A 10-year time series analysis of plankton in the Belgian part of the North Sea

Phytoplankton and zooplankton form the base of marine trophic webs. Therefore, factors that affect plankton dynamics often have far-reaching implications for the broader ecosystem, including higher trophic levels such as fish and marine mammals. In recent years, marine heatwaves (MHWs) have become more common, with MHW duration and frequency increasing globally over the last century 1. In this study, we investigated the effects of MHWs on the plankton community in the coastal area of the Belgian part of the North Sea (BPNS) between 2014-2024. The analysis was conducted in three parts: (1) a long-term analysis of MHW detections and characteristics, (2) two generalized additive mixed models (GAMMs) of temperature components and their effects on plankton densities, and (3) a dynamic structural equation model (DSEM) to examine structural linkages between groups of interest from the GAMMs. These analyses allowed us to characterize MHWs, unravel community dynamics related to temperature, and define likely pathways of MHW effects on the trophic web in the coastal BPNS. For MHW characterization, MHWs were defined as events during which the sea surface temperature (SST) surpasses the 90th percentile threshold of the seasonal climatology for ≥5 days 2. SST data was taken from the European Space Agency Climate Change Initiative and Copernicus Climate Change Service reprocessed SST analyses, and the climatology baseline was established using 10 years of data (2014-2024). Long-term trends were extracted using seasonal trend decomposition with a 4-year smoothing window. Using the heatwaveR package 3, MHWs were investigated at nine coastal stations, aligning with available plankton datasets. For each MHW event, duration, intensity, and time of onset were calculated. MHW frequency displayed an increasing trend at most stations over the 10-year period. Stations that were more coastal as well as further from the Scheldt estuary were the most prone to heatwave events, and MHWs were most likely to occur in winter and summer. MHW events by station, seasonal climatology, and long-term trend were then incorporated into a multivariate GAMM model of plankton densities. Plankton data was collected from 2014-2024 across the nine coastal stations in the BPNS that are monitored monthly as part of the Belgian contribution to LifeWatch. A Working Party 2 net (mesh size 200-µm) and a Apstein net (55-µm) were used to target mesozooplankton and phytoplankton, respectively. Samples were processed with imagers and automatically classified (followed by manual validation), yielding densities for detected taxa 4,5. Taxa with highest abundance were included as response variables and modeled using a hurdle model where zeroes were modeled using a negative binomial distribution and positive densities were modeled using log-transformed values on a Gaussian distribution. MHWs were accounted for using a rolling index (intensity x duration). The hurdle model highlighted that the MHW index had no effect on presence/absence. However, once taxa were present, the MHW effect on plankton abundance displayed high spatial heterogeneity. Weak negative correlations were found between the MHW index and Calanoida, Appendicularia, and Noctiluca spp. Finally, a subset of the biological and MHW data from the GAMM model was used in a DSEM, where plankton were grouped by trophic niche (e.g., primary producers, herbivores, carnivores) and linked accordingly. First DSEM analysis results supported that SST likely impacts the ecosystem at multiple levels (i.e. copepods, appendicularians, and Noctiluca spp.). This study indicates that SST seasonality and long-term SST trend are most related to plankton trends, and that MHWs are more likely to affect specific taxa at a scale that is highly local and context-dependent. Furthermore, we highlight a few key trophic links to consider when investigating the mechanism of temperature effects on marine plankton. 1. E. C. J. Oliver et al., Nat Commun. 9, 1324 (2018). 2. A. J. Hobday et al., Progress in Oceanography. 141, 227–238 (2016). 3. R. W. Schlegel, A. J. Smit, Journal of Open Source Software. 3, 821 (2018). 4. R. Lagaisse et al., Sci Data (2025), doi:10.1038/s41597-025-06278-w. 5. J. Mortelmans et al., Geoscience Data Journal. 6, 76–84 (2019).