 This study compared the performance of various trend estimation methods when applied to long-term normalized difference vegetation index, NDVI, time series. The authors found that the best performing methods were those which removed the seasonality of the data before estimating the trend. These methods produced more accurate results than other methods which did not account for the seasonality of the data. Additionally, the authors found that the presence of fire events could lead to false positive detections of breakpoints in the data. This highlights the importance of understanding the underlying causes of any changes in vegetation productivity, as well as the potential impacts of these changes on ecosystem services. This article was authored by Marcus Reichstein, Miguel de Mahetcha, Christopher S. R. May, and others.