全球变化知识资源中心

Global satellite observations document expansions of the low-chlorophyll central ocean gyres and an overall inverse relationship between anomalies in sea surface temperature and phytoplankton chlorophyll concentrations. These findings can provide an invaluable glimpse into potential future ocean changes, but only if the story they tell is accurately interpreted. Chlorophyll is not simply a measure of phytoplankton biomass, but also registers changes in intracellular pigmentation arising from light-driven (photoacclimation) and nutrient-driven physiological responses. Here, we show that the photoacclimation response is an important component of temporal chlorophyll variability across the global ocean. This attribution implies that contemporary relationships between chlorophyll changes and ocean warming are not indicative of proportional changes in productivity, as light-driven decreases in chlorophyll can be associated with constant or even increased photosynthesis. Extension of these results to future change, however, requires further evaluation of how the multifaceted stressors of a warmer, higher-CO₂ world will impact plankton communities.

Ocean warming has been implicated as causing an expansion of the low-chlorophyll, low-productivity central ocean gyres^{1, 2, 3}. Satellite observations have also shown that broad ocean regions exhibit an inverse relationship between interannual and interdecadal changes in sea surface temperature (SST) and surface phytoplankton chlorophyll concentrations^{4, 5, 6, 7}. In other words, chlorophyll tends to decrease when temperatures increase, or increase when temperatures decrease. If chlorophyll concentration is simplistically taken as a measure of phytoplankton biomass, then these findings imply a warmer future ocean may be accompanied by decreased phytoplankton stocks and productivity^{2, 4, 8, 9, 10}. However, the chlorophyll signal is not so easy to interpret. Chlorophyll concentrations also register physiological adjustments in cellular pigmentation arising from changes in upper ocean light and nutrient conditions^{11, 12, 13, 14}. These physiological responses can potentially undermine earlier interpretations of the SST–chlorophyll relationship, and thus its extension to future ocean warming impacts^{5, 7}.

Here, we show that physiological changes in cellular pigmentation are often the dominant cause of satellite-observed interannual variations in chlorophyll. This physiological signal includes a clear signature of phytoplankton responses to changing mixed-layer light conditions (that is, ‘photoacclimation’). Our findings imply that temperature-correlated decreases in chlorophyll over large ocean areas are often not synonymous with equivalent decreased productivity. To arrive at this conclusion, it was necessary to first revisit mechanisms of light-regulated chlorophyll synthesis to construct a photoacclimation model applicable to the dynamic light environment of the upper ocean. The model was then evaluated against global phytoplankton carbon-to-chlorophyll ratio data (θ, a standard metric of cellular pigmentation), before assessing the contribution of photoacclimation to the satellite chlorophyll record. Our results yield a less dire interpretation of contemporary ocean phytoplankton changes, provide new insight into chlorophyll synthesis regulation, and offer a revised description of photoacclimation that can benefit ocean ecosystem modelling, global productivity estimates, and evaluations of photoprotection in the light-saturated upper ocean.

Oxygenic photosynthesis involves coupled electron transport between two pigmented light-harvesting photosystems, termed PSII and PSI (Fig. 1a). An inner-membrane pool of plastoquinone (PQ) molecules functions as the electron shuttle between these photosystems, while simultaneously transporting protons across the photosynthetic membrane (which drives ATP synthesis). When phytoplankton are exposed to a range of increasing light intensities, their photosynthetic rate initially increases with light, but then saturates because ATP and reductant (NADPH) turnover become rate limiting (Fig. 1b). Consequently, electron transport between PSII and PSI progressively ‘backs-up’ as light increases, causing the PQ pool to become increasingly reduced (that is, in the form of plastoquinole, PQH₂ (Fig. 1b)). The ratio of these oxidized to reduced molecules (that is, the PQ pool ‘redox state’) thus provides a sensitive measure of the balance between the cell’s light-harvesting ‘machinery’ and its capacity to utilize its photosynthetically generated ATP and NADPH. Phytoplankton use this signal (along with other correlated signals) to regulate chlorophyll synthesis^{15, 16, 17, 18} (Supplementary Discussion). A predominantly oxidized PQ pool indicates that light harvesting is insufficient, so chlorophyll synthesis is upregulated. Conversely, a pool that is predominantly reduced indicates that chlorophyll concentration is too high, so synthesis is downregulated. This relationship between PQ redox state and chlorophyll synthesis is fundamental to photoacclimation models applied to satellite observations of surface ocean mixed-layer phytoplankton.

a, Electron transport in photosynthetic membranes between water splitting at PSII and reductant (NADPH) formation after PSI. Also shown is electron transport in darkness from NADPH to the PQ pool through Ndh. PSII, photosystem II. PSI, photosystem I. Cyt b₆f, cytochrome b₆f complex. PQ, plastoquinone. PQH₂, plastoquinole. Fdx, ferredoxin. PC, plastocyanin. NADPH, nicotinamide adenine dinucleotide phosphate. NDH, NADPH dehydrogenase. Yellow ‘lightning bolts’ signify light absorption by PSII and PSI. b, Relationship between photosynthesis and incident irradiance (black line; relative). Photosynthesis initially increases in proportion to irradiance and then becomes light-saturated. The PQ pool is initially reduced in the dark, is oxidized on exposure to light, and then becomes progressively reduced with increasing light (red line; right axis). c, Changes in phytoplankton carbon-to-chlorophyll ratio (θ; green symbols; left axis; g C (g Chl)⁻¹) and cell division rate (red symbols; right axis; d⁻¹) as a function of growth irradiance (mol photon m⁻² h⁻¹) as typically observed in nutrient-replete laboratory cultures. Data for Dunaliella tertiolecta from ref. 5. d, Relationship between θ and cell division rate for nutrient-replete phytoplankton based on data from c. e, Relationship between θ and cell division rate for steady-state nutrient-limited phytoplankton. Circles: Thalassiosira weisflogii under (blue) NO₃, (red) NH₄ and (green) PO₄ limitation. Crosses: NO₃-limited cultures of (blue) Dunaliella tertiolecta, (purple) Thalassiosira weisflogii and (orange) Ostreococcus tauri. Data from refs 11,47.

Photoacclimation has been thoroughly examined in laboratory phytoplankton cultures^{11, 12, 14, 25, 26}, albeit under light conditions that rarely mimic natural deep-mixing layers. During these experiments, phytoplankton grown over a range of different light levels typically exhibit a strong decrease in θ with decreasing growth irradiance (that is, they become more pigmented; Fig. 1c, green symbols). This response, regulated by PQ redox sensing, is insufficient to prevent reductions in cell division rates at low light (Fig. 1c, red symbols). These parallel responses to changing growth irradiance can be summarized by replotting θ as a function of growth rate (Fig. 1d), thus illustrating how light-driven increases in division rate correspond to strong increases in θ. The opposite relationship is observed in laboratory studies of nutrient stress (Fig. 1e). Here, increases in division rate are associated with strong decreases in θ because faster-growing cells need more chlorophyll (that is, lower θ) to meet demands for photosynthetic ATP and NADPH production.

The dependencies of θ on light and nutrient conditions are critical for understanding why satellites often observe chlorophyll to decrease with increasing temperature and stratification, and what these changes mean in terms of ocean production. More specifically, chlorophyll can decrease because of a decrease in biomass (lower productivity) or an increase in θ. If the change in θ is significant, it can result from an increase in nutrient stress (lower productivity; Fig. 1e) or an increase in growth irradiance (same or increased productivity; Fig. 1d). Interpreting the satellite chlorophyll record therefore requires separating the effects of biomass, nutrients and light, which in turn requires a robust description of photoacclimation for the surface ocean mixed-layer light environment.

Over the photoperiod, incident sunlight in the mixed layer is attenuated (that is, decreases) exponentially with depth. During midday, photosynthesis near the surface is often light-saturated, meaning that the PQ pool is reduced and the signal for chlorophyll synthesis is ‘off’ (Fig. 2a). Supersaturating light levels within this ‘high light zone’ provide no additional redox information, but they do impact calculated values of the average light in the mixed layer. This discrepancy is one reason why average daily irradiance is not an appropriate descriptor of photoacclimation. Recognizing this limitation, an alternative approach has been to describe photoacclimation as a function of the median light level within a well-mixed surface layer (I_ML; refs 7,13,27).

where PAR is the photosynthetically active radiation (400–700 nm; mol photons m⁻² h⁻¹), K_d is the attenuation coefficient (m⁻¹) for downwelling PAR, and MLD is the mixed-layer depth (m). This approach was intended to capture the midpoint light level where the signal for chlorophyll synthesis is ‘on’ for half of the mixing cycle and ‘off’ for the other half. However, neither the median nor the average mixed-layer light level account for the chlorophyll synthesis signal being ‘off’ in the dark (Fig. 2a). Consequently, these descriptors of photoacclimation can overestimate cellular chlorophyll levels when mixing depths exceed the photic zone depth (Z_eu; the surface layer of the ocean supporting net photosynthesis).

a, Schematic of the mixed-layer (indicated on left) light environment across the photoperiod (top). Redox state of the plastoquinone (PQ) pool is an important regulator of chlorophyll synthesis. Elevated incident sunlight during midday results in light-saturated photosynthesis in the upper photic layer, a reduced PQ pool (Fig. 1), and thus downregulation of chlorophyll synthesis (that is, ‘signal off’; area above black dashed line). Similarly, exposure to darkness below the photic depth (white dashed line) results in a biochemical reduction of the PQ pool (Fig. 1) and downregulation of chlorophyll synthesis. In between these two conditions, the PQ pool is oxidized and signals an upregulation of chlorophyll synthesis. Phytoplankton carbon-to-chlorophyll ratios (θ) reflect the balance between these ‘on’ and ‘off’ signals, which depends on mixed-layer depth (MLD; m), incident sunlight (PAR; mol of photons m⁻² h⁻¹), and the attenuation coefficient for PAR (K_d; m⁻¹). b, Steady-state values of θ (g g⁻¹) from the new photoacclimation model as a function of MLD (m). Solid symbols, deep-mixing scenarios where MLD > photic depth (Z_eu; m). Open symbols, shallow-mixing scenarios where MLD < Z_eu. c, Relationship between model values of θ and PAR^0.45/K_d (mol photon (m h)⁻¹). Black line, equation (2). d, Relationship between θ values for the full model (left axis) and from equations (2) and (3) (bottom axis). b–d, Symbol shapes and colours indicate model combinations of PAR and K_d, as defined in the key at the bottom of the figure.

Satellite measurements provide information on ocean ecosystem properties representative of the actively mixing surface layer, which can vary from tens to hundreds of metres. Retrieved chlorophyll concentrations reflect both phytoplankton abundance and cellular pigmentation (θ). An ability to partition chlorophyll into these two components has arisen from the development of spectral ocean colour inversion algorithms^{30, 31, 32} and subsequent assessments of phytoplankton carbon (C_phyto) from retrieved particulate backscattering coefficients (b_bp; refs 13,27). Initially, C_phyto products were evaluated against indirect biomass proxies^{13, 27, 33, 34}, but recent analytical field measurements of C_phyto (ref. 35) have now directly validated the satellite algorithm³⁶. Simultaneous retrieval of C_phyto and Chl allows global evaluations of phytoplankton θ variability.

MODIS Aqua 8-day resolution θ data were aggregated into 37 regional bins¹³ (Fig. 3a) and initially compared with median mixed-layer light values (I_ML; Fig. 3b). The resultant relationship shows strong regional dependencies that illustrate how fundamental aspects of photoacclimation are insufficiently accounted for by I_ML alone. In contrast, the photoacclimation model developed here (equations (2) and (3)) gives predicted values of θ that are far more consistent (r² = 0.92) with satellite observations for nearly all of the regional bins (Fig. 3c). High-latitude Southern Ocean data are clear outliers in this relationship (Fig. 3c; red symbols), probably reflecting biases in satellite ocean retrievals for this region that are well documented^{37, 38}. The model also consistently underestimates observed θ values by ~20% for the lowest-variance regions of the South Pacific Gyre (SPG in Fig. 3a; black symbols in Fig. 3c), probably reflecting a particularly high Raman scattering contamination in b_bp retrievals for this region of the clearest natural waters on Earth³⁹.

a, The 11-year record of MODIS Aqua 8-day resolution θ data were aggregated into 37 regional bins based on annual variability in chlorophyll, following ref. 13 (right-hand legend; L0, lowest variability (that is, most stable) waters; L4, highest variability waters). Basin designations are: NA, North Atlantic; CA, Central Atlantic; SA, South Atlantic; NP, North Pacific; CP, Central Pacific; SP, South Pacific; CI, Central Indian; SI, South Indian; SO, Southern Ocean; Med., Mediterranean; and Excluded, near-shore and polar waters excluded from the analyses. Central ocean gyres are indicated by the addition of ‘G’ to two-letter basin designation (for example, SPG, South Pacific Gyre). b, Relationship between satellite-observed values of θ (g C (g Chl)⁻¹) and median mixed-layer light levels (I_ML, mol photons m⁻² h⁻¹). Colours, ocean basin as defined in the key at the bottom. South Pacific data for the low-variance L0 and L1 bins (black) are separated from higher-variance bins (brown) (see discussion in the main text). c, Relationship between satellite-observed θ and model values from equations (2) and (3).

Satellite observations of global ocean phytoplankton began in 1978. From the