The protracted domestication model posits that wild cereals in southwest Asia were cultivated over millennia before the appearance of domesticated cereals in the archaeological record. These ‘pre-domestication cultivation’ activities are widely understood as entailing annual cycles of soil tillage and sowing and are expected to select for domestic traits such as non-shattering ears. However, the reconstruction of these practices is mostly based on indirect evidence and speculation, raising the question of whether pre-domestication cultivation created arable environments that would select for domestic traits. We developed a novel functional ecological model that distinguishes arable fields from wild cereal habitats in the Levant using plant functional traits related to mechanical soil disturbance. Our results show that exploitation practices at key pre-domestication cultivation sites maintained soil disturbance conditions similar to untilled wild cereal habitats. This implies that pre-domestication cultivation did not create arable environments through regular tillage but entailed low-input exploitation practices oriented on the ecological strategies of the competitive large-seeded grasses themselves.