Hydrographic fronts shape productivity, nitrogen fixation, and microbial community composition in the southern Indian Ocean and the Southern Ocean

Abstract. Biogeochemical cycling of carbon (C) and nitrogen (N) in the ocean
depends on both the composition and activity of underlying biological
communities and on abiotic factors. The Southern Ocean is encircled by a
series of strong currents and fronts, providing a barrier to microbial
dispersion into adjacent oligotrophic gyres. Our study region straddles the
boundary between the nutrient-rich Southern Ocean and the adjacent
oligotrophic gyre of the southern Indian Ocean, providing an ideal region to
study changes in microbial productivity. Here, we measured the impact of C
and N uptake on microbial community diversity, contextualized by
hydrographic factors and local physico-chemical conditions across the
Southern Ocean and southern Indian Ocean. We observed that contrasting
physico-chemical characteristics led to unique microbial diversity patterns,
with significant correlations between microbial alpha diversity and primary
productivity (PP). However, we detected no link between specific PP (PP
normalized by chlorophyll-a concentration) and microbial alpha and beta
diversity. Prokaryotic alpha and beta diversity were correlated with
biological N2 fixation, which is itself a prokaryotic process, and we detected
measurable N2 fixation to 60∘ S. While regional water masses
have distinct microbial genetic fingerprints in both the eukaryotic and
prokaryotic fractions, PP and N2 fixation vary more gradually and
regionally. This suggests that microbial phylogenetic diversity is more
strongly bounded by physical oceanographic features, while microbial
activity responds more to chemical factors. We conclude that concomitant
assessments of microbial diversity and activity are central to understanding
the dynamics and complex responses of microorganisms to a changing ocean
environment.


mm pre-combusted GF/F filter (<10 kPa). Filters were snap-frozen in liquid nitrogen and stored at -80° C while at sea. Filters 121 with enriched (T24) and unenriched (T0) samples were acidified and dried overnight at 60° C. Analysis of 15 N and 13 C 122 incorporated was carried out by the Isotopic Laboratory at the UC Davis, California campus, using an Elementar Vario EL 123 Cube or Micro Cube elemental analyzer (Elementar Analysensysteme GmbH, Hanau, Germany). 124 Carbon assimilation rates were calculated according to Knap et al. (1996), excluding the 14 C -12 C conversion factor; and N2 125 fixation was calculated according to Montoya et al. (1996), respectively. The minimum quantifiable rate was calculated 126 according to Gradoville et al. (2017). 127

Pigment analysis 128
For pigment analyses, 4 L of seawater were filtered (< 10 kPa) on 47 mm Whatman GF/F filter and stored at -80° C until 129 further analysis. High-Performance Liquid Chromatography (HPLC) was carried out as described in Kilias  respectively (S1_code_archive/dada2/dada2_16S.R L88:104; S1_code_archive/dada2/dada2_18S.R L92:104). Primer-165 trimmed fastq files were quality trimmed with a minimum sequence length of 50 bp, and checked by inspection of the average 166 sequence length distribution (for both the 16S rRNA gene and 18S rRNA gene sequences). Samples within forward and reverse 167 fastq files were dereplicated and merged with a minimum overlap of 20 bp. ASV tables were constructed and potential chimeras 168 were identified de-novo and removed using the removeBimeraDenovo command. Sequencing statistics for removed reads and 169 sequences in each step can be found in Table S3 Kerguelen island, we consider the SAF as part of the convergence zone between the SO and IO, the South Subtropical 216 Convergence province (SSTC), rather than as a Southern Ocean frontal system. At 7 m depth, we noted clear shifts in 217 temperature (SST), salinity, and dissolved inorganic nutrient (NO3 -, PO4 3-, Si) concentrations when crossing the STF. The STF 218 is described as a circumpolar frontal zone creating the boundary between our measurements of the warm (20-25 °C), saline 219 (>35), and oligotrophic (NO3 -< 0.03 µM; PO4 3-: 0.04 -0.21 µM) subtropical waters (STW) of the Indian South Subtropical 220 Gyre (ISSG) and the cold (3-6 °C), macro-nutrient rich (NO3 -: 19.2 -24.9 µM; PO4 3-: 1.43 -1.71 µM) SO (Fig. 1, Fig. 2, Fig.  221 S3). In the context of this study, STW and ISSG could be used interchangeably; we refer henceforth to ISSG. 222

Primary productivity (PP) 223
Maximum primary productivity (PP) within our dataset were measured near the Kerguelen plateau in the Polar Front Zone 224 (PFZ) at Station 9 (3236.8 and 3553.3 µmol L -1 d -1 , respectively) and Station E (2212.4 -2688.1 µmol L -1 d -1 , n = 6). Comparing 225 all PP measurements across water masses, we found relatively high PP in other stations of the PFZ (Stations 6, 7; Fig. 3a; 226 Overall, the variation of specific primary productivity (P B ) did not show great variations between provinces, with maximum 231 rates at station 11 (Table 1; Fig. 3b). We did not find a significant correlation between mixed layer depth and P B (Pearson 232 correlation; r = 0.21, p = 0.47, n = 12). 233

Phytoplankton pigment analyses 242
Photosynthetic pigment concentrations showed a clear separation between the oligotrophic ISSG and the nutrient-rich SO ( S5a). Across water masses, fucoxanthin concentration was slightly higher in the AZ (0.06 -0.5 mg m -3 , n = 4) than in all other 255 water masses (0 -0.13 mg m -3 , n = 10). 256 The distribution of potential phytoplankton size classes (PSCs; pico-nano-and microplankton), calculated from diagnostic 257 pigments (Supplementary A), showed a clear pattern over temperature variations (Fig. S5b). The pigment data suggested that 258 picoplankton dominated warm water in the ISSG, picoplankton abundance sharply decreased (second-order polynomial fit R 2 259 = 0.98, p <0.001, n = 14) at lower values of SST. Pigment data also suggested that microplankton showed a contrary trend to 260 the relative fraction of picoplankton, having high abundance in cold-water and decreasing at higher values of SST, with a 261 minimum at 20°C SST and a slight increase (14% microplankton of all phytoplankton size classes) towards 25°C SST (second-262 order polynomial fit R 2 = 0.77, p <0.001, n = 14). Nanoplankton showed a maximum at 12°C SST and decreased both towards 263 warmer and colder waters (second-order polynomial fit, R 2 = 0.58, p < 0.01, n = 14). 264

Eukaryotic planktonic community composition 265
For each station, except station 4, the V4 region of the small subunit ribosomal RNA gene (18S rRNA) was amplified and 266 sequenced to determine the community composition of micro-, nano-, and pico-eukaryotes in all three oceanic provinces. We 267 recovered a total of 2618 ASVs. After removing sequences annotated to metazoans, 2501 ASVs remained (4.4% of ASVs 268 removed). 269 We found a strong correlation between both eukaryotic richness and diversity (Inverse Simpson Index) with SST (Pearson 270 correlation, r = 0.85, p < 0.001 for Richness, and r = 0.82, p = 0.001 for Inv. Simpson, n = 12, respectively; Fig. S7a, S7c). 271 Overall, eukaryotic diversity was negatively correlated with PP (r = -0.66, p = 0.02, n = 12; Fig. S7e) and significantly and 272 positively associated with N2 fixation (r = 0.74, p = 0.01, n = 12; Fig. S7g). However, a strong correlation between rate 273 measurements (PP, N2 fixation) and eukaryotic diversity was only apparent in the ISSG, and no significant across other 274 provinces (Pearson correlation after removal of ISSG samples from dataset: PP r = 0.47, p = 0.24, and N2 fixation, r = -0.48, Our RDA constrained 81% of the variance in the ASV table, with a p-value of 0.095 (Permutations = 999, n = 12). Sites were 277 well separated between Longhurst provinces along the first two RDA axes (capturing 52.67% constrained variance, Fig 4a). 278 Our PERMANOVA, which tested the province-based separation, produced moderate but significant results (Permutations = 279 999, R 2 = 0.54, p = 0.001; n = 12). An additional PERMANOVA grouping sites by water masses produced similar results 280 (Permutations = 999, R 2 = 0.67, p = 0.001, n = 12; Fig. 4a). We found that more ASVs only occurred in one province, rather 281 than in two or more provinces (Fig. 4e). Sites within the ISSG province were associated with SST and N2 fixation. Sites in the 282 SSTC were associated with high NH4 + concentrations. Sites belonging to the SO were associated with dissolved inorganic 283 nutrients (NO3 -, PO4 3-, Si), dissolved oxygen, and chl a concentrations as well as high PP. Linear relationships between beta 284 diversity and rates were only weak for PP (PERMANOVA; Permutations = 999, R 2 = 0.27, p = 0.004, n = 12) and both weak 285 and insignificant between beta diversity and N2 fixation (PERMANOVA; Permutations = 999, R 2 = 0.13, p = 0.14, n = 12). 286 Investigating whether and at which magnitude environmental parameters have an effect on microbial community dissimilarity, 287 our general dissimilarity model (GDM) showed the expected curvilinear relationship between the predicted ecological distance 288 and community dissimilarity (Fig. 4c I). Based on I-spline magnitudes of all tested environmental variables, geographic 289 distance had little effect on community dissimilarity (Fig. S11a). Community dissimilarity changed most notably in response 290 to variability in low magnitudes of PP (i.e. ISSG and STF; 17% of total community dissimilarity, n =12) and plateaued with 291 PP above 1100 µmol C L -1 d -1 (Fig. 4c III). A community dissimilarity change occurred most notably when N2 fixation when 292 rates were above 2 nmol L -1 d -1 (~ 19% of change in total community dissimilarity associated to changes in N2 fixation rates. 293 Fig. 4c IV). Among all tested environmental parameters, our I-spline results showed that community dissimilarity increased 294 most in response to variability in MLD and PO4 3concentrations (49% of change in total community dissimilarity associated 295 to MLD variability, and 63% to PO4 3variability, respectively, n = 12; Fig. S11a). 296 Significant differences in community dissimilarity structure between Longhurst provinces were associated with high-297 pseudocount taxa, dominated by dinoflagellates (Dinophyceae) and diatoms (Bacillariophyta; SIMPER analysis; Table S6). 298 The pseudocount of ASVs belonging to the phylum Ochrophyta (Bacillariophyta_X) contributed to differences between ocean 299 provinces (contributing to at least 9.51% of the differences in community dissimilarity between the SO and ISSG). Moreover, 300 4.79% of the differences in community dissimilarity between the SO and the SSTC were associated with a higher ASV count 301 of Bacillariophyta_X ASVs in the SO. Further, we identified ten ASVs belonging to the phylum Dinophyceae contributing 302 with 2.1% to the community dissimilarity structure between the SO and ISSG; and with 5.79% to the community dissimilarity 303 structure between the SSTC and ISSG. This was further supported by relatively high concentrations of the photosynthetic 304 pigments chl c3 and peridinin (both indicative pigments for dinoflagellates) in the SO and SAF. We found a relatively high 305 number of ASV94 and ASV23 (Chloroparvula pacifica) in the SSTC, contributing 3.07% to the community dissimilarity 306 between the SSTC and the ISSG.

Prokaryotic community composition 308
From each of 14 stations, a fragment of the small subunit ribosomal RNA gene (16S rRNA) was amplified and sequenced to 309 obtain insights into the diversity and community composition of prokaryotes. A total of 1308 ASVs was recovered from which 310 we removed 267 ASVs annotated as chloroplasts and 68 ASVs annotated as mitochondria. Prokaryotic richness increased with 311 increasing sea surface temperature (Pearson correlation: r = 0.65, p-value = 0.03, n = 11; Fig. S7a). Maximum alpha diversity 312 (Inverse Simpson) estimate was found in the SAF (81.92, Station 15; Fig. S7d). Prokaryotic alpha diversity (Inverse Simpson) 313 was positively (but not significantly) linked to primary productivity (r = 0.36, p = 0.55, n = 11; Fig. S7f) but showed a 314 significant negative correlation with N2 fixation (r = -0.7, p = 0.05, n = 11; Fig. S7h). 315 Our RDA of the prokaryotic ASV table captured 90% of the total variance with a p-value of 0.06 (Permutations = 999, n = 316 11). Sites clustered into Longhurst provinces along the first two RDA axes (62.48% of variance constrained; Fig 4b). This was 317 also shown in the PERMANOVA solution for Longhurst provinces (Permutations = 999, R 2 = 0.62, p < 0.001, n = 11) and our 318 PERMANOVA grouping into water masses (Permutations = 999, R 2 = 0.74 p < 0.001, n = 11; Fig. 4b). We found more ASVs 319 occurring in either the ISSG or the SO provinces rather than across all provinces (Fig. 4f). Further, the ISSG and the SO shared 320 Alphaproteobacteria (5.69% of the total difference in community dissimilarity, SIMPER analysis, Table S6). The ISSG was 340 characterized by a high number of Cyanobacteria and some Actinobacteria. The cyanobacterial fraction was dominated by 341 Prochlorococcus and Synechococcus, respectively. 342 Within the class level, all stations were dominated by Alpha-, and Gammaproteobacteria, Bacteroidia, Oxyphotobacteria 343 (Cyanobacteria), and Verrucomicrobia. Within the Alphaproteobacteria, we found a great dominance of ecotype I, II, and IV 344 of SAR11 clade throughout all samples (Table S4). The relative number of pseudocounts of bacteria belonging to the phylum 345 Bacteroidetes decreased towards warmer SST in the ISSG, with significant differences between the SO and ISSG Our results suggest that regional N2 fixation was not limited by the presence of other sources of bioavailable N (Fig. S10) Our results also strongly suggest that prokaryotic community structure and composition (beta diversity) were strongly impacted 395 by the presence of biological N2 fixation, itself a prokaryotic process (Karl et al., 2002). For example, the N2-fixing 4.2 Total and specific primary productivity differentially affect microbial diversity 400 We found PP was highest in the PFZ and decreased towards higher latitudes in the SO (Fig. 3a) between oceanic provinces (n = 12, nISSG= 4, nSSTC = 3, nSO = 4), it did suggest that further validation of this assumption is 421 needed. We observed that PP changed gradually across the sampling region, and that local variability in PP was high between 422 samples taken ~15 km apart within the SSTC and SO (Fig. 3a) (2018), we found a negative correlation between eukaryotic alpha diversity and PP within the ISSG. Further, we found no 425 correlation between eukaryotic diversity and PP within the SSTC and SO and none between prokaryotic alpha diversity across 426 all provinces (Fig. S7). 427 In terms of beta diversity, we observed a structuring effect of PP for both pigment-, 16S rRNA gene-, and 18S rRNA gene-428 derived diversity profiles (Fig. 4 a,b, Fig. S5). Pigment analysis revealed that photosynthetic prokaryotic diversity is strongly 429 impacted by the relative abundance of Prochlorococcus, which does not generally occur in cold, high-latitude waters (>40°S/N; 430 3) both Prochlorococcus and Synechococcus, were not detected in the SO (Table S4, Table S6). In the SSTC and SO, 434 phytoplankton communities had high proportions of dinoflagellates (Dinophyceae) and diatoms (Bacillariophyta) (up to 74% 435 of diatom diagnostic pigment concentrations), which are known as essential contributors to marine PP and microbial diversity 436 Further, our results show that phytoplankton community structure appears to be tightly coupled to the occurrence of specific 439 heterotrophic organisms (Table S6)

Implications for microbial regionality 444
fraction of total biomass in the ISSG than in the SO. However, we did not measure zooplankton biomass or grazing rates, so 462 this remains speculative. 463

Conclusion & outlook 464
Our study leads us to conclude that simultaneous assessment of microbial diversity, biogeochemical rates, and the physical 465 partitioning of the ocean (provincialism) is central to the understanding of microbial oceanography. 466 Each water mass in our study had a distinct microbial fingerprint, including unique communities in frontal regions. Microbial 467 alpha diversity and community dissimilarity correlated with biogeochemical rate measurements; however, mechanisms driving 468 this association need further investigation through high-resolution sampling across spatial and temporal scales. Our results also 469 indicate that high-latitude N2 fixation could meaningfully contribute to the global and regional N-pool (as reported for Arctic 470 N2 fixation by Sipler et al., 2017), which may become especially significant as global stratification (and concomitant 471 restrictions in deep water replenishment of nutrients) intensifies. 472 While our sampling is too limited to conclude the point, our observations that phylogenetic diversity is constrained by 473 hydrographic properties and province boundaries, but biogeochemical rates and nutrient concentrations are changing more 474 gradually suggests that trans-province functional redundancy is present despite strong biogeographic separation in 475 phylogenetic terms. As an outlook, we therefore encourage examining both phylogenetic and functional diversity to assess 476 how functional groups and guilds contribute to the major biogeochemical (C, N) cycles across provinces and other 477 biogeographic regions. Coordinated studies across ocean provinces are key to establishing the baselines we need to monitor 478 the rapidly changing properties of the Southern high-latitudes in the face of rising temperature, acidification, and perturbations 479 in regional currents.