Planting of native species is becoming extensive in fragmented agricultural landscapes in both rehabilitation and restoration programs. It is important to have knowledge of patterns of gene flow since it may be beneficial where the objective is to restore native populations and landscape processes, or may be undesirable if particular plantings are undertaken for more specific objectives such as hydrological management. Eucalyptus loxophleba and Acacia saligna are two woody species being planted in agroforestry programs to address amelioration of dryland salinity in the agricultural region of Western Australia. Both species have widespread distributions in Western Australia and their morphological and genetic variation is represented in subspecies. Planting of one subspecies within the natural range of another may lead to genetic contamination of natural remnant populations if pollen dispersal is extensive and the subspecies are inter-fertile. Pollen dispersal between planted and native remnant populations was investigated in both species where they occur in a mosaic of agricultural production. Extensive levels of long distance pollen dispersal were observed in both species over distances of up to 2 km. In addition the significantly greater fecundity of the planted subspecies of A. saligna lead to high levels of pollen immigration in the less fecund natural subspecies. The extensive pollen dispersal observed in these two species indicates that the risk of genetic contamination of remnant populations is significant. A risk management framework is required to ensure that agroforestry programs can be developed to achieve rehabilitation outcomes without negative impacts on remnant patches of biodiversity.