Seawater samples were collected at 7 depths (from 50 to 10,918 m) of the CD (Table 1). Our results show that DIP concentration was relatively low (0.11 µmol L− 1) in surface water (50 m). This concentration is comparable to those reported for surface waters in the Pacific subtropic gyres6, the HOT station2, and the Arabian Sea 5, but higher than those in central, eastern parts of the North Atlantic Gyre22. However, DIP concentrations were rather uniformly significantly higher in the deep waters (1,000–10,918 m), ranging between 2.22 and 2.91 µmol L− 1 and averaging 2.49 ± 0.25 µmol L− 1. These results are comparable to those for the Mariana Trench reported in an early study13.
Different than DIP distribution, dissolved organic P showed a stratified pattern of distribution in the CD water column, with high and low concentrations alternating between successive sampling depths (Fig. 1). DOP concentrations varied from 79 to 463 nmol L− 1 (averaging 214.0 ± 152.3 nmol L− 1) and differed between consecutive depths by up to one order of magnitude. Furthermore, DOP concentrations were 1–2 orders of magnitude lower than DIP, opposite to that commonly observed in non-trench waters23. Notably, DOP represented on average 46.1% of the total dissolved phosphorus (TDP) in surface water and only 8.4% in the deep waters (Table S1), much lower than those (55–92%) reported for non-trench waters in the Pacific and Atlantic oceans6,24. Furthermore, the turnover time of the DOP pool was 13.0 months in surface water and from 5.1 to 6.9 months in the deep waters (Table S1), much less than those previously reported for the Atlantic Ocean25.
We measured the potential activity of alkaline phosphatase using fluorogenic substrate 4-methylumbelliferyl-phosphate, extending our previous work to the full ocean depth15. The most striking observation was the contrasting patterns in DIP and APA between surface water and deep waters. The CD surface water had the highest APA with the lowest measured DIP and DOP concentrations (Fig. 1). On the contrary, the CD deep waters had high concentrations of DIP and concurrently, elevated levels of APA. APA averaged 17.0 ± 1.1 nmol h− 1, which is 2–3 orders of magnitude higher than those reported for non-trench waters in the Pacific Ocean4. Similarly, exceptionally high Km values (60.3–221.6 mmol, averaging 71.7 mmol) were measured in the water column, suggesting that the CD microbial communities had great potential to maximize APA for utilizing DOP. Thus, our results showed that two distinct P-APA regimes existed in the CD water column, one in the surface water having low DIP and high APA, and the other in the deep waters with high DIP and elevated APA (Fig. 1).
Alkaline phosphatase, which releases P bound in DOP, is a large group of phosphomonoesterases with a relaxed substrate specificity 2. APA can be induced by DIP-stressed or -limited microbial communities in environments where DIP concentrations are low25,26. Therefore, APA is commonly used as an indicator for P-limit conditions in the aquatic system. The detection of elevated concentrations of DIP in the deep waters (Fig. 1) of the Challenger Deep suggests that microbial communities in the deep waters of the trench were not P-limited. However, different from the non-trench deep waters, e.g., the central Pacific Ocean4 and central Atlantic Ocean27, the CD deep waters retained both elevated concentrations of DIP and exceptionally high levels of APA. The measured APA were orders of magnitude higher than those measured in other oceans, like the Atlantic Ocean27,28, the Mediterranean Sea29, and the Tyrrhenian Sea30.
The exceedingly high concentrations of DIP with concurrent high levels of APA in the CD deep waters contradicts with the common notion that high abundances of DIP repress microbial synthesis of AP9,26,31,32. We further examined the nutrient chemistry of the water column, showing that the CD microorganisms were periodically carbon-starved. This is evidenced by the depth-alternating C/P ratios in the water column (Table 1). The surface water had a C/P ratio of 603:1, much higher than the canonical Redfield ratio, whereas the deep waters exhibited alternating low and high C/P ratios, which were, respectively, lower and higher than the Redfield ratio (Table 1). Thus, the two P-APA regimes in the CD water column exhibited different patterns in distribution of DIP, DOP, APA, Km and microbial community compositions (described below), reflecting the different nutrient utilization strategies of the microorganisms in the CD water column. Specifically, microorganisms in the surface-water regime produced APA in response to P-stress, whereas those in the deep-water regime responded in an adaptive manner to C-limitation, implicating the intimate linkage between P cycle and C cycle in the trench waters. We posit that the deep-water microbes likely utilized a piggyback strategy to liberate organic carbon (OC) from breaking DOP by expressing high levels of APA, concurrently leading to high abundances of DIP in the deep33,34. In fact, the proportions of the regenerated DIP (surface water, 92.2%; deep waters, average 65.2%) from DOP breakdown in the CD water column were much higher than the global average (36%; Table S1)35. Microbial expression of enhanced AP activity in the deep waters represents a strategy for microbial liberation and consumption of organic moieties derived from DOP, irrespective of P availability in the environment26,36. We interpret the contrasting patterns of APA, DIP and DOP in the surface water and deep waters reflecting microbial niche partitioning and adaptation to the two different P-APA regimes in the water column. This argument is also supported by the calculated potential in situ hydrolysis rate of DOP (VDOP)6, which showed a strong linear relationship with DOP (R2 = 0.983; p < 0.003, data not shown). Viewed together, these results demonstrated that microbial APA in the CD deep waters was responding to carbon starvation and access to OC, rather than to phosphorus depletion.
It appears that input of AP-hydrolyzable DOP to the deep trench water was in a pulsed mode, the dynamics of which are well illustrated by the pattern of alternating high and low values of DOP, C/P ratio, and VDOP (Fig. S1). Additionally, the measured Km values were 2–3 orders of magnitude higher than the measured DOP concentrations, implying that the CD microorganisms must be adapted to use the pulsed AP-hydrolyzable DOP in the water column as microorganisms can adjust their utilization of available DOP in the environment6. Thus, we hypothesize that the intermittent pulse inputs of DOP from the surface to the deep water provided a mechanism of P and C supply and the resultant APA impulses for regeneration of DIP and DOC. Consequently, the deep trench waters exhibited high levels of APA and the accumulated DIP, which correspondingly resulted in alternating high and low C/P ratios and DOC concentrations. Our results further suggest that cautions must be exercised in using APA as an indicator of P stress or limitation in ecological interpretations4–6, 37. High APA values may not necessarily reflect a P-limiting or -stressed condition to the microbial community2.
We used 16S rDNA and rRNA to determine, respectively, the compositions of bulk and active assemblages of the resident microbial communities in the CD water column. Microbial communities consisted of taxa from 21 dominant classes in 16 phyla, among which 19 classes were bacteria and 2 archaea (Marine Group I and Thermoplasmata) (Fig. S2). Notably, microbial communities in the CD water column were dominated by lineages that had been shown in actively producing AP and utilizing both DIP and DOP, including the oligotrophs SAR11 clade, mixotrophs cyanobacteria like Procholorcoccus, and oppotunitrophs like Gammaproteobacteria (mainly Alteromonas) (Fig. 2). Our results showed that the SAR11 clade accounted for 48.5% (32.0%) of bulk (active) bacterial OTUs in surface water, 31.3% (12.4%) and 68.2% (8.9%) at 7,000 and 10,918 m, respectively. The SAR11 clade is the dominant chemoheterotrophic bacteria and efficient competitors for nutrient resources in the global oceans38,39. Members of the SAR11 clade are frugal and typically have low P requirements40. The capacity of taking up DIP and DOP has been demonstrated experimentally22,40,41. Given the exceptionally high DIP concentrations in the CD deep waters, it is reasonable to posit that SAR11 used DOP for producing organic carbon (OC), rather than for DIP.
Strikingly, active cyanobacteria were detected in the whole water column, with high abundances in surface water and in the hadal waters (7,000 and 10,918 m) (Fig. 2). In particularly, Prochlorococcus comprised 53.4% of the bacterial OTUs in surface water, 30.6% and 9.1% in 7,000 and 10,918 m deep waters, respectively. Many cyanobacteria are mixotrophs and good competitors of heterotrophic bacteria for DOP 42,43. Previous studies have shown that cyanobacteria were able to mineralize DOP by producing alkaline phosphatases44. Another study revealed that Prochlorococcus and Pelagibacterales bacteria together accounted for ca. 90% of P uptake in the North Atlantic Ocean22. Taken together, our results suggest that these microorganisms likely utilized DOP as a source for producing organic moieties for sufficing the carbon requirements, especially in the hadal waters where organic matter may be more refractory and less abundant.
Microbial dichotomy in the ocean has been well studied. Our results showed that particle-associated microbial AP probably have played a more important role than the free-living AP in P cycling in the trench. This is evidenced by the observed higher particle-associated APA (Vmax-PA) and higher Vmax/Km ratios than the corresponding values for free-living AP (Table 1; Table S1). The entire water column exhibited a high proportion of cell-associated APA (averaging 60.7% of the total APA) and higher APAPA/Km−PA (0.47) than APAFL/Km−FL (0.13) (Table S1). This conclusion is also supported by the calculated VDOP. VDOP for APAPA is on average 1.7 times greater than APAFL (Table S1), indicating a greater role of APAPA in producing DIP from DOP. This finding is also in accordance with those reported by Malfatti et al. (2014)45 and Davis and Mahaffey (2017)46, in contrast to findings of Thomson et al. (2019)32 and others.
We developed a priori model using path analysis modeling, to test our hypothesized relationships among APA, Km, DOC, DIP, and active microbial communities in the water column. The model showed a very good fit with our data and explained 90% of DOC variability in the CD waters (Fig. 3). Both Km and Vmax contributed directly to the increases in DOC, and a significant positive correlation (R2 = 0.976) was observed between Km and DOC (data not shown). These results revealed that microbial alkaline phosphatase-induced enzymatic decomposition of DOP provided microorganisms with both P and OC in the trench waters.
In summary, our results represent the first dataset of microbial alkaline phosphatase activities, and provide a direct observational basis for understanding of microbial strategies for resource utilization and linking microbial-mediated P cycle and C cycle in the deepest ocean. Our findings show that carbon cycle in the deep ocean is intricately tied to cycles of nutrients, particularly that of phosphorus.