Fig. 2: Convergence cross mapping networks of microeukaryotes at mooring F4 from 2016-08-01 to 2020-09-17.

Each node in the CCM network represents an ASV, and each edge represents the causal influences. The edge weight corresponds to the Normalized Mutual Information determined from the comparison of the individual ASV and their predicted representation in the shadow manifold. ASVs are connected if the smoothed p value of the weight is p < 0.05 (Supplementary Information Fig. S3). A Node color reflects the month in which the ASV exhibits maximal abundance, calculated from the maximum abundance mode for each year ranging from January to December. B In this representation, nodes are colored based on the community membership that was determined by the Louvain community detection algorithm. “HS” labels denote high light summer, “LW” represents low light winter, “TS” corresponds to transition spring, and “TA” indicates transition autumn. C The Normalized Mutual Information aggregated across the edges between the clusters, visually represented by thickness of the arrows corresponding to their respective values. Colors visually represent the clusters. D Interaction analysis between taxonomic clusters. For each of the ten clusters, the interactions between ASV groups are examined at class level, considering “Syndiniales”, “Dinophyceae”, “Bacillariophyta” and “MAST”. The cluster assignments are marked by different colors. The thickness of the arrows denotes the strength of the interaction, while the shapes represent the various taxa groups at the class level.