Intraspecific genotypic variability determines concentrations of key truffle volatiles

Aroma variability in truffles has been attributed to maturation (Tuber borchii), linked to environmental factors (Tuber magnatum), but the involvement of genetic factors has been ignored. We investigated aroma variability in Tuber uncinatum, a species with wide distribution. Our aim was to assess aroma variability at different spatial scales (i.e. trees, countries) and to quantify how aroma was affected by genotype, fruiting body maturity, and geographical origin. A volatile fingerprinting method was used to analyze the aroma of 223 T. uncinatum fruiting bodies from seven European countries. Maturity was estimated from spore melanization. Genotypic fingerprinting was performed by amplified fragment length polymorphism (AFLP). Discriminant analysis revealed that, regardless of the geographical origin of the truffles, most of the aroma variability was caused by eight-carbon-containing volatiles (C8-VOCs). In an orchard of T. uncinatum, truffles producing different concentrations of C8-VOCs clustered around distinct host trees. This clustering was not associated with maturity, but was associated with fungal genotype. These results indicate that the variation in C8-VOCs in truffles is most likely under genetic control. They exemplify that understanding the factors behind aroma variability requires a holistic approach. Furthermore, they also raise new questions regarding the ecological role of 1-octen-3-ol in truffles.

Under equilibrium extraction with SPME, samples of 300 mg or 400 mg only presented a negligible variability in terms of peak intensity (shown here for 4 major volatiles, n = 3 replicates). This variability was comparable to the variability observed in between different gleba samples of the same truffle, but negligible compared to the variability observed among all the Tuber uncinatum samples analyzed in this study (i.e. 1-octen-3-ol TIC peak area ranged from zero to 1×10 9 ).

Fig. S5. Comparing SPME-GC/MS and OLS-GC/FID data sets to estimate the matrix effect
A total of 32 Tuber uncinatum samples (400 mg gleba) from Switzerland, Italy and France were analyzed by SPME-GC/MS (no internal standard) and OLS-GC/FID (with internal standard) for C8-VOC. 1-Octen-3-ol, 3-octanone and 3octanol could be detected with the OLS-GC/FID. (A) Comparing both methods highlights a good agreement (R 2 = 0.859) between the intensities of the C8-VOC marker [m/z 57; Rt415-523] (SPME data) and the quantification of 1-octen-3-ol (OLS data). (B) Comparing the SPME data to the sum of all C8-VOCs detected by with the OLS further improved the correlation between both methods (R 2 = 0.863). This indicates that the matrix effect is negligible compared to the variability observed in the T. uncinatum samples. (C) Of the three C8-VOCs detected by OLS-GC/FID, 1-octen-3-ol was by far the dominant one; 3-octanone and 3-octanol only represented a minor fraction of the quantity of 1-octen-3-ol (max. 10%, median <1%).
Values are expressed as peak area percent of the total ion chromatogram. 1-Octen-3-ol could be detected in all species except the white truffle Tuber magnatum. Note the important variability in 1-octen-3-ol within a single species.

Fig S7 Season variability of volatile markers within single sites.
The heatmaps above were generated from Table S1 by selecting volatile markers that presented significant seasonal differences for the sites in France (Auvergnes, n = 4 collection times) and Switzerland (Wallis, n = 6 collection times) (P < 0.01, Kruskal Wallis test

Fig S9 1-Octen-3-ol intra and interspecific variability in mycelia of Tuber borchii and T. melanosporum. (A)
Scatterplot of factor loadings (principal factor analysis) of the volatile profiles of mycelia of T. borchii (two strains) and T. melanopsorum (one strain) indicates that C8 producers can be distinguished from non producers. (B) Boxplot of the the major C8 volatile 1-octen-3-ol (4 replicates per strain) highlights differences in C8 levels between the two isolates of T. borchii, confirming the strain specificity of C8 biosynthesis.

Predicted nb of samples in each regions
Regions can be distinguished by stepwise discriminant analysis Discriminant analysis was performed on Table S2 using stepwise forward discriminant analysis (F to remove = 0.05; F to enter = 0.1). Calssification scores and the number of samples in each regions are listed in the table above.

Table S4 1-Octen-3-ol biosynthesis in Tuber borchii is not maturity dependent.
Zeppa et al (2004) reported 1-octen-3-ol to be a volatile marker only present in very ripe (stage III) T. borchii fruiting bodies (stages III = 71-100% mature; stage II = 31-70% mature; stage I = 6-30% mature; stage 0 = 0-5% mature). Out of 16 T. borchii samples analyzed in this study it is noteworthy that 1-octen-3-ol could also be detected at maturity stage II. Furthermore the same volatile was also missing in some fruiting bodies at maturity stage III. Taken together this suggests that maturity is not the major factor controlling 1-octen-3-ol biosynthesis.