The purpose of this paper is to model farmer behavior with respect to in situ conservation, extending the existing literature beyond competition within the principal crop to encompass a broader definition of on-farm diversity and testing the hypothesis that factors affecting the principal crop explain overall crop diversity on the farm . Primary survey data is used from a rural area of Puebla state, Mexico. A Poisson regression is run on the total number of species in the milpa system as explained by cultural, agricultural and economic variables. A set of Poisson regressions, one for each crop group, is run in order test whether factors affect different crops in different ways. Policy implication of the findings are discussed for an in situ conservation program.