Stratification of spatial data into management classes is a common way of interpreting and managing spatial agricultural data. High-resolution environmental and crop production information was collected within a 2.2-ha apple orchard (Malus domestica cv. Gala) near Sydney, NSW, Australia. Classifying the block into management units using the environmental data did not help to interpret the observed apple spatial variation in apple production. A backwards modelling approach was subsequently undertaken, effectively analyzing whether the crop production explains environmental variation. The backwards management unit model identified a different spatial patterning that revealed both an environmental and managerial effect on fruit production in different sections of the orchards. The unknown management effect had masked any spatial environmental-crop interactions in the initial (forward) management unit analysis. The combined forward and backward mod-eling approach significantly assisted the interpretation of the spatial variation in production in the orchard and identified a previously unknown management effect. When using management units to assess spatial relationships, the inclusion of a backwards modeling approach may help industry and growers to better understand and interpret spatial relationships and facilitate orchard management.