Soil pollution by arsenic (As) has become one of the greatest environmental problems. Heavy metal pollution can lead to a decrease in soil quality and functional degradation, resulting in a decrease in soil enzyme activities. Soil enzyme activity has been proposed as an appropriate indicator for assessing metal pollution since it is sensitive to microbial changes induced by heavy metal stress and is strongly linked to soil nutrient cycling. Therefore, early detection and evaluation systems monitoring soil pollution are significantly critical. Hence, in long-term As-polluted soils the activity of four soil enzymes, such as dehydrogenase (DHA), fluorescein diacetate (FDA), acid phosphatase (AcP) and urease (URE) was investigated. Functional diversity indexes of soil enzymes were evaluated by geometric mean (GM), weighted mean (WM) and total enzyme index (TEI). Results showed that GM, WM and TEI were better correlated with soil characteristics than single enzyme activities. Only in Hunan Shimen soils, total As content shows a significant negative correlation with all soil enzyme activities and functional diversities. Results of RDA analysis showed that AN, clay and SOM contents were the dominant soil factors affecting soil enzymes and functional diversities in the Yunnan soils. In contrast, TN, total As, AN and AP were dominant factors in the Hunan Shimen soils. The ecological dose (ED10, ED50) values varied widely. URE and DHA are indicated as the most sensitive soil enzymes for As pollution. Except for GM, two functional diversities' (TEI and WM), ED10 and ED50 values showed higher results than single soil enzyme. The results demonstrate that soil properties strongly influence soil enzyme activities in both soils. In contrast, soils with high SOM content can reduce the effect of As on them by covering them from the inhibition effect. URE and DHA soil enzymes illustrated the lowest ED values. Consequently, they can be used as bioindicators in As-contaminated soils. [ABSTRACT FROM AUTHOR]