Climate change and fish reproduction: A meta-analysis of global laboratory-based experimental evidence across species and ecosystems

Abstract

Climate change is altering fish reproduction by disrupting spawning cycles, reducing egg viability, and altering reproductive behaviours essential for population sustainability, to our knowledge, this is the first synthesis to integrate laboratory-based evidence for temperature, hypoxia, ocean acidification, and their combined exposure across reproductive performance, physiological, developmental, and molecular endpoints, and across freshwater, brackish, and marine habitats. Following PICOS and PRISMA criteria, we synthesised 779 effect sizes (740 retained after outlier exclusion) from 80 controlled laboratory studies (38,500 observations; 54 species; 28 countries). A three-level multilevel model accounting for non-independence among effect sizes drawn from the same study showed a significant overall negative effect of climate stressors on fish reproduction (Hedges' g = −0.421, 95% CI: −0.631 to −0.210, p < 0.001; I² = 94.9%, decomposed into 36.3% between-study and 58.6% within-study heterogeneity). Hypoxia was the dominant and statistically distinct driver (g = −1.145, 95% CI: −1.648 to −0.643, p < 0.0001), most severely impairing gonadal development and sperm quality. The overall warming effect did not reach significance once non-independence was accounted for (g = −0.221, 95% CI: −0.448–0.006, p = 0.057), an uncertain negative trend rather than confirmed impairment, although tropical freshwater species showed significant subgroup-specific declines in reproductive physiology and genetic parameters. Ocean acidification showed a small positive effect (g = 0.199, 95% CI: 0.056–0.343, p = 0.006) that was not robust under robust variance estimation (p = 0.133) and is interpreted with caution. Combined multi-stressor exposure showed a negative trend (g = −0.901, 95% CI: −1.992–0.190, p = 0.106) that did not reach significance, based on only three studies. A formal omnibus test confirmed that stressor categories differ significantly in their effects (QM = 18.61,df = 3, p = 0.0003), with hypoxia statistically distinct from both temperature and ocean acidification. Leave-one-out and cumulative meta-analyses confirmed the robustness of the overall finding.