In nature we do not observe boundary curves or states. Rather we observe behaviors manifested in time. One way of capturing these beha- vioral observations is through the simulation of a large number of indivi- duals that follow the rules generated by the dynamic programming equation. Colin Clark and I (Mangel and Clark 1988 , Clark and Mangel 2000 ) called such individual based models forward iterations (see Connections for more about these), to distinguish them from the backward iterations that generate the decision rules. To implement them, we envision simulating a large number, N , of individuals in which the egg complement of individual i at time t is denoted by X i ( t ). We then use the random number generator to connect the state of each individual at time t to time t 1. If we let the state 1 correspond to death, the forward state dynamics associated with Eq. ( 4.31 ) are: X ( t þ 1) ¼ 1 if the parasitoid does not survive from t to t þ 1; X ( t þ 1) ¼ X ( t ) if no host is encountered or an inferior host is encoun- tered and the parasitoid survives from t to t þ 1; and X ( t þ 1) ¼ X ( t ) 1 if a superior host is encountered or an inferior host is encountered and accepted. By simulating forward, we are able to track variables that are measurable in the field or laboratory such as behaviors, mean egg complements, and survival. Sometimes these can even be done by purely analytical (Markov Chain) methods; see Mangel and Clark ( 1988 ) and Houston and McNamara ( 1999 ) for examples, but many times simulation is required because the analytical methods are simply too hard.