Genotypic differences of Cd concentration and nutrition traits of 23 main celery cultivars of China in a Cd contaminated greenhouse

Due to the increasing concerns of heavy metal contamination in greenhouse eld, safety production of vegetables especially leafy vegetables was largely limited. In this study, the cadmium (Cd) concentration and major nutrition qualities of 23 main celery cultivars of China were compared in a greenhouse experiment. Great genotypic differences in biomass, cadmium accumulation and nutrition traits were observed. The biomass of cultivars HQ, JZ, JH and SQ were signicantly higher than those of the others. And Cd concentration in edible part ranged from 0.53 to 2.56 mg·kg − 1 DW, of which SQ exhibited the lowest Cd concentration. In addition, SQ has the lowest Cd transport factor (TF) and bio-concentration factor (BCF), followed by another genotype LF. Simultaneously, they have relatively higher chlorophyll content and vitamin C concentration, lower cellulose content. Therefore, the two genotypes SQ and LF were selected as promising candidates to grow on a mid Cd-contaminated greenhouse eld to achieve safety production. Further correlation analysis showed that Cd concentration in edible part is positively correlated with cellulose content, but negatively correlated with vitamin C concentration. The results of celery variety screening provide a safe production strategy of moderately polluted greenhouse vegetable elds. Data presented as means ± standard error (SE) with three replicates. All data were means of three replications. Different letters indicate statistically signicant differences at P < 0.05 as detected by Duncan’s Multiple Range Test (DMRT).


Introduction
With the rapid development of cities, atmospheric deposition, mining and smelting activities, wastewater irrigation, sewage irrigation, fertilizer and pesticide application have led to heavy metal pollution in the soil environment and agriculture (Elgallal et al., 2016;Rai et al., 2019;Ma et al., 2020). Cadmium (Cd) is classi ed as a Group 1 human carcinogen (Cancer IAFO, 2018) and is considered an important pollutant with great concern in agricultural ecosystems due to its high toxicity and bioavailability Ma et al., 2021). Plants grown in Cd-contaminated soil readily accumulated Cd in their edible parts, and continuous dietary intake of Cd could result into serious health problems such as prostate, lung cancers and bone disorders (Zhang et al., 2014;Tang et al., 2019a;Yang et al., 2020).
In China, the rapid industrialization and ignorance of environmental protection during the past 30 years led to a serious heavy metal contamination of soil. It is reported that there is about 13000 hectares of Cd polluted land, and the Cd content in the polluted soil is 2.5-23 mg·kg − 1 (Liu et al., 2009). According to the national soil pollution survey, more than one-third of the self-use land and the selected sites around the heavily polluted enterprises exceeded the standard (MEE and MNR, 2014). Therefore, the food safety and related health risks were raising more and more attention. And Cd accumulated most in vegetables and rice elds . In 2019, China's vegetable planting area reached 20.87 million ha, the output exceeded 700 million tons, the per capita share reached more than 500 kg, and the output value exceeded 2 trillion RMB . Since vegetable crops play an important role in the dietary structure of Chinese residents, the safe use of heavy metal pollution vegetable land is of great signi cance to ensure the safety of production and human health (Feng et al., 2018).
The absorption and accumulation of Cd in soil by different kinds of vegetables are obviously different, which are generally manifested as leaf vegetables > root vegetables > fruit vegetables (Zhong et al., 2017). According to the degree and difference of Cd absorption and enrichment of different vegetable crops, the production can be reasonably arranged. For example, in the heavy metal polluted areas, it is necessary to plant less leafy vegetables which are easy to accumulate heavy metals, and plant weakly enriched fruits and root vegetables. And Cd accumulation mechanism of different leafy vegetables was different, which led to signi cant differences in cadmium uptake and accumulation of different species (Ding et al, 2013b;Nabulo et al, 2011). Different genotypes of the same leafy vegetables also had different responses to cadmium uptake and accumulation in soil (Xiao et al., 2015;Zhou et al., 2012). Therefore, low accumulation genotype can be screened and used as a major safety production strategy of polluted land (Tang et al, 2016;Tang et al, 2019a;Tang et al, 2019b).
Celery (Apium graveolens L.) is a kind of leafy vegetable which is conventionally planted in China with edible stems and leaves (Yan et al., 2020). Celery is rich in vitamins, protein, cellulose and other nutrients (Li et al., 2017). Based on the results of meta-analysis, it was found that celery is not sensitive to cadmium pollution in soil, and can be used as a candidate leaf vegetable for safe production in moderately and slightly polluted vegetable elds . However, a few reports showed that celery was vulnerable to soil Cd contamination (Ni et al., 2002;Wang et al., 2013;Fang et al., 2019), and long-term ingestion of celery grown in Cd-contaminated eld would pose an adverse effect on human health . As the previous studies of celery were mostly conducted in the form of pot experiments with arti cial cadmium addition and few researches were conducted in the eld condition, genotypic difference of celery on heavy metal accumulation in mild to mid contaminated eld is urgently needed to be studied.
Here, 23 main cultivars of China were grown in a natural Cd contaminated greenhouse. The objectives were to compare their (i) growth; (ii) Cd uptake, enrichment, and translocation; and (iii) nutrition qualities, to screen out the Cd low-accumulated celery cultivars for directly large-scale application in local greenhouse.

Experimental site
A greenhouse experiment was conducted in Guoxiang farm (30°3′16 N and 119° 54′ 34 E) in Hangzhou, Zhejiang Province, China. The experimental site is a vegetable greenhouse which has been planting vegetables for decades. The basic soil properties were determined as follows: pH = 5.77 (soil: water = 1: 2.5), total N content 3.24 g·kg − 1 , total P content 2.78 g·kg − 1 , total K content 0.37 g·kg − 1 , available P content 67.35 mg·kg − 1 , available K content 94.62 mg·kg − 1 and total Cd content 0.87 mg·kg − 1 .

Plant Materials And Experimental Design
A total of 23 cultivated genotypes of celery were purchased from local seed company. The detailed information of the seeds was listed in Table S1. The experiment was carried out with random plot layout, and each genotype was repeated three times. Therefore, there were totally 69 experiment plots of about 6 m 2 (2 m width × 3 m length) each. Seeds were sowed directly in October 2018, followed by the thorough irrigation of the eld to facilitate seed germination. After emergence of celery seedlings, about 250 celery plants were left in each plot, and the interval of each celery was about 15 cm. During the whole growing season, regular eld management was conducted by the farmers. Plants were sampled at the harvested stage after ten weeks.

Determination Of Plant Biomass And Chlorophyll Content
For each genotype, plant samples were divided into roots and shoots, then washed with a 20 mM Na 2 -EDTA solution for 20 min followed with deionized water for 3 times. After mixing the cleaned plant samples according to the root, stem and leaf, take out half of the samples and dry them in the oven at 105 ℃ for 30 minutes, then dry them at 65 ℃ until a constant weight is obtained.
After harvesting, leaves of fresh samples were collected for chlorophyll content determination. Chlorophyll content was measured according to Wang et al. (2019) using acetone-ethyl alcohol colorimetry. Brie y, 0.50 g fresh leaf sample of different genotypes was collected and put into a mixture of acetone and ethanol (2:1) for dark treatment. After 24 h, the absorbance at 663 nm and 645 nm wavelength by an ultraviolet spectrophotometer (Lambda350V-vis, PerkinElmer, Singapore). Then the chlorophyll content was determined with the equation below: Total chlorophyll content (mg·g − 1 ) = 20.2 × A 645nm + 8.02 × A 663nm Determination of soluble protein, soluble sugar, vitamin C and cellulose Soluble protein concentration in plants were determined using Coomassie brilliant blue colorimetry method (Li, 2000). 1.00 g fresh celery sample, add 2 mL distilled water to grind into homogenate, use deionized water to transfer all to 15 mL centrifuge tube, and let it stand at room temperature for 30 minutes. The suspension was then centrifuged (3500 rpm) for 15 minutes and the supernatant was transferred to a 10 mL volumetric ask. The sample solution (0.1 mL) and Coomassie brilliant blue G-250 protein reagent (5 mL) were shaken and mixed in a test tube with a stopper. After 2 minutes of reaction, the absorbance value was determined at 595 nm wavelength by ultraviolet spectrophotometer (lambda 350V vis, Perkin Elmer, Singapore) Soluble sugar concentration was determined using anthrone-sulfuric acid colorimetry at 620 nm wavelength (Bao, 2008). 1.00 g fresh celery sample was put into a large test tube, added with 15 mL distilled water, and boiled in boiling water bath for 20 min. Then samples were ltered into a 100 mL volumetric ask, washed the residue with distilled water for several times, constant volume to scale, take 1.0 mL of sample extraction solution to be tested, add 5 mL of anthrone reagent, shake and mix quickly, then boil for 10 min in boiling water bath after cooling, the optical density was measured with ultraviolet spectrophotometer (lambda 350V vis, Perkin Elmer, Singapore) at 620nm wavelength. The determination of Vitamin C concentration in plants was performed using 2, 6-dichloro indophenol titration (Tang et al, 2016). 2.00 g fresh celery samples and 3 mL 2% oxalic acid were ground to homogenate in mortar, and then the contents were transferred to a 50 mL volumetric ask containing 1% oxalic acid; 1 mL 30% ZnSO 4 and 1 mL 15% K 4 Fe (CN) 6 were added; after mixing, the suspension was diluted to 50 ml with 1% oxalic acid in the volumetric ask. Transfer 5 mL of sample solution into a conical ask and titrate with 2, 6-dichloroindole phenol (brown burette) until the sample solution turns reddish and does not fade within 15 s. Cellulose concentration was determined using anthrone-sulfuric acid colorimetry at 620 nm wavelength (Tang et al, 2016). Take celery sample (0.10 g, DW) into 60% H 2 SO 4 (30 mL) for 30 min in cold water bath, then dilute with 60% H 2 SO 4 in a volumetric ask to 50 mL, mix evenly, and lter with Buchner funnel. Dilute the ltrate sample (2.5 mL) to 50 mL with deionized water, add 2% anthrone (0.5 mL) and concentrated H 2 SO 4 (2 mL) and mix well. Let the solution stand for 12 min and cool to room temperature. Measure the absorbance at 620 nm with ultraviolet spectrophotometer (lambda 350V Vis, PerkinElmer, Singapore).

Determination Of Cd Uptake And Accumulation
The dried plant samples (0.20mg) were weighed into a PTFE digestion tube, and digested at 150°C for 5 h with 1 mL HClO 4 and 5 mL HNO 3 . After ltration to 30 mL, Cd concentrations were determined using an inductively coupled plasma-mass spectrometry (ICP-MS, Plasma Quant, Germany). Plant standard reference material (GBW (E) 100348, the National Research Center for Certi ed Reference Materials of China) was used for quality control (Bao, 2008).

Determination Of Soil PH And Rhizosphere Soil Cd Content
Shake off the soil that is easy to adhere to the root, and gently brush off the soil that is closely attached to the root with a brush, and collect it as rhizosphere soil.
After harvesting, soil samples were collected from eld blocks accordingly. Then soil samples were thoroughly mixed and air-dried in the laboratory, after which soils were ground and passed through 16 and 100 mesh sieves. After 30 minutes of preparation of soil suspension, the pH of soil was analyzed by pH meter in 1:2.5 soil water suspension.0.20 g soil sample was weighed and digested with 5 mL HNO 3 , 1 mL HClO 4 and 1 mL HF at 180°C for 10 h, and total Cd concentration in soil was determine by ICP-MS (Plasma Quant, Germany) (Bao, 2008). Soil standard reference material (GBW07429) was used as quality control sample.
Statistical analysis SPSS statistical 20.0 software (SPSS Inc., Chicago, IL.USA) was adopted for data analysis and graphical work was accomplished with Origin Pro 8.5 (Northampton, MA 01060 USA). Data were analyzed using oneway analysis of variance (ANOVA) at a signi cant level of p < 0.05 indicated by Duncan's test. The correlation was performed using bivariate analysis.

Biomass, moisture content and Chlorophyll concentration varied in different genotypes
After harvesting, biomass of all samples was determined. Our results showed that the biomass in edible part were largely different among all 23 genotypes (Table 1). Edible part biomass ranged from 0.14 (BL) to 1.19 (HQ) g·plant − 1 . Speci cally, cultivar HQ, JZ and JH was much heavier (> 1.0 g·plant − 1 ) than others, while BL exhibited the lowest biomass. Plant biomass in HQ, JZ and JH was 8.50-fold, 7.36-fold and 7.21fold of that in BL, accordingly. For moisture content, ranges differed between 94.03% (SQ) and 97.10% (BL) ( Table 1).  (Table 1).
Cadmium concentration, translocation factor and bio-concentration factor were different in different Genotypes As shown in Table 1 Translocation factor (TF) is an important index to evaluate a plant's ability to transport trace elements from roots to shoots. TF was calculated as the ratio of Cd concentration in shoots to the Cd concentration in roots. Cd TF in the celery showed sharp uctuations among different genotypes (Fig. 2), which varied from 0.40 (SQ) to 1.23 (OQ). In general, Cd TF in genotype OQ, BLC, JH and SH were obviously higher (TF > 1.0), while Cd TF in genotype SQ was the lowest (TF < 0.5) ( Fig. 2; Fig. 4).
Bio-concentration factor (BCF) is used to evaluate the ability of celery to accumulate Cd, which is calculated as the ratio of Cd concentration in the shoots (DW basis) to the Cd concentration in the soil. Cd BCF of 23 genotypes uctuated sharply. Cd BCF in XS, OQ and JH was obviously higher (> 3.0) than others, whereas genotype SQ showed a lowest Cd BCF (< 1.0) ( Fig. 2; Fig. 4). Cd BCF in in XS, OQ and JH was 6.52fold, 5.71-fold and 5.03-fold of that in SQ (Fig. 2).
Cd concentration of rhizosphere soil and pH of soil varied in different Genotypes As shown in Fig. 3, Cd concentration in rhizosphere soil after harvesting of all 23 genotypes ranged from 0.67 (JH) to 0.87 (LF) mg·kg − 1 , with a mean value of 0.75 mg·kg − 1 . Speci cally, soil Cd concentration in genotype LF was signi cantly higher than others, while genotype JH exhibited a lowest soil Cd concentration (0.67 mg·kg − 1 ). Soil Cd concentration in genotype LF was 1.30-fold of that in JH.
For soil pH, cultivation of SQ induced a reduced pH while BL cultivation resulted in a highest pH value ( Fig. 3; Fig. 4). However, tiny differences were detected among all the soil samples.   Fig. 4). Soluble sugar concentration in HQ was 11.81-fold of ZYX (Table 2).

Correlation Between Plant Cd Concentration And Nutrition Qualities
Celery biomass was signi cantly (r = 0.637; p < 0.01) and positively correlated with soluble sugar concentration in edible part (Table 3). And edible part Cd concentration was signi cantly (r = 0.713; p < 0.01) and positively correlated with cellulose concentration in edible part, but signi cantly (r = 0.571; p < 0.01) and negatively correlated with Vitamin C concentration (Table 3). While it was negatively correlated with biomass, chlorophyll concentration and soluble sugar concentration (r = 0.311, 0.131 and 0.254, respectively; p > 0.05) ( Table 3). Cellulose concentration in edible part was signi cantly (r = 0.540; p < 0.01) and negatively correlated with Vitamin C concentration (Table 3).

Discussion
Cd contamination in greenhouse elds has been a large limit for the sustainable production of leafy vegetables. It is reported that plant genotype was the most important plant factor affecting heavy metals uptake of plants (Bhargava et al., 2012). Speci cally, heavy metal uptake and accumulation varied among the genotypes of the same species (Ding et al., 2013a;Nabulo et al., 2012). Different Cd accumulation among different genotypes have been investigated in Chinese cabbage (Brassica pekinensis L.) (Wang et al., 2015;Tang et al., 2016;Wang et al., 2017), welsh onion (Allium stulosum L.) (Li et al., 2012;Li et al., 2016), spinach (Spinacia oleracea L.) (Tang et al., 2019a) and lettuce (Lactuca sativa L. var. ramosa Hort.) (Dala-Paula et al., 2019). Therefore, growing metal-excluding or low heavy metal accumulating genotypes of vegetables in the contaminated elds could be a simple but effective strategy to mitigate the transport of heavy metals to edible parts, and further effectively guarantee human health (Feng et al., 2018). Liu et al. (2009) recommended 4 identi cation standards of safe genotypes in Cd polluted areas, including (1) the concentration of Cd in the edible part should be lower than MPC; (2) BCF < 1.0; (3) TF < 1.0; and (4) when growing in the polluted soil, the aboveground biomass did not decrease signi cantly. We observed that there were signi cant differences in Cd concentration and distribution among 23 celery genotypes (Table 1), which were in consistence with previous literatures (Zhang et al., 2013). The genotypic difference of Cd content in edible parts among different genotypes of celery provides a potential possibility for screening low Cd accumulating celery genotypes. In this study, the content of Cd in the tested soil was 0.87 mg·kg − 1 , which belonged to moderate pollution soil according to Soil Environmental Quality-Risk Control Standard for Soil Contamination of Agricultural Land (GB15618-2018). However, Cd concentration in edible parts of all genotypes in this eld is lower than MPC (< 0.2 mg·kg − 1 FW), that is all the celery genotypes were quanti ed with the National Food Safety Standards (GB2072-2017), indicating that celery was a relatively safety leaf vegetable species for moderate polluted greenhouse. Among the 23 genotypes of celery, 19 genotypes had a lower TF than 1.0 (except for SH, JH, BLC and OQ) (Fig. 2), among which only SQ had a lower BCF which was less than 1.0 (Fig. 2). Similar research has reported that celery cultivar/genotype Hongchengsiji was considered a low Cd accumulating genotype due to its low Cd uptake and transport (Zhang et al., 2013). Our results suggested that SQ and LF were suitable to cultivate in Cdcontaminated elds to guarantee food safety.
The growth condition of different genotypes may vary signi cantly in the same eld (Table 1, Table 2). For example, Mi et al. (2019) reported that the fresh weight of 35 Chinese cabbage genotypes showed obvious differences. Another investigation also elucidated that different genotypes of celery showed discrepant growth condition (Golubkina et al., 2020). Similarly, our results showed that the edible biomass of HQ (the largest) was almost 8.30-fold of BL (the lowest), while aforementioned SQ and LF ranked sixth and eighth which is higher than the mean value of celery edible part (0.57g·plant − 1 ) ( Table 1). This could be partly explained by the higher chlorophyll content of SQ and LF (Table 1). Moreover, soluble protein concentration and vitamin C concentration of SQ and LF were higher than the mean value of 23 genotypes, respectively ( Table 2), suggesting that SQ and LF maintained a better nutrition level even on Cd-contaminated eld. Taken together, due to their larger biomass, slight Cd uptake, lower TF and BCF, together with better performance of nutrition indexes, SQ and LF were therefore selected as promising candidates to grow on a Cd-contaminated greenhouse eld to guarantee both the food production and human health.
The contents of vitamin C (Table 2) and cellulose (Table 2) in vegetables are important indexes to evaluate the nutritional quality of vegetables (Adenusi et al., 2015). Vitamin C is one of the most important nutrients to maintain the physiological function of human body. It is a kind of vitamin which human body needs the most and cannot synthesize itself (Ivey et al., 2012). Edible part Cd concentration was signi cantly and negatively correlated with Vitamin C concentration, Fu et al. (2009) and Hou et al. (2018) also showed that Cd has a signi cant limit on vitamin C in plants. When people cannot get enough vitamin C and other nutrition through vegetables, eating vegetables contaminated by heavy metals poses a threat to human health. Under the in uence of cadmium pollution in soil, celery growth was inhibited, enzyme activity decreased, ligni cation accelerated, senescence accelerated, cellulose content increased, celery stem and leaf ligni cation brosis degree increased, which would seriously affect the taste of celery, thereby reducing the quality of vegetables (Dureková et al., 2007;Rahoui et al., 2016). Cellulose is the main component of plant cell wall (Kumar et al., 2016). Studies have shown that plant tissue cell wall plays an important role in cadmium accumulation (Konno et al., 2010;50. Vatehová-Vivodová et al., 2018;Wu et al., 2020). There was a signi cant positive correlation between the cadmium content in the edible part of celery and the cellulose content in the edible part of celery (Table 3). It can be inferred that after cadmium enters celery, it might be mainly stored in the cell wall.

Conclusions
In this study, the biomass, Cd uptake and accumulation and nutrition traits of celery of different genotypes were compared in a greenhouse eld experiment. We concluded that more lights should be shed on SQ and LF to be cultivated in Cd-contaminated sites since they had both a heavier biomass and a lower Cd content in the edible part, which provided promising celery resources for safe production and human health of heavy metal polluted areas. Further correlation analysis showed that cadmium content in celery shoot was signi cantly and negatively correlated with vitamin C, but signi cantly and positively correlated with cellulose.

Declarations
Ethical approval: Not applicable. Competing interests: The authors declare that they have no competing interests.
Availability of data and materials: All data used or analyzed during this study are available from the corresponding author on reasonable request. Figure 1 Edible part Cd concentration of 23 celery genotypes grown on the Cd contaminated soil. Data were presented as means ± standard error (SE) with three replicates. Bars indicate signi cant difference at P < 0.05

Figure 2
Translocation factor and bio-concentration factor of 23 celery genotypes grown on the Cd contaminated soil. Data were presented as means ± standard error (SE) with three replicates. Bars indicate signi cant difference at P < 0.05 Figure 3 Cd concentration of rhizosphere soil and pH of soil of 23 genotypes of celery. Data were presented as means ± standard error (SE) with three replicates. Bars indicate signi cant difference at P < 0.05

Figure 4
The heatmap overview of all 23 celery respectively. The heatmap drawing was processed using an online platform (http://www.omicshare.com/tools).

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.