COS 44-3 - Identification of species traits and quantification of functional diversity by aggregation of monitoring datasets

Background/Question/Methods

Global change is increasingly impacting species distributions, changing macroecological patterns and reshuffling natural communities. This highlights biodiversity monitoring as an essential task to follow and analyze the ongoing changes. Although the biodiversity of a particular region can be quantified in terms of the number of species present, it is often more informative to consider functional diversity, i.e. the diversity of ecological strategies found in a system. Functional diversity is relatively easy to quantify when the traits of the organisms under consideration are known. However, for many important groups of organisms, information on trait values and even our understanding on nature and relative importance of trait axis is woefully incomplete.

Recently we proposed that diffusion maps, a manifold learning method based on principled harmonic analysis, can be used to reconstruct trait axes and values from monitoring data. This method can thus be used to quantify the functional diversity in a system based solely on monitoring data. Here we focus particularly on phytoplankton, where extensive monitoring timeseries are available from national monitoring programs. However, data from different monitoring programs often differs in terms of methodology and the set of species covered which complicates the aggregation of insights from such datasets.

Results/Conclusions

Here we propose a new method for aggregating phytoplankton monitoring datasets. We illustrate the application of this method on the example of Dutch and German phytoplankton monitoring data for the North Sea. We then use diffusion maps to reconstruct important trait axes, infer species traits and quantify functional-diversity based solely on the co-occurrence of species in samples.

Our results show that analyzing the aggregated dataset yields qualitatively and quantitatively better results than analyzing both datasets of their own. The analysis of the combined dataset identifies plausible trait axes and reveals the extent of spatial and temporal variation of functional diversity with much greater precision than either dataset on its own.

The results presented establish an important proof of principle, providing further evidence that a) traits and functional diversity can be robustly reconstructed from monitoring data alone, and showing that b) data aggregation can greatly increase the accuracy of this reconstruction, and c) heterogeneous datasets can be aggregated by the method proposed here.

In summary these insights suggest that diffusion maps applied to large-scale aggregated data-sets could greatly advance our understanding of phytoplankton diversity and even lead to a global standard for functional diversity of phytoplankton.