Contrasting life-history responses to climate variability in eastern and western North Pacific sardine populations

Sardines (Sardinops, Sardina spp.) are small pelagic fish exhibiting large multidecadal fluctuations in abundance with significant ecological and economic impacts. They occur in temperate oceans with shallow genetic divergence across regions characterized by distinct oceanographic settings: warm western boundary currents and cool eastern boundary upwelling systems. Historical catch records, stock assessments, and paleo-records indicate repeated boom–bust cycles linked to basin-scale climate variability such as the Pacific Decadal Oscillation (PDO). Notably, sardine populations in the western and eastern North Pacific respond oppositely to decadal temperature anomalies: the Pacific subpopulation of Japanese sardine (JP; Sardinops sagax melanostictus) tends to increase during cooler periods, whereas the northern subpopulation of Pacific sardine in the California Current (CA; Sardinops sagax sagax) increases during warmer periods. The mechanisms connecting physical forcing to early life growth, survival, and recruitment remain unresolved. Since early life growth strongly influences recruitment and survival, understanding how temperature and metabolism shape growth during larval and juvenile stages is critical to explain contrasting population responses. This study investigates early life history traits, thermal and metabolic histories, and growth–temperature relationships of JP and CA sardines to elucidate mechanisms underlying their opposite responses to climate variability, with comparisons to South African sardine populations in analogous boundary current systems.

Previous work documents global-scale sardine fluctuations over centuries to millennia and correlations with climate indices such as PDO (e.g., Chavez et al.). Studies link habitat temperature to biomass, recruitment, or early life survival in both western and eastern boundary systems, yet the opposite biomass–temperature responses of JP versus CA sardines have remained puzzling. Theories emphasize temperature effects on metabolic capacity (aerobic scope) and trophodynamic mediation via prey fields and environmental stressors. South African sardines also exhibit regionally distinct growth associated with differing boundary current systems. Traditional limitations included the inability to reconstruct fine-scale environmental and physiological histories during early life, prompting the use of otolith microstructure and stable isotopes to bridge this gap.

Samples: Otoliths were collected from 420 age-0 JP sardines (10–15 cm SL) in the Oyashio region (2006–2010, 2014–2015) and 136 age-1 CA sardines (10–16 cm SL) from the Southern California Bight (1987, 1991–1998, 2005–2007). Sampling years included biomass increase periods. Daily otolith increments and δ13C, δ18O were measured from hatch to 120–150 days post hatch (dph), encompassing larval through most juvenile stages. Otolith microstructure and growth back-calculation: Daily increment widths were measured along the postrostrum axis. First daily increment was assumed at 3 dph (JP) and 8 dph (CA). Otolith radius-at-age was summed from increment widths. Standard length-at-age was back-calculated by biological intercept method, with Monte Carlo simulations (10,000 runs) to account for uncertainty in initial lengths. Where linearity of otolith size–fish length was not assured (e.g., large South African adults), otolith radii and increment widths were used directly as growth proxies. Stable isotope analyses: Otolith powder was obtained by high-precision micromilling in stage-specific age windows (JP: 0–120 dph in 7 bins; CA: 0–150 dph in 5 bins). δ18O and δ13C were analyzed using MICAL 3c and DELTA V + GAS Bench systems; analytical precisions were better than ±0.10–0.17‰ for δ18O and ±0.10–0.15‰ for δ13C. Corrections for acid fractionation and instrument temperature were applied. Estimating field metabolic proxy (M_oto): M_oto, the fraction of metabolically derived carbon in otolith carbonate, was estimated as M_oto = (δ13C_oto − δ13C_DIC) / (δ13C_diet − δ13C_DIC) + ε (ε=0). δ13C_diet ranges were derived from literature values of sardine muscle and zooplankton, considering diet–tissue enrichment (~1.5‰). δ13C_DIC was taken from observational databases. Temperature reconstruction: For CA sardine, otolith δ18O was converted to temperature using δ18O_otolith = δ18O_seawater − 0.18×T + 2.69 with fixed δ18O_seawater = −0.32‰ (±0.12‰ variability implying <0.7 °C error). For JP sardine, due to larger δ18O_seawater variability, temperature–salinity relationships were derived monthly using Argo float data (surface layer, 130–180°E, 30–45°N) and a regional δ18O_seawater–salinity regression, then fitted with quadratic functions to convert otolith δ18O to temperature (average RMSE ~1.0 °C). Environmental indices: Year-class mean experienced temperatures during early days post-hatch were compared with PDO (March–May for JP; April for CA) and with the Coastal Upwelling Transport Index (CUTI; 33–36°N, April) for CA. SST–survival analyses used satellite SST (OSTIA) and log recruitment residuals from Ricker models to assess basin-scale correlations. Statistical analyses: Linear mixed-effects models tested differences in increment widths (age×region, random fish ID), M_oto, and experienced temperatures between JP and CA across life stages. Relationships between M_oto and temperature used generalized linear models with interactions among temperature, region, and life stage. Optimal temperature was inferred as the temperature maximizing the gap between modeled 95th and 5th percentile M_oto across 1 °C bins (proxy for aerobic scope). Growth–temperature relationships employed Pearson correlations and quadratic models where appropriate, with Benjamini–Hochberg corrections. Comparisons to South African sardines used previously published otolith data (up to 100 dph).

- Early life growth and metabolism differ between JP (western boundary) and CA (eastern boundary) sardines: - Otolith radii were larger in JP than CA throughout 10–150 dph; JP resembled south-east South Africa sardines, whereas CA resembled west coast South Africa sardines up to ~100 dph. - M_oto peaked at 45–60 dph in JP and 60–90 dph in CA; JP had significantly higher M_oto during 0–120 dph. - Experienced temperatures: JP declined from ~19 °C to ~16 °C over the first 120 d; CA remained ~15–16 °C over 150 d. JP temperatures were significantly higher during 0–90 dph. - Environmental controls captured by otolith proxies: - JP experienced temperature (0–60 dph) correlated negatively with PDO (Mar–May): r = −0.80, p = 0.03, n = 7. - CA experienced temperature (0–30 dph) correlated negatively with CUTI (April, 33–36°N): r = −0.67, p = 0.02, n = 11. - Metabolic response and optimal temperatures: - M_oto increased with temperature in all stages, indicating higher metabolic rates at warmer temperatures. - After accounting for temperature, JP had higher M_oto in early and late juvenile stages than CA, with gentler temperature-dependent slopes for JP in these stages. - Inferred optimal temperature (from widest M_oto range/aerobic scope) differed: JP >20 °C in larvae, declining to ~15 °C by late juvenile; CA ~14–15 °C across stages. - Growth–temperature relationships: - Larval stage: positive length–temperature correlations in both JP (r = 0.46, pc = 0.045, n = 26) and CA (r = 0.42, p = 0.046, n = 21). - Early juvenile: JP showed dome-shaped relationship at 75 dph (quadratic model significant); CA positive (r = 0.47, p = 0.045, n = 21). - Late juvenile: JP negative (r = −0.58, pc = 0.011, n = 26 at 105 dph); CA positive (r = 0.43, p = 0.046, n = 21 at 120 dph). - Temperatures of highest median growth for JP declined from >20 °C (larvae) to ~15 °C (late juveniles), matching the shift in optimal temperature; CA remained ~14–16 °C. - Cross-system consistency: Patterns in South African sardines mirror JP/CA contrasts, supporting generality across western (warmer) vs eastern (cooler) boundary systems. - Basin-scale SST–survival patterns (supplementary): Negative correlations in western and positive in eastern North Pacific; regional mean SSTs correlate with PDO or CUTI over longer scales.

The study addresses why western and eastern North Pacific sardine populations respond oppositely to climate variability. Otolith-derived thermal and metabolic histories reveal that JP and CA sardines differ in metabolic capacity (M_oto), thermal optima, and growth–temperature relationships across ontogeny. For JP (western boundary systems), cooler regimes likely expand prey-rich, less stressful habitats and align with a declining optimal temperature through juvenile stages, widening aerobic scope and facilitating higher growth and survival under cooler conditions. For CA (eastern boundary upwelling system), warmer regimes may reduce environmental stressors (e.g., turbulence, low oxygen, feeding constraints) commonly associated with strong upwelling and small-plankton-dominated food webs, enabling greater utilization of aerobic scope and higher growth under warmer conditions. These findings synthesize and reconcile two frameworks—the optimal temperature hypothesis and the trophodynamics hypothesis—by showing that temperature sets metabolic capacity while prey availability and habitat stressors regulate its utilization, jointly determining growth and recruitment prospects. Cross-system comparisons with South African sardines suggest these mechanisms are general features of sardines inhabiting contrasting boundary current systems.

This work combines otolith microstructure with high-resolution δ18O and δ13C analyses to reconstruct temperature and metabolic histories of early life stages of sardines, demonstrating distinct thermal optima, metabolic intensities, and growth–temperature relationships between JP and CA populations. The results offer a mechanistic explanation for their opposite biomass responses to climate variability and show similar patterns in South African boundary current systems. Under projected climate change, growth of JP sardine may decline with warming in the Kuroshio–Oyashio region, whereas CA sardine growth may increase with warming and potential enhancements in nutrient supply and productivity in the California Current. Future research should integrate later life stages, phenology, spatial distributions, maternal condition, and fishing pressure, and expand isotopic reconstructions across more populations and years to refine projections of population dynamics under changing climates.

- Small sample sizes for some interannual correlations (e.g., n = 7 for JP PDO relation; n = 11 for CA CUTI), limiting statistical power and generalizability. - Estimation of M_oto depends on literature-based δ13C_diet ranges, assumed ε = 0, and δ13C_DIC variability; although sensitivity analyses suggest limited error, uncertainties remain. - Back-calculation of length assumes linear otolith–body size relations and constant intercepts within individuals; physiological or environmental variability can violate these assumptions. - CA otolith increments were difficult to count to the edge due to winter growth slowdown; stage binning and resolution differed between regions. - Temperature reconstructions for JP rely on monthly temperature–salinity relationships derived from Argo and regional δ18O_seawater–salinity regressions, introducing model error (average RMSE ~1.0 °C) and assuming coherent T–S structure. - Use of regional mean SST as a proxy for habitat temperature and survival introduces mismatch with individual otolith-derived temperatures due to patchy distributions and variable dispersal. - Optimal temperature inference via M_oto range (95th–5th percentile gap) is an indirect proxy for aerobic scope and assumes adequate sampling across temperature bins; bins with sparse data were excluded.