Correlation of metabolites within the whole RIL population
Spearman correlation analysis was performed and a correlation matrix was generated between all pairs of known metabolites across the whole RIL population. The results revealed that some metabolites are highly correlated with each other (Figure 6 ). Except a few exceptions, all amino acids cluster together. They showed a high degree of correlation of mostly greater than 0.5 with p -values of less than 0.001 (Table S7 ). Such a high positive significant correlation could be an indication of a preserved metabolism of amino acids in seeds. It has previously been reported that the metabolism of amino acids in seeds might be regulated by post-transcriptional regulators in order to regulate the distribution of nitrogen (Kazmi et al.2017; Toubiana et al. 2012). We also observed a high number of significant correlations between amino acids and TCA cycle intermediates such as citrate, malate, fumarate and succinate. Such a correlation between amino acids as a nitrogen (N) source and TCA intermediates as carbon (C) metabolites, indicates a maintained crosstalk between N and C metabolism in the seeds (Figure 6, Table S7 ). A similar crosstalk has been previously suggested for different species including Arabidopsis and tomato (GutiƩrrez et al. 2007; Kazmi et al. 2017; Nunes-Nesi, Fernie & Stitt 2010; Stitt & Fernie 2003).
We have also detected significant correlation between galactinol and myo-inositol (R=0.53 and p -value=1.06E-6) (Figure 6, Table S7 ). These metabolites are classified as sugar alcohols which have been reported to be involved in responses of seeds to stressful environments, such as low temperature (He et al. 2016).