Is our understanding of aquatic ecosystems sufficient to quantify ecologically driven climate feedbacks?
Corday R. Selden, Richard LaBrie, and Laura C. Ganley should be considered joint first author.
Abstract
enThe Earth functions as an integrated system—its current habitability to complex life is an emergent property dependent on interactions among biological, chemical, and physical components. As global warming affects ecosystem structure and function, so too will the biosphere affect climate by altering atmospheric gas composition and planetary albedo. Constraining these ecosystem-climate feedbacks is essential to accurately predict future change and develop mitigation strategies; however, the interplay among ecosystem processes complicates the assessment of their impact. Here, we explore the state-of-knowledge on how ecological and biological processes (e.g., competition, trophic interactions, metabolism, and adaptation) affect the directionality and magnitude of feedbacks between ecosystems and climate, using illustrative examples from the aquatic sphere. We argue that, despite ample evidence for the likely significance of many, our present understanding of the combinatorial effects of ecosystem dynamics precludes the robust quantification of most ecologically driven climate feedbacks. Constraining these effects must be prioritized within the ecological sciences for only by studying the biosphere as both subject and arbiter of global climate can we develop a sufficiently holistic view of the Earth system to accurately predict Earth's future and unravel its past.
Résumé
frLa Terre fonctionne comme un système intégré—son habitabilité pour une vie complexe est une propriété émergente qui dépend des interactions entre les composantes biologiques, chimiques et physiques. Le réchauffement climatique affecte la structure et la fonction des écosystèmes, et en retour, la biosphère affecte également le climat en modifiant la composition des gaz atmosphériques et l'albédo planétaire. Il est essentiel de quantifier ces rétroactions entre les écosystèmes et le climat afin de prédire avec précision les changements futurs et élaborer des stratégies d'atténuation; cependant, l'interaction entre les processus écologiques complique l'évaluation de leurs impacts. Dans cet article, nous examinons l'état des connaissances sur la façon dont les processus écologiques et biologiques (par exemple, la concurrence, les interactions trophiques, le métabolisme, l'adaptation) affectent la directionnalité et l'ampleur des rétroactions entre les écosystèmes et le climat à l'aide d'exemples issus du monde aquatique. Nous soutenons que, malgré les nombreuses preuves de l'importance de plusieurs de ces rétroactions, notre compréhension limitée des effets additifs des processus écosystémiques empêche de faire une quantification robuste de la plupart des rétroactions climatiques d'origine écologique. Circonscrire ces effets doit être une priorité pour les sciences aquatiques, car ce n'est qu'en étudiant la biosphère en tant que sujet et arbitre du climat planétaire que nous pourrons développer une vision suffisamment holistique du système terrestre pour prédire avec précision l'avenir de la Terre et élucider son passé.
Zusammenfassung
deDie Erde funktioniert als einheitliches System—ihre derzeitige Bewohnbarkeit für komplexes Leben ist eine Eigenschaft, die sich entwickelt hat und von Wechselwirkungen zwischen biologischen, chemischen und physikalischen Komponenten abhängt. So wie die globale Erwärmung die Struktur und Funktion des Ökosystems beeinflusst, wird auch die Biosphäre das Klima beeinflussen, indem sie die Zusammensetzung der atmosphärischen Gase und die Albedo des Planeten verändert. Diese Ökosystem-Klima-Rückkopplungen zu bewerten ist von entscheidender Bedeutung für die genaue Vorhersage zukünftiger Veränderungen und zur Entwicklung von Minderungsstrategien. Das Zusammenspiel ökologischer Prozesse erschwert jedoch die Bewertung dieser Rückkopplungen. Anhand anschaulicher Beispiele aus aquatischen Lebensräumen, erläutern wir hier den Stand des Wissens darüber, wie ökologische und biologische Prozesse (z. B. Konkurrenz, trophische Interaktionen, Stoffwechsel, Anpassung) die Richtung und das Ausmaß von Rückkopplungen zwischen Ökosystemen und Klima beeinflussen können. Wir argumentieren, dass trotz zahlreicher Belege für die wahrscheinliche Bedeutung von ökologisch bedingten Klimarückkopplungen, unser derzeitiges Verständnis der additiven Effekte von Ökosystem Dynamiken eine zuverlässige Quantifizierung vieler ausschließt. Die Abschätzung dieser Auswirkungen sollte stärker in den Fokus innerhalb der Umweltwissenschaften rücken. Denn nur wenn wir verstehen, wie die Biosphäre und das globale Klima einander beeinflussen, können wir eine ausreichend ganzheitliche Betrachtung der Erde als integriertes System entwickeln. Nur mit dieser ganzheitlichen Betrachtung, werden wir in der Lage sein, die Zukunft der Erde genau vorherzusagen und ihre Vergangenheit aufzuklären.
תקציר
heהמדע חייב לתעדף את אבחון ואפיון השפעות אלו משום שרק על ידי הסתכלות הוליסטית על מצב הביוספירה כסיבה ומסובב של האקלים העולמי ניתן יהיה לחזות במדויק את עתיד כדור הארץ ולפענח את עברו הראיות הרבות לחשיבותם האפשרית של רבים מתהליכי המערכת האקולוגית, ההבנה הקיימת אינה מספקת לכימות סך השפעתם על שינויי האקלים.כדור הארץ פועל כמערכת משולבת — יכולתו לתמוך בחיים מורכבים הוא מאפיין התלוי ביחסי הגומלין בין רכיביו הביולוגיים, הכימיים והפיזיקליים. ככל שההתחממות הגלובלית משפיעה על המבנה ופעולתה של המערכת האקולוגית, כך גם הביוספירה תשפיע על האקלים על ידי שינוי הרכב הגזים האטמוספריים ובאלבדו הפלנטרי. אפיון לולאות משוב אלו חיוני לחיזוי מדוייק של שינויים עתידיים ולפיתוח אסטרטגיות להפחתת נזקים; אולם, קשרים בין תהליכים אקולוגיים שונים והשפעתם ההדדית מקשים על הערכת סך השפעתם הכוללת. כאן, אנו חוקרים את הידע הקיים על האופן בו תהליכים אקולוגיים וביולוגיים (לדוגמא, תחרות, יחסים טרופיים, מטבוליזם, התאמה) משפיעים על כיווניות וגודל לולאות המשוב בין מערכות אקולוגיות והאקלים, באמצעות דוגמאות ממערכות מיימיות. אנו טוענים כי, למרות
1 INTRODUCTION
In a 1972 letter to the editor of Atmospheric Environment, James Lovelock (1972) posited that Earth's habitability to extant life was not a fortunate coincidence of nature but, rather, an emergent property inextricably linked to life's evolution. By conceptualizing the Earth as a “super-organism” (Lovelock, 1995) capable of maturation, homeostasis, and inevitably death, Lovelock and his collaborator, microbiologist Lynn Margulis, challenged scientists to explore how seemingly disparate biological communities might interact as tissues in a body to regulate the physical state of the whole (Lovelock & Margulis, 1974a, 1974b; Margulis & Lovelock, 1974). Fifty years later, the so-called Gaia hypothesis—at once both controversial and foundational—remains inspirational. Yet, critical examination suggests that it is incomplete (Tyrrell, 2013).
Biological climate feedbacks (Figure 1) are cycles in which an environmental force alters biological function, which, in turn, amplifies (amplifying/positive feedbacks) or lessens (stabilizing/negative feedbacks) the stimulus. They arise across spatial scales, generating emergent properties. At the population and community scales, niche construction and the development of alternative biome states arise due to stabilizing feedbacks (Pausas & Bond, 2022); the biogeochemical cycles that maintain planetary habitability have similarly emerged at the global scale (Falkowski et al., 2008). Yet, life can also drive its environment into a less favorable state. The advent of oxygenic photosynthesis ultimately cloistered its anaerobic progenitors to deep recesses of the Earth (Canfield, 2014; Falkowski & Godfrey, 2008). Similarly, land plant evolution led to a decline in atmospheric carbon dioxide (CO2), which cooled the Earth to an unfavorable degree (Tyrrell, 2013 and references therein). Such observations spurred a revised view of biosphere-geosphere interactions: life and the planet have coevolved through reciprocal influence but not necessarily to the boon of biology (Schneider, 1986; Schneider & Londer, 1984).
As we cannot rely on the Earth system to self-stabilize, we must move quickly to better understand the processes that regulate global climate and develop effective stewardship practices. Despite the significance of biological feedbacks and their potential use in geoengineering schemes, they are omitted from General Circulation Models (Prentice et al., 2015). Current efforts are underway to parameterize these processes in Earth Systems Models, yet significant hurdles remain (Bonan & Doney, 2018; Heinze et al., 2019). In particular, large uncertainties arise as ecosystem functional responses are extrapolated from short-term measurements at local scales to longer-term global estimates (Heinze et al., 2019). Moreover, biological adaptation and evolution are generally not considered but may significantly alter climate projections (Ward et al., 2019).
Constraining the directionality and magnitude of ecosystem-climate feedbacks is essential to predicting Earth's climate future and unraveling its past, but is our understanding of the Earth systems presently sufficient to quantify these processes? Here, we consider the state-of-knowledge on how ecosystem processes affect climate by altering albedo (Section 2), modulating greenhouse gas (GHG) fluxes (Section 3), and redistributing carbon (C) through long-term reservoirs (Section 4) using illustrative examples drawn from aquatic ecosystems. Our goal is not to provide a comprehensive review of ecologically driven climate feedbacks but, rather, to highlight key uncertainties (see Boxes) and identify knowledge gaps pertinent to improving our understanding of Earth system function. While physical processes can directly affect climate (e.g., Kang et al., 2023), purely physical climate feedbacks are beyond our scope. We hope that this article offers fertile grounds for identifying and prioritizing new areas of research in aquatic ecology and inspires cross-disciplinary problem-solving to address today's climate crisis.
2 ALBEDO ALTERATION
The quintessential example of ecologically driven climate feedback comes from the ability of certain marine phytoplankton to influence atmospheric concentrations of cloud condensation nuclei (CCN, Figure 2), aerosols on which water vapors condense to form cloud droplets (Charlson et al., 1987). Cloud formation impacts Earth's albedo, that is, its propensity to reflect incoming solar radiation back to space, which in turn affects Earth's heat budget by altering the balance of incoming and outgoing radiation (McNeill, 2017). Indeed, aerosol-cloud interactions remain the largest uncertainty in Earth's heat budget due to difficulties constraining the dependence of CCN activity on aerosol concentration, size, and chemical composition (Forster et al., 2021). In the marine environment, phytoplankton community composition and metabolic state, among other ecological factors, impact aerosol production and chemical composition, influencing regional and global climate (Brooks & Thornton, 2018).
2.1 Dimethyl sulfide production: The classic feedback
Sulfate aerosols are an important CCN in the marine atmosphere that can regulate cloud formation, coverage, and lifetime (Andreae, 1990). The largest nonanthropogenic source of marine sulfate aerosols comes from atmospheric oxidation of the biogenic, volatile organosulfur compound dimethyl sulfide (DMS) (Fitzgerald, 1991; Kilgour et al., 2022). DMS is an algal metabolite derived from the enzymatic cleavage of dimethylsulfoniopropionate, which is used as an osmolyte (Vairavamurthy et al., 1985), antioxidant (Sunda et al., 2002) and grazing defense agent (Wolfe et al., 1997; Yoch, 2002) in many eukaryotic algae including Prymnesiophyceae (e.g., Phaeocystis) and Dinophyceae (e.g., Gyrodinium) (Keller et al., 1989; Yoch, 2002). Climate models indicate that DMS production cools the climate by 1.7–2.3 W/m2 (Hopkins et al., 2023)—a value on par with the increased radiative forcing from contemporary (2019) anthropogenic CO2 emissions (2.16 W/m2) (Forster et al., 2021).
Early work on DMS, as well as popular discourse on Gaia, prompted the formulation of the CLAW hypothesis (Charlson et al., 1987) which states that, by modulating DMS production in response to changes in temperature, marine ecosystems alter global climate. Subsequent work has shown that DMS production by phytoplankton increases with solar radiation and bacterial dimethylsulfoniopropionate degradation increases with temperature (Levine et al., 2012; Toole & Siegel, 2004), representing a stabilizing feedback. However, DMS release also depends on numerous ecological processes, including multitrophic predator–prey interactions (Dove, 2015; Nevitt et al., 1995; Shemi et al., 2021; Thewissen et al., 2011) and nutrient satiety (Sunda et al., 2007), the effects of which are not yet quantitatively constrained. Moreover, increasing temperatures may reduce biomass of eukaryotic phytoplankton (e.g., Phaeocystis), decreasing global DMS production and flux by an estimated 15% and 8%, respectively (Wang et al., 2018). This decrease is expected to lower sulfate aerosol concentrations in the atmosphere (Wang et al., 2018), but the subsequent impact on marine cloud coverage may not be as significant as originally predicted by the CLAW hypothesis, which underestimated the contributions from additional marine CCN sources such as sea spray aerosols (Fossum et al., 2020; Quinn & Bates, 2011). Finally, the consequences of phenological shifts among DMS-producing algae, and potential match/mismatches with grazers, remain largely unresolved (Box 1).
BOX 1. Does organismal phenology affect how ecosystems respond to climate?
The study of temporally recurrent life history events (phenology) highlights the effect of seasonal shifts on ecosystem dynamics (Edwards & Richardson, 2004). Climate change has modified both seasonal environmental characteristics and organismal traits—including metabolic rates, distribution, and trophic interactions—which influences their phenology (e.g., Beaugrand, 2009; Edwards & Richardson, 2004; Pendleton et al., 2022). Marine phytoplankton blooms, for example, now start earlier and last longer due primarily to ocean warming, with increased and decreased intensities observed at higher and lower latitudes, respectively (Friedland et al., 2018). Shifts in phenology can also be associated with potential mismatches in limnic and marine food webs (Ji et al., 2010; Thackeray et al., 2013; Visser & Both, 2005), significantly altering ecosystem function. Recent modeling efforts from the Southern Ocean indicate that temperature-driven changes in zooplankton community structure delay bloom initiation by 3 weeks and decrease C export (Karakuş et al., 2022). However, the magnitude and direction of phenological shifts can differ markedly across ecosystems. For example, in the English Channel, the phenological mismatch between zooplankton and phytoplankton growth appears unrelated to temperature and does not alter species abundance/production (Atkinson et al., 2015), suggesting that future phenological shifts may disrupt food webs in highly seasonal, pulsed ecosystems like in the Southern Ocean, but that other factors may compensate elsewhere. Significantly more work is needed to understand how changing phenology will alter primary productivity and C export efficiency (Section 4) in aquatic ecosystems.
2.2 Biologically enhanced sea spray aerosols regulate cloud formation and DMS-climate feedback
Sea spray aerosols (SSAs), which are formed by bubbles bursting at the ocean surface, contain a complex mixture of inorganic ions, organic molecules, and biological components, with their composition strongly influenced by the interactions of marine primary and secondary producers (Crocker et al., 2022; O'Dowd et al., 2015). Due to their salt content, SSAs are also highly effective CCN, and their competition for water vapor can suppress the activation of sulfate aerosols into cloud droplets, decreasing cloud formation (Fossum et al., 2020). On the other hand, under conditions of low sulfate aerosol concentrations, SSA serves as a secondary CCN source to facilitate cloud formation (Fossum et al., 2020). Through the combination of these two feedbacks, SSAs modulate the impact of changes in DMS emissions on marine cloud coverage and climate. Moreover, the presence of biological exudates enhances the ability of SSAs to act as heterogeneous ice-nucleating particles (INPs) in mixed-phase and cirrus clouds (Alpert et al., 2022; Hill et al., 2023). Recent modeling work found large, seasonally dependent effects of marine INPs on mixed-phase global net cloud forcing with changes of −0.21 W/m2 in wintertime and +0.48 W/m2 in summertime (Zhao et al., 2021). Marine INPs have also been observed in high-altitude cirrus clouds, which are important to Earth's radiative balance because they contribute to planetary warming by trapping outgoing longwave radiation that would otherwise escape to space (Cziczo et al., 2004, 2013). Although warming is expected to reduce the global abundance of some well-known ice-nucleating algae such as diatoms, significant uncertainties exist regarding how temperature-driven shifts in ecosystem structure and environment will impact future SSA INP concentrations, especially at a regional level (Tesson & Šantl-Temkiv, 2018). For example, sea ice retreat at high latitudes—where diatoms are abundant (Pautova et al., 2023)—will lead to exposure of below-ice algal blooms that may contribute to higher local SSA INP concentrations. Future studies accounting for the impacts of phytoplankton community shifts on SSA INP concentrations may provide critical insight into how climate-driven ecological changes will affect cloud formation and, ultimately, Earth's radiative balance.
3 GREENHOUSE GAS PRODUCTION
GHGs insulate Earth by impeding the escape of outgoing (longwave) radiation. In addition to geochemical sources, several GHGs are released as metabolic waste products. These waste products include CO2, which is generated through multiple aerobic and anaerobic respiratory pathways, and methane (CH4) and nitrous oxide (N2O), which are released during the anaerobic respiration of small C molecules (methanogenesis) and nitrate (denitrification), respectively [although N2O is also produced as a by-product of other nitrogen-cycling pathways (Bange et al., 2010)]. CO2, with an atmospheric residence time of ~1000 years (Lacis et al., 2010), is by far the most impactful GHG and accounts for 2.16 W/m2 of Earth's greenhouse effect (Forster et al., 2021). Nevertheless, shorter-lived CH4 [~10.5 years (Mayer et al., 1982)] and N2O [~117 years (Prather et al., 2015)] are also impactful due to their much higher sustained-flux global warming potential of 45 and 270, respectively, over a 100-year horizon (Neubauer & Megonigal, 2015). In turn, this represents an additional 0.54 and 0.21 W/m2 of radiative forcing (Forster et al., 2021). Respiration contributes the vast majority of nonanthropogenic CO2 emissions, ~50% of CH4 (Rosentreter et al., 2021) and ~25% of N2O emissions (Tian et al., 2020). Biological CO2, CH4, and N2O fluxes are, however, subject to change as Earth warms due to temperature-driven shifts in metabolic rates, the extent of anaerobic habitat, and resource competition, among other factors (Figure 3).
3.1 Warming favors CO2 production over consumption due to the temperature dependence of metabolism
Metabolic rates depend on temperature (Brown et al., 2004). This dependence can be parameterized using Q10, the proportional increase in metabolism under a 10°C warming. Theory predicts that, due to their different activation energies, respiration rates (Q10 = 2–3 at 15°C; Berggren et al., 2022) rise more rapidly with warming than production rates (1.88 at 19°C; Bissinger et al., 2008; Harris et al., 2006; López-Urrutia et al., 2006). In situ observations support these predictions (Regaudie-de-Gioux & Duarte, 2012), with mesocosm experiments suggesting a decrease in ecosystem C sequestration potential by ~13% by 2100 (Yvon-Durocher et al., 2010). The temperature dependence of metabolism should thus, theoretically, serve to amplify warming. However, several important unknowns remain. The net trophic status of Earth's diverse aquatic ecosystems is not fully resolved (e.g., the oligotrophic ocean; Duarte et al., 2013; Williams et al., 2013), nor can we effectively model how different systems of differing net trophic status will spread or retract in the coming century. Moreover, other biological factors, including acclimation and growth limitation, may dampen the realized effects of temperature on metabolism (Section 4), meaning that temperature alone is likely insufficient to predict changes in the trophic status of aquatic ecosystems.
3.2 Warming will expand anaerobic environments, controlling biogenic CH4 and N2O emissions
Methanogens and denitrifiers are largely anaerobic. The spatial/temporal extent of anoxic waters/sediments, and the supply of suitable oxidants and reductants, are thus important controls on their CH4 and N2O production rates. The volume of O2 deficient zones (ODZs) depends on the balance between O2 resupply and consumption. In non-photic littoral and deep waters, O2 is resupplied by vertical mixing and lateral advection, whereas its removal is mediated primarily by heterotrophic microbes. Rising temperatures change the extent of O2 deficiency globally by promoting stratification, altering circulation patterns, reducing O2 solubility, and increasing respiration rates (Diaz & Rosenberg, 2008; Oschlies et al., 2018). Anthropogenic nutrient (Howarth et al., 2011; Jenny et al., 2016) and organic matter (e.g., Brothers et al., 2014) loadings further exacerbate this spread in inland and coastal waters (Diaz & Rosenberg, 2008; LaBrie & Maranger, 2024). The net result has been a fourfold increase in ocean ODZ volume since the 1960s (Schmidtko et al., 2017) and an exponential increase in the number of impacted coastal water areas (Diaz & Rosenberg, 2008). The spread of ODZs will likely reinforce climate change by expanding anaerobic habitat and GHG emissions. However, the magnitude of this effect is still unpredictable (LaBrie et al., 2023) and will depend on several ecological factors.
The balance between CH4 production (methanogenesis) and consumption (methanotrophy) determines if a water body emits or consumes CH4 (Bastviken, 2009). Methanogenesis further depends on resource competition between methanogens and sulfate reducers, with sulfate reducers outcompeting methanogens for substrates like hydrogen and acetate (Oremland & Polcin, 1982). As for CO2 production/consumption, methanogenesis, methanotrophy, and sulfate reduction depend on temperature. If all other factors are held constant, rising temperatures will favor CH4 production because methanogenesis Q10 (4.1 ± 0.4, Bastviken, 2009) is greater than that of methanotrophy (2.4 ± 1.4, Thottathil et al., 2019) and sulfate reduction (1.5–3.9, Isaksen & Jorgensen, 1996). However, largescale patterns in substrate availability in ODZs may change under warming conditions, and the effects on this potentially amplifying feedback remain unconstrained.
Microbial community dynamics play perhaps an even larger role in aquatic N2O emissions by denitrifiers. Denitrification becomes energetically favorable following O2 consumption (Bange et al., 2010). Under very low O2 conditions, energetics favor partial denitrification, resulting in the production of N2O rather than N2, which is the terminal product favored under anoxia (Richardson et al., 2009). However, microbes using anaerobic ammonium oxidation (anammox) compete for similar resources, albeit at lower rates, representing about 28% of N removal in the oceans (Babbin et al., 2014). In contrast to denitrification, anammox does not produce N2O, thereby dampening the denitrification feedback loop on climate. However, low C:N ratios in dissolved/particulate organic matter favor anammox (Babbin et al., 2014), suggesting that variable phytoplankton stoichiometry may alter this feedback. How the balance between denitrification and anammox will change in the future also remains unconstrained.
4 C REDISTRIBUTION
Ecosystems can serve as both sources and sinks of GHGs. Due to the longevity of atmospheric CO2 and the ubiquity of CO2 fixation in nature, the biological redistribution of C from the atmosphere to other reservoirs profoundly influences global climate. Two examples of this redistribution in aquatic ecosystems have garnered much attention: sequestration in the deep sea via the biological C pump (see Siegel et al., 2023) and in “blue C ecosystems,”, defined here as coastal ecosystems that are net C sinks over periods >100 years (Hurd et al., 2022) and contain high vegetative structural complexity/biomass (Figure 4).
4.1 C uptake and storage potential of vegetated coastal ecosystems compromised by climate change
Much research and public attention has focused on blue C ecosystems—for example, mangroves, salt marshes, kelp forests, and seagrass meadows—due to their potential utility in climate mitigation/adaptation schemes (Gonneea et al., 2019; Perry et al., 2022; Wigand et al., 2017). Blue C ecosystems sequester more C per surface area than terrestrial systems (Mcleod et al., 2011). For example, seagrasses and kelp sequester an estimated 48–112 and 61–268 TgC, respectively, from the atmosphere annually, compared to an estimated 181 TgC year−1 sequestered by terrestrial forests (Krause-Jense and Duarte 2016; Mcleod et al., 2011). The values of C removed by seagrasses and kelp forest ecosystems are equivalent to radiative forcings of 0.36–0.84 and 0.46–2.02 mW m−2 year−1, respectively. However, global warming and related phenomena (e.g., heat waves, sea level rise, and increased flooding duration) will fundamentally alter these ecosystems in some regions and may cause increased anaerobic rhizosphere, reduced belowground biomass, plant stress and mortality, and loss of vascular plant cover, which could drive C emission rather than storage.
Climate change-induced sea level rise threatens the C storage capacity of coastal ecosystems by causing increased flooding and erosion. In salt marsh ecosystems, excess flooding oversaturates soils, resulting in loss of marsh plant cover and a corresponding increase of barren peat flats and open water (Raposa et al., 2017). Similarly, in mangrove ecosystems, sea level rise decreases tree cover, thus reducing C uptake and increasing the erosion potential of sedimentary C stores (Waycott et al., 2011). As plant mortality and erosion increases, more stored C may be released to the atmosphere or surrounding water as CH4 or CO2 (Reddy et al., 2022). Conversely, flooding of mangroves and marshes may permanently bury the peat layers facilitating long-term C storage (Mcleod et al., 2011). Ocean warming is further expected to reduce C burial by some blue C ecosystems in some regions. For example, gradual warming and prolonged heatwaves have caused poleward range contraction of kelp forests and the expansion of warm affinity herbivores in some regions (Vergés et al., 2019). These changes in kelp forest ecosystems reduce their resilience to climate change (Johnson et al., 2011; Wernberg et al., 2016) and can impact their ability to store and turnover C (Vergés et al., 2019; Peleg et al., 2020). The loss of these blue C ecosystems may exacerbate climate change, but a greater understanding of the synergistic effects of climate stressors is necessary to quantify both loss rates and climate impacts.
4.2 Biological C pump drives climate over short geologic timescales
Marine primary producers take up ~50 Pg C-CO2 year−1, a value comparable to that of terrestrial ecosystems (Field et al., 1998). Of this material, ~10 Pg C year−1, equivalent to a radiative forcing of 75.3 mW m−2 year−1 (Nowicki et al., 2022), is exported to the deep sea by a network of ecosystem processes described collectively as the biological C pump. These processes include aggregation and sinking (gravitational flux) and active transport via animal migration, which contribute about 60% and 40% of total particle export, respectively (Boyd et al., 2019; Ducklow et al., 2001; Le Moigne, 2019). Once in the deep sea, this C can remain out of contact with the atmosphere for decades to centuries (Nowicki et al., 2022), although large uncertainties exist regarding the fate of the dissolved material (Box 2). Consequently, the biological pump is thought to control atmospheric CO2 on short geologic timescales (Sarmiento & Gruber, 2002) and to be a major cause of shifts in the atmospheric CO2 reservoir associated with glacial/interglacial cycles (Gruber, 2004; Sigman & Boyle, 2000). The impact of the biological pump depends on the myriad factors modulating either CO2 drawdown (primary production) or sequestration efficiency (transfer to the deep sea). Controls on these processes are complex, and many remain poorly constrained, causing significant disagreement among Earth system models on how global export flux will change by 2100; estimates currently range from −41% to +1.8% (Henson et al., 2022).
BOX 2. What drives the microbial carbon pump?
Heterotrophic microbes rely on dissolved organic matter (DOM) as their respiratory substrate (Azam et al., 1983; Pomeroy, 1974). As this material is transformed, certain classes of C compounds (e.g., carboxyl-rich alicyclic molecules) accumulate, with residence times of hundreds to thousands of years (Legendre et al., 2015). This mechanism, coined the microbial C pump (Jiao et al., 2010), sequesters ~0.2 Pg C year−1 (1.51 mW m−2 year−1) (Legendre et al., 2015); the resulting C pool is of similar magnitude to atmospheric C-CO2 (Hansell & Carlson, 2015). There are two main hypotheses to explain these long DOM residence times. First, the dilution hypothesis (Arrieta et al., 2015; Jannasch, 1967) states that molecules exist at such a low concentration that their consumption by prokaryotes is energetically unfavorable. Second, the structural recalcitrance hypothesis (Barber, 1968) states that chemical properties, likely the presence of oxygen-rich groups (Hertkorn et al., 2006; LaBrie et al., 2022), protect molecules from further biodegradation. There are currently many unresolved questions regarding the microbial C pump, including how production rates and residence times for this unused organic C will be altered by climate change. A recent modeling study suggested that this pathway will become more important in the oligotrophic ocean (Polimene et al., 2017). However, if the dilution hypothesis is the main driver of DOM preservation, global change should not impact the amount of stored C through this mechanism. The microbial C pump is a young hypothesis compared to the biological C pump, and much remains to be unraveled.
4.2.1 Significant uncertainties in CO2 drawdown
The primary productivity of a marine ecosystem sets an upper limit on the amount of C that can be exported to the deep sea or sediments. The effects of climate change on primary production, therefore, have the potential to mediate long-term dynamics via both stabilizing and amplifying feedbacks, depending on whether productivity increases or decreases, respectively. Current Earth system models yield highly variable results, with estimated shifts in net primary production (NPP) of −15 to +30% by 2100 under high CO2 emissions [RCP8.5; (Tagliabue et al., 2021)]. These discrepancies are driven both by differences among models in how growth rates and grazing pressure respond to changing conditions (Laufkötter et al., 2015; Rohr et al., 2023) and by regional uncertainties in resource availability/demand (Tagliabue et al., 2021).
Elucidating how phytoplankton community growth rates shift with changing physico-chemical conditions (e.g., temperature, pCO2, dissolved oxygen availability) will be essential to accurately predict changes in net production (for a nuanced discussion; see Martiny et al., 2022). Increases in both temperature, which directly affects metabolic rates (Brown et al., 2004; Gillooly et al., 2001), and CO2 availability (Paul & Bach, 2020; Riebesell et al., 2007) should increase photosynthetic yields. Indeed, while the Q10 of respiration is higher than that of photosynthesis (see Section 2), long-term culturing experiments suggest that thermal adaptation in phytoplankton will increase net C fixation rates under warming (Barton et al., 2020). Ocean deoxygenation, too, may enhance primary production by improving the efficiency of CO2-concentrating mechanisms and quantum yields, as observed recently in a common marine diatom (Sun et al., 2022). Together, these changes in ocean physicochemistry should enhance future CO2 drawdown; however, shifts in resource availability and grazing pressures may counteract these gains.
Increased phytoplankton growth rates under warming can be suppressed by nutrient limitation (Marañón et al., 2018), light limitation (Edwards et al., 2016), and grazing (Rohr et al., 2023). Nutrient upwelling from deep to well-lit surface waters is expected to decline at low latitudes due to thermal stratification (Rykaczewski et al., 2015). Accordingly, satellite-based estimates suggest that global NPP may have declined by 2.1% per decade from 1998 to 2015, following modeled reductions in nitrate (Gregg & Rousseaux, 2019); however, interannual variability in phytoplankton physiology, among other factors, creates significant uncertainty in calculating NPP from ocean color data and complicates its assessment through time (Behrenfeld et al., 2015). If real, reduced aeolian iron fluxes under warming may exacerbate this trend by intensifying iron-limitation of N2 fixation (Falkowski et al., 1998; Weber & Deutsch, 2014). In regions where zooplankton grazing is tightly coupled to phytoplankton growth, increased herbivory is also expected to offset gains in production at the community scale (Hunter-Cevera et al., 2016; Ribalet et al., 2015; Sherman et al., 2016), barring decoupling due to phenological shifts (Box 1). In contrast, diminished ice cover may enhance NPP at high latitudes by relieving light limitation, although the magnitude of this effect remains poorly constrained due to model discrepancies regarding nutrient limitation severity (Laufkötter et al., 2015; Vancoppenolle et al., 2013). The impact of these controls will depend, too, on adaptation at cellular and community scales. For example, the reduced cost of photosynthesis may facilitate the reallocation of energy to nutrient uptake (Van de Waal & Litchman, 2020), with potential ramifications for higher trophic levels (Box 3). Indeed, model and in situ data indicate that nutrient uptake plasticity in the subtropics may fully compensate for projected declines in 21st-century NPP (Kwon et al., 2022). Ultimately, predicting future change will require not only the assessment of changing growth rates and limiting factors but also the interplay among these variables.
BOX 3. Will stoichiometric mismatch between food and consumer promote elemental imbalances?
Rising temperatures promote increased C fixation (Toseland et al., 2013; Zvereva & Kozlov, 2006) and shifts toward more C-rich species (Woods et al., 2003; Yvon-Durocher et al., 2015, 2017) in phytoplankton communities (De Senerpont Domis et al., 2014). Such shifts can lead to stoichiometric mismatches between phytoplankton and zooplankton (Elser et al., 2000; Sterner & Hessen, 1994), limiting energy transfer to higher trophic levels. If zooplankton adapt and alter their elemental ratios, they themselves become nutritionally inadequate for their predators, potentially affecting the strength of top-down control (Declerck et al., 2015). However, many species across the food web maintain elemental homeostasis (Hessen & Anderson, 2008) and should excrete excess C. Heterotrophic bacteria, which also compete for inorganic nutrients, should be favored over phytoplankton, further enhancing the stoichiometric distortion in consumer food (Hall, 2009). Since herbivorous zooplankton stoichiometry determines nutrient recycling rates (Elser & Urabe, 1999; Sterner, 1990), it reinforces the existing stoichiometric selection pressure. If consumers are able to adapt to the stoichiometric mismatch, for example, with higher relative assimilation and/or retention of the limiting element (Frisch et al., 2014), the selection pressure is further enhanced, representing stabilizing eco-evolutionary feedback on climate (Declerck et al., 2015), as the excess C stored during photosynthesis is not released through grazer respiration. However, the adaptive potential of consumers might be limited (Teurlincx et al., 2017), potentially resulting in the crash of these populations (Diehl et al., 2022).
4.2.2 C export depends on evolving and poorly constrained food web dynamics
The proportion of C exported to the deep sea depends largely on ecosystem structure and trophic interactions. Phytoplankton community composition, for example, affects the likelihood that organic matter is aggregated and exported rather than passing through the food web (Guidi et al., 2009). Trophic processes like grazing, algal mixotrophy (i.e., mixed autotrophy/heterotrophy), viral infection, and parasitism affect the flux of organic C through the microbial loop and the classical food web, its export, and its remineralization (Brussaard, 2004; Calbet & Landry, 2004; Park et al., 2004). In a recent synthesis, Henson et al. (2022) identify zooplankton migration, particle and phytoplankton size, and temperature-dependent remineralization as key variables for understanding C export in a changing climate. Here we focus on the feedbacks and uncertainties associated with grazing only to illustrate the importance of understanding how C flows through an ecosystem.
Zooplankton grazing is the largest source of uncertainty across marine biogeochemical models (CMIP6), with grazing rates varying 70-fold depending on their formulation (Rohr et al., 2023). Zooplankton annually consumes approximately two-thirds of primary production (Steinberg & Landry, 2017); of this, 0.7–2.7 PgC year−1 is transferred to the deep sea as dissolved inorganic C due to vertical migration and bacterial degradation of sinking fecal pellets/carcasses (Pinti et al., 2023). Uncertainties in CMIP6 arise primarily due to differences in grazing dynamics, including trophic status (Rohr et al., 2023). For example, warmer waters favor higher zooplankton grazing rates relative to phytoplankton growth rates, which should favor a shift toward net heterotrophy and thus drive an amplifying feedback loop. However, the temperature dependence of grazing versus growth is dampened under oligotrophic conditions (Chen et al., 2012), the extent of which is expected to increase with ocean warming (Agusti et al., 2017), thus mitigating the potential amplifying feedback. Moreover, metabolic thermal performance curves are prime targets for selection in a warming ocean, and adaptation may “undo” the short-term effects of increased temperatures (Data S1). Additionally, increased oligotrophy will also favor mixotrophs, which directly compete with microzooplankton but increase the mean size of prey available to larger grazers; consequently, including mixotrophy in models increases global carbon export from 7.2 to 9.8 PgC year−1 (Ward & Follows, 2016), which should have a stabilizing effect on climate. This example illustrates some of the challenges associated with predicting ecological feedbacks on climate systems—the balance between primary production and grazing, and the trophic mode and sizes of grazers, may determine the fate and flux of fixed C and, thus, the direction and magnitude of this feedback on climate.
5 CONCLUSIONS
Our understanding of aquatic ecosystems remains insufficient to robustly quantify even the most well-studied ecologically driven climate feedbacks. Unraveling these loops—their directionality, magnitude, and dependencies—must be prioritized among ecologists just as the inclusion of well-constrained feedbacks must be prioritized in climate models. As outlined in Figure 1, climate change is altering the physico-chemical environment of ecosystems, with cascading effects on ecosystem structure and function. While unraveling these changes is important, so too is better constraining the reciprocal response of ecosystems. The case studies detailed above showcase the probable significance of several ecological factors, each of which has far-reaching influence. Chief among these are (1) the community composition of primary producers, including biogenic ecosystem engineers, (2) the composition/availability of nutrients/food at all ecosystem levels, and (3) the relative temperature dependence of metabolic rates. While assessment of these functions is underway across myriad ecosystems, we must prioritize quantification of their present and future impact on local resource availability, and on Earth's albedo and atmosphere.
Yet, underpinning all ecological processes and the climate feedbacks that they control is the role of adaptation. Adaptations can occur at the level of genomes (via evolution) or phenomes (via altered expression) and have the potential to initiate, amplify, and dampen feedbacks. Adaptation may alter organismal phenology (Box 1), nutritional requirements/elemental stoichiometry (Box 3; Data S1), physiology, and ecosystem interactions (e.g., chemical warfare, grazing). While Lovelock and Margulis may have envisioned a homeostatic home-world, selective pressures act on individuals within the system, not the system itself. The resiliency of a fully evolved multicellular organism is driven by strong feedbacks among its constituent actors, which were forged and reinforced by successive selection events until the good of each unit aligns with that of the whole. The interdependencies among constituents of the biosphere, in contrast, remain tenuous—adaptations that favor the individual under changing conditions are consequently more likely to destabilize the whole than in a multicellular organism. Only by elucidating feedbacks between the biosphere and geosphere can we begin to understand how the Earth system functions as a whole and forecast its future in the face of today's climate crisis.
AUTHOR CONTRIBUTIONS
Corday R. Selden: Conceptualization; writing – original draft; writing – review and editing. Richard LaBrie: Writing – original draft; writing – review and editing. Laura C. Ganley: Project administration; writing – original draft; writing – review and editing. Daniel R. Crocker: Writing – original draft; writing – review and editing. Ohad Peleg: Writing – original draft; writing – review and editing. Danielle C. Perry: Writing – original draft; writing – review and editing. Hannah G. Reich: Writing – original draft; writing – review and editing. Matthew Sasaki: Writing – original draft; writing – review and editing. Patricia S. Thibodeau: Writing – original draft; writing – review and editing. Jana Isanta-Navarro: Visualization; writing – original draft; writing – review and editing.
ACKNOWLEDGEMENTS
We wish to thank the Ecological Dissertations in the Aquatic Sciences (Eco-DAS) XIV program, organized by Paul Kemp and Krissy Remple, the National Science Foundation (NSF, award OCE-1925796), and the Association for the Sciences of Limnology and Oceanography (ASLO) for funding and fostering this collaboration. The figures for this article were created with BioRender.com. We thank Paul Falkowski for offering an external perspective on this work prior to review, and to the reviewers for their insight. This work is dedicated to James Lovelock (1919–2022), a creative and holistic thinker. Fifty years after its introduction, the Gaia hypothesis—despite its flaws—continues to inspire systems-level problem-solving among Earth scientists from diverse disciplines.
CONFLICT OF INTEREST STATEMENT
The authors declare no conflict of interest.
Open Research
DATA AVAILABILITY STATEMENT
Data sharing not applicable to this article as no datasets were generated or analyzed during the current study.
