The long-tailed duck (Clangula hyemalis) is a vulnerable and declining species wintering in the Baltic Sea. The introduction of the invasive fish, the round goby (Neogobius melanostomus), dramatically impacted the benthic macrofauna in hard-bottom, while no significant changes occurred in soft-bottom benthic macrofauna. Therefore, we aimed to assess the extent to which the diet of long-tailed duck changed in two different bottom types. We analysed the stomach content of 251 long-tailed ducks bycaught in gillnets from 2016 to 2020 in hard- and soft-bottom habitats and compared these results with those published by Žydelis and Ruškyte (2005). The results show that the long-tailed duck experienced a change in diet in hard-bottom habitats, shifting from the blue mussel to Hediste diversicolor, barnacles and fish. In soft-bottom habitats, their diet remained similar over time and was based on H. diversicolor, a few bivalve species and Saduria entomon. There was no evidence of significant differences in diet neither between sex nor age. Despite the above-mentioned changes in diet, the average body condition of the species did not change neither over time nor between habitats. This confirms that long-tailed ducks have high feeding flexibility and quick species response to changes in prey availability, as they are capable of shifting their diet to new prey.

The stomach content of 267 long-tailed duck individuals accidently bycaught in fishermen’s gillnets between 2016 and 2020 from November to May.
Before the collection of stomach samples, long-tailed ducks were weighted and dissected for the determination of sex, age and body index.
Stomach content samples were frozen at -20 °C after individuals were removed from the bycatch until the analysis.
Before the analysis, stomach samples were defrosted and rinsed with clean water.
The prey items were removed and sorted for taxonomic identification to the lowest possible taxonomic unit and for the determination of abundance and biomass.
The ability to count prey individuals depended on the level of digestion. The intact prey items were counted and weighted.
Bivalve shells, fish otoliths and polychaete jaws, were paired and used to estimate the preys’ abundance as explained in Camphuysen and Leopold (2007).
All taxonomically identified prey fragments were weighted and used for characterisation of the biomass of a given prey taxa.
The biomass was measured as wet weight with an accuracy of ± 0.0001 g after excess water was removed using filter paper.
This approach ensured comparability of our results with those obained in the study of Žydelis and Ruškyte (2005).
The body index was estimated by evaluating and summing intestinal fat, subcutaneous fat and the condition of the pectoral muscle. 
Fat deposits were scored from 0 (no fat) to 3 (very fat).
The same procedure was used to determine the condition of pectoral muscle, where 0 was a strongly emaciated pectoral muscle and 3 was a muscle in good condition.

Few data on body condition are missing. ReadMe file is available with all the information about the dataset.