Introduction: Antimony (Sb) is a naturally occurring metalloid. Anthropogenic activities have caused elevated levels of Sb in soil. Sb is a toxic element and Sb(III) is much more toxic than Sb(V). However, Sb(V) is the dominant speciation in field soil. Thus using Sb(III) to study Sb toxicity will greatly overrate its toxicity and has a little ecological relevance for assessing hazard posed by Sb in real life scenarios. The springtail Folsomia candida has been extensively used as an ecologically relevant model organism because its high sensitivity. The interest on toxicity of Sb has increased in recent decades, however, data on toxicity of Sb to springtails are fairly limited and existing studies did not consider effects of various soil properties and long aging time. As all know, these two factors could significantly influence metal toxicity and should be considered for the purpose of toxicity assessment. To address this, the chronic toxicity of Sb(V) to Folsomia candida were evaluated following ISO guideline in the laboratory studies.

Materials and methods: Ten types of soil, representing typical agricultural soils in China, were used as test soils. The Sb(V) were artificially added into soils and nominal concentrations were set up as 400–12800 mg/kg. After aging Sb-treated soils for 150 d, the total Sb and water soluble Sb concentrations were measured, and springtails were exposed to soils for 28 d. At the end of test, the number of reproduction were counted. The EC₅₀ (concentrations causing 50% inhibition) values are calculated using Logistic curve.

Results: The toxicity of Sb significantly differed between ten soils, closely relating to distinct variations in water soluble Sb. The EC₅₀ values based measured total Sb ranged from 1735 mg/kg to 13680 mg/kg, which corresponded to a 8-fold variation among ten soils. Correlation analysis showed that EC₅₀ values correlated significantly and positively with clay, organic matter, free Fe and Al oxides, amorphous Fe oxides, indicating decreased toxicity with increasing contents of these soil constituents. The EC₅₀ values significantly and negatively correlated with pH and CaCO₃. The prediction models were further developed using regression analysis. The results indicated that free Fe oxides was the most important single factor in predicting Sb toxicity, explaining 89.3% of toxicity variance, and the inclusion of clay in this model could best explain the toxicity variance (95.9%). Soil pH and organic matter could also well explain the toxicity variance.

Conclusions: This study not only demonstrated the significant toxicity differences of Sb to springtails among soils, but also developed quantitative relationships using key soil properties. It should be noted that the prediction models was based on soils aged for 150 d which made Sb equilibrate sufficiently in soil, which allow prediction of Sb toxicity in a wide range of soils and are helpful to establish soil-specific quality criteria.