全球变化知识资源中心

A changing climate is altering many ocean properties that consequently will modify marine productivity. Previous phytoplankton manipulation studies have focused on individual or subsets of these properties. Here, we investigate the cumulative effects of multi-faceted change on a subantarctic diatom Pseudonitzschia multiseries by concurrently manipulating five stressors (light/nutrients/CO₂/temperature/iron) that primarily control its physiology, and explore underlying reasons for altered physiological performance. Climate change enhances diatom growth mainly owing to warming and iron enrichment, and both properties decrease cellular nutrient quotas, partially offsetting any effects of decreased nutrient supply by 2100. Physiological diagnostics and comparative proteomics demonstrate the joint importance of individual and interactive effects of temperature and iron, and reveal biased future predictions from experimental outcomes when only a subset of multi-stressors is considered. Our findings for subantarctic waters illustrate how composite regional studies are needed to provide accurate global projections of future shifts in productivity and distinguish underlying species-specific physiological mechanisms.

An ongoing major challenge is to grasp how climate-change-mediated alteration of environmental conditions will influence biota across different scales, from organismal health to community structure^{1, 2}. Oceanographers have employed climate-change models^{3, 4}, time-series observations⁵ and manipulation experiments⁶ to understand the biological ramifications of global change. Phytoplankton manipulation studies reveal how alteration of individual properties, such as CO₂, affects physiology^{2, 6, 7}. However, the validity of such single-parameter findings^{6, 8, 9}, in the context of complex ocean change^{1, 2}, is challenged by research that reveals interactive effects between multi-stressors on phytoplankton physiology^{10, 11}. We need to diagnose and understand the physiological mechanisms that underpin interconnected responses to multi-stressors, which together set the cumulative response of phytoplankton species to changing conditions^{4, 6, 8}.

Understanding the combined effects, across the global ocean, of complex change on phytoplankton physiology requires a gradualist approach^{12, 13}. Individual provinces will encounter different permutations of multi-faceted change¹⁴, and each is characterized by a range of resident phytoplankton groups (termed biomes⁵). Earth System models provide a framework of projections of regional change¹⁴ that stimulate improved experimental design to understand the biological effects of oceanic change. In return, a new generation of manipulation studies must deliver estimates of the combined effects of complex change on many phytoplankton species, and distinguish the underlying mechanisms that underpin these physiological outcomes.

Here, we target subantarctic diatoms, which are ubiquitous and bloom-formers¹⁵. We experimentally manipulate a representative species^{6, 15} (Pseudonitzschia multiseries) under year 2100 conditions to quantify its response to ocean change. For simplicity, owing to the complex nature of our multi-stressor experiment, we chose batch cultures that permit initial (high nutrient) conditions to be modified biologically but require careful monitoring. Our experimental design, along with physiological metrics and comparative proteomics, enables diagnosis of individual and interactive effects of ocean properties on diatom physiology. Thus, regionally we can quantify the cumulative effect of complex change, and begin to identify underlying physiological mechanisms, as a first step towards re-evaluation of climate-change biogeochemical model parameterizations and experimental designs^{3, 4, 13}.

We commence by outlining a new experimental design that relies on recognition of the controlling physiological variable for the study organism. Fullest interpretation of results requires the application of many physiological diagnostics, together with a statistical approach that is powerful enough to unravel the relative contribution of individual and interactive environmental effects on our diatom. At present, even sophisticated experiments^{10, 11} manipulate only subsets of properties projected to change by 2100 (refs 1, 2, 3, 4). To address the dual issues of quantification of the cumulative effects of complex change and its mechanistic underpinning, we require an experiment that supersedes present-day single- or 2–3-parameter manipulations. We employed a collapsed factorial design that provides a tractable, efficient, approach while concurrently manipulating the stressors that exert major physiological controls (temperature/CO₂/nutrients/iron/light⁶). This streamlined design requires identification of the dominant physiological control¹⁶ before grouping (that is, collapsing) the remaining stressors into one combined factor (Methods).

In the Southern Ocean, temperature is recognized as setting the upper bound on diatom growth¹⁷. For P. multiseries, we used a literature-based physiological ranking⁶ to identify temperature as the (putative) dominant control. Its pivotal role for our subantarctic diatom was substantiated by a reaction norm that revealed twofold higher growth (Fig. 1), on 3 °C warming projected for 2100 (refs 3, 4, 18); note, this corroboration is contingent on our selection criteria (Fig. 1 caption) and different outcomes are possible if other metrics are applied. The remaining parameters (CO₂/nutrients/iron/light) were then grouped into a combined factor. Next, we employed a 2² factorial design with four treatments (Fig. 2): (A) control; (B) 2100 warming only¹⁸; (C) 2100 conditions without warming; (D) 2100 conditions¹⁸ (Table 1). Our approach balances the needs of predicting cumulative physiological effects of future conditions with identifying the nature of environmental forcing. This method led to improved efficiency in experimental design (2² compared with 2⁵ treatments for 5-factors), and the orthogonality of the dominant physiological control with the collapsed stressors permits identification of how much variation is explained by temperature alone, the collapsed stressors, and their interplay.

The norm substantiated temperature as the dominant physiological control on subantarctic diatoms following our arbitrary selection criteria. Error bars denote ± s.d. of the mean (n = 3), and the temperature bounds of the purple rectangle denote the mean (10.6 °C) for subantarctic surface waters for present-day spring/summer^{29, 46} used in treatments A and C, and that projected (13.7 °C) for this province by 2100 (ref. 18) and used in B and D. Doubling of growth from 10.6 to 13.7 °C reveals that selection of these temperatures cause the largest increase in growth rate per °C warming, relative to present-day temperature. Thus, the role of warming in setting growth of this subantarctic diatom will be time-dependent and may decrease beyond 2100 unless diatoms can respond by altering their thermal traits⁴⁷.

Figure 3 reveals wide-ranging physiological responses across treatments A–D. As anticipated (Fig. 1), the individual effect of warming (B) significantly increased (chlorophyll-based) growth from 0.49 ± 0.05 to 0.75 ± 0.05 d⁻¹. However, the effect of warming alone on other metrics was either small (cellular chlorophyll, Si) or inconclusive. In C, despite higher iron, p_CO2 and irradiance, contrasting trends were evident: growth was 0.37 ± 0.05 d⁻¹, cellular chlorophyll increased sixfold, and cellular N and P increased fivefold. In D, where warming was one of five altered multi-stressors, more pronounced patterns emerged. Growth (0.99 ± 0.09 d⁻¹) doubled and substantial decreases were observed for cellular P and Si (compared with treatment A, Fig. 3). Model-averaged estimates of inter-comparisons of physiological effects show striking differences for individual and interactive effects of warming and other multi-stressors across these metrics (Supplementary Table 2).

Error bars are the s.d. of the mean (n = 3).^∗Denotes data from one C treatment where some assays were at the limit of detection (that is, lower than the blank) owing to the weak physiological response. No significant change in cell size was observed across treatments A–D. Note, batch cultures for treatments A–D were each sampled/collected during exponential growth (Supplementary Fig. 1) and hence cells were not resource-limited. At the experiment’s conclusion, pronounced decreases in phosphate and increases in pH were evident in D, suggesting resource limitation (Supplementary Table 1). However, the physiological ramifications of these shifts are probably minor on the basis of reported plasticity of laboratory-cultured P. multiseries in response to increased pH (to 8.4; ref. 48), and bloom-forming diatoms that were not P-limited in batch cultures at ~0.5 μmol PO₄ l⁻¹ (ref. 49).

Individual versus interactive effects.

The individual effect of warming on subantarctic P. multiseries physiology was compared with published studies (Supplementary Table 5). Warming influences subpolar diatoms in different ways, including increasing cellular chlorophyll^{22, 23}, carbon^{22, 23}, C/N ratios²² and growth²³, but decreasing both cellular P (ref. 24) and cell size^{25, 26}. Here, warming resulted in increased growth rate and cellular chlorophyll (Fig. 3). The differentially abundant proteins in treatments A versus B were enriched for B in processes involved mostly in intracellular transport (for example, vesicle coat complex AP-2, clathrin adaptor complex, and inner membrane protein translocase, Data set 1). This is indicative of greater intracellular protein transport and turnover when diatoms are exposed to warming²⁷.

The metric for maximum growth (μ_max) as a function of temperature²⁸ provides insights into the relative roles of warming and other stressors for our diatom species. Although temperature sets the upper bound on Southern Ocean phytoplankton growth¹⁷, iron-enrichment^{6, 29} and/or nutrient stress or limitation³⁰ can modify rates and influence μ/μ_max. As expected, our low-iron control showed submaximal growth (0.5 d⁻¹ compared with 1.1 d⁻¹ predicted μ_max (ref. 28) at 10.6 °C). Future warming in B increased growth to submaximal rates (0.75 d⁻¹ compared with 1.3 d⁻¹μ_max at 13.7 °C) revealing that other factors either individually/or interactively restricted growth. The influence of these stressors on growth is evident from statistical modelling (Fig. 4), and from D, which had the highest rate (0.99 ± 0.05 d⁻¹). Despite using batch cultures, submaximal growth in D is more likely to be influenced by the individual/interactive effects of 30% less macronutrients for future subantarctic waters¹⁸ than higher CO₂/irradiances that each have either negligible²⁵ or positive²³ effects on diatom growth, respectively.

Our collapsed factorial design enables quantification of individual and interactive effects of temperature, but requires inferences to be made into the interplay of temperature and other stressors. Diagnosis of this interplay relied on a comparison of physiological trends from treatments C and D with studies where individual²⁶ and/or several²⁵ properties (temperature/p_CO2, iron/p_CO2 were manipulated (Supplementary Table 5). For example, iron enrichment increases diatom growth^{26, URL:}