Public CPR data replicate the 1958–2002 dinoflagellate and meroplankton timing advances in the central North Sea, which stayed to 2019 as an SST model predicted, but not the copepod advance

Question: do the plankton timing shifts that Edwards & Richardson (2004) reported for the central North Sea in 1958–2002 replicate in public CPR Survey v1.3 data, and did 1958–2002 trends or a 1958–2002 SST relation predict 2003–2019 timing? Data: 33,503 CPR samples (54–58 N, 3 W–8 E, 1958–2019) via OBIS, 71 taxa in five groups selected by coverage before analysis, ERSSTv5 SST. Method: central tendency (CT, abundance-weighted mean month), plus circular phase and median month; group means of taxon anomalies; moving-block bootstrap; four out-of-sample predictors of the 2003–2019 mean; 25 variants (region, three treatments of the 19% of records without counts, log and occurrence scales, three indices). Results: in 1958–2002 the dinoflagellate CT moved −20 days [−31, −7] (reported −23) and the meroplankton CT −24 days [−38, −10] (reported −27); circular phase gave the same direction, so these are peak shifts. Copepods (+3 [−6, +17]; reported −10) and other holozooplankton (+6.5 [−5, +16]; reported −10) did not replicate in any variant, although copepods were earlier in warm years (−8 days per °C, detrended). The diatom CT moved −16 days but its phase did not change; the spring share of diatom abundance rose in 21 of 29 taxa, so CT recorded bloom balance, not earlier blooms. In 2003–2019 (SST +0.76 °C over 1958–2002), dinoflagellates were −17 days (SST model −15) and meroplankton −18 days (SST model −10, trend −17). Over the five groups, the SST model had a mean absolute error of 4.2 days, versus 5.7 for trend extrapolation and 8.5 for no change, and it was best in 24 of 25 variants. For single taxa the SST prediction correlated with observed 2003–2019 timing (r = 0.38 [0.12, 0.59]) but did not clearly reduce error; 17-year taxon trends had reliability 0.06, so trend persistence cannot be tested. Limitations: taxon list and box differ from the original study; missing counts; only five groups and 17 test years; SST is a correlate, not a demonstrated cause.

Continuous Plankton Recorder · phenology · North Sea · dinoflagellates · meroplankton · copepods · diatoms · central tendency · circular statistics · trophic mismatch · sea surface temperature · out-of-sample test · replication · climate change