Coupling between upper ocean layer variability and size-fractionated phytoplankton in a non-nutrient-limited environment

MARINE ECOLOGY PROGRESS SERIES Mar Ecol Prog Ser Vol. 499: 35–46, 2014 doi: 10.3354/meps10668 Published March 3 INTRODUCTION The dynamics of phytoplankton in the upper ocean is linked to both the upper mixed layer (UML) and to mesoscale and submesoscale variability. The UML is the result of complex interactions between atmospheric forcing and this variability (e.g. Ferrari & Boccaletti 2004). For example, mesoscale eddies may cause deepening or shoaling of the UML, thereby influencing the turbulent mixing and affecting the © Inter-Research 2014 · www.int-res.com*Corresponding author: [email protected] Coupling between upper ocean layer variability and size-fractionated phytoplankton in a non-nutrient-limited environment Pablo Sangrà1,*, Cristina García-Muñoz2, Carlos M. García3, Ángeles Marrero-Díaz4, Cristina Sobrino5, Beatriz Mouriño-Carballido5, Borja Aguiar-González4, Cristian Henríquez-Pastene6, Ángel Rodríguez-Santana4, Luis M. Lubián2, Mónica Hernández-Arencibia4, Santiago Hernández-León1, Elsa Vázquez5, Sheila N. Estrada-Allis4 1Instituto Universitario de Oceanografía y Cambio Global, Universidad de Las Palmas de Gran Canaria, 35017 Las Palmas de Gran Canaria, Spain 2Departamento de Ecología y Gestión Costera, Instituto de Ciencias Marinas de Andalucía (ICMAN-CSIC), 11510 Puerto Real, Cádiz, Spain 3Departamento de Biología, Facultad de Ciencias del Mar y Ambientales, Universidad de Cádiz, 11510 Puerto Real, Cádiz, Spain 4Departamento de Física, Universidad de Las Palmas de Gran Canaria, 35017 Las Palmas de Gran Canaria, Spain 5Departamento de Ecoloxía e Bioloxía Animal, Universidade de Vigo, 36200 Vigo, Pontevedra, Spain 6Departamento de Geofísica, Universidad de Concepción, 160-C Concepción, Chile ABSTRACT: We describe the coupling between upper ocean layer variability and size-fractionated phytoplankton distribution in the non-nutrient-limited Bransfield Strait region (BS) of Antarctica. For this purpose we use hydrographic and size-fractionated chlorophyll adata from a transect that crossed 2 fronts and an eddy, together with data from 3 stations located in a deeply mixed region, the Antarctic Sound (AS). In the BS transect, small phytoplankton (<20 µm equivalent spherical diameter [ESD]) accounted for 80% of total chl aand their distribution appeared to be linked to cross-frontal variability. On the deepening upper mixed layer (UML) sides of both fronts we observed a deep subducting column-like structure of small phytoplankton biomass. On the shoaling UML sides of both fronts, where there were signs of restratification, we observed a local shallow maximum of small phytoplankton biomass. We propose that this observed phytoplankton distribution may be a response to the development of frontal vertical circulation cells. In the deep, turbulent environment of the AS, larger phytoplankton (>20 µm ESD) accounted for 80% of total chl a. The proportion of large phytoplankton increases as the depth of the upper mixed layer (ZUML), and the corresponding rate of vertical mixing, increases. We hypothesize that this change in phytoplankton composition with varying ZUML is related to the competition for light, and results from modification of the light regime caused by vertical mixing. KEY WORDS: Physical-biological coupling · Mesoscale · Submesoscale · Vertical mixing · Phytoplankton composition · Antarctica Resale or republication not permitted without written consent of the publisher F REE REE A CCESS CCESS Mar Ecol Prog Ser 499: 35–46, 2014 rate of vertical mixing of phytoplankton (e.g. Thompson et al. 2007). One major source of submesoscale variability is small filamentous fronts that originate at eddy boundaries or at mesoscale fronts. Intense vertical velocity associated with vertical circulation cells may cause the fronts to slump, resulting in restratification and enhancement of turbulent mixing (Thomas et al. 2008, D’Asaro et al. 2011). Vertical circulation cells related to meso scale and submesoscale variability, mixing processes, and their interaction may modify the vertical ex changes of tracers and hence influence phytoplankton dynamics (e.g. Nagai et al. 2006, Klein & Lapeyre 2009). Vertical mixing may influence light availability, nutrient entrainment and uptake, zooplankton en - counter probability, and settling velocity (e.g. Peter - sen et al. 1998, Huisman et al. 2002, Man & Lazier 2006, Kiørboe 2007, Behrenfeld 2010). Turbulence is not always active in the UML and therefore it is important to distinguish between the terms UML and ‘mixing layer’. The mixing layer is also a uniform density layer but where turbulence is active throughout (Brainerd & Gregg 1995). The vertical extent of the upper mixing layer is particularly important for phytoplankton modulation by turbulence, because it determines the rate of vertical mixing of the cells. In a non-nutrient-limited environment it may be ex pec - ted that this rate will essentially control the position and movement of the cells in the light gradient, as well as the zooplankton-phytoplankton encounters that can reduce/increase the grazing pressure. Since the early Sverdrup (1953) critical depth hypo - thesis, many studies have used either the mixing layer depth or the upper mixed layer depth (ZUML) as proxies for turbulence forcing. The effects of this forcing in less complex non-nutrient-limited environments are twofold: modulation of the population net growth rate, and modulation of the composition of the plankton community (e.g. Petersen et al. 1998, Huisman et al. 2004, Behrenfeld 2010). Hewes et al. (2008) studied the relationship between hydrographic properties, nutrients and phytoplankton biomass using data collected during a summer cruise in a region near the South Shetland Island that includes our study region (Fig. 1). In areas of high iron concentrations they observed an inverse relationship between ZUML and total chl a, but with a very low coefficient of determination (R2 = 0.176). When climatological data for the region are considered (Hewes et al. 2009), this relationship is much more robust (R2= 0.73). Climatological data analysis also showed that blooms in the region co-vary with shallow UML depths (e.g. Mitchell & Holm-Hassen 1991, Hewes et al. 2009). Contrary to these observations, Holm-Hansen et al. (2004) did not observe any signi ficant correlation between ZUML and phytoplankton abundance in a much broader survey. They concluded that low iron concentrations are the major factor controlling phytoplankton biomass. Recent observations by Mendes et al. (2012) in the vicinity of James Ross Island (Weddell Sea) show bloom levels of phytoplankton (~5 mg chl am−3) in deep mixed layers (ZUML ~ 80 m) and poorly stratified water. Several studies in our surveyed region have also examined the possible relationship between the mixing layer depth/water stratification and phytoplankton composition (e.g. Kopczynska & Ligowsky 1985, Kang & Lee 1995, Kang et al. 2001, Mendes et al. 2012, 2013). Kopczynska & Ligowsky (1985) first suggested a connection between variability of phytoplankton composition and water mass properties for this region. Kang & Lee (1995) observed the dominance of nano-sized flagellates in the more stratified waters of the Bransfield Strait, whereas nanoand 36 –59.5 –59 –58.5 –58 –57.5 –57 –56.5 –56 –62 –62.5 –63 –63.5 –64 Longitude Latitude Fig. 1. Location of sample stations for the study of upper ocean layer variability and size-fractionated phytoplankton distribution in the Bransfield Strait (Stns 1 to 12) and Antartic Sound (Stns 19, 21 and 22), Antarctica. Blue dots: stations where standard measurements were taken; red dots: stations additionally sampled with a microstructure turbulence profiler. Gray lines are isobaths with a contour interval of 100 m Sangrà et al.: Ocean surface layer and phytoplankton dynamics micro-sized dia toms dominate in the more homogeneous waters of Drake Passage. Recent results from 3 summer data collections by Mendes et al. (2013) showed that dia toms dominate in deeper mixed layer conditions whereas nano-sized cryptophytes dominate in more stratified conditions. Their February 2010 observations across Bransfield Strait are of particular interest because they almost coincided in time and location with the present study. As depicted in their Fig. 6, there is a clear correlation between the variability of the ZUML and phytoplankton composition along the cited transect. For stratified environments (ZUML ~ 15 to 25 m) nano-sized phytoplankton dominated. In a homogeneous water column with a very deep mixed layer (ZUML > 160 m) diatoms dominated instead. This study aimed to explore observationally how the distribution and composition of phytoplankton are connected with the variability of the ocean sur face layer. The observations by Mendes et al. (2013), although made in the same area surveyed by us, were, however, mainly decoupled from the mesoscale variability, as their spatial resolution well exceeded the mean local first baroclinic Rossby radius of deformation (Rd) (~10 km; Chelton et al. 1998). Therefore, these observations do not resolve the effects of mesoscale variability and their interaction with the UML on the plankton dynamics. In January 2010, during the mid-austral summer, an interdisciplinary survey was made over a region located near the South Shetland Island at mesoscale resolution (stations ~9 km apart, Fig. 1). The general coupling between the mesoscale and phytoplankton distributions and their taxonomic composition has already been outlined in the study by García-Muñoz et al. (2013) from a more detailed biological perspective. In the present study we focused attention on the processes underlying this coupling. Particular em pha sis was given to the Bransfield Strait region (Fig. 1), where distinct water masses meet and there is a rich mesoscale variability (Sangrà et al. 2011). Moreover nutrient concentrations in this southern region were above limiting thresholds for phytoplankton growth, both during this cruise and in past observations (Hewes et al. 2009, Teira et al. 2012). This allowed our analysis of the effects of the physical environment to focus on light rather than nutrient availability as the main factor affecting phytoplankton growth in conditions of vertical mixing. Both circumstances make the selected region suitable for the study of the coupling between ocean surface layer variability and phytoplankton dynamics. MATERIALS AND METHODS Physical data collection Data were acquired during the interdisciplinary cruise COUPLING conducted in January 2010 on board RV ‘Hespérides’, during the austral midsummer near the South Shetland Islands. This survey formed part of the project COUPLING (Physical-biological coupling at the mesoscale range around the South Shetland Islands) of the Spanish Antarctic Program. For this study we selected 1 transect of 12 stations across the Central Bransfield Strait and 3 stations (Stns 19, 21 and 22) in the Antarctic Sound (Fig. 1). Distinct water mass compositions and rich mesoscale variability were the motivation for this selection. In order to resolve the mesoscale, stations were sampled every 5 nautical miles (~9 km). The transect sampling can be viewed as quasi-synoptic as <3 d (8 to 11 January 2010) were required to complete the transect. At each station, vertical profiles of temperature, salinity and in situ fluorescence were obtained using a Seabird 911plus combined conductivity, temperature and depth sensor (CTD) with a Seapoint fluorescence sensor attached to a rosette system of 24 oceanographic 12 l Niskin bottles. We also recorded microstructure profiles with a micro - structure turbulence profiler (TurboMap; Wolk et al. 2002) at 3 stations of the transect (Stns 2, 6 and 9), and at the 3 stations (13 to 14 January 2010) in the Antarctic Sound (Fig. 1). Three casts were conducted at each station and data were depth averaged within 8 m bins. The dissipation rate of turbulent kinetic energy εwas estimated from the measured vertical microstructure shear and derived from the shear variance by integrating the power spectrum of the velocity shear between length scales of 2 cm and half the Kolmogorov scale. A correction was made to recover the unresolved variance using the Nasmyth empirical spectrum (Oakey & Elliott 1982). As an indicator of turbulent mixing we computed the Thorpe scales (Thorpe 1977) from the CTD casts using a method analogous to those of Gargett & Garner (2008). We first removed spikes in salinity, re - moved pressure reversal, and depth-averaged data at each half meter. Second, we sorted the density profiles in order to obtain statically stable profiles without inversions. We then calculated the Thorpe displacements as the vertical shifts of data points needed to achieve static stability. Finally the Thorpe scale was calculated from the root mean square of an ensemble of Thorpe displacements as computed at successive non-zero Thorpe displacements. 37 Mar Ecol Prog Ser 499: 35–46, 2014 ZUML was inferred from CTD profiles using the Kara et al. (2000) algorithm. First, the potential density anomaly at 10 m is chosen as an initial reference density σθref. Second, a search is made of the potential density anomaly profile for regions of uniform potential density anomaly. These regions were de - fined as any pair of adjacent values where the potential density anomaly variation Δσθ, corresponding to a change of potential temperature of 0.8°C, was <0.1. If a uniform region is found, σθref is updated with the values corresponding to the shallower depth of the pair of profile points. This procedure is applied for every occurrence of pairs of points within the first uniform potential density anomaly region so that σθref is that at the base of the mixing layer. Then, the mixing layer depth is the depth at which the potential density anomaly has changed by an absolute value of Δσθfrom this reference value. For each station a vertical light attenuation coefficient kd(m−1) was calculated by measuring photo - synthetically available radiation (PAR, wavelength 400–750 nm) values at 1 m depth intervals in the water column with a hemispherical quantum sensor (CI PAR, Chelsea Instruments, relative spectral sensitivity flat to +3% from 450 to 700 nm). The euphotic layer depth (ZEU) was defined as the depth at which the light intensity was attenuated to 1% of its value just below the surface, and was calculated as ZEU = ln(0.01)/kd. Values of kdwere calculated over the depth that PAR measurements are reliable as shown by Vaillaincourt et al. (2003). Current velocities were measured continuously using a hull-mounted 75 KHz acoustic Doppler current profiler (ADCP; Teledyne RD Instruments). We use broadband raw data with 2 min ensembles from the surface to ~500 m depth with a bin size of 8 m. Raw current velocity data were processed using the CODAS software package (Firing 1991); ship mo - tions were removed. At each station in the Bransfield Strait we averaged the ensembles along the first 10 min and last 10 min of the CTD-rosette complete cast, whereas for the Antarctic Sound we used continuously recorded data along the 30 h sampling interval. We decided not to remove the signal of the barotropic tide as the tidal model used predicts that it is very low in this area (Padman et al. 2002). Chlorophyll adata Chl aconcentrations were determined fluorometrically; 250 ml water samples were sequentially filtered through 20, 2 and 0.2 µm polycarbonate membrane filters and pigments were extracted overnight in 90% acetone at −20°C. We discriminate between small and large phytoplankton as a function of their corresponding size-fractionated chl aconcentrations resulting from the above sequential filters. Small phytoplankton includes concentrations for size fractions <20 µm (0.2 to 20 µm) and large phytoplankton concentrations for size fractions >20 µm. Fluorescence was measured using a Turner TD-700 fluoro meter which had been calibrated with pure chl afollowing UNESCO (1994) standard protocol. To calibrate fluorescence measurements we followed Cuttelod & Hervé (2010). Fluorescence values used for calibration were recorded at the closing of each sampling bottle used for chl adetermination operated from the deck computer. The average value of fluorescence in the layer 350 to 400 m was thus first subtracted at all depths, to produce the depth corrected fluorescence. Then we obtained the following linear regression relation that was used to transform depth-corrected relative fluorescence units (Fc) to chl a(mg chl am−3) for this cruise: chl a= 0.596 Fc + 0.064, n = 118, R2= 0.734. As fluorometer data were not reliable near the surface, when obtaining the vertical section of total chl afrom the fluorometer calibrated data we used the values of extracted total-chlorophyll for the near surface layers (5, 10 and 20 m). RESULTS AND DISCUSSION Mesoscale variability Fig. 2 shows the distribution of isopycnals along the transect crossing the Bransfield Strait (Fig. 1). We observed the same water masses and mesoscale structures as those previously reported for the region by Sangrà et al. (2011). The main water masses are the relatively warm and fresh Transitional Zonal Water with Bellingshausen influence (TBW) that enters the Strait from the west and the relatively cold and salty Transitional Zonal Water with Weddell Sea influence (TWW) that inflows from the Weddell Sea and occupies the main body of the Strait (Sangrà et al. 2011). Mesoscale variability is mainly represented by the Bransfield Front, an anticyclonic eddy, and the Peninsula Front. These structures are very closely related and form part of what Sangrà et al. (2011) named the Bransfield Current System (BCS). In Fig. 2 we can clearly recognize the signal of the Bransfield Front by the steeply tilted isopycnals at the subsurface layers between Stns 1 and 3. The front extends until ~300 m, which is in the range of TWW. Current velocities from 38 Sangrà et al.: Ocean surface layer and phytoplankton dynamics the ADCP indicate a jet-like structure that is associated with the Bransfield Current; its axis is centered at Stn 2, where the current velocity can reach up to 0.4 m s−1. This current is the major component that drives the BCS, transporting TBW along the slope of the South Shetland Islands. In Fig. 2 we can also identify the signature of an anticyclonic eddy through the deepening of the isopycnals be tween Stns 3 and 6 just south of the Bransfield Front. The signal of the eddy is visible until ~240 m depth within the TBW range. Due to geostrophic adjustment there is a deepening of the isopycnals at the center of the eddy that induces an increase of ZUML between Stns 3 and 6. Eddy sampling took 28 h; during this period it is unlikely that the eddy displacement was greater than 10 km and thus its sampling can be considered to be quasi-synoptic. This is supported by the fact that westward shelf-advection speeds of the eddies at high latitudes are very low (Chelton et al. 2007); advection by the mean flow can be discounted as outside the Bransfield Front it does not exceed 0.1 m s−1 (Sangrà et al. 2011). Finally, close to the Antarctic Peninsula, we can recognize the signal of a shallow frontal region through the steeply tilted isopycnals between Stns 7 and 11 that Sangrà et al. (2011) named the Peninsula Front. TBW and TWW converge at the surface of the Peninsula Front. The above structures have been detected in all mesoscale resolution surveys of the region and, hence, they can be considered as permanent features of the circulation of this region during the austral summer (Sangrà et al. 2011). From the combined distribution of fluorometer calibrated total chl aand isopycnals show in Fig. 2, we can extract signs of relatively strong subduction regions at both fronts. These regions are indicated by rather high values of total chl awell below ZUML, and the euphotic layer forming deep column-like structures. At the Bransfield Front, values of total chl a> 0.2 mg m−3 were observed down to 180 m depth, indicating phytoplankton subduction on the light (anticyclonic) side of the front at Stn 1 (Fig. 2). At the Peninsula Front heavy (cyclonic) side (between Stns 9 and 11) there is also clearly evidence of subducting phytoplankton. The distribution of total chl ahas a column-like structure between Stns 9 and 11, that is sub ducted down as far as 300 m at Stn 10 (Fig. 2). Models and observations (e.g. Nagai et al. 2006, Pallàs-Sanz et al. 2010) have related these frontal subduction regions with the descending part of vertical circulation cells linked to ageostrophic secondary circulation of the front. Although our data set is too incomplete (i.e. 2dimensional and with insufficient resolution at both fronts) to diagnose the vertical circulation, the occurrence of deep chl acolumns supports, at least qualitatively, a downward motion at one side of both fronts. Observations in other frontal regions (e.g. Pollard & Regier 1992, Pallàs-Sanz et al. 2010, Thomas & Joyce 2010), show that the subducting region on one side of a front is accompanied by an upwelling region on the other side. On the upwelling side of the front restratification may take place due to slumping of the isopycnals induced by the ascending water (Nagai et al. 2006, Thomas et al. 2008). At the heavy (cyclonic) 39 Chl a (mg m–3) Stn number 0 20 40 60 80 100 Distance (km) Depth (m) 0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320 340 1 2 3 4 5 6 7 8 9 10 11 12 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 BF AE PF Fig. 2. Mesoscale variability in the Bransfield Strait (Stns 1 to 12): Thin black lines are isopycnals (potential density anomaly σθ,kg m−3); the color filled contour plot shows total chl a (mg m−3) calculated from fluorometer calibrated data (color filled contour plot); the bold black line shows the euphotic layer depth (ZEU) and bold red line shows the upper mixed layer depth (ZUML). Circled stations on the top axis indicate Stns 2, 6 and 9 where turbulence profiles were recorded. The locations of the Bransfield Front (BF), the anticyclonic eddy (AE) and the Peninsula Front (PF) are indicated on the top axis. The bottom topography is indicated by the shaded grey area at the right lower corner Mar Ecol Prog Ser 499: 35–46, 2014 side of the Bransfield Front (between Stns 2 and 3) there are signs of restratification evidenced by the squeezing of isopycnals at the depth range 30 to 60 m and by the slumping of the 2 nearest surface isopycnals’ accompanied by a shoaling of the UML. On the light (anticyclonic) side of the Peninsula Front, be - tween Stns 8 and 9, the UML also shoals, indicating restratification. This shoaling is due to the decreased tilt of the 27.5 and 27.55 kg m−3 isopycnals near the surface (20 to 40 m) and the opposing tilt of the 27.6 and 27.65 kg m−3 isopycnals at deeper layers (60 to 80 m). Although the restratification observed on the heavy (cyclonic) side of the Bransfield Front and on the light (anticyclonic) side of the Peninsula Front may be due to internal motions, the occurrence of an upwelling region accompanying the subducting region should not be disregarded. Upper mixed layer (UML) variability As shown in Fig. 2, the ZUML is modulated by the presence of mesoscale structures. It is deeper on the light (anticyclonic) side of the Bransfield Front at Stn 1, in the center of the eddy (Stns 4 and 5), on the heavy (cyclonic) side of the Peninsula Front (Stns 10 and 11), and at TWW (Stn 12), coinciding with geo - strophically depressed isopycnals and/or with the subducting regions. At Stn 12 the ZUML reaches the bottom because the whole water column is occupied by TWW which is homogeneous. The ZUML is shallower at the axis and at the heavy side of the Bransfield Front (between Stns 2 and 3) and at the axis and the light side of the Peninsula Front (Stns 7, 8 and 9). It is important to investigate whether turbulent mixing is active over the whole depth range of the observed UMLs and thus if the UMLs are actually mixing layers. Regarding this question, the ε profile at Stn 6 located inside the eddy shows a turbulent upper mixing layer throughout the depth range of the UML, where values of εare maximum (Fig. 3b). If we choose the depth of this mixing layer as the depth where the value of εdrops by one order of magnitude (Brainerd & Gregg 1995), this depth coincides ap proximately with the ZUML as indicated by the corresponding density profile 40 Fig. 3. Averaged vertical profiles of kinetic energy dissipation rate ε(bars, W kg−1), potential density anomaly σθ(black lines, kg m−3), and fluorescence (green lines, relative units) as obtained from microstructure turbulence profilers at (a) the Bransfield Front (Stn 2), (b) the anticyclonic eddy (Stn 6), (c) the Peninsula Front (Stn 9) and (d) the Antarctic Sound (based on average values measured at Stns 19, 21 and 22) Sangrà et al.: Ocean surface layer and phytoplankton dynamics (Fig. 3b). The profile for Stn 2 also shows a maximum of εin the UML portion of the water column (Fig. 3a). Vertical profiles of εshow a deeper mixing layer at the Peninsula Front axis (Stn 9) when compared with those at the Bransfield Front axis (Stn 2) and at the eddy (Stn 6) (Fig. 3a,b,c). This deepreaching mixing may be related to frontal in - stabilities that may en hance mixing as observed by D’Asaro et al. (2011) in a small submesoscale front. Stn 9 was located at the mouth of a submarine canyon. This complex bathymetry may favor the generation of internal motion that could be also responsible for the ob served deep mixing. Unfortunately we do not have direct measurements of turbulence from all the stations. An indirect ap - proach to infer whether the mixed layer corresponds to an active turbulent layer is to quantify the Thorpe scale (Thorpe 1977). The Thorpe scale measures the inversions in a vertical density profile (e.g. Gargett & Garner 2008). Thorpe (1977) related these inversions to the 3-dimensional turbulent eddy field overturning. Therefore, the appearance of the inversion in a vertical density profile is an indicator of turbulent mixing (Gargett & Garner 2008). Fig. 4 shows the distribution of the Thorpe scale along the Bransfield Strait transect. In general, the ZUML effectively matches the base of an upper active turbulent layer where inversions are frequent. The most active turbulent layer is observed within the UML of the anticyclonic eddy between Stns 3 and 6. Another active turbulent region corresponds to Stns 10, 11 and 12, where the inversion reaches the ZUML indicating that the whole UML is actually an active turbulent layer. Although intermittent, the density profile of Stn 1 also shows inversions down to the ZUML, indicating turbulent activity at the depth range of the UML. At the axis of the Peninsula Front (Stn 9) there are intense inversions below the UML; this indicates that the UML and the mixing layer are decoupled, as the turbulence profiles further point out. Antarctic Sound turbulent environment Fig. 3d shows mean profiles of density, fluorescence and εas obtained from the microstructure turbulence profiler in the Antarctic Sound. Notice the higher levels of turbulence throughout the water column in comparison with the Bransfield Strait profiles. In fact the mean value of the 9 profiles recorded in the Antarctic Sound for the depth range 20 to 100 m, ε= 5.73 ×10−7 W kg−1, was ~5-fold higher than those recorded along the Bransfield Strait transect, ε= 1.22 ×10−7 W kg−1. Antarctic Sound density profiles were homogeneous over the full depth range (Fig. 3d), and composed of Weddell Sea Shelf Water (WSSW). As TWW (potential temperature, θ = −1.30 to −0.50°C, salinity, S = 34.37 to 34.5) is a modification of WSSW (θ= −1.73 to −0.93°C, S = 34.37 to 34.5), the 2 water masses are not very distinct; they are cold, salty and homogeneous, with WSSW being slightly colder. Therefore the water mass properties are very similar south from the axis of the Peninsula Front and at the Antarctic Sound. Because the density profiles are homogeneous, instead of defining a ZUML we computed the mixing layer depths as the depth where εis reduced by an order of magnitude: hence 350 m for Stn 19, 220 m for Stn 21, and 320 m for Stn 22. Notice that those mixing layer depths are more than 3 fold higher than the observed mean ZUML (60 m) at the Bransfield Strait. 41 0 20406080100 Distance (km) 0 20 40 60 80 100 120 140 Depth (m) 12 Stn number 345 6 789101112 0.5 0.9 1.3 1.7 2.1 2.5 2.9 3.3 3.7 4.1 4.5 4.9 5.3 5.7 6.1 6.5 6.9 7.3 7.7 8.1 8.5 8.9 9.3 9.7 Thorpe scale (m) Fig. 4. Thorpe scale distribution (filled color contour plot) in the Bransfield Strait. Thorpe scale units are in meters and provide a measure of vertical overturning and thus an indicator for turbulent mixing. Turbulent mixing will be more active for large Thorpe values and less active for low Thorpe values. Note that the largest values are observed in the UML, coinciding with the largest values of ε(see Fig. 3). See Fig. 2 legend for explanation of superimposed black and red lines Mar Ecol Prog Ser 499: 35–46, 2014 Coupling total chl awith mesoscale variability and mixing Maximum values of total chl a, as obtained from the fluorometer calibrated data (Fig. 2b), were ob - served near the surface on the light (anticyclonic) side and at the axis of the Peninsula Front (Stns 7, 8, and 9), and at the edge of the Anticyclonic Eddy (Stn 6). At the Peninsula Front stations these maxima coincide with the shallowest UML (54, 34, and 18 m res pectively) and are thus located on the restratified side of the front. From this maximum, the phytoplankton is subducted and mixed within the heavy (cyclonic) side of the front (between Stns 9 and 11). This subducting region acts as a sink for the cells that are transported to the deep ocean interior down to at least 300 m. Therefore the cross-frontal distribution of total chl ais asymmetric, with a shallow local maxi - mum on the restratified side of the front and a deep column-like structure on the subducting side. At TWW (Stn 12), total chl adistribution is also diluted over the whole water column, probably due to a high rate of vertical mixing as suggested by an UML reaching the bottom (115 m), and by the high density of inversions (Fig. 4). As shown in Fig. 2, the maximum at the Peninsula Front diffuses towards the eddy, attenuating its intensity and diluting in the UML. This suggests that the eddy entrains the phytoplankton from the Peninsula Front source, although in situ net growth should not be disregarded. The entrained phytoplankton will be subject to mixing and accumulate at the eddy UML. This can explain the relatively high values of total chl aat subsurface levels (30 to 60 m) inside the eddy. The largest density inversions were observed at the eddy, suggesting strong mixing (Fig. 4). This is confirmed by direct measurements of turbulence at Stn 6 (Fig. 3b). At the axis and at the heavy (cyclonic) side of the Bransfield Front (Stns 2 and 3) a relative maximum of total chl aat the base of the UML was observed but it had lower concentrations than those at the Peninsula Front. As for the case of the Peninsula Front, this coincides with the restratified side of the front evidenced by a local minimum of the ZUML. On the light (anticyclonic) side located at Stn 1, the concentration of total chl ais again diluted in the water column, coinciding with a deep UML. As already mentioned, significant values of total chl awere observed until 180 m, well below the ZUML and ZEU, indicating subduction. Similarly to the case of the heavy (cyclonic) side of the Peninsula Front, this subduction and high rate of vertical mixing may be responsible for the observed total chl adilution throughout the water column. Again, as was the case for the Peninsula Front, the cross-frontal distribution of total chl awas asymmetric, with a deep column-like structure indicating subduction on one side, and a shallow local maximum on the restratified side. A plausible explanation for this asymmetric distribution may be related to the occurrence of vertical circulation cells as observed in other frontal regions (e.g. Pallàs-Sanz et al. 2010) as previously suggested. The descending part of these cells may cause the subducting, column-like structure of chl ain the deep UML part of the fronts. At the restratified part of the fronts, the shallow local maximum may be related to the ascending part of the vertical cell that will transport phytoplankton cell to the surface where higher light levels will increase their net growth rate. Other plausible mechanisms that may explain the observed local maxima on the sides of the fronts are the occurrence of a confluent flow and/or the advection of phytoplankton from an upstream source. Preference of large plankton for well-mixed environments When correlating phytoplankton size fraction variability with the physical environment, the first no - ticeable point is that in the relatively high turbulent environment of the Antarctic Sound large phytoplankton dominate, whereas in the low turbulent region of the Bransfield Strait small phytoplankton dominate (Fig. 5). In the Antarctic Sound, where the upper turbulent active layer is very deep (>300 m, Fig. 3d), the chl aprofiles are homogeneous, with a mean value of 1.8 mg m−3 for larger phytoplankton and one order of magnitude less, 0.3 mg m−3, for small phytoplankton. In the Bransfield Strait (lower average turbulence) the picture is reversed. There, the maximum values for small phytoplankton largely exceed 1 mg m−3 whereas the values for large phy - toplankton do not exceed 0.5 mg m−3 (Fig. 5a,b). García-Muñoz et al. (2013), using Flow-CAM and CHEM TAX software, observed that the nanophytoplankton size range is the most abundant along the Bransfield Strait transect. It is mainly composed of small diatoms followed by haptophytes, prasinophytes and cryptophytes. At the Antarctic Sound, they observed that phytoplankton is largely composed of microplanktonic diatoms, mainly Thalassio - sira sp. (García-Muñoz et al. 2013). The general pattern described above suggests that large phytoplankton have a preference for those environments where the rate of vertical mixing is 42 Sangrà et al.: Ocean surface layer and phytoplankton dynamics high, with very deep active turbulent upper layers such as those we observed at the Antarctic Sound stations. In this regard, if we inspect in more detail the Bransfield Strait section corresponding to the large phytoplankton (Fig. 5b) we can appreciate that, except for the near surface at Stn 2, clearly higher (>0.25 mg m−3) concentrations of large phytoplankton chl aare found only at those stations coinciding with the deepest UMLs. This is the case of the light side of the Bransfield Front (Stn 1, ZUML = 91 m), the heavy side of the Peninsula Front (Stn 11, ZUML = 117 m) and the TWW (Stn 12, ZUML = 115 m). The Thorpe scale distributions at these stations indicates that the UML coincides with an active turbulent upper layer (Fig. 4), suggesting a high rate of vertical mixing. Therefore, a close inspection of the Bransfield Strait transect supports the above hypothesis concerning the preference of large phytoplankton for those environments where the rate of vertical mixing associated with deep UMLs is high. In contrast to the large phytoplankton distribution, size-fractionated chl afor small phytoplankton shows appreciable values (>0.25 mg m−3) all along the Bransfield Strait transect. As the phytoplankton are mainly composed of small phytoplankton (80%) in this transect, their distribution mirrors the distribution for total chl a (Fig. 2b). Thus the phytoplankton distribution res - ponds to the same physical forcing as suggested above for total chl a, which is mainly coupled with mesoscale structure variability. Phytoplankton composition and ZUML variability The above observations of size-fractionated chl a distribution suggest a covariance between large phy - to plankton and the ZUML, and hence a relationship between this component and the rate of vertical mixing. To explore this, in Fig. 6 we plotted the percentage of large phytoplankton versus ZUML for all our stations. Fig. 6 also shows the mean values of εfor the Bransfield Strait and Antarctic Sound regions. There is a clear linear relationship (R2 = 0.86) be - tween the ZUML/mixing layer depth and the percentage of large phytoplankton (Fig. 6), indicating that the proportion of large phytoplankton increases with the deepening of the ZUML. For the shallowest ZUML, as is the case on the light side of the Peninsula Front (Stns 8 and 9) and the edge of the Bransfield Front (Stn 3), the percentage of large phytoplankton is less than 15%. In those stations of the Bransfield Strait where the ZUML is deep and well below the ZEU, as is 43 Chl a < 20 µm mg m–3 mg m–3 Antarctic Sound ab c Stn number Chl a > 20 µm Chl a < 20 µm Chl a > 20 µm Chl a (mg m–3) Stn number 0 20 40 60 80 1000 20 40 60 80 100 Distance (km)Distance (km) 0 20 40 60 80 100 120 140 0 20 1.65 1.55 1.45 1.35 1.25 1.15 1.05 0.95 0.85 0.75 0.65 0.55 0.45 0.35 0.25 0.15 40 60 80 100 120 140 10 0.5 1 1.5 2 2.5 2 3 4 5 6 7 8 9 10 11 1212 345 6 789101112 0.15 0 25 50 75 100 0.2 0.25 0.3 0.35 0.4 Depth (m) Fig. 5. Distributions of (a) small phytoplankton (chl a< 20 µm equivalent spherical diameter [ESD]) and (b) large phytoplankton (chl a> 20 µm ESD) in the Bransfield Strait (filled contour plots). See Fig. 2 legend for explanation of superimposed black and red lines. (c) Depth profiles of small (blue) and large (red) phytoplankton at the Antarctic Sound stations. The bold black line shows the euphotic layer depth (ZEU)