All together 136 CTD casts were performed, from which 18 were catalogued as isotopic (a full suite of observations is given in Table 1a and b), 65 as chemical (i.e., GO-SHIP level 1 variables) and 59 as physical (i.e., only sampling for salinity). Due to the water amount needed, two casts were performed on most of the isotopic stations, the first cast was a full profile and the second a shallow one. During the physical stations, water samples at three levels were taken for salinity analysis. The samples were then analyzed on board using a Guildline Autosal Salinometer. A total of 162 samples in 59 stations were taken during the cruise with an offset with respect to standard water varying from 0.0002 to 0.0030 depending on the laboratory temperature. The samples were taken at depth with a constant salinity gradient to ensure that no natural changes in salinity affect the comparison between sample and sensor.

The primary CTD system (for specifications, see Table 2) initially used on board was a Seabird SBE9plus + CTD s/n 0285 from the University of Hamburg connected to a SBE11 deck unit, configured with a 24-position SBE-32 pylon (from GEOMAR) with 10 L Niskin bottles. The position of bottle no. 23 and no. 24 was occupied by the lADCP (for specifications, see Table 3). Initially, the CTD was set up with two sensors for temperature and conductivity, an oxygen sensor, a fluorometer, and an altimeter. To test the configuration and performance of the instrument a station was carried out on the Cretan Sea at the start of the cruise. Unfortunately, we had countless problems with instruments, sensors, cables and the rosette during most of the campaign that forced us to change them very often with others available on board, resulting in a continuous change of system configuration. Thus, all different configurations were carefully considered when post-processing the CTD data.

Temperature, salinity and pressure data were post-processed by applying Seabird software and MATLAB^® routines. At this stage, spikes were removed and 1 dbar averages calculated. A first attempt to assess the performance of the conductivity sensors installed on the CTD rosette was done by comparing the salinity data with the bottle samples analyzed with the salinometer. The different hardware setups and configurations are taken carefully into account during post-processing. Overall accuracies are within the expected range of salinity (0.003).

Underway CTD measurements (uCTD; for specifications, see Table 4) provide high-resolution profiles of temperature, conductivity and depth, which allow us to characterize the upper-ocean properties and identify the position and characteristics of mesoscale structures. The advantage of this type of measurement is that it is not required to stop the vessel, it is only necessary to maintain lower velocities (about 3 kn) during the deployments to reach greater depths. These measurements were made with an Ocean Science uCTD system.

The first uCTD deployment was done on 5 March, between CTD 015 and 016 stations, and we continued with this type of sampling between each CTD station to increase the sampling resolution. Unfortunately, several deployments were canceled due to severe weather conditions and no uCTD cast was performed when the depth was shallower than 500 m. All together, 176 casts were taken with depths ranging from 557 to 864 m.

Two probes were used during the cruise with a no-time-limit mode configuration (apart from the first cast configured to stop recording after 600 s, reaching 616 m depth) in order to get longer records. The probe tail spools were attached to the winch through a rope loop that was made new every day in the morning. Despite the probes being able to record several casts, data were downloaded right after each cast using a SBE software in order to avoid losing the data in case the probe was lost and to free up the memory. The probes were exchanged when the battery was running low (around 3.8 V). On three occasions no data were recorded because the magnet was taken off twice before deployment.

For calibration purposes, some additional casts were done right after the CTD cast in order to compare the data sets. The probes were also sent down with the starboard CTD in station 130.

Data files were processed using a set of MATLAB^® routines. After extracting the downcast data, the first correction was done to remove inaccuracies in the descent rate, based on the work of Ullmann and Hebert (2014). Additionally, the data were aligned to the comparable CTD data sets.

Ocean currents were studied by means of vertical profiles made with a lADCP-2 system (Workhorse RD Instruments type, Table 3) which included two ADCPs operating at a frequency of 300 kHz, one looking upward and the other one looking downward. The system was placed in the rosette occupying the position of Niskin bottles 23 and 24. During the cruise, the lADCP batteries were changed twice: the first time on 17 March in Station 58 and the second time on 27 March in Station 105. Except for three stations (station 73, 74, 80) with water depths less than 500 m, lADCP measurements were done at all CTDs. For these stations, the currents were observed by the ship-mounted ADCP. At isotope stations, lADCP profiles were only recorded from the deep cast. The gained data were processed with LDEO MATLAB^® lADCP processing system version 10.15 (Turnherr, 2014). This software uses the raw lADCP data, processed CTD data and navigational data from the CTD. The resulting data are the u- and v- velocities at the depth. The bin size was set to 8 m.

During the whole campaign, underway current measurements were taken with two vessel-mounted Ocean Surveyors (ADCP) manufactured by RDI. The first, with a working frequency of 75 kHz, covered approximately the top 500–700 m of the water column. The number of bins was set to 100 with bin size of 8 m. The second, with a working frequency of 38 kHz, has a depth range of about 1600 m, set with the same bin number as the previous one and bin size of 16 m. Both instruments run in narrowband mode and were controlled by computers using the conventional RDI VMDAS software under a Microsoft Windows system with a pinging set to fast as possible. No interferences with other used acoustical instruments were observed. The ADCP data was post-processed afterwards with the CODAS3 Software System (https://currents.soest.hawaii.edu/docs/adcp_doc/, last access: 10 November 2020), which allows extracting data, assigning coordinates, and editing and correcting velocity data. Moreover, the data were corrected for errors in the value of sound velocity in water, and misalignment of the instrument with respect to the axis of the ship (about -2.8∘ for 75 kHz ADCP and about -0.15∘ for 38 kHz ADCP).

Underway (UW) measurements of partial pressure of CO2 (pCO2), and dissolved oxygen partial pressure (pCO2 the corresponding data set in Table 1b only contains pCO2) in seawater were carried out by means of a Contros HydroC pCO2 analyzer for pCO2 and an Aanderaa optode for oxygen.

The instruments were placed in a cooling box in the hangar. Seawater was drawn from the ship's centrifugal pump for clean seawater that was continuously flowing through the cooling box with the inlet close to the instruments. Water was pumped through a SeaBird 5 salinity and temperature sensor and onto the HydroC instrument (Gerke, 2020).

The system operated reliably throughout the cruise, except when data acquisition was interrupted for the pCO2 instrument for 2 d directly after the ship's centrifugal pump was switched off. This led to a 5 d period without data between 5 and 10 March. During the cruise, 13 samples were taken from the cooling box for discrete measurements of pH and total alkalinity. The UW measurements started on 2 March at 20:20 and stopped on 1 April 2018, at 14:00 UTC.

The underway oxygen measurements were calibrated by comparing them to the Winkler measurements taken for surface samples at the chemical CTD stations.

Dissolved oxygen in seawater was not only measured with the CTD, but samples were also taken at every station and depth along the cruise and reported in µmolkg-1. GO-SHIP guidelines recommend Winkler measurements on all samples, in addition to sensor measurements on the CTD package, and we largely followed those recommendations. Unfortunately, we had to mark large numbers of oxygen values determined with the CTD as questionable due to the several technical problems with the CTDs and sensors. Usually, samples were taken at standard depths, but, especially at the surface and at the bottom, the depths were varied according to the requirements of the other biogeochemical parameters. Oxygen was measured following the automatic Winkler potentiometric method, modified following Langdon (2010). Titrations were done within the sampling calibrated flasks using an Automatic Titrator Mettler Toledo T50 with a platinum combined electrode.

Reagents of blank and thiosulfate standardization were done daily by means of potassium iodate standard 1.667 mmol by OSIL, UK. About 1400 samples were analyzed on board. The precision of dissolved oxygen measurements was determined on five replicates at the beginning and at the end of the cruise (Table 5).

In addition, during the cruise 46 duplicates were analyzed. The results are given in Table 6.

Discrete CO2 variables were measured on board, i.e., dissolved inorganic carbon (DIC), pH, total alkalinity (TA) and carbonate ion (CO32-); these variables were measured at selected stations and depths (Table 9). In addition, discrete samples for DIC, pH and TA were analyzed specifically from surface Niskin bottles to be compared with the continuous water supply feeding the pCO2 system in determined stations. For further details, especially about the on-board procedure for the measurement of samples, see Hainbucher et al. (2018).

During the cruise, one gas chromatograph purge-and-trap (GC/PT) system was used for the measurements of the transient tracers CFC-12 and SF6. The system is modified versions of the setup normally used for the analysis of CFCs (Bullister and Weiss, 1988). All samples were collected in 250 mL ground glass syringes, of which an aliquot of about 200 mL was injected to the purge-and-trap system, normally within 5 h from sampling.

The traps consisted of 100 cm 1/16 in. tubing packed with 70 cm Heysep D kept at temperatures between -70 and -75 ∘C during trapping. The traps were desorbed by heating to 120 ∘C and passed onto the pre-column. The pre-column consisted of 20 cm Porasil C, followed by 20 cm Molsieve 5A in a 1/8 in. stainless steel column. The main column was a 1/8 in. packed column consisting of 180 cm Carbograph 1AC (60–80 mesh) and a 50 cm Molsieve 5A post-column. Both columns were kept isothermal at 60 ∘C. Detection was performed on an electron capture detector (ECD).

Standardization was performed by injecting small volumes of gaseous standard containing CFC-12 and SF6. This working standard was prepared by the company Dueste-Steiniger (DS1). The CFC-12 and SF6 concentrations in the working standard has been calibrated vs. a reference standard obtained from R.F Weiss group at SIO, and the CFC-12 data are reported on the SIO98 scale. Calibration curves were measured roughly once a week in order to characterize the nonlinearity of the system, depending on workload and system performance. Point calibrations were always performed between stations to determine the short-term drift in the detector. Replicate measurements were taken except for near coastal stations due to high workload. To assess the reproducibility of the setup, 50 replicate samples were run, and this resulted in a reproducibility of 1.0 % or 0.01 pmolkg-1 for CFC-12 and 2.3 % or 0.03 fmolkg-1 for SF6. In total, we successfully measured 1084 samples on 68 stations for transient tracers. The results are discussed in Li and Tanhua (2020).

In addition to the on-board analysis, at three stations (no. 52, no. 84 and no. 106) 1500 ml glass ampoules were flame-sealed for later analysis in the lab in Kiel for the detection of novel halogenated tracers such as HFC134a and HCFC22 (Li and Tanhua, 2019).

Seawater samples for DOC were collected from the CTD Rosette into 250 mL polycarbonate Nalgene bottles. Samples were filtered through a 0.2 µm Nylon filter under high-purity air pressure. Filtered samples were collected in 60 mL Nalgene bottles, acidified and stored at 4 ∘C in the dark.

DOC measurements were carried out with a Shimadzu total organic carbon analyzer (TOC-Vcsn) via high-temperature catalytic oxidation. Samples were acidified with HCl 2N and sparged for 3 min with CO2-free pure air, in order to remove inorganic carbon. From three to five replicate injections were performed until the analytical precision was lower than 1 % (±1 µM). A five-point calibration curve was done by injecting standard solutions of potassium hydrogen phthalate in the expected concentration range of the samples. At the beginning and end of each analytical day the system blank was measured using low carbon water (LCW), and the reliability of measurements was controlled by a comparison of data with a DOC reference (CRM) seawater sample kindly provided by Dennis A. Hansell of the University of Miami (https://hansell-lab.rsmas.miami.edu/consensus-reference-material/index.html, last access: 10 November 2020).

In total 650 samples were collected at 38 stations. Samples were collected at the following depths: 10, 25, 50, 75, 100, 150, 200, 300, 400, 500, 750, 1000, and every 250 m until reaching the ocean floor.

Chromophoric dissolved organic matter (CDOM) is the fraction of DOM that absorbs light at visible and ultraviolet (UV) wavelengths. It plays a key role in the marine ecosystem by regulating light penetration into the water column (Nelson and Siegel, 2013) and preventing cellular DNA damage (Herndl et al., 1993; Häder and Sinha, 2005). A fraction of CDOM re-emits part of the absorbed light and is called fluorescent DOM (FDOM). The study of the absorption properties of CDOM, together with the analysis of the excitation–emission matrixes (EEMs) through the parallel factorial analysis (PARAFAC) can give qualitative information on the different groups of chromophores (protein-like, humic-like and PAH-like) present in the DOM pool, their changes due to photodegradation and/or microbial transformation, the main sources of CDOM and an indirect estimation of its molecular weight and aromaticity degree (Stedmon and Nelson, 2015; Retelletti et al., 2015; Gonelli et al., 2016; Margolin et al., 2018). The CDOM data collected during the MSM72 cruise will represent an unique opportunity to (i) compare CDOM optical properties in the different water masses of the Mediterranean Sea with those collected in the GEOTRACES cruise (spring–summer 2013) and relate them to the different trophic conditions of the basin and to (ii) study the relationship between DOC and CDOM in the surface, intermediate, and deep waters.

Samples for the determination of stable carbon isotopes (δ13C) of dissolved inorganic carbon (DIC) were taken on 11 stations (the “isotope stations”, normally performed as a double cast) in the various basins along the cruise track. In total, 214 samples were taken in 100 mL dark glass bottles immediately poisoned with 100 µL saturated mercury chloride. The samples were measured off-line during the fall of 2018 at the Centre for Isotope Research (CIO), Energy and Sustainability Research Institute Groningen (ESRIG), University of Groningen.

Samples for nitrogen (N) and oxygen (O) isotopes in nitrate (NO3-) and nitrate + nitrite (NO3-+NO2-) analysis were collected at 44 stations evenly distributed along the transect. In total, 790 samples have been collected. High-resolution NO3-δ15N and δ18O measurements represent a powerful tool to unravel the sources and sinks of reactive (i.e., fixed) N at the scale of the Mediterranean Sea. Complemented with coral-bound δ15N records covering the last few centuries, these measurements may also shed light on the contribution of industrially fixed N to the reactive N budget by revealing the large-scale systematics required to interpret the records back in time.

Unfiltered samples for N and O isotopic composition of NO3- were collected in 60 mL plastic bottles and stored frozen (-20 ∘C) until analysis. NO3-+NO2-δ15N and δ18O will be measured (2019–2020) at the Max Planck Institute using the denitrifier method (Sigman et al., 2001; Casciotti et al., 2002). Briefly, 3–20 nmol of NO3-+NO2- is quantitatively converted to N2O gas by denitrifying bacteria (Pseudomonas aureofaciens) that lack an active N2O reductase. The N2O is then analyzed by gas chromatography–isotope ratio mass spectrometer (GC-IRMS; MAT253, Thermo) with online cryo-trapping (Weigand et al., 2016). Measurements are referenced to air N2 for δ15N and VSMOW for δ18O using the nitrate reference materials IAEA-NO3 and USGS-34. For NO3-δ15N and δ18O analysis, NO2- is removed with the sulfamic acid method prior to the isotopic analysis (Granger and Sigman, 2009). The reproducibility is generally better than 0.1 ‰ for δ15N and δ18O.

The LISST-Deep instrument obtains in situ measurements of particle size distribution, optical transmission, and the optical volume scattering function (VSF) at depths down to 3000 m. It is manufactured by Sequoia Inc. and owned by the Hellenic Centre for Marine Research (HCMR), Greece.

Using a red 670 nm diode laser and a custom silicon detector, small-angle scattering from suspended particles is sensed at 32 specific log-spaced angle ranges. This primary measurement is post-processed to obtain sediment size distribution, volume concentration, optical transmission, and volume scattering function. The LISST-Deep s/n 4004 is categorized as a type B instrument, which means that the range of particles it measures ranges from 1.25 to 250 µm. The LISST-Deep must be powered externally at all times. This is typically achieved by connecting it to a rosette, getting power from the main CTD unit.

Parameters measured during the cruise were as follows:

particle size distribution from 1.25–250 or 2.5–500 µm;

For the cruise MSM72 the sampling of these optical estimates is in itself an important achievement because, for the first time, LISST-DEEP was used to record data in a transect over the full length of the Mediterranean Sea. Furthermore, the estimation of these parameters, combined with Particulate Organic Carbon - Particulate Organic Nitrogen (POC-PON) estimation and other physical and chemical parameters, improves the study of the dynamics of the Mediterranean Sea.

In general, the use of LISST-DEEP during the cruise follows the standard methods provided by Sequoia Inc. but with one important difference. For the estimation of the above parameters the use of a background file is required for normalization purposes. This file is normally produced in laboratory conditions with MilliQ 2 filtered water. However, experience until now has proved that the use of this background file leads us to an overestimation of the parameters and especially of the beam attenuation coefficient, especially in the eastern Mediterranean Sea (which is characterized as ultra-oligotrophic). Therefore, during this cruise we used a sampled in situ background file chosen as the minimum of the sum of the digital counts in the 32 rings and where the LaserPower to LaserReference (Lp/Lr) ratio is at a maximum.

The main problem, which we faced, was the frequent change of the CTD main unit and the different cables that we had to use for the instrument connection to the CTD. Fortunately, with the most valuable help of the cruise technician, we managed to deploy the LISST-DEEP as much as possible. Additionally, the maximum depth limitation of the instrument (3000 m) forced us to remove it in deep casts, achieving a total of 54 stations.

The west–east section (Fig. 2) is a typical example for the distribution of temperature and salinity in the Mediterranean Sea showing the different heat and salt content between the western and eastern basin. A clear intrusion of the salty Levantine Intermediate Water (LIW) from east to west in the first 500 m is depicted, while the low-salinity Atlantic Water (AW) protrudes eastwards creating a front at about 20–22∘ E.

The underway CTD data are a valuable addition to the classical CTD data. They enhance the resolution of data in the horizontal scale and give insight into eddy activity. Although the data do not reach to the bottom, the vertical resolution with about 1000 m is useful to characterize scales relevant for the LIW transport.

The uCTD stations along the easternmost part of the northward transit in the Ionian Sea are taken every 5 nm on average. Larger gaps in the line were essentially caused by the deployment of CTD stations. The uCTD salinity distribution of Fig. 3 shows that the Pelops gyre is well resolved.

Considering the route of the ship during the cruise, it was possible to identify different ADCP transects that correspond to areas with the most important water mass dynamics. In particular, the most important sections were gyre activity in the area west of Crete and south of the Peloponnese; the west Cretan, Otranto (Fig. 4), and Sicilian straits; the eastern boundary of the Ionian Sea; and the west–east Mediterranean transect. The north–south current component (Fig. 4) in the Otranto Strait clearly shows the outflow of the Adriatic Deep Water (AdDW) along the western part, while in the upper and intermediate layer of the central part the inflow of the Levantine Intermediate Water (LIW) proceeds.

The vertical distribution of dissolved oxygen along a section from the Cretan Sea to Gibraltar, including part of the Cretan Passage and the southern Ionian Sea, is shown in Fig. 5. This section shows the Oxygen Minimum Layer (<180 µmoleskg-1), which occupies the layer 500–1500 m. Increased oxygen towards the bottom indicate the ventilation of deep water in the Mediterranean. The western part of the Ionian Sea appears to be better oxygenated than the eastern part due to the spreading of newly ventilated dense water from the Adriatic Sea via the Otranto Strait, a feature that is observed in the transient tracer section as well.

Figure 6 illustrates the distribution of nitrate along the quasi-zonal section. Interesting features include the maximum nutrient layer in the range of depth of 500–1500 m, which is co-located to the minimum of transient tracers. The deepest layer shows an homogeneous distribution of nutrients and the nutrient impoverished upper layer is not yet completely depleted of nutrients, likely due to mesoscale dynamics (as is the case, for example, south of Crete).

The DOC data collected during the MSM72 cruise represents an unique opportunity to (i) investigate the long-term variation in DOC distribution in intermediate and deep waters on a basin scale, (ii) quantify the role of DOC in C export and sequestration in the Mediterranean Sea, (iii) estimate DOC mineralization rates, and (iv) assess the functioning of microbial loop in the different areas of the Mediterranean Sea.

DOC concentrations range between 34 and 80 µM (Fig. 7). The highest values (>50 µM) were observed in the upper 200 m, with a marked increase moving eastward. The lowest concentrations (<40 µM) are between 1000 and 2000 m; in the bottom waters a slight increase in DOC can be observed. This feature, already reported for the Mediterranean Sea, can be explained by the export of the DOC accumulated in the surface layer by deep water formation (Santinelli, 2015, and references therein). The high stratification, occurring in the easternmost stations, makes DOC accumulation there more visible. A different functioning of the microbial loop has been reported for the western and eastern Mediterranean Sea and these data support that DOC dynamics in the surface layer of the two sub-basins is different.