Aim : The goal of the study was to quantify changes in the extent of mangroves since the mid-twentieth century and to test the hypothesis that these changes are driven by ocean waves.Location : The pristine 300-km-long coast of French Guiana, South America.Methods : We produced time series of mangrove maps using archival remote sensing images. We retrieved significant wave heights (HS), mean wave periods (TM) and mean wave directions (θM) from the European Centre for Medium-Range Weather Forecasts reanalysis products. We used complex empirical orthogonal function (CEOF) decomposition to extract the main mode of mangrove surface area (MS) variability and singular value decomposition (SVD) to test the relationships between MS and HS, TM and θM.Results : The leading mode of variability extracted from the CEOF decomposition of MS captured approximately 78% of the total auto-covariance and revealed multi-decadal fluctuations in MS that were on the order of 10,000 ha. The SVD results indicated that the multi-decadal fluctuation in MS cross-covaried with HS, TM and θM over the North Atlantic sector, particularly in the region immediately off the French Guiana coast that is remotely forced by the North Atlantic Oscillation (NAO) during the winter season.Main conclusions : We provide evidence based on linear statistics that variations in the extent of mangroves are driven by large-scale, low-frequency changes in North Atlantic waves that are related to the NAO. Such a relationship is hypothesized to operate through wave pounding, which alters the mud substrates on which mangroves thrive and which varies with the phase of the NAO. In addition to long-term trends due to anthropogenic climate change, our results stress the importance of studying low-frequency modes of climate variability to understand changes in the extent of mangroves.