River deltas are diverse, economically and ecologically important ecosystems that are increasingly vulnerable to environmental change. A recent study reported that global-scale human impact on delta morphology has led to net land area gain (Nienhuis et al. 1 ). However, we argue that an unconventional definition of deltas, miscellaneous coastal features unduly characterized as deltas and misidentified delta area changes led to spurious statistics in the study by Nienhuis et al. 1 and that a recently published correction 2 does not address our concerns. We call for the rigorous verification and analysis of large environmental datasets.
Delta identificationNienhuis et al. 1 defines marine river mouths with fluvial water and sediment discharge thresholds (respectively >1 m 3 /s and > 0.01 kg/s) as deltas. This definition runs counter to standard delta 3-6 geoscience definitions, but nevertheless uses morphological terms that are specific to deltas 3,4 as distinct from estuaries, such as shoreline protrusion and triangular shape. Simplified assumptions on river-mouth sediment flux redistribution by river, waves and tides used by Nienhuis et al. 1 neglect the sediment-dispersal role of these agents that render most river mouths devoid of deltas 3,5,6 . These assumptions also neglect sedimentation pathways interlinking the connected upbuilding and outbuilding components of deltas (subaerial delta-plain, subaqueous delta-front, pro-delta) that differentiate them from simple estuaries. Nienhuis et al. 1 uses automatically-generated 'buffer' areas to identify land change from global spatial data 7 , arguing for the exclusion of areas distant from channel banks and shorelines. We find that this automatic identification of buffers, without verification of their accuracy, leads to thousands of features wrongly identified as "deltas" by Nienhuis et al. 1 , including estuaries, built-