Research Article

Niche Partitioning Among Prochlorococcus Ecotypes Along Ocean-Scale Environmental Gradients

See allHide authors and affiliations

Science  24 Mar 2006:
Vol. 311, Issue 5768, pp. 1737-1740
DOI: 10.1126/science.1118052

Abstract

Prochlorococcus is the numerically dominant phytoplankter in the oligotrophic oceans, accounting for up to half of the photosynthetic biomass and production in some regions. Here, we describe how the abundance of six known ecotypes, which have small subunit ribosomal RNA sequences that differ by less than 3%, changed along local and basin-wide environmental gradients in the Atlantic Ocean. Temperature was significantly correlated with shifts in ecotype abundance, and laboratory experiments confirmed different temperature optima and tolerance ranges for cultured strains. Light, nutrients, and competitor abundances also appeared to play a role in shaping different distributions.

The marine cyanobacterium genus Prochlorococcus (1) is the numerically dominant phytoplankter in tropical and subtropical oceans (25). It is distributed throughout the illuminated surface waters from about 40°N to 40°S, can be found as deep as 200 m, and sometimes reaches abundances greater than 105 cells per ml (2). Prochlorococcus can account for 21 to 43% of the photosynthetic biomass in oligotrophic oceans and 13 to 48% of the net primary production (46). As such, its abundance and dynamics have a substantial impact on open ocean ecosystems and global biochemical cycles.

In the tropical and subtropical Atlantic Ocean, Prochlorococcus maintains its numerical dominance throughout most of the year, despite 10-fold changes in depth-integrated abundance (3, 6). It reaches its maximum (6, 7) when surface waters are highly stratified and devoid of major nutrients (8) and its minimum during deep winter mixing when blooms of Synechococcus and larger eukaryotic phytoplankton occur (3, 6, 7). Prochlorococcus abundance falls off sharply at higher latitudes, suggesting that temperature, or some related variable, is an important factor in shaping its global distributions (2).

Phylogenetic trees constructed with several marker genes, including small subunit (SSU) ribosomal RNA (rRNA), the rRNA internal transcribed spacer (ITS) regions, psbB, petB/D, and rpoC1, as well as whole-genome analyses, show that flow-cytometrically defined (9) Prochlorococcus is a genetically diverse group with several distinct clades, which have SSU rRNA gene sequences that differ by less than 3% (1014). Cultured strains belonging to different clades have different pigmentation, maximum growth rates, metal tolerances, nutrient utilization, and photophysiological characteristics, and thus these clades have been called ecotypes (1517). The ecotypes have recently been named after type strains with the prefix “e” (18, 19). Cells belonging to different ecotypes are distributed differently along environmental gradients. For example, high light–adapted ecotypes [sensu (12)] such as eMIT9312 and eMED4 are more abundant near the surface than low light–adapted ecotypes such as eSS120, eMIT9313, eNATL2A, and eMIT9211 (1820), which are typically most abundant in the deeper waters.

To better understand the drivers of ecotypic differentiation between Prochlorococcus clades, we measured, by using quantitative polymerase chain reaction (qPCR), the abundance of cells belonging to the six known ecotypes (fig. S1) as a function of depth along a meridional transect in the Atlantic Ocean (AMT) (21) (Fig. 1A). We then analyzed the distribution and abundance of ecotypes in the context of relevant environmental variables and the physiological properties of type strains.

Fig. 1.

(A) False-color image of sea surface temperature during the Atlantic Ocean meridional transect cruise (AMT13) from September to October 2003, overlaid with cruise track and sampled stations (boxes). Reds are warm temperatures; blues are cold temperatures. (Right) In situ measured surface red fluorescence (relative) (Rel. Fls), indicative of chlorophyll concentrations and temperature. The cruise track goes through the coastal Canary Current upwelling feature around 20°N. (B) Depth profiles of the abundance of four dominant Prochlorococcus ecotypes (red, eMIT9313; blue, eNATL2A; green, eMED4; and yellow, eMIT9312) and total Prochlorococcus cell concentrations (solid black) from three representative stations [stars in (A)]. The 1°N station is typical of equatorial region stations, 25°N is typical of stratified subtropical stations, and 48°N is typical of high-latitude (not polar) stations. Note that the abundance scale is logarithmic. Error bars indicate one standard deviation of the mean. (C) (Top) Depth-integrated (0 to 200 m) ecotype abundance and total Prochlorococcus abundance over the transect. (Bottom) Meridional distribution of the ratio of the summed ecotype abundances measured by qPCR and total Prochlorococcus abundance determined by flow cytometry. (D) Cross sections along the transect (upper 200 m) of nitrate concentration (μmol kg–1); total Prochlorococcus (log cells per ml); dominant ecotype clade abundances (log cells per ml) for eMIT9312, eMED4, eNATL2A, and eMIT9313; and Synechococcus (log cells per ml). Temperature contour lines are shown in white.

The AMT transect covered a broad range of biogeochemical provinces (22), with surface temperatures ranging from 5° C to 29°C (Fig. 1A). The availability of major phytoplankton nutrients, including nitrate, nitrite, ammonium, phosphate, and silicate, also varied, reaching maxima in the south, near the coastal Canary Current upwelling system, and around the equatorial upwelling region (Fig. 1D and fig. S2). For all locations, nitrate, phosphate, and silicate concentrations were correlated (r2 = 0.9), following near Redfield ratios (15 N/1 P/1 to 10 Si), whereas nitrite and ammonium were localized to specific regions and depths and did not follow strict relationships with other nutrients (figs. S2 and S3). Consistent with these nutrient patterns, surface chlorophyll fluorescence, an indicator of autotrophic biomass, was elevated at high latitudes and in the Canary Current upwelling system but remained low in the severely nutrient-limited subtropical gyres (Fig. 1A).

Ecotype distributions. The depth distributions of Prochlorococcus ecotypes at three stations representing different biogeochemical provinces revealed distinctly different patterns (Fig. 1B). Although total Prochlorococcus abundance (which was measured independently by using flow cytometry) was maximal at the surface and decreased with depth at each station, the ecotypes comprising this total had different distributions. In well-stratified waters (1°N and 25°N in Fig. 1B), ecotype depth distributions had a stacked nature typical of these environmental conditions (18, 19). The high light–adapted clades eMIT9312 or eMED4 dominated near the surface (∼0 to 50 m), and low light–adapted eNATL2A increased in abundance at middepths (∼75 m), followed by eMIT9313, another low-light clade. Closer examination of the patterns at these two stations, however, revealed subtle differences that suggest that light adaptation was not the only variable shaping distributions. At 1°N, for example, eNATL2A and eMED4, low and high light–adapted clades, respectively, were roughly equal in abundance at the surface, whereas at 25°N eNATL2A was orders of magnitude less abundant than eMED4 near the surface. At high latitudes such as 48°N (Fig. 1B), one of the high light–adapted clades, eMED4, outnumbered the other, eMIT9312, by orders of magnitude in surface waters. Furthermore, the low light–adapted clade eMIT9313, which displayed deep subsurface maxima (∼100 m) at low and midlatitudes, was so reduced in numbers at high latitudes that it was near the limit of detection. As has been previously reported (18, 19), two of the clades, eMIT9211 and eSS120, were present at low concentrations (<50 cells per ml) in all samples and thus were not included in this part of our analysis (fig. S4).

Similar shifts in ecotype dominance were also revealed by ecotype depth-integrated abundances along the entire transect (Fig. 1C), despite relatively constant (∼1 × 107 to 2 × 107 cells per mm2) total Prochlorococcus concentrations. Overall, high light–adapted eMIT9312 was the dominant ecotype, exceeding the abundance of other ecotypes by two orders of magnitude between 15°N and 15°S. Low light–adapted ecotypes eNATL2A and eMIT9313 had roughly equal abundance in this part of the transect, and high light–adapted eMED4 was the least abundant. By contrast, in cooler, well-mixed, nutrient-rich, high-latitude waters where eMED4 achieved numerical dominance, eMIT9313 disappeared, and eNATL2A achieved the second highest abundance. eMIT9312, the most abundant clade in the tropical and subtropical oceans, was below the detection limit at latitudes above 42°S. Our qPCR primers, designed from sequences of cultured isolates, captured on average ∼75% (range from 35 to 118%) of the total Prochlorococcus population counted by using flow cytometry (Fig. 1C), indicating that there are likely additional members not quantified in this study. However, there was no pattern to the residuals, suggesting that our observations were not greatly affected by culture bias (23).

Environmental factors and ecotype abundance. By using nitrate concentration to represent a suite of covarying nutrients (figs. S2 and S3), depth as related to light intensity, and plotting temperature contours on a slice through the upper 200 m along the transect, we visualized how ecotype abundances map on to environmental gradients (Fig. 1D). As expected, total Prochlorococcus cell concentrations were typically highest at or near the surface and decreased with depth. They were also highest in warm waters, which here are defined as waters above the 17°C isotherm, except below 150 m, where populations were likely limited by light availability. Lastly, total Prochlorococcus abundances were highest where nutrients were lowest, consistent with findings in other tropical and subtropical oligotrophic regions (4, 6, 7, 24).

These patterns of total Prochlorococcus abundance emerge from the distributions of ecotypes, which display more complex behavior when viewed individually. For example, the relative abundances of the high light–adapted ecotypes (eMIT9312 and eMED4) depended on latitude and closely followed nutrient and temperature contours (Fig. 1D). eMIT9312 dominated at low and midlatitudes but was progressively replaced by eMED4 at latitudes above 30°. Thus, these two ecotypes, which differ in their SSU rRNA sequences by less than 1% (12), had overlapping but distinct niches in the Atlantic Ocean and comprised the majority (93%) of the Prochlorococcus cells over the transect.

Ecotypes eMIT9313 and eNATL2A had more restricted distributions. eNATL2A was most prevalent along the trailing edge of the eMIT9312 abundance contour and was most abundant in waters between the 15°C and 23°C isotherms. This distribution started at the surface waters of 45°N, meandered in the water column in mid- and low latitudes, and reached higher abundances again at the surface waters near 30°S. eNATL2A only reached high concentrations in deeper waters, where eMIT9312 and eMED4 numbers were reduced. Similarly, eMIT9313's distribution was more patchy and localized to specific latitudes or depths, closely following the vertical gradients of eMIT9312 and eMED4, and was maximal where total Prochlorococcus abundance decreased the most sharply with depth. Thus, although eNATL2A and eMIT9313 were minor contributors to overall Prochlorococcus numbers in the Atlantic Ocean, they dominated at certain depths and latitudes.

Regression analysis. The geographic patterns of Prochlorococcus and its ecotypes suggest causal relationships with environmental variables (Fig. 1D). We used stepwise regression analyses to investigate relationships with temperature, light, nitrate, nitrite, ammonia, phosphate, and Synechococcus (a potential competitor) (table S1). As has been observed previously (25), a significant fraction (26%, P < 0.01) of the variability in the total Prochlorococcus population could be explained by temperature, which was largely driven by significant correlations between the numerically dominant ecotype, eMIT9312, and temperature (56%, P < 0.01). By using laboratory isolates belonging to the two dominant (and high light–adapted) ecotypes, we found that, although both had similar temperature optima for growth, members of eMIT9312 grew faster at this temperature (25°C) and could tolerate warmer temperatures than eMED4. The latter could grow at temperatures below 15°C, however, whereas eMIT9312 could not (Fig. 2 and fig. S5), thus confirming the importance of temperature in defining ecotype niche space.

Fig. 2.

Maximum growth rate as a function of temperature derived from the analysis of cultured strains belonging to the eMIT9312 clade (yellow) and the eMED4 clade (green), the two dominant near-surface ecotypes. (Specific strain data are shown in fig. S5.)

Prochlorococcus ecotype abundances are related to other variables as well. For example, eMED4 was correlated (9%, P < 0.01) with light, which is consistent with its adaptations to high light (16). Nutrient relationships were complex (often being anticorrelated with eco-types) (table S1) and were generally not consistent with the ability of strains to use nutrient sources (17, 26), suggesting that colinear variables not measured here were responsible for these relationships. Integrated abundances of Synechococcus (Fig. 1D), which unlike Prochlorococcus can use nitrate (17), were also inversely related to Prochlorococcus (fig. S6). This relationship has been found before for seasonal cycles in the Atlantic Ocean (3, 6) and implies that at high nutrient levels Synechococcus and larger phytoplankton may exclude Prochlorococcus. But this relationship with Synechococcus was complex: eMIT9313 was anticorrelated (22% of the variability), eMED4 (18%) and eMIT9312 (10%) were correlated, and the abundance of eNATL2A was uncorrelated.

The stepwise regressions, which had varying degrees of success in explaining the overall abundance patterns, also demonstrated that there were additional factors that were important in determining Prochlorococcus ecotype abundance. This complexity is likely governed by differential susceptibility to viruses (27) and protozoan grazers (28) as well as other factors, such as metal tolerances (15).

The relationships we have established here represent only a broad-brush assessment, to be refined as we learn more about Prochlorococcus microdiversity. By design, multiple ribotypes (based on ITS sequences) are represented in the ecotype populations that are enumerated by our qPCR primers (18), and it is likely that each of these is a physiological and genomic variant. This prompts the question: If one designed primers to enumerate ribotypes within the ecotype clusters, what would their distributions look like along these gradients? Would they be more similar to each other than to those belonging to other clades? Furthermore, although the four dominant ecotypes account for the majority of the total Prochlorococcus population, our analyses (Fig. 1), as well as evidence from clone libraries (18) and Sargasso Sea shotgun sequencing (29), reveal the existence of additional genotypes that do not group with any of the six defined genetic clades and thus would not be captured by our approach. These additional clades likely account for the difference between the total Prochlorococcus population measured by flow cytometry and the summed clades measured by qPCR (Fig. 1C), particularly in the deeper waters (Fig. 1B).

Supporting Online Material

www.sciencemag.org/cgi/content/full/311/5768/1737/DC1

Materials and Methods

Figs. S1 to S6

References and Notes

View Abstract

Navigate This Article