Abstract Background The UN-Decade on Restoration declared that ‘invest in research’ is a critical strategy, as restoration is complex and practices that work in one ecosystem may have adverse impacts in another. However, large-scale restoration has been implemented quickly across the neglected and misunderstood tropical savannas without a corresponding advance in evidence-based guidelines. In other words, we do not know what we are doing. Scope Savannas are not degraded forests. They are mosaics of more-or-less-open pristine ecosystems, maintained by natural disturbances (fire, herbivory). Restoring savannas is fundamentally different from restoring forests. Misunderstandings have led to mistakes that start when mapping degraded land, predicting resilience, prioritizing restoration, and setting target ecosystems and restoration goals. Afterwards, there comes a plethora of inadequate ‘restoration’ techniques that either fail or follow unexpected trajectories. Even worse, these ‘Frankenstein-ecosystems’ have inadvertently replaced valuable secondary savannas and pristine grasslands. Eventually, the wrong indicators (e.g. tree cover and carbon storage) applied to assess Cerrado restoration success drive to wrong conclusions. Restoration, therefore, has not provided the expected ecosystem services, neither has it increased the habitat suitable for Cerrado-endemic plants and animals. Conclusions Cerrado restoration urges caution. ‘If it ain’t broke, don’t fix it!’ When mapping, we must clearly separate at different scales: pristine savannas/grasslands, secondary savannas (naturally regenerating), and both from effectively degraded land. Prioritization of restoration should aim at: (i) larger additionality in ecosystem services and biodiversity, (ii) reducing inequalities in native vegetation cover between regions, and (iii) directly benefiting larger human population. Because re-creating fire-resilient Cerrado vegetation with its dynamic structure and huge biodiversity of species and plant forms is still a mirage, finding a way to pause land conversion is urgent. For effectively degraded areas, where restoration is needed, the major restoration goal should be re-establishing savanna vegetation structure and critical ecosystem services.
Giselda Durigan (2025) studied this question.