Modeling selective pressures on phytoplankton in the global ocean

Abstract

Our view of marine microbes is transforming, as culture-independent methods facilitate rapid characterization of microbial diversity. It is difficult to assimilate this information into our understanding of marine microbe ecology and evolution, because their distributions, traits, and genomes are shaped by forces that are complex and dynamic. Here we incorporate diverse forces--physical, biogeochemical, ecological, and mutational--into a global ocean model to study selective pressures on a simple trait in a widely distributed lineage of picophytoplankton: the nitrogen use abilities of Synechococcus and Prochlorococcus cyanobacteria. Some Prochlorococcus ecotypes have lost the ability to use nitrate, whereas their close relatives, marine Synechococcus, typically retain it. We impose mutations for the loss of nitrogen use abilities in modeled picophytoplankton, and ask: in which parts of the ocean are mutants most disadvantaged by losing the ability to use nitrate, and in which parts are they least disadvantaged? Our model predicts that this selective disadvantage is smallest for picophytoplankton that live in tropical regions where Prochlorococcus are abundant in the real ocean. Conversely, the selective disadvantage of losing the ability to use nitrate is larger for modeled picophytoplankton that live at higher latitudes, where Synechococcus are abundant. In regions where we expect Prochlorococcus and Synechococcus populations to cycle seasonally in the real ocean, we find that model ecotypes with seasonal population dynamics similar to Prochlorococcus are less disadvantaged by losing the ability to use nitrate than model ecotypes with seasonal population dynamics similar to Synechococcus. The model predictions for the selective advantage associated with nitrate use are broadly consistent with the distribution of this ability among marine picocyanobacteria, and at finer scales, can provide insights into interactions between temporally varying ocean processes and selective pressures that may be difficult or impossible to study by other means. More generally, and perhaps more importantly, this study introduces an approach for testing hypotheses about the processes that underlie genetic variation among marine microbes, embedded in the dynamic physical, chemical, and biological forces that generate and shape this diversity.

Figure 1. Diagrams illustrating our modeling approach.

(A) Marine Synechococcus and Prochlorococcus isolates vary in their abilities to grow on NH₄, NO₂ and NO₃ , , as indicated by the boxes next to the strain names and key below , . These nitrogen use abilities are reflected in the genomes of the isolates, including the presence of genes needed for assimilating NH₄ (glnA), NO₂ (nirA) and NO₃ (narB) (solid dots indicate the presence of a particular gene –[15]). (B) All picophytoplankton in the model initially can use NO₃, NO₂ and NH₄. After three years, three different types of mutations occur, each at the same rate, Λ (see Text S1). The different types of mutants produced are (i) mutants that cannot use NO₃; (ii) mutants that cannot use NO₂ or NO₃; (iii) null mutants that retain the ability to use NO₃, NO₂ and NH₄.

Figure 2. The distribution of phytoplankton biomass and accumulation of NO₃/NO₂ loss mutants in modeled surface waters.

The global distribution of phytoplankton (phosphorus biomass [log₁₀(µM P)], 0 to 10 m, annual average) is plotted for (A) picophytoplankton and (B) large phytoplankton. The distribution of mutant picophytoplankton (phosphorus biomass [log₁₀(µM P)], 0 to 10 m, annual average) is plotted for (C) null mutants and (D) NO₃/NO₂ loss mutants. Data shown are for the fifth year of one integration, out of an ensemble of ten integrations that were initialized with different, randomly generated, phytoplankton communities. In panel (D), the cruise track of Atlantic Meridional Transect 13 (AMT 13) is indicated with a black line, and the location 35°N, 22°W is indicated with a cross.

Figure 3. Observations and model predictions along a latitudinal transect.

The distribution of Prochlorococcus and Synechococcus picocyanobacteria along AMT13, and model predictions for the accumulation of NO₃/NO₂ loss mutants in their model analogs. (A) The distribution of two Prochlorococcus ecotypes and Synechococcus along AMT13 [log₁₀(cells ml⁻¹)]. (B) The abundance of three model phytoplankton types along AMT13 that have similar distributions to real-world picocyanobacteria (converted to log₁₀(cells ml⁻¹) assuming 1 fg P cell⁻¹). (C) The abundance of NO₃/NO₂ loss mutants of these modeled ecotypes (converted to log₁₀(cells ml⁻¹) assuming 1 fg P cell⁻¹). (D, E) Indices of the disadvantage associated with losing the ability to use nitrate (D) and nitrite (E) for each ecotype. The disadvantage associated with losing the ability to use nitrate is calculated as log₁₀[ (null)/(NO₃ loss mutant)], and the disadvantage associated with losing the ability to use nitrite is calculated as log₁₀[(NO₃ loss mutant)/(NO₃/NO₂ loss mutant)]. These indices are plotted for each ecotype in locations where the parent is abundant (greater than 10⁻³ times its maximum value).

Figure 4. Seasonal patterns in the abundances of nutrients and phytoplankton.

Seasonal patterns in the availability of (A) ammonium and nitrate, and the abundance of (B) modeled ecotypes E2 and E3 (ecological analogs of Prochlorococcus eMED4 and Synechococcus, respectively), and their (C) null mutants and (D) NO₃/NO₂ loss mutants, at the location 35°N, 22°W, near the surface.