Atterberg limits fail in recognizing fragipan horizons

Fragipan is a subsurface soil horizon characterized by high bulk density, coupled with an abundance of fine, blind pores. We hypothesized that this distinctive porosity pattern could affect the Atterberg limits, with sig-nificant deviations from those of otherwise comparable soil horizons. Additionally, longer pre-wetting times might be needed to determine the liquid and plastic limits in fragipan, due to restrictions in water movements. Based on these hypotheses, our research focused on (1) assessing the most suitable fragipan-specific sample preparation time for liquid and plastic limit determination; (2) evaluating if the Atterberg limits reflect the presence of fragic properties; (3) exploring the relationships between Atterberg limits and other relevant soil properties through Artificial Neural Networks. Fragipan horizons (F) had a slightly higher liquid limit, much higher bulk density, and specific surface area than nonfragipan (NF). The plastic limit was however comparable for F and NF. Thus, fragipans had a larger interval of plasticity, which is in contrast with slaking tests and well- known field evidence. Despite the expected differences between F and NF, three hours’ pre-wetting was enough for all samples. This might be due to the size fraction used in the analysis, more sensitive to collapse upon water saturation than bigger clods. Atterberg limits did not help in discriminating F from NF, despite the relevant differences in other soil physical properties. Neural Networks showed that the transition from the plastic to the liquid state was mainly driven by texture and specific surface area, while the plasticity interval was mainly affected by the bulk density. These results provided further insights into the mechanical properties of fragipan horizons.