 So the final step in ProAct is trade-offs. This is how do we make choices between different alternatives? This is a particularly challenging problem because we care about lots of different things. So we have multiple objectives. We are measuring those objectives. We oftentimes have a number of different choices. And then we have all of this uncertain scientific information. So this can be pretty tricky. However, there's a range of different methods that have been developed. And again, this is where I encourage you to collaborate with a decision scientist or an operations research expert who can really help to apply these methods in a robust way and think about the opportunities to link ecological forecasts with decision models. So there's three major trade-off methods that we'll talk about. Dominance methods, robust methods, and multi-attribute criteria methods. The first approach is a suite of dominance methods. These methods are approaches where what we want to do is eliminate choices that will never be the best choice. There's always going to be another option that is better. The second method that we can use to think about trade-offs are robust methods. These are approaches where when there's large uncertainties in the future, what are some of the options that allow you the flexibility to make changes into the future and are probably not going to be bad? So how can we narrow the set of options to those that might be better than not? Multi-attribute methods are particularly important for environmental management decisions because we oftentimes care about more than two things. And that means that we have to think about ways of being able to compare across a range of different objectives. One of the ways that we do that is by thinking about how across all of those performance measures, we normalize those. So we're not thinking about performance measures in their original units. Instead, we're translating that into an index of desirability. These are oftentimes called utility functions. This is a way of mapping a value that is measured as a performance measure for the particular range of that performance measure onto that index of desirability that then allows us to understand which of the levels of that particular indicator is the most desirable, which is the least? And how do the points map in between? That allows us to create some type of functional form that creates a normalization for each particular performance measure. And because each of those is a unitless value, we're able to combine that by combining those normalized utility functions with a weight. Now the weight ends up being really important because that gives us a sense of how much we care about one objective over another objective. With stakeholders, this means that you can have multiple different analyses that represent more than one stakeholder perspective. So you can see for a particular stakeholder what their preferred option might be. Options that are least desirable. And you can also identify across the stakeholders if there's some options that are never going to be considered because they're not ones that perform well for any stakeholder, or ones that require additional conversations because they're in a set where they're most desirable or within a grouping that is more desirable for a particular stakeholder group. There's rigorous ways to assess these utilities because these are preference elicitations. This is different than expert elicitations because you're not developing a probability distribution. You're mapping between values and some level of desirability. Additionally, there's ways of constructing the weights using rigorous approaches such as swing weighting methods. And employing those methods appropriately is critical to being able to get output from a decision model that is relevant, useful, and helps to progress the way that we think about making environmental management decisions and considering the different options that we care about.