Beyond seed counts: What two decades of seed testing reveal about European beech in Poland

Forest regeneration depends on a long chain of processes, of which the number of seeds produced is only the first link. Climate-driven changes in seed production and masting in temperate trees are well documented, whereas long-term changes in the quality of the seeds produced have been examined far less often. Using a nationwide operational dataset for European beech ( Fagus sylvatica L.) from Poland (13,387 seed lots from 380 forest districts, 1996–2024; temporal analyses on 5381 freshly harvested lots from 352 districts, 2004–2023), we followed three dimensions of seed quality: seed mass, the proportion of filled seeds (purity) and the viability of filled seeds. The three dimensions were nearly uncorrelated (ρ ≤ 0.11 at the district-year level) and changed over time in different ways. Adjusted annual mean seed mass declined by 1.48 g per year (29.6 g, or about 10%, over 20 years), with the whole distribution shifting downwards uniformly across the country. The median proportion of filled seeds remained stable, but the frequency of lots with fewer than 85% filled seeds rose from about 1% to 14% between 2006 and 2023. The viability of filled seeds showed no temporal trend, although it was the only dimension associated with seasonal climate: warmer previous summers and harvest-year autumns were associated with higher viability, and warmer harvest-year springs and summers with lower viability, and these associations were unchanged after controlling for annual crop size. Seed fill was not associated with seasonal climate, and seed mass showed at most a weak negative association with seasonal temperature that could not be assigned to any seasonal window. The decline in seed mass was not related to the masting history of the districts, nor to the local rate of decline in the interannual variability of seed production reconstructed from collection records. In cold storage, viability declined with storage duration and the rate of loss varied among harvest years more than the mean rate itself, but climate during seed development did not modify the rate of loss. Summer and autumn temperatures increased over the study period whereas spring temperature did not; the dimension of quality that responds to climate therefore remained stable, while the two dimensions that changed were at most weakly related to seasonal climate. It seems that the decline in seed mass and the increasing frequency of fill failure are driven by processes that seasonal climatic means do not capture, and their consequences for regeneration remain to be tested.