For ATP measurements, 1 l of seawater was prefiltered through a 250 µm net and then filtered through a 0.22 µm membrane filter. The filter was immediately plunged into 3 ml boiling TRIS–EDTA phosphate buffer (pH 7.75) and the ATP was extracted at 10 °C for 3 min and kept frozen (-20 °C) until laboratory analysis in Italy (Holm-Hansen and Paerl 1972). The filtrate, from the ATP sample, was stored in sterile polycarbonate bottles. Extracts for ATP determination were prepared according to Holm-Hansen and Paerl (1972) and analysed by measuring the peak height of the firefly bioluminescence with a Lumat LB9507 luminometer by EG&G Berthold. The conversion factor, C/ATP = 250, was adopted to convert ATP values into carbon biomass units (C-ATP) according to Karl (1980). The accuracy of this conversion is ± 20% (Skjoldal and Båmstedt 1977). According to ATP concentrations, Karl (1980) classified the trophic status of marine systems as follows: oligotrophy when ATP is < 100 ng l^-1; moderate trophism ATP > 100 < 500 ng l^-1 and eutrophy ATP < 500 ng l^-1.

CHLa concentrations, as an index of phytoplankton biomass, were determined in the water column from surface to a maximum of 160 m depth. The water samples (1 l) were filtered on Whatman GF/F glass-fibre filters, according to Lazzara et al. (1990). After filtration, the filters were immediately stored at -20 °C. CHLa was extracted in 90% acetone and read before and after acidification. Determinations were carried out with a Varian Eclipse spectrofluorometer. Maximum excitation and emission wavelengths (431 and 667 nm, respectively) were selected on a pre-scan with a solution of CHLa from Anacystis nidulans (Sigma-Aldrich Co). The conversion factor of C-CHL = 100 was adopted to convert CHLa values into carbon biomass units (Smith et al. 1998).

The Trophic State Index (TSI), applied to classify the stations according to their algal biomass, was calculated from the chlorophyll measurements (Carlson 1983: TSI (CHLa) = 9.81 ln(CHLa) + 30.6).

Samples for prokaryotic abundance (PA; including bacteria, archaea and cyanobacteria) were collected into sterile Falcon vials (50 ml). Each sample was immediately fixed in pre-filtered formaldehyde (0.2 µm porosity; 2% final concentration) and stored at 4 °C until analysis. Within three months, two replicates of each sample were filtered through polycarbonate black membrane filters (0.2 µm porosity; GE Water & Process Technologies) and stained for 10 min with 4’,6-diamidino-2-phenylindole (DAPI, Sigma, final concentration 10 µg ml^-1) according to Porter and Feig (1980). Stained cells were counted under a Zeiss AXIOPLAN 2 imaging epifluorescence microscope (magnification: Plan-Neofluar 100× objective and 10× ocular; HBO 100 W lamp; filter sets: G365 excitation filter, FT395 chromatic beam splitter, LP420 barrier filter) equipped with the digital camera AXIOCAM-HR. Images were captured and digitised on a personal computer using the AXIOVISION 3.1 software. Linear dimensions of cells were measured and their shapes equated to standard geometric figures according to Lee and Fuhrman (1987), Fry (1990) and Massana et al. (1997). Volume of single cells (VOL, as µm³) was calculated according to Bratbak (1985). Cell carbon content (CCC, as fg C cell^-1) was derived from cell volumes according to Loferer-Krößbacher et al. (1998). Total prokaryotic biomass (PB, as µg C l^-1) was calculated by multiplying PA by CCC, locally derived from single cell VOL (La Ferla et al. 2015).

ETS measurements and relative rates

Respiratory rates were quantified according to the tetrazolium reduction technique (Packard 1971, 1985) as modified for the microplankton community (Kenner and Ahmed 1975). The ETS assay allowed an estimation of the maximum velocity (Vmax) of the dehydrogenases transferring electrons from their physiological substrates (NADH, NADPH and succinate) to a terminal electron acceptor (O₂) through their associated electron transfer system. Details on our ETS measurements in Antarctic and peri-Antarctic areas have been described previously (Azzaro et al. 2006, Crisafi et al. 2010, Misic et al. 2017). Briefly, subsamples (from 2 to 20 l) were pre-filtered through a 250-μm mesh-size net and concentrated on GF/F-glass-fibre filters (nominal pore diameter 0.7 μm) at reduced pressure (< 1/3 atm). Although the filter porosity was specific for microplankton, GF/F filters retain also particles colonised by very small heterotrophs. The filters were immediately folded into cryovials and stored in liquid nitrogen to prevent the enzymatic decay (Ahmed et al. 1976) until they were analysed in the laboratory (< 3 months). The assays were performed in duplicate and homogenates were incubated for 30 min in the dark at the in situ temperature (± 0.5 °C) of the sample. The ETS was corrected for in situ temperature with the Arrhenius equation using a value for the activation energy of 11.0 kcal mol^-1 (Packard et al. 1975, Arístegui and Montero 1995).

The specific standard deviation (i.e. the percentage of the standard deviation of the replicates on the average value of the same replicates), due to the analytical procedures and sample handling, was about 35%.

ETS (μl O₂ l^-1 h^-1) was considered equal to the respiration rate in the epipelagic zone and converted to respiratory Carbon Dioxide Production Rates (CDPR) (μg C l^-1 d^-1) by using the following (Eq.1):

CDPR = (ETS*12/22.4) * (122/172) (Eq.1)

where 12 is the C atomic weight, 22.4 the O₂ molar volume and 172/122 the Takahashi oxygen/carbon molar ratio (Takahashi et al. 1985). Real respiratory rates have been calculated using the conversion factors from ETS Vmax to CDPR as referred by La Ferla and Azzaro (2001b) and Azzaro et al. (2006), in the epi- and mesopelagic layers, respectively.

Cell specific respiratory rates (CSRR) were calculated by dividing normalised CDPR values to the normalised cell abundance values in each station by adopting a prokaryotic contribution of 50% and 80% at epipelagic and mesopelagic layers, respectively, assuming that the activities we measured were mainly due to the prokaryotic fraction and that all the cells have similar activity levels.

Heterotrophic energy production (HEP) determination

Today, the P/O ratio is thought to be closer to 2.5 rather than 3.0 (Ferguson 2010, Moran et al. 2012). With this ratio and using ETS activities to compute respiration, one can derive ATP production rates in a particular oceanic region through the determination of the HEP (Packard et al. 2015). HEP, in micro Joules (µJ), can be calculated from seawater respiration (R) in µmol O₂ h^-1 l¹ using the following equation (Eq.2):

HEP (µJ h^-1 l^-1) = R (µmol O₂ h^-1 l^-1) * 2 * 2.5 * 0.048 * 10^-6 (Eq.2)

where 2 is the number of electron pairs that participate in reduction of one molecule of O₂ to two molecules of H₂O; 2.5 is the modified P/O of Ochoa (1943), the ATP produced by the flow of one electron pair; 0.048 is the Gibbs Free Energy (∆G) per micro mol ATP in J·(µmol ATP)^-1 (Hinkle 2005, Ferguson 2010, Moran et al. 2012, Packard et al. 2015); and 10^-6 converts Joules to micro Joules. If, HEP in milli Joules (mJ) is desired, simply replace the factor, 10^-6, in Eq. 2, with the factor, 10^-3. The turnover time (τ) of ATP in the microplankton community, from which came the sample, is the molar ratio of the ATP concentration to the ATP production rate, in the microplankton. The calculation steps are shown in Table 2.

Table 2.Download as CSV 

Heterotrophic Energy production (HEP) and adenosine triphosphate (ATP) turnover time in microplankton in the Ross Sea water column.

Pelagic Zone	Depth Interval	Potential Respiration (Φ)	ATP	Respiration	HEP (ATP Production)	HEP (Energy Production)	ATP Turnover Time (τ)
	(m)	(µmol O2 l-1 h-1)	(ng l-1)	(µmol O2 l-1 h-1)	(µmol l-1 h-1)	(mJ l-1 h-1)	(min)
Euphotic	2–100	9.13 ± 5.93	123 ± 72.2	2.38 ± 1.54	11.88 ± 7.71	570 ± 370	1.25 ± 0.76
Epipelagic	2–160	8.97 ± 3.27	110 ± 71.0	2.33 ± 0.85	11.67 ± 4.25	560 ± 204	1.15 ± 0.77
Mesopelagic Upper (A)	100–500	4.02 ± 3.49	30.35 ± 31.68	1.05 ± 0.91	5.23 ± 4.54	251 ± 218	0.74 ± 0.49
Mesopelagic Upper (B)	160–500	3.03 ± 3.19	27.49 ± 32.58	0.86 ± 0.83	4.31 ± 4.15	207 ± 199	0.78 ± 0.53
Mesopelagic lower	500–800	1.65 ± 0.88	13.48 ± 6.76	0.43 ± 0.23	2.15 ± 1.15	103 ± 55	0.76 ± 0.53

Prokaryotic heterotrophic activity (PHA) and production (PHP)

PHA was evaluated by ³H-leucine incorporation rate assay using the microtubes method described by Smith and Azam (1992). Briefly, triplicate 1.7 ml subsamples and duplicate zero-time killed (trichloroacetic acid-TCA, 5% final concentration) blanks were incubated in 2 ml polypropylene microcentrifuge tubes (Safe-lock, Eppendorf) with L-[4,5^-3H] leucine (Amersham, GE Healthcare, SA 61Ci/mmol) giving final concentration of 20 nM. Microtubes were distributed in floating racks into darkened containers with seawater disposed in a refrigerated box. Samples were incubated for 170–190 min at a temperature between -0.5 and -0.8 °C. Incubation was stopped with TCA (5% final concentration). Pellets were washed twice with 5% TCA and 80% ethanol and finally supplemented with 1 ml of liquid scintillation cocktail (Ultima Gold MV Perkin Elmer Life and Analytical Sciences). The liquid-scintillation analysis was effected in a Perkin Elmer Model Wallac 1414 WIN Spectral counter using the internal quenching control. Saturation curves analysis and time-course experiments were carried out at four different stations (from 2 to 300 m depth) with hourly controls for up to six hours of incubation. This procedure was established in a previous cruise in the RS, carried out within the framework of the Victoria Land Transect Project – VLTP-2004 (PNRA XIX Expedition). During PHA time-course experiments, linearity in the ³H-leucine uptake was observed in one sample (25 m depth) between 1 and 6 hours of incubation (r = 0.96), while in the other three samples (2, 270 and 310 m depths), linearity was observed only between 1 and 3 hours of incubation (r = 0.83 and 0.88, respectively).

Prokaryotic heterotrophic production (PHP) was calculated from the ³H-leucine incorporation rate (PHA) expressed in moles incorporated per unit time and volume (Kirchman et al. 1985) according to the equation PHP = PHA*CF where CF is a conversion factor expressed in kg C mol^-1. CF were determined following a “semi-theoretical” approach using the molecular weight of leucine, the leucine content of cellular protein and the cellular carbon equivalent of the protein according to Simon and Azam (1989) and the isotope dilution (ID) in situ experimentally determined according the rectangular hyperbola fitting method of van Looij and Riemann (1993), as described by Pedrós-Alió et al. (2002). In particular, ID was determined from two samples collected at 20 and 45 m depth, respectively and two at 300 m depth. It varied between 1.04 and 1.37. The mean value of 1.25 was used to calculate the CF that was equivalent to 1.94 kg C (mol leucine)^-1 (Table 3).

Table 3.Download as CSV 

Isotopic Dilution (ID) detected from leucine incorporation rates (Vmax) in samples Abio19 (45 and 300 m depth) and Abio05 (20 and 300 m depth), respectively. ID mean value 1.25 ± 0.14.

Station	depth (m)	V max ± SD p mol l-1 h-1	ID	confidence interval 95%
Abio19	45	18.058 ± 0.870	1.06	0.95–1.18
	300	0.357 ± 0.069	1.24	1.00–1.48
Abio05	20	9.785 ± 4.813	1.37	0.69–2.04
	300	0.932 ± 0.225	1.34	1.01–1.66

PHP was expressed as production of biomass (as C) per time unit and volume.

Derived parameters as Specific Growth Rate d^-1 [SGR (µ) = PHP/PB (Prokaryotic biomass)] and Biomass Turnover Time (days) [BTT (g) = ln(2)/µ] were calculated according to Kirchman (2001).

In comparison, PHP_D, PB_D, SGR_D and BTT_D were calculated using 107 fg C µm^-3 cell and an ID = 1 according to Ducklow et al. (2001). Cell specific incorporation rates (CSIR) were calculated by dividing normalised PHP values to the normalised cell abundance values in each station.

The prokaryotic C requirement was computed as Prokaryotic Carbon Demand (PCD), i.e. PHP+CDPR by using normalised data, assuming the contribution of prokaryotes to total microbial community respiration as 50% in the epipelagic layer and a contribution of 80% in the mesopelagic ones (La Ferla et al. 2005, del Giorgio et al. 2011, Baltar et al. 2015). The Prokaryotic Growth Efficiency (PGE) was calculated as PHP/PCD and the isotope dilution (ID) in situ was experimentally determined according the rectangular hyperbola fitting method of van Looij and Riemann (1993), as described by Pedrós-Alió et al. (2002) using HYPER 32 fitting software.

In order to detect possible influences between environmental factors and microbial variables, Spearman Rank correlation coefficients were calculated for the microbiological data and environmental parameters using the SigmaStat software V3.0 and the Mann and Whitney test using the PAST.exe (Hammer et al. 2001). The analyses of variance (one-way ANOVA and Kruskal-Wallis) were applied to some parameters to assess the significance of the differences between depth layers and stations.

Data were integrated with depth according to the trapezoidal method and normalised to the depth: from 2 to 100 m for the epipelagic layer; from 100 to 800 m for the mesopelagic layer.

The depth-integrated rate (ʃR dz in mg C m² d^-1 l^-1) for the water column was calculated within the depth interval between Z₁ and Z₂ using the following equation (Eq. 3):

ʃR dz = y (Z₂^(x+1) – Z₁^(x+1) / (x+1) (Eq.3)

The environmental assessment revealed a general picture of low trophism on a spatial scale as exemplified by the CHLa concentration. Analysis of the ATP concentrations also indicated modest or poor trophism. Considering ATP as a quantitative proxy for total living biomass (Holm-Hansen and Paerl 1972, Karl 1980), a relatively homogeneous biomass occurred throughout the RS. However, according to Karl’s classification, most stations manifested moderate trophism or oligotrophy. In a previous study carried out in January-February 2001 in the RS, an extensive algal biomass and variable ATP estimates corresponding to different trophic statuses were observed (Azzaro et al. 2006, La Ferla et al. 2015). In particular, marked peaks of ATP were found (up to 1752 ng l^-1), revealing strong eutrophy according to Karl (1980). In our study, ATP determinations confirmed the trophic evaluation derived by TSI. In the 0–100 m depth layer, the autotrophic biomass (in terms of C-CHLa) accounted for 20–59% (mean value 44 ± 19%) of the total biomass (in terms of C-ATP). The highest and the lowest ratios of autotrophic biomass occurred at stations Abio09-D and Abio20, respectively, thus corroborating the previous statements about TSI and ATP estimates. In the epipelagic layer, the integrated CHLa values revealed a low phytoplankton standing stock varying from 4.76 to 23.68 mg m^-2. In contrast, in austral summer 2014, Mangoni et al. (2017) detected high autotrophic biomass (integrated CHLa up to 371 mg m^-2) in the epipelagic layer of the RS.

PA was in a range comparable to similar measurements made in several Antarctic marine environments (see table S1 in La Ferla et al. 2015). Comparison to previous studies in the RS confirmed the results of Monticelli et al. (2003) in Terra Nova Bay and Celussi et al. (2009) in Cape Adare.

PB decreased with depth by a factor of 1.7. However, it was higher than other prokaryotic measurements made previously in the RS (Buitenhius et al. 2012, Carlson et al. 1999, Steward and Fritsen 2004). The utilisation of a variable cell carbon content (from 11 to 45) to calculate the biomass in each sample could partially explain this pattern. Generally, two standard PA to biomass conversion factors are utilised in ecological marine studies: 20 fg C cell^-1 (Lee and Fuhrman 1987) or 9.1 fg C cell^-1 (Buitenhuis et al. 2012). Since cell carbon content is directly linked to cell-size variability (Young 2006), the utilisation of unvarying factors to convert abundance to biomass might distort the evaluations by overestimating or, in this case, underestimating the actual carbon amounts (La Ferla et al. 2015).

The ETS assay, originally designed by Packard (1971), continues to be successfully adopted in oceanic regions because of its high sensitivity and resolution levels that are not attainable with other methods based on sample incubation (del Giorgio and Williams 2005). Moreover, the response to the bias derived by the utilisation of empirical conversion factors from the ETS Vmax into actual rates of O₂ consumption and metabolic CO₂ production has largely been discussed and countered (del Giorgio and Williams 2005, La Ferla et al. 2010, Packard et al. 2015, Filella et al. 2018). Independently, good correlations between ETS and in vivo respiration rates were obtained in surface seawater samples in the framework of the ABIOCLEAR cruise. In the aphotic zone between 100 and 600 m depth of the RS, Azzaro et al. (2006) reported decreasing ETS activity throughout the water column and, despite the algal bloom in 2001, the ETS activity fell in a narrower range (0.017–0.170 µl O₂ l^-1 h^-1 than our data in the mesopelagic layer. In Summer 2014, Misic et al. (2017) observed different ETS - POM relationships, but ones consistent with the characteristics of a phytoplanktonic bloom of the Phaeocystis type. Moreover, their results featured averaged ETS values twice ours in the epipelagic and mesopelagic layers. CDPR was also twice ours in the upper layers and 4.6 times higher than ours in the mesopelagic one. The comparison between these findings evidenced great differences in the metabolic rates on an inter-annual scale and corroborated the importance of heterotrophic signals in understanding climate trends.

HEP calculations in the microplankton quantify the energy generation due to the decomposition of ATP by a group of enzymes (ATPases) in plasmalemma membranes of constituent bacteria and archaea as well as in constituent eukaryote mitochondrion. It represents a new metric in oceanographic analysis. The only other oceanic region, for which HEP has been calculated, is the Peru Current Upwelling at 15° S (Pisco, Peru). The HEP calculations for the epipelagic and mesopelagic waters of the Peru Upwelling at 15° S are given in Packard et al. (2015). The average HEP (and standard deviation) in the epipelagic zone down to 150 m was 24 ± 30 × 10³ µJ h^-1 l^-1. In a transect (C-Line) for 185 km across the Peru Upwelling, epipelagic HEP ranged from a high of 108 × 10³ µJ h^-1 l^-1, at station C10, over the Peru Trench, 71 km from the coast, to a low of 2 × 10³ µJ h^-1 l^-1, 22 km further offshore at station C12. In mesopelagic waters below 150 m down to 1000 m, HEP was 96% lower than it was in the epipelagic zone, averaging only 84 ± 59 µJ h^-1 l^-1. These values, from the Peru Upwelling system (Packard et al. 2015), are more than an order of magnitude higher than the values from the RS.

Time course experiments on PHA showed results in agreement with those detected in the VLTP-2004 project (Monticelli, personal communication). Linearity occurred within 1 and 6 hours for samples collected at 25 m depth and in a smaller time lapse for the others. Along the water column, higher PHA was always observed in the photic layers while reduced activity in the aphotic waters occurred. An increase in activity was always observed in the bottom samples, i.e. those taken a few metres from the sea floor. The increase of heterotrophic bacteria metabolisms in benthic boundary layers is a known phenomenon observed in other water columns (Packard and Christensen 2004).

At station Abio22-A, PHA was particularly high at all depths, with the highest normalised rates (12.442 nmol m^-3 h^-1) in the photic layer. This was one order of magnitude higher than equivalent normalised PHA calculations observed at the other four stations. The mean leucine incorporation rate observed in the 0 - 50 m layer was 8.05 pmol l^-1 h^-1 (sd = 4.52, n = 15), the same order of magnitude as observed by Ducklow et al. (2001) in the RS during late spring period. The PHA observed in our cruise was also in accordance with that observed by Pedrós-Alió et al. (2002) in the Gerlache strait (Antarctic Peninsula) in late spring and summer cruises. The cell specific incorporation rate (CSIR) strictly reflected the distribution of PHA throughout the water column with the highest values in the surface to 50 m depth layer.

In our experiment, an isotope dilution of 1.25 was used to calculate the CF. It was equivalent to 1.94 kg C mol leu^-1. Considering the variability observed in ID determinations and the coefficient of variation (CV%) detected in the triplicate samples for leucine incorporation-rate analysis (mean = 14.3%, sd = 11.3%, n = 84), the ID should be not too far from 1 (assuming no isotope dilution). That value corresponds to a theoretical CF = 1.55 kg C mol leu^-1 (Simon and Azam 1989). Similar ID values (mean = 1.27 corresponding to CF = 1.96 kg C mol leu^-1) were detected in the Antarctic Peninsula area in late spring-summer where the empirical method carried out simultaneously produced on average CF = 0.81 kg C mol leu^-1 (Pedrós-Alió et al. 2002). In the framework of experiments conducted in subtropical northeast Atlantic Ocean, Baltar et al. (2010) discussed the incongruence often observed between empirical and theoretical CFs estimates as well as their variability in the mesopelagic water column (range 0.13–0.85 kg C mol^-1 leu). They argued that, in the deep domain, the carbon limitation and the slower cell growth take place, further reducing the deep water CFs as compared to the theoretical ones. The choice of theoretical, semi-theoretical or empirical CF can markedly affect the PHP estimation and, consequently, the derived parameters. In the case of a significantly different CF, it would be appropriate to use both values to furnish the best information about the carbon flux in the prokaryotic compartment. PHP followed the same distribution of PHA throughout the water column with a maximum value corresponding to a production peak at 25 m depth at station Abio22-A.

In the euphotic layer, SGR d^-1 calculated with our CFs, resulted 2.4 times lower than SGR_D d^-1 calculated using the conversion factors used by Ducklow et al. (2001). Contrary to the SGR, the BTT (days) was 2.4 times higher than that calculated with Ducklow’s CFs. In mesopelagic waters, SGR and BTT were the opposite of those observed in the upper 100 m of the epipelagic zone.

From mean hourly values, the CDPR/PHP ratios in the epi- and mesopelagic layers were 11.75 and 0.80 µg C l^-1 h^-1, respectively.

Prokaryotic (bacterial and archaeal) activity is often measured using the PGE that defines the balance between catabolic and anabolic prokaryotic processes (Baltar et al. 2015). It corresponds to the proportion of dissolved organic carbon (DOC) that is converted by microorganisms into biomass and might be consumed by higher trophic levels (Eichinger et al. 2010). PGE is considered the most important factor affecting the C budget (Van Wambeke et al. 2002) and it indicates the efficiency of organic substrates recycling by prokaryotes (Mazuecos et al. 2015). The determination of PGE depended on the choice of the methodological procedures and approach, i.e. mostly respiration estimates and leucine to carbon CF. Indirect respiration estimates were often obtained from the sinking biogenic particles (Sweeney et al. 2000), from the bacterial production experimental data and from an empirical contribution of respiration (Ducklow et al. 2001; Ducklow 2003) or by sediment traps (DiTullio et al. 2000, Langone et al. 2000, Langone et al. 2003, Nelson et al. 1996). Sometimes, PGE is arbitrarily considered to be 30% or 36% (Manganelli et al. 2009) and a few papers have simultaneously identified the actual respiration and heterotrophic production rates (La Ferla et al. 2005, Zaccone et al. 2003, del Giorgio et al. 2011, Baltar et al. 2009, 2015 and references herein) in pelagic waters. In deep waters in the Atlantic Ocean, Baltar et al. (2010) determined PGE variations in the range < 1–34%. This high variability resulted in being highly correlated to the empirical conversion factors determined and adopted to calculate PHP (Baltar et al. 2010). High variability of PGE, on both space and time scales in ocean samples, has often been assessed (Lemée et al. 2002, Reinthaler and Herndl 2005) and low values of PGE (< 15%) have been associated with oligotrophic conditions (Biddanda et al. 2001, del Giorgio et al. 2011). In short, at low growth efficiency rates, more dissolved organic matter is remineralised, keeping the nutrients cycling within the microbial cycle; conversely, at high growth efficiency rates, the dissolved organic matter is more efficiently transferred into the particulate phase thus strengthening the carbon distribution throughout the trophic food web (Cajal-Medrano and Maske 2005). In our study, the highest PGE values were determined at stations where oligotrophic conditions at the sea-surface were evidenced by TSI. Typically, most of the primary production in low-productivity environments is respired by bacteria (Biddanda et al. 2001). In epipelagic layers, the prokaryotic respiratory processes exceeded heterotrophic production, whilst high PGE was surprisingly calculated in the mesopelagic layer showing consistent differences with the upper layer. In addition, excluding the Abio16 and Abio22-A stations, where high PHPs were determined, the averaged PGE value in the mesopelagic layer surpassed the value in the epipelagic one. These findings were consistent with the high values detected in the deep layers of the Mediterranean Sea (La Ferla et al. 2005 and 2010) where PGE data did not correlate with primary production, but rather confirmed that the PCD could not be sustained solely by the DOC of autochtonous origin and/or by phytoplankton exudation. In the RS, Azzaro et al. (2006) compared estimates of carbon flux by sediment traps and found that about 63% of organic carbon, remineralised by respiration, was derived from the POC pool, confirming the decomposition rates of Ducklow et al. (2001) in the vertical POC flux. Weak bacteria-primary production coupling has also been ascribed to temperature restriction of metabolic utilisation (Pomeroy and Wiebe 2001) or to grazing together with the lack of bio-available dissolved organic matter (Bird and Karl 1999). The relatively high PGE probably also reflected the capability of prokaryotic cells to individually divide (fast growing communities) or to increase in size. We found that, in our samples, volumetric determinations in the mesopelagic layer were higher than in the epipelagic layer. They fell into the range of 0.038 to 0.196 µm³ (data not shown). Furthermore, the circulatory dynamics of the water masses could also explain the high PGE at depth. When vertical convective processes occur with the sinking of surface water masses, a consequent enrichment of fresh organic matter in the deep layers happens (La Ferla and Azzaro 2001a, Azzaro et al. 2012). In temperate seas, the lateral advection of newly-formed water masses (both intermediate and deep) from convective regions as well as the lateral injection in the winter, of organic matter from the canyons and shelves, enhanced C respiration in deep layers (Packard et al. 2008, La Ferla et al. 2010). The horizontal PGE variability in the water layers suggested that, in mesopelagic waters, prokaryotes are able to use the available organic matter and convert it into biomass more efficiently than in epipelagic ones. This finding could be interpreted as an adaptive physiological response. It reflects the ability of deep-water microbes to efficiently exploit the available DOC at great depths. Placenti et al. (2018) suggested this mechanism for the deep Mediterranean Sea. Another aspect, that was unfortunately not considered, concerns the maintenance of the prokaryotic biomass and metabolism in terms of energetics when assessing the role of microbes in oceanic carbon cycles (Eichinger et al. 2010). According to Baltar et al. (2015), a combination of environmental stressors could enhance the proportion of the energy flux devoted to cell maintenance, inducing increases in cell specific respiration and decreases in PGE. In incubation experiments, slow-growing bacterial communities tended to have low PGE and to respire a high portion of the secondary production in terms of leucine uptake (del Giorgio et al. 2011). Moreover, in oligotrophic systems, low PGE may result from the maintenance of active transport and from the production of exoenzymatic hydrolysis with high bacterial energy demand (del Giorgio and Cole 1998).

However, using mean hourly normalised values, the ratio CDPR/PHP (µg C l^-1 h^-1) in the epipelagic layer was 3 times higher than in the mesopelagic one, presumably due to the occurrence of autotrophic respiration (Marra and Barber 2004). The prokaryotic carbon requirement (PCD) was particularly low in the mesopelagic layer where autotrophic production was lacking.

The significant relationship between CDPR and other physical and chemical parameters measured, suggests that respiration is strictly interconnected with environmental forces. Respiration varied in response to changes in hydrology according to Rivkin and Legendre (2001). In addition, it co-varied with different microbial parameters showing the consistent patterns of diverse aspects of microbial metabolism, as previously postulated by del Giorgio et al. (2011). The close link between the diverse aspects of prokaryotic patterns would imply that changes in the metabolic variables synergistically mediate the fate of organic matter by influencing the composition of organic material reaching the sediments (Catalano et al. 2006), the distribution of particulate and dissolved matter with depth (Carlson et al. 2000, Fabiano et al. 2000, Misic et al. 2017) and the remineralisation through the water column (Azzaro et al. 2006, 2012).

In all the stations, CSRR was surprisingly higher in the epipelagic layer than in the mesopelagic one. This suggests that a valuable contribution of organic matter of phytoplanktonic origin might sustain the heterotrophic metabolism in the upper layer. When we calculate CSRR by ETS Vmax, i.e. without utilisation of ETS to carbon conversion factors, almost all the mesopelagic values would be lower than surface ones with averaged CSRR value of 0.34 and 0.29 fg C cell^-1 in the epi- and mesopelagic layers, respectively (data not shown). Although different from other reports from temperate seas (Placenti et al. 2018, Baltar et al. 2009), the CSRR in epipelagic water was higher than in the mesopelagic one, suggesting more actively respiring cells in the upper layers. Nevertheless, high CSRR values were found at stations where TSI was low, suggesting the importance of cell-specific respiration in oligotrophic conditions, in agreement with the findings of Baltar et al. (2015) for subantarctic waters. Conversely, CSRR in Summer 2014 was higher in the deeper layers than in the surface ones (Misic et al. 2017). Nevertheless, high CSRR values were found at stations where TSI was low, which also corroborates the importance of prokaryotic respiration in the surface layer.

The cell-specific incorporation rate (CSIR) strictly reflected the distribution of PHA throughout the water column with the highest values in the surface to 50 m depth layer. The average CSIR was similar to that detected by Ducklow et al. (2001) and Pedrós-Alió et al. (2002) in the upper 50 m layer of the RS during Summer 1997 as well as in the Antarctic Peninsula, respectively. A decreased availability of organic carbon for synthesising new biomass could explain this finding (Baltar et al. 2015). The CSIR-temperature Spearman-rank relationships were positive in the photic layer and negative in the aphotic one (data not shown). The positive correlation observed in the epipelagic layer amongst heterotrophic rates, CHLa and respiration, allowed us to consider a direct or indirect (previous exoenzymatic hydrolysis) flux of labile DOC from phytoplankton biomass and detritus towards the new prokaryotic biomass. In the XIX PNRA expedition (VLTP-2004 project), a mean of 2.35 µg C l^-1 h^-1 potentially mobilised by leucine aminopeptidase + ß-glucosidase activities, was detected in the photic layer off Victoria Land (Monticelli, personal communication). In the photic zone of Terra Nova Bay (RS), during summer 2000, the daily flow of C towards new prokaryotic biomass was equivalent to 0.2% of C, potentially mobilised by exoenzymatic activities (Monticelli et al. 2003).

Overall, in Summer 2005, the investigated area of the RS contributed in different ways to the epi- and mesopelagic layer carbon metabolism. Per sea-surface area, the autotrophic (by C-CHLa), prokaryotic (PB) and total standing stocks (C-ATP) amounted to 1545, 1681 and 4605 mg C m^-2, respectively. The remaining heterotrophic component (HB) presumably accounted for a biomass of 1379 mg C m^-2. The prokaryotic biomass appeared to be predominant in the mesopelagic layer with respect to the epipelagic one (depth integrated PB ratio epi/meso: 0.4). The entire heterotrophic production accounted for 1.697 mg C m^-2 h^-1 with a similar weight in the epi- and mesopelagic layers (depth integrated PHP ratio epi/meso was 1.03). Respiration remineralised 8.279 mg C m^-2 h^-1 with higher rates in the upper layers (depth integrated CDPR ratio epi/meso was 2.7).