A Lagrangian approach to the Atlantic Jet entering the Mediterranean Sea: Physical and biogeochemical characterization

Journal of Marine Systems 226 (2022) 103652 Available online 14 October 2021 0924-7963/© 2021 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Contents lists available at ScienceDirect Journal of Marine Systems journal homepage: www.elsevier.com/locate/jmarsys ALagrangian approach to the Atlantic Jet entering the Mediterranean Sea: Physical and biogeochemical characterization Iria Salaa,b,∗, Marina Bolado-Penagosc,∗∗, Ana Bartuala, Miguel Brunoc, Carlos M. Garcíaa, Ángel López-Urrutiad, Cristina González-Garcíae,f, Fidel Echevarríaa aDepartamento de Biología, Facultad de Ciencias del Mar y Ambientales, Instituto Universitario de Investigaciones Marinas (INMAR), Campus de Excelencia Internacional del Mar (CEI-MAR), Universidad de Cádiz, Puerto Real, 11510 Cádiz, Spain bDepartment of Mathematics and Statistics, University of Strathclyde, G1 1XH Glasgow, United Kingdom cDepartamento de Física Aplicada, Facultad de Ciencias del Mar y Ambientales, Instituto Universitario de Investigaciones Marinas (INMAR), Campus de Excelencia Internacional del Mar (CEI-MAR), Universidad de Cádiz, 11510 Puerto Real, Cádiz, Spain dInstituto Español de Oceanografía (IEO), Centro Oceanográfico de Gijón, 33212 Gijón, Asturias, Spain eDepartamento de Ecología y Gestión Costera, Instituto de Ciencias Marinas de Andalucía (ICMAN-CSIC), Puerto Real, 11510 Cádiz, Spain fDepartamento de Ecología y Biología Animal, Universidad de Vigo, 36210 Vigo, Spain ARTICLE INFO Keywords: Submesoscale processes Chlorophyll patchiness Phytoplankton Lagrangian experiment Strait of Gibraltar Atlantic Jet ABSTRACT The temporal evolution of the physical and biogeochemical properties of the Atlantic Jet (AJ) along the first ∼75 m of the water column during a 4-day journey was analysed by following the trajectory of a drifter dragged by the jet from the Strait of Gibraltar towards the Alborán Sea. Three stages were differentiated based on the evolution of several variables (e.g., velocity, temperature, nutrients, fluorescence). (i) Within the Strait of Gibraltar, the water column was primarily influenced by the tidal cycle, leading to a nutrientenrichment of surface waters. However, due to the short residence time, the phytoplankton community that was mainly dominated by diatoms, did not demonstrate significant changes. (ii) Once outside the Strait, the drifter trajectory was mainly influenced by the frontal dynamics associated with the AJ. The drifter moved forward along the jet but also laterally across it and was continuously attracted to the mainstream (maximum current speed) or detached to its southern edge (minimum current speed). Due to the associated upwelling processes induced by the intensification of the current along the mainstream, the water column was characterized by colder, nutrient-richer water and lower fluorescence values. Conversely, along the southern edge of the jet, the water column was characterized by higher temperature, low nutrient concentration, and higher fluorescence. Along the first stations of this stage, diatom total abundance and biovolume continuously increased, reaching values ∼12-times higher than the initial concentrations. (iii) In the last stage, the water parcel was still influenced by the frontal dynamics but with less intensity. Additionally, the colder and denser water of the AJ and the associated phytoplankton community subducted progressively as it moved into the region surrounded by warmer waters. Concomitantly, fluorescence and diatoms total abundance and biovolume decreased and were influenced by the decline of nutrient availability and the increase of mesozooplankton. Our results reveal the coupled processes induced by the entrance of the AJ in the Alborán Sea and highlight the strong control of the physical environment over the ecological processes in this region. 1. Introduction The Strait of Gibraltar is the narrow exchange passage connecting the Alborán Sea, the westernmost basin of the Mediterranean Sea, and the Gulf of Cádiz which opens to the north Atlantic Ocean. Water exchange is usually described as a typical two-layer inverse-estuarine circulation that is induced by the negative hydrological budget of the ∗Corresponding author at: Departamento de Biología, Facultad de Ciencias del Mar y Ambientales, Universidad de Cádiz, Puerto Real, 11510 Cádiz, Spain. ∗∗ Corresponding author at: Departamento de Física, Aplicada, Facultad de Ciencias del Mar y Ambientales, Universidad de Cádiz, Puerto Real, 11510 Cádiz, Spain. E-mail addresses: [email protected] (I. Sala), [email protected] (M. Bolado-Penagos). Mediterranean Sea (e.g., Armi and Farmer,1988). This circulation pattern leads to a natural tendency for oligotrophy in the Mediterranean basin that is associated with the surface flow towards the Alborán Sea of the relatively fresh and nutrient-poor Atlantic water (i.e., the Atlantic Jet) and a deep outflow of salty and nutrient-rich Mediterranean water towards the Atlantic Ocean (e.g., Armi and Farmer,1988). Three water masses are involved in this water exchange through the Strait https://doi.org/10.1016/j.jmarsys.2021.103652 Received 1 September 2020; Received in revised form 28 September 2021; Accepted 6 October 2021 Journal of Marine Systems 226 (2022) 103652 2 I. Sala et al. of Gibraltar: the Surface Atlantic Water (SAW: temperature ∼ 22.60 ◦C and salinity ∼36.40), the North Atlantic Central Water (NACW: temperature ∼ 13.60 ◦Cand salinity ∼35.80), and the Mediterranean Water (MW: temperature ∼ 13.50 ◦Cand salinity ∼38.40) (Gascard and Richez,1985;Gómez et al.,2001;Macías et al.,2008b). The Atlantic Jet (AJ) is the primary driver of the complex circulation of the Alborán Sea (e.g., Bormans and Garrett,1989;Renault et al., 2012), with an average velocity of ∼1 m s−1(García-Lafuente et al., 2000). The inflowing jet feeds the quasi-permanent Western Alborán Gyre (WAG; Fig. 1a) and the less intense and persistent Eastern Alborán Gyre (EAG) (e.g., Crepón,1965;Lacombe,1971;Lanoix,1974; Viúdez et al.,1996). Furthermore, the AJ-WAG system breeds a Coastal Cyclonic Gyre (CsCG) that arises close to the Spanish coast (Fig. 1a) and is associated with the upwelling of nutrient-rich waters (Sarhan et al.,2000;Macías et al.,2008a). This scheme, however, is far from being steady. In particular, the frequent occurrence of strong wind events (e.g., Vargas et al.,2003) and the high energetic tidal processes described for this area (e.g., García-Lafuente and Cano,1994) strongly modify this general pattern (Bolado-Penagos et al.,2021). Briefly, when westerly winds prevail (low sea level pressure over the Mediterranean Sea), the AJ increases its velocity and travels nearer to the Spanish coast (Candela et al.,1990;García-Lafuente et al.,2002b). In this situation, the horizontal extension of the WAG is enhanced, and the CsCG is constrained which reinforces the coastal upwelling (Macías et al.,2008a). In the opposite situation with prevailing winds from the east (high sea level pressure over the Mediterranean Sea), the AJ velocity decreases and travels farther from the Spanish coast (GarcíaLafuente et al.,1998). This southward displacement of the AJ results in the increase of the horizontal size of the CsCG and the development of an offshore upwelling (Sarhan et al.,2000;Macías et al.,2008a). On the other hand, the high energetic tidal processes that strongly influence the Strait of Gibraltar could also increase/decrease the magnitude of the AJ during the eastward/westward phase of the tidal cycle (e.g., GarcíaLafuente et al.,2002a;Romero-Cózar et al.,2021). Additionally, tides can destabilize and disengage the AJ-WAG system by enhancing the positive vorticity in the easternmost side of the Strait (Sánchez-Garrido et al.,2013). In an area as dynamic as the Strait of Gibraltar and the Alborán Sea, biological processes are tightly controlled by hydrodynamics. In situ (e.g., Ruiz et al.,2001) and remote sensing studies (e.g., Macías et al.,2007b;Navarro et al.,2011) have revealed enhanced levels of chlorophyll concentration along the AJ related to mesoscale and submesoscale frontal processes. Oguz et al. (2016), using a coupled physical–biological model that did not compute wind forcing nor tidal dynamics, confirmed that the intense frontal activity associated with the Atlantic inflow constitutes a major source of biological enrichment in the Alborán Sea. This is intensified by the tidal cycle that plays an important role in the nutrient supply of the Atlantic water inflow (Macías et al.,2007a;Ramírez-Romero et al.,2014;Sánchez-Garrido et al., 2015). The interaction of the tidal current with the sharp topography of the Strait of Gibraltar generates undulatory features at the Atlantic– Mediterranean Interface (AMI) that modify the AJ properties with a tidal periodicity (e.g., Macías et al.,2006;Bartual et al.,2011; Ramírez-Romero et al.,2012). During spring tides when the AMI is shallower, internal waves lead to the recirculation of nutrients from deep to shallower waters (Wesson and Gregg,1994;Macías et al., 2007a,2008a). During neap tides when the interface is deeper, a major contribution of nutrient-enriched NACW is expected (Gómez et al.,2001;Macías et al.,2006). Additionally, the influence of the tidal current could increase the phytoplankton biomass and primary productivity in the Alborán Sea by 40% (Sánchez-Garrido et al.,2015). All of these processes play an important role in the nutrient and biomass input to the Mediterranean Sea which, as previously mentioned, has a natural tendency to oligotrophy. Previous studies that have focused on the AJ have been performed using satellite imagery (e.g., Macías et al.,2007a;Renault et al.,2012), numerical modelling (e.g., Oguz et al.,2014;Sotillo et al.,2016), and in situ data collected with a Eulerian perspective (e.g., García-Lafuente et al.,1998;Gómez et al.,2001;Ruiz et al.,2001). There are also a number of studies with a Lagrangian perspective that analyse the path followed by the AJ. They are based on the release and monitoring of one or more drifting buoys that can be simple and record only position and time data or more complex and incorporate one or more sensors that record the variables of interest (e.g., temperature) depending on the objective of the study. García-Lafuente and Delgado (2004) analysed the trajectory, the velocity, and the surface temperature (∼ 1 m depth) of a mooring array that was accidentally released and was advected with the AJ around the WAG. Vélez-Belchí et al. (2005) released ten neutrally buoyant floats advected with the AJ to provide direct estimates of vertical velocity, temperature, and pressure. Sotillo et al. (2016) used the trajectory followed by several drifting buoys to evaluate the performance of several modelling systems in reproducing the circulation patterns of the Strait of Gibraltar and the Alborán Sea. More recently, Bolado-Penagos et al. (2020) employed Coastal Nomad drifters to characterize the surface circulation patterns from Cape Trafalgar (NW limit of the Strait) towards the Alborán Sea. However, to the best of our knowledge, there is no study that is focused on the AJ with a Lagrangian perspective that, in addition to analysing the trajectory followed by the drifter, studies the temporal and spatial evolution of the physical and biological properties along the water column. For that purpose, we have designed a Lagrangian experiment that involved the release of a drifter in the Strait of Gibraltar closely followed by an oceanographic vessel on its journey towards the Alborán Sea. The main goals of this study are (i) to describe and analyse in detail the changes in both water attributes and plankton community during this journey and (ii) to infer the physical processes that control the phytoplankton dynamics. 2. Data and methodology Sampling was conducted during the MEGAN cruise from October 1st to 6th, 2015, on board the R.V. Sarmiento de Gamboa during the last days of the spring tide (full moon on September 28th). During this sampling period, a drifting buoy (Offshore Nomad, SouthTEK Sensing Technologies Ltd) with a drogue at ∼5 m depth was released at the Strait of Gibraltar (5.61 ◦W – 35.99 ◦N) to identify the path followed by the inflowing AJ (Fig. 1). The drifter position was obtained from a GPS device, and its position was sent by Iridium satellite communications using the application LDManager (https://ldmanager.southteksl.com). Its position was recorded every ∼30 min, being possible to estimate its mean velocity from its displacements. The vessel closely followed this drifter, and a sampling station was demarcated each ∼2 h except at the end of the sampling period when the drifter velocity considerably decreased, and the stations were demarcated every ∼6 h (stations J43 to J46). On this 110-h journey, a total of 46 sampling stations were carried out (J01 to J46, Fig. 1a) covering a total of 315.62 km (see detailed information in Table A.1). During the entire cruise, current-velocity data were acquired using a vessel-mounted 75-kHz Teledyne RD Instrument Acoustic Doppler Current Profiler Ocean Surveyor. The temporal resolution of these measurements was 60 seconds, and the bin size was 8 m from a 16 to 800 m depth. This database was filtered (i) to exclude those values registered more than 1 km away from the trajectory of the drifter and (ii) to consider only data shallower than 120 m depth. Additionally, to accord with the drifter temporal resolution, the current-velocity data were averaged every 30 minutes. Wind data were also recorded throughout the entire cruise at the meteorological station on board the research vessel. Wind velocity and direction were recorded every minute. To fit the drifter temporal resolution, both were averaged every 30 minutes. Temperature,salinity,fluorescence (previously calibrated to measure chlorophyll in mg m−3) and photosynthetically active radiation (PAR) Journal of Marine Systems 226 (2022) 103652 3 I. Sala et al. Fig. 1. (a) Location of the different sampling stations (circles) with overlaid mean chlorophyll-aconcentration (mg m−3) for the sampling period (from October 1st to 6th, 2015). Black dotted contours display the mean sea surface temperature (◦C) computed for the sampling period. Both satellite data were downloaded from the Copernicus Marine Environmental Monitoring Service. The black square encloses a (b) zoom of the Strait of Gibraltar region. The white line represents the trajectory followed by the drifting buoy. The star represents the first sampling station of the survey (J01). Yellow diamonds depict the first station of each sampling day (J04, October 2nd; J15, October 3rd; J25, October 4th; J38, October 5th; J35, October 6th). Black diamonds indicate some reference stations (J10, J20, J30 and J40). The red line represents the section analysed in Section 3.3. CsCG: Coastal Cyclonic Gyre. WAG: Western Alborán Gyre. were recorded down to a depth of 200 m (except for shallower coastal stations J03 and J04, see Table A.1) in each of the 46 sampling stations using a combined CTD probe (Conductivity, Temperature and Depth; Seabird SBE 911 plus). However, only the first 120 m of the water column were analysed which ensured the observation of the AMI boundary that separates the inflowing Atlantic water and the outflowing Mediterranean water. Here, the AMI was associated with the 37.30 isohaline corresponding with the salinity value ascertained at the pycnocline depth from the CTD casts. From the different temperature and salinity profiles, the percentage of the NACW and the MW present in the water column were estimated based on a triangulation method following Bray et al. (1995). Moreover, the euphotic layer depth (Z𝑒𝑢) was defined as the depth at which the light intensity was attenuated to 1.00% of the value just below the surface. Thus, Z𝑒𝑢 was calculated as: 𝑍𝑒𝑢 =ln(0.01) 𝐾𝑑 (1) where K𝑑(m−1) is the vertical light attenuation coefficient calculated using the Beer–Lambert Law equation (Kirk,1994): 𝐼𝑧=𝐼0⋅𝑒−𝐾𝑑⋅𝑧(2) where I𝑧(μEinstein m−2s−1) is the intensity of radiation at any depth, I0(μEinstein m−2s−1) is the intensity of radiation at the surface, and z (m) is the depth. Due to the low variability of the PAR in this study, a mean value of Z𝑒𝑢 was computed for the whole period considering only the PAR profiles registered during the daylight hours. Additionally, discrete water samples were collected at each sampling station using a rosette system of 24 Niskin bottles with a volume capacity of 12-L each; this was accomplished at four different depths: 5, ∼50 and 75 m, and at the Depth of the Chlorophyll Maximum (DCM). The depth limit (i.e., 75 m) was selected considering our interest in studying the inflowing Atlantic water. Several variables related to the dynamics of the pelagic ecosystem were analysed as follows. For inorganic nutrients concentration, two replicates of 5 mL of filtered seawater (0.70 μm Whatman GF/F pre-combusted filters) were collected at each sampling station and depth and stored at -20 ◦C. Nitrate (NO3), nitrite (NO2), and phosphate (PO4) concentrations were quantified in the laboratory using an autoanalyzer (Technicon AAII-TRACS 800) following the techniques of Strickland and Parsons (1972). Samples for the estimation of picoand nanoplankton by cytometry were collected at each sampling depth and at all of the stations within the Strait of Gibraltar (J01 to J10) and thereafter only at the odd stations except the last three (J44 to J46) when the jet velocity considerably decreased. Samples were fixed with a solution of 1.00% glutaraldehyde and 0.05% formaldehyde and were immediately counted using a flow cytometer (Becton–Dickinson FacsAria). Laser calibration settings that were employed for the identification of picocyanobacteria and picoeukaryotes were set using natural algae cultures from the microalgae collection of the Instituto de Ciencias Marinas de Andalucía (ICMAN-CSIC). Large nanoplankton and microplankton samples (16 to 250 μm) were alternatively collected at odd stations along the entire cruise with the exception of the last three (J44 to J46) at the surface (5 m) and the DCM depths. The analysis of this size fraction was performed on two different water samples. For the smaller fraction (16 - 100 μm), 5 L were filtered and concentrated through a 10-μm mesh to a final volume varying between 100 and 150 mL and subsequently filtered through a 100-μm mesh to discard larger organisms. This fraction was analysed in vivo, on board, after water sampling using a FlowCAM®(Fluid Imaging Technologies) to count and measure all organisms. Following the same methodology for the larger fraction (100 – 250 μm), 10 L were filtered and concentrated to a final volume varying between 94 and 134 mL and discarding organisms larger than 250 μm. This fraction was fixed with buffered formaldehyde (4.00% final concentration), stored in amber glass flasks, and later analysed in the laboratory with FlowCAM. For both fractions, the FlowCAM was used in auto-trigger mode (Álvarez et al.,2011) with a chamber of 2.00 mm 𝑥0.10 mm and a x10 objective for the smaller fraction and a chamber of 3.00 mm 𝑥0.30 mm and a x4 objective for the larger fraction. In both cases, the entire sample was completely processed. The obtained images were sorted automatically applying a neural network with an Inception Resnet v2 architecture (Szegedy et al.,2016) using KERAS (https://keras.io/). Three neural networks based on the size fraction (i.e., cells with a diameter from 16 to 20 μm, 20 to 100 μm, and 100 to 250 μm) were trained to automatically classify the images obtained through the FlowCAM. Journal of Marine Systems 226 (2022) 103652 4 I. Sala et al. Fig. 2. (a) Temporal evolution of mean wind velocity (m s−1, black line) and direction (◦N; black dots) during the entire survey. (b) Temporal evolution of mean drifter velocity (m s−1; black line) and mean current velocity (m s−1) at ∼50 m depth (red line) during the whole survey. (c) Temporal evolution of the zonal component of the current velocity (u-vel; m s−1) during the entire survey. Positive/negative values indicate the current flows eastward/westward towards the Alborán Sea/Gulf of Cádiz. Black vertical lines represent the first station of each sampling day. The black horizontal line in (c) depicts the Atlantic–Mediterranean Interface (associated with the 37.30 isohaline). For each neural network, more than 20,000 images were manually classified by an expert to create each training set. Although each set had a different number of categories, in summary, the three size fractions were classified into 29 ‘functional types’ of living particles and three groups of non-living particles (Sala,2021). The global accuracy of the 16 to 20 μm training set was 71.05% (78.01% for living particles, 66.64% for non-living particles), 72.59% for the 20 to 100 μm training set (75.68% for living particles, 55.39% for non-living particles), and 72.39% for the 100 to 250 μm training set (74.92% for living particles, 70.71% for non-living particles). From a taxonomic perspective, these groups are merged into five ‘functional classes’ (diatoms, silicoflagellates, dinoflagellates, ciliates, and other unidentified living particles). Each category was associated with a simple geometrical shape thus the individual cell biovolume was calculated using the dimensions (i.e., diameter, length, and width) measured by the FlowCAM software. A previous volume correction was performed for fixed samples (100 - 250 μm) (Menden-Deuer et al.,2001). Total abundance and biovolume at 5 m depth and the DCM were averaged to obtain a mean value for surface layers. Finally, mesozooplankton samples were collected using a Bongo net (mouth opening 0.125 m2, mesh size 200 μm) by double-oblique tows at odd stations along the entire cruise except the last three (J44 to J46). Samples were collected down to a depth of ∼75 m (with the exception of the shallower coastal station J03 with 35 m). The net was towed for an average time of 7.16 ±1.34 minutes at ∼2 knots, filtering a mean volume of 60.15 ±8.50 m3. The retained material was fixed with formaldehyde (4.00% final concentration), stored in plastic bottles, and later analysed in the laboratory. Each mesozooplankton sample was strained onto a 200-μm mesh screen and split to obtain fractions with homogeneous abundances. Subsamples were spilt into a tray manually avoiding overlapping of organisms and photographed with a Nikon D810 camera (36-megapixel sensor) equipped with macro 1:1 lens. Zooplanktonic organisms were counted and measured by image analysis using ImageJ software (https://imagej.nih.gov/ij/). The biovolume of each organism was assumed to be equivalent to that of the corresponding ellipsoid (Jimenez et al.,1987;Alcaraz et al.,2003). A previous volume correction due to shrinkage was performed (Black and Dodson,2003). 3. Results 3.1. The drifter trajectory The drifter travelled with an average velocity of 0.51 ±0.37 m s−1with a maximum value of 3.78 m s−1at the exit of the Strait of Gibraltar (stations J06 to J08) and a minimum <0.01 m s−1before leaving the Strait (between stations J06 and J07) and in the vicinity of the Djibouti Plateau (J46, Fig. 2b). During the first ∼10 h of the survey under easterly wind conditions, the drifter described a 360-degree counterclockwise gyre (J01 to J06, Fig. 1b) before leaving the Strait of Gibraltar. From J02 to J04, the zonal component of the current velocity (u-vel) was negative (westward direction) evidencing the entrance of the drifter into the Strait (Fig. 2c). In the next hours, drifter and water column velocity progressively increased and achieved the maximum positive values (eastward direction) of the entire survey (>2.00 m s−1; J06 to J08, Fig. 2b,c) that drove the drifter out of the Strait. Once the drifter left the Strait, its average velocity did not exceed 1.00 m s−1but varied throughout its journey (Fig. 2b). Between stations J08 and J28, the drifter’s average velocity fluctuated between ∼1.00 and 0.50 m s−1 and thereafter fluctuated between ∼0.50 and <0.01 m s−1during the Journal of Marine Systems 226 (2022) 103652 5 I. Sala et al. Fig. 3. Temporal evolution of (a) temperature (◦C) and (b) salinity during the entire survey. The horizontal line (white in (a) and black in (b)) represents the Atlantic–Mediterranean Interface (associated with the 37.30 isohaline). Black vertical lines indicate the first station of each sampling day. Dotted white lines represent CTD vertical sampling profiles. Fig. 4. Temporal evolution of the proportion of (a) North Atlantic Central Water (NACW, %) and (b) Mediterranean Water (MW, %) during the entire survey. The black line represents the Atlantic–Mediterranean Interface (associated with the 37.30 isohaline). Black vertical lines depict the first station of each sampling day. Dotted white lines indicate CTD vertical sampling profiles. rest of the journey. Five clear peaks of maximum drifter velocity were observed (Fig. 2b). The first three peaks reached maximum values of approximately 1.00 m s−1and the next two had a maximum velocity of ∼0.50 m s−1(Fig. 2b). The water column also experienced a progressive decrease in its current velocity along the survey as the drifter moved away from the Strait of Gibraltar. The upper ∼75 m of the water column (AMI depth = 72.26 ±13.28 m) corresponding with the inflowing Atlantic water flowed with a mostly positive velocity towards the Alborán Sea (red values, Fig. 2c). Meanwhile, the deepest layer corresponding with the MW flowed with a primarily negative velocity (blue values, Fig. 2c). However, the waving pattern outlined by the drifter velocity was not easily appreciable through the water column. Compared to previous and subsequent values, u-vel between J15 and J18 stations strongly decreased and reached values of <0.25 m s−1. This zonal velocity Journal of Marine Systems 226 (2022) 103652 6 I. Sala et al. decrease corresponded with an abrupt change in the trajectory of the drifter (Fig. 1a) that mainly moved in a N-NE direction. 3.2. Physical structure of the water column As the drifter advanced towards the Alborán Sea, the temperature of the inflowing Atlantic water moderately increased, reaching maximum values at the end of the journey (∼22 ◦C) while no variation of temperature of the MW was noticeable (<16 ◦C; Fig. 3a). On the other hand, the salinity of the inflowing Atlantic water varied little (disregarding J46) but slightly increased on the MW towards the Alborán Sea (Fig. 3b). The temporal evolution of temperature and salinity vertical profiles showed an undulatory pattern from station J10 (outside the Strait) to station J46. This behaviour was also appreciable with the variation of the AMI depth which fluctuated during the whole survey with an average depth of 72.26 ±13.28 m. As observed in Fig. 4a, the NACW was mainly distributed in the first 60 – 80 m of the water column with a maximum contribution inside the Strait (>60%, J01). Once outside the Strait (J11 to J46), its contribution was fairly constant with a mean value of 40.53 ±2.97%. Although, at several stations, spots with a slightly higher percentage (∼ 55%) were observed. Additionally, the percentage of MW demonstrated a progressive increase from the surface to deep waters. Along the first ∼75 m of the water column, its percentage varied from ∼5 to 50% close to the AMI (Fig. 4b). 3.3. Physical description of the AJ frontal dynamics The visual analysis of the u-vel variability along the across-jet section indicated in Fig. 1 (red line), allows identifying the location and width (7.78 km) of the AJ (Fig. 5a) to describe the characteristic frontal dynamics. Moreover, the across the jet relative vorticity (𝜁, s−1) was estimated as: 𝜁≡𝑣𝑥−𝑢𝑦(3) where u and v are the velocity components in 𝑥and 𝑦(i.e., east and north) directions. 𝜁was positive (anticlockwise motion) to the north and negative (clockwise motion) to the south (Fig. 5b). 𝜁reached absolute values as high as 2 ⋅10−4s−1. Consequently, the local Rossby number (R0, dimensionless) estimated as: 𝑅0=𝜁 𝑓(4) with fdenoting the planetary vorticity (Coriolis parameter, 8.5 ⋅10−5 s−1for 36 ◦N), yields a value of 2.35. With this R0value, the jet was prone to lose geostrophic balance. Therefore, to restore this frontal geostrophic balance, a secondary ageostrophic circulation is developed along the across-jet dimension (Mahadevan,2016). This secondary ageostrophic circulation is associated with the local increasing of vertical velocity (W, m s−1) along the edges of the jet. Thus, vertical velocity is scaled in accordance to: 𝑊∼𝑅0⋅𝛿⋅𝑈(5) with Ubeing the characteristic current intensity of the jet (u-vel, m s−1) and 𝛿, the small ratio of the Coriolis parameter (f) to the buoyancy frequency (N, s−1), arising from density stratification: 𝛿=𝑓 𝑁(6) 𝑁=√−𝑔 𝜌0 ⋅ 𝑑𝜌 𝑑𝑧 (7) With gbeing the gravitational acceleration (9.80 m s−2), 𝜌0the initial density of the water mass, and d𝜌/dz the relationship between density and depth along the first 100 m of the water column from J10 to J42 (see Fig. A.12), the resulting Nhas a value of 0.016 s−1. Considering U= 0.82 m s−1(maximum u-vel along this across section, Fig. 5a) and N= 0.016 s−1, the estimated vertical velocity is as large as 0.01 m s−1. For the AJ, moving eastward, this secondary circulation induces upward/downward vertical velocities in the southern/northern edge of the jet (Mahadevan,2016). Another characteristic behaviour of frontal dynamics associated with the secondary ageostrophic circulation is the development of baroclinic instabilities that cause the generation of frontal undulations (i.e., meanders). These meanders reinforce the horizontal density gradients across the jet (isopycnals become closer) intensifying the jet current speed and favouring the loss of the geostrophic balance. For the AJ, flowing eastward, due to vorticity readjustment, a water parcel Fig. 5. Latitudinal variation of (a) the zonal component of current velocity (u-vel, m s−1) and (b) the vertical component of relative vorticity (s−1) at 16 m depth along the across-jet section identified in Fig. 1a (red line). The red vertical line identifies the point of maximum u-vel (0.82 m s−1). Journal of Marine Systems 226 (2022) 103652 7 I. Sala et al. moving near the southern edge will tend to be attracted towards the mainstream of the jet when isopycnals are bent to the north and will gain velocity. However, a water parcel will tend to be detached from the jet when the isopycnals are bent to the south and lose velocity (Cushman-Roisin and Beckers,2011). This behaviour may explain the oscillatory changes in the drifter speed (Fig. 2b); higher/lower drifter speeds would indicate closer/farther distances to the mainstream of the jet. 3.4. Dynamics of nutrients and the pelagic community To study the dynamics of the water attributes and the pelagic community, we considered the first ∼75 m of the water column aiming to sample only the inflowing Atlantic water. However, the AMI was regularly observed affecting the deepest water samples at several stations (black horizontal line, Fig. 6). The NO3+ NO2concentration (hereafter NO𝑥,μM) varied from < 0.01 to ∼ 7.00 μM except for station J04 at 25 m depth where the maximum concentration for the entire survey was registered (9.42 μM) (Fig. 6a). As expected, the NO𝑥increased with depth along the survey with a slight decrease as the drifter advanced in space and time. The analysis of comparative availability of nutrients in surface layers has been made by tracking the position of the NO𝑥-nutricline depth; the first depth at which the NO𝑥>1.00 μM (Fig. 6a; Morán et al.,2001). The first stations in the survey had a clear availability of nutrients (J01 to J10). However, from station J11 onwards, a surface water layer with the NO𝑥concentration <1.00 μM was observed (above the white line on Fig. 6a), fluctuating and deepening with time, and achieving its maximum depth at station J44 (∼35 m depth). This undulating nutrient pattern is associated with the rise of isotherms and isohalines that results in the presence of the NO𝑥-rich waters in the most superficial layers of the water column. These inputs were less evident at the final stations of the survey. The PO4concentration was clearly higher inside the Strait of Gibraltar at the most coastal stations (J02 to J04, Fig. 6(b) with values of > 0.60 μM. However, once the drifter left the Strait, the PO4concentration decreased, varying between <0.01 and ∼ 0.50 μM although with a heterogeneous pattern. The PO4-nutricline is defined as the first depth at which the PO4>0.10 μM (Morán et al.,2001). In this case, concentration on surface waters reached values of <0.10 μM PO4 (above the white line on Fig. 6b) from station J14 onwards. The depth reached by this layer was close to the depth of the euphotic layer (52.66 m) at stations J32 and J45. The PO4concentration in the water column considerably decreased from J37 to the end, a fact that could indicate a transition to more oligotrophic waters. The association of the previously described undulatory pattern to the isotherms rising was more difficult to appreciate for the PO4except at stations J23, J29, and J35. Fig. 7 illustrates the temporal evolution of fluorescence on the Atlantic water layer as a proxy for chlorophyll concentration. To more effectively note the undulatory pattern previously observed, mean fluorescence for the first ∼75 m of the water column is represented in Fig. 7a while, in Fig. 7b, its depth distribution can be examined. Inside the Strait of Gibraltar (J01 to J08), fluorescence was low with mean values of <0.80 mg m−3except at coastal station J03 with 1.12 mg m−3(Fig. 7a). The vertical distribution of fluorescence showed a higher concentration between stations J12 and J43 in the first ∼50 m of the water column limited by the euphotic layer (52.66 m) where values between 1.50 and 3.00 mg m−3were recorded (Fig. 7b). During this period (J12 to J43), mean fluorescence described an oscillation with time, although it is from station J19 that the undulatory pattern varied according to the variables previously described (Figs. 3 and 6). However, fluorescence tended to decrease locally when the isotherms ascended, and there was a higher nutrient concentration in shallower layers. Thus, the maximum mean fluorescence peaks and the high fluorescence spots were observed when the isotherms descended and nutrient concentration in surface layers were low. These high fluorescence spots deepened as the drifter progressed into the Alborán Sea, coinciding with the progressive disappearance of the NO𝑥in surface waters (Figs. 6a and 7b). The two groups of autotrophic picoplankton analysed in this study showed, as expected, an opposite distribution (see Fig. 8). Prochlorococcus was the most abundant group (Fig. 8b) with values of >10,000 cells mL−1during most of the survey. Several pulses of abundance maxima (>25,000 cells mL−1) were recorded along the temporal evolution of the water parcel and at different depths (i.e., J06, J21, J23, J25, and J43; Fig. 8a). Some of them (e.g., J25 and J43) coincided with fluorescence maxima (Fig. 7b). Synechococcus was the less abundant group with values of ∼5,000 cells mL−1for almost the entire survey (Fig. 8b). Although, at coastal station J04 and the exit of the Strait of Gibraltar (i.e., J06 to J09), it reached values of ∼15,000 cells mL−1 between 5 and ∼40 m depth. In the surface waters of J46 station, Synechococcus attained its maximum abundance (19,385.54 cells mL−1; Fig. 8b). However, an undulatory pattern similar to that observed for the previous variables was not clearly detected for the two groups of autotrophic picoplankton. Regarding large-sized phytoplanktonic fractions, diatom abundance was the highest with at least 69% of the total abundance in all sampling stations while its contribution to the total biovolume was >96% (see Fig. A.13). Thus, hereafter, only the diatoms details are shown. Among the diatoms, the 15 identified categories (Sala,2021) were classified into six different classes according to their abundance: Chaetoceros spp.,Guinardia spp.,Leptocylindrus spp.,Skeletonema spp., other centric diatoms (i.e., Thalassiosira spp.,Rhizosolenia spp.), and pennate diatoms (i.e., Asterionellopsis spp.,Pseudo-nitzschia spp.). The contribution of the different groups to the total diatom population are shown in Fig. 9a. The two most abundant groups were ‘other centric diatoms’ ranging from 34.37% (J17) to 75.35% (J43) of total diatom abundance and Chaetoceros spp., varying between 8.86% (J43) and 39.56% (J25) (disregarding J01). The contribution of the remaining classes of diatoms varied between quite similar ranges but without a clear trend. Both diatoms’ total abundance and biovolume showed a similar trend as the drifter travelled into the Alborán Sea (Fig. 9b) and, as observed previously with other variables, there was an oscillation between minimum and maximum values. Inside the Strait of Gibraltar (J01 to J09), the maximum value of both variables was observed at coastal station J03 (maximum PO4concentration, Fig. 6b) with a value ∼5 times higher than that observed in J01 and ∼6-times higher than that observed in J07 (minimum value for the entire experiment). Once outside the Strait, the diatom abundance and biovolume continuously increased during the next ∼20-hours of the survey, achieving the maximum value at J17 (∼12-folded the initial value). During the remainder of the survey, from J17 to J46, a decreasing trend was observed with two sharp lows at J19 and J46. As previously observed with mean fluorescence, this decreasing trend matches with the progressive disappearance of the NO𝑥in surface waters (Fig. 6a). Finally, both mesozooplankton total abundance and biovolume demonstrated the same trend, gradually increasing from the beginning to the end of the Lagrangian experiment (Fig. 9c), 4-folding the initial values. This trend showed a series of continuous ups and downs that do not seem to be related to the daily vertical migrations of these organisms. 4. Discussion To adequately interpret the results obtained in this study, it should be emphasized that the experimental approach did not allow the same water mass to be followed in space and time. Notwithstanding, it did afford us the opportunity to analyse the temporal and spatial evolution of the water properties being introduced through the Strait of Gibraltar with a common origin. Furthermore, it is necessary to take into account that, when the Atlantic water is driven into the Alborán Journal of Marine Systems 226 (2022) 103652 8 I. Sala et al. Fig. 6. Temporal evolution of (a) nitrate + nitrite (NO𝑥), and (b) phosphate (PO4) concentration (μM) during the entire survey. The black line represents the Atlantic–Mediterranean Interface (associated with the 37.30 isohaline). The white line indicates the minimum depth in which (a) the NO𝑥concentration is >1.00 μM, and (b) the PO4concentration is >0.10 μM. Black vertical lines represent the first station of each sampling day. Dotted white lines identify CTD vertical sampling profiles. White circles represent the sampling depths. White asterisks depict the depth of the chlorophyll maximum. Fig. 7. Temporal evolution of (a) mean fluorescence (mg m−3) computed for the first 75 m of the water column and (b) vertical fluorescence distribution (mg m−3) during the entire survey. Black vertical lines represent the first station of each sampling day. In (b), the white line indicates the Atlantic–Mediterranean Interface (associated with the 37.30 isohaline), the black line signifies the minimum depth in which nitrate + nitrite concentration is >1.00 μM, and dotted white lines represent CTD vertical sampling profiles. Sea by the AJ, the intensity of the jet stream decreases significantly with depth (Cushman-Roisin and Beckers,2011). Consequently, the drifter displacement induced by the jet dynamics will not coincide with those experienced by subsurface water layers. In Fig. 2b, the drifter speed (black line) can be compared with the speed registered at ∼ 50 m depth (red line) along the entire study. Within the Strait at both depths (5 and ∼50 m), the speed was quite similar, however, once in the Alborán Sea, both values differed during the remaining journey. On the other hand, the Atlantic water would be continuously modified by lateral mixing with the surrounding Mediterranean waters. Mixing processes will be more evident along the edges of the jet. Hence, from a combined physical and biochemical perspective, the temporal and spatial evolution of the AJ resembles, to a certain extent, the progression of an upwelling filament: an intrusion of upwelled water into the ocean separated from the surrounding water by fronts (e.g., Wasmund et al.,2016). Three stages were differentiated according to the most influential processes that induced the temporal and spatial evolution of the physical and biochemical characteristics of the AJ. Journal of Marine Systems 226 (2022) 103652 9 I. Sala et al. Fig. 8. Temporal evolution of (a) Prochlorococcus, and (b) Synechococcus abundance (cells mL−1) during the entire survey. The black line represents the Atlantic–Mediterranean Interface (associated with the 37.30 isohaline). The white line indicates the minimum depth in which nitrate + nitrite concentration is >1.00 μM. Black vertical lines depict the first station of each sampling day. Dotted white lines identify CTD vertical sampling profiles. White circles signify the sampling depths. White asterisks represent the depth of the chlorophyll maximum. Fig. 9. Temporal evolution of (a) the biovolume composition of the diatom assemblage (%), (b) diatoms total abundance (ind. m−3; red line) and biovolume (μm−3m−3; black line), and (c) zooplankton total abundance (ind. m−3; red line) and biovolume (μm−3m−3; black line) during the entire survey. Black vertical lines in (b) and (c) represent the first station of each sampling day. Black stars in (c) indicate the Bongo samples collected overnight.