Erroneous information considering Chironomidae and Chaoboridae accumulation was given in Figure 4 published in Luoto et al. (2017). Therefore, it cannot be stated as in the results (page 5) that the chironomid flux increased towards the present. This data is not interpreted or discussed in the manuscript any further, but as a result from the incorrect accumulation unit, the midge accumulation data cannot be used for reference in other research. The volumetric midge sample size given in the methods (page 3) is the maximum sample size used. Also the midge percentage information in Figure 4 is incorrect. However, the correct percentages were otherwise used throughout the manuscript, including the midge-based reconstructions.
This is figure 4 from the original paper
Apparently, the accumulation rates in figure are wrong (the figure shows concentrations not accumulation rates), the amount of sediment used differed from that reported, and, added almost as an after thought, the midge assemblage data are also wrong.
Fortunately however, the correction assures the reader that all the rest of the paper, with the exception of a sentence about the increasing flux, was based on the correct data. Obviously, it unreasonable to expect the correction to include a figure showing the correct data or an explanation of why the wrong data were published: there is just not enough space on the internet.
Curiously, the description of the stratigraphy on page five of the original paper, which the correction assures the reader is for the the correct data, perfectly matches the incorrect data.
The incorrect data in the stratigraphy are easy to digitise and are entirely consistent with count sums of between 7 and 21 midges. Counts so low that no honest scientist would knowingly try to portray them as counts of at least 50 midges.
The Chaoborus:Chironomidae ratio calculated from the digitised data matches the published curve; the N2 curve can also be reproduced (if the data are first square root transformed). Even the ordination can be perfectly reproduced.
What are the odds of an ordination on the incorrect data being identical to an ordination on the correct data?
I cannot reproduce the reconstructions of dissolved oxygen or total phosphorous as the calibration set data have not been archived. (The total phosphorous reconstruction uses a four component WAPLS model which apparently has an R2 of 0.92. Without the raw data, I am disinclined to believe this; it could be another transfer function where the apparent performance is reported as the cross-validated performance.)
While it is good to see the authors taking responsibility for problems in the Hiidenvesi paper, I am not persuaded that the correction is a full account of its problems. The authors need to archive the raw data, explain how the incorrect data came to be published, and how the results that they report to have used the correct data can be reproduced with the incorrect data.