POC and PON are considered the main resource supporting diverse microbial communities in both the suspended and sinking particle-pools . The range of activities associated with this diverse community of microbes alters the nature of the particulate and dissolved pool and thus contributes to the quantity of carbon or nitrogen, which is effectively exported below the seasonal mixed layer. As such, an understanding of the prokaryotic community composition and their functional capacity on both sinking and suspended particles may facilitate a more mechanistic understanding of microbial contributions to POC/PON degradation and/or synthesis. This study expands our understanding of the prokaryotic community contributions to organic carbon and nitrogen export by providing the first metagenome assembled genomes from suspended and sinking particle pool fractions in the Southern Ocean.
Differences in prokaryotes may explain the divergence in POC and PON concentration in both the sinking and suspended particle pools
Recent studies have shown a positive relationship between phytoplankton biomass and the magnitude of POM export [32, 33]. These findings were corroborated by our data showing a positive relationship between MLD integrated chlorophyll and POM flux. However, this relationship was poor (r2 = 0.33 for POC and r2 = 0.06 for PON) suggesting that phytoplankton biomass may only account for ~30% of carbon flux variability and as little as ~6% of nitrogen flux variability. This is perhaps unsurprising considering the many factors that influence the concentration of POM, which is effectively exported out of the surface layer (e.g. sinking rates and prokaryotic activity etc.) [34, 35]. Previous studies have demonstrated that POM content (labile, semi-labile, recalcitrant, or refractory), rather than POM concentration, is the main driver of prokaryotic community structure [36, 37]. There is also evidence showing that as POM sinks through the mesopelagic zone it is subjected to degradation by several free-living and particle-associated prokaryotes . As such, prokaryotes alter the chemical constituency of POM and may in turn also contribute as secondary drivers of change, altering their community structure [22, 38]. Labile or semi-labile , recalcitrant or refractory DOM production  from prokaryotic degradation of POM includes the production of so called ‘sticky polysaccharides’, which form part of the aggregation of DOM into suspended POM and sinking POM . In this instance, both the suspended and sinking POM serve as the main source of carbon and energy for prokaryotes in the mesopelagic [42, 43]. Since POC flux to the mesopelagic is insufficient to support the carbon demand of prokaryotes, suspended particles are considered a major sustaining source of organic carbon for microbes in the mesopelagic . While the concentration of POC in sinking particles decreases exponentially with depth, the concomitant POC concentration in suspended particles remains largely constant and is typically ~1-2 orders of magnitude higher than that of sinking particles . This finding is similar to our POC data, which was substantially higher in the suspended vs sinking fraction. However, the same was not true for PON with more PON concentration at stations 1-3 and similar PON concentration at stations 4 and 5 in the sinking material when compared to suspended.
Several factors may account for the widespread variability in the POC:PON ratio observed on suspended and sinking samples in this study. These include i) the preferential degradation of nitrogen rich POM  (most notably in the suspended samples at stations 1 and 2 where POC:PON ratios were >30), ii) the synthesis of refractory POC resistant to further degradation , which would also drive high POC:PON ratios, iii) chemoautotrophic microbial activity on the POM, increasing the POC:PON ratio  and iv) the oxidation of sinking POC by marine microbiota , which may drive a preferential reduction in POC relative to PON, thereby decreasing POC:PON ratios as evidenced in all sinking samples where POC:PON ratios were less than the Redfield ratio of 6.6. Despite the large differences in the distribution of POC:PON ratios between suspended and sinking samples, and between stations, the bacterial community composition was very similar. This suggests that variability in POC:PON ratios may be a product of variability of microbial activity despite similarities in bacterial community structure. On the other hand, differences were observed in archaeal communities such that the PON content might have been different between stations. Alternatively, the composition of the source material may have been similar but may be acted upon differently by the bacteria and archaea driving secondary changes in their community structure [12, 22, 48]. Evidence for this argument can be seen in the relative abundance of bacterial MAGs, which demonstrates variability in prokaryotic activity between suspended and sinking pools, and between stations, despite similarities in community composition. As such, our results suggest that differences in prokaryotic activity rather than diversity, particularly in the case of bacteria, impact the signature of POC and PON in both sinking and suspended material. Nevertheless, examples of the impact of archaeal diversity on POC:PON variability (although most likely secondary) are evident when observing Nitrososphaeria, which were highest at stations 1 and 2 where the highest POC:PON ratios were encountered, together with the highest POC flux (most notably at station 2). This implies that Nitrososphaeria may be actively involved in PON degradation, a notion that is corroborated by the presence of nitrogen metabolism in Nitrososphaeria at stations 1 and 2. Conversely, station 3 had the least amount of Nitrososphaeria and observed a particularly high PON flux relative to all other stations.
Prokaryotic ecological strategists based on POC and PON content
Prokaryotes may exhibit different ecological strategies in response to POC content [49, 50]. Previous studies suggest that r-strategists may be more prevalent in sinking particle-pools where they degrade transient POM [50, 51], whereas K-strategists appear to exploit complex compounds (e.g. RDOM) from the suspended particle-pool [29, 50]. As such, a niche differentiation would be expected in prokaryotic community and functional activity between sinking and suspended pools. However, prokaryotes are also known to detach from sinking particles, potentially enhancing the suspended particle-pool with microbiota that are similar to the sinking particle-pool . Metabolic activities associated with K-strategists include CO2 fixation via the Wood-Ljungdahl pathway, the Calvin cycle, Arnon-Buchanan cycle and the Hydroxypropionate-hydroxybutyrate cycle by condensing two molecules of CO2 as electron acceptor and hydrogen as electron donor into Acetyl-CoA as building blocks for biosynthesis. Prokaryotes exhibiting these metabolic activities are typically chemoautotrophs, which synthesise complex organic carbon such as RDOM or polysaccharides polymers from CO2 .
In our results, these chemoautotrophic bacteria MAGs, typical of K-strategists, were more prevalent in the sinking particle pool than the suspended pool, which is in contrast to previous studies . However, it is likely that any metabolic activity which uses POM to form polymers may consequently initiate aggregation and subsequent sinks enhancing POM export flux, thus accounting for their presence on sinking material. On the other hand, there was no discernible difference in the functional profiles between suspended and sinking material associated with r-strategists and K-strategist, respectively. For example, bacterial MAGs from Gammaproteobacteria were present in both the sinking and suspended sample at station 2 and possessed CAZymes involved in the degradation of labile POM such as diatom-derived POM , grass POM , and virus-induced POC from picocyanobacterial and polysaccharides , whose expected role would be to reduce carbon flux via particle degradation while sinking into the mesopelagic. Similarly, all bacterial MAGs were associated with the degradation of chitin, regardless of their association with sinking or suspended material. Chitin is rich in both carbon and nitrogen and can be reintegrated into biomass forming polysaccharide polymers or mineralized to enrich the water column with inorganic carbon and nitrogen, reducing both the POC and PON export flux . As such, our results suggest that a combination of microbial driven transition between suspended particles and the formation of aggregates (e.g. via the synthesis of sticky polysaccharides) and the dissociation of microbes from the sinking particle pool to the suspended particle pool  make it difficult to discern any specific bacterial preference of r-/K-strategists for one particle type over another. This is contrary to some studies which suggest specific biogeochemical roles for prokaryotes in suspended and sinking particle-pools in the marine carbon cycle .
On the other hand, our chemoautotrophic archaea MAGs mostly Nitrososphaeria were determined to be more prevalent at station 1 and 2 (suspended only) and relatively less abundant at stations 3 to 5 (both suspended and sinking fractions). As with bacteria, chemoautotrophic archaeal MAGs appear to be K-strategists and are expected to dominate the suspended particle pool, which was indeed the case for our samples at stations 1 and 2. This is more in-line with predicted archaeal MAGs distribution that is said to be more dominant in r-strategists where they scavenge for PON in the suspended particle-pool . The ammonia oxidizing archaea (AOA) present on sinking samples at stations 1 and 2 (Nitrososphaeria) and the suspended sample at station 1 (Poseidoniia) directly utilise simple and complex organic nitrogen as their main source of ammonia and nitrite . Based on the POC:PON ratio it appears that the suspended particle-pool might be more influenced by AOA resulting in low concentrations of PON relative to POC in the suspended fraction, driving ratios in the suspended material that far exceed Redfield at all stations reaching >30 at stations 1 and 2. Surprisingly, Cyanobacteriia had no genes linked to the nitrogen metabolism pathway. However, the sinking sample at station 2 possessed genes for nitrogen metabolism and also had the highest POC:PON ratio. Nitrogen metabolism was also more prevalent in archaeal compared to bacterial MAGs. The presence of AOA (Nitrososphaeria and Poseidoniia), NOB (Gammaproteobacteria) and NOA (Nitrososphaeria) MAGs at station 1 (suspended and sinking) and station 2 (suspended) had the highest POC:PON ratio, the highest POC flux and the lowest PON flux. These NOB/NOA and AOA are obligatory partners where the NOB/NOA catalyse the degradation of PON to ammonia for the nitrite producing AOA partner . This may explain the decrease in PON export flux observed at station 1 and 2. NOB/NOA, and AOA are also key players in the removal of nitrogen from PON, increasing the POC:PON ratio at station 1 and 2, thereby increasing the POC export flux relative to PON export flux. In addition to preferential degradation of PON, Archaeal MAGs may also be involved in the synthesis of RDOC, enriching the water column with organic carbon particles, increasing POC export flux relative to the PON flux .
Although Cyanobacteriia are well known photosynthetic microbes involved in nitrogen fixation, the fact that our MAGs showed no evidence for nitrogen fixation is surprising [61, 62]. A possible reason for this may be the presence of non-cyanobacterial diazotrophs (NCD), such as dinitrogen (N2) fixing bacteria and archaea . Indeed, Gammaproteobacteria (on the sinking sample at station 2) were the only bacterial MAG containing nitrogen metabolism, while Nitrososphaeria were also present at station 1 and 2. Poseidoniia MAG at station 1 (suspended) also had metabolic capacity for nitrogen metabolism. Coincidentally, these were the two stations with the highest POC:PON ratio (and highest carbon flux) indicative of preferential nitrogen uptake by the prokaryotic community. Since phytoplankton biomass can account for only ~6% of the nitrogen flux, the high PON flux observed in station 3 to 5 might be due to prokaryotic activity on PON by assimilating the available inorganic nitrogen into its biomass . The dissimilation of inorganic nitrogen from PON  which favours PON export is thus more likely to be a result of prokaryotic activity than phytoplankton biomass.