 So we have a scintillating set of presentations coming up, six presentations on food security and nutrition followed by an exciting 10 table discussion and in between lunch. So lots of excitement coming up. The talks are coming. I want to make a few key points that I want you to bear in mind as we move through. First thing is that forests and trees in terms of their contribution to food and nutrition are only part of most people's livelihood. It's very rare that people's entire livelihood is based on trees or forests. It's a contribution to a much more complex system and so it's all in the interactions often at a livelihood level. And livelihoods are embedded in landscapes. And by the way, we're not still in the Sentinel Landscape Workshop. That was yesterday and the day before for those of us who are in it. Landscapes and central landscapes remain important now but we're working in a broader range of environments than just the Sentinel Landscapes. The key point is that you've got lots of different livelihood systems interacting at a landscape scale with bridging and binding social capital being important in terms of linking those different livelihood systems together and linking amongst very different livelihood systems like pastoralists moving their livestock through agricultural areas and very much the interlocking livelihoods become very significant. No more important than in Nepal where you can't understand what's going on unless you realize and bringing from the 10-year session that we've just had collectively used, individually used land, interacting via livestock and trees being hugely important because forest communal, commonly owned forest areas increasingly degraded and then under community control meaning that fodder no longer available from those farmers have to grow the trees on their farms. The whole thing changes. Now the language that we tend to use is tree-based solutions but quite often once we see the trees it's not the tree that needs to actually change. So we were able to increase maize production by 30% in the mid-hills in Nepal once you realize that the trees are there and interacting with the maize and if you select through farmer participatory selection varieties that suit the complex environment that farmers are using then they're much more productive. And what is going on here? The farmers plant at way, way higher densities than recommended. They think to way lower densities than recommended and they do all this relay cropping with millet on terraces with trees. Simit was selecting maize in beautiful terraces with no trees at the correct densities. Why is the thinning being done to feed animals? Because the objective of the farmer is not just to produce maize but to run their whole system which is maize, a tree fodder providing a winter feed for cattle and the whole thing has to work and of course the manure from the cows goes back onto the land. The trees critical in that role but tree-based not necessarily in terms of the solutions. Let us have an open mind. Another approach is to food security analysis. Keeping the livestock element, we're working with Mark Van Beek at Illry to look at trees and food security. And if we do a very simple analysis now, I don't know about what you think about one of these questionnaires that are going on in the CG. SeaCaps have one that takes about two days or something to administer. There are others I've seen which are equally problematic. And I've been hearing a number of people talking about the sensor landscapes. Yesterday they said there's real farmer fatigue, small holder fatigue in interacting with us. Well I'm not surprised. I don't want to sit down for two days and talk about my beauty of my livelihood. So we need much more simple straightforward quick tools. Half an hour questionnaire on food security can produce really exciting data. And here again let's get away from means. We all know the problem of mean values and looking at differences between means. One foot in the fire, one foot in the freezer and I'm a comfortable temperature. In reality of course I'm not. And we really have a problem if we don't actually look at distributions. Here you can see the green is food consumed, the red livestock consumed, the blue crops sold and then livestock sold. And you can see for four places, Malcasa and Baco and Ethiopia, Bukesira and Gishwati in Rwanda, we get very different distributions. This red line shows where we're at food security value of index of one. So below that you've got food security problems. What we can see here is that the makeup is very different. The numbers of people with serious problems are very different and the way in which they are building their food security is very different. And there are variations within and between the sites. That leads us to not just options by context, minus theory of place, but to embedding options by context within a co-learning framework. And that is realizing there are no silver bullets. But what we need to do is to be developing interventions and the evidence of what interventions are most cost effective in different contexts through systematic testing and feedback learning in which we try out different options across contexts. And we have to embed our research in development if we're going to do this properly because we're never getting enough money in our research budgets. The people who got the money are the development partners, ten times the magnitude at least. And the problem is often development organisations are not using any evidence at all in what they're trying out to improve food security and nutrition. So what we need to be doing is embedding evidence in systematic testing and then feeding back to make that development dollar more effectively spent.