Figure 1.
Fourier-transform infrared (FT−IR) raw spectra of A. thaliana powder samples.
Figure 1.
Fourier-transform infrared (FT−IR) raw spectra of A. thaliana powder samples.
Figure 2.
Standard normal variate (SNV) preprocessed FT−IR spectra (a). (a1,a2) are the extended spectral regions related to phenolic compounds.
Figure 2.
Standard normal variate (SNV) preprocessed FT−IR spectra (a). (a1,a2) are the extended spectral regions related to phenolic compounds.
Figure 3.
Fourier-transform near-infrared (FT−NIR) raw spectra of A. thaliana powder samples.
Figure 3.
Fourier-transform near-infrared (FT−NIR) raw spectra of A. thaliana powder samples.
Figure 4.
FT−NIR SNV preprocessed spectra (a). (a1) is the extended spectral regions related to phenolic compounds.
Figure 4.
FT−NIR SNV preprocessed spectra (a). (a1) is the extended spectral regions related to phenolic compounds.
Figure 5.
A total of 100 mixed samples FT−IR spectra of Arabidopsis powder samples for one concentration created by Dirichlet distribution (a). FT-IR spectra developed between two replicates, i.e., red drought_1 and red non-drought_1 (RD_1 and RND_20) for one variety of concentrations (b).
Figure 5.
A total of 100 mixed samples FT−IR spectra of Arabidopsis powder samples for one concentration created by Dirichlet distribution (a). FT-IR spectra developed between two replicates, i.e., red drought_1 and red non-drought_1 (RD_1 and RND_20) for one variety of concentrations (b).
Figure 6.
A total of 100 mixed samples FT-NIR spectra of Arabidopsis powder samples for one concentration created by Dirichlet distribution (a). FT-NIR spectra developed between two replicates i.e., white drought_1 and white non-drought_1 (WD_1 and WND_10) for one concentration (b).
Figure 6.
A total of 100 mixed samples FT-NIR spectra of Arabidopsis powder samples for one concentration created by Dirichlet distribution (a). FT-NIR spectra developed between two replicates i.e., white drought_1 and white non-drought_1 (WD_1 and WND_10) for one concentration (b).
Figure 7.
Principal component analysis of Arabidopsis thaliana powder samples for (a) FT−IR and (b) FT-NIR spectroscopy under different stress conditions. Here, the abbreviations RBD, RBND, RD, RND, WD, and WND stand for red+ blue drought, red+ blue non-drought, red drought, red-blue non-drought, white drought, and white non-drought.
Figure 7.
Principal component analysis of Arabidopsis thaliana powder samples for (a) FT−IR and (b) FT-NIR spectroscopy under different stress conditions. Here, the abbreviations RBD, RBND, RD, RND, WD, and WND stand for red+ blue drought, red+ blue non-drought, red drought, red-blue non-drought, white drought, and white non-drought.
Figure 8.
Actual and predicted concentration values for the phenolic compounds in A. thaliana leaf powder samples using the PLSR model for (a) calibration and (b) prediction datasets. Here, RMSEC and RMSEP represent root-mean-square error for calibration and prediction.
Figure 8.
Actual and predicted concentration values for the phenolic compounds in A. thaliana leaf powder samples using the PLSR model for (a) calibration and (b) prediction datasets. Here, RMSEC and RMSEP represent root-mean-square error for calibration and prediction.
Figure 9.
PCR graphs of actual and predicted concentration values for the phenolic compounds in Arabidopsis powder samples using (a) calibration and (b) prediction datasets, respectively.
Figure 9.
PCR graphs of actual and predicted concentration values for the phenolic compounds in Arabidopsis powder samples using (a) calibration and (b) prediction datasets, respectively.
Figure 10.
Hybrid linear analysis (HLA/GO) model for actual and predicted concentration values for the phenolic compounds in A. thaliana powder samples for (a) calibration and (b) prediction datasets.
Figure 10.
Hybrid linear analysis (HLA/GO) model for actual and predicted concentration values for the phenolic compounds in A. thaliana powder samples for (a) calibration and (b) prediction datasets.
Figure 11.
Actual and predicted concentration values for the phenolic compounds in A. thaliana leaf powder samples using the PLSR model for (a) calibration and (b) prediction datasets. Here, RMSEC and RMSEP represent the root-mean-square error for calibration and prediction, respectively.
Figure 11.
Actual and predicted concentration values for the phenolic compounds in A. thaliana leaf powder samples using the PLSR model for (a) calibration and (b) prediction datasets. Here, RMSEC and RMSEP represent the root-mean-square error for calibration and prediction, respectively.
Figure 12.
PCR graphs of actual and predicted concentration values for the phenolic compounds in A. thaliana leaf powder samples using (a) calibration and (b) prediction datasets, respectively.
Figure 12.
PCR graphs of actual and predicted concentration values for the phenolic compounds in A. thaliana leaf powder samples using (a) calibration and (b) prediction datasets, respectively.
Figure 13.
Actual and predicted concentration values for the phenolic compounds in A. thaliana leaf powder samples using the hybrid linear analysis (HLA/GO) model for (a) calibration and (b) prediction datasets. Here, RMSEC and RMSEP represent the root-mean-square error for calibration and prediction, respectively.
Figure 13.
Actual and predicted concentration values for the phenolic compounds in A. thaliana leaf powder samples using the hybrid linear analysis (HLA/GO) model for (a) calibration and (b) prediction datasets. Here, RMSEC and RMSEP represent the root-mean-square error for calibration and prediction, respectively.
Figure 14.
Beta coefficient plots developed through the PLSR model: (a) FT−IR spectroscopy and (b) FT−NIR spectroscopy.
Figure 14.
Beta coefficient plots developed through the PLSR model: (a) FT−IR spectroscopy and (b) FT−NIR spectroscopy.
Figure 15.
Arabidopsis Col-10 seeds (a), seeds grown in a Petri dish (b) under different lighting conditions: (c) red + blue, (d) red, and (e) white.
Figure 15.
Arabidopsis Col-10 seeds (a), seeds grown in a Petri dish (b) under different lighting conditions: (c) red + blue, (d) red, and (e) white.
Figure 16.
RGB images of non-drought and drought conditions of A. thaliana plants under different LED lightning conditions. Here, RBND and RBD (a,b) represents red + blue non-drought and red + blue drought, RND and RD (c,d) represents red non-drought and red drought, and WND and WD (e,f) represents white non-drought and white drought respectively.
Figure 16.
RGB images of non-drought and drought conditions of A. thaliana plants under different LED lightning conditions. Here, RBND and RBD (a,b) represents red + blue non-drought and red + blue drought, RND and RD (c,d) represents red non-drought and red drought, and WND and WD (e,f) represents white non-drought and white drought respectively.
Figure 17.
Flowchart for A. thaliana powder samples spectral data analysis.
Figure 17.
Flowchart for A. thaliana powder samples spectral data analysis.
Table 1.
FT-IR spectral vibrations of phenolic compounds observed in A. thaliana leaf powder samples.
Table 1.
FT-IR spectral vibrations of phenolic compounds observed in A. thaliana leaf powder samples.
Spectroscopic Technique | Absorption Frequency, ν (cm−1) | Assignment |
---|
FT-IR spectroscopy | 3500–2500 | O-H stretching |
1700–1600 | C=O stretching |
2954 and 2850 | C-H stretching |
1505 | C=C stretching |
1600 | Benzene ring skeleton |
1500 | C=C aromatic stretching |
Table 2.
FT-NIR spectral vibrations of phenolic compounds observed in A. thaliana leaf powder samples.
Table 2.
FT-NIR spectral vibrations of phenolic compounds observed in A. thaliana leaf powder samples.
Spectroscopic Technique | Absorption Frequency, ν (cm−1) | Assignment |
---|
FT-NIR spectroscopy | 8350 | Second overtone of C-H stretching |
6000–7000 | First overtone of the O-H and N-H stretching |
5172 | Combination of O-H and C-O stretching |
4813 | Combination of O-H bending and C-O stretching |
4450 to 4285 | Combination band region |
4450 and 4410 | O-H bond combined with the C-O bond |
4380 and 4315 | the C-H bond |
4285 | The C-H bond combined with the C-H bond |
Table 3.
The reference values of phenolic compounds (mg/g dry weight (DW)) obtained from the HPLC analysis.
Table 3.
The reference values of phenolic compounds (mg/g dry weight (DW)) obtained from the HPLC analysis.
Phenolics | Red + Non-Drought | Red + Drought | Red-Blue + Non-Drought | Red-Blue + Drought | White + Non-Drought | White + Drought |
---|
Gallic acid | ND | ND | ND | ND | 0.043 ± 0.010 a 1 | 0.017 ± 0.002 b |
Catechin | 0.141 ± 0.003 a | 0.165 ± 0.021 a | 0.130 ± 0.014 a | 0.134 ± 0.014 a | 0.150 ± 0.011 a | 0.145 ± 0.026 a |
Chlorogenic acid | 0.123 ± 0.002 a | 0.119 ± 0.011 a | ND | ND | 0.123 ± 0.005 a | 0.137 ± 0.009 a |
Caffeic acid | 0.049 ± 0.010 b | 0.060 ± 0.007 b | ND | ND | 0.059 ± 0.005 b | 0.050 ± 0.011 b |
(-)-Epicatechin | ND | ND | ND | ND | 0.055 ± 0.011 b | 0.037 ± 0.004 b |
Epicatechin gallate | 0.124 ± 0.006 c | 0.255 ± 0.030 b | ND | ND | 0.743 ± 0.023 a | 0.302 ± 0.030 b |
Ferulic acid | 0.033 ± 0.013 cd | 0.053 ± 0.001 c | ND | ND | 0.138 ± 0.014 b | 0.384 ± 0.029 a |
Sinapic acid | ND | 0.015 ± 0.002 b | ND | ND | 0.032 ± 0.002 a | 0.035 ± 0.007 a |
Benzoic acid | 0.136 ± 0.002 b | 0.138 ± 0.009 b | ND | 0.135 ± 0.010 b | ND | ND |
Rutin | 0.340 ± 0.005 b | 0.339 ± 0.018 b | 0.464 ± 0.132 ab | 0.600 ± 0.168 a | 0.39 ± 0.040 ab | 0.390 ± 0.043 ab |
Quercetin | 0.281 ± 0.004 a | 0.339 ± 0.049 a | 0.283 ± 0.016 a | 0.354 ± 0.085 a | 0.287 ± 0.018 a | 0.259 ± 0.014 a |
Kaempferol | 0.085 ± 0.008 b | 0.098 ± 0.017 b | 0.100 ± 0.014 b | 0.086 ± 0.013 b | 0.174 ± 0.038 a | 0.104 ± 0.016 b |
TOTAL | 1.311 ± 0.013 cd | 1.582 ± 0.063 bc | 0.977 ± 0.136 d | 1.309 ± 0.241 cd | 2.194 ± 0.053 a | 1.859 ± 0.084 b |
Table 4.
Datasets used for FT-IR and FT-NIR spectroscopy.
Table 4.
Datasets used for FT-IR and FT-NIR spectroscopy.
Technique (n = 600) | Samples | Number of Samples (Calibration) | Number of Samples (Prediction) |
---|
FT-IR spectroscopy | Arabidopsis powder samples | 360 | 240 |
Table 5.
Results from the developed PLSR, PCR, and HLA/GO models for the prediction of phenolic compounds in A. thaliana leaf powder samples using FT-IR spectroscopy.
Table 5.
Results from the developed PLSR, PCR, and HLA/GO models for the prediction of phenolic compounds in A. thaliana leaf powder samples using FT-IR spectroscopy.
Region | Model/Preprocessing | | RMSEC (mg/g) | | RMSEP (mg/g) | LVs |
---|
FT-IR spectroscopy | PLSR/Mean norm. | 0.983 | 0.051 | 0.978 | 0.058 | 8 |
PLSR/MSC | 0.981 | 0.054 | 0.981 | 0.056 | 8 |
PLSR/SNV | 0.981 | 0.053 | 0.980 | 0.055 | 8 |
PLSR/SG-1 | 0.969 | 0.053 | 0.975 | 0.063 | 8 |
PLSR/SG-2 | 0.968 | 0.071 | 0.972 | 0.066 | 5 |
PLSR/Raw | 0.970 | 0.069 | 0.968 | 0.070 | 8 |
PCR/SG-1 | 0.949 | 0.089 | 0.963 | 0.077 | 6 |
HLA/GO/SNV | 0.929 | 0.109 | 0.941 | 0.100 | 8 |
Table 6.
Results from the developed PLSR, PCR, and HLA/GO models developed for the prediction analysis of phenolic compounds in A. thaliana leaf powder samples using FT-NIR spectroscopy.
Table 6.
Results from the developed PLSR, PCR, and HLA/GO models developed for the prediction analysis of phenolic compounds in A. thaliana leaf powder samples using FT-NIR spectroscopy.
Region | Model/Preprocessing | | RMSEC (mg/g) | | RMSEP (mg/g) | LVs |
---|
FT-NIR spectroscopy | PLSR/Mean norm. | 0.943 | 0.094 | 0.931 | 0.104 | 5 |
PLSR/MSC | 0.999 | 0.003 | 0.999 | 0.003 | 7 |
PLSR/SNV | 0.999 | 0.003 | 0.999 | 0.004 | 7 |
PLSR/SG-1 | 0.993 | 0.031 | 0.991 | 0.036 | 6 |
PLSR/SG-2 | 0.993 | 0.032 | 0.991 | 0.037 | 5 |
PLSR/Raw | 0.927 | 0.107 | 0.912 | 0.118 | 6 |
PCR/MSC | 0.999 | 0.004 | 0.999 | 0.003 | 6 |
HLA/GO/SNV | 0.922 | 0.116 | 0.897 | 0.131 | 5 |