Anthropogenic activities are stressing ocean ecosystems through pollution, overfishing, warming, deoxygenation, increased stratification, and acidification. The mesopelagic ('twilight') zone (TZ), spanning 200–1000 m depth and comprising about a quarter of the ocean's volume, has been considered relatively insulated from these changes due to the time lag in propagating surface perturbations downwards. However, this overlooks the near-instantaneous impact of changes in marine productivity and organic matter export from the surface, crucial for TZ organisms reliant on sinking particulate organic matter via the Biological Carbon Pump (BCP). Warming has already reached TZ depths, affecting both surface ecosystems and the BCP's operation throughout the upper water column. The ecological impact of ocean warming on the delivery of organic matter to subsurface habitats has been largely neglected. The BCP involves phytoplankton removing dissolved inorganic carbon (DIC) and nutrients in the photic zone (primary production), followed by the gravitational settling of particulate organic matter (POM) to the ocean interior. As POM sinks, microbes and other organisms degrade it through remineralization, converting it back into dissolved inorganic constituents. Only 0.5–2% of the original surface export flux reaches the ocean floor. Ocean mixing and upwelling return DIC and nutrients to the surface, completing the cycle. The BCP governs deep ocean carbon storage, nutrient release proximity to the surface, and consequently, feedback on surface productivity, atmospheric CO2, and global climate. Temperature significantly affects metabolic activity and the BCP. A 10 °C temperature increase roughly doubles metabolic rates. A cool ocean interior preserves sinking particles, facilitating efficient organic matter transfer to depth. Warming reduces this efficiency, decreasing food availability at depth. This is particularly crucial in the TZ, where the majority of sinking organic matter is processed and transformed biochemically. The TZ is also a major reservoir of biodiversity and biomass, supporting various organisms, including diving predators and diel vertical migrants. Despite its importance, TZ ecology and its response to warming remain poorly understood. The geological past offers a valuable perspective on these questions.

While most TZ organisms have a poor fossil record, planktonic foraminifera are an exception. These microscopic protists inhabit depth-stratified habitats, secreting calcium carbonate shells that accumulate on the seafloor. Their shells reveal information about their life environment over millennia. Deep-sea core drilling provides a valuable archive for investigating TZ history, showing that both biotic and abiotic factors have influenced foraminifera evolution over millions of years. The vertical and horizontal distribution of species has undergone significant changes linked to long-term climate trends, transitions, and transient warming events. Global numerical modeling of past climates and ocean states helps disentangle the dominant drivers of these changes. This study leverages past observations and modeling to infer future changes, focusing on low-latitude TZ habitats because high-latitude oceans are affected by mixing processes obscuring temperature effects and possess a less complete paleo record. The study contrasts past, present, and future ocean conditions, discussing the applicability of past steady-state conditions to a dynamically changing present and future.

The study examines two periods: the early Eocene (extreme warmth) and the mid-Miocene to present (global cooling). The early Eocene exhibited significantly higher surface temperatures than present, with largely ice-free poles. Estimated atmospheric CO2 concentrations (600–2500 ppm) suggest CO2 was the primary driver of long-term climate change during the Cenozoic. The warmest period, the Early Eocene Climatic Optimum (EECO), featured ocean bottom water temperatures of 10–12 °C (compared to 2 °C today) and global mean sea surface temperatures (SSTs) 10–16 °C warmer than preindustrial levels. High-latitude temperatures were disproportionately elevated, resulting in reduced temperature gradients. Olivarez Lyle and Lyle noted a striking lack of organic carbon in Eocene sediments despite high biological productivity, suggesting that elevated temperatures increased microbial metabolic rates, accelerating the respiration of sinking particulate organic carbon (POC) and leaving less to be buried. John et al. supported this hypothesis using δ¹³C-depth profiles, showing sharper-than-modern gradients consistent with more efficient POC recycling and shallower remineralization depths. The early Eocene pelagic ecosystem was concentrated near the surface, suggesting TZ habitats were inhospitable due to reduced food supply and an oxygen minimum zone (OMZ). Conversely, cooling after the EECO led to the opening of deeper niches and increased biodiversity. The mid-Miocene differed from the early Eocene by the presence of a continental-scale Antarctic ice sheet. The Miocene Climatic Optimum (MCO) was warmer than today, and associated with a lower meridional temperature gradient. Deep-sea sediment records from the Miocene are more abundant, offering a more complete picture of the BCP and TZ ecology. Deep-dwelling planktonic foraminifera were rare in the mid-Miocene, with a possible shift to deeper habitats around 13 Ma associated with the Middle Miocene Climatic Transition (MMCT). Cooling from the mid-Miocene onwards saw an increase in the number and abundance of TZ species, with some species migrating from the euphotic zone to the TZ and others moving deeper. The late Miocene and Pliocene witnessed further diversification. This diversification, along with increased depth distribution of foraminifera, indicates a more efficient BCP operating in cooler oceans. A greater difference between surface and benthic δ¹³C in foraminifera tests and a weakening of near-surface δ¹³C-depth gradients further support this. To evaluate the temperature-dependence of the past BCP, the study used the cGENIE Earth system model, calibrated against modern and paleo observations. The model incorporates temperature-dependency of nutrient uptake rates at the surface and the rate of sinking POM remineralization. While the model reasonably reproduces large-scale ocean circulation, it has low resolution (10° longitude) and lacks a coupled dynamical atmosphere. The modeled early Eocene climate showed a global mean SST of 27 °C, a mid-Miocene SST of around 24 °C, and a preindustrial SST of 19 °C. Corresponding benthic temperatures were 10 °C, 4.6 °C, and 1.5 °C, respectively. Although the model doesn't perfectly reproduce meridional temperature gradients, simulated low-latitude surface and deep ocean temperatures generally agree with paleotemperature data. Ocean warming in the model increased metabolic rates, accelerating remineralization and shoaling the depth of organic matter remineralization. While global export increased during the early Eocene, BCP efficiency in transferring organic matter to the deep was reduced. This is because reduced transfer efficiency dominated over higher POC export, leading to a weaker BCP and lower POC fluxes at depth. At low latitudes, the warmest (early Eocene) simulation showed the lowest POC delivery to TZ depths, the shallowest and most severe OMZ, and the sharpest near-surface δ¹³C gradients. The model's δ¹³C-depth profiles agreed with those reconstructed from offshore Tanzania. For the mid-Miocene, the simulated warmer-than-modern temperatures resulted in reduced POC delivery and a shoaled OMZ, but less severely than for the early Eocene. Modeled δ¹³C-depth profiles agreed with those reconstructed from planktonic foraminifer data. The geological record demonstrates that carbon cycling and deep-ocean communities differed in warmer oceans, a consequence of the temperature-dependence of the BCP.

The study's key findings demonstrate a strong relationship between ocean temperature and the efficiency of the biological carbon pump (BCP), impacting the ocean's twilight zone (TZ). Analysis of past warm periods (Early Eocene and mid-Miocene) revealed that warmer temperatures led to a less efficient BCP, resulting in less organic matter reaching the TZ and a shallower, more intense oxygen minimum zone (OMZ). This resulted in decreased abundance and diversity of TZ plankton communities, which were concentrated closer to the surface in warmer climates. Using an Earth system model (cGENIE), the researchers projected the effects of anthropogenic warming on the TZ. Even under a low-emission scenario, the model predicts a significant (more than 20%) decrease in food supply to the mid-TZ by 2100, with more severe reductions under mid-range and high-emission scenarios. The model also showed a significant reduction in oxygen levels in the TZ due to increased metabolic rates at warmer temperatures and the decrease in transfer efficiency of the BCP. An empirical model was created based on the relationship between planktonic foraminifera abundance (as a proxy for overall TZ community) and modeled POC fluxes over the mid-Miocene to present. This empirical model confirmed the predictions from the cGENIE simulations, showing a decline in abundance in the mid-TZ in response to decreasing POC delivery and increasing local water temperature. The decline in abundance is predicted to occur at a faster rate and be more significant under higher emission scenarios. Under the high emissions scenario, this could result in up to a 70% decline by 2200. The recovery of the TZ ecosystem is predicted to take several thousand years, even after atmospheric CO2 levels decline. While the model implicitly accounts for other environmental factors, it is limited by its reliance on food supply as the primary factor, neglecting the complex interactions within the ecosystem and other effects such as ocean acidification.

The findings highlight the critical role of food supply changes in shaping future TZ habitats. The projected decreases in food supply, even under a low-emission scenario, underscore the potential for widespread ecological disruption in the TZ. While motile species might adapt by migrating, the combined effects of temperature, oxygen, and food supply changes could create novel and challenging environmental conditions. Past TZ ecosystems evolved gradually in response to slow environmental changes, but the rapid nature of future anthropogenic changes may limit such adaptations. The empirical model provides a best-case scenario, illustrating how steady-state geological analogs differ from the transient nature of anthropogenic change. The study emphasizes temperature as a key determinant of food and oxygen distribution with depth, affecting species abundance and diversity in the TZ. This relationship likely holds across Earth's history, as suggested by studies linking evolving ocean temperature and organic matter flux to marine sediment composition and carbon isotopic composition over at least half a billion years. A comprehensive understanding of temperature-dependent processes related to the BCP is essential for understanding the drivers of evolutionary and extinction events in the geological record.

This study demonstrates a critical link between ocean temperature, the biological carbon pump, and the health of the ocean's twilight zone. Past warm periods serve as analogs for predicting future impacts, revealing that even moderate warming can lead to significant declines in food supply and oxygen in the TZ. The research emphasizes the urgent need for strong emissions mitigation to prevent widespread and potentially irreversible ecological damage. Further research should focus on refining empirical models and incorporating additional factors, such as ocean acidification and changes in the intensity of vertical mixing in the water column to create more robust and accurate future projections of ecosystem health in the TZ.

The study's reliance on the cGENIE Earth system model, which has limitations in resolution and the representation of some processes, represents a significant limitation. Additionally, the empirical model simplified the complex interactions within the TZ ecosystem, focusing primarily on the relationship between food supply and plankton abundance. The model does not account for potential extinction events or other factors, such as changes in ocean chemistry and the combined effects of temperature, food supply, and oxygen. Finally, the low spatial resolution of the model and the lack of inter-annual variability mean that the future projections should be considered as broad illustrative scenarios rather than robust projections.