The decline effect is generally regarded as a measure of poor reproducibility of scientific results, where an initially exciting discovery with large effects become increasingly difficult to reproduce over the years. While the detection of the decline effect is often considered a setback or even a reproducibility crisis, it has been widely reported in various disciplines, including marine science. Yet, the causes of the decline effect are rarely explored. By conducting a meta-analysis of 373 studies across 24 years, this study tested the reproducibility of the negative effect of ocean acidification on calcification of marine organisms. We found that the pioneering studies tended to report large negative effects of ocean acidification on calcification, but as more researchers assimilated their results into this research in subsequent years, the size of negative effects declined. This decline effect was driven by three types of bias, including experimental design bias (e.g., exposure duration and pH manipulation method), species selection bias (e.g., selection of sensitive species in the early studies) and publication bias (i.e., the file-drawer effect). Based on the findings, we advocate that the decline effect represents a scientific process by which discoveries are initially assimilated, and their limitations are subsequently explored. Where context-dependency of a discipline is high, the lack of reproducibility may not represent a crisis, but a part of theory development and eventual gravitation towards a consensus position.

Ocean acidification,Decline effect,Marine organisms,Meta-analysis,Calcification