Discrimination of individual seed viability by using the oxygen consumption technique and headspace-gas chromatography-ion mobility spectrometry

TU Ke-ling ,YIN Yu-lin ,YANG Li-ming ,WANG Jian-hua ,SUN Qun#

1 Department of Seed Science and Biotechnology,College of Agronomy and Biotechnology,China Agricultural University/The Innovation Center (Beijing) of Crop Seeds Whole-Process Technology Research,Ministry of Agriculture and Rural Affairs/Beijing Key Laboratory of Crop Genetic Improvement,Beijing 100193,P.R.China

2 College of Arts and Science of Hubei Normal University,Huangshi 435109,P.R.China

3 College of Science,China Agricultural University,Beijing 100083,P.R.China

Abstract Identifying and selecting high-quality seeds is crucial for improving crop yield. The purpose of this study was to improve the selection of crop seeds based on separating vital seeds from dead seeds,by predicting the potential germination ability of each seed,and thus improving seed quality. The methods of oxygen consumption (Q) of seeds and the headspace-gas chromatography-ion mobility spectrometry (HS-GC-IMS) were evaluated for identifying the viability of individual seeds. Firstly,the oxygen consumption technique showed clear differences among the values related to respiratory characteristics for seeds that were either vital or not,and the discrimination ability of final oxygen consumption(Q120) was achieved not only in sweet corn seeds but also in pepper and wheat seeds. Besides,Qt was established as a new variable to shorten the measuring process in the Q2 (oxygen sensor) procedure,which was significantly related to the viability of individual seeds. To minimize seed damage during measurement,the timing for viability evaluation was pinpointed at the 12,6 and 9 h for pepper,sweet corn,and wheat seeds based on the new variables concerning oxygen consumption (i.e.,Q12,Q6 and Q9,respectively). The accuracies of viability prediction were 91.9,97.7 and 96.2%,respectively. Dead seeds were identified and hence discarded,leading to an enhancement in the quality of the seed lot as indicated by an increase in germination percentage,from 86.6,90.9,and 53.8% to all at 100%. We then used the HS-GC-IMS to determine the viability of individual sweet corn seeds,noting that corn seed has a heavier weight so the volatile gas components are more likely to be detected. A total of 48 chromatographic peaks were identified,among which 38 target compounds were characterized,including alcohols,aldehydes,acids and esters. However,there were no significant differences between the vital and dead seeds,due to the trace amount volatile composition differences among the individual seeds. Furthermore,a PCA based on the signal intensities of the target volatile compounds obtained was found to lose its effectiveness,as it was unable to distinguish those two types of sweet corn seeds. These strategies can provide a reference for the rapid detection of single seed viability.

Keywords: headspace-gas chromatography-ion mobility spectrometry,oxygen consumption,seed respiration,sweet corn,tomato,wheat

1.Introduction

Sweet corn (Zeamayscv.saccharatavar.rugosa),wheat (TriticumaestivumL.),and tomato (Lycopersicon esculentumMill.) are highly important agricultural products all over the world. As the basis of agricultural production,seeds greatly affect seedling emergence and further influence field yield. Even under a safe moisture content,the seeds would gradually lose vigor,ultimately leading to the loss of vitality (Thu-Phuonget al.2015;Michalaket al.2021). Hence,the vitality of seeds has attracted greater attention. However,the routine methods for monitoring seed viability such as the germination test,tetrazolium staining,and conductivity testing have drawbacks since they are time-consuming,require large numbers of seeds,or are destructive to the seeds (Michalaket al.2021). Therefore,new reliable tools for seed viability determination are needed which are both time-efficient and non-destructive (Maet al.2020;Michalaket al.2021).

Various methods and techniques for evaluating seed viability have been developed,and one good prospect among them is Q2 (oxygen sensor) technology. The Q2 instrument (www.astec-global.com),which is now called the Seed Respiration Analyzer,measures oxygen concentration with the same time interval to reflect the respiration rates of tested seed kernels in order to indicate different types of respiratory patterns (Bello and Bradford 2016,2021). Oxygen is essential for seed germination because the seeds need the energy produced by oxidative respiration. Respiration capability is an integrative indication of seed vigor. Seeds with higher vigor are more likely to be rehabilitated from dehydration when they are put through the imbibition process,and they subsequently start a larger amount of respiration at an earlier time (Hourmant and Pradet 1981;Alaniet al.1985). In contrast,seeds with lower vigor,such as those that are damaged and undeveloped,remain at a lower level of respiration,and those which are dead lack any ability to respire. Such variations in the requirements of oxygen during imbibition and germination can be caused by dormancy,after-ripening or priming,and they reflect the status of the seed (Pataneet al.2006;Bradfordet al.2007,2008).

Hence,the measurement of the oxygen related to germination contributes much to the evaluation of seed kernel quality. Previous measurements of respiratory activities relative to germination have confirmed a relationship between respiration and seed quality(Baalbakiet al.2009;Chenet al.2010;Wei 2014),although the variable change in results across seed species has raised concerns (Eliaset al.2012). These results may indicate that Q2 technology is expected to become a powerful tool for seed viability identification,due to its labor-saving and interpretation efficiency. However,evaluations based on the Q2 procedure may last for days,leading to the irreversible germination of the seeds by the end of the procedure (Bewleyet al.2013).

The analysis of volatiles is a potential non-invasive method for predicting seed viability. Gas chromatographyion mobility spectrometry (GC-IMS) is a gas-phase separation technology that allows for pre-separation of samples relying on gas chromatography,after which the compounds can be characterized according to their gas phase ion mobility (Liuet al.2020). GC-IMS does not require the time-consuming pretreatment of samples,so it provides a rapid detection method for the analysis of trace volatile compounds and establishing a comprehensive profile of them. As an emerging technology,GCIMS is currently most commonly used in food quality determination through the analysis of volatile compounds(Karpas 2013;Arroyo-Manzanareset al.2018;Geet al.2020;Liuet al.2020,2021). However,at this point,only a small amount of literature has reported on the relationship between volatile compounds and seed viability,but most of these studies used aged seeds or those stored under high humidity (Miraet al.2010;Colvilleet al.2012;Michalaket al.2021;Zhanget al.2021),and there is a lack of studies on naturally untreated and individual crop seeds.

Therefore,this study aims: (1) to verify the relationship between original Q2 datasets and seed viability,other than the systematically generated ASTEC values(Increased Metabolism Time,IMT;Oxygen Metabolism Rate,OMR;Critical Oxygen Pressure,COP;Relative Germination Time,RGT),and to shorten the evaluation process by exploring highly correlated variables at an early stage;and (2) to identify the volatile compounds involved and establish the flavor fingerprints of vital and dead seeds based on headspace-gas chromatographyion mobility spectrometry (HS-GC-IMS).

2.Materials and methods

2.1.Seed respiration measurements

A total of 175 sweet corn,149 pepper,and 206 wheat seeds were used as the materials to measure single seed respiration by using the oxygen consumption technique.The seeds for each crop came from the same variety and were randomly selected. Samples were planted in microtiter plates with 48 wells (one seed per well),with 1 000 μL of 0.5% agar in a 2.0-mL tube. Measurements were performed for 120 h with an interval of 0.33 h(20 min) under a constant temperature of 25°C. The information and codes of the samples were inputted into the Q2 Operation Software. The samples were sealed air-tight with photosensitive material and the data of oxygen content (relative value,percent of initial value)were recorded by the Q2 instrument (ASTEC-Global,the Netherlands). Datasets were then output as Excel files.

Seed viability was evaluated and categorized as vital or dead,depending on whether the seeds germinated at the end of the measurement. ASTEC values were generated automatically by the Q2 Software,but oxygen consumption (Qt) at timetwas calculated as follows:

whereYtis the recorded oxygen concentration atth with an interval of 0.33 h (20 min);Y0.33is the first data point obtained in the procedure;Qtrefers to the oxygen concentration when the measuring time reachesth,and hence the values of Qtare mostly positive;the value oftis supposed to be as low as possible to keep the samples from germinating irreversibly.

The data generated after viability evaluation,together with the corresponding respiratory data,were randomly separated into a calibration set (75%) and a test set (25%).The calibration set was used to conduct further analysis in terms of variable development,and the testing set was used to validate the results. The accuracies of individual seed viability predictions were calculated and are presented here in the form of percentages to indicate the ultimate values of the new variable(s). Statistical analysis was conducted using Microsoft Excel 2019 and SPSS 25.0.Graphs were made using Microsoft Excel 2019 and Origin 2021.

2.2.GC-IMS analysis

To investigate whether HS-GC-IMS can be used as a non-destructive method in seed viability determination according to the volatile compounds released by vital and dead seeds,150 sweet corn seeds of the same variety were tested. Because the weight of a single sweet corn seed is the heaviest compared with wheat and pepper seed,the volatile gas components are most likely to be detected. The GC-IMS device (FlavourSpec?,G.A.S.Instrument,Germany) was used to identify the volatile compounds of the sweet corn seeds. The instrumental analysis software includes Laboratory Analysis View(LAV),Reporter plug-ins,Gallery Plot plug-ins,Dynamic PCA plug-ins,and a GC×IMS Library Search.

A single seed of sweet corn was weighed,placed into the 20-mL headspace glass vial and then incubated at 40°C for 15 min. After this process,500 μL of the headspace gas was injected into an FS-SE-54-CB-1(15 m×0.53 mm,0.5 μm) capillary column by nitrogen(99.999% purity),with a heated syringe at a temperature of 85°C. Nitrogen was used as carrier gas at a flow rate of 150 mL min-1. The flow rate was set to 2 mL min-1for the first 2 min,then linearly increased to 10 mL min-1within 8 min,further increased linearly to 100 mL min-1within 10 min,and finally to 150 mL min-1for 5 min. The column and IMS ionization chamber were kept at 60 and 45°C,respectively. Each seed was analyzed according to this process.

The n-ketones C4-C9 (Sinopharm Chemical Reagent Beijing Co.,Ltd.,China) were used as external standards to calculate the retention index (RI) of volatile compounds.By the comparison of RI and the drift time,the volatile compounds were tentative identified. Immediately afterward,the qualitative analysis of the substance was carried out according to the 2014 NIST database and IMS database built into the GC×IMS Library Search.

After the nondestructive GC-IMS analysis,a singlekernel germination test was conducted to obtain the viability of each of the 150 sweet corn seeds. Seeds that germinated within 7 d were considered vital,and those that did not germinate were considered to be dead seeds.

3.Results

3.1.Q2 technology

Final oxygen consumption (Q120) as a substitute for ASTEC valuesThe oxygen consumption curves of pepper,sweet corn,and wheat are shown in Fig.1. Vital seeds had a rapid imbibition process and consumed a lot of oxygen,yielding a sharply decreasing oxygen consumption curve. In contrast,there was no observable change in the dead seeds. Through 120 h of oxygen consumption measurement,the 175 sweet corn seeds were divided into 159 vital and 16 dead seeds,and a similar division was obtained for the 149 pepper seeds(129 vital,20 dead) and the 206 wheat seeds (111 vital,95 dead).

Fig.1 Oxygen consumption curves for individual seeds of pepper (A),sweet corn (B),and wheat (C). The typical curves are shown for a rapidly germinating seed (left curve) and a slowly germinating seed (right curve). Seeds generally exhibit the two-stage or sigmoidal patterns shown here,but some seeds have linear patterns or other variations.

As shown in Fig.2,the various standard deviations and arithmetic means indicate inconsistent and different distribution patterns amongst the ASTEC-viability results within a given crop species,while those of the Q120value were comparably consistent. Q120was found to be significantly different between the dead and vital seeds in each crop species. Some studies have shown that part of the ASTEC values were correlated with some characteristics during seed germination. For example,seed viability of conventional rice could be well established based on RGT (Chenet al.2013). A certain range in the values of increased metabolism time (IMT),oxygen metabolism rate (OMR),and relative germination time (RGT) were suggested to indicate vitality in waxy maize (Wanget al.2016). Critical oxygen pressure (COP) was found to be highly correlated with field germination percentage in China fir (Zhaoet al.2012). The results of this study verified the differences in ASTEC values among crop species,but conflicted with those in previous studies,possibly due to the differences in relation to the numerical versions of the measured objects. This study focused specifically on identifying the viability of individual seeds.

Compared with the four ASTEC values,the data for the final oxygen consumption of vital and dead seeds were found to be aggregated separately within a particular range,with little overlap. Such distribution patterns of the Q120values would contribute to identifying the viability of a single tested seed of the pepper,sweet corn,or wheat species. Moreover,the Q120values had the highest correlation with seed viability (Fig.3-A-C). Hence,the regressive functions linking ASTEC values to seed viability could not be established,but it was still possible to differentiate the individual seed viability states for the individual pepper,sweet corn,and wheat seeds according to the distribution of the Q120values. In other words,seeds identified as vital had a larger Q120value,while those identified as dead had a lower Q120value.

Due to the small proportion of overlapping viability states,it would be acceptable if a few vital seeds were discarded to guarantee the higher quality of the seed lots.To guarantee the maximum enhancement of the quality,a new Q2 variable concerning the viability was set up as the max Q120value of dead seeds. According to the data distribution of the calibration set,the separation point for vital pepper seed was Q120＞8.2 (Fig.3-D),that for vital sweet corn seed was Q120＞20.9 (Fig.3-E) and that for vital wheat seed separation was Q120＞20.6 (Fig.3-F). The accuracies of the test set for predicting both vital and dead of pepper,sweet corn,and wheat seeds were found to be high,at 91.9,100 and 100%,respectively (Fig.3-G-I).All of their germination percentages eventually increased from 86.6,90.9 and 53.8% to 100% after seed selection by Q120.

Fig.3 Pearson correlation matrix of Q2 (oxygen sensor) parameters and seed viability (A,B and C),scatter plots of the final oxygen consumption (Q120) values for individual vital and dead pepper (D),sweet corn (E),and wheat (F) seeds,and confusion matrices of the prediction results in the test set using Q120 values (G,H and I). Each column represents a crop,namely pepper,sweet corn and wheat. IMT,increased metabolism time;OMR,oxygen metabolism rate;COP,critical oxygen pressure;RGT,relative germination time.

Early viability evaluation based on oxygen consumptionIn order to shorten the whole process of oxygen consumption measurement to prevent the irreversible germination of seeds,we explored the variables of early viability evaluation. An initially increasing pattern in the correlation coefficients between oxygen consumption atth (Qt) was discovered in the Q2 procedure. In the later stages when the measurement time reached 12,6 and 9 h for pepper,sweet corn,and wheat seeds,the values of the correlation coefficients remained stable (Fig.4).Such a trend suggested the possibility for an early viability assessment. Therefore,times were pinpointed mostly based on the curve,and this suggested that viability assessment could be done within several hours rather than delaying it until the end of the procedure. Similar to the analysis of Q120,these results revealed the value of oxygen consumption relative to identifying the viability states for tomato,sweet corn,and wheat seeds. Pepper seeds with Q12greater than 4.6 were identified as vital,sweet corn seeds with Q6greater than 3.6 were identified as vital,and wheat seeds with Q9greater than 1.9 were identified as vital(Fig.5-A-C). All three thresholds were obtained with high accuracy,at 91.9,97.7 and 96.2%,respectively (Fig.5-D-F). Their germination percentages also increased to 100%after seed selection by Qt,with some acceptable loss of vital seeds. The results in this section suggested that rather than completing the entire 120-h procedure,it could be shortened to 12 h for pepper seeds,6 h for sweet corn seeds and 9 h for wheat seeds.

Fig.4 Scatter plot of spearman correlation coefficients of oxygen consumption (Q) at t h and viability for pepper,sweet corn and wheat seeds. Note that all correlation coefficients were significant at the 0.01 level.

Fig.5 Scatter plots of the oxygen consumption (Qt) value for individual vital and dead pepper (A),sweet corn (B) and wheat (C)seeds,and confusion matrices of the prediction results in the test set using oxygen consumption at 12 h (Q12) for pepper (D),Q6 for sweet corn (E),and Q9 for wheat (F).

3.2.Viability evaluation for individual seeds based on GC-IMS

In order to develop a nondestructive seed viability detection method,we also used GC-IMS technology to determine the volatile gas components of sweet corn seeds in order to find the differences between individual vital and dead seeds. In this part,after the GC-IMS non-destructive testing,the germination test was carried out for the 150 sweet corn seeds,113 of which germinated and were considered as vital seeds,while the remaining 37 ungerminated seeds were considered as dead seeds. The HS-GC-IMS spectra of vital and dead seeds are presented in Fig.6,from which the information regarding their volatile compounds can be obtained. Thex-axis represents the ion relative drift time of the volatile flavor compounds,and they-axis represents the gas phase retention time,in both Fig.6-A and B,while the third-dimension coordinate represents the peak intensity in Fig.6-A. It was usually difficult to discern the diversity of volatile flavor compounds of sweet corn seeds through three-dimensional spectroscopy(Fig.6-A),therefore two-dimensional spectroscopy was adopted for further comparisons (Fig.6-B). In Fig.6-B,the red vertical line with an abscissa of 1.0 represents the reactive ion peak (RIP),and each point on the right represents a volatile flavor compound. The intensity information is represented by color,in which the white corresponds to a relatively low concentration of volatiles and red to a higher concentration (Liuet al.2020).

Fig.6 Comparison of the volatile profiles of vital and dead sweet corn seeds with three-dimension spectra (A) and two-dimension spectra (B). The data for one vital seed and one dead sweet corn seed were randomly selected for presentation. RIP,reactive ion peak. V,intensity.

The changes in fingerprint volatiles of more vital and dead sweet corn seeds are shown in Fig.7. A single seed is displayed in each row of the fingerprint,and each volatile flavor compound is represented by a column.Therefore,it was feasible to roughly determine the content of volatile compounds by the color of each square (Liuet al.2021). A total of 38 volatile flavor compounds were identified in sweet corn seeds,mainly including alcohols,ketones,aldehydes and esters. Compounds numbered 1-10 are the volatile flavor compounds that were not identified. The results showed that all seeds possessed similar volatile flavor compounds,whether vital or dead.There were some differences in the peak signal intensities among the samples,which indicated that the contents of volatile flavor compounds revealed some diversity among the different seeds. However,it was still difficult to distinguish the vital seeds from the dead seeds in the absence of a regular pattern.

Fig.7 Changes in the fingerprint volatiles of vital and dead sweet corn seeds. Numbers 1-10 are the unidentified volatile compounds in sweet corn seeds.

To establish the relationship between seed viability and the volatile composition of sweet corn,a principal component analysis (PCA) was applied. As shown in Fig.8,the accumulative variance contribution (57.6%)indicates that the three principal components cannot reveal most of the information of the original data. The seed samples that were vital and dead were not well separated,and exhibited no obvious division. The results showed that this approach was insufficient to distinguish the viability of a single seed according to its volatile flavor compounds.

Fig.8 Principal component analysis (PCA) results for the volatile flavor compounds of sweet corn seeds.

4.Discussion

4.1.Early oxygen consumption makes rapid seed viability assessment possible

Seed viability is routinely evaluated for commercial seed lots,and the priority for vegetable technology development should focus on seed quality enhancement (Qu 2003).For precise agronomy and food security issues,it is well established that the viability of individual seeds should be identified more strictly at a lower cost of testing resources,especially for crops like pepper (Tuet al.2018) and maize. Fortunately,Q2 provides an option for detecting seeds with different viability levels,by enlarging the gap by imbibition,which is of great significance for the selection of high value-added vegetable seeds. The Q2 instrument measures the respiration of individual seeds,then the software automatically calculates a number of parameters,such as the IMT,OMR,RGT,and COP,called ASTEC values,which can be used as relatively rapid seed quality assessment indices (Chenet al.2010;Bradfordet al.2013;Bello and Bradford 2021). Although it has been accepted as a fast and efficient method (Heet al.2019;Bello and Bradford 2021),the tested seeds may still irreversibly germinate when the procedure lasts for 120 h.This study aimed to explore the relationship between seed respiration and seed viability in an individual version and therefore proposes a better approach to secure the viability evaluation in a non-destructive way,namely to prevent samples (or most samples) from germinating.

In practical applications,the vital and dead seeds can be distinguished according to their differential early oxygen consumption (a clustering strategy can be adopted) of seed lots,to improve seed quality. In this way,selected vital samples could be reused in practical production,which would reduce the amount of time and energy consumed during the experimental evaluation process. To that end,we also made some attempts,to see whether a certain time after imbibition was required for early oxygen consumption evaluation. The seeds did not germinate,and the seeds were allowed to recover to the original water content,which did not lead to an abnormal performance in the subsequent germination test. On the contrary,a few hours of imbibition played a role of hydro-priming,which would lead to more uniform germination,especially for vegetable seeds. However,it should be noted that due to the different states of seeds and the change in the environment during imbibition,naturally the Q2 technology will be affected,and that change will be greater among seeds for different crops due to differences in respiratory intensity. Therefore,the methods we explored are more suitable for the distinction between dead and vital seeds from the same seed lot. In further studies,we will also apply this strategy to more crops and more varieties of the same crop,to find more stable and universal parameters or indices.

4.2.The preliminary attempt to evaluate individual seed viability by GC-IMS

Inspired by Q2 technology,we tested the application of the non-destructive and rapid profiling GC-IMS technology to single seed viability discrimination for the first time. According to the characteristics of GCIMS technology,detecting the volatile gas composition of samples requires a certain concentration of those gases,so the weight of samples is a key factor. Thus,we first chose sweet corn seed,which has the greatest weight of a single seed compared with wheat and pepper seed,so its volatile gas components are the most likely to be detected. Accordingly,38 target volatile compounds were characterized,including alcohols,aldehydes,acids,and esters,which was consistent with previous studies(Zhanget al.2020,2021;Michalaket al.2021). However,further analysis showed that this profile was insufficient to distinguish the viability of a single sweet corn seeds according to their volatile flavor compounds. The failure of this approach might be due to the relatively light weight of a single sweet corn seed (about 0.2 g) and the faint flavor of dry seeds. According to previous research,5 g samples were needed for characterizing white and deteriorated yellow rice (Zhanget al.2020). In other studies on the detection of different quality levels in seeds,～1 g seeds with high moisture content were stored in headspace vials for 24 h to allow the volatile compounds trapped within the seed to be fully released prior to detection (Michalaket al.2021),and 3 g seed samples were needed for predicting seed vigor of naturally-aged seed lots (Zhanget al.2021).

Given the above results,we have some ideas for future improvements and applications of this method.For example,we could use the non-destructive GC-IMS method to identify seed lots with differences in viability or monitor the viability changes of seed lots during storage.In this way,sufficiently large seed samples could be obtained for testing,instead of a single seed,to make the best use of this non-destructive technology. For one single seed,the headspace vial (which is now 20 mL)needs to be smaller in the future,or the peculiar flavor seeds of traditional Chinese medicine could be tested as an alternative,in order to increase the concentration of volatile gas components and better detect the subtle differences in volatile flavor compounds between single vital and dead seeds. We remain optimistic that with the further development of this technology,more volatile gas components can be determined,and trace amount differences may be detected in the future.

5.Conclusion

The limitations in the routine methods for detecting seed viability have inspired us to explore new reliable tools for seed viability determination. Our results indicate that the oxygen consumption technique can meet the needs of many applications,as the final oxygen consumption(Q120) has successfully distinguished dead seeds from vital seeds with the highest (100%) accuracy for both sweet corn and wheat seeds. In addition,an earlier stage new variable,Qt,has proven to be significantly related to the viability of individual crop seeds,which could not only minimize seed damage during measurement,but also achieve high accuracy for viability prediction(～96%). Moreover,after the seed lots were selected by the corresponding variable Qt,the final germination percentages all increased to 100%,whether involving sweet corn,wheat,or pepper seeds,which leads to an enhancement in crop seed quality. Furthermore,to our knowledge,this is the first study to use the headspacegas chromatography-ion mobility spectrometry (HS-GCIMS) and early oxygen consumption for individual seed viability identification. Although the determination of vital and dead seeds based on volatile compounds has not achieved satisfactory results,it is still a pioneering exploration with great hope for further improvements.These strategies can provide a reference for the reliable and rapid detection of single seed viability.

Acknowledgements

This work was supported by the National Key Research and Development Program of China (2018YFD0100903).

Declaration of competing interest

The authors declare that they have no conflict of interest.