Article pubs.acs.org/JAFC
Metabolite Profiling of Barley Grain Subjected to Induced Drought Stress: Responses of Free Amino Acids in Differently Adapted Cultivars Alexandra Lanzinger,† Thomas Frank,† Gabriela Reichenberger,§ Markus Herz,§ and Karl-Heinz Engel*,† †
Lehrstuhl für Allgemeine Lebensmitteltechnologie, Technische Universität München, Maximus-von-Imhof-Forum 2, D-85350 Freising-Weihenstephan, Germany § Bavarian State Research Center for Agriculture (LfL), Institute for Crop Science and Plant Breeding, Am Gereuth 8, D-85354 Freising, Germany S Supporting Information *
ABSTRACT: To investigate cultivar-specific metabolite changes upon drought stress in barley grain, differently adapted cultivars were field-grown under drought conditions using a rain-out shelter and under normal weather conditions (2010−2012). The grain was subjected to a gas chromatography−mass spectrometry-based metabolite profiling approach allowing the analyses of a broad spectrum of lipophilic and hydrophilic low molecular weight constituents. Multi- and univariate analyses demonstrated that there are grain metabolites which were significantly changed upon drought stress, either decreased or increased in all cultivars. On the other hand, for proteinogenic free amino acids increased concentrations were consistently observed in all seasons only in cultivars for which no drought resistance/tolerance had been described. Consistent decreases were seen only in the group of stress tolerant/resistant cultivars. These cultivar-specific correlations were particularly pronounced for branched-chain amino acids. The results indicate that free amino acids may serve as potential markers for cultivars differently adapted to drought stress. KEYWORDS: metabolite profiling, barley grain, drought stress, Hordeum vulgare L., free amino acids
■
INTRODUCTION In 2013, coarse grain world production yielded approximately 1300 million tons. Barley (Hordeum vulgare L.) amounted to 145 million tons worldwide and is thus the second largest coarse grain after maize.1 As feed, either the whole barley plant or only the grain is used. For humans, the grain is malted or used directly as food. Because barley plants show high adaptation to a wide range of environmental conditions, cultivation takes place even under conditions when other cereals cannot be grown, for example, due to high evaporation and/or lack of rainfall.2 However, owing to global climate change water deficit is also becoming an increasing challenge for barley cultivation.3 Levels of plant metabolites are affected by drought stress to protect plants’ cellular turgor and to regulate ion transport, synthesis and activation of enzymes, gene expression, and redox-homeostasis.4−6 Comprehensive approaches such as metabolite profiling have been shown to be suitable tools for the investigation of drought stress metabolism. For example, the impact of water deficit has been studied in Arabidopsis and in the model legume Lotus japonicus.7,8 Furthermore, the metabolite profiles of drought-stressed plants from different types of cereals such as Zea mays L., Oryza sativa L., and Triticum L. have been analyzed.9−11 A recent study investigated the impact of induced drought stress on the metabolite profiles of barley grain.12 The experimental setup was based on six cultivars field-grown under natural weather conditions as control and under induced drought conditions, using a rain-out shelter in three consecutive growing seasons. It was shown that there are metabolites which © 2015 American Chemical Society
significantly changed, either increased or decreased, due to drought stress for all investigated cultivars. The study also indicated differences for the factor growing conditions × cultivars; however, the data did not allow conclusions to be drawn regarding cultivar-specific responses. There are several comprehensive studies focusing on metabolite changes upon drought stress in general;7,12,13 however, metabolite profiling studies on cultivar-specific responses are rare. For example, in source tissues from Glycine max and Triticum L., the metabolite responses upon drought were observed to vary between cultivars differing in adaptation.11,14 Similarly, only a few drought-responsive metabolites exhibited changes in the plant tissue of droughtstressed model and forage legumes for all species.8 In sink organs, extensive metabolite investigations on the impact of water deficit in light of differences between cultivars regarding their adaptation to drought are also rare. For example, Semel et al. investigated drought effects in pericarp tissue of an F1 generation and its parental line from tomato plants by metabolite profiling.15 Similarly, Nam et al. investigated compositional changes of rice kernels in transgenic lines and their corresponding wild type subjected to drought by using a rain-out shelter.16 The authors found that stress affects levels of amino acids and fatty acids regardless of the genotype with different adaptation to drought stress. For maize, compositional Received: Revised: Accepted: Published: 4252
March 2, 2015 April 13, 2015 April 13, 2015 April 13, 2015 DOI: 10.1021/acs.jafc.5b01114 J. Agric. Food Chem. 2015, 63, 4252−4261
Article
Journal of Agricultural and Food Chemistry Table 1. Compilation of the Investigated Barley Cultivars cultivar
characteristicsa
origin
Argentinian Mutant 6519 Mackay
Argentina; chemically induced mutant of Quilmes Paine (1993) Australia; Cameo × Coru (2001)
Palmella Blue (CIho 3609) IPZ 24727
Ethiopian landrace Germany; Maresi × Omega
stress tolerance; ramularia tolerance; nonparasitic leaf spot tolerance; mildew susceptibility
Marnie Iron Barke Steina
mildew resistance; leaf rust resistance; net blotch resistance26 leaf rust resistance stress susceptibility; leaf disease resistances; mildew resistance leaf rust resistance
Streif Emperor
Germany; 6429f Havanna × (Prisma × Br 4714a) (2003) Germany; Marnie × LP 813.6.98 (2009) Germany; Libelle × Alexis (1996) Germany; (Sultan × St.434/62) × (Voldagsen × Carlsberg × Volla) (1980) Germany; Pasadena × Aspen (2007) Canada; HB320 × Maresi
Polygena
Germany; 46401-80 × 45465-78 (1994)
a
drought resistance; salt stress resistance drought tolerance;27 powdery mildew resistance; net blotch susceptibility; spot blotch susceptibility28 drought tolerance
mildew resistance; leaf rust resistance net blotch susceptibility; leaf rust moderate susceptibility; powdery mildew resistance leaf rust resistance
According to descriptions provided by the plant breeders or in the literature. 0.8 mg/mL phenyl-β-D-glucopyranoside in distilled water (fraction 3); 0.4 mg/mL p-chloro-L-phenylalanine in distilled water (fraction 4). The retention time mixture consisted of 1.5 mL of undecane (2 mg/ mL in hexane), 2.5 mL of hexadecane (1.5 mg/mL in hexane), 4 mL of tetracosane (1.5 mg/mL), and 4 mL of triacontane (1.5 mg/mL) dissolved in 15 mg of octatriacontane. Internal standards and retention time mixture standards were purchased from Fluka (Buchs, Switzerland). Sample Preparation. Barley plants were harvested and samples were prepared as previously described.12 Barley kernels from each biological replicate were ground with a cyclone mill (Cyclotec 1093, Foss Tecator, Rellingen, Germany) using a 500 μm sieve and immediately freeze-dried for 48 h. Samples were transferred to LDPE bottles (Kautex Textron, Bonn, Germany) and stored at −18 °C until analysis. Sample Workup. Freeze-dried barley samples were subjected to an extraction and fractionation scheme resulting in four fractions containing nonpolar (fractions 1 and 2) and polar constituents (fractions 3 and 4) as previously described in detail for maize.19 Flour (200 mg) was weighed into a cartridge, and 100 μL of methanol was added for cell disruption. After 20 min, methanol was removed by application of vacuum (30 mbar, 30 min), and nonpolar and polar compounds were extracted. One hundred microliters of standard solutions 1 and 2 as well as 300 μL of standard solutions 3 and 4 was added to the nonpolar and polar extracts, respectively. The nonpolar extract was transesterified with sodium methoxide and separated by solid phase extraction (SPE) into fractions 1 and 2 containing fatty acid methyl esters (FAME) and hydrocarbons as well as phytosterols, free fatty acids (FFA), and fatty alcohols, respectively. An aliquot of the polar extract was silylated and purified by selective hydrolysis to obtain fraction 3 containing sugar and sugar alcohols. Fraction 4 containing acids, amino acids, and amines was obtained by subjecting a second aliquot of the polar extract to silylation, oximation, and selective hydrolysis. For gas chromatography−mass spectrometry (GC-MS) analysis, 1 μL of fractions 1 and 2 as well as 0.5 μL of fractions 3 and 4, respectively, was injected. In addition, 1 μL of a 1:10 dilution of fraction 1 was injected to quantitate constituents present in high concentrations. The GC-MS conditions were as previously described.12 Metabolite Identification. Metabolites were identified by comparison of mass spectra and retention indices with those of reference compounds, with data from the NIST08 mass spectra library, the GOLM metabolome database, and the literature.20−22 Quantitation of Free Amino Acids. Quantitation of proteinogenic free amino acids was based on the equation
studies on grain from plants grown under water limitation were performed to compare different hybrids.9 Additionally, there are some studies that investigated the cultivar-specific impact of drought stress on metabolites via targeted analyses. For example, Thakur and Rai studied the alterations of levels of amino acids in maize shoots from differently drought-resistant cultivars.17 Singh et al. specifically analyzed the accumulation of proline in leaves of barley varieties with different adaptions to drought.18 The objective of the present study was to apply metabolite profiling to identify metabolites that might serve as markers for barley cultivars differently adapted to drought stress. Therefore, a set comprising both stress resistant/tolerant cultivars and cultivars, for which no drought resistance/tolerance had been described, were field-grown in three consecutive growing seasons under normal weather conditions as control and under induced drought stress using a rain-out shelter. The same barley cultivars were also grown under natural weather conditions in a second field trial at another location. Using metabolite profiling in combination with multivariate and univariate analyses, this setup should enable screening for cultivar-specific responses of barley grain to drought stress.
■
MATERIALS AND METHODS
Sample Material. Eleven two-row spring barley cultivars were field-grown under natural weather conditions at two control locations in Bavaria, Germany, and under drought stress induced by using a rainout shelter in the growing seasons 2010−2012; information on the cultivars is provided in Table 1. The first control location (Freising, C1) was situated next to the rain-out shelter, whereas the second control location (Frankendorf, C2) was 30 km distant. GPS coordinates of the field sites were as follows: C1 and rain-out shelter, latitude 48.3924, longitude 11.7569, elevation 448 m above normal null (NN); C2, latitude 48.3667, longitude 12, elevation 455 m above NN. Two biological replicates were taken from field trials with totally randomized field plot design and analyzed in triplicate (laboratory replicate samples). For further information regarding growing periods, soil quality, and weather conditions, see Wenzel et al.12 Metabolite Profiling. Preparation of Standard Solutions. The following internal standards were used for nonpolar (fractions 1 and 2) and polar (fractions 3 and 4) extracts of barley (see later for fractionation protocol): 0.5 mg/mL heptacosane in hexane (fraction 1); 0.6 mg/mL 5α-cholestan-3β-ol in dichloromethane (fraction 2); 4253
DOI: 10.1021/acs.jafc.5b01114 J. Agric. Food Chem. 2015, 63, 4252−4261
Article
Journal of Agricultural and Food Chemistry c(A) =
analyzed in triplicate and one quality control. In total, there were 66 batches (11 cultivars × 2 field replicates × 3 seasons). The GC-MS chromatograms were integrated by means of Xcalibur 2.1 (Thermo Fisher Scientific, Waltham, MA, USA). Standardization of peak heights and retention time matching were performed using Chrompare 1.1.25 For retention time matching characteristic compounds of the fractions 1−4 (fraction 1, myristic acid methyl ester, palmitic acid methyl ester, stearic acid methyl ester, arachidic acid methyl ester, and internal standard 1; fraction 2, silylated myristic acid, palmitic acid, stearic acid, internal standard 2, and β-sitosterol; fraction 3, silylated fructose, myo-inositol, internal standard 3, sucrose, and raffinose; fraction 4, silylated alanine, phosphoric acid, malic acid, internal standard 4, and citric acid) were used. Normalization of peak heights was done by using the internal standards 1−4. On the basis of noise levels of 1 and 2% peak heights relative to the internal standards in fractions 1/2 and 3/4, respectively, peaks below this threshold were discarded. For multivariate data analysis, principal component analysis (PCA) was performed using XLSTAT version 2013.3.02 (Addinsoft, Paris, France). Analysis of variance (ANOVA) was undertaken within GenStat 16 (VSN International Ltd., Hemel Hempstead, UK). The following described models were fit for each metabolite from polar fractions 3 and 4 across 11 cultivars, or for each cultivar, to evaluate the effect of growing across 11 cultivars or within each cultivar, respectively:
height(A) × c(IS) × Rf × 100 × 5 (mg/kg flour) height(IS) × W
where c is the concentration (μg/g), height the average absolute peak height of three repetitions, IS the internal standard, A the analyte, Rf the response factor, W the recovery rate, and 5 the dilution factor. A mixture containing 0.1 mg/mL alanine, glycine, tyrosine, leucine, lysine, glutamine, isoleucine, proline, valine, phenylalanine, and tryptophan, 0.5 mg/mL serine, threonine, and asparagine, 0.3 mg/ mL aspartic acid, and glutamic acid in MeOH/H2O (80:20, v/v) was prepared for validation. Response Factor. For calculating response factors (Rf), 1 mL of amino acid standard mixture was silylated and analyzed in triplicate. Response factors were calculated by using the equation
Rf =
height(IS) × c(A) height(A) × c(IS)
where Rf is the response factor, height the averaged absolute peak height of three repetitions, IS the internal standard, A the analyte, and c the concentration. Repeatability. Repeatability was determined by triplicate analysis of one barley sample. Relative standard deviations (RSD) of peak heights normalized to internal standard 4 were calculated for each amino acid. Recovery. Recovery rates of amino acids were determined by analyzing nine aliquots of a polar extract. The first three aliquots of the extract were spiked with 50 μL of amino acid standard mixture at the beginning of the analytical workup, whereas three further aliquots were spiked at the end of the procedure prior to GC-MS measurement. The last three aliquots were analyzed to calculate peak heights of the amino acids naturally occurring in the unspiked flour. Recovery rates were calculated by using the equation
vijkl = i ̀ + aî + sj + nk + tl + (sn)jk + (st )jl + (nt )kl + åijkl (i)
vijk = i ̀ + aî + sj + nk + (sn)jk + åijk
(ii)
vijkl (or vijk, respectively) is the response for the ith block of the jth growing conditions/location and the kth season (and the lth cultivar for vijkl), μ is the overall mean, βi is the random effect of the ith block, sj is the effect of the jth growing conditions/location, nk is the effect of the kth season, tl is the effect of the lth cultivar, (sn)jk is the effect of the interaction between the jth growing conditions/location and the kth season, (st)jl is the effect of the interaction between the jth growing conditions/location and the lth cultivar, (nt)kl is the effect of the interaction between the kth season and the lth cultivar, and eijkl (or eijk, respectively) is the random error. Differences for the effect of growing conditions/location were considered to be statistically significant if the effect was significant by ANOVA, and changes were consistent for (i) each cultivar and season or (ii) for each season. The effect of growing conditions/location was additionally statistically tested by Tukey’s test. The significance level was set to p < 0.01 for all statistical comparisons.
H − HP3 W = P1 × 100% HP2 − HP3 where W is the recovery rate, HP1 the average relative peak height from three aliquots of the extract spiked at the beginning of the analytical procedure, HP2 the average relative peak height from three aliquots of the extract spiked prior to GC-MS measurement, and HP3 the average relative peak height from three unspiked aliquots. Relative peak heights were normalized to the internal standard. Linearity. Linearity was determined for each amino acid by analyzing different amounts of the amino acids standard mixture (corresponding to 10−2500 mg amino acid/kg barley flour) after silylation in triplicate. Calibration curves were plotted, and the respective regression coefficients were calculated. Analysis of Protein. Analysis of protein contents was performed by near-infrared transmission spectroscopy (NIT) according to the method of Grausgruber and Vollmann and calculated as percent in grain.23 Amino acid spectra of free and protein-derived amino acids were analyzed in duplicate according to Commission Regulation (EC) No. 152/2009, III, F and G.24 Quality Control. To ensure constant analysis conditions, reference material (commercially available barley kernels) was analyzed in triplicate at the beginning and end of each sample set (2 field replicates × 3 seasons) and regularly together within each batch. Data were considered acceptable when the variation coefficient of representatives of each compound class did not exceed 25%. The representatives were as follows: FAME (C16:0, C18:0, C20:0), hydrocarbons (squalene), free fatty acids (C16:0, C18:0), sterols (campesterol, β-sitosterol, βsitostanol), monosaccharides (fructose, glucose, galactose), trisaccharides (raffinose), sugar alcohols (xylitol, myo-inositol), organic acids (citric acid, malic acid), and amines (γ-aminobutyric acid). Furthermore, data obtained by GC-MS analysis of the reference material were checked for additional peaks to control for potential laboratory contamination. Data Processing and Statistical Analysis. Freeze-dried flour from each biological replicate was analyzed in triplicate. For analysis, the samples were grouped into randomized batches; each batch consisted of three biological replicates (C1, C2, rain-out shelter)
■
RESULTS AND DISCUSSION To reveal changes in the metabolite profiles reflecting the adaptation of cultivars to drought stress, the set of barley grain previously investigated by Wenzel et al. was further extended.12 The material comprised two groups of spring barley assigned according to the descriptions provided by the plant breeders or in the literature.26−28 The first consisted of three cultivars (‘Argentinian Mutant 6519’, ‘Mackay’, ‘Palmella Blue’) from countries with hot and dry summers reported to be drought resistant/tolerant and of one cultivar (‘IPZ 24727’) from Germany described to be generally stress tolerant. The second consisted of seven cultivars originating from countries with moderate climates being either stress susceptible (‘Barke’) or without specific reports on drought resistance/tolerance (‘Marnie’, ‘Iron’, ‘Steina’, ‘Streif’, ‘Polygena’, ‘Emperor’) (Table 1). PCA based on agronomic parameters (e.g., yield, grain size, thousand kernel weight) confirmed the different performances of the two groups upon drought stress in the three growing seasons. 4254
DOI: 10.1021/acs.jafc.5b01114 J. Agric. Food Chem. 2015, 63, 4252−4261
Article
Journal of Agricultural and Food Chemistry
Figure 1. PCA of GC-MS metabolite profiling data of combined polar fractions 3 and 4 from stress resistant/tolerant barley cultivars (A−C, black symbols) and from cultivars reported to be either stress susceptible or without specific reports on drought resistance/tolerance (A−C, gray symbols) grown under drought stress in a rain-out shelter (squares) and under natural weather conditions at control location 1 (circles) and control location 2 (triangles) in (A) 2010, (B) 2011, and (C) 2012 and in (D) 2010 (black symbols), 2011 (white symbols), and 2012 (gray symbols). Circles encompass samples from the drought-stress conditions. Two field replicates were analyzed each in triplicate.
The 11 cultivars were field grown under induced drought stress using a rain-out shelter and under control, that is, natural weather conditions (C1 and C2), in three consecutive seasons. The GC/MS-based metabolite profiling was performed as previously described; the procedure results in four fractions containing FAME and hydrocarbons (fraction 1); FFA, sterols, and fatty alcohols (fraction 2); sugars and sugar alcohols (fraction 3); and amino acids, acids, and amines (fraction 4).12,19 A total of 214 peaks (90 polar and 124 lipophilic) were detected; 60 polar and 77 lipophilic metabolites were identified by comparison of retention indices and mass spectra to those of reference compounds or from MS libraries.12 Changes Due to Drought Stress Irrespective of Cultivars. PCA of the GC-MS metabolite profiling data obtained for fractions 1 and 2 from the extended set of cultivars grown under induced drought stress and control conditions, respectively, confirmed the results from the previous study: Differences in lipophilic metabolites were mainly due to seasonal impact rather than to drought stress.12 In contrast, PCA of polar fractions 3 and 4 demonstrated that in each season samples subjected to drought stress were clearly separated from those grown under control conditions (Figure 1). The PCA plots are in agreement with those previously obtained for the smaller set of cultivars.12 The best separation was observed in 2011 (Figure 1B). Again, the clustering in 2010
turned out to be influenced by a distinct aridity period in that season, and the clustering in 2012 also revealed separations according to growing location under the control conditions (Figure 1A,C). Importantly, the separation due to the factor growing conditions was still observed even when the data from all seasons were combined (Figure 1D). These multivariate analyses revealed that the polar fractions are the most prominent metabolites contributing to a differentiation of barley grown under control conditions and induced drought, respectively. Therefore, the polar constituents were additionally subjected to univariate statistical analysis on the basis of the mean values of all cultivars. On the basis of the employed cutoff threshold, 49 components were assessed via ANOVA and Tukey’s test. There were metabolites for which the responses were either significantly decreased or increased upon drought stress compared to C1 and C2, irrespective of the cultivar. They were the same 16 as those previously reported for the smaller set of cultivars;12 the 13 identified metabolites including, for example, the monosaccharides fructose and glucose, the trisaccharide raffinose, the sugar alcohol mannitol, several organic acids, the biogenic amine γ-aminobutyric acid (GABA), and the amino acid glutamic acid are listed in Table 2. They comprise many metabolites commonly reported to be involved in adjustment mechanisms to drought, such as 4255
DOI: 10.1021/acs.jafc.5b01114 J. Agric. Food Chem. 2015, 63, 4252−4261
Article
Journal of Agricultural and Food Chemistry
maize were in accordance with those found in the present study. For some of the constituents shown in Table 2, changes in concentrations upon stress have been reported to depend on the adaptation of the plant. For example, Sanchez et al. found species-specific differences for plant tissue of legumes with regard to their quantitative response to drought stress for fructose.8 Levels of tricarboxylic acid cycle intermediates increased in leaves of a tolerant barley cultivar due to salinity stress, but were unchanged in a sensitive cultivar.35 Analyzing drought-stressed rice grain, Nam et al. found increasing GABA, fructose, and glucose levels especially in the drought-tolerant genotypes.10 For barley grain such correlations between the drought stress-induced responses of these metabolites and the degree of adaptation of the different cultivars could not be found. Cultivar-Dependent Changes Due to Drought Stress. To reveal cultivar-specific changes in the responses of polar metabolites upon drought stress, statistical analysis (ANOVA, p < 0.01) of the polar metabolites was subsequently performed not on the basis of the mean values of all cultivars but for each individual cultivar. For 19 constituents the responses were not statistically significantly different; 30 metabolites including those from Table 2 showed significant results for at least one cultivar. For two of the constituents presented in Table 2, mannitol and succinic acid, significant decreases and increases, respectively, were found for each cultivar. For the remaining metabolites, there were individual cultivars that did not show significant changes. However, despite these cultivar-specific results, the data of the 13 metabolites from Table 2 did not allow a correlation to be established between the number of significant differences and the degree of drought stress adaptation. On the other hand, the components shown in Table 3 turned out to exhibit statistically significant differences corresponding to groups of cultivars. The stress-susceptible cultivar ‘Barke’ and the two cultivars ‘Marnie’ and ‘Iron’, for which no drought resistance/tolerance had been described, were characterized by significant changes, particularly for the branched-chain amino
Table 2. Relative Mean Responses of Identified Metabolites from 11 Cultivars Found To Be Statistically Significant for the Factor Growing Conditions/Location across Three Consecutive Seasons relative mean responsea metabolite
rain-out shelter
C1
C2
0.026 (a) 0.052 (a) 0.465 (a)
0.215 (b) 0.070 (b) 0.692 (b)
0.165 (b) 0.065 (b) 0.574 (ab)
0.674 0.775 3.826 1.000 0.330 4.378 0.176 0.070 0.358 0.152
0.386 0.417 2.562 0.681 0.117 3.361 0.077 0.031 0.190 0.102
0.353 0.428 2.279 0.626 0.106 3.396 0.066 0.021 0.160 0.086
decreasing responses mannitol 2-glycerophosphoric acid glutamic acid increasing responses glucose fructose raffinose citric acid isocitric acid malic acid succinic acid threonic acid GABA glycine
(a) (a) (a) (a) (a) (a) (a) (a) (a)b (a)
(b) (b) (b) (b) (b) (b) (b) (b) (b)b (b)
(b) (b) (b) (b) (b) (b) (b) (b) (b)b (b)c
a
Relative mean response = response (metabolite)/response (internal standard) × 100. bExcept for ‘Palmella Blue’. cRain-out shelter versus C2: not consistent across all growing seasons.
fructose, glucose, raffinose, mannitol, several organic acids, and the well-known stress marker GABA.4,5,12,29−34 The data demonstrate that responses to drought stress commonly known for plants and their vegetative tissues are not readily transferable to grain. For instance, levels of mannitol and glutamic acid decreased due to drought stress in barley grain, but have been described to be increased in vegetative tissues of several plants.4,30,32 These differences in responses of metabolites to drought stress depending on the investigated tissue are in line with the results reported by Harrigan et al. in grain of maize hybrids grown under different water regimens.9 Changes in the contents of raffinose, mannitol/sorbitol, and several organic acids observed in grain from drought-stressed
Table 3. p Values of Identified Metabolites from Each Cultivar Found To Be Statistically Significant for the Factor Growing Conditions/Location across Three Consecutive Seasons p value metabolite
Arg. Mutant 6519
proteinogenic free amino alanine valine isoleucine leucine glutamine asparagine serine threonine further metabolites pyroglutamic acid putrescine myo-inositol
Mackay
Palmella Blue
IPZ 24727
acids −a − − − − − − −
Marnie
Iron
− − − − − −