment, that's why the line is different. But that gap really shows that there is a fundamental difference between seasonal and climate type changes which also calls into question some of these empirical approaches where you fit your climate response to recent seasons and then just assume that that applies to climate change. So we have to be careful there and this points us towards mechanistic models. I'll also just do a quick look when we look at frost versus heat. There is a temptation to think that heat might be expanding but at least frost is moving away. That means that we don't have to worry so much about frost damages. Now we worry about heat damages. Maybe that's a good trade-off. But in reality what we're seeing at least in North America is all of these red areas are places where extreme heat is going to reach either by 2050 or by 2100. So you can see this dramatic expansion including many agricultural zones. I'm looking at Jeff here because there's a lot of Michigan in this chart. The heat is expanding up the mountains and up North. The frost is really only going away in purple here in a few smaller places that are not even as agriculturally profound. So part of this is that the heat is coming but the frost is still variable enough that you still get frosts even in many of the places with heat which means now we have a middle season limit as well as an end of season limit. We did of course in our IPCC work split hydrological drought and agricultural drought. Hydrological drought I think of as the water resources for irrigation and your surface ponds or anything like that whereas the agricultural and ecological drought is really about your availability of soil moisture and I'll just note that there are many, many indices here that we could look at. We've looked in particular at SPEI but there are many that you could examine. In general I would say that the models do pretty well. With that our crop models have irrigation in it but oftentimes the models assume that irrigation is available. So we need to connect it more with the water resource models so that we can have a better set of responses. One other thing that I have to mention is that drought and CO2 actually interact very strongly in our crop models. An example of this is we have experiments with that elevated CO2 done experimentally and this is maize yields and during a wet year the Y-axis here is the response to climate change during I'm sorry is the response to CO2. So in a wet year there's a very low response whether it's irrigated or rain fed but in a dry year the irrigated response is not very strong whereas the rain fed response is very large. Another way of understanding this is that when conditions are dry the crops close their stomata. They make it so that there is less exchange with the environment because they want to hold on to that moisture. That also means that when they close those stomata they are not taking in CO2 in the same way. However, when there is a high CO2 environment even when they're holding on to that water they can still get enough CO2. So the carbon dioxide is especially beneficial during drought years and we can see that in the models and in the experiments. All right, last couple of months I'm looking at of course we have river flood and pluvial flood. Pluvial flood is like heavy precipitation events. The models do a pretty good job with pluvial flood but with river flood many of our models don't even know what is happening elsewhere in the basin. They are effectively single column models so if there's a flood coming down the Mississippi River that site you have in Missouri does not know about that flood. So we are often missing that and in the United States the worst year for us is almost always 1993. Big flood events, huge losses of crops and the models often miss it. All right, so here's the summary of all of this. I've skipped a couple of them but I wanted to kind of provide this and talk about in green here are the CIDs that I think our models do a pretty good job of responding to. Blue are the ones where some of the models include a response or maybe there's more work that we think we can do. The darker blue here are things that we think we can add in a soft coupling. So for example, if you give me maps of coastal areas that are flooding we can add that additional factor to our results. But you'll notice that there are several columns here where we have pretty much no response at all. That's things like fire affecting agricultural zones, hail, heavy storms, some of these things we don't do as well with. And then I put this little red dot on some places just to call out explicitly things like water logging, impacts on agricultural laborers, health, pests and diseases and sequential extremes. These are all areas of active research. And I should also note that when we're providing climate information we have to be careful because we don't always provide changes in every one of these CIDs. So if you have used an approach where the only difference between the baseline climate and the future climate is the average temperature, well you're not gonna see changes in rainfall, you're not gonna see changes in the extreme characteristics, for example. All right, we're doing a lot of work in AgMap right now to build machine learning models and this is a very complicated figure. But the bottom line of it is that we've taken the top 16 maize producing countries of the world. So the number one producer is the US, number two is China, gone across this list. And we've taken many, many different climate features including cool temperatures, freezing temperatures, hot temperatures at two different levels, the average temperature, the total rainfall and then some drought characteristics like consecutive dry days or the number of rainy days. And then we've taken all of our crop models as an additional set of information and we've asked the machine learning model to create the best possible predictor for each of these countries national yields. And when you see a big circle here that means that was the top feature and then you get down to the smallest circle which is elsewhere, you know, top five features. So one way to think about this is if we look at the USA, the top features that were selected were two crop models. So that's probably a good sign for the crop modeling community that the machine learning model said let's start with the crop models and then augment more information. But what they choose to augment is the mean precipitation. They've added information about the total rainfall which we understand to be a way of representing that 1993 flood that the models are missing. They've added additional information, maybe that very heavy rainfall season shows up through that characteristic instead. And then the final one up here, there's a little bit on the number of wet days and frost as another thing that maybe the models are not capturing well enough. Another example here is South Africa. The number one feature is a crop model but then they need more information about extreme heat and cool days. So this is directly pointing us to where we need to improve our models. I already mentioned we have to think about food systems. That means not just the crops but the heat tolerance of the agricultural labors. We heard about this from Robert and Galadio yesterday. So I'm going to go fast. The other message that I really wanted to say loud and clear here today is that when we form adaptations we are not generically adapting. We are targeting some specific climatic impact driver or some specific vulnerability or exposure to a hazard or something that we care about. So here is an example of an adaptation that we have explored in our models to that growing degree days, the number of kind of heat units that we get. So this one is specifically targeting the mean temperature changes of climate change. And what we've done is we've looked all around the world at the different seeds that are grown. In this case I think it's maize that we're looking at. And we have asked ourselves what is the longest growing variety that we can find which means the most heat units, the most ready for a warmer climate. And then we've compared it against the number of heat units that are typical for a growing season in each part of the world. And what you'll see here is in oranges and reds are places where the future growing season is so warm that there are no seeds today that can actually meet that demand. There's no genetic material out there that we can use to have the same growing season at the warmer climate. And you'll notice that this is not the same map as a mean temperature map because what people don't realize is that the United States here is a very productive region not just because we have a nice temperature but we can grow out in the field for a long time. We can spend a long time out in the field collecting that sunshine, making better grain. But when it gets warmer that means that the United States agricultural zone has more days to accumulate that heat and therefore has a stronger impact than a place with a shorter growing season. So this combination of length of the growing season and the growing temperatures combined with genetic information this is really I think very fascinating stuff. The last one I'm going to show here as a scientific plot I believe is thinking about how the global economic models view this same problem. This is a very dry figure but I will talk you through it. We have several different variables on the x-axis here and what we're plotting on the y-axis is the change in 2050 of a climate scenario compared to a future where there is no climate change. And what you'll see is the first thing that comes in is this variable called y-exo for exogenous. So this is what comes from the crop model and basically says here is the yield change that the crop models predict. And the very first thing that the economic models do is respond to that by saying well if the yield is going down dramatically here they're not going to grow it anymore. They're going to move the agriculture somewhere else. So the economic models reshuffle agriculture internally so the actual effect on the agricultural production is less than what the crop models say because they've optimized just a little bit. So the actual crop production changes show up in this one and you can see it's less. Then what happens is in response to the lowered agricultural yields they have to expand the agricultural area. That's what this is. This is that reshuffling has caused new area to be brought under cultivation which leads to other questions about ecosystems and other things that we know encroachment can cause. In general that allows us to maintain production which is necessary because the consumption demand stays very strong. We have to meet that consumption. That can cause big changes in exports and imports and overall the price increases. So this is the overall story. Yields change. We have to shift the production regions. We have to expand area to maintain that same total amount of production to meet high demands. That's going to require new trade and generally higher prices. So this is the kind of thing we're building in our models. And yeah, so it's not just climate impacts. It's also things like land use, dietary demand, trade policy, food waste on the field and beyond and then the role of agro technology. So one of the major efforts that we do within AgMIP is we work with local stakeholders to develop scenarios of future agricultural systems. So here are some of our stakeholder engagements in different parts of Africa. And the overall approach that we're taking is that we assess the climate change risks and engage the stakeholders and say, are you preparing for warmer temperatures, different types of droughts, higher carbon dioxide environments? That usually gets people's attention and then we can have a discussion where we think about what the future agricultural pathways are for a region. That's another way of saying how will agriculture develop so that here in Italy, will they still be growing the same foods in the same ways in 50 years? We could have that conversation and build a scenario of agricultural change even before we bring in the climate itself. Once we have the system changes and the climate changes, we can design agricultural adaptations, especially packages. So not just one at a time, but multiple adaptations. And that allows us to run through models and evaluate the impact and come back to the top and loop through this several different times so that we can iterate and make the best adaptations. And then of course, in the end, we're discussing with this scenario and policy process. All right, so last slide I have here is just a summary. Hopefully I've shown that agriculture is responsive to many climatic impact drivers and we're only tracking a subset. We need to do better and more and that requires better models but also fundamental agricultural research and experimentation. The modeling approaches allow us to capture specific responses and adaptation options. And as I said, we need more data, more models. Adaptations are targeting specific climatic impact drivers and we need more work to identify the specific indices and thresholds and adaptation technologies that move our tolerance around. And that we need to have this co-development process to really make it all work. And I think with that, I'm on my schedule here. So maybe what I'll do is I'll take just any burning questions in the room before we move on to the next speakers. Any questions? Thank you very much for a nice presentation. I have a question about your crop model. Is it include the other factors that could affect the crop yield like the quality of seed or soil moisture content availability and pest and diseases and also the land area. I mean, in a few cases, we have a larger area for, for example, a specific yield and then after that, after a few years, the farmer changed their mind. And they harvested another crop. So we tend to use maps of crop areas and estimates of crop areas. So that is, it's not predictive as much as we track and try to follow. It's basically a different type of modeling that would help us understand why the farmer made that change. And the behavioral aspect of that, when you ask how many bad years would a farmer need before they change, is very unknown. I think that's something that people really don't spend enough time thinking about and we need to do more. Soil moisture, yes, we handle that. We can do better, but we do have soil moisture as a fundamental property of the model. Pest and diseases, we have a whole agnep team working on that, but it's very, very challenging because there are thousands of pests, worms, bugs, insects, things like that. And then pathogens, there are thousands of those. And then when you combine them with every crop species, now you have in combination even more. So we're developing generic ways that we can say that you don't have to characterize every single insect, but you might characterize the way they attack the plant. Are they attacking the roots or the leaves or the stem? Are they eating the live fruits or are they eating the flowers? Like that kind of thing we can figure out. I think I answered the question, but let me know if I missed something. Erica, the question, okay? Yeah. And I need to figure out how to get back to Zoom so I can see first. Thank you, Alex. I just wanted to know, there's another factor, nutrients availability in the soil. So is it possible to include that as well as nutrients availability? Yeah. Yeah, sorry, I just remembered the other question was the quality, so I want to come back to that. Go ahead. Okay, and there's another factor, like when you were saying like the changes in crop type, like people were maybe cultivating some other crop and then later they had transferred to another crop. So in that case, is it possible like, maybe there are some socioeconomic factors because I've seen in my country that people used to like cultivate rice more, but then the prices of maize, when they sell maize, they get like better price. So they shifted to maize. So there are like factors, like socioeconomic factors. So is it possible to include those as well or have these been considered? Yeah, thanks for the question. So in terms of the nutrient quality of the soils, yes, we do track that. The soil databases can be good or bad, but we are generally able to find soil information. And then monitoring the long-term soil health is a challenge, but we are trying. And soil ecology is particularly challenging, but we have nitrogen, phosphorus, potassium, those types of things, and that does affect the quality of the grains. The grain quality also depends on carbon dioxide concentrations. So we know that there are things like that, but in general, we do a better job with cereal crops than plantation crops like fruits and grapes, things like that, where the quality is more important than the quantity. And then your other question about the economic side, we do have farm household models that are looking at the overall economic net returns. And we are trying to track other trends, like in your country, the introduction of irrigation means the whole borough season is now fair play, right? And if you're growing rice then, you can grow other crops in different times. So we try to follow those changes. There's a question online. Yes, so there is one that many would like to answer. All right, looks like Vincent has asked a question here. It's an observation. Vincent, do you wanna say this out loud rather than me try to read it here? Oh, sure. Can you hear me? Yeah, please go ahead. Yes, I have an observation from your earlier statement. You said that when it comes to adaptation, we are not really adapting, but telegating a specific CID. But while working with you from where you've shown that we're working with some communities in Senegal and I think in Mozambique, basically it's like you're coming up with a package of an adaptation. So my concern, don't you think that there's a contradiction when you are talking about adaptation and when you are working with a community? Yeah, that's a good question. And yeah, so when we do an adaptation package, typically there are several elements that show up in an adaptation package. It could be a change of a system where basically there is a determination that a crop is no longer as suitable or likely to have a strong economic return as something else. So you wanna shift towards a ground nut system or towards something else. But the other thing is if you are using some kind of improved drought resistant variety or some kind of improved growing degree day seed selection, the company that is making that or the government subsidy that is enabling that may also couple it with a different fertilizer or chemical regime or may couple it with requirements around planting dates or some kind of crop insurance or who knows what. But the idea is they are trying to change multiple elements of the economic system but each element is still oriented around specific types of extremes. But I think it's a fair point there which is to what level does the package blur those lines is something we should track more. Thank you for that point. All right, one last question and then I wanna make sure we get to our other speakers. I'm just curious if you also consider companion cropping. Like there are some things like a climate adopted companion cropping. Yeah, so there's a broad category that I'll call multi cropping that includes intercropping where you have multiple crops on the same field and then sequences of crops. And we are looking into some of those but that is also quite challenging and maybe I'll just say one last word and you can call it a little bit selfish if you'd like but the overall level of investment in the crop community is much, much, much smaller than the level of investment in something like the climate community. So as the overall focus of the attention moves from the climate questions towards the climate impact questions and the interventions and the adaptation, we need a little bit of rebalancing so that we can get the models that can actually do some of these things that we're asking because I think the attention is calling shedding light on some of these areas that we need better models but they won't just appear if we ask for them we have to make tangible investments to make that happen. All right, thank you for that. Let's go on. Our next speaker is Benjamin Sultan and he's right there to get started. So Benjamin, I'm gonna hand it over to you. Thank you for joining us.