Evaluating sampling schemes for quantifying seed production in beech (Fagus sylvatica) forests using ground quadrats

Accurate estimates of seed production are central for understanding mast seeding mechanisms at tree and forest scales, and for designing sustainable management strategies. As trees are long-lived organisms, a long-term perspective is required to understand how reproduction acts during the life cycle of a tree. However, longterm series of seed production are challenging to obtain, as the available seed count procedures strictly rely on field methods, which are cost- and time-consuming, inherently limiting their widespread use at extensive spatial and temporal scales. In this study, we proposed a simple, rapid and flexible field method based on counting the seed in mobile ground quadrats (GQ), which was tested in beech forests. Quadrat measurements were first validated against reference measurements obtained from litter traps (LT) in three permanent plots. Results indicated that GQ provides robust and reliable estimates of seeds, which are not affected by seed predation occurring at the forest floor. Additional quadrat measurements were performed to evaluate the influence of sampling schemes (random, regular, systematic) on the estimation of mean seed production at the plot scale. One hundred quadrats were collected in 0.25 ha beech plots and considered as a reference for evaluating the different sampling schemes and sampling sizes. Measurements were performed in October (three plots), which represented the peak of seed fall, and November (two plots). Results indicate that about 25 randomly located measurements allowed to characterize plot-level mean seed production with an acceptable error below 20%, regardless of the different mean seed production observed between the studied plots and the sampling periods. If the 25 sampling points are arranged in a grid, the obtained mean estimates are within the confidence interval of the reference plot-level values.

Automated summary

Accurate estimates of seed production are crucial for understanding mast seeding mechanisms and designing sustainable management strategies. A new field method for counting seeds in beech forests was proposed and tested, providing reliable estimates unaffected by seed predation. Sampling schemes were evaluated to characterize plot-level mean seed production, suggesting that a grid of 25 random measurements can accurately estimate the plot-level values.