A comparison of ground-based count methods for quantifying seed production in temperate broadleaved tree species

Abstract • Key message Litter trap is considered the most effective method to quantify seed production, but it is expensive and time-consuming. Counting fallen seeds using a quadrat placed on the ground yields comparable estimates to the litter traps. Ground quadrat estimates derived from either visual counting in the field or image counting from quadrat photographs are comparable, with the latter being also robust in terms of user sensitivity. • Context Accurate estimates of forest seed production are central for a wide range of ecological studies. As reference methods such as litter traps (LT) are cost- and time-consuming, there is a need of fast, reliable, and low-cost tools to quantify this variable in the field. • Aims To test two indirect methods, which consist of counting the seeds fallen in quadrats. • Methods The trial was performed in three broadleaved (beech, chestnut, and Turkey oak) tree species. Seeds are either manually counted in quadrats placed at the ground (GQ) or from images acquired in the same quadrats (IQ) and then compared against LT measurements. • Results GQ and IQ provide fast and reliable estimates of seeds in both oak and chestnut. In particular, IQ is robust in terms of user sensitivity and potentially enables automation in the process of seed monitoring. A null-mast year in beech hindered validation of quadrats in beech. • Conclusion Quadrat counting is a powerful tool to estimate forest seed production. We recommend using quadrats and LT to cross-calibrate the two methods in case of estimating seed biomass. Quadrats could then be used more routinely on account of their faster and simpler procedure to obtain measurements at more spatially extensive scales.