Speaker
Description
Soil ecotoxicological test is an essential tool for risk assessment of various xenobiotic chemicals. Such tests can be performed using soil invertebrates by exposing them to specified soil contaminating chemicals. Soil invertebrates provide various ecosystem services (i.e., soil transformations are beneficial for mankind). For example, soil invertebrates may influence the mineralization of nutrients in soil organic matter (SOM), affecting the "soil fertility’’ – an important factor from the agricultural aspect. Hence, soil invertebrates serve as an outstanding biological indicator of the terrestrial ecosystem and overall soil quality, considering their high sensitivity compared to other soil quality indicators (physical/chemical). Therefore, laboratory tests using invertebrates can be considered as the mainstay of ecotoxicological impact assessment. Quantitative and/ or qualitative results elicited from such tests help several regulatory authorities across the globe to determine the ecological risk level of substances and safe exposure limits for human and soil biota. Thus, such valuable information enables governmental regulatory authorities to control manufacturing output and sale of pesticides, to decide threshold limit for safe application of residues to agricultural soils, etc. However, laboratory tests (both in vivo and in vitro) are costly and time-consuming affairs and involve extensive use of animals. Hence, such tests cannot be extended entirely for predicting the toxicity of novel compounds. As a result, an alternative in-silico approach of quantitative structure-activity relationships (QSARs) is used for environmental risk assessment for novel compounds, free of the exhaustive use of test animals under the REACH regulations in the EU. In this background, necessary limited data available from laboratory tests on the soil invertebrate Folsomia candida (C. name: Springtail) were collated from the database of ECOTOX (cfpub.epa.gov/ecotox). Data is collected for the endpoint - pEC$_{50}$ only. Samples of 45 chemical compounds were selected, for which chemical descriptors were calculated for each compound. Then the whole dataset is split into a test dataset (11 compounds) and a training dataset (34 compounds), based on Euclidean Distance based approach. Using genetic algorithm, significant descriptors out of all descriptors’ pools were selected. Using these selected set of descriptors both in test set and training set, the Best Subset Selection software (http://teqip.jdvu.ac.in/QSAR_Tools/) was run, which gave the best possible combinations of a limited number of descriptors based on desired linear model equation length from which four best models were selected based on their internal and external validation metrics. Four partial least squares (PLS) models were built based on those four multiple linear regression (MLR) models. These four models were then used in Intelligent Consensus Predictor version 1.2 (PLS version) to get the final consensus model, built using the best selection of predictions (compound-wise) from four ‘qualified’ individual models. Both internal and external validations metrics of this consensus model are well-balanced and within the acceptable range as per the OECD criteria, which is reconfirmed by predictions made through the Chemical Read-Across method. From the aforementioned parameters, a certain conclusion on general contribution can be made: Firstly, with an increase in the number of electron donor groups and consequent hydrogen bond interactions, the toxicity (pEC$_{50}$) increases. The soil toxicity can increase with an increase in the aqueous solubility of compounds. However, on the contrary, if bulky groups are present amidst the polar groups, toxicity is observed to decrease due to steric hindrance and increased lipophilicity. Secondly, it was seen from the data that when the number of five-membered rings increases, then toxicity (pEC$_{50}$) increases. Lastly, a negative correlation was observed between the number of substitutions to aromatic rings and toxicity (pEC$_{50}$).