One reason often advanced for the conservation of crop genetic resources is that landraces and other cultivated and non-cultivated materials can offer a rich source of resistance to agricultural diseases and pests. In this paper, we investigate this hypothesis empirically.

Using data from agricultural experiments, we estimate the distributions of pathogen resistance across large populations of wheat varieties. This gives us information about the probabilities of finding useful material in a search of the gene bank. Using this information, we use a search model and Monte Carlo simulations to answer four questions: What is the value of specialized knowledge about which genetic resources are most likely to display resistance? How should search resources be allocated across two or more types of germplasm? How are the chances of a successful search affected by the size of the collection? What is the aggregate value of a collection? Preliminary findings suggest that current scientific strategies are consistent with the optimal strategies identified by the model.