Zooplankton samples were collected between 03/26/2018 and 04/27/2018 around the northern tip of the Antarctic Peninsula (63° 0' 1.843'' S, 60° 0' 16.901''W) onboard the RV Polarstern during the PS112 campaign in order to identify spatial distribution in response to environmental variables (CTD raw data files from POLARSTERN cruise PS112, https://doi.org/10.1594/PANGAEA.895969) and the abundance of krill (Euphausia superba) and salps (Salpa thompsoni). Samples were taken using a Bongo net with a mesh size of 150 µm. The net was equipped with a flowmeter (HydroBios) to measure the filtered volume. On board, the net sample was sieved over a 2000 µm mesh in order to separate organisms >2000 µm. The smaller fraction (150 – 2000 µm) was homogenized in 200 mL 0.2 µm filtered seawater and equally split into 4 x 50 mL by using a Hensen-Stempel pipette. The mesozooplankton size range of 150 – 2000 µm was defined according to Atkinson et al. (2012). Two parts were then filtered on 47 mm GF/C Whatman filters (precombusted, acidified and weighed) for analysis of dry weight (DW), bulk carbon (C), nitrogen (N) and phosphorus (P) content, while the third part was preserved in 4 % formalin for abundance, biovolume and size structure analysis. The C/N filters were sealed in tin capsules and analyzed using a CHN analyzer (Thermo, Flash EA 1112). Prior to the analysis, filters for particulate phosphorus were combusted at 450 °C for 5 hours. Particulate organic phosphorus (POP) was measured photometrically as orthophosphate (PO4) by molybdate reaction after sulfuric acid and heat digestion at 90 °C, modified after (Grasshoff et al., 2009). Another filter containing the 4th part served as a back-up. Mesozooplankton bulk stoichiometry data are shown in dataset one.The zooplankton subsamples for taxonomic analysis were scanned using the ZooScan digital imaging system (Model Biotom, Hydroptic Inc., France), a water-proof scanner with a resolution of 2400 dpi (Gorsky et al., 2010; doi:10.1093/plankt/fbp124). Prior to scanning, the formalin preserved samples were rinsed and five samples were further subdivided with a Motoda splitter to reduce the number of organisms per scan and avoid overlapping in the scanning frame. The splits were then placed on the scanner and overlapping organisms were separated manually. Subsequently, the obtained scanning image was processed with ZooProcess, a macro of the image processing software ImageJ (Rasband, 2012) to allow automated processing and measurement of images. These single object images and their metadata were uploaded to the web-based application EcoTaxa (https://ecotaxa.obs-vlfr.fr/prj/2529). Manual validation of the results was required to ensure correct classification. The images were identified to the lowest taxonomic level possible. Prior to quantitative analysis of the obtained data, the image categories containing no zooplankton organisms such as “detritus”, “fiber”, “bubbles” etc. were removed. Abundance of zooplankton taxa was calculated based on the number of images per taxonomic category. Zooplankton organisms were identified to the lowest possible taxonomical level. Whenever identification to species level was not possible, the sample was identified to the next identifiable taxonomical category and assigned a putative species name. The abundance and biovolume data are shown in dataset two and three.The metadata of each image also contain the estimates for body size (body length: major axis of the best fitting ellipse; body width: minor axis) that were used to calculate the biovolume of each object. For the biovolume per size class, the biovolume (mm³/m³) was sorted in octave-scale size class intervals given as individual biovolume (mm3). The lowest limit of the first size class corresponded to the smallest detected ellipsoidal biovolume of 0.00025 mm³. Each size class was then doubled with respect to the previous one. Consequently, the resulting intervals were narrow for small body sizes and became progressively wider with increasing body size. The largest size class was determined by the largest individuals in each sample. As a result, the lower boundary of each size class equaled the interval width. The biovolume (mm³/m³) was then summed for each size class interval. The size distribution (mm³) with total biovolume (mm³/m³) per size bin is given in dataset four.