Reconstructing relative sea level (RSL) is essential for understanding coastal evolution and mitigating the impacts of climate change. Foraminiferal environmental DNA (eDNA) has recently been validated as a robust proxy for reconstructing past sea levels and has the potential to overcome limitations of traditional morphological approaches associated with the poor or selective preservation of foraminiferal tests. However, eDNA assemblages are influenced not only by the spatiotemporal variability of foraminiferal communities but also by factors affecting DNA transport and preservation. Moreover, unlike morphological assemblages, eDNA comprises multiple DNA sources, including propagules, juveniles, adults, and extracellular DNA. The spatiotemporal variability of intertidal foraminiferal eDNA assemblages, differences in taxonomic composition among life-stage-derived eDNA fractions, and their influence on RSL reconstruction remain poorly understood.

To address these uncertainties, we conducted a two-year eDNA monitoring study across three stations spanning a tidal elevation gradient in mangrove and mudflat environments in Hong Kong, with replicate sampling during both dry and wet seasons. In addition, foraminiferal eDNA was analyzed from bulk sediment and the 500–63 μm and <63 μm sediment fractions collected during the dry season. Our results showed that foraminiferal eDNA assemblages exhibited environment-specific temporal variation but no significant small-scale spatial variation, with significant seasonal differences detected only in the mudflat–mangrove transitional zone. Size-fraction analysis further revealed that the propagule- and juvenile-dominated <63 μm fraction closely resembled bulk eDNA assemblages in the mudflat environment but diverged from bulk assemblages in mangroves, suggesting a greater contribution of propagule- and juvenile-derived eDNA to the total eDNA pool in mudflat sediments. Bayesian transfer function (BTF) modelling demonstrated that, in mangrove environments, elevation estimates derived from bulk sediment and the 500–63 μm fraction accurately reproduced observed elevations, whereas estimates from the <63 μm fraction showed tendency toward underprediction but generally remained within the model uncertainty. In contrast, all eDNA assemblages from the mudflat–mangrove transitional zone, including bulk samples collected in both seasons and both size fractions, consistently overpredicted elevation. These findings indicate that stable mangrove environments provide the most suitable settings for establishing reliable elevation–assemblage relationships, whereas transitional environments require greater caution because of their higher environmental variability. Our study provides practical guidance for applying foraminiferal eDNA as a proxy for reconstructing past sea levels in dynamic coastal settings.