The twin crises of climate change and biodiversity loss define a strong need for functional diversity monitoring. While the availability of high-quality ecological monitoring data is increasing, the quantification of functional diversity so far requires the identification of species traits, for which data is harder to obtain. However, the traits that are relevant for the ecological function of a species also shape its performance in the environment and hence should be reflected indirectly in its spatio-temporal distribution. Thus it may be possible to reconstruct these traits from a sufficiently extensive monitoring dataset. Here we use diffusion maps, a deterministic and de-facto parameter-free analysis method, to reconstruct a proxy representation of the species' traits directly from monitoring data and use it to estimate functional diversity. We demonstrate this approach both with simulated data and real-world phytoplankton monitoring data from the Baltic sea. We anticipate that wider application of this approach to existing data could greatly advance the analysis of changes in functional biodiversity.