2 Um 
Q. Devresse et al.: Eddy-enhanced primary production sustains heterotrophic microbial activities 
surface winds occurred offshore (Supplement Fig. S5), which 
might explain the high Chl a concentration (>0.5ugL7!) 
that we found at the surface (5m) of all stations within the 
CE. Within the eddy, we observed that Chl a was higher in 
the western than in the eastern part of the eddy (Figs. 3b and 
4a). Chelton et al. (2011) showed based on satellite observa- 
:jons that due to the rotational flow and the westward propa- 
gation of CEs, Chl a tends to accumulate in their southwest 
quadrants while being lower in their northeast quadrants. To 
the best of our knowledge, this is the first time that high- 
resolution in situ sampling could demonstrate this specific 
submesoscale Chl a distribution within a CE. Outside of the 
CE boundaries, we noticed a thermal front with colder sur- 
face water. Thermal fronts have been detected outside of the 
periphery of eddies and are interpreted to result from eddy— 
eddy interaction (see review by Mahadevan, 2016) and/or 
eddy—-wind interaction (Xu et al., 2019). In this frontal zone, 
we observed higher nutrient concentrations than in the adja- 
cent stations including the western part of the CE periphery 
and a doming of the nutrient isolines, which indicates up- 
welling (see Fig. 2). Consequently, Chl a was elevated and 
‘compressed” to the surface in this area similar to in the CE 
(Fig. 4a). 
Our flow cytometry data (Fig. S6) showed that cyanobac- 
teria (Synechococcus) and eukaryotic pico- and nanoplank- 
con within the CE were unevenly distributed. This suggests 
that the phytoplankton community of the CE was likely dis- 
änct from the surrounding waters but also variable on the 
submesoscale within the CE. This is consistent with previous 
studies on phytoplankton distributions in eddies (e.g. Lochte 
and Pfannkuche, 1987; Lasternas et al., 2013; Hernändez- 
Hernändez et al., 2020). Moreover, the mixed layer was also 
nighly variable within the CE and so were PPrTor rates (Ta- 
ole S1, Figs. 3 and 6). We observed a 3-fold variation in 
depth-integrated PPror rates over 100m depth (Table 1) 
within the CE which is consistent with earlier observations 
of a 5-fold variation in primary production integrated over 
ihe euphotic zone in a CE in the subtropical Pacific Ocean 
(Falkowski et al., 1991). Overall, primary productivity was 
enhanced within the CE and the frontal zone with an av- 
erage 4-fold increase in depth-integrated PPrTor rates over 
100m depth than in the open ocean and coastal stations. 
This is consistent with Löscher et al. (2015), who found that 
depth-integrated primary productivity over the Chl a maxi- 
mum of a CE in the Mauritanian upwelling system was 3-fold 
higher than in the surrounding waters. Extracellular release 
rates (PPpoc) were also enhanced within the eddy, but PER 
was slightly lower at the eddy surface (Fig. 4d, e). We pro- 
pose two hypotheses regarding this distribution: (1) the lower 
PER was due to a higher proportion of larger phytoplankton 
(e.g. diatoms), which have lower turnover rates and there- 
fore lower PER (Malinsky-Rushansky and Legrand, 1996) 
and/or (2) the upwelling of nutrients generated by the CE 
might have enhanced the physiological health of the phyto- 
plankton community (Agusti and Duarte 2013). 
4.2 Variations in heterotrophic bacterial abundance 
and activity associated with a cyclonic eddy 
Along the zonal transect, in the stations not affected by the 
eddy (open ocean + coastal stations), a significant positive 
correlation was observed between HB abundance and PPrTot 
rates (Fig. 7a). Those variables were rather uniformly dis- 
tributed from the coast to the offshore waters excluding sam- 
ples influenced by the eddy, which is in agreement with ear- 
lier findings by Bachmann et al. (2018) for the Mauritanian 
upwelling system during summer. Both our BR and BP were 
also within the range of reported rates for coastal and off- 
shore waters of the CanUS (Reinthaler et al., 2006; Alonso- 
Säez et al., 2007; Vaque et al., 2014). BP rates slightly de- 
creased from the coast to the open ocean when samples from 
the eddy were not considered. Similar trends were found 
in the CanUS with different upwelling intensities and dur- 
ing different seasons (Alonso-Säez et al., 2007; Vaque et al., 
2014). The distinet distribution of BP and BR rates affected 
the distribution of the BGE, which was higher in the coastal 
than in the open ocean stations. Overall, our BGESs represent 
the lower end of global ocean values, but similarly low BGEs 
have been observed for other EBUS, such as the CanUS 
(Alonso-Säez et al., 2007), the California upwelling system 
(del Giorgio et al., 2011), and the Humboldt upwelling sys- 
tem (Maßmig et al., 2020). Yet, we report an average BGE 
2 times lower than Alonso-Säez et al. (2007), which may 
be due to differences in upwelling intensity. Indeed, Kim et 
al. (2017) denoted that BGE increased with increasing up- 
welling intensity in the Ulleung Basin. At the coast, PPDoc 
rates were sufficient to compensate for the BCD, indicating 
a strong trophic dependence of bacteria on phytoplankton, 
whereas in the open ocean PPpoc rates covered between 
2.6 % to 78 %, indicating a much lower trophic dependence 
of bacteria on phytoplankton. Therefore, in the open ocean, 
other carbon sources (i.e. PPpoc, SL-DOC) must have been 
used to compensate the BCD. SL-DOC compounds have a 
turnover of weeks to months, which allows them to escape 
rapid microbial degradation (Hansell et al., 2009). Conse- 
quently, we hypothesise that the BCD in the open ocean was 
sustained through SL-DOC produced in excess near the coast 
and transported offshore. Indeed, in the CanUS, currents and 
eddies have been shown to laterally transport DOC offshore 
up to 2000 km (Lovecchio et al., 2018). 
Within the CE-influenced stations (CE + frontal zone), HB 
abundance was disconnected from the PPrTor rates (Fig. 7b). 
For example, in the southwestern periphery and the frontal 
zone HB abundances were relatively low, while both PPTorT 
rates and Chl a concentrations were relatively high (Fig. 4a, 
c). Hernändez-Hernändez et al. (2020) reported a similar ob- 
servation with a strong heterogeneity of HB biomass distri- 
bution within a CE in the CanUS. Attachment to particles, 
viral lysis, or grazing by nanoflagellates might have led to a 
selective reduction in HB abundance. However, the exact rea- 
sons for the low HB occurrence at the eddy periphery and the 
Biogeosciences, 19, 51995219. 202. 
https://doi.org/10.5194/bg-19-5199-2022