Cadmium toxicity to and accumulation in a soil collembolan (Folsomia candida): major factors and prediction using a back-propagation neural network mode

Cadmium toxicity to and accumulation in a soil collembolan (Folsomia candida): major factors and prediction using a back-propagation neural network mode | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Cadmium toxicity to and accumulation in a soil collembolan (Folsomia candida): major factors and prediction using a back-propagation neural network mode Simin Li, Zhu Li, Xin Ke, Worachart Wisawapipat, Peter Christie, and 1 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-3740915/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 02 Mar, 2024 Read the published version in Environmental Science and Pollution Research → Version 1 posted 5 You are reading this latest preprint version Abstract Accurate prediction of cadmium (Cd) ecotoxicity to and accumulation in soil biota is important in soil health. However, very limited information on Cd ecotoxicity on naturally contaminated soils. Herein, we investigated soil Cd ecotoxicity using Folsomia candida , a standard single-species test animal, in 28 naturally Cd-contaminated soils, and the back-propagation neural network (BPNN) model was used to predict Cd ecotoxicity to and accumulation in F. candida . Soil total Cd and pH were the primary soil properties affecting Cd toxicity. However, soil pH was the main factor when the total Cd concentration was ˂ 3 mg kg − 1 . Interestingly, correlation analysis and the K-spiked test confirmed nutrient potassium (K) was essential for Cd accumulation, highlighting the significance of studying K in Cd accumulation. The BPNN model showed greater prediction accuracy of collembolan survival rate (R 2 = 0.797), reproduction inhibitory rate (R 2 = 0.827), body Cd concentration (R 2 = 0.961), and Cd bioaccumulation factor (R 2 = 0.964) than multiple linear regression models. Then the developed BPNN model was used to predict Cd ecological risks in 57 soils in southern China. Compared to multiple linear regression models, the BPNN models can better identify high-risk regions. This study highlights the potential of BPNN as a novel and rapid tool for the evaluation and monitoring of Cd ecotoxicity in naturally contaminated soils. back-propagation neural network model Cd soil collembolan bioaccumulation metal toxicity Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 1. Introduction Soil cadmium (Cd) pollution is a major concern globally. Combustion, traffic, metal industries, landfills, sewage sludges, mining, and fertilizers are primary sources of soil Cd contamination, which represents a high risk to soil biota (Zhao et al., 2015 ; Ali et al., 2019 ). Soil-animal-based soil ecotoxicology methods are commonly used to predict metal ecological risks. Collembolans are widespread in soils and play an important role in nutrient cycling, energy transfer, and maintenance of biodiversity (Fountain and Hopkin, 2005 ; Howcroft et al., 2009 ). They generally live near the soil surface and are vulnerable to soil disturbance. Several collembolan species have been used as standard soil ecotoxicological test animals. Notably Folsomia candida , because of its sensitivity to pollutants, ease of culture, short life cycle, and high reproduction rate et al. (Coyle et al., 2017 ). Laboratory studies show that high Cd concentrations significantly decreased collembolan survival and reproduction rates, growth (body weight and body length), and locomotor behavior (Lock and Janssen, 2001a , b ; Kim and An, 2014 ; Liu et al., 2019 ). Therefore, the endpoints (survival, reproduction, and growth rates) in soil ecological tests commonly be used as bioindicators to indicate soil ecological risks. However, those endpoints are not trustworthy in naturally contaminated soils. Firstly, the determined effective concentration that decreased the reproduction rate by 50% (EC 50 ) obtained in labs (newly spiked artificial soils) was 158 (137–184) mg kg − 1 (Lock and Janssen, 2001a ), a significantly higher value than those in most naturally contaminated agricultural soils examined in China (Zhao et al., 2015 ). In addition, Zhang et al. ( 2021 ) propose that soil pH, rather than soil total metal concentration, is the main factor restricting collembolan survival and reproduction rates in naturally contaminated agricultural soils. Thus, in natural soils, studies on the survival and reproduction of collembolans in naturally metal-contaminated soils may not accurately reflect metal availability, while studies on metal accumulation in collembolans are increasing because they are indicative of metal bioavailability in soils and the potential risks to the human food chain (Sahraoui et al., 2021 ). The internal pollutant concentration in biota is a good biological indicator of organisms exposed to polluted areas, and methods that allow the prediction of metal bioavailability to soil biota are needed for soil health and soil ecological risk assessment. Studies indicate that soil organic matter content, soil pH, and clay, iron, and manganese oxide contents, etc. have significant relationships with Cd toxicity to and accumulation in collembolans (Lock and Janssen, 2001a , b ; van Gestel, 2008 ; Liu et al., 2019 ). In addition, these soil properties are commonly incorporated into empirical models (single and multiple linear regression equations) to build empirical equations for the prediction of Cd toxicity to and accumulation in collembolans (Liu et al., 2019 ). However, most conclusions are generally drawn in newly spiked artificial soils and/or standard LUFA 2.2 soil, which set soil parameters as a single variable with different ranges (Speyer, Germany). These results can be inconsistent when compared with naturally contaminated soils that show large ranges of soil properties (Lock and Janssen, 2001a , b ; Liu et al., 2019 ). Metal bioavailability may be lower in natural field soils due to aging effects (Wang et al., 2022 ), and the empirical equations obtained are conditional and not applicable to other soil types (Liu et al., 2019 ; Wang et al., 2021 ). In addition, soil properties are associated with each other, linear regression analysis cannot allow for their complex and nonlinear calculations, and they may therefore be inefficient under a wide range of conditions (De’ath and Fabricius, 2000). Therefore, factors influencing Cd toxicity to and accumulation in collembolans under naturally Cd-contaminated soils, and accurate, rapid, and universal models for the prediction of Cd toxicity to and accumulation in collembolans under a wide range of soil properties require further development. The back-propagation neural network (BPNN) model has recently been proven to be a rapid, accurate, universal, and complexity-solved model (Wang et al., 2016 ). It is one of the most extensively used artificial neural network (ANN) models and is a multi-layer feed-forward neural network model trained according to the error reverse propagation algorithm (Tumbo et al., 2002 ; Noh et al., 2006 ). The BPNN comprises numerous neurons and simple units of parallel algorithms and thus has a strong fitting ability and allows complex calculations in practical applications (Noh et al., 2006 ; Wang et al., 2016 ). This model has been successfully used in numerous fields of research. For example, Li et al. ( 2019 ) identified the interactive relationships between meteorological factors and bioaerosols to effectively forecast the bioaerosol concentration in south central China by the BPNN model. In the soil-plant system, the BPNN model has successfully calculated and predicted the complex internal factors of metal migration and accumulation from soil to plant tissues. Liu et al. ( 2011 ) used an ANN model and estimated Cd and copper (Cu) concentrations in rice grains by integrating the spectral indices with environmental parameters. Wang et al. ( 2021 ) used the BPNN model to predict the Zn bioaccumulation factor (BAF) in rice grains based on soil properties and safe utilization in a large-scale field study, and the prediction performance was significantly higher than that of multiple linear regression models. Numerous studies confirm that BPNN is a useful tool for minimizing errors of prediction and allowing calculations under complicated conditions than multiple linear regression models (Liu et al., 2011 ; Li et al., 2019 ; Wang et al., 2021 ). In addition, the BPNN model calculation sets training, validation, and testing data to avoid overfitting situations, and is thus also available for untested data, increasing efficiency and cost-effectiveness (Wang et al., 2016 ). However, to our knowledge, there is little or no information on its application to soil-animal systems, which may help us better predict metal bioavailability and ecotoxicity to soil biota in naturally contaminated soils with a wide range of soil properties. The objectives here were therefore to (1) identify the Cd toxicity effects and accumulation characteristics of F. candida in a wide range of naturally Cd-contaminated soils, (2) identify the main factors affecting Cd toxicity to and accumulation in F. candida , (3) develop BPNN models to predict Cd toxicity to and accumulation in F. candida and compare the prediction performance with multiple linear regression models, and (4) further use the developed BPNN models in Cd bioaccumulation prediction and monitoring. The aim was to build accurate, rapid, and widely applicable methods to predict and assess Cd ecotoxicity and accumulation in naturally contaminated soils with a wide range of soil properties. 2. Materials and methods 2.1 Test soils and organism A total of 28 naturally Cd-contaminated topsoil samples (0–20 cm) with a wide range of soil properties were collected from eight Chinese provinces, in southern China. The basic soil properties were determined using standard methods (Lu, 2000 ) and are shown in Table S1 , namely soil pH (soil: water ratio 1:2.5 (w/v)), soil organic carbon content (SOC), cation exchange capacity (CEC), soil texture (clay), total and available nitrogen contents (TN, AN), total and available phosphorus contents (TP, AP), total and available potassium contents (TK, AK), and total Cd, Fe, and Mn contents. The ranges of soil pH, SOC content, and soil total Cd content were 4.73–8.4, 6.55–41.6 g kg –1 , and 0.54–25.7 mg kg –1 , respectively. After air-drying and sieving through a 2-mm mesh, the soils were equilibrated for 7 days at 50% water holding capacity before testing. F. candida is widespread in soils and has been widely used in ecotoxicity tests (OECD, 2009 ). Individuals were originally obtained from Aarhus University, Denmark, and were subsequently cultured in our laboratory for > 10 years. They are reared in Petri dishes (height 10 mm, diameter 90 mm) with a layer of moist plaster mixed with activated charcoal (9:1 w/w) at 20 ± 1°C and 75% relative humidity with a photoperiod of 16:8 h (light/dark). The animals were fed dried baker's yeast (Angel Yeast Co., Ltd., Yichang, Hubei, China). Distilled water was added to maintain the moisture content. F. candida individuals synchronized for 10–12 days were used in the tests (OECD, 2009 ). 2.2 Single-species toxicity test The test procedure was based on the OECD guidelines with some modifications (OECD, 2009 ). Ten cleaned and gut-vomited F. candida individuals were introduced into each test vessel (5.5 cm diameter, 250 mL volume) containing 30 g soil. There were three replicates of each treatment. The test vessel was covered with plastic film with small holes punched through. Field conditions were mimicked by providing no extra food. The animals were cultured in the same conditions as described above. The containers were weighed twice weekly to maintain water balance. After 28 days, soil animals were extracted using a controlled temperature gradient extractor. The numbers of juveniles and adults were counted under a stereomicroscope (Leica S8 APO, Wetzlar, Germany). 2.3 Chemical analysis The extracted F. candida individuals were starved for 2 days to remove interfering substances from their gut contents. They were then rinsed with ultrapure water and freeze-dried (− 80°C). Aliquots (~ 20–300 µg) of the animals were weighed (precision ± 1 µg, XS3DU, Mettler Toledo, Columbus, OH) and digested with 300 µL of a mixture of HNO 3 (65%; ultrapure) and H 2 O 2 (30%; ultrapure) (1:1 v/v) in 5-mL Teflon plastic digestion vessels at 105°C which were sealed and dried in an oven for 6 h. After digestion and cooling, the vessels were transferred to an electric heating plate and heated at 120°C to dryness. The samples were stored in a refrigerator at 4°C until analysis. Before analysis, 2 mL 5‰ HNO 3 was added to dissolve the residue, and the samples were analyzed by ICP-MS (NeXion 2000, Perkin Elmer, Waltham, MA) after filtration. A certified reference material (GBW08551, pork liver (Institute of Geophysical and Geochemical Exploration, Langfang, Hebei, China)) was included for quality control, with recoveries ranging from 89.4 to 108%. 2.4 Description of the BPNN model The BPNN model is a feed-forward neural network consisting mainly of input layers, multiple hidden layers, and output layers. Each layer consists of several nodes (Wang et al., 2016 ). The number of soil property factors is selected as the input layer based on the correlation between soil properties and output target (linear regression and redundancy analysis (RDA)), the output layer is 1, and the number of hidden layers is determined based on the empirical formula: n1= \(\sqrt{ n +m}\) + k (1) where n1 is the number in the hidden layer, n is the number in the input layer, m is the number in the output layer, and k is a regulation constant from 1 to 10. Seventy percent of samples are used for training, 15% for validation, and 15% for testing (Tumbo et al., 2002 ; Noh et al., 2006 ). 2.5 Statistical analysis All data are expressed as mean ± standard error (SE, n = 3). Statistical analysis was conducted with Microsoft Excel 2016 and the SPSS 20.0 software package, using one-way analysis of variance to compare the differences among treatments, and Duncan’s multiple range test was at the 5% protection level. The neural network toolbox in MATLAB was used for the BPNN model, and the Levenberg-Marquardt training algorithm was used for calculation. The lethal concentration that reduced survival by 50% (LC 50 ) and EC 50 were calculated by probit regression modeling at the 5% protection level. ArcMap 10.8 software was used to draw sensitive maps. In soil 23, the soil total Cd concentration was 0.76 mg kg − 1 , soil pH was 5.05. F. candida had the highest survival and reproduction numbers in this soil and, therefore, was used as the max reproduction number to calculate the reproduction inhibitory rate (Zhang et al., 2015 ). Reproduction inhibitory rate = \(\frac{\text{max reproduction number - reproduction number}}{\text{max reproduction number}}\) × 100% (2) The bioaccumulation factor (BAF) is commonly used to indicate the accumulation of metals and is calculated by: BAF = C body / C soil (3) where C body is the concentration of Cd in F. candida body tissue, and C soil is the total Cd concentration in the soil. Log-transformed soil parameters (except soil pH), body Cd concentrations in F. candida , and BAF values were used to ensure the normality of data and homogeneity of the variance and used for linear regression and RDA. The relationships between survival and reproduction number, body Cd concentration in F. candida , BAF, and soil properties were conducted using Pearson’s correlation analysis. The RDA analysis of internal Cd concentration in F. candida (Log [Cd]) and BAF-Cd (Log [BAF]) were conducted using Canoco 5.0 software, with Log [Cd] and Log [BAF] as species variables and the soil properties as environmental variables. The reproduction inhibitory/survival rate, Log [Cd], and Log [BAF] were also examined using multiple linear regression analysis with different soil properties: Survival rate = a 1 log ( soil properties 1 ) + a 2 log ( soil properties 2 ) + …… a n log ( soil properties n ) + b (4) Reproduction inhibitory rate = a 1 log ( soil properties 1 ) + a 2 log ( soil properties 2 ) + …… a n log ( soil properties n ) + b (5) Log [Cd] = a 1 log ( soil properties 1 ) + a 2 log ( soil properties 2 ) + …… a n log ( soil properties n ) + b (6) Log [BAF-Cd] = a 1 log ( soil properties 1 ) + a 2 log ( soil properties 2 ) + …… a n log ( soil properties n ) + b (7) where [Cd] was the Cd in F. candida , a1, a2, … and a n were the regression coefficients, and b was the constant coefficient. The accuracy of the predictions was evaluated based on mean absolute error (MAE), mean relative error (MRE), the root-mean-square error (RMSE), and determination coefficient (R 2 ). The higher R 2 value and the smaller MAE, MRE, and RMSE values, the higher accuracy of the model (Wang et al., 2016 ; 2021 ). 3. Results and discussion 3.1 Survival and reproduction of F. candida The soil samples used in this study were generally representative of the Cd concentrations found in most Chinese Cd-contaminated agricultural soils (Zhao et al., 2015 ), and the Cd concentrations in all soils were higher than the current national soil Cd risk screening value (GB 15618 − 2018). An average of 76.1% of adults were alive in the 28 Cd-contaminated soils (Table 1 ), indicating that the death rate of F. candida was relatively low in the naturally Cd-contaminated soils. There was no significant relationship between the survival number and soil total Cd concentration or any other soil properties by Pearson’s analysis (Fig. 1 ), and the Cd LC 50 was not able to be calculated in these soils. The scatter plot of the survival rate in different Cd-contaminated soils is shown in Fig. 2. The survival rate appeared to increase with Cd concentrations in soils ˂ 3 mg Cd kg − 1 concentration (r = 0.528, P 3 mg kg − 1 (r = – 0.645, P < 0.01). These trends were opposite to those of previous studies, showing a significant negative relationship between soil Cd concentrations and the survival rate, and soil pH, SOC, and soil CEC etc that were essential parameters for the survival rate in laboratory studies (Luo et al., 2014 ; Sahraoui et al., 2021 ). This may be because the previous conclusions were drawn mainly from ranging and finding tests, those tests set the Cd concentration, soil pH, SOC, and CEC, etc as a single variable in a specific soil (artificial soils, standard soil LUFA, or newly Cd spiked soils) (Lock and Janssen, 2001a , b ; Liu et al., 2019 ). Therefore, it is easy to find the main factors and the most appropriate soil conditions. However, the naturally contaminated soils studied here were more complex, and the soil properties were closely associated with each other, and it was, therefore, difficult to obtain these relationships. Notably, the current study showed that survival number decreased with increasing soil pH (r = 0.585, P < 0.05) in low Cd concentration soils (< 3 mg kg − 1 ) (Fig. S1 ), suggesting that instead of Cd, pH was the major factor constraining survival in soils with low Cd concentrations, on the contrary, Cd was likely to have a promotive effect on survival rate in low Cd concentration soils. Studies indicate that collembolans are more appropriate living in acid and circumneutral soils, and few survive in soils with pH values > 8.0, indicating that soil pH plays a major role in collembolan survival (Fountain and Hopkin, 2005 ; Howcroft et al., 2009 ). Therefore, the standard single-species test should not be generalized for a wide range of soil pH values. Instead, the standard single-species test animal should be specified for certain ranges of pH values (Fountain and Hopkin, 2005 ; Howcroft et al., 2009 ; Liu et al., 2019 ). An alternative explanation is hormesis effects, demonstrating stimulatory or beneficial effects occurring at a lower dose, but inhibitory or toxic effects at a higher dose (Mark, 2008 ; Gospodarek et al., 2020 ). Hormesis should therefore be considered in the experimental design, selection of biological markers and endpoints, and risk model development et al. in assessing pollutants (Christiani and Zhou, 2016 ). Overall, the relationships between survival rate and soil properties are intricate and are challenging to predict using traditional statistical methods. The BPNN model was therefore used here to resolve the complex situation between survival rate and soil properties, in which soil total Cd concentration and soil pH were selected as input layers in the neural network mode. Table 1 Survival number and rate, juvenile number and reproduction inhibitory rate, Cd concentration in body tissue, and Cd bioaccumulation factor (BAF) of F. candida in 28 naturally Cd-contaminated soils. Survival number Survival rate (%) Juvenile number Reproduction inhibitory rate (%) Cd concentration in tissue (mg kg − 1 ) BAF Min 5 50 61 0 2.44 0.40 Max 9.3 93 352 82.7 85.7 36.0 Median 7.3 73.3 172 51.1 10.1 3.16 Mean ± SE 7.6 ± 0.2 76.1 ± 2.10 181 ± 16 48.5 ± 4.4 18.2 ± 3.8 5.00 ± 1.46 Coefficient of variation (%) 20.6 20.6 46.3 46.3 112 146 Reproduction number showed significant negative relationships with soil pH and total Cd concentration based on Pearson’s correlation analysis (Fig. 1 ). With the total soil Cd concentration, the obtained EC 50 value was 8.00 (5.92–24.1), a lower value than that in previous studies using artificial soils (Lock and Jassen, 2001a). This also indicates that Cd is not the only factor limiting F. candida reproduction in soils. The scatter plot of the reproduction number in different naturally Cd-contaminated soils is shown in Fig. S2. In the soils with lower Cd concentrations, the reproduction number was quite low, possibly due to inappropriate alkaline soil pH conditions for F. candida reproduction (Fountain and Hopkin, 2005 ). For example, in soil No. 2, the total Cd concentration was 0.62 mg kg –1 but the reproduction inhibition rate was 50%, resulting from the high pH value (8.47) (Table S1 ). Zhang et al. ( 2021 ) obtained similar results in which soil pH had more adverse effects on F. candida reproduction than metals in naturally contaminated soils, especially in calcareous soils. In addition, only when the metal detoxification level exceeds its capacity, it may cause adverse effects on individual and population numbers (Lock and Janssen, 2001b ; Howcroft et al., 2009 ). The Cd accumulated in soil animals may be formed as non-toxic substances by forming metal-proteins complexes. This metal-protein complex was observed by Morgan et al. ( 2004 ) and Vijver et al. ( 2006 ), who found that 70% of Cd in earthworms was sequestered by metallothionein (MT). Therefore, some Cd levels may be accumulated in the animal’s body with minimum toxic effects on the survival and reproduction of the testing organisms. Toxicity endpoint survival is complex according to soil characteristics and low sensitivity; therefore, further studies are required to better understand metal accumulation in soil animals. 3.2 Cadmium accumulation in F. candida The internal Cd concentrations in F. candida and the BAF values in the 28 naturally Cd-contaminated soils were 18.2 ± 3.83 mg kg − 1 (2.44–85.7 mg kg − 1 ) and 5.00 ± 1.46 (0.398–36.0) respectively, showing significant differences among different soils ( P 100, indicating a wide variation in soil properties and affecting Cd accumulation in F. candida . The Cd BAF values in 92.9% of the soil samples were > 1.0, suggesting a high ecological risk of Cd in soils and potential food chain risks, which should be a concern in these soils. Pearson correlation analysis (Fig. 1 ) shows that soil pH was the primary factor that negatively influenced the accumulated Cd concentration in F. candida and BAF-Cd values. Conversely, silt, total K, and total Cd concentrations in soils showed positive associations with the internal Cd concentration. Soil total K and available K concentrations also had positive relationships with the BAF. However, total Cd concentration in soils showed a negative association with the BAF. RDA analysis gave similar results (Fig. 3 ). A Monte Carlo permutation test in the RDA analysis shows that Log [Cd] and Log [BAF] had significant correlations with the selected soil properties (Pseudo-F = 10.5, P < 0.005). The first and second RDA components (RDA1 and RDA2) explained 50.4 and 39.7%, respectively, of the total variation, suggesting that the soil attributes were significantly highly diverse. Soil total Cd concentration was the dominant factor significantly restraining Cd accumulation, contributing 43.6% of the variance. Soil pH, K, and SOC made lesser contributions of 22.0, 12.6, and 6.2% of the total variation, respectively. Nutrient elements have been given less consideration in previous studies on metal accumulation (van Gestel, 2008 ; Liu et al., 2019 ). However, here, Pearson’s correlation analysis and RDA analysis both confirm that K was essential for Cd accumulation by F. candida . These results may be explained as follows. Firstly, K may exchange with Cd on soil particle surfaces, thereby increasing exchangeable/available Cd concentration in soils with further soil animal Cd assimilation (Wu et al., 2015 ; Wang et al., 2019 ). Wang et al. ( 2019 ) found that K addition had a significantly positive relationship with available Cd content in soils, and K increased Cd accumulation in plant tissues. Secondly, K generally tends to be deficient in Chinese agricultural soils, with an area of deficiency of up to 60% of agricultural soils (Chen and Zheng, 2004 ; Römheld and Kirkby, 2010 ). In this study, the average total K levels in the selected soils were 15.1 (3.16–26.9 g kg –1 ), slightly lower than the average K content in Chinese soils (Chen et al., 2020 ). The deficiency of K may therefore be a constraining factor in the selected soils for collembolan. Thirdly, K is an essential nutrient for the growth and development of organisms and is involved in numerous processes, such as the regulation of enzyme activity, membrane potential, cellular homeostasis, and stable protein synthesis (Tester, 2001 ; Luersen et al., 2016 ). Thus, K may increase the resistance of F. candida under metal exposure, further increasing Cd accumulation in the body. Studies also confirm that K channels control neuromuscular, mating, and locomotion behaviors in Caenorhabditis elegans (Abraham et al., 2010 ; LeBoeuf and Garcia, 2012 ; Luersen et al., 2016 ). Potassium channels have secretory and excretory functions and thus may promote the adsorption and accumulation of Cd (Piermarini et al., 2013 ; Wu et al., 2015 ). More importantly, Cd 2+ binds to K channels, and this influences the transfer and cytotoxicity of Cd (Wang et al., 2017 ). Thus, K is essential for Cd accumulation in F. candida . To provide direct evidence of the effects of soil K on Cd accumulation in F. candida , a K-spiked single-species test was conducted based on the OECD guidelines with some modifications (OECD, 2009 , Supplementary materials). The results confirm that soil K showed a positive relationship with Cd accumulation in F. candida (R 2 = 0.964, P < 0.001), with the body Cd concentration significantly increasing with increasing soil K concentration (Fig. S3, P < 0.001). The current study indicates the necessity of studying the effects of nutrient elements on metal accumulation in biota. 3.3 Comparison of Cd toxicity prediction performance between multiple linear regression and BPNN models Multiple linear regression models are commonly used to predict and estimate the dependent variable by the optimum combination of multiple independent variables (De’ath and Fabricius, 2000). Here, the survival rate of F. candida in naturally Cd-contaminated soils cannot be predicted by multiple linear regression analysis, therefore, the BPNN model was used. The reproduction inhibitory rate was predicted and compared by the BPNN and multiple linear regression models. All soil properties were input as variables and the reproduction inhibitory rate was well interpreted by stepwise multiple linear regression equations in the tested soils, but only soil total Cd concentration and soil pH were valid parameters. The stepwise regression equation was developed as follows: Reproduction inhibitory rate = 0.180 Log Cd T + 0.073 pH – 0.097 (R 2 = 0.302, P > 0.05) (8) Based on Pearson’s correlation analysis, soil total Cd concentration and soil pH had significant relationships with survival and reproduction inhibitory rate, thus, soil total Cd concentration and soil pH were set as input variables in the BPNN model, and survival/reproduction inhibitory rate was the output target. The number of the input layer was 2, the output layer was 1, the hidden layer was 5 based on the empirical formula, and the number of network iterations was 5000. In the BPNN model, 20 sets of samples were used for training, 4 were used for validation, and 4 for testing. Specific training, validation, and testing results are shown in Table S2 and the predictions of all the samples were significant ( P < 0.001). The prediction results were compared with the multiple linear regression model, as shown in Table 2 and Fig. 4 . The survival and reproduction inhibitory rates were successfully predicted by the BPNN model with R 2 = 0.797 and 0.930, respectively. The prediction accuracy of the reproduction inhibitory rate by the BPNN model was much higher than by multiple linear regression, with a higher R 2 and lower MAE, MRE, and RMSE. This indicates that the BPNN model is superior to the stepwise regression model. In addition, the larger soil total Cd concentration level (0.54 − 25.7 mg kg –1 ) indicated that the BPNN model was sensitive and suitable for wide soil metal concentrations. Therefore, the newly developed BPNN model can be useful for predicting Cd toxicity to F. candida in naturally contaminated soils. Table 2 Prediction performance parameters of the BPNN and stepwise multiple linear regression models for the estimation of Cd concentration in F. candida and Cd bioaccumulation factor (BAF). Target Model R 2 MAE MRE (%) RMSE Survival rate BPNN 0.797 3.27 0.042 4.69 Reproduction inhibitory rate BPNN 0.827 8.24 0.194 10.3 Multiple linear regression 0.302 16.3 0.318 20.7 Log [Cd] BPNN 0.961 0.003 0.014 0.077 Multiple linear regression 0.638 0.002 0.012 0.201 Log [BAF] BPNN 0.964 0.002 0.014 0.075 Multiple linear regression 0.591 0.002 0.011 0.194 3.4 Comparison between multiple linear regression and BPNN models in predicting Cd accumulation in collembolan The Cd accumulation in collembolan was interpreted well by stepwise multiple linear regression equations in the tested soils, and only soil total Cd and pH were valid parameters selected through stepwise multiple linear regression analysis, and the developed equations for Log [Cd] and Log [BAF] were as follows: Log [Cd] = 0.568 Log Cd T – 0.154 pH + 1.748 (R 2 = 0.638, P < 0.001) (9) Log [BAF] = – 0.432 Log Cd T – 0.154 pH + 1.748 (R 2 = 0.591, P < 0.001 (10) Based on Pearson’s correlation analysis and RDA analysis, soil total Cd concentration, soil pH, SOC, and soil K play significant roles in Cd bioaccumulation, therefore, were selected as input variables in the BPNN model. Log [Cd] or Log [BAF] was the output target. The number of input layer was 4, the number of output layer was 1, the number of hidden layers was 5 based on the empirical formula, and the number of network iterations was 5000. In the BPNN model, 20 sets of samples were used for training, 4 sets for validation, and 4 sets for testing. Specific results on training, validation, and testing results are also shown in Table S2 and the predictions of all samples were significant ( P < 0.001). To retain the consistency of selected soil properties in all the models, the multiple linear regression equations entered were as follows: Log [Cd] = 0.623 Log Cd T – 0.144 pH + 0.228 Log K T – 0.319 Log SOC + 1.815 (R 2 = 0.687, P < 0.001) (11) Log [BAF] = – 0.377 Log Cd T – 0.144 pH + 0.228 Log K T – 0.319 Log SOC + 1.815 (R 2 = 0.646, P < 0.001) (12) The prediction performance of entered multiple linear regression models slightly improved to the stepwise multiple linear regression models, therefore, entered multiple linear regression equations were not compared with the BPNN model. Although the MAE and MRE of the BPNN and multiple linear regression models were similar (Table 2 ), the BPNN model had higher R 2 and lower RMSE than the multiple linear regression model. The prediction accuracy of the BPNN model of Log [Cd] and Log [BAF] was superior to the multiple linear regression model. Notably, the correlation between the Log [Cd] predicted by the BPNN model and the measured Log [Cd] was stronger than that by the multiple linear regression model, with the predicted and measured values very close to the 1:1 line among a wide range of soil Cd concentrations (R 2 = 0.910), but the correlation by multiple linear regression model was lower (R 2 = 0.638, Fig. 4 C). A similar trend was found in Log [BAF] (Fig. 4 D). The correlation between the Log [BAF] predicted by the BPNN model and the measured Log [BAF] was higher than that by the multiple linear regression model, which the correlation between the predicted and measured values was very close to the 1:1 line (R 2 = 0.979). In addition, with increasing Log [BAF], the predicted values by the multiple linear regression model gradually deviated from the 1:1 line, especially when the measured Log [BAF] exceeded 1.00. Recently, Wang et al. ( 2021 ) obtained similar results showing that the multiple linear regression model cannot predict well the Zn BAF in rice grain when the BAF exceeds 0.40. Thus, the multiple linear regression model is not suitable for the prediction of high Cd BAF and is limited to a wide range of soil properties. The BPNN model accounted for the complex nonlinear relationships between Cd accumulation in F. candida and soil properties and showed superiority in predicting internal Cd concentrations and Cd BAF values in F. candida over a wide range of Cd concentrations. 3.5 Application of the BPNN models for predicting Cd ecotoxicity in Southern Chinese soils With the developed BPNN models, the predicted Cd toxicity to and accumulation in soil collembolans can be assessed without conducting toxicity tests. Here, 57 samples were also collected from southern China with a wide range of soil properties (soil pH 4.14–8.36, soil total Cd 0.15–25.5 mg kg –1 , SOC 5.94–47.8 g kg –1 , Table S3), and the accumulated Cd concentrations in F. candida and BAF values were predicted based on the BPNN models with the input layers soil total Cd, soil pH, SOC, and K, the number of input layer was 4, the number of output layer was 1, the number of hidden layer was 5, and the number of network iterations was 5000. Multiple linear regression models were also used for comparison with the valid parameters of the total Cd and pH in soils (Fig. 5 ). The predicted values were visualized on sensitive maps, and different colors were assigned to indicate the level of risk. The predicted values of the internal Cd concentration in F. candida and BAF-Cd values, as observed by the BPNN models, were generally higher than those by the multiple linear regression models. Notably, the BPNN models better identified the regions of higher risk. This was consistent with our previous results in which multiple linear regression underestimated Cd accumulation in F. candida (Fig. 4 ). The mathematic models are also beneficial in the quantification of influencing factors and the development of further strategies to control metal risks in soils. There is a tendency for soil pH to decrease and SOC to increase in agricultural soils (Guo et al., 2010 ; Sun et al., 2012 ). Therefore, the effects of changes in soil properties on the evaluation of Cd ecological risks in soils need to be considered. Three soils (Table S1 , Soil No. 15, 25, and 6) with different pH values (4.73, 5.85, and 7.36) were used to demonstrate the effects of changes in pH (± 0.5 and 1 unit) and SOC (+ 1% and 3%) on the BAF-Cd values. The results showed that soil pH was more important than SOC in Cd accumulation in F. candida (Fig. 6 ). Increases in SOC of 1% and 3% had no significant effect on the BAF-Cd values, but pH was significantly negatively associated with the BAF-Cd values. This is probably because decreasing soil pH increased Cd bioavailability, further increasing the accumulated Cd in the collembolan bodies. Thus, remediation methods should be taken to increase soil pH to lower the ecological risks of Cd in soils. The input data from the BPNN model is easily derived from soil chemical analyses or obtained from databases of national soil surveys (Zhao et al., 2015 ), and thereby the Cd toxicity to and accumulation in F. candida in soils can be predicted by the BPNN model without conducting toxicity tests. 4. Conclusions Cadmium toxicity to soil animals and accumulation in their bodies are intricate and were substantially affected by soil properties in naturally contaminated soils, and soil pH and soil Cd concentration were the main factors driving Cd toxicity to and accumulation in soil collembolan. When the total Cd concentration was < 3 mg kg − 1 , the high soil pH was the main factor restricting collembolan survival rather than soil total Cd concentration. Pearson’s correlation, RDA analysis, and the K-spiked test all confirm that K was essential for Cd accumulation in F. candida . This supports the necessity of taking the effects of nutrient elements (K) on metal (Cd) accumulation in soil biota into account. Compared with the multiple linear regression model, the BPNN model developed showed higher performance in estimating the toxicity to and accumulation in F. candida under a wide range of soil properties and better identified high-risk regions of Cd ecotoxicology in southern China. The developed BPNN model successfully recognized and resolved complex nonlinear relationships between Cd ecotoxicology in F. candida and soil properties in naturally contaminated soils and can be used to predict Cd ecotoxicology generally in untested soils. It can therefore be used as an alternative tool to evaluate and monitor Cd ecotoxicity in practice. Declarations Funding This study was funded by the National Natural Science Foundation of China (41977136). Author contributions: Simin Li: Writing original draft, Investigation, Methodology, Data curation, Visualization, Formal analysis; Zhu Li: Review & editing, Funding acquisition; Xin Ke: Review & Editing; Worachart Wisawapipat: Review & Editing; Peter Christie: Review & Editing; Longhua Wu: Review & Editing. Data Availability: Not applicable Ethical approval: The authors hereby approve that principles of ethical and professional conduct have been followed in the work. Consent to participate: The present research does not involve any human or animal participation. Consent for publication: The authors and the responsible authorities at the institute/organization where this work has been carried out give their explicit consent to submit and publish the work in ESPR if found suitable. Competing interests: The authors declare no competing interests. References Abraham, L. S., Oh, H. J., Sancar, F., Richmond, J. E., Kim, H., 2010. An alpha-catulin homologue controls neuromuscular function through localization of the dystrophin complex and BK channels in Caenorhabditis elegans . PLoS Genetics 6, 1001077. Ali, H., Khan, E., Ilahi, I., 2019. Environmental chemistry and ecotoxicology of hazardous heavy metals: Environmental persistence, toxicity, and bioaccumulation. Journal of Chemistry 6730305. Chen, F., Zheng, S. X., 2004. Research progress of efficient potassium application technology for crops in South China. Soils and Fertilizers 6, 29–32. Chen, X. Q., Li, T., Lu, D. J., Cheng, L., Zhou, J. M., Wang, H. Y., 2020. Estimation of soil available potassium in Chinese agricultural fields using a modified sodium tetraphenyl boron method. Land Degradation & Development 31, 1737-1748. Christiani, D. C., Zhou, W., 2016. Hormesis: The new approach in risk assessment? Human & Experimental Toxicology 21, 399–400. Coyle, D. R., Nagendra, U. J., Taylor, M. K., Campbell, J. H., Cunard, C. E., Joslin, A. H., Jr, M. A. C., 2017. Soil fauna responses to natural disturbances, invasive species, and global climate change: Current state of the science and a call to action. Soil Biology & Biochemistry 110, 116‒133. De'ath, G., Fabricius, K. E., 2000. Classification and regression trees: A powerful yet simple technique for ecological data analysis. Ecology 81, 3178‒3192. Fountain, M. T., Hopkin, S. P., 2005. Folsomia candida (collembola): A standard soil arthropod. Annual Review of Entomology 50, 201‒222. Gospodarek, J., Boliglowa, E., Glen-Karolczyk, K., 2020. Impact of Nonchemical Protection of Broad Bean on Epigeic and Soil Arthropodofauna-Analysis in Field-Realistic Conditions. Agronomy-Basel 10, 211. Guo, J. H., Liu, X., Zhang, Y., Shen, J. L., Han, W. X., Zhang, W. F., Christie, P., Goulding, K. W., Vitousek, P. M., Zhang, F. S., 2010. Significant acidification in major Chinese croplands. Science 327, 1008–1010. Howcroft, C. F., Amorim, M. J. B., Gravato, C., Guilhermino, L., Soares, A. M. V. M., 2009. Effects of natural and chemical stressors on Enchytraeus albidus : Can oxidative stress parameters be used as fast screening tools for the assessment of different stress impacts in soils ?. Environment International 35, 318‒324. Kim, S. W., An, Y. J., 2014. Jumping behavior of the springtail Folsomia candida as a novel soil quality indicator in metal-contaminated soils. Ecological Indicators 38, 67–71. LeBoeuf, B., Garcia, L. R., 2012. Cell excitability necessary for male mating behavior in Caenorhabditis elegans is coordinated by interactions between big current and ether-A-Go-Go Family K + Channels. Genetics 3, 1025–1041. Li, X. N., Cheng, X., Wu, W. J., Wang, Q. H., Tong, Z. Y., Zhang, X. Q., Deng, D. H., Li, Y. H., 2019. Forecasting of bioaerosol concentration by a Back Propagation neural network model. Science of the Total Environment 698, 134315. Liu, H., Xuan, L., Zhou, J., Zhou, D. M., Wang, Y. J., 2019. Effects of soil properties on cadmium toxicity to Folsomia candida (Collembola). Bulletin of Environmental Contamination and Toxicology 103, 90–97. Liu, M. L., Liu, X. N., Wu, M. X., Li, L. F., Xiu, L. N., 2011. Integrating spectral indices with environmental parameters for estimating heavy metal concentrations in rice using a dynamic fuzzy neural-network model. Computers & Geosciences 37, 1642–1652. Lock, K., Janssen, C. R., 2001a. Cadmium toxicity for terrestrial invertebrates: taking soil parameters affecting bioavailability into account. Ecotoxicology 10, 315–322. Lock, K., Janssen, C. R., 2001b. Zinc and cadmium body burdens in terrestrial oligochaetes: use and significance in environmental risk assessment. Environmental Toxicology and Chemistry 20, 2067–2072. Lu, R. K., 2000. Analytical Methods for Soil and Agrochemistry. Agricultural Science and Technology Press, Beijing. Luersen, K., Gottschling, D. C., Doring, F., 2016. Complex locomotion behavior changes are induced in Caenorhabditis elegans by the lack of the regulatory leak K+ Channel TWK-7. Genetics 2, 683–701. Luo, W., Verweij, R. A., van Gestel, C. A. M., 2014. Assessment of the bioavailability and toxicity of lead polluted soils using a combination of chemical approaches and bioassays with the collembolan Folsomia candida . Journal of Hazardous Materials 280, 524–530. Mark, P. M., 2008. Hormesis defined. Ageing Research Reviews 7, 1–7. Morgan, A. J., Stu¨rzenbaum, S. R., Winters, C., Grime, G. W., Aziz, N. A., Kille, P., 2004. Differential metallothionein expression in earthworm ( Lumbricus rubellus ) tissues. Ecotoxicology and Environmental Safety 57, 11–19. Noh, H., Zhang, Q., Shin, B., Han, S., Feng L., 2006. A neural network model of maize crop nitrogen stress assessment for a multi-spectral imaging sensor. Biosystems Engineering, 94, 477–485. OECD, 2009. Guidelines for the testing of chemicals no. 232, collembolan reproduction test in soil. Organization for Economic Cooperation and Development. Piermarini, P. M., Rouhier, M. F., Schepel, M., Kosse, C., Beyenbach, K. W., 2013. Cloning and functional characterization of inward-rectifying potassium (Kir) channels from Malpighian tubules of the mosquito Aedes aegypti . Insect Biochemistry and Molecular Biology, 43, 75–90. Römheld, V., Kirkby, E. A., 2010. Research on potassium in agriculture: needs and prospects. Plant and Soil 335, 155–180. Sahraoui, A. S., Verweij, R. A., Belhiouani, H., Cheriti, O., van Gestel, C. A. M., Sahli, L., 2021. Dose-dependent effects of lead and cadmium and the influence of soil properties on their uptake by Helix aspersa : an ecotoxicity test approach. Ecotoxicology 30, 331–342. Sun, X. L., Zhao, Y. G., Wu, Y. J., Zhao, M. S., Wang, H. L., Zhang, G. L., 2012. Spatio-temporal change of soil organic matter content of Jiangsu Province, China, based on digital soil maps. Soil Use and Management 28, 318–328. Tester, M., 2001. Partitioning of nutrient transport processes in roots. Journal of Experimental Botany 52, 653–653. Tumbo, S. D., Wagner, D. G., Heinemann, P. H., 2002. Hyperspectral–based neural network for predicting chlorophyll status in corn. Transactions of the ASAE, 45, 825–832. van Gestel, C. A. M., 2008. Physico-chemical and biological parameters determine metal bioavailability in soils. Science of the Total Environment 406, 385–395. Vijver, M. G., van Gestel, C. A. M., van Straalen, N. M., Lanno, R. P., Peijnenburg, W. J. G. M., 2006. Biological significance of metals partitioned to subcellular fractions within earthworms ( Aporrectodea caliginosa ). Environmental Toxicology and Chemistry 25, 807–814. Wang, L. H., He, J. F., Xia, A., Cheng, M. Z., Yang, Q., Du, C. L., Wei, H. Y., Huang, X. H., Zhou, Q., 2017. Toxic effects of environmental rare earth elements on delayed outward potassium channels and their mechanisms from a microscopic perspective. Chemosphere 181, 690–698. Wang, K., Fu, G. P., Yu, Y., Wan, Y. A., Liu, Z., Wang, Q., Zhang, J. S., Li, H. F., 2019. Effects of different potassium fertilizers on cadmium uptake by three crops. Environmental Science and Pollution Research 26, 27014–27022. Wang, Q. Y., Sun, J. Y., Hu, N. W., Wang, T. Y., Yue, J., Hu, B., Yu, H. W., 2022. Effects of soil aging conditions on distributions of cadmium distribution and phosphatase activity in different soil aggregates. Science of the Total Environment 834, 155440. Wang, S. X., Zhang, N., Wu, L., Wang, Y. M., 2016. Wind speed forecasting based on the hybrid ensemble empirical mode decomposition and GA-BP neural network method. Renewable Energy 94, 629–636. Wang, Y. Z, Yu, T., Yang, Z. F., Bo, H. Z., Lin, Y., Yang, Q., Liu, X., Zhang, Q. Z., Zhuo, X. X., Wu, T. S., 2021. Zinc concentration prediction in rice grain using back-propagation neural network based on soil properties and safe utilization of paddy soil: A large-scale field study in Guangxi, China. Science of the Total Environment 798, 149270. Wu, Y., Baum, M., Huang C. L., Rodan, A. R., 2015. Two inwardly rectifying potassium channels, Irk1 and Irk2, play redundant roles in Drosophila renal tubule function. American Journal of Physiology-Regulatory Integrative and Comparative Physiology 309, R747–R756. Zhang, X. Y., Zhang, Y. W., Zhang, H. Y., Yang, Q. Y., Wang, H. Y., Zhang, G. C., 2015. Preparation, characterization, and antibacterial activity of octenyl succinic anhydride modified inulin. International Journal of Biological Macromolecules 78, 79–86. Zhang, Y. B., Li, Z., Ke, X., Wu, L. H., Christie, P., 2021. Multigenerational exposure of the collembolan Folsomia candida to soil metals: Adaption to metal stress in soils polluted over the long term. Environmental Pollution 292, 118242. Zhao, F. J., Ma, Y., Zhu, Y. G., Tang, Z., McGrath, S. P., 2015. Soil contamination in China: Current status and mitigation strategies. Environmental science & technology 49, 750–759. Supplementary Files 5Supportinginformation1.docx Cite Share Download PDF Status: Published Journal Publication published 02 Mar, 2024 Read the published version in Environmental Science and Pollution Research → Version 1 posted Reviewers agreed at journal 15 Jan, 2024 Reviewers invited by journal 15 Jan, 2024 Editor invited by journal 15 Jan, 2024 Editor assigned by journal 04 Jan, 2024 First submitted to journal 28 Dec, 2023 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-3740915","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":267321340,"identity":"22b9e640-dfd2-48bf-b9c2-13a60aa7db53","order_by":0,"name":"Simin Li","email":"","orcid":"","institution":"Institute of Soil Science Chinese Academy of Sciences","correspondingAuthor":false,"prefix":"","firstName":"Simin","middleName":"","lastName":"Li","suffix":""},{"id":267321341,"identity":"d3723ca8-e312-43b6-9122-3544aee5ba4c","order_by":1,"name":"Zhu Li","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAuUlEQVRIiWNgGAWjYBACPmYGNgaGCgkStLCBtZwhSQsIMbaR4jA2dvZnj3nnWdjLtx9/wPCjhkHenAiHpRvzbpNI3HAmx4Cx5xiD4c4GwlqOSQO1JBgw5DAw8DYwJBgcIKiFsU2ad46EvXz/8weMf4nTwswmzdsgwdhwI8GAmUhb2Ngk5xwD+uXGG4PDMsckDDcQ0sLPf/yZxJuaOqDD0h8+fFNjI0/QFhQAVExKnI6CUTAKRsEowAkAbykylLEWMacAAAAASUVORK5CYII=","orcid":"","institution":"Institute of Soil Science Chinese Academy of Sciences","correspondingAuthor":true,"prefix":"","firstName":"Zhu","middleName":"","lastName":"Li","suffix":""},{"id":267321342,"identity":"6e17bcc6-dbb2-4fac-9f79-242332a9eb80","order_by":2,"name":"Xin Ke","email":"","orcid":"","institution":"CAS Center for Excellence in Molecular Plant Sciences: Institute of Plant Physiology and Ecology Shanghai Institutes for Biological Sciences","correspondingAuthor":false,"prefix":"","firstName":"Xin","middleName":"","lastName":"Ke","suffix":""},{"id":267321343,"identity":"484f061f-86c1-49ad-a282-3a03a3add239","order_by":3,"name":"Worachart Wisawapipat","email":"","orcid":"","institution":"Kasetsart University","correspondingAuthor":false,"prefix":"","firstName":"Worachart","middleName":"","lastName":"Wisawapipat","suffix":""},{"id":267321344,"identity":"1590e570-e6ca-4a09-9a89-a9ad8a4b899a","order_by":4,"name":"Peter Christie","email":"","orcid":"","institution":"Institute of Soil Science Chinese Academy of Sciences","correspondingAuthor":false,"prefix":"","firstName":"Peter","middleName":"","lastName":"Christie","suffix":""},{"id":267321345,"identity":"451b1c17-d27f-46b9-b476-55b4a3c56ef6","order_by":5,"name":"Longhua Wu","email":"","orcid":"","institution":"Institute of Soil Science Chinese Academy of Sciences","correspondingAuthor":false,"prefix":"","firstName":"Longhua","middleName":"","lastName":"Wu","suffix":""}],"badges":[],"createdAt":"2023-12-12 02:26:03","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-3740915/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-3740915/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1007/s11356-024-32638-x","type":"published","date":"2024-03-02T15:02:14+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":49741365,"identity":"e4d450bc-628e-4eeb-b1b8-5fd9c40c924a","added_by":"auto","created_at":"2024-01-17 09:25:30","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":84291,"visible":true,"origin":"","legend":"\u003cp\u003ePearson’s correlation analysis among different soil properties and the survival and reproduction number, Log-Cd in body tissue, and Log-BAF of \u003cem\u003eF. candida \u003c/em\u003ein 28 naturally Cd-contaminated soils. All soil properties except soil pH are log-transformed.\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-3740915/v1/cbeebe84243bbbcb9332e5d2.png"},{"id":49741082,"identity":"4e1ea76b-a9e2-44af-815d-8d419efb6797","added_by":"auto","created_at":"2024-01-17 09:17:30","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":57508,"visible":true,"origin":"","legend":"\u003cp\u003eSee image above for figure legend\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-3740915/v1/2ea14e1c94f2c9e2e3d48a86.png"},{"id":49741971,"identity":"54a1f880-9eeb-4c02-90c5-2815e8428704","added_by":"auto","created_at":"2024-01-17 09:33:30","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":46495,"visible":true,"origin":"","legend":"\u003cp\u003eRedundancy analysis between Log [Cd] and Log [BAF] and selected soil properties. Soil CEC was excluded due to its collinearity with the selected variables. Blue crosses, Cd accumulation characteristics of \u003cem\u003eF. candida\u003c/em\u003e; red crosses, selected soil property variables.\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-3740915/v1/fb46550ca32e254524afaca5.png"},{"id":49741087,"identity":"46d1b9a8-01d6-45de-93d5-41e19c81407b","added_by":"auto","created_at":"2024-01-17 09:17:30","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":251523,"visible":true,"origin":"","legend":"\u003cp\u003eLinear correlations between the measured and predicted survival rates by the BPNN model (A), the measured and predicted reproduction inhibitory rate by the BPNN model and stepwise multiple regression model (B), the measured and predicted Cd concentration in body tissues by the BPNN model and stepwise multiple regression model (C) and BAF of Cd by the BPNN and stepwise multiple regression model (D). Dark and light red (or blue) shaded areas represent 95% confidence intervals of regression and prediction, respectively.\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-3740915/v1/85748069d43beaaf2d27689a.png"},{"id":49741083,"identity":"db29786d-128e-4e96-adb0-dc61e14071ba","added_by":"auto","created_at":"2024-01-17 09:17:30","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":133475,"visible":true,"origin":"","legend":"\u003cp\u003ePredicted body Cd concentration (A, B) and BAF-Cd (C, D) in \u003cem\u003eF. candida\u003c/em\u003e by the multiple linear regression models (A, C) and the BPNN models (B, D) in 57 soil samples.\u003c/p\u003e","description":"","filename":"5.png","url":"https://assets-eu.researchsquare.com/files/rs-3740915/v1/aea2efd739a4d95de7a13bac.png"},{"id":49741086,"identity":"62786da2-756e-4b01-aeee-6e441d6b3f97","added_by":"auto","created_at":"2024-01-17 09:17:30","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":157345,"visible":true,"origin":"","legend":"\u003cp\u003ePredicted Log BAF-Cd values in\u003cem\u003e F. candida\u003c/em\u003e by the BPNN model in three soils with different pH values when soil pH (± 0.5 and 1 unit) and SOC (+1% and 3%) were changed.\u003c/p\u003e","description":"","filename":"6.png","url":"https://assets-eu.researchsquare.com/files/rs-3740915/v1/bfc216d0134a686614aed90e.png"},{"id":51958700,"identity":"e1e1069e-57f2-481a-b2bc-dc44697b364a","added_by":"auto","created_at":"2024-03-04 15:18:02","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1087255,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-3740915/v1/41977fdb-a2dc-4195-91bb-6cd8918865dc.pdf"},{"id":49741088,"identity":"f581900c-6f28-4505-be64-f60ddaeb6001","added_by":"auto","created_at":"2024-01-17 09:17:30","extension":"docx","order_by":5,"title":"","display":"","copyAsset":false,"role":"supplement","size":161274,"visible":true,"origin":"","legend":"","description":"","filename":"5Supportinginformation1.docx","url":"https://assets-eu.researchsquare.com/files/rs-3740915/v1/73de395f1c7842c4a45f7ea5.docx"}],"financialInterests":"","formattedTitle":"Cadmium toxicity to and accumulation in a soil collembolan (Folsomia candida): major factors and prediction using a back-propagation neural network mode","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eSoil cadmium (Cd) pollution is a major concern globally. Combustion, traffic, metal industries, landfills, sewage sludges, mining, and fertilizers are primary sources of soil Cd contamination, which represents a high risk to soil biota (Zhao et al., \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2015\u003c/span\u003e; Ali et al., \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). Soil-animal-based soil ecotoxicology methods are commonly used to predict metal ecological risks. Collembolans are widespread in soils and play an important role in nutrient cycling, energy transfer, and maintenance of biodiversity (Fountain and Hopkin, \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2005\u003c/span\u003e; Howcroft et al., \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2009\u003c/span\u003e). They generally live near the soil surface and are vulnerable to soil disturbance. Several collembolan species have been used as standard soil ecotoxicological test animals. Notably \u003cem\u003eFolsomia candida\u003c/em\u003e, because of its sensitivity to pollutants, ease of culture, short life cycle, and high reproduction rate et al. (Coyle et al., \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2017\u003c/span\u003e). Laboratory studies show that high Cd concentrations significantly decreased collembolan survival and reproduction rates, growth (body weight and body length), and locomotor behavior (Lock and Janssen, \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2001a\u003c/span\u003e, \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003eb\u003c/span\u003e; Kim and An, \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e2014\u003c/span\u003e; Liu et al., \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). Therefore, the endpoints (survival, reproduction, and growth rates) in soil ecological tests commonly be used as bioindicators to indicate soil ecological risks. However, those endpoints are not trustworthy in naturally contaminated soils. Firstly, the determined effective concentration that decreased the reproduction rate by 50% (EC\u003csub\u003e50\u003c/sub\u003e) obtained in labs (newly spiked artificial soils) was 158 (137\u0026ndash;184) mg kg\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e (Lock and Janssen, \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2001a\u003c/span\u003e), a significantly higher value than those in most naturally contaminated agricultural soils examined in China (Zhao et al., \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2015\u003c/span\u003e). In addition, Zhang et al. (\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e2021\u003c/span\u003e) propose that soil pH, rather than soil total metal concentration, is the main factor restricting collembolan survival and reproduction rates in naturally contaminated agricultural soils. Thus, in natural soils, studies on the survival and reproduction of collembolans in naturally metal-contaminated soils may not accurately reflect metal availability, while studies on metal accumulation in collembolans are increasing because they are indicative of metal bioavailability in soils and the potential risks to the human food chain (Sahraoui et al., \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). The internal pollutant concentration in biota is a good biological indicator of organisms exposed to polluted areas, and methods that allow the prediction of metal bioavailability to soil biota are needed for soil health and soil ecological risk assessment.\u003c/p\u003e \u003cp\u003eStudies indicate that soil organic matter content, soil pH, and clay, iron, and manganese oxide contents, etc. have significant relationships with Cd toxicity to and accumulation in collembolans (Lock and Janssen, \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2001a\u003c/span\u003e, \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003eb\u003c/span\u003e; van Gestel, \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e2008\u003c/span\u003e; Liu et al., \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). In addition, these soil properties are commonly incorporated into empirical models (single and multiple linear regression equations) to build empirical equations for the prediction of Cd toxicity to and accumulation in collembolans (Liu et al., \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). However, most conclusions are generally drawn in newly spiked artificial soils and/or standard LUFA 2.2 soil, which set soil parameters as a single variable with different ranges (Speyer, Germany). These results can be inconsistent when compared with naturally contaminated soils that show large ranges of soil properties (Lock and Janssen, \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2001a\u003c/span\u003e, \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003eb\u003c/span\u003e; Liu et al., \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). Metal bioavailability may be lower in natural field soils due to aging effects (Wang et al., \u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e2022\u003c/span\u003e), and the empirical equations obtained are conditional and not applicable to other soil types (Liu et al., \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; Wang et al., \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). In addition, soil properties are associated with each other, linear regression analysis cannot allow for their complex and nonlinear calculations, and they may therefore be inefficient under a wide range of conditions (De\u0026rsquo;ath and Fabricius, 2000). Therefore, factors influencing Cd toxicity to and accumulation in collembolans under naturally Cd-contaminated soils, and accurate, rapid, and universal models for the prediction of Cd toxicity to and accumulation in collembolans under a wide range of soil properties require further development.\u003c/p\u003e \u003cp\u003eThe back-propagation neural network (BPNN) model has recently been proven to be a rapid, accurate, universal, and complexity-solved model (Wang et al., \u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e2016\u003c/span\u003e). It is one of the most extensively used artificial neural network (ANN) models and is a multi-layer feed-forward neural network model trained according to the error reverse propagation algorithm (Tumbo et al., \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e2002\u003c/span\u003e; Noh et al., \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e2006\u003c/span\u003e). The BPNN comprises numerous neurons and simple units of parallel algorithms and thus has a strong fitting ability and allows complex calculations in practical applications (Noh et al., \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e2006\u003c/span\u003e; Wang et al., \u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e2016\u003c/span\u003e). This model has been successfully used in numerous fields of research. For example, Li et al. (\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2019\u003c/span\u003e) identified the interactive relationships between meteorological factors and bioaerosols to effectively forecast the bioaerosol concentration in south central China by the BPNN model. In the soil-plant system, the BPNN model has successfully calculated and predicted the complex internal factors of metal migration and accumulation from soil to plant tissues. Liu et al. (\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2011\u003c/span\u003e) used an ANN model and estimated Cd and copper (Cu) concentrations in rice grains by integrating the spectral indices with environmental parameters. Wang et al. (\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e2021\u003c/span\u003e) used the BPNN model to predict the Zn bioaccumulation factor (BAF) in rice grains based on soil properties and safe utilization in a large-scale field study, and the prediction performance was significantly higher than that of multiple linear regression models. Numerous studies confirm that BPNN is a useful tool for minimizing errors of prediction and allowing calculations under complicated conditions than multiple linear regression models (Liu et al., \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2011\u003c/span\u003e; Li et al., \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; Wang et al., \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). In addition, the BPNN model calculation sets training, validation, and testing data to avoid overfitting situations, and is thus also available for untested data, increasing efficiency and cost-effectiveness (Wang et al., \u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e2016\u003c/span\u003e). However, to our knowledge, there is little or no information on its application to soil-animal systems, which may help us better predict metal bioavailability and ecotoxicity to soil biota in naturally contaminated soils with a wide range of soil properties.\u003c/p\u003e \u003cp\u003eThe objectives here were therefore to (1) identify the Cd toxicity effects and accumulation characteristics of \u003cem\u003eF. candida\u003c/em\u003e in a wide range of naturally Cd-contaminated soils, (2) identify the main factors affecting Cd toxicity to and accumulation in \u003cem\u003eF. candida\u003c/em\u003e, (3) develop BPNN models to predict Cd toxicity to and accumulation in \u003cem\u003eF. candida\u003c/em\u003e and compare the prediction performance with multiple linear regression models, and (4) further use the developed BPNN models in Cd bioaccumulation prediction and monitoring. The aim was to build accurate, rapid, and widely applicable methods to predict and assess Cd ecotoxicity and accumulation in naturally contaminated soils with a wide range of soil properties.\u003c/p\u003e"},{"header":"2. Materials and methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003e2.1 Test soils and organism\u003c/h2\u003e \u003cp\u003eA total of 28 naturally Cd-contaminated topsoil samples (0\u0026ndash;20 cm) with a wide range of soil properties were collected from eight Chinese provinces, in southern China. The basic soil properties were determined using standard methods (Lu, \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2000\u003c/span\u003e) and are shown in Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e, namely soil pH (soil: water ratio 1:2.5 (w/v)), soil organic carbon content (SOC), cation exchange capacity (CEC), soil texture (clay), total and available nitrogen contents (TN, AN), total and available phosphorus contents (TP, AP), total and available potassium contents (TK, AK), and total Cd, Fe, and Mn contents. The ranges of soil pH, SOC content, and soil total Cd content were 4.73\u0026ndash;8.4, 6.55\u0026ndash;41.6 g kg \u003csup\u003e\u0026ndash;1\u003c/sup\u003e, and 0.54\u0026ndash;25.7 mg kg \u003csup\u003e\u0026ndash;1\u003c/sup\u003e, respectively. After air-drying and sieving through a 2-mm mesh, the soils were equilibrated for 7 days at 50% water holding capacity before testing.\u003c/p\u003e \u003cp\u003e \u003cem\u003eF. candida\u003c/em\u003e is widespread in soils and has been widely used in ecotoxicity tests (OECD, \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2009\u003c/span\u003e). Individuals were originally obtained from Aarhus University, Denmark, and were subsequently cultured in our laboratory for \u0026gt;\u0026thinsp;10 years. They are reared in Petri dishes (height 10 mm, diameter 90 mm) with a layer of moist plaster mixed with activated charcoal (9:1 w/w) at 20\u0026thinsp;\u0026plusmn;\u0026thinsp;1\u0026deg;C and 75% relative humidity with a photoperiod of 16:8 h (light/dark). The animals were fed dried baker's yeast (Angel Yeast Co., Ltd., Yichang, Hubei, China). Distilled water was added to maintain the moisture content. \u003cem\u003eF. candida\u003c/em\u003e individuals synchronized for 10\u0026ndash;12 days were used in the tests (OECD, \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2009\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003e2.2 Single-species toxicity test\u003c/h2\u003e \u003cp\u003eThe test procedure was based on the OECD guidelines with some modifications (OECD, \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2009\u003c/span\u003e). Ten cleaned and gut-vomited \u003cem\u003eF. candida\u003c/em\u003e individuals were introduced into each test vessel (5.5 cm diameter, 250 mL volume) containing 30 g soil. There were three replicates of each treatment. The test vessel was covered with plastic film with small holes punched through. Field conditions were mimicked by providing no extra food. The animals were cultured in the same conditions as described above. The containers were weighed twice weekly to maintain water balance. After 28 days, soil animals were extracted using a controlled temperature gradient extractor. The numbers of juveniles and adults were counted under a stereomicroscope (Leica S8 APO, Wetzlar, Germany).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003e2.3 Chemical analysis\u003c/h2\u003e \u003cp\u003eThe extracted \u003cem\u003eF. candida\u003c/em\u003e individuals were starved for 2 days to remove interfering substances from their gut contents. They were then rinsed with ultrapure water and freeze-dried (\u0026minus;\u0026thinsp;80\u0026deg;C). Aliquots (~\u0026thinsp;20\u0026ndash;300 \u0026micro;g) of the animals were weighed (precision\u0026thinsp;\u0026plusmn;\u0026thinsp;1 \u0026micro;g, XS3DU, Mettler Toledo, Columbus, OH) and digested with 300 \u0026micro;L of a mixture of HNO\u003csub\u003e3\u003c/sub\u003e (65%; ultrapure) and H\u003csub\u003e2\u003c/sub\u003eO\u003csub\u003e2\u003c/sub\u003e (30%; ultrapure) (1:1 v/v) in 5-mL Teflon plastic digestion vessels at 105\u0026deg;C which were sealed and dried in an oven for 6 h. After digestion and cooling, the vessels were transferred to an electric heating plate and heated at 120\u0026deg;C to dryness. The samples were stored in a refrigerator at 4\u0026deg;C until analysis. Before analysis, 2 mL 5\u0026permil; HNO\u003csub\u003e3\u003c/sub\u003e was added to dissolve the residue, and the samples were analyzed by ICP-MS (NeXion 2000, Perkin Elmer, Waltham, MA) after filtration. A certified reference material (GBW08551, pork liver (Institute of Geophysical and Geochemical Exploration, Langfang, Hebei, China)) was included for quality control, with recoveries ranging from 89.4 to 108%.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec6\" class=\"Section2\"\u003e \u003ch2\u003e2.4 Description of the BPNN model\u003c/h2\u003e \u003cp\u003eThe BPNN model is a feed-forward neural network consisting mainly of input layers, multiple hidden layers, and output layers. Each layer consists of several nodes (Wang et al., \u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e2016\u003c/span\u003e). The number of soil property factors is selected as the input layer based on the correlation between soil properties and output target (linear regression and redundancy analysis (RDA)), the output layer is 1, and the number of hidden layers is determined based on the empirical formula:\u003c/p\u003e \u003cp\u003en1= \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\sqrt{ n +m}\\)\u003c/span\u003e\u003c/span\u003e+ k (1)\u003c/p\u003e \u003cp\u003ewhere n1 is the number in the hidden layer, n is the number in the input layer, m is the number in the output layer, and k is a regulation constant from 1 to 10. Seventy percent of samples are used for training, 15% for validation, and 15% for testing (Tumbo et al., \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e2002\u003c/span\u003e; Noh et al., \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e2006\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec7\" class=\"Section2\"\u003e \u003ch2\u003e2.5 Statistical analysis\u003c/h2\u003e \u003cp\u003eAll data are expressed as mean\u0026thinsp;\u0026plusmn;\u0026thinsp;standard error (SE, n\u0026thinsp;=\u0026thinsp;3). Statistical analysis was conducted with Microsoft Excel 2016 and the SPSS 20.0 software package, using one-way analysis of variance to compare the differences among treatments, and Duncan\u0026rsquo;s multiple range test was at the 5% protection level. The neural network toolbox in MATLAB was used for the BPNN model, and the Levenberg-Marquardt training algorithm was used for calculation. The lethal concentration that reduced survival by 50% (LC\u003csub\u003e50\u003c/sub\u003e) and EC\u003csub\u003e50\u003c/sub\u003e were calculated by probit regression modeling at the 5% protection level. ArcMap 10.8 software was used to draw sensitive maps.\u003c/p\u003e \u003cp\u003eIn soil 23, the soil total Cd concentration was 0.76 mg kg\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e, soil pH was 5.05. \u003cem\u003eF. candida\u003c/em\u003e had the highest survival and reproduction numbers in this soil and, therefore, was used as the max reproduction number to calculate the reproduction inhibitory rate (Zhang et al., \u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e2015\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eReproduction inhibitory rate = \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\frac{\\text{max reproduction number - reproduction number}}{\\text{max reproduction number}}\\)\u003c/span\u003e\u003c/span\u003e \u0026times; 100% (2)\u003c/p\u003e \u003cp\u003eThe bioaccumulation factor (BAF) is commonly used to indicate the accumulation of metals and is calculated by:\u003c/p\u003e \u003cp\u003eBAF\u0026thinsp;=\u0026thinsp;\u003cem\u003eC\u003c/em\u003e\u003csub\u003ebody\u003c/sub\u003e/\u003cem\u003eC\u003c/em\u003e\u003csub\u003esoil\u003c/sub\u003e (3)\u003c/p\u003e \u003cp\u003ewhere \u003cem\u003eC\u003c/em\u003e\u003csub\u003ebody\u003c/sub\u003e is the concentration of Cd in \u003cem\u003eF. candida\u003c/em\u003e body tissue, and \u003cem\u003eC\u003c/em\u003e\u003csub\u003esoil\u003c/sub\u003e is the total Cd concentration in the soil.\u003c/p\u003e \u003cp\u003eLog-transformed soil parameters (except soil pH), body Cd concentrations in \u003cem\u003eF. candida\u003c/em\u003e, and BAF values were used to ensure the normality of data and homogeneity of the variance and used for linear regression and RDA. The relationships between survival and reproduction number, body Cd concentration in \u003cem\u003eF. candida\u003c/em\u003e, BAF, and soil properties were conducted using Pearson\u0026rsquo;s correlation analysis. The RDA analysis of internal Cd concentration in \u003cem\u003eF. candida\u003c/em\u003e (Log [Cd]) and BAF-Cd (Log [BAF]) were conducted using Canoco 5.0 software, with Log [Cd] and Log [BAF] as species variables and the soil properties as environmental variables. The reproduction inhibitory/survival rate, Log [Cd], and Log [BAF] were also examined using multiple linear regression analysis with different soil properties:\u003c/p\u003e \u003cp\u003eSurvival rate\u0026thinsp;=\u0026thinsp;a\u003csub\u003e1\u003c/sub\u003e log (\u003cem\u003esoil properties 1\u003c/em\u003e)\u0026thinsp;+\u0026thinsp;a\u003csub\u003e2\u003c/sub\u003e log (\u003cem\u003esoil properties 2\u003c/em\u003e) + \u0026hellip;\u0026hellip; a\u003csub\u003en\u003c/sub\u003e log (\u003cem\u003esoil properties n\u003c/em\u003e)\u0026thinsp;+\u0026thinsp;b (4)\u003c/p\u003e \u003cp\u003eReproduction inhibitory rate\u0026thinsp;=\u0026thinsp;a\u003csub\u003e1\u003c/sub\u003e log (\u003cem\u003esoil properties 1\u003c/em\u003e)\u0026thinsp;+\u0026thinsp;a\u003csub\u003e2\u003c/sub\u003e log (\u003cem\u003esoil properties 2\u003c/em\u003e) + \u0026hellip;\u0026hellip; a\u003csub\u003en\u003c/sub\u003e log (\u003cem\u003esoil properties n\u003c/em\u003e)\u0026thinsp;+\u0026thinsp;b (5)\u003c/p\u003e \u003cp\u003eLog [Cd]\u0026thinsp;=\u0026thinsp;a\u003csub\u003e1\u003c/sub\u003e log (\u003cem\u003esoil properties 1\u003c/em\u003e)\u0026thinsp;+\u0026thinsp;a\u003csub\u003e2\u003c/sub\u003e log (\u003cem\u003esoil properties 2\u003c/em\u003e) + \u0026hellip;\u0026hellip; a\u003csub\u003en\u003c/sub\u003e log (\u003cem\u003esoil properties n\u003c/em\u003e)\u0026thinsp;+\u0026thinsp;b (6)\u003c/p\u003e \u003cp\u003eLog [BAF-Cd]\u0026thinsp;=\u0026thinsp;a\u003csub\u003e1\u003c/sub\u003e log (\u003cem\u003esoil properties 1\u003c/em\u003e)\u0026thinsp;+\u0026thinsp;a\u003csub\u003e2\u003c/sub\u003e log (\u003cem\u003esoil properties 2\u003c/em\u003e) + \u0026hellip;\u0026hellip; a\u003csub\u003en\u003c/sub\u003e log (\u003cem\u003esoil properties n\u003c/em\u003e)\u0026thinsp;+\u0026thinsp;b (7)\u003c/p\u003e \u003cp\u003ewhere [Cd] was the Cd in \u003cem\u003eF. candida\u003c/em\u003e, a1, a2, \u0026hellip; and a\u003csub\u003en\u003c/sub\u003e were the regression coefficients, and b was the constant coefficient.\u003c/p\u003e \u003cp\u003eThe accuracy of the predictions was evaluated based on mean absolute error (MAE), mean relative error (MRE), the root-mean-square error (RMSE), and determination coefficient (R\u003csup\u003e2\u003c/sup\u003e). The higher R\u003csup\u003e2\u003c/sup\u003e value and the smaller MAE, MRE, and RMSE values, the higher accuracy of the model (Wang et al., \u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e2016\u003c/span\u003e; \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e2021\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e"},{"header":"3. Results and discussion","content":"\u003cdiv id=\"Sec9\" class=\"Section2\"\u003e \u003ch2\u003e3.1 Survival and reproduction of \u003cem\u003eF. candida\u003c/em\u003e\u003c/h2\u003e \u003cp\u003eThe soil samples used in this study were generally representative of the Cd concentrations found in most Chinese Cd-contaminated agricultural soils (Zhao et al., \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2015\u003c/span\u003e), and the Cd concentrations in all soils were higher than the current national soil Cd risk screening value (GB 15618\u0026thinsp;\u0026minus;\u0026thinsp;2018). An average of 76.1% of adults were alive in the 28 Cd-contaminated soils (Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e), indicating that the death rate of \u003cem\u003eF. candida\u003c/em\u003e was relatively low in the naturally Cd-contaminated soils. There was no significant relationship between the survival number and soil total Cd concentration or any other soil properties by Pearson\u0026rsquo;s analysis (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e), and the Cd LC\u003csub\u003e50\u003c/sub\u003e was not able to be calculated in these soils. The scatter plot of the survival rate in different Cd-contaminated soils is shown in Fig.\u0026nbsp;2. The survival rate appeared to increase with Cd concentrations in soils ˂ 3 mg Cd kg\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e concentration (r\u0026thinsp;=\u0026thinsp;0.528, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05) but showed a significant negative relationship with soil Cd concentrations in soils\u0026thinsp;\u0026gt;\u0026thinsp;3 mg kg\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e (r = \u0026ndash; 0.645, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.01). These trends were opposite to those of previous studies, showing a significant negative relationship between soil Cd concentrations and the survival rate, and soil pH, SOC, and soil CEC etc that were essential parameters for the survival rate in laboratory studies (Luo et al., \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2014\u003c/span\u003e; Sahraoui et al., \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). This may be because the previous conclusions were drawn mainly from ranging and finding tests, those tests set the Cd concentration, soil pH, SOC, and CEC, etc as a single variable in a specific soil (artificial soils, standard soil LUFA, or newly Cd spiked soils) (Lock and Janssen, \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2001a\u003c/span\u003e, \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003eb\u003c/span\u003e; Liu et al., \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). Therefore, it is easy to find the main factors and the most appropriate soil conditions. However, the naturally contaminated soils studied here were more complex, and the soil properties were closely associated with each other, and it was, therefore, difficult to obtain these relationships. Notably, the current study showed that survival number decreased with increasing soil pH (r\u0026thinsp;=\u0026thinsp;0.585, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05) in low Cd concentration soils (\u0026lt;\u0026thinsp;3 mg kg\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e) (Fig. \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e), suggesting that instead of Cd, pH was the major factor constraining survival in soils with low Cd concentrations, on the contrary, Cd was likely to have a promotive effect on survival rate in low Cd concentration soils. Studies indicate that collembolans are more appropriate living in acid and circumneutral soils, and few survive in soils with pH values\u0026thinsp;\u0026gt;\u0026thinsp;8.0, indicating that soil pH plays a major role in collembolan survival (Fountain and Hopkin, \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2005\u003c/span\u003e; Howcroft et al., \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2009\u003c/span\u003e). Therefore, the standard single-species test should not be generalized for a wide range of soil pH values. Instead, the standard single-species test animal should be specified for certain ranges of pH values (Fountain and Hopkin, \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2005\u003c/span\u003e; Howcroft et al., \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2009\u003c/span\u003e; Liu et al., \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). An alternative explanation is hormesis effects, demonstrating stimulatory or beneficial effects occurring at a lower dose, but inhibitory or toxic effects at a higher dose (Mark, \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2008\u003c/span\u003e; Gospodarek et al., \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). Hormesis should therefore be considered in the experimental design, selection of biological markers and endpoints, and risk model development et al. in assessing pollutants (Christiani and Zhou, \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e2016\u003c/span\u003e). Overall, the relationships between survival rate and soil properties are intricate and are challenging to predict using traditional statistical methods. The BPNN model was therefore used here to resolve the complex situation between survival rate and soil properties, in which soil total Cd concentration and soil pH were selected as input layers in the neural network mode.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eSurvival number and rate, juvenile number and reproduction inhibitory rate, Cd concentration in body tissue, and Cd bioaccumulation factor (BAF) of \u003cem\u003eF. candida\u003c/em\u003e in 28 naturally Cd-contaminated soils.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"7\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eSurvival number\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eSurvival rate\u003c/p\u003e \u003cp\u003e(%)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eJuvenile\u003c/p\u003e \u003cp\u003enumber\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eReproduction inhibitory rate\u003c/p\u003e \u003cp\u003e(%)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eCd concentration\u003c/p\u003e \u003cp\u003ein tissue\u003c/p\u003e \u003cp\u003e(mg kg\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003eBAF\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMin\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e50\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e61\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e2.44\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.40\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMax\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e9.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e93\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e352\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e82.7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e85.7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e36.0\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMedian\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e7.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e73.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e172\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e51.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e10.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e3.16\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMean\u0026thinsp;\u0026plusmn;\u0026thinsp;SE\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e7.6\u0026thinsp;\u0026plusmn;\u0026thinsp;0.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e76.1\u0026thinsp;\u0026plusmn;\u0026thinsp;2.10\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e181\u0026thinsp;\u0026plusmn;\u0026thinsp;16\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e48.5\u0026thinsp;\u0026plusmn;\u0026thinsp;4.4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e18.2\u0026thinsp;\u0026plusmn;\u0026thinsp;3.8\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e5.00\u0026thinsp;\u0026plusmn;\u0026thinsp;1.46\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCoefficient of variation (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e20.6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e20.6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e46.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e46.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e112\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e146\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eReproduction number showed significant negative relationships with soil pH and total Cd concentration based on Pearson\u0026rsquo;s correlation analysis (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). With the total soil Cd concentration, the obtained EC\u003csub\u003e50\u003c/sub\u003e value was 8.00 (5.92\u0026ndash;24.1), a lower value than that in previous studies using artificial soils (Lock and Jassen, 2001a). This also indicates that Cd is not the only factor limiting \u003cem\u003eF. candida\u003c/em\u003e reproduction in soils. The scatter plot of the reproduction number in different naturally Cd-contaminated soils is shown in Fig. S2. In the soils with lower Cd concentrations, the reproduction number was quite low, possibly due to inappropriate alkaline soil pH conditions for \u003cem\u003eF. candida\u003c/em\u003e reproduction (Fountain and Hopkin, \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2005\u003c/span\u003e). For example, in soil No. 2, the total Cd concentration was 0.62 mg kg\u003csup\u003e\u0026ndash;1\u003c/sup\u003e but the reproduction inhibition rate was 50%, resulting from the high pH value (8.47) (Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e). Zhang et al. (\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e2021\u003c/span\u003e) obtained similar results in which soil pH had more adverse effects on \u003cem\u003eF. candida\u003c/em\u003e reproduction than metals in naturally contaminated soils, especially in calcareous soils. In addition, only when the metal detoxification level exceeds its capacity, it may cause adverse effects on individual and population numbers (Lock and Janssen, \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2001b\u003c/span\u003e; Howcroft et al., \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2009\u003c/span\u003e). The Cd accumulated in soil animals may be formed as non-toxic substances by forming metal-proteins complexes. This metal-protein complex was observed by Morgan et al. (\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e2004\u003c/span\u003e) and Vijver et al. (\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e2006\u003c/span\u003e), who found that 70% of Cd in earthworms was sequestered by metallothionein (MT). Therefore, some Cd levels may be accumulated in the animal\u0026rsquo;s body with minimum toxic effects on the survival and reproduction of the testing organisms. Toxicity endpoint survival is complex according to soil characteristics and low sensitivity; therefore, further studies are required to better understand metal accumulation in soil animals.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec10\" class=\"Section2\"\u003e \u003ch2\u003e3.2 Cadmium accumulation in \u003cem\u003eF. candida\u003c/em\u003e\u003c/h2\u003e \u003cp\u003eThe internal Cd concentrations in \u003cem\u003eF. candida\u003c/em\u003e and the BAF values in the 28 naturally Cd-contaminated soils were 18.2\u0026thinsp;\u0026plusmn;\u0026thinsp;3.83 mg kg\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e (2.44\u0026ndash;85.7 mg kg\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e) and 5.00\u0026thinsp;\u0026plusmn;\u0026thinsp;1.46 (0.398\u0026ndash;36.0) respectively, showing significant differences among different soils (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001) (Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). The coefficient of variation was \u0026gt;\u0026thinsp;100, indicating a wide variation in soil properties and affecting Cd accumulation in \u003cem\u003eF. candida\u003c/em\u003e. The Cd BAF values in 92.9% of the soil samples were \u0026gt;\u0026thinsp;1.0, suggesting a high ecological risk of Cd in soils and potential food chain risks, which should be a concern in these soils.\u003c/p\u003e \u003cp\u003ePearson correlation analysis (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e) shows that soil pH was the primary factor that negatively influenced the accumulated Cd concentration in \u003cem\u003eF. candida\u003c/em\u003e and BAF-Cd values. Conversely, silt, total K, and total Cd concentrations in soils showed positive associations with the internal Cd concentration. Soil total K and available K concentrations also had positive relationships with the BAF. However, total Cd concentration in soils showed a negative association with the BAF. RDA analysis gave similar results (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e3\u003c/span\u003e). A Monte Carlo permutation test in the RDA analysis shows that Log [Cd] and Log [BAF] had significant correlations with the selected soil properties (Pseudo-F\u0026thinsp;=\u0026thinsp;10.5, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.005). The first and second RDA components (RDA1 and RDA2) explained 50.4 and 39.7%, respectively, of the total variation, suggesting that the soil attributes were significantly highly diverse. Soil total Cd concentration was the dominant factor significantly restraining Cd accumulation, contributing 43.6% of the variance. Soil pH, K, and SOC made lesser contributions of 22.0, 12.6, and 6.2% of the total variation, respectively.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eNutrient elements have been given less consideration in previous studies on metal accumulation (van Gestel, \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e2008\u003c/span\u003e; Liu et al., \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). However, here, Pearson\u0026rsquo;s correlation analysis and RDA analysis both confirm that K was essential for Cd accumulation by \u003cem\u003eF. candida\u003c/em\u003e. These results may be explained as follows. Firstly, K may exchange with Cd on soil particle surfaces, thereby increasing exchangeable/available Cd concentration in soils with further soil animal Cd assimilation (Wu et al., \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e2015\u003c/span\u003e; Wang et al., \u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). Wang et al. (\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e2019\u003c/span\u003e) found that K addition had a significantly positive relationship with available Cd content in soils, and K increased Cd accumulation in plant tissues. Secondly, K generally tends to be deficient in Chinese agricultural soils, with an area of deficiency of up to 60% of agricultural soils (Chen and Zheng, \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2004\u003c/span\u003e; R\u0026ouml;mheld and Kirkby, \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e2010\u003c/span\u003e). In this study, the average total K levels in the selected soils were 15.1 (3.16\u0026ndash;26.9 g kg\u003csup\u003e\u0026ndash;1\u003c/sup\u003e), slightly lower than the average K content in Chinese soils (Chen et al., \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). The deficiency of K may therefore be a constraining factor in the selected soils for collembolan. Thirdly, K is an essential nutrient for the growth and development of organisms and is involved in numerous processes, such as the regulation of enzyme activity, membrane potential, cellular homeostasis, and stable protein synthesis (Tester, \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2001\u003c/span\u003e; Luersen et al., \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e2016\u003c/span\u003e). Thus, K may increase the resistance of \u003cem\u003eF. candida\u003c/em\u003e under metal exposure, further increasing Cd accumulation in the body. Studies also confirm that K channels control neuromuscular, mating, and locomotion behaviors in \u003cem\u003eCaenorhabditis elegans\u003c/em\u003e (Abraham et al., \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2010\u003c/span\u003e; LeBoeuf and Garcia, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2012\u003c/span\u003e; Luersen et al., \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e2016\u003c/span\u003e). Potassium channels have secretory and excretory functions and thus may promote the adsorption and accumulation of Cd (Piermarini et al., \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e2013\u003c/span\u003e; Wu et al., \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e2015\u003c/span\u003e). More importantly, Cd\u003csup\u003e2+\u003c/sup\u003e binds to K channels, and this influences the transfer and cytotoxicity of Cd (Wang et al., \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e2017\u003c/span\u003e). Thus, K is essential for Cd accumulation in \u003cem\u003eF. candida\u003c/em\u003e. To provide direct evidence of the effects of soil K on Cd accumulation in \u003cem\u003eF. candida\u003c/em\u003e, a K-spiked single-species test was conducted based on the OECD guidelines with some modifications (OECD, \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2009\u003c/span\u003e, Supplementary materials). The results confirm that soil K showed a positive relationship with Cd accumulation in \u003cem\u003eF. candida\u003c/em\u003e (R\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;=\u0026thinsp;0.964, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001), with the body Cd concentration significantly increasing with increasing soil K concentration (Fig. S3, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001). The current study indicates the necessity of studying the effects of nutrient elements on metal accumulation in biota.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003e3.3 Comparison of Cd toxicity prediction performance between multiple linear regression and BPNN models\u003c/h2\u003e \u003cp\u003eMultiple linear regression models are commonly used to predict and estimate the dependent variable by the optimum combination of multiple independent variables (De\u0026rsquo;ath and Fabricius, 2000). Here, the survival rate of \u003cem\u003eF. candida\u003c/em\u003e in naturally Cd-contaminated soils cannot be predicted by multiple linear regression analysis, therefore, the BPNN model was used. The reproduction inhibitory rate was predicted and compared by the BPNN and multiple linear regression models. All soil properties were input as variables and the reproduction inhibitory rate was well interpreted by stepwise multiple linear regression equations in the tested soils, but only soil total Cd concentration and soil pH were valid parameters. The stepwise regression equation was developed as follows:\u003c/p\u003e \u003cp\u003eReproduction inhibitory rate\u0026thinsp;=\u0026thinsp;0.180 Log Cd\u003csub\u003eT\u003c/sub\u003e + 0.073 pH \u0026ndash; 0.097 (R\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;=\u0026thinsp;0.302, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026gt;\u0026thinsp;0.05) (8)\u003c/p\u003e \u003cp\u003eBased on Pearson\u0026rsquo;s correlation analysis, soil total Cd concentration and soil pH had significant relationships with survival and reproduction inhibitory rate, thus, soil total Cd concentration and soil pH were set as input variables in the BPNN model, and survival/reproduction inhibitory rate was the output target. The number of the input layer was 2, the output layer was 1, the hidden layer was 5 based on the empirical formula, and the number of network iterations was 5000. In the BPNN model, 20 sets of samples were used for training, 4 were used for validation, and 4 for testing. Specific training, validation, and testing results are shown in Table S2 and the predictions of all the samples were significant (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001). The prediction results were compared with the multiple linear regression model, as shown in Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e and Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e4\u003c/span\u003e. The survival and reproduction inhibitory rates were successfully predicted by the BPNN model with R\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;=\u0026thinsp;0.797 and 0.930, respectively. The prediction accuracy of the reproduction inhibitory rate by the BPNN model was much higher than by multiple linear regression, with a higher R\u003csup\u003e2\u003c/sup\u003e and lower MAE, MRE, and RMSE. This indicates that the BPNN model is superior to the stepwise regression model. In addition, the larger soil total Cd concentration level (0.54 \u0026minus;\u0026thinsp;25.7 mg kg \u003csup\u003e\u0026ndash;1\u003c/sup\u003e) indicated that the BPNN model was sensitive and suitable for wide soil metal concentrations. Therefore, the newly developed BPNN model can be useful for predicting Cd toxicity to \u003cem\u003eF. candida\u003c/em\u003e in naturally contaminated soils.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003ePrediction performance parameters of the BPNN and stepwise multiple linear regression models for the estimation of Cd concentration in \u003cem\u003eF. candida\u003c/em\u003e and Cd bioaccumulation factor (BAF).\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"6\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTarget\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eModel\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eR\u003csup\u003e2\u003c/sup\u003e\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eMAE\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eMRE (%)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eRMSE\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSurvival rate\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eBPNN\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.797\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e3.27\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.042\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e4.69\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eReproduction inhibitory rate\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eBPNN\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.827\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e8.24\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.194\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e10.3\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMultiple linear regression\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.302\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e16.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.318\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e20.7\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eLog [Cd]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eBPNN\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.961\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.003\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.014\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.077\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMultiple linear regression\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.638\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.002\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.012\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.201\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eLog [BAF]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eBPNN\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.964\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.002\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.014\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.075\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMultiple linear regression\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.591\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.002\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.011\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.194\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003e3.4 Comparison between multiple linear regression and BPNN models in predicting Cd accumulation in collembolan\u003c/h2\u003e \u003cp\u003eThe Cd accumulation in collembolan was interpreted well by stepwise multiple linear regression equations in the tested soils, and only soil total Cd and pH were valid parameters selected through stepwise multiple linear regression analysis, and the developed equations for Log [Cd] and Log [BAF] were as follows:\u003c/p\u003e \u003cp\u003eLog [Cd]\u0026thinsp;=\u0026thinsp;0.568 Log Cd\u003csub\u003eT\u003c/sub\u003e \u0026ndash; 0.154 pH\u0026thinsp;+\u0026thinsp;1.748 (R\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;=\u0026thinsp;0.638, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001) (9)\u003c/p\u003e \u003cp\u003eLog [BAF] = \u0026ndash; 0.432 Log Cd\u003csub\u003eT\u003c/sub\u003e \u0026ndash; 0.154 pH\u0026thinsp;+\u0026thinsp;1.748 (R\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;=\u0026thinsp;0.591, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001 (10)\u003c/p\u003e \u003cp\u003eBased on Pearson\u0026rsquo;s correlation analysis and RDA analysis, soil total Cd concentration, soil pH, SOC, and soil K play significant roles in Cd bioaccumulation, therefore, were selected as input variables in the BPNN model. Log [Cd] or Log [BAF] was the output target. The number of input layer was 4, the number of output layer was 1, the number of hidden layers was 5 based on the empirical formula, and the number of network iterations was 5000. In the BPNN model, 20 sets of samples were used for training, 4 sets for validation, and 4 sets for testing. Specific results on training, validation, and testing results are also shown in Table S2 and the predictions of all samples were significant (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001).\u003c/p\u003e \u003cp\u003eTo retain the consistency of selected soil properties in all the models, the multiple linear regression equations entered were as follows:\u003c/p\u003e \u003cp\u003eLog [Cd]\u0026thinsp;=\u0026thinsp;0.623 Log Cd\u003csub\u003eT\u003c/sub\u003e\u0026ndash; 0.144 pH\u0026thinsp;+\u0026thinsp;0.228 Log K\u003csub\u003eT\u003c/sub\u003e \u0026ndash; 0.319 Log SOC\u0026thinsp;+\u0026thinsp;1.815\u003c/p\u003e \u003cp\u003e(R\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;=\u0026thinsp;0.687, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001) (11)\u003c/p\u003e \u003cp\u003eLog [BAF] = \u0026ndash; 0.377 Log Cd\u003csub\u003eT\u003c/sub\u003e \u0026ndash; 0.144 pH\u0026thinsp;+\u0026thinsp;0.228 Log K\u003csub\u003eT\u003c/sub\u003e\u0026ndash; 0.319 Log SOC\u0026thinsp;+\u0026thinsp;1.815\u003c/p\u003e \u003cp\u003e(R\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;=\u0026thinsp;0.646, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001) (12)\u003c/p\u003e \u003cp\u003eThe prediction performance of entered multiple linear regression models slightly improved to the stepwise multiple linear regression models, therefore, entered multiple linear regression equations were not compared with the BPNN model. Although the MAE and MRE of the BPNN and multiple linear regression models were similar (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e), the BPNN model had higher R\u003csup\u003e2\u003c/sup\u003e and lower RMSE than the multiple linear regression model. The prediction accuracy of the BPNN model of Log [Cd] and Log [BAF] was superior to the multiple linear regression model.\u003c/p\u003e \u003cp\u003eNotably, the correlation between the Log [Cd] predicted by the BPNN model and the measured Log [Cd] was stronger than that by the multiple linear regression model, with the predicted and measured values very close to the 1:1 line among a wide range of soil Cd concentrations (R\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;=\u0026thinsp;0.910), but the correlation by multiple linear regression model was lower (R\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;=\u0026thinsp;0.638, Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e4\u003c/span\u003eC). A similar trend was found in Log [BAF] (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e4\u003c/span\u003eD). The correlation between the Log [BAF] predicted by the BPNN model and the measured Log [BAF] was higher than that by the multiple linear regression model, which the correlation between the predicted and measured values was very close to the 1:1 line (R\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;=\u0026thinsp;0.979). In addition, with increasing Log [BAF], the predicted values by the multiple linear regression model gradually deviated from the 1:1 line, especially when the measured Log [BAF] exceeded 1.00. Recently, Wang et al. (\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e2021\u003c/span\u003e) obtained similar results showing that the multiple linear regression model cannot predict well the Zn BAF in rice grain when the BAF exceeds 0.40. Thus, the multiple linear regression model is not suitable for the prediction of high Cd BAF and is limited to a wide range of soil properties. The BPNN model accounted for the complex nonlinear relationships between Cd accumulation in \u003cem\u003eF. candida\u003c/em\u003e and soil properties and showed superiority in predicting internal Cd concentrations and Cd BAF values in \u003cem\u003eF. candida\u003c/em\u003e over a wide range of Cd concentrations.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003e3.5 Application of the BPNN models for predicting Cd ecotoxicity in Southern Chinese soils\u003c/h2\u003e \u003cp\u003eWith the developed BPNN models, the predicted Cd toxicity to and accumulation in soil collembolans can be assessed without conducting toxicity tests. Here, 57 samples were also collected from southern China with a wide range of soil properties (soil pH 4.14\u0026ndash;8.36, soil total Cd 0.15\u0026ndash;25.5 mg kg\u003csup\u003e\u0026ndash;1\u003c/sup\u003e, SOC 5.94\u0026ndash;47.8 g kg\u003csup\u003e\u0026ndash;1\u003c/sup\u003e, Table S3), and the accumulated Cd concentrations in \u003cem\u003eF. candida\u003c/em\u003e and BAF values were predicted based on the BPNN models with the input layers soil total Cd, soil pH, SOC, and K, the number of input layer was 4, the number of output layer was 1, the number of hidden layer was 5, and the number of network iterations was 5000. Multiple linear regression models were also used for comparison with the valid parameters of the total Cd and pH in soils (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e5\u003c/span\u003e). The predicted values were visualized on sensitive maps, and different colors were assigned to indicate the level of risk. The predicted values of the internal Cd concentration in \u003cem\u003eF. candida\u003c/em\u003e and BAF-Cd values, as observed by the BPNN models, were generally higher than those by the multiple linear regression models. Notably, the BPNN models better identified the regions of higher risk. This was consistent with our previous results in which multiple linear regression underestimated Cd accumulation in \u003cem\u003eF. candida\u003c/em\u003e (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e4\u003c/span\u003e).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThe mathematic models are also beneficial in the quantification of influencing factors and the development of further strategies to control metal risks in soils. There is a tendency for soil pH to decrease and SOC to increase in agricultural soils (Guo et al., \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2010\u003c/span\u003e; Sun et al., \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e2012\u003c/span\u003e). Therefore, the effects of changes in soil properties on the evaluation of Cd ecological risks in soils need to be considered. Three soils (Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e, Soil No. 15, 25, and 6) with different pH values (4.73, 5.85, and 7.36) were used to demonstrate the effects of changes in pH (\u0026plusmn;\u0026thinsp;0.5 and 1 unit) and SOC (+\u0026thinsp;1% and 3%) on the BAF-Cd values. The results showed that soil pH was more important than SOC in Cd accumulation in \u003cem\u003eF. candida\u003c/em\u003e (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e6\u003c/span\u003e). Increases in SOC of 1% and 3% had no significant effect on the BAF-Cd values, but pH was significantly negatively associated with the BAF-Cd values. This is probably because decreasing soil pH increased Cd bioavailability, further increasing the accumulated Cd in the collembolan bodies. Thus, remediation methods should be taken to increase soil pH to lower the ecological risks of Cd in soils. The input data from the BPNN model is easily derived from soil chemical analyses or obtained from databases of national soil surveys (Zhao et al., \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2015\u003c/span\u003e), and thereby the Cd toxicity to and accumulation in \u003cem\u003eF. candida\u003c/em\u003e in soils can be predicted by the BPNN model without conducting toxicity tests.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e"},{"header":"4. Conclusions","content":"\u003cp\u003eCadmium toxicity to soil animals and accumulation in their bodies are intricate and were substantially affected by soil properties in naturally contaminated soils, and soil pH and soil Cd concentration were the main factors driving Cd toxicity to and accumulation in soil collembolan. When the total Cd concentration was \u0026lt;\u0026thinsp;3 mg kg\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e, the high soil pH was the main factor restricting collembolan survival rather than soil total Cd concentration. Pearson\u0026rsquo;s correlation, RDA analysis, and the K-spiked test all confirm that K was essential for Cd accumulation in \u003cem\u003eF. candida\u003c/em\u003e. This supports the necessity of taking the effects of nutrient elements (K) on metal (Cd) accumulation in soil biota into account. Compared with the multiple linear regression model, the BPNN model developed showed higher performance in estimating the toxicity to and accumulation in \u003cem\u003eF. candida\u003c/em\u003e under a wide range of soil properties and better identified high-risk regions of Cd ecotoxicology in southern China. The developed BPNN model successfully recognized and resolved complex nonlinear relationships between Cd ecotoxicology in \u003cem\u003eF. candida\u003c/em\u003e and soil properties in naturally contaminated soils and can be used to predict Cd ecotoxicology generally in untested soils. It can therefore be used as an alternative tool to evaluate and monitor Cd ecotoxicity in practice.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study was funded by the National Natural Science Foundation of China (41977136).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor contributions:\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eSimin Li: Writing original draft, Investigation, Methodology, Data curation, Visualization, Formal analysis; Zhu Li: Review \u0026amp; editing, Funding acquisition; Xin Ke: Review \u0026amp; Editing; Worachart Wisawapipat: Review \u0026amp; Editing; Peter Christie: Review \u0026amp; Editing; Longhua Wu: Review \u0026amp; Editing.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData Availability:\u003c/strong\u003e Not applicable\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthical approval:\u0026nbsp;\u003c/strong\u003eThe authors hereby approve that principles of ethical and professional conduct have been followed in the work.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent to participate:\u0026nbsp;\u003c/strong\u003eThe present research does not involve any human or animal participation.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication:\u0026nbsp;\u003c/strong\u003eThe authors and the responsible authorities at the institute/organization where this work has been carried out give their explicit consent to submit and publish the work in ESPR if found suitable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests:\u0026nbsp;\u003c/strong\u003eThe authors declare no competing interests.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eAbraham, L. S., Oh, H. J., Sancar, F., Richmond, J. E., Kim, H., 2010. An alpha-catulin homologue controls neuromuscular function through localization of the dystrophin complex and BK channels in \u003cem\u003eCaenorhabditis elegans\u003c/em\u003e. PLoS Genetics 6, 1001077. \u003c/li\u003e\n\u003cli\u003eAli, H., Khan, E., Ilahi, I., 2019. Environmental chemistry and ecotoxicology of hazardous heavy metals: Environmental persistence, toxicity, and bioaccumulation. Journal of Chemistry 6730305.\u003c/li\u003e\n\u003cli\u003eChen, F., Zheng, S. X., 2004. Research progress of efficient potassium application technology for crops in South China. Soils and Fertilizers 6, 29\u0026ndash;32.\u003c/li\u003e\n\u003cli\u003eChen, X. Q., Li, T., Lu, D. J., Cheng, L., Zhou, J. M., Wang, H. Y., 2020. Estimation of soil available potassium in Chinese agricultural fields using a modified sodium tetraphenyl boron method. Land Degradation \u0026amp; Development 31, 1737-1748.\u003c/li\u003e\n\u003cli\u003eChristiani, D. C., Zhou, W., 2016. Hormesis: The new approach in risk assessment? Human \u0026amp; Experimental Toxicology 21, 399\u0026ndash;400.\u003c/li\u003e\n\u003cli\u003eCoyle, D. R., Nagendra, U. J., Taylor, M. K., Campbell, J. H., Cunard, C. E., Joslin, A. H., Jr, M. A. C., 2017. Soil fauna responses to natural disturbances, invasive species, and global climate change: Current state of the science and a call to action. Soil Biology \u0026amp; Biochemistry 110, 116‒133. \u003c/li\u003e\n\u003cli\u003eDe\u0026apos;ath, G., Fabricius, K. E., 2000. Classification and regression trees: A powerful yet simple technique for ecological data analysis. Ecology 81, 3178‒3192. \u003c/li\u003e\n\u003cli\u003eFountain, M. T., Hopkin, S. P., 2005. \u003cem\u003eFolsomia candida\u003c/em\u003e (collembola): A standard soil arthropod. Annual Review of Entomology 50, 201‒222. \u003c/li\u003e\n\u003cli\u003eGospodarek, J., Boliglowa, E., Glen-Karolczyk, K., 2020. Impact of Nonchemical Protection of Broad Bean on Epigeic and Soil Arthropodofauna-Analysis in Field-Realistic Conditions. Agronomy-Basel 10, 211.\u003c/li\u003e\n\u003cli\u003eGuo, J. H., Liu, X., Zhang, Y., Shen, J. L., Han, W. X., Zhang, W. F., Christie, P., Goulding, K. W., Vitousek, P. M., Zhang, F. S., 2010. Significant acidification in major Chinese croplands. Science 327, 1008\u0026ndash;1010.\u003c/li\u003e\n\u003cli\u003eHowcroft, C. F., Amorim, M. J. B., Gravato, C., Guilhermino, L., Soares, A. M. V. M., 2009. Effects of natural and chemical stressors on \u003cem\u003eEnchytraeus albidus\u003c/em\u003e: Can oxidative stress parameters be used as fast screening tools for the assessment of different stress impacts in soils ?. Environment International 35, 318‒324.\u003c/li\u003e\n\u003cli\u003eKim, S. W., An, Y. J., 2014. Jumping behavior of the springtail \u003cem\u003eFolsomia candida\u003c/em\u003e as a novel soil quality indicator in metal-contaminated soils. Ecological Indicators 38, 67\u0026ndash;71.\u003c/li\u003e\n\u003cli\u003eLeBoeuf, B., Garcia, L. R., 2012. Cell excitability necessary for male mating behavior in Caenorhabditis elegans is coordinated by interactions between big current and ether-A-Go-Go Family K\u003csup\u003e+\u003c/sup\u003e Channels. Genetics 3, 1025\u0026ndash;1041. \u003c/li\u003e\n\u003cli\u003eLi, X. N., Cheng, X., Wu, W. J., Wang, Q. H., Tong, Z. Y., Zhang, X. Q., Deng, D. H., Li, Y. H., 2019. Forecasting of bioaerosol concentration by a Back Propagation neural network model. Science of the Total Environment 698, 134315. \u003c/li\u003e\n\u003cli\u003eLiu, H., Xuan, L., Zhou, J., Zhou, D. M., Wang, Y. J., 2019. Effects of soil properties on cadmium toxicity to \u003cem\u003eFolsomia candida \u003c/em\u003e(Collembola). Bulletin of Environmental Contamination and Toxicology 103, 90\u0026ndash;97. \u003c/li\u003e\n\u003cli\u003eLiu, M. L., Liu, X. N., Wu, M. X., Li, L. F., Xiu, L. N., 2011. Integrating spectral indices with environmental parameters for estimating heavy metal concentrations in rice using a dynamic fuzzy neural-network model. Computers \u0026amp; Geosciences 37, 1642\u0026ndash;1652. \u003c/li\u003e\n\u003cli\u003eLock, K., Janssen, C. R., 2001a. Cadmium toxicity for terrestrial invertebrates: taking soil parameters affecting bioavailability into account. Ecotoxicology 10, 315\u0026ndash;322.\u003c/li\u003e\n\u003cli\u003eLock, K., Janssen, C. R., 2001b. Zinc and cadmium body burdens in terrestrial oligochaetes: use and significance in environmental risk assessment. Environmental Toxicology and Chemistry 20, 2067\u0026ndash;2072.\u003c/li\u003e\n\u003cli\u003eLu, R. K., 2000. Analytical Methods for Soil and Agrochemistry. Agricultural Science and Technology Press, Beijing.\u003c/li\u003e\n\u003cli\u003eLuersen, K., Gottschling, D. C., Doring, F., 2016. Complex locomotion behavior changes are induced in Caenorhabditis elegans by the lack of the regulatory leak K+ Channel TWK-7. Genetics 2, 683\u0026ndash;701.\u003c/li\u003e\n\u003cli\u003eLuo, W., Verweij, R. A., van Gestel, C. A. M., 2014. Assessment of the bioavailability and toxicity of lead polluted soils using a combination of chemical approaches and bioassays with the collembolan \u003cem\u003eFolsomia candida\u003c/em\u003e. Journal of Hazardous Materials 280, 524\u0026ndash;530.\u003c/li\u003e\n\u003cli\u003eMark, P. M., 2008. Hormesis defined. Ageing Research Reviews 7, 1\u0026ndash;7.\u003c/li\u003e\n\u003cli\u003eMorgan, A. J., Stu\u0026uml;rzenbaum, S. R., Winters, C., Grime, G. W., Aziz, N. A., Kille, P., 2004. Differential metallothionein expression in earthworm (\u003cem\u003eLumbricus rubellus\u003c/em\u003e) tissues. Ecotoxicology and Environmental Safety 57, 11\u0026ndash;19.\u003c/li\u003e\n\u003cli\u003eNoh, H., Zhang, Q., Shin, B., Han, S., Feng L., 2006. A neural network model of maize crop nitrogen stress assessment for a multi-spectral imaging sensor. Biosystems Engineering, 94, 477\u0026ndash;485.\u003c/li\u003e\n\u003cli\u003eOECD, 2009. Guidelines for the testing of chemicals no. 232, collembolan reproduction test in soil. Organization for Economic Cooperation and Development.\u003c/li\u003e\n\u003cli\u003ePiermarini, P. M., Rouhier, M. F., Schepel, M., Kosse, C., Beyenbach, K. W., 2013. Cloning and functional characterization of inward-rectifying potassium (Kir) channels from Malpighian tubules of the mosquito \u003cem\u003eAedes aegypti\u003c/em\u003e. Insect Biochemistry and Molecular Biology, 43, 75\u0026ndash;90.\u003c/li\u003e\n\u003cli\u003eR\u0026ouml;mheld, V., Kirkby, E. A., 2010. Research on potassium in agriculture: needs and prospects. Plant and Soil 335, 155\u0026ndash;180.\u003c/li\u003e\n\u003cli\u003eSahraoui, A. S., Verweij, R. A., Belhiouani, H., Cheriti, O., van Gestel, C. A. M., Sahli, L., 2021. Dose-dependent effects of lead and cadmium and the influence of soil properties on their uptake by \u003cem\u003eHelix aspersa\u003c/em\u003e: an ecotoxicity test approach. Ecotoxicology 30, 331\u0026ndash;342.\u003c/li\u003e\n\u003cli\u003eSun, X. L., Zhao, Y. G., Wu, Y. J., Zhao, M. S., Wang, H. L., Zhang, G. L., 2012. Spatio-temporal change of soil organic matter content of Jiangsu Province, China, based on digital soil maps. Soil Use and Management 28, 318\u0026ndash;328.\u003c/li\u003e\n\u003cli\u003eTester, M., 2001. Partitioning of nutrient transport processes in roots. Journal of Experimental Botany 52, 653\u0026ndash;653.\u003c/li\u003e\n\u003cli\u003eTumbo, S. D., Wagner, D. G., Heinemann, P. H., 2002. Hyperspectral\u0026ndash;based neural network for predicting chlorophyll status in corn. Transactions of the ASAE, 45, 825\u0026ndash;832.\u003c/li\u003e\n\u003cli\u003evan Gestel, C. A. M., 2008. Physico-chemical and biological parameters determine metal bioavailability in soils. Science of the Total Environment 406, 385\u0026ndash;395.\u003c/li\u003e\n\u003cli\u003eVijver, M. G., van Gestel, C. A. M., van Straalen, N. M., Lanno, R. P., Peijnenburg, W. J. G. M., 2006. Biological significance of metals partitioned to subcellular fractions within earthworms (\u003cem\u003eAporrectodea caliginosa\u003c/em\u003e). Environmental Toxicology and Chemistry 25, 807\u0026ndash;814.\u003c/li\u003e\n\u003cli\u003eWang, L. H., He, J. F., Xia, A., Cheng, M. Z., Yang, Q., Du, C. L., Wei, H. Y., Huang, X. H., Zhou, Q., 2017. Toxic effects of environmental rare earth elements on delayed outward potassium channels and their mechanisms from a microscopic perspective. Chemosphere 181, 690\u0026ndash;698. \u003c/li\u003e\n\u003cli\u003eWang, K., Fu, G. P., Yu, Y., Wan, Y. A., Liu, Z., Wang, Q., Zhang, J. S., Li, H. F., 2019. Effects of different potassium fertilizers on cadmium uptake by three crops. Environmental Science and Pollution Research 26, 27014\u0026ndash;27022. \u003c/li\u003e\n\u003cli\u003eWang, Q. Y., Sun, J. Y., Hu, N. W., Wang, T. Y., Yue, J., Hu, B., Yu, H. W., 2022. Effects of soil aging conditions on distributions of cadmium distribution and phosphatase activity in different soil aggregates. Science of the Total Environment 834, 155440. \u003c/li\u003e\n\u003cli\u003eWang, S. X., Zhang, N., Wu, L., Wang, Y. M., 2016. Wind speed forecasting based on the hybrid ensemble empirical mode decomposition and GA-BP neural network method. Renewable Energy 94, 629\u0026ndash;636. \u003c/li\u003e\n\u003cli\u003eWang, Y. Z, Yu, T., Yang, Z. F., Bo, H. Z., Lin, Y., Yang, Q., Liu, X., Zhang, Q. Z., Zhuo, X. X., Wu, T. S., 2021. Zinc concentration prediction in rice grain using back-propagation neural network based on soil properties and safe utilization of paddy soil: A large-scale field study in Guangxi, China. Science of the Total Environment 798, 149270. \u003c/li\u003e\n\u003cli\u003eWu, Y., Baum, M., Huang C. L., Rodan, A. R., 2015. Two inwardly rectifying potassium channels, Irk1 and Irk2, play redundant roles in \u003cem\u003eDrosophila\u003c/em\u003e renal tubule function. American Journal of Physiology-Regulatory Integrative and Comparative Physiology 309, R747\u0026ndash;R756.\u003c/li\u003e\n\u003cli\u003eZhang, X. Y., Zhang, Y. W., Zhang, H. Y., Yang, Q. Y., Wang, H. Y., Zhang, G. C., 2015. Preparation, characterization, and antibacterial activity of octenyl succinic anhydride modified inulin. International Journal of Biological Macromolecules 78, 79\u0026ndash;86.\u003c/li\u003e\n\u003cli\u003eZhang, Y. B., Li, Z., Ke, X., Wu, L. H., Christie, P., 2021. Multigenerational exposure of the collembolan\u003cem\u003e Folsomia candida \u003c/em\u003eto soil metals: Adaption to metal stress in soils polluted over the long term. Environmental Pollution 292, 118242. \u003c/li\u003e\n\u003cli\u003eZhao, F. J., Ma, Y., Zhu, Y. G., Tang, Z., McGrath, S. P., 2015. Soil contamination in China: Current status and mitigation strategies. Environmental science \u0026amp; technology 49, 750\u0026ndash;759.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"environmental-science-and-pollution-research","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"espr","sideBox":"Learn more about [Environmental Science and Pollution Research](https://www.springer.com/journal/11356)","snPcode":"11356","submissionUrl":"https://submission.nature.com/new-submission/11356/3","title":"Environmental Science and Pollution Research","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"back-propagation neural network model, Cd, soil collembolan, bioaccumulation, metal toxicity","lastPublishedDoi":"10.21203/rs.3.rs-3740915/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-3740915/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eAccurate prediction of cadmium (Cd) ecotoxicity to and accumulation in soil biota is important in soil health. However, very limited information on Cd ecotoxicity on naturally contaminated soils. Herein, we investigated soil Cd ecotoxicity using \u003cem\u003eFolsomia candida\u003c/em\u003e, a standard single-species test animal, in 28 naturally Cd-contaminated soils, and the back-propagation neural network (BPNN) model was used to predict Cd ecotoxicity to and accumulation in \u003cem\u003eF. candida\u003c/em\u003e. Soil total Cd and pH were the primary soil properties affecting Cd toxicity. However, soil pH was the main factor when the total Cd concentration was ˂ 3 mg kg\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e. Interestingly, correlation analysis and the K-spiked test confirmed nutrient potassium (K) was essential for Cd accumulation, highlighting the significance of studying K in Cd accumulation. The BPNN model showed greater prediction accuracy of collembolan survival rate (R\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;=\u0026thinsp;0.797), reproduction inhibitory rate (R\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;=\u0026thinsp;0.827), body Cd concentration (R\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;=\u0026thinsp;0.961), and Cd bioaccumulation factor (R\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;=\u0026thinsp;0.964) than multiple linear regression models. Then the developed BPNN model was used to predict Cd ecological risks in 57 soils in southern China. Compared to multiple linear regression models, the BPNN models can better identify high-risk regions. This study highlights the potential of BPNN as a novel and rapid tool for the evaluation and monitoring of Cd ecotoxicity in naturally contaminated soils.\u003c/p\u003e","manuscriptTitle":"Cadmium toxicity to and accumulation in a soil collembolan (Folsomia candida): major factors and prediction using a back-propagation neural network mode","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-01-17 09:17:25","doi":"10.21203/rs.3.rs-3740915/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"reviewerAgreed","content":"","date":"2024-01-16T00:52:01+00:00","index":0,"fulltext":""},{"type":"reviewersInvited","content":"","date":"2024-01-15T18:57:46+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"Environmental Science and Pollution Research","date":"2024-01-15T15:10:55+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2024-01-05T04:58:11+00:00","index":"","fulltext":""},{"type":"submitted","content":"Environmental Science and Pollution Research","date":"2023-12-29T02:31:53+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"environmental-science-and-pollution-research","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"espr","sideBox":"Learn more about [Environmental Science and Pollution Research](https://www.springer.com/journal/11356)","snPcode":"11356","submissionUrl":"https://submission.nature.com/new-submission/11356/3","title":"Environmental Science and Pollution Research","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"519bf4ab-d24e-4646-af40-5400a404da62","owner":[],"postedDate":"January 17th, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2024-03-04T15:13:22+00:00","versionOfRecord":{"articleIdentity":"rs-3740915","link":"https://doi.org/10.1007/s11356-024-32638-x","journal":{"identity":"environmental-science-and-pollution-research","isVorOnly":false,"title":"Environmental Science and Pollution Research"},"publishedOn":"2024-03-02 15:02:14","publishedOnDateReadable":"March 2nd, 2024"},"versionCreatedAt":"2024-01-17 09:17:25","video":"","vorDoi":"10.1007/s11356-024-32638-x","vorDoiUrl":"https://doi.org/10.1007/s11356-024-32638-x","workflowStages":[]},"version":"v1","identity":"rs-3740915","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-3740915","identity":"rs-3740915","version":["v1"]},"buildId":"qtupq5eGEP_6zYnWcrvyt","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}