Enhanced fish production during a period of extreme global warmth

Fish are key upper-trophic-level consumers and a major protein source globally. Climate-driven changes in primary production are expected to cascade to fish biomass, productivity, community composition, and size structure. Earth system models and fisheries data indicate that anthropogenic ocean warming intensifies stratification, reduces nutrient supply, decreases primary production, and thereby reduces fish productivity—especially in low-latitude subtropical systems. However, geological records show mixed relationships between temperature and marine productivity over long timescales, and the sedimentary record primarily preserves a limited subset of organisms, leaving quantitative links between primary production and higher trophic levels poorly constrained. Beyond changes in primary production, temperature can alter food-web process efficiencies (e.g., metabolic rates, grazing, predator attack rates), potentially producing positive or negative impacts on secondary production. This study asks how pelagic fish production responded to long-term, extreme global warmth in the Early Paleogene, and which ecological mechanisms (primary production, trophic transfer efficiency, predator–prey size structure) best explain the observed patterns.

Modeling and observational studies predict declines in marine primary and higher trophic-level production with anthropogenic warming due to increased stratification and reduced nutrient supply. Recent decades show declines in fish productivity consistent with these predictions, with subtropical ecosystems being most affected. Geological records, however, provide conflicting evidence on the relationship between temperature and marine productivity over long timescales, with some proxies showing positive, negative, or no correlation. Physiological theory suggests higher temperatures increase metabolic costs and may reduce body sizes and trophic energy transfer, yet empirical observations and experiments indicate that warming can increase grazing rates, predator attack rates, and foraging success. Phytoplankton niche models suggest picophytoplankton may respond positively to warming in subtropical pelagic regions. Overall, the literature highlights multiple, potentially opposing mechanisms by which temperature affects energy flow and trophic structure, motivating a data-driven assessment using fossil records and simple ecological modeling.

Data: The study analyzed isolated fish microfossils (ichthyoliths: teeth and scales) from DSDP Site 596 (South Pacific gyre). Sediment samples were wet-sieved (38 μm), ichthyoliths >106 μm were picked, mounted, imaged, and measured (ImageJ; calibrated with an ocular micrometer). Ichthyolith accumulation rate (IAR; ichthyoliths/cm²/Myr) was computed using the site age model. The interval 62–46 Ma was chosen to follow post-K–Pg fish evolution and avoid ice-volume effects in δ18O temperature estimates. Data were grouped into 1 Myr bins (average 4 measurements/bin; intra-sample SD 13.8%). The area under the per-Myr tooth-size density curve (total IAR) was used as a proxy for total fish community production. Temperature comparison used an independent benthic δ18O-based bottom-water temperature reconstruction to reflect whole-ocean climate relevant to mesopelagic fish.

Statistical analysis: IAR time series were compared to reconstructed temperature. A quadratic regression of IAR vs. temperature quantified the relationship and was tested against a linear model (F-test).

Food-web model: A minimal size-structured trophic transfer model was constructed. Key elements: (1) Allometric scaling between tooth size and body size; (2) Primary production represented by a size distribution; (3) Trophic transfer across discrete trophic levels using Gaussian predator–prey size preference kernels (parameters for mean predator–prey size ratio and variance) and a fractional trophic transfer efficiency; (4) Propagation of size-specific primary productivity to higher trophic levels to predict an upper-trophic productivity size distribution comparable to the observed tooth-size distribution (assuming constant tooth-to-biomass conversion and fish fraction of the predator community through time). Model complexity was constrained by the trophically aggregated ichthyolith data.

Model fitting and experiments: A reference model with time-constant parameters was first calibrated by least squares to minimize the misfit between observed and modeled tooth-size distributions across the record. Then, in separate experiments, individual parameters (trophic transfer efficiency, total primary productivity magnitude, mean predator–prey size ratio, predator–prey size ratio variance) were allowed to vary independently by 1-Myr bin to assess which time-varying mechanism best reproduces the observed temporal dynamics. Sensitivity tests limited per-Myr parameter changes to 5–100%. An additional model enforced a positive linear scaling between primary productivity magnitude and phytoplankton mean cell size. Model performance was evaluated using RMSE across size distributions. Calculations were performed in R 3.6.1.

• Ichthyolith accumulation rate (IAR), a proxy for pelagic fish production, increased nearly tenfold during peak Early Eocene Climate Optimum (EECO) warmth, reaching ~300 ich/cm²/Myr around ~50 Ma from lows near ~30 ich/cm²/Myr at ~60 Ma.
• IAR was strongly and positively related to reconstructed bottom-water temperature over 62–46 Ma; a quadratic regression explained 75% of IAR variance, with the quadratic term significantly improving fit over a linear model (F-test p < 0.05), indicating a positive nonlinear temperature–production relationship.
• The mean and standard deviation of ichthyolith size distributions varied modestly (~5% and ~20%, respectively) without significant temporal trends or temperature relationships, implying changes were driven by production magnitude rather than size structure shifts.
• Data-constrained modeling showed that time variation in energy-flow mechanisms reproduced observations best: • Allowing trophic transfer efficiency to vary through time reduced model RMSE by a factor of 13.7 relative to a time-constant reference. • Allowing total primary productivity to vary reduced RMSE by a factor of 13.6. • Predator–prey size structure changes were much less effective: time-varying mean predator–prey size ratio and its variance reduced RMSE by factors of 2.4 and 3.7, respectively.
• To match observations, time-varying primary productivity required up to 100% change per Myr, whereas trophic transfer efficiency required only ~10% change per Myr, indicating small efficiency shifts can strongly impact fish production.
• Enforcing positive size–productivity scaling (larger phytoplankton with higher productivity) degraded fits, increasing optimized RMSE by ~3× compared to the time-varying productivity model without size–productivity scaling, and predicted larger fish inconsistent with the ichthyolith record.
• Results indicate increased trophic transfer efficiency and, secondarily, increased primary production drove enhanced fish production during EECO warmth, while predator–prey structural changes did not.

The ichthyolith record from a subtropical South Pacific gyre site reveals a strong positive, nonlinear relationship between ocean temperature and pelagic fish production over million-year timescales during the Early Paleogene. This contrasts with short-term, anthropogenic-scale projections and observations that generally show declines in primary and fish production with warming due to stratification and nutrient limitation. Modeling constrained by the fossil size-distribution data indicates that mechanisms increasing total energy flow—particularly higher trophic transfer efficiency, and to a lesser extent primary production—best explain the observed production increases, whereas changes in predator–prey size structure cannot. Potential processes consistent with higher trophic transfer efficiency at warmer temperatures include increased grazing/attack rates and enhanced microbial loop activity that channels dissolved organics to higher trophic levels, though these would need to offset increased metabolic costs. The findings emphasize temporal-scale dependence: gradual, multi-million-year warming may allow ecosystems to adjust, potentially enhancing biomass transfer without the rapid reorganization and extinctions associated with abrupt anthropogenic perturbations. These insights refine our understanding of upper-trophic-level responses to climate over geological times and provide a framework for integrating ecological theory with paleontological evidence.

This study provides evidence that during a period of extreme global warmth (EECO), subtropical pelagic fish production increased markedly and exhibited a positive nonlinear relationship with ocean temperature. A simple, data-constrained, size-structured trophic model shows that modest increases in trophic transfer efficiency—and, secondarily, increases in primary production—can account for the fossil record, whereas alterations in predator–prey size structure cannot. The work bridges ecological modeling with the ichthyolith fossil record to reveal upper-trophic-level dynamics over long timescales, challenging the assumption that warming universally depresses fish productivity. Future research should test these mechanisms across additional sites and oceanographic regimes, better constrain the physiological and ecological drivers of trophic efficiency under warming, quantify tooth-to-biomass relationships, and integrate more detailed, trophically resolved models as paleontological data allow.

• Spatial scope: Analyses are from a single subtropical South Pacific gyre site (DSDP Site 596); results may not generalize globally.
• Proxy constraints: IAR is an indirect proxy for fish community production and relies on assumptions of constant tooth-to-biomass conversion and stable representation of fish within the broader predator community through time.
• Data aggregation: The ichthyolith record is trophically aggregated and limits model complexity; multi-parameter time-varying fits are underdetermined.
• Model assumptions: Simplified size-structured trophic model with Gaussian predator–prey kernels and constant allometric relationships; unknown trade-offs among parameters (e.g., tooth–body allometry vs. phytoplankton size distribution) limit inference of true underlying size spectra.
• Temporal resolution and sampling: 1 Myr binning with limited samples per bin (mean ~4) and intra-sample variability; smoothing choices may affect fits.
• Temperature reconstruction: Use of benthic δ18O-based bottom-water temperatures as a climate proxy may introduce uncertainties relative to local surface conditions, though justified for mesopelagic fish.
• Mechanistic ambiguity: While trophic transfer efficiency emerges as parsimonious, specific physiological or community-level mechanisms driving efficiency changes over geological times remain unresolved.