Main

We collected colonies of Trichodesmium spp. using trace-metal-clean methods in the subtropical western Atlantic Ocean in April 2003 (Supplementary Table 1), and analysed them for surface-adsorbed and intracellular P using the oxalate surface wash method8. We also measured total and surface-adsorbed C, N, Fe, Mn and Mo. Additional laboratory experiments determined the removal efficiency of the surface-adsorbed P, C and N by the oxalate reagent, and examined phosphate-uptake kinetics and C:P ratios in adsorbed and internal pools of Trichodesmium cultures grown under P-limited or P-replete conditions. The removal efficiency of surface-adsorbed 33PO4 (labelled with a single 5-min pulse) from the surface of Trichodesmium cultures by the oxalate reagent was ∼90% (Supplementary Fig. 1). Our laboratory results also indicated that none of the C and N in phytoplankton was surface-adsorbed, because their levels in the total and intracellular pools were essentially the same (Supplementary Fig. 2). Our previous results have shown that application of the oxalate reagent does not produce any cell breakage or lysis, or affect the physiological status of the phytoplankton in any way8.

The molar C:N:surface-adsorbed P ratios (median and mean) (80–105 C:15–18 N:1 P) of the field-collected Trichodesmium colonies were not significantly different from the elemental ratios calculated using the total (= surface-adsorbed + intracellular) P pool (71–80 C:13–14 N:1 P; Fig. 1 and Supplementary Table 1).These results suggest that the Redfield ratio based on total P measurements in plankton largely reflects the contribution of the surface-adsorbed P pool. In contrast, internal nutrient ratios (316–434 C: 59–83 N:1 intracellular P) in the same colonies were four to five times higher than in the surface-adsorbed and total fractions, and do not conform with the Redfield model (Fig. 1).

Figure 1: Box plots of elemental composition (C, N, P) in the total, surface-adsorbed and intracellular pools in field-collected Trichodesmium colonies from the western subtropical Atlantic Ocean.
figure 1

a, C:P ratios. b, N:P ratios. The line within the box is the median, and the boundary of the boxes indicates the 25th and 75th percentiles. Error bars to the left and right of the box indicate the 10th and 90th percentiles. Filled circles show outlying points. A t-test showed that elemental compositions in the surface-adsorbed and total pools were not significantly different (C:P; t = 1.64, P > 0.05, N:P; t = 1.34, P > 0.05).

Full size image

Our study focused on P partitioning in the biogeochemically critical diazotroph Trichodesmium, because Redfield hypothesized that the constant elemental ratios in the ocean are a reflection of the chemical composition of N-fixing organisms1. Our data suggest that that is indeed the case in the surface-adsorbed pool. Our results also showed that surface scavenging of P is not unique to Trichodesmium. High levels of surface-adsorbed P were also measured in Pseudo-nitzschia sp. and Hemiaulus hauckii blooms from two locations in the western Atlantic Ocean, as well as in the diatom Thalassiosira weissflogii (Fig. 2), and four other algal taxa in the laboratory (Supplementary Fig. 2).

Figure 2: Cellular partitioning of phosphorus in phytoplankton.
figure 2

a, Fraction of intracellular and surface-adsorbed P measured in field-collected samples from blooms of P. nitzschia sp. and H. hauckii. Samples were collected on 23 April 2003 at 06° 53.06′ N, 52° 01.95′ W and on 26 April 2003 at 09° 48.55′ N, 54° 04.95′ W. As observed for Trichodesmium, more than 60% of the total P measured in those blooms was in the surface-adsorbed pool. b, Fraction of intracellular and surface-adsorbed P measured in the centric diatom Thalassiosira weissflogii under different growth (senescent versus actively growing) and substrate (DIP versus DOP) conditions in the laboratory. DIP, dissolved inorganic phosphate; DOP, dissolved organic phosphate (in this case glycerophosphate). The amount of surface-adsorbed P (90%) in the senescent diatoms was considerably higher than the amount measured in the exponential growth phase (30%). These results suggest that a fraction of the surface-adsorbed pool is being internalized and used for growth. In contrast to the substantial removal of surface-adsorbed P observed in the culture grown on DIP (senescent and actively growing), none of the cellular P in cultures grown on DOP was removed by the reagent. This is consistent with the tendency of oxyanions such as phosphate to form, by ligand exchange, surface complexes with hydrous oxides in biogenic surfaces and minerals. This process allows removal of surface-adsorbed P by the oxalate wash because it is effective at removing hydrous oxides.

Full size image

These results indicate that P, like scavenged cations such as iron8,9, is present in phytoplankton in both surface-adsorbed and intracellular pools. The ratios of these pools in phytoplankton are variable, and do not always reflect Redfield stoichiometry. Although previous studies have also reported large deviations in Redfield ratios10,11,12, those studies have not addressed the impact of different cellular pools of P.

In P-replete and P-limited laboratory Trichodesmium cultures, C:P ratios showed the same trends (Fig. 3a), although the magnitude of differences between pools was somewhat lower than in field samples. The total cellular ratios of P-replete cultures were at near-Redfield values of 95, but intracellular ratios of these same cultures were 1.7 times higher, at 161. The P-limited cultures had higher total cellular C:P ratios of 207, but intracellular ratios were about 1.4 times greater at nearly 300—very close to the increased intracellular values in the field collections (Fig. 1). These high intracellular C:P ratios in both natural and cultured Trichodesmium were consistent with calculated cellular C:P values (range 190–390 mol mol-1), based on RNA as the major intracellular P compound under P-limited conditions2,13.

Figure 3: Effects of P-limited and P-replete growth on intracellular (oxalate-washed) and total cellular (seawater-washed) P content and P uptake rates in laboratory cultures of Trichodesmium IMS101.
figure 3

Values are the means and error bars are the standard deviations of triplicate cultures. a, Molar C:P ratios in both cellular pools in P-limited and P-replete cultures. The dashed line represents the Redfield ratio. b, Phosphate uptake rates into total cellular and intracellular pools as a function of phosphate concentration, in the same P-replete cultures shown in a. Shown are half-saturation constants (Ks, in µM) and maximum uptake rates (Vmax, in mol P cell-1 h-1) for both cellular pools. c, As in b, but for P-limited cultures.

Full size image

The existence of two distinct cellular P pools could have significant implications for studies using the Redfield model to define phytoplankton nutrient limitation. For example, whereas total and surface-adsorbed elemental ratios in the Trichodesmium colonies suggest P sufficiency (Redfield ratio; N:P range = 10–20), the intracellular pool was depleted in P (higher than Redfield ratio: N:P range = 50–300; Fig. 4a). The current Redfield model based on total P levels may thus not adequately reflect the physiological status of some phytoplankton species. Consistent with that hypothesis, N fixation by Trichodesmium in the western subtropical Atlantic Ocean during our study was suppressed at two-thirds of our stations, despite the fact that total elemental ratios were close to the Redfield model (Figs 1 and 4a). Nutrient ratios in the intracellular pool suggested that N fixation was low because those Trichodesmium colonies were P-deficient (Fig. 4a), as previously reported for the same region of the Atlantic Ocean14,15.

Figure 4: Relationship between different phosphorus pools and nitrogen fixation (a) and surface-adsorbed manganese (b).
figure 4

a, Relationship of N2 fixation (reported as nmol of C2H2 per µg of chlorophyll a per hour) with total (open circles) and intracellular (filled circles) N:P ratios measured in the field-collected Trichodesmium colonies. Solid line at N:P = 50 is the critical ratio that marks the transition of P-limitation in phytoplankton2. b, N:surface-adsorbed P ratios versus surface-adsorbed Mn measured in field-collected Trichodesmium colonies. Regression equation (solid line): N:surface-adsorbed P = -12.1 × [surface-adsorbed Mn] + 44.5; r2 = 0.81. Dashed lines indicate 95% confidence limits.

Full size image

Whereas the particle reactivity of P in the environment has been studied by geochemists16, no previous study has addressed the impact of that mechanism on Redfield stoichiometry and phytoplankton physiology. The existence of the surface-adsorbed and intracellular P pools in phytoplankton suggests that P uptake could be a two-step kinetic process; that is, adsorption to the cell surface, followed by internalization, similar to that reported for iron17. The different amounts of surface-adsorbed P in senescent (∼ 90%) and in exponentially growing (30%) cultures suggest that phytoplankton have developed mechanisms to access at least some of the surface-adsorbed pool (Fig. 2). However, further studies are needed to evaluate whether surface-adsorbed P can be internalized fast enough to make a significant difference in the nutritional status of different types of phytoplankton.

Our results suggest that kinetic constants for phytoplankton P uptake need to be re-evaluated using intracellular levels of P instead of total concentrations. Removal of the surface-adsorbed P pool with the oxalate reagent results in substantially lower half-saturation constants for uptake (Ks)—88% and 51% of those measured in total cellular pools for P-replete and P-limited cultures, respectively (Fig. 3b, c). Maximum uptake rates (Vmax) are even more strongly affected, and oxalate-washed values are only 51% (P-replete) and 42% (P-limited) as high as in total cellular pools. Previous physiological P-uptake and growth assessments and N:P requirements of phytoplankton that considered only the total cellular pool18,19 may therefore need re-interpretation in relation to intracellular P pools.

Surface scavenging of P by Trichodesmium colonies appears to be mediated by the amount of surface-adsorbed manganese (Supplementary Fig. 3). This is consistent with the covariance between total P and Mn concentrations previously reported in biogenic particles from the North Atlantic20. The surface scavenging of phosphate is explained by the tendency of oxyanionic elements to form, by ligand exchange, surface complexes with hydrous oxides on biogenic surfaces and minerals21. This mechanism of P adsorption in Trichodesmium colonies is further substantiated by the positive relationship between surface-adsorbed Mn and other oxyanions such as molybdate (Supplementary Fig. 3). This adsorption trend suggests that the surface-adsorbed enrichment of phosphate is caused by inorganic scavenging processes, rather than by active biogenic uptake. This is consistent with our laboratory results showing that the oxalate reagent did not remove any of the cellular P from cultures grown on a dissolved organic P source, in contrast to phosphate-grown cultures (Fig. 2b).

Our results showed that the amount of surface-adsorbed manganese on the Trichodesmium colonies seems to be the main factor determining the adsorption capacity for P (Supplementary Fig. 3). Therefore, N:P ratios in the colonies were inversely related to levels of surface-adsorbed manganese (Fig. 4b). This relationship suggests that deviations in Redfield stoichiometry (N:P range = 10–40; Fig.\ 4b) observed in the Atlantic Ocean could be partially explained by the adsorption of P onto cell-surface-bound metal hydroxides and oxides, rather than being attributed solely to N fixation22 or N pollution23.

Significant decoupling of elemental ratios between the surface-adsorbed and intracellular pools could have far-reaching implications for marine biogeochemistry. For example, current Redfield ratios normalized to total cellular P may not adequately reflect the nutritional status of phytoplankton, as they appear to be largely determined by the adsorption of this nutrient onto reactive chemical surface sites of plankton. Furthermore, models that use Redfield stoichiometry to couple changes in nutrient inventories and marine biogeochemical processes24,25, should take into consideration the existence of the two different P pools in phytoplankton.

Methods

Trichodesmium colonies were collected using a Zodiac inflatable deployed from the RV Seward Johnson from 18 April to 20 May 2003 in the subtropical North Atlantic Ocean (Supplementary Table 1). Colonies were collected at a depth of ∼5 m by towing an acid-washed 102-µm plankton net. In a class-100 bench, ∼200 colonies per station were hand-picked from the acid-washed cod-end using a plastic inoculating loop, deposited into 3-ml acid-washed Teflon vials and digested in a combination of quartz-distilled HCl, HNO3 and HF. Fe, Mn and Mo content in the Trichodesmium colonies were then measured by inductively coupled plasma mass spectrometry (ICP-MS). P concentrations in the digested colonies were determined by spectrophotometric techniques developed for small volumes published elsewhere14. Particulate organic C and N levels in the colonies were determined using a Carlo Erba NA1500 NCS system. Intracellular levels of P, Fe, Mn and Mo were determined as described above after washing ∼100 colonies with the oxalate reagent8. The redox potential of oxalic acid (EH0 = -631 mV) indicates that this reagent reduces most of the iron (EH0 = -295 mV) and manganese (EH0 = -50 mV) hydroxides bound to phytoplankton. The surface-adsorbed pool is calculated by the difference between the total and intracellular determinations. Nitrogen fixation was determined on isolated colonies using the protocols described in ref. 14.

Removal efficiency of adsorbed phosphate by the oxalate reagent was tested on laboratory cultures of Trichodesmium IMS101 and the centric diatom Thalassiosira weissflogii following 5-min-pulse labelling of extracellular pools using 33PO43- (Supplementary Fig. 1), as previously described for iron (refs 8, 9). For uptake kinetics determinations, cultures of the Atlantic isolate Trichodesmium IMS101 were grown in P-replete (20 µM) or P-limited (0.2 µM) seawater medium, and harvested in early exponential phase to generate short-term (1 h) Michaelis–Menten PO43- uptake curves using the tracer 33PO43- (ref. 18). Elemental ratios in the cultures were determined using the same protocols described above for the field-collected Trichodesmium colonies.