Principal component analysis
In this study we used 100 lines of a tomato recombinant inbred line (RIL) population derived from a cross between Solanum lycopersicum (cv. Moneymaker) (MM), and Solanum pimpinellifoliumaccession G1.1554 (PI) (Voorrips et al. 2000). The RILs, together with the parental lines, were grown in two different nutritional environments; high phosphate and low nitrate. The harvested seeds were used to measure the metabolite content in the dry mature seeds. The metabolites were measured by gas chromatography-time of flight-mass spectrometry (GC-TOF-MS) analysis in a specific GGG design. In total 118 primary metabolites were detected from which 58 could be identified. These identified metabolites were classified as amino acids, organic acids, sugars, sugar alcohols and some other compounds (Table S1 ).
Principal component analysis of the primary metabolites in the parental lines (MM and PI) indicated clear genetic effects as the two genotypes displayed different accumulation of metabolites in each maternal environment. In addition, maternal environmental effects were also observed within each genotype as metabolites accumulated differently in HP and LN (the two different maternal environments) (Figure 1 ).
Furthermore, analysis of significant changes in dry seed metabolites of the two species (MM and PI) between the different environments revealed that the metabolite contents were influenced to a higher extent in MM as compared to PI. In MM the level of 36 metabolites, mostly including amino acids and organic acids, had significantly changed between the environments, while for PI this number decreased to 26 from which 19 were common between the two species (Table S2 ). In general, PI as a wild tomato species is more tolerant to suboptimal environments (Kumar 2006; Rao et al. 2013; Rodríguez-López et al. 2011) which might be the reason why it does not need to modify its metabolites to a high extent to cope with a changing environment. While in the case of MM, which is considered as a domestic sensitive accession, many metabolites may have to be altered in order to allow it to deal with environmental stresses.
Metabolites are not only influenced by the genetic background of the seeds but also by the environment under which seeds develop and mature and the interaction between genotype and environment (G×E) (He et al. 2016). In our study, a wide range of the metabolites in the parental lines were also significantly influenced by the genotype, nutritional environment and their interaction (Table S3 ). In total 15% of the annotated metabolites did not significantly change in our analysis and 57, 60 and 53% of the metabolites were influenced by genotype, environment and G×E, respectively.