Many species have adapted successfully to traditionally cultivated agricultural environments but, as production systems are intensified, this adaptation is reaching its limits. Conflicting facets of sustainability compound the problem. Here we describe how reductions in the use of water in rice fields is compromising the persistence of the largest known breeding population of the Australasian Bittern (Botaurus poiciloptilus), a globally endangered waterbird. In fields with traditional, early permanent water, bitterns began nesting around 77 days after inundation, with 65% of nests having sufficient time for all chicks to fledge before harvest. Our breeding success model showed that all nests could potentially be successful if permanent water was applied by early November, with a ponding period – the phase when fields are flooded – of at least 149 days. The modelling suggests that successful bittern breeding was unlikely where rice was grown using new water-saving methods – drill-sown and delayed permanent water – because the ponding period is too short. These methods have become the rice industry standard in Australia, rising from 34% of fields in 2014 to 91% in 2020. While this saved 1.5–4.5 megalitres/ha per year, it has undermined the habitat value of these agricultural wetlands. ‘Bittern-friendly’ rice growing incentives could encourage timely nesting and maximise breeding success. Early and sufficient ponding can be complemented by establishing adjacent wetland habitat refuges, maintaining grassy banks, and creating dedicated patches to fast-track nesting. Increasing water-use efficiency in agro-ecosystems is widely touted as being beneficial to the environment, but our research demonstrates the urgent need to manage trade-offs with biodiversity conservation.