Identifying maternal trees is one of the major challenges in both studying species’ spatial patterns and collecting seeds from maternal trees for reforestation, especially in dioecious species where only half of reproductive-sized individuals may produce seeds. To identify the maternal trees, reproductive events need to be observed, but not all individuals reproduce every year and individual-based phenology data is rarely available. To solve this challenge, we conceived a novel method to predict individual sex using the potential difference in soil pollen concentrations beneath the crowns of female and male plants, based on the fact that only males produce pollen. Although pollen has generally been considered to be decomposed unless in sediments, we found a massive number of pollen grains in surface soil, and mean pollen concentration under males was significantly higher than under females as hypothesized. The overall accuracy of the method was 85% (Sugiyama et al. 2017).