基于农田土壤镉污染修复后糙米镉的指标筛选

目前农田土壤镉污染修复以降低糙米镉含量为标准，而忽略了其变化的关键性的土壤物理、化学及生物指标。基于此，采用方差分析、多元回归分析、通径分析、随机森林和结构方程模型等统计方法分析了土壤pH值、有机质、容重、阳离子交换量、速效养分、质地、微团聚体、酶活性、微生物数量等23个土壤指标对土壤有效态Cd和糙米Cd含量重要性排序。结果表明，总体上施用土壤调理剂提高了土壤pH值、容重和阳离子交换量，改变了土壤质地组成，增加了大粒级团聚体，并影响了微生物环境，有效降低了土壤有效态Cd和糙米Cd含量，但抑制了部分酶活性和微生物数量。通过逐步回归分析，表明土壤pH值和放线菌可以解释土壤有效态Cd 64.32%的变化，阳离子交换量、速效钾、有效磷和蔗糖酶可以解释糙米Cd 82.10%的变化；通径分析表明大粒径团聚体、有机质、黏粒、砂粒、放线菌和真菌对土壤有效态Cd和糙米Cd的直接效应较大；随机森林分析表明土壤pH值是影响土壤有效态Cd和糙米Cd含量的主导因素；结构方程模型表明糙米Cd含量主要受土壤pH值、有机质、阳离子交换量和有效磷的影响，土壤有效态Cd主要受土壤酶活性和微生物数量的影响。不同分析方法侧重点不同，总体上对糙米Cd含量影响较大的是pH值、阳离子交换量、有效磷和有机质等，对土壤有效态Cd含量影响较大的是pH值、放线菌和有机质等。

Screening of Indicators in Brown Rice Cadmium after Remediation of Cadmium Pollution in Farmland Soil

At present, the restoration of cadmium pollution in farmland soil is based on reducing the cadmium (Cd) content in brown rice, while ignoring the key soil physical, chemical and biological indicators of its change. Based on statistical methods including analysis of variance, multiple regression analysis, path analysis, random forest and structural equation model, the importance rankings of 23 soil indices to soil available Cd and brown rice Cd were analyzed. The results showed that the application of soil conditioners improved soil pH, bulk density and CEC, changed soil texture composition, increased the proportion of large-size aggregates, affected the microbial environment, and effectively reduced the contents of soil Cd and brown rice Cd, but inhibited a part of enzyme activities and the number of microorganisms. Through stepwise regression analysis, soil pH and actinomycetes explained 64.32% changes in soil available Cd, CEC, available potassium, available phosphorus (P) and invertase explained 82.10% changes in brown rice Cd. Path analysis showed that large-size aggregates, SOM, clay, sand, actinomycetes and fungi greatly directly affected soil available Cd and brown rice Cd. Random forests showed that soil pH was the dominant factor affecting soil available Cd and brown rice Cd. The structural equation model showed that the brown rice Cd was mainly affected by soil pH, SOM, CEC and available P, and the soil available Cd was mainly affected by soil enzyme activities and microbial quantity. Different analysis methods had different focuses. In general, pH, CEC, available P and SOM showed the greatest influences on the Cd content of brown rice. While, pH, actinomycetes and SOM had the greatest influences on soil available Cd.