Life histories determine divergent population trends for fishes under climate warming

Ocean warming, alongside acidification and expanding hypoxia, is expected to strongly affect marine fauna, particularly in warm regions due to oxygen-related metabolic constraints. Anticipated outcomes include poleward shifts and reduced catches in subtropical and tropical regions, with additional impacts on population demography and abundance. Yet demographic impacts of warming across marine fishes remain mixed and mechanisms underlying differential population responses are poorly understood. Temperature drives within-species variation in life-history traits—warmer environments often yield smaller body sizes, higher mortality, faster growth, and earlier maturation. Because life-history traits shape population growth and resilience, temperature-induced life-history changes likely affect population sustainability. Prior work often evaluated single traits (e.g., size declines), but trade-offs among traits mean population-level consequences require a multi-trait perspective. To address this gap, the authors compiled life-history and temperature data for Indo-Pacific fishes across 55° S to 65° N, a region including warming hotspots, to evaluate within-species life-history responses to temperature and the resulting implications for population growth across habitat-related groups.

The study builds on evidence that climate change affects marine organisms via thermal and oxygen constraints and that warming can cause range shifts and altered demography. Prior research documents temperature-mediated changes in individual traits (growth, maturation timing, mortality) and size-temperature relationships, often in single-species or limited taxonomic scopes (e.g., Atlantic cod, tunas, reef fishes). The importance of life-history traits for resilience and fisheries reference points is established, but most assessments assume uniform warming effects and focus on single traits like body size. Recent work emphasizes natural mortality and fecundity in determining population recovery and growth. The authors position their work to generalize temperature–life history relationships across many species and traits, and to integrate multiple traits to infer population-level consequences under warming.

Data compilation: The authors screened ~8000 references (1958–2017) and extracted population-specific life-history data from 440 sources (peer-reviewed papers, assessments, theses, reports) for 1402 population records, covering 332 species (83 families) in the Indo-Pacific. For 1268 population records (321 species), they matched temperature data. Extracted traits included von Bertalanffy growth coefficient K (yr−1), asymptotic length L∞ (cm), age and length at 50% maturity (A50 in years, L50 in cm), length–weight exponent b, lifespan Amax (years), and natural mortality M (yr−1). Length measures were standardized to total length. When both combined- and sex-specific data were available, combined-sex data were prioritized; otherwise female data were used. Temperature data: Long-term mean decadal sea temperatures were taken from NOAA World Ocean Atlas 2013 (WOA13) with 0.25° resolution and depth profiles to 5500 m. Two habitat temperature variables were derived per population: sea surface temperature (SST, 0 m) and bottom temperature (BT, at the maximum population depth). For each, minimum, mean, maximum, and coefficient of variation were computed over spatial extents reflecting sampling locations and depth ranges. Natural mortality estimation: To ensure comparability, M was re-estimated for each population using Gislason et al. (2010) Model II: ln(M) = 0.55 − 1.61 ln(L) + 1.44 ln(L∞) + ln(K), where L is the midpoint of the length range. Because length-range data were unavailable, L was taken as length at A50, leveraging invariant relationships among maturation, L50, L∞, and K. Missing A50 values (available for only 119 populations) were restored using a theoretical linear relationship between A50 and L∞ (calibrated with 70 populations), approximating L50 as 2/3 L∞, and then computing L from the von Bertalanffy growth curve. Outliers with M > 25 yr−1 (15 populations) were excluded. Statistical modeling of temperature effects: Linear mixed-effects models (LMEs; R packages lme4, lmetest) related ln-transformed life-history traits to species mean-centered temperature metrics (SST and BT; min/mean/max/CV), with species and family as random effects (random intercepts and/or slopes). Likelihood ratio tests guided model parsimony. For group-wise analyses, six FishBase habitat/phylogeny-associated groups (reef, demersal, pelagic, benthopelagic, bathydemersal, elasmobranch) were analyzed with mean-centered SST as fixed effect and ln-traits as responses; slopes (β1) and intercepts (β0) with 95% CIs were estimated. Principal component analysis (PCA) summarized intercepts and slopes for K and L∞ across groups. Life-table modeling for population growth sensitivity: An age-structured life-table integrated growth, maturation, fecundity, and survivorship to estimate annual population growth sensitivity to a 1 °C SST increase. Length-at-age Lt followed von Bertalanffy growth (L∞, K, t0=0). Weight-at-age Wt = αLt^β (α=0.02, β=3). Fecundity mt = γWt^δ (γ=1.18, δ=2.93), with mt=0 for t < A50 and positive thereafter. Survivorship lt = exp(−Σ Mt), where age-specific M was derived from the Gislason model substituting Lt for L. Lifetime reproduction R0 = Σ lt mt over ages 0–50 years (fixed maximum age). Temperature sensitivities of K, L∞, and A50 (from LME slopes) were applied to project trait changes under +1 °C, and the log2 ratio of annual population growth (+1 °C vs baseline) quantified sensitivity. Two datasets were used: (i) 100 populations with empirical K, L∞, and A50; (ii) 1265 populations with empirical K and L∞ and model-derived A50.

- Across all species, higher mean SST was associated with faster life histories: increased K and M, decreased L∞, and earlier A50. Fixed-effect slopes: K = 0.05 yr−1 °C−1 (P = 0.001); M = 0.05 yr−1 °C−1 (P = 0.001); L∞ = −0.02 cm °C−1 (P < 0.001); A50 = −0.04 yr °C−1 (P = 0.03). No significant effects were found for b, L50, or Amax, likely due to fewer observations. - Temperature effects were approximately linear over temperature anomalies of about ±5 °C for most traits, suggesting populations were within thermal tolerance ranges. - Species and family effects were significant, indicating heterogeneous species responses and phylogenetic structure in life-history–temperature relationships. - Group differences: All six habitat/phylogeny groups showed consistent effect directions (positive for K and M, negative for L∞), but sensitivities varied. Reef and demersal fishes had stronger temperature sensitivities; pelagic and bathydemersal fishes showed weaker or more variable sensitivities. Elasmobranchs had larger L∞ intercepts and smaller reductions in body size with temperature than other groups. Pelagic fishes were centrally located in PCA space, indicating low sensitivity of K and L∞ to temperature. - Life-table projections under +1 °C warming showed divergent population growth responses along the fast–slow life-history continuum. With empirical A50 (n=100), most populations had negative sensitivity (log2 ratio < 0) except a few with short generation times. With modeled A50 (n=1265), sensitivities were variable but predominantly positive overall; 41% (n=519) of populations showed negative sensitivity. Negative sensitivity was more common in short-generation (fast life-history) populations. - Among groups, pelagic fishes had the highest proportion of populations with negative sensitivity (60%). Slow life-history groups (elasmobranch, demersal, bathydemersal) more often showed positive sensitivities, especially at longer generation times. Overall within the dataset, only about 25–30% of slow life-history fishes showed declines in population growth under +1 °C, whereas 42–60% of fast life-history fishes did.

Warming in the Indo-Pacific is projected to accelerate fish life histories by increasing growth rates (K) and natural mortality (M), while decreasing asymptotic size (L∞) and age at maturation (A50). However, these trait changes translate into divergent population growth responses depending on position along the fast–slow life-history continuum. Slow life-history species (long generation times) are predicted to benefit overall through increased early-life fecundity and reduced mortality costs relative to later-life losses. Fast life-history species (short generation times) are predicted to experience reduced population growth due to fewer early-life gains but persistent later-life costs. These patterns vary among habitat/phylogeny groups: elasmobranch, demersal, and bathydemersal fishes tend to benefit; benthopelagic and reef fishes show mixed outcomes; pelagic fishes are most likely to be negatively affected. The findings align partially with previous observations (e.g., reduced sustainability of fast life-history populations under stress, size changes differing by species) but contrast with some interspecific results (e.g., tunas). The predictions for many pelagic fishes remain to be empirically verified. The study underscores the importance of considering multiple life-history traits (not just body size) and highlights the roles of natural mortality and fecundity in shaping population growth. It also challenges assumptions of uniform warming effects across species used in fisheries impact assessments, advocating for differential life-history responses in forecasting fisheries production under climate change. While mobility may buffer some populations from fixed-location warming and uncertainties remain due to data limitations, the life-table approach provides a tractable proxy for temperature sensitivity in data-poor contexts.

This study synthesizes extensive Indo-Pacific fish life-history data with environmental temperatures to show that warming generally accelerates life histories but yields divergent population growth outcomes: likely benefits for slow life-history species and risks for fast life-history species, especially pelagic fishes. By integrating multiple traits within a life-table framework, the work advances beyond single-trait assessments and provides evidence that differential life-history sensitivities should be incorporated into fisheries and climate impact models. Future research should: (1) empirically test predicted declines in pelagic populations and verify group-specific sensitivities; (2) jointly evaluate positive and negative temperature effects across life stages and traits; (3) incorporate life-history evolution and plasticity; (4) refine recruitment and early-life process data to link trait changes to realized population growth; and (5) integrate movement and distributional shifts with demographic responses for more accurate projections.

- Recruitment processes and early-life histories were not modeled explicitly due to data limitations; the life-table approach uses R0-based proxies rather than realized recruitment and growth rate r. - Age at 50% maturity (A50) was missing for most populations and was modeled from relationships with L∞ and K; errors in A50 estimates propagate to M and fecundity schedules. - Natural mortality M was re-estimated via an empirical invariant model (Gislason Model II), which may not capture species-specific mortality processes; some outliers were removed. - Several model constants were assumed (t0=0; α, β for length–weight; γ, δ for fecundity–weight), and maximum age was fixed at 50 years for all populations, potentially biasing life-table outputs. - SST was the primary driver; BT effects were weaker, and thermal exposure mismatches may exist for demersal/bathydemersal species. - Temperature inputs were long-term means from WOA13; temporal mismatches with life-history sampling and local variability were not accounted for. - Trait coverage was uneven; fewer observations for b, L50, and Amax limited detection of temperature effects for these traits. - Species mobility and potential range shifts were not incorporated; predicted sensitivities are proxies for fixed-location warming exposure rather than actual future trajectories.