Technical Comments

Response to Comment on “Patterns of Diversity in Marine Phytoplankton”

See allHide authors and affiliations

Science  30 Jul 2010:
Vol. 329, Issue 5991, pp. 512
DOI: 10.1126/science.1190048

Abstract

Huisman argues that environments of intermediate variability promote coexistence of model phytoplankton, apparently contrasting our hypothesis that stability allows for greater diversity of equivalent competitors in the ocean. We argue that our original interpretations of the mechanisms governing model diversity patterns remain valid and that Huisman’s results are complementary to our hypotheses.

Species diversity in marine and terrestrial ecosystems often varies strongly in space. In a recent study (1), we reported that marine phytoplankton diversity decreased with increasing latitude in a global phytoplankton community model. Niche differentiation determined phytoplankton biogeography and enhanced overall diversity but could not by itself explain the large-scale spatial diversity patterns. We hypothesized that stable conditions (as in subtropical and tropical oceans) promoted coexistence of phytoplankton with similar competitive abilities, as measured by R* (the lowest environmental nutrient concentration at which the growth and mortality are in balance) (2). This neutral coexistence broke down under seasonally variable conditions (as experienced at higher latitudes), where the fastest-growing phytoplankton outcompeted all others, leading to lower diversity. This broad division between “gleaners” (low R*) and “opportunists” (high growth rate) was the result of physiological tradeoffs imposed on the model phytoplankton types, and this mechanism has strong underpinnings (35). Huisman’s (6) two phytoplankton types are, in essence, a “gleaner” and an “opportunist,” and this strategic division is apparent in his results. Huisman also argues that the two phytoplankton types can coexist in intermediate levels of environmental variability. This result for two particular phytoplankton, although correct, does not appear to be likely when a more diverse phytoplankton community is considered, as in Barton et al. (1). However, Huisman describes a scenario that highlights generalizations about stable and variable ocean environments and the types of phytoplankton that dominate these habitats.

The resource competition models of Barton et al. (1) and Huisman (6) define phytoplankton types by three traits: the maximum growth rate μi, the half-saturation nutrient concentration ki, and the mortality rate m, which is held constant (7). For the following discussion, we parameterize the two-dimensional trait space in terms of Ui (Ui=μi/m) and the resource requirements Ri* (Ri*=mki/μim). When we considered the maintenance of diversity in a system with many possible phytoplankton types (1), we were essentially asking this question: When can differing phytoplankton with a broad range of these traits coexist for many years? Insights can be gained by using a model with many more than two species [e.g., (8)]. Physiological tradeoffs constrain the possible phytoplankton to lie in distinct regions of trait space. We expect phytoplankton nutrient requirements to increase with U, and for small U (U1), R* should also increase [after Ri*=ki/(Ui1), for a given ki]. In the three-dimensional global model (1), we found that many phytoplankton types had similar low R* values over a range of U. Considering these factors, we represent here a physiologically possible trait space that also includes Huisman’s two phytoplankton types (Fig. 1, A and B).

When all phytoplankton types compete for a single resource, we can determine which (R*, U) type(s) win the competition for resources. We first consider a single phytoplankton type in trait space (Huisman’s phytoplankton type 1) and solve (numerically), with this being the only phytoplankton present, to find the temporal cycles of phytoplankton biomass P1 and nutrient N1. For a given amplitude (A) and period of nutrient delivery, we then determine the neutral growth curve, which is the set of trait space points (Ri*, Ui) such that an infinitesimal seed population of phytoplankton i would neither grow nor decay over a cycle. When the amplitude of variation is zero and N is steady, the neutral curve isRi*=N1, and we obtain the steady state solution of Tilman (2) (vertical blue line, Fig. 1A). When A is small, the neutral curves approach the average value of N for rapidly growing phytoplankton and curve to the left as Ui decreases (A = 0.2 curve, Fig. 1A). Phytoplankton to the left of the neutral curve grow, and they decay to the right. When we draw the neutral curve for phytoplankton type 1 at intermediate variability (A = 0.6), we see that there are more competitive phytoplankton types in trait space (to the left), and the winning type (Pi) would be the one with a neutral curve tangent to the trait space boundary. If the trait space boundary remains close to the tangent neutral curve over a range of U, we expect that these phytoplankton types will be excluded very slowly and in the more complex three-dimensional model can persist with only small amounts of input by transport and mixing. In other words, the level of environmental variability selects for a characteristic subset of phytoplankton on the trait space boundary.

Fig. 1

(A) Physiologically possible trait space (gray shading) that includes points P1 (Huisman’s phytoplankton type 1; Embedded Image mmol N m−3; U1 = 2.0) and P2 (Huisman’s phytoplankton type 2; Embedded Image mmol N m−3; U2 = 1.6). The thick black line shows the left edge of the trait space. The blue lines show the neutral curves for P1 in stable (A = 0) and variable (A = 0.2, 0.6) conditions. Under intermediate variability (A = 0.6), point Pi will outcompete P1. (B) Same trait space as in (A). The red line is the tangent neutral curve for P2 in stable conditions (A = 0). The blue line is the tangent neutral curve for P1 in highly variable conditions (A = 0.95). The dashed black line is the tangent neutral curve intersecting the trait space boundary at Pi for an intermediate variability (A = 0.60). (C) Phytoplankton biomass per unit U as a function of amplitude (A) and U for the phytoplankton on the trait space boundary in (B), discretized with dU = 0.017. A broader spread in biomass, increasing to the left in the graph, implies greater diversity. (D) Huisman’s case with only two phytoplankton types. Neutral curves for P1 (blue line) and P2 (red line) for intermediate levels of variability (A = 0.64) are drawn. The dashed black line shows the neutral curve that intersects both points.

If we now consider a range of variabilities (amplitudes from A = 0 to A = 1), we can evaluate which types of phytoplankton along the trait space boundary (points in the shaded interior will be excluded by species directly to the left) will dominate for different environmental variabilities (Fig. 1B). Gleaners (i.e., Huisman’s phytoplankton type 2) are successful in steady conditions (A = 0) and opportunists (i.e., Huisman’s phytoplankton type 1) in variable conditions (A = 0.95). The neutral curve for the gleaner lies close to the trait space boundary for a range of organisms, rather than just a point as for more variable conditions, suggesting that extended coexistence is possible among these low R* phytoplankton types. We ran the resource competition model (7) for the phytoplankton types on the trait space boundary for a range of amplitudes and found that a diverse set of gleaners in stable conditions is replaced by a less diverse set of opportunists in more variable conditions (Fig. 1C).

Considering this framework, we argue that Huisman’s experiment with two phytoplankton is a special case within this broader possible range of organisms. In Fig. 1D, the neutral curves for Huisman’s two phytoplankton at intermediate variability (A = 0.64) show that the two phytoplankton can coexist under these environmental conditions, as reported by Huisman (6). If the trait space boundary (Fig. 1, A and B) were to curve in the same manner as the neutral curve for intermediate disturbance levels (dashed black line, Fig. 1D), high model diversity would be possible with that level of variability. However, mechanistic models of tradeoffs in traits suggest a trait space shape similar to ours (Fig. 1, A and B) (9). Rather, we believe Huisman’s example describes where different survival strategies, gleaners or opportunists, will be competitive, and even coexist, in the ocean. A previous ocean modeling study by Dutkiewicz et al. (10) conducted experiments similar to Huisman’s and found that gleaners dominate the stable tropics, opportunists dominate the variable high latitudes, and both strategies were viable in a narrow region in between. The zonal mean diversity of gleaners and opportunists for the 10 ensemble members of the global phytoplankton community model used in Barton et al. (1) shows a similar pattern, with a diverse group of gleaners in the lower latitudes, relatively few opportunists in the higher latitudes, and a narrow band in mid-latitudes where both strategies coexist (40°N and S) (Fig. 2). Thus, we agree that the intermediate disturbance hypothesis may have relevance to these models, but primarily in determining where different strategies co-occur.

Fig. 2

Zonal mean total (black), gleaner (red), and opportunist (blue) diversity for the 10 ensemble members of the global phytoplankton community model used in (1). The diversity is defined here as those species with greater than 0.1% of the total phytoplankton biomass.

We conclude that Huisman’s example (6) does not invalidate our hypothesis concerning the broader diversity pattern: that of a diverse, gleaner-dominated community existing in stable ocean environments and a less diverse, opportunist-dominated community existing in more variable, higher latitudes. We echo the call for additional experimental, theoretical, and field observational work toward improving the understanding of patterns of marine diversity and biogeography.

Supporting Online Material

www.sciencemag.org/cgi/content/full/329/5991/512-d/DC1

SOM Text

References

References and Notes

  1. Supporting material is available on Science Online.
View Abstract

Subjects

Navigate This Article