 My name is Spencer Wood, I'm a scientist at the Natural Capital Project, talking today about how tourism and recreation depend on ecosystems. When we talk about recreation, we're talking about activities that people do in their leisure time. So for example, people might like to go hiking or kayaking or fishing, and we need models that can help us determine where those people go and what influences their decision to go to a particular place. So a person's decision to visit any particular place may be influenced by many different factors. So it could be influenced by the proximity to a road, so their ability to access the site. It might be influenced by the cultural history of the place. It could be that they're looking for places that are safe to visit. There's also some number of environmental factors that will determine a person's decision to visit a place. So for example, a kayaker might like to go somewhere with clean water, or a fisher might like to go somewhere with lots of fish. And typically the way people model these decisions is with a linear regression, where these predictors, things like water quality, or access, or the safety of a place, are used as predictors of the number of visits to each place. If we look across different studies where people have used this regression approach to predicting visitors, we find that there's a long list of predictors that might be useful. In a study of shellfish collectors in Puget Sound in Washington in the United States, people found that the development of the area, along with the quality of the water, the access, the abundance of shellfish, were all important factors. In other places, people find completely different factors are important. In some studies of national parks, the visitation rate was related to the fees, the number of substitute sites. In other places, it's a different mix of cultural and natural and other factors that are important for determining visitation. So unfortunately what we find is that in order to study visitation, we need to know the effect of each of those predictors in each place. So in other words, in each place that you study visitation, people's decisions about where to recreate and why are different. And as a result, we need to have data for a place that we can use to measure the effect of those different predictors on the visitation rate. So at the Natural Capital Project, we've been looking into proxy data that we can use to measure visitation to any place. And we've discovered that the number of photographs that people take and post to the online website Flickr are a good substitute for data on visitation to any particular site. So if we look, for example, at the coast of Belize on the left or downtown Minneapolis and the United States on the right, we can see that people are uploading photos of their trips and each of these photos is given a GPS coordinate and a date and the time that they were taken. And we can use this information to estimate visitation rates all around the world. So what the Invest Recreation Model allows you to do is use those densities of photos to plot current visitation in terms of the number of user days across the landscape. And it also lets you relate that density of visitors back to attributes of the landscape that might be predicting where people go. Those can be any number of predictors that are cultural or natural, perhaps information about the infrastructure, and allows you to build our regression model, which is the customary approach, and determine the effect size of each of those predictors. So you can see in this diagram that a cell with lots of photos or lots of visitors also happens to be a place with lots of the brown covered attribute. And in other cells with low visitation or low numbers of photos, that might be related to the blue habitat or the blue activity. So using the correspondence between the visitation rate and the activity, the Invest Model allows you to determine the effect of each on visitation. The Invest Model will present you with a screen where you can specify the grid cell size, so the size of the cells on the left hand diagram on the screen, as well as your input data layers, the predictors that you want to use to estimate the effects on visitation. And you have to specify an area of interest, so the region that you want to study. This tool will output shape files, maps of expected visitation, along with the effect of each predictor that you include in your model. There's also an option in the Invest GUI here to include several global datasets that we've included with the distribution of the tool. And those include data on locations of habitats, as well as locations of infrastructure and other types of activities around the world. So some of the limitations and assumptions of this model that might be important to keep in mind are that, like you've heard, we're using photographs as a proxy measure of where people go. This is an estimate of the visitation rate in any given place. We also assume that people's preferences don't change over time. So in other words, we calculate the effect of each of those predictors on visitation, and we use that effect size as our basis for determining how changes in those layers might result in changes in visitation in the future. The model uses a linear regression, but there are, of course, other more complex models out there that you might think about using. So of course, this model is available as part of the Invest package, which can be downloaded from the web page. And it's important to note that the recreation model does require a connection to the internet, and that's because our database of photographs lives on a server that the Invest tool connects to to process the results. Of course, like you've already heard, there's more information available about this model in the documentation and plenty of help online at the NetCap Forum.