Tuber melanosporum outcrosses: analysis of the genetic diversity within and among its natural populations under this new scenario


Author for correspondence:
Francesco Paolocci
Tel: +390755014861
Fax: +390755014869


  • • Tuber melanosporum is an ectomycorrhizal ascomycete producing edible ascocarps. The prevalent view is that this species strictly selfs, since genetic analyses have never detected heterozygotic profiles in its putatively diploid/dikaryotic gleba. The selfing model has also forged the experimental approaches to assess the population genetic variability. Here, the hypothesis that T. melanosporum outcrosses was tested.
  • • To this end, SSR (simple sequence repeats) and ITS (internal transcribed spacer) markers were employed to fingerprint asci and the surrounding gleba within single ascocarps. The distribution of genetic variability was also investigated at different geographical levels using single (SSR and ITS) and multilocus (AFLP, amplified fragment length polymorphism) markers.
  • • It is shown that T. melanosporum outcrosses since asci display additional alleles besides those present in the surrounding, uniparental, gleba. Furthermore, SSR and AFLP data reveal a high rate of intrapopulation diversity within samples from the same ground and root apparatus and the highest rate of genetic variability within the southernmost populations of the distributional range.
  • • These data call for a profound re-examination of T. melanosporum mating system, life cycle and strategies for managing man-made plantations. They also strongly support the idea that the last glaciation restricted the species distribution to the Italian and Spanish peninsulas.


The growth and survival of many forest trees and shrubs rely heavily on root colonization by ectomycorrhizal fungi that mediate nutrient and water uptake in exchange for photosynthetically derived carbon compounds. Most of the fungi involved in ectomycorrhizal symbioses are basidiomycetes and a few are ascomycetes, among these Tuber spp. Some species of this genus are also economically important since they produce high-value edible mushrooms, known as truffles, with unique organoleptic qualities. Edible European truffle species are either ubiquitous with a large genetic and morphological variability such as T. aestivum Vittad. and T. mesentericum Vittad., or patchy, confined to the southern European countries such as the most prestigious white and black truffle species, T. magnatum Pico and T. melanosporum Vittad., respectively. The former species is naturally found in Italy and some Balkan areas, the latter in Italy, Spain and France. Bertault et al. (1998, 2001) explained the limited distributional range of T. melanosporum and T. magnatum and their lower genetic diversity with respect to other truffle species as the result of a bottleneck suffered by these two species during the last ice age. Furthermore, the reproductive mode may have contributed to further exacerbating the low amount of genetic variability of these species. Although the difficulties of growing and the impossibility of mating these fungi under controlled experimental conditions have prevented researchers from gaining direct insight into the truffle life cycle and reproductive mode, it has, in fact, been inferred that Tuber are self-fertilizing fungi (Bertault et al., 1998, 2001; Murat et al., 2004; Mello et al., 2005). This is because codominant markers have never detected heterozygotic profiles in the putatively diploid/dikaryotic hyphae of fruiting bodies and mycorrhizas. More recent data, however, have challenged the hypothesis of selfing in truffles. In a departure from this reproductive model, simple sequence repeat (SSR) analysis in T. magnatum samples have substantiated the view that the gleba and mycorrhizas of this species are haploid structures while proving the presence of alleles of paternal origin among meiotic spores inside the asci (Paolocci et al., 2006; Rubini et al., 2007). These findings reinforced the hypothesis, suggested by a large-scale truffle screening with SSR markers, that outcrossing is the prevailing reproductive mode in T. magnatum.

In fact, from this survey, consistent gene flow within and among geographically close T. magnatum populations emerged (Rubini et al., 2005).

The genetic diversity in T. melanosporum is strikingly low (Gandeboeuf et al., 1997; Bertault et al., 1998). Nevertheless, a pattern of genetic differentiation among French T. melanosporum populations has been inferred by virtue of ITS/SNP (internal transcribed spacer/single nucleotide polymorphism) markers and of a sampling strategy based on the assumption that, because of selfing, a few individuals can reflect the regional populations better than extensive sampling per locality (Murat et al., 2004). However, conclusive evidence that T. melanosporum is a strictly selfing species and its fruit bodies are formed by diploid/dikaryotic hyphae has still to be provided. Clarifying the reproductive mode of the species is also a key biological issue in correctly modelling the pattern of distribution of genetic diversity within and among natural populations.

Therefore, we aimed to verify whether T. melanosporum is a strictly selfing species, and evaluate the amount and distribution of its genetic variation. To fulfill the first aim, SSR and ITS/SNP markers were employed to separately genotype the DNAs recovered from asci and from the surrounding gleba within a number of selected truffles. To study the degree and distribution of genetic variability, SSR, ITS/SNP as well as the multilocus AFLP (amplified fragment length polymorphism) markers were used to fingerprint samples collected both on the local scale – that is, fruiting bodies from the same root apparatus and from single natural truffle grounds – and over the species’ distributional range.

To meet the increasing worldwide demand for truffles and counterbalance the drop in natural production (Hall et al., 2003), man-made T. melanosporum plantations are also being set up in countries where this species is not endemic. Thus, understanding the reproductive mode and the distributional pattern of genetic variability in T. melanosporum natural populations will have an impact on both basic and applied issues.

Materials and methods

Sample sources and DNA isolation

Two hundred and ten Tuber melanosporum ascocarps were sampled in the years 2000–2006 from natural truffle grounds located throughout the species’ distributional range. To study the pattern of genetic diversity, the entire sample set was grouped into 13 populations according to geographical criteria (Table 1 and Supporting Information, Table S1). Among the entire truffle collection, seven ascocarps (279–282, 284, 285, 287) were collected from the roots of a single oak tree (Quercus pubescens) located in a natural truffle ground in S. Maria di Reggiano, Umbria (central Italy); 12 (192–198, 377–381) and seven samples (149–153, 155 and 156), respectively, were collected from two confined truffle grounds in Capodacqua di Foligno and Cerreto di Spoleto, both located in Umbria (Table S1). All the ascocarps were first macro- and microscopically checked for species determination according to Montecchi & Sarasini (2000), then cut into slices, frozen in liquid nitrogen and stored at –70°C before DNA extraction.

Table 1.  Analyzed Tuber melanosporum populations and diversity parameters
PopulationProvenanceITS and SSR sample sizeITS haplotypesSSR haplotypesSSR allelic richnessAFLP sample sizeAFLP haplotypesHS
  1. ITS, internal transcribed spacer; SSR, simple sequence repeats; HS: Shannon's diversity index.

 1M. Subasio (Italy)16121.13 (0.08)16120.85
 2Spoleto (Italy)53271.31 (0.04)32290.96
 3Abruzzo (Italy)32261.48 (0.08)16161
 4Ascoli (Italy)12221.15 (0.11) 6 50.87
 5Lazio (Italy) 5121.17 (0.18) 4 30.75
 6Piemonte (Italy) 6321.17 (0.17) 6 50.87
 7Yonne (France) 6121.17 (0.17) 6 50.87
 8Dordogne (France)11221.16 (0.11)10 90.94
 9Gap (France) 8221.15 (0.13) 9 70.86
10Provence (France)18241.21 (0.08)10 80.88
11Castellon (Spain)13171.64 (0.10)11 90.87
12Navarra (Spain)15291.97 (0.11)13131
13Sardinia (Italy)11161.49 (0.16) 4 41

Total genomic DNA was isolated from approx. 0.3 g of gleba from each ascocarp according to Paolocci et al. (1999). In addition, pools of asci, free of hyphal fragments, were recovered from 58 T. melanosporum ascocarps (Table 4) and DNA was isolated from these pools using the FastPrep apparatus (Q-BIOgene, Irvine, CA, USA) according to Paolocci et al. (2006).

Table 4.  Simple sequence repeat (SSR) and internal transcribed spacer (ITS) patterns of gleba and asci from single Tuber melanosporum ascocarps
SamplePopulationSSR LocusITS Nucleotide position (bp)ITS haplotype
  1. DNA source: g, gleba; a, asci. −, missing data; m, maternal haplotype; p, paternal haplotype.The SSR allele size is expressed in base pairs; polymorphic patterns are reported in bold.


Isolation and analysis of SSR loci

Tuber melanosporum SSR loci were isolated following the enriched genomic library and the inter simple sequence repeat (ISSR) methods reported in Rubini et al. (2004). The enrichment procedure was performed using the motifs (CA)n and (GA)n as baits on DNA isolated from truffle 79 (population 12, Table S1). The ISSR fingerprints were obtained using either the single primers BDB(CA)7 and CAA(GA)5 or the primer pair HVH(TG)7/(AG)8T on samples 366, 368 (population 12) and 378 (population 1). Amplifications were performed as reported in Rubini et al. (2004). The annealing temperature was 50°C (45°C for primer CAA(GA)5 only). The resulting fragments were ligated into pGEM T-easy vector (Promega, Madison, WI, USA) and cloned into competent E. coli JM83 cells. Plasmid DNA from positive clones was isolated using the JetQuick plasmid purification kit (Genomed, Löhne, Germany) and sequenced using the BigDye Terminator V 3.1 cycle sequencing kit and an ABI Prism 310 genetic analyzer (Applied Biosystems, Foster City, CA, USA).

For all fragments containing SSR, locus-specific primers were designed and preliminarily tested under the PCR conditions and cycling parameters reported in Rubini et al. (2004). The primer sets producing single amplicons were further used to PCR-amplify the respective loci on a large set of T. melanosporum samples. The following PCR cycling conditions were used: a denaturation step at 94°C for 2 min and 30 s, 35 cycles consisting of 30 s at 94°C, 30 s at the respective annealing temperature (Table 2) and 30 s at 72°C, followed by a final extension at 72°C for 20 min. Either forward or reverse primers were 5′-labeled with a fluorescent dye (6-FAM, VIC, NED or PET) and the resulting amplicons were analyzed by capillary electrophoresis in an ABI 3130 Genetic Analyzer with a Genescan 500 LIZ size standard (Applied Biosystems). Sizing of amplicons was performed using GeneMapper software version 3.7 (Applied Biosystems).

Table 2.  Characteristics of simple sequence repeats (SSR) loci isolated from Tuber melanosporum
Locus nameGenBankRepeat motifAmplification patternPrimer namePrimer sequence (5′–3′)Ta (°C)Allele size range (bp)NA
  • a

    Loci analyzed in all 206 individuals.

  • b

    Loci from Bertault et al. (2001).

  • c

    Irregular microsatellite, the consensus motif is reported.

  • d

    Microsatellite interruption by a nonrepeated region.

  • S, single band; m, multiple bands. Ta, temperature of annealing. NA, number of alleles.

F12Ia,b (GTTA)8s   215–2234
H1ba,b (GAGT)14s   90–1063

All 11 SSR loci isolated and characterized by Bertault et al. (2001) were also analyzed using the primer pairs and the PCR procedures reported by the authors.

PCR amplification and sequence analysis of the ITS

The PCR amplification of ITS was performed using the universal ITS1-ITS4 primer pair (White et al., 1990), as previously described (Paolocci et al., 1997). The ITS amplicons were either purified using a JetQuick PCR purification kit (Genomed) and directly sequenced or ligated into pGEM T-easy vector (Promega), cloned into competent E. coli JM83 cells and then sequenced. Sequences were obtained using the primers ITS1, ITS4 and 5.8sf, 5.8sb (Rubini et al., 1998). ITS sequences were aligned and analyzed for the presence of mutations using Vector NTI software version 10.3 (Invitrogen, Carlsbad, CA, USA).

The ITS sequences were deposited in GenBank under the following accession numbers: EU200420-EU200422 (samples 84, 281 and 121 corresponding to the haplotypes XI, XII, XIII); EU200433, EU200434, EU200410 (samples 178, 199 and 298); EU555383-EU555385 (cloned ITS from sample 178); EU555386 (cloned ITS from sample 199); EU555387 (cloned ITS from sample 298); EU555388-EU555391 (cloned ITS from sample 437); EU555392 (cloned ITS from sample 282).

AFLP analyses

Amplified fragment length polymorphism analyses were carried out on a subsample of 143 ascocarps, selected on the basis of tissue integrity and availability to ensure the isolation of high-quality DNA for two or more restriction reactions, using the AFLP Plant Mapping Kit, Small Genomes (Applied Biosystems) according to the manufacturer's protocol. Selective amplifications were performed using the MseI primer with two 3′ selective nucleotides (CT), combined with four different EcoRI primers having TA, TG, AA or TC as 3′ selective nucleotides. EcoRI primers were 5′-labeled with a fluorescent dye (6-FAM or JOE) for high-throughput analysis using the ABI Prism 310 Genetic Analyzer. Gene Scan 500 ROX was used as internal size standard. The GeneScan, version 2.1, and Genotyper, version 2.5, programs (Applied Biosystems) were used to analyze AFLP data. Polymorphic AFLP bands in the range 50–500 bp were visually scored as present (1) or absent (0) across the samples regardless of the peak height and compiled in a data matrix. Polymorphisms based on the presence or absence of a fragment in only one out of the 143 samples were not considered. On 20 randomly selected specimens, the entire AFLP analysis, from DNA isolation down to capillary electrophoresis of the products resulting from the selective PCR step, was repeated, at least once, to confirm the reproducibility of the AFLP patterns generated.

Data analyses

The genetic profile exhibited by each individual on the basis of SSR, ITS or AFLP markers was treated as a haplotype on the assumption that T. melanosporum is a haploid fungus (see the Discussion section).

For SSR, ITS and AFLP markers, the number of haplotypes detected in the entire sample collection and at the population level was calculated using the Arlequin software, version 3.1. (Excoffier et al., 2005). SSR allelic richness was estimated using FSTAT software version 2.9 (Goudet, 1995) according to the rarefaction method (Petit et al., 1998). Two-locus linkage disequilibrium (LD) on SSR loci was performed using Genepop, version 3.2a, across all populations. Within single populations, the AFLP haplotype diversity was calculated by a normalized Shannon's diversity index HS = −ΣPiln Pi/ln N, where Pi is the frequency of the ith haplotype and N is the sample size (Goodwin et al., 1992).

For the combined analyses of all the markers used, the AFLP data matrix was integrated with the ITS and SSR data, scored using a binary code.

The analysis of molecular variance (AMOVA) (Excoffier et al., 1992) and FST calculation were performed with Arlequin, version 3.1. (Excoffier et al., 2005).

A pairwise genetic distance matrix among individuals was generated for haploid binary data using GenAlEx software, v.6.1 (Peakall & Smouse, 2006). The same software was used to perform the principal coordinate analysis (PCO).

A FST genetic distance matrix among populations was obtained with the software AflpSurv 1.0 (Vekemans, 2002). Cluster analysis was performed with the neighbor-joining method using Phylip, v.3.66 (Felsenstein, 2005).

Spatial analysis of molecular variance (SAMOVA) was performed using a simulated annealing procedure described by Dupanloup et al. (2002) and using SAMOVA, version 1.0 ( The program was run seven times with a number of population groups (K) ranging from two to eight. We used 100 simulated annealing steps and a distance based on pairwise difference between haplotypes. For each K, the configuration with the highest FCT values, after the 100 independent simulated annealing processes, was retained as the best grouping of populations.

The existence of genetic structure was also tested with a Bayesian approach using the program STRUCTURE, v.2.2 (Pritchard et al. 2000). The program was used to estimate the likelihood of a given number of clusters (K). Five independent runs of K = 1–10 were performed at 200.000 MCMC (Markov Chain Monte Carlo) repetitions and 200.000 burn-in period using no prior information and assuming correlated allele frequencies and admixture.

The pattern of isolation by distance was evaluated according to Rousset (1997). A Mantel test with 1000 random permutations was performed with the matrix of pairwise FST distances among populations and a matrix of the ln(geographic distance) using the software GenAlEx, v.6.1 (Peakall & Smouse, 2006).


Isolation and characterization of polymorphic microsatellite loci

Using the microsatellite-enrichment procedure previously described (Rubini et al., 2004), 12 SSR-containing clones, called ME1-06ME39, were characterized (Table 2). Eight additional SSR loci, named 07ISSR1-07ISSR14, were isolated following the ISSR-based approach. None of the sequences flanking the SSR loci showed similarity with GenBank sequences. The primer pairs specific to the 20 selected SSR loci, their optimized annealing temperature along with the SSR core motif, the allelic size range and the number of alleles per locus are given in Table 2.

Following a preliminary PCR screening performed on 30 ascocarps of different geographic origin, only five (ME2, ME4, 07ISSR9, 07ISSR10 and 07ISSR14) out of the 20 selected loci were retained. The other loci were excluded from the analyses, since they either produced a multiple banding pattern or were monomorphic (Table 2). Furthermore, out of the 11 SSR-containing loci described in Bertault et al. (2001), only two (F12I and H1b) displayed polymorphism among individuals. To summarize, the following seven SSR loci were selected to analyze the entire T. melanosporum sample collection: ME2, ME4, 07ISSR9, 07ISSR10, 07ISSR14, F12I and H1b.

When all truffles were considered, only the loci 07ISSR9 and 07ISSR10 resulted in significant LD (P < 0.002381; Bonferroni-adjusted significance level). Therefore, to avoid any bias, the locus 07ISSR10 was not considered in the statistical analysis.

Patterns of genetic diversity across the T. melanosporum distributional range

SSR polymorphism  The number of alleles per locus ranged from two to four when scored on the 206 ascocarps that yielded positive SSR amplification (Table 2, Fig. 1). Remarkably, the different alleles per locus generally were not distributed evenly within and among populations, with one allele appearing more frequently than the others. More specifically, most of the rare alleles were present exclusively or with the highest frequency in the southernmost populations. In particular, the alleles 244 at locus ME2, 208 at 07ISSR14, and 94 at H1b were only found in population 12 (Navarra, Spain), whereas the other Spanish population (11) was the only one displaying the allele 106 at locus H1b. The alleles 217 and 223 at F12I were only found in populations 3 (Abruzzo, Italy) and 2 (Spoleto, Italy), respectively. The allele 215 at locus F12I was found in all individuals of population 5 (Lazio, Italy) and, although at low frequency, in populations 2, 11, 12 and 13 (Sardinia, Italy) only. Allele 247 at locus 07ISSR9 and allele 264 at locus 07ISSR10 were particularly abundant (frequency 0.72 and 0.54) in population 13. They were also found in populations 2, 3, 10 (Provence, France), 11, 12 and allele 247 also in population 9 (Gap, France), but with a frequency < 0.4.

Figure 1.

Distribution of allelic frequencies of the seven Tuber melanosporum simple sequence repeat (SSR) loci analyzed in the entire sample collection. Each figure section represents a locus, named as in Table 2. Squares indicate the color and amplicon size (bp) of each allele. Numbers within the maps refer to populations listed in Table 1. Diameters of the circles are proportional to the sample sizes.

Altogether, the seven polymorphic SSR loci identified 22 haplotypes within the truffle collection analyzed (Fig. 2a). Haplotype 11 emerged as the most common (0.64), while the frequencies of the others were much lower, ranging from 0.005 to 0.07 (data not shown).

Figure 2.

Distribution of simple sequence repeats (SSR) (a) and internal transcribed spacer (ITS) (b) haplotypes across the Tuber melanosporum populations analyzed. Squares indicate color and number of each haplotype. Numbers within the maps refer to populations listed in Table 1. Diameters of the circles are proportional to the sample sizes.

The number of haplotypes per population ranged between two and nine (Table 1). It is worth mentioning that the highest SSR-haplotype/sample-size ratios were not specific to populations represented by the highest number of individuals. Furthermore, the highest allelic richness (Table 1) was found in the three southernmost populations (11 and 12 from Spain and 13 from Sardinia) along with the southern population (3) from Abruzzo, Italy (Table 1). Based on AMOVA, most of the genetic variation (78%) was found within populations, whereas a significant percentage (22%) was found among populations (FST = 0.215, P < 0.01).

ITS polymorphism  Successful ITS PCR amplification was obtained for 205 T. melanosporum samples. With the exception of samples 178, 199 and 298, the direct sequencing of the ITS amplicons produced electropherograms with no ambiguities. However, when the amplicons were cloned and several randomly picked clones were sequenced, single point mutations among the repeated units occasionally emerged (see later). For this reason, the ITS sequence obtained by the direct sequencing was hereafter considered as a consensus sequence. The consensus sequences exhibited several SNPs among individuals. The most frequent SNPs were detected at positions 7 (T/G), 369 (C/T) and 535 (G/C), corresponding to the haplotypes I, II and III described in Murat et al. (2004) (Table 3). While none of the remaining seven, rare haplotypes detected by the same authors were observed in our sample set, the present ITS survey permitted the identification of three new ITS haplotypes named XI, XII and XIII, whose point mutations with respect to haplotype I are given in Table 3.

Table 3.  rDNA internal transcribed spacer (ITS) haplotypes of Tuber melanosporum and single nucleotide polymorphism (SNP) positions
HaplotypesNucleotide position
  • Haplotype I was used as the reference sequence. Dots indicate identical nucleotides.

  • a

    Haplotypes described by Murat et al. (2004).

  • b

    Gleba samples showing sequence ambiguity.


Concerning sample 178, repeated DNA isolation and amplification steps followed by the direct ITS sequencing showed the occurrence of G/T (K) and T/C (Y) ambiguities at positions 7 and 369, respectively, with a marked prevalence of G at position 7 and T at position 369. These ambiguities were congruent with the presence, among the rDNA repeated units of the gleba, of units harboring either haplotype II or III, with the former being the most abundant. The sequencing of ITS clones from sample 178 confirmed this hypothesis, in that out of the 32 clones, 23 displayed haplotype II, six haplotype III and each of the remaining three clones a single random point mutation. Direct sequence analyses of ITS from samples 199 and 298 showed a C/T (Y) ambiguity at position 496 and 252, respectively. The sequencing of independent clones confirmed that along with ITS units harboring the haplotype I, the samples 199 and 298 also harbored units showing T at position 496 and 252, respectively. Owing to the ambiguous ITS genotyping, samples 178, 199 and 298 were excluded from the analysis of ITS haplotype frequencies.

The distribution of ITS haplotypes across the T. melanosporum populations is reported in Fig. 2b. Haplotype I was the most frequent in all populations, with the exception of population 6 (Piemonte) characterized by a high frequency (0.5) of haplotype III. Haplotype II was present in most of the Italian and French populations, with the exception of populations 1, 5, 13 and 7, but in none of the 28 individuals from Spain. Haplotypes XI, XII and XIII were rare and only recorded in Spanish, Italian and French populations, respectively. Genetic differentiation among populations was detected by AMOVA (FST = 0.12, P = 0.001).

AFLP polymorphism  The ascocarps were genotyped using four AFLP primer combinations. For each primer pair, 75–97 fragments in the 50–450 bp range were scored. Fifty-one (15%) fragments were polymorphic across all the ascocarps (Table S1), with the primer combination CT/AA showing the highest ratio of polymorphic bands. Overall, the AFLP fingerprint identified 110 genetic patterns (hereafter AFLP haplotypes), with the number of haplotypes per population ranging from three to 29 (Table 1). Remarkably, Shannon's diversity index (HS) indicated a high degree of diversity in all populations, with values ranging from 0.75 to 1 across the 13 regional populations, irrespective of their sample sizes (Table 1). Additionally, only eight of the 110 haplotypes were shared by two or more populations. AMOVA further confirmed this pattern: 85% of the total genetic variation was found within populations, whereas the remaining 15% was attributable to differences among populations (FST = 0.153, P < 0.01).

Analysis of population structure

Analysis of population structure was performed on a combined dataset consisting of 137 individuals for which data relative to all the markers used were available (Table S1). A first inference was made at the individual level. PCO analysis showed the tendency of the Spanish individuals to separate from French samples (Fig. 3a). At the same time, however, neither the Spanish nor the French individuals were clearly differentiated from the Italian ones. Interestingly, the same analysis revealed a genetic differentiation of few samples collected in the confined truffle ground of Cerreto di Spoleto (population 2) from all the others. According to AMOVA, most of the genetic variation (82%) resided within populations, while a significant percentage (18%) was found among populations (FST = 0.177, P < 0.0001). Although supported by a low bootstrap value, the dendrogram based on FST distances among populations revealed a clustering of the southernmost populations (5, 12, 11, 13, 3, 4) (Fig. 3b). However, the analysis performed with SAMOVA resulted in a continuous decrease of FCT value for K, ranging from 2 to 8 (Fig. S1a) and failed to detect population clustering. On the basis of STRUCTURE analysis, the inferred number of clusters (K) showing the highest value of ln P(D) (−1789.06 ± 9.45) was 7 (Fig. S1b). Despite the fact that FST indicated the existence of genetic differentiation among populations, neither SAMOVA nor STRUCTURE analysis revealed a clear geographic distribution of the genetic variability. Finally, no isolation by distance pattern was detected using Mantel test (R2 = 0.015, P = 0.250).

Figure 3.

Distance analyses at individual (a) and population (b) levels based on simple sequence repeats (SSR), internal transcribed spacer (ITS) and AFLP datasets of Tuber melanosporum. (a) Two-dimensional plot of the principal coordinate analysis. Most of the Spanish and French samples are grouped within the circles. (b) Neighbor-joining tree of T. melanosporum populations. Numbers near the branches are the bootstrap values (%) resulting from 1000 replicates.

Analyses performed on the datasets relative to each molecular marker (AFLP, ITS and SSR) did not provide any additional information about the genetic structure of populations (data not shown).

Patterns of T. melanosporum genetic diversity within single truffle fields and root apparatus

For a closer look at the genetic diversity of truffles on a local scale, 11 and seven samples harvested during the collecting year 2000 from two natural truffle fields located in Umbria at Capodacqua and Cerreto di Spoleto, respectively, were genotyped using SSR, ITS/SNP and AFLP markers. According to the ITS, all 11 individuals from Capodacqua showed haplotype I, whereas two different SSR and nine AFLP haplotypes were found.

The seven samples from Cerreto di Spoleto shared haplotype I, but a single truffle differed from the others for the SSR pattern, whereas none of them showed the same AFLP profile (Table S1).

The same markers were also employed to analyze seven fruit bodies harvested during the collecting year, 2001, from roots of a single Quercus pubescens plant located in S. Maria di Reggiano. The combination of the ITS and SSR markers identified three haplotypes among the seven individuals. Furthermore, none of these individuals showed the same AFLP haplotype; rather, they were differentiated from each other by a number of polymorphic bands ranging from 1 to 15 (Table S1).

Inferring the T. melanosporum reproductive system using SSR and ITS markers and differential screening of gleba and asci within each ascocarp

The absence, on the entire sampling scale, of significant SSR linkage disequilibrium and the evidence that, according to AFLP markers, a high number of genotypes were also detected within confined truffle grounds suggest that T. melanosporum is not highly inbred. To test this hypothesis, 58 ascocarps randomly selected within five polymorphic populations according to SSR and/or ITS/SNP markers were treated as reported in Paolocci et al. (2006) to purify the asci from the surrounding gleba and separately recover the DNA contributed by these two structures. For each ascocarp, equal amounts of DNA from the gleba and from the purified asci were then amplified using the primer pairs specific to one or more polymorphic SSR and/or the ITS1/ITS4 primer pair. Repeated experiments of DNA isolation and amplification from different portions of the gleba from the 58 selected ascocarps did not reveal sequence ambiguities in the ITS region or the presence of more than one peak (allele) for each of the SSR loci analyzed. The pools of asci from 58 ascocarps were analyzed with SSR markers, and seven showed two peaks (alleles) at at least one locus (Table 4). More specifically, the asci from ascocarp 440 showed two alleles at loci ME4 and 07ISSR10, those from 368 showed two alleles at locus ME2, and those from samples 295, 299, 364, 366 and 80 showed two alleles at locus 07ISSR14. In all these truffles, one of the two peaks displayed by the asci was always recorded in the corresponding gleba and the other found in the reference population. The asci from the remaining 30 ascocarps always presented a single allele of identical size to that in the corresponding gleba, at all SSR loci.

Furthermore, the DNA from the gleba and asci from 44 ascocarps was amplified using the ITS-specific primers and the resulting amplicons were directly sequenced. Sequence ambiguities were never detected in the gleba, whereas eight of the 44 pools of asci showed ambiguities (Table 4). Notably, one of the two alternative nucleotides at each position was shown by the corresponding gleba, whereas the other was always present in at least one truffle within the same population. More specifically, the pools of asci from samples 437, 438, 440, 475, 479, 179 and 181 showed three ambiguities at positions 7 (G/T), 369 (T/C) and 535 (C/G). These results are congruent with a fertilization event occurring between strains carrying different haplotypes, such as between strains with haplotype I and II. The asci from sample 282, a specimen collected in a truffle ground located at S. Maria di Reggiano, showed a single (G/C) ambiguity at position 66 (Table 4), whereas its gleba showed haplotype I. This is consistent with a mating between a maternal strain displaying haplotype I and a paternal one with haplotype XII (Table 3). Haplotype XII was shown as maternal strain in truffles 281, 284 and 287, from the same collection site.

The amplicons yielded by the asci from samples 282 and 437 were cloned and multiple clones per sample were sequenced. Concerning sample 437, nine of the 21 randomly peaked clones displayed haplotype I and eight displayed haplotype II. Out of the 20 clones sequenced from sample 282, 10 showed haplotype I and eight haplotype XII. Each of the remaining clones from both samples displayed randomly distributed point mutations with respect to the consensus sequences. Thus, on asci of both 282 and 437 samples, the two putative parental haplotypes were the most abundant among the rDNA units.

All in all, either the SSR or ITS/SNP markers, or both, such as in sample 440, allowed us to detect alleles of paternal origin in the asci of 24% of the fruit bodies analyzed.


In the present study we provide evidence that T. melanosporum is an outcrossing rather than a strictly selfing species, as has been believed up to now. In fact, molecular data based on SSR and ITS/SNP markers revealed the presence of alleles of paternal origin in the asci of fruit bodies whose gleba is made up of homokaryotic, uniparental (maternal) hyphae. Concomitantly, AMOVA analyses on SSR, ITS and AFLP data from samples collected throughout the species’ distributional range converged to show that most of the genetic variation was found within populations, while a significant FST value for all the markers suggests a geographic differentiation among populations. Finally, the finding that the southernmost populations showed the highest values of allelic richness provides the first genetic evidence in support of the hypothesis that potential T. melanosporum refugia were located in the Italian, and probably Spanish, peninsulas and that, after the glaciation, the species range expanded northward (Bertault et al., 1998, Murat et al., 2004).

Tuber melanosporum mating system

The strikingly low genetic variability displayed by T. melanosporum has been attributed to the concerted effects of a population bottleneck suffered by this and other Tuber species during the last glaciation and the strictly selfing reproductive mode of this species. To overcome the difficulties of completing the truffle life cycle under controlled conditions and test whether or not T. melanosporum is a selfing species, in this study we employed a strategy based on the comparison of the molecular profiles exhibited by asci and those shown by the surrounding gleba within each given truffle fruit body. A similar approach based on SSR markers has recently been pursued to prove that T. magnatum outcrosses (Paolocci et al., 2006). In this study we took advantage of nuclear SSR and ITS/SNP markers, either identified previously or characterized in the present study, to compare the genetic patterns of gleba and asci in 58 T. melanosporum fruit bodies. In 24% of the fruit bodies analyzed, the asci displayed either SSR or ITS alleles, or both, in addition to those observed in the corresponding gleba, which in turn always displayed a single allele per locus. The presence of the additional alleles in the asci with respect to the gleba suggests that fertilization occurred between two genetically polymorphic strains and that the hyphae of the gleba are uniparental. The dual screening of asci/gleba can allow us to infer the ITS/SSR haplotype of each mating partner as well. As an example, the observation that the asci from sample 437 displayed the ITS sequence ambiguities G/T, T/C and C/G at positions 7, 369 and 535, respectively, whereas the ITS units of their respective gleba showed haplotype II makes it conceivable that the crossing partner harbored the ITS haplotype I. The presence of units showing the paternal haplotype was confirmed by sequencing the ITS clones amplified from the asci of sample 437.

In our model, each T. melanosporum fruit body results from a mating event between two haploid mycelia and the fertilization step precedes and promotes the maternally driven development of the gleba. The data reported in the present study showed that, while outcrossing takes place, the paternal contribution is only detectable in the asci, or better in the spores within them, when specific procedures to break spores and isolate their DNA are followed. On the contrary, when non-ascus-enriched portions of fruit bodies are sampled, regardless of their position in the ascocarp and the ascocarp development stage, the resulting DNA is mostly of uniparental origin. Thus, since the paternal partner does not participate in building up the gleba, which represents the most abundant tissue of the ascocarp, it must contribute its genome at very early stages of truffle development and in a spatially-temporally very narrow window. This reproductive model also characterizes most of the ascomycetes, including T. magnatum (Rubini et al., 2007), and substantially differs from the most accredited selfing model that has guided all the population biology studies in T. melanosporum to date. Thus, gleba is not a diploid structure, as supposed by Bertault et al. (2001). In turn, it is not the selfing but the prevalence of haploid tissue in T. melanosporum fruit bodies that explains the lack of heterozygous individuals detected by Bertault et al. (1998, 2001) and Murat et al. (2004) in their SSR- and ITS/SNP-based truffle screenings, respectively. With this reproductive model in mind, we can challenge the view of self-perpetrating, a sort of clonal propagation by sex, mycelial strains in the ground. The condition of sharing the same genetic profile in the gleba can, in fact, no longer suffice to signify that ascocarps from the same stand are, and will give rise to, clones.

Having shown that the gleba of T. melanosporum truffles is formed by uniparental-haploid hyphae and that this species outcrosses, we expected to observe a high number of haplotypes per population as the result of recombination. This expectation was fully confirmed by our AFLP and SSR data, which showed a high degree of haplotype diversity within regional populations and, according to AFLP data, within truffles from the same stands and from the same root apparatus. Concerning ITS, this marker turned out to be less informative than SSR and AFLP in detecting polymorphism within populations. Additionally, the direct ITS sequencing showed that with the exception of samples 178, 199 and 298, which exhibited two prevalent ITS haplotypes, all the other truffles displayed only one prevalent ITS haplotype in the gleba. The extremely low rate of crossovers between the rDNA loci of two parental strains harboring different ITS haplotypes has been largely documented in yeast, fruit fly and humans, and finds its most plausible explanation with a meiotic mechanism that, at the rDNA loci, privileges a between-sister-chromatids rather than an interchromosomal recombination (for a recent review, see Eickbush & Eickbush, 2007).

Outcrossing rates within truffle grounds

Whether or not T. melanosporum is an obligate randomly mating species remains to be addressed. The ongoing genome sequencing effort will help us to disclose the organization of the mating genes in T. melanosporum in the near future. At present, additional alleles in the asci with respect to those shown by the gleba were not detected in all the truffles examined. This could be because of the relatively low number of polymorphic single locus markers so far isolated in this species. Furthermore, the exiguous amount of DNA recoverable from the pool of spores prevents us from making a side-by-side comparison between the DNA patterns of the gleba and those from the corresponding asci using multilocus informative markers such as AFLPs.

In this study, we show that, moving from single-locus to multilocus markers, it becomes more and more feasible to bring to light differences among truffles from the same regional populations, the same truffle stands and even from the same root apparatus. While SSR loci proved to be more informative, in terms of the number of haplotypes per population, than the ITS locus in most of the 13 regional populations (9/13), the AFLPs always revealed the highest amount of genetic polymorphism at any geographical scale (Table 1). As an example, while the combination of the ITS and SSR markers allowed us to identify three haplotypes among the gleba of the seven truffles collected on the same plant roots, AFLPs resolved each gleba as a particular genotype, corroborating the view that each gleba derives from a different mycelial strain. This finding, incidentally, broadens the results of previous studies showing the co-existence of different T. melanosporum strains on a single root system (Bertault et al., 2001), although, differing from these authors, we interpret this high amount of genetic diversity as resulting from an extensive outcrossing rather than from simultaneous presence of different self-perpetrating clones. In keeping with this inference, AMOVA of SSR, ITS and AFLP data proved that most of the genetic variation (78, 88 and 85%, respectively) was found within populations, with a smaller, albeit significant, percentage among populations.

Glacial refugia in southern Europe?

Neither RAPD nor SSR markers proved the presence of a phylogeographic signal in populations from the northernmost T. melanosporum distributional range (Bertault et al., 1998), whereas such a signal discriminating eastern from western French populations was claimed by Murat et al. (2004) on the basis of ITS haplotype distribution. Embracing the glaciation hypothesis, the more southerly populations are expected to show greater genetic diversity. With the notable exception of population 6 from Piemonte (northern Italy), which showed 50% of the truffles with haplotype III, our ITS survey documented a large prevalence of truffles with haplotype I in any population. Conversely, in keeping with the hypothesis that the maximum allelic richness is expected to be found where populations were confined during the last ice age (Comps et al., 2001), SSR data documented a higher allelic richness in the southernmost populations of the species’ range: populations 11 and 12 from Spain, population 13 from Sardinia and population 3 from Abruzzo.

It is worth mentioning that populations 3, 12 and 13 also displayed the highest ratio of AFLP diversity, with a HS value of 1 (Table 1). Furthermore, this evidence nicely fits with the observation that the main glacial refugia for some of the host species (Quercus pubescens, Tilia and Corylus spp.) with which T. melanosporum has to establish a symbiotic relationship were located in the Italian, Iberian and Balkan peninsulas (Brewer, 2002; Petit et al., 2003). Hence we are tempted to speculate that populations 3, 11, 12 and 13 hold part or most of the genetic diversity surviving from the glaciation. Interestingly, cluster analysis performed on pairwise FST distances among populations showed a tendency of the southernmost populations (5, 12, 11, 13, 3, 4) to be grouped together.

Although AMOVA indicated the existence of a significant population differentiation and PCO seems to depict a separation between French and Spanish populations, SAMOVA and STRUCTURE analyses did not support a clear geographic distribution of the genetic diversity. Additionally, the Mantel test did not reveal an increase in genetic differentiation with geographic distance.

Future investigation using a larger sample collection and multilocus species-specific markers such as retrotransposon-based markers (Riccioni et al., 2007) will help us to shed more light on the amount and distribution of genetic polymorphisms throughout the T. melanosporum geographical range. Until now, the evidence that no isolation by distance was found within the sampling area is congruent with the hypothesis of a drastic restriction of the species range in the Italian and Spanish peninsulas during the last glaciation, followed by a rapid northward expansion of a few strains that had to intercross to fruit and propagate.

In conclusion, we have provided data that lead to a profound re-interpretation of the T. melanosporum reproductive mode and, consequently, of the methodology for studying and modeling the distribution of its genetic variability at different geographical levels. In addition, our findings call for a reconsideration of agronomic practices for T. melanosporum cultivation, spanning from the methodology of host plant inoculation to the correct management of man-made truffle plantations to sustain a sufficient degree of fungal biodiversity to allow strains to mate and fructify.


This research was partially funded by the Region of Umbria and by Fondazione Cassa di Risparmio – Perugia. The authors are grateful to Profs G. Chevalier (Department of Micology, INRA, Clermont-Ferrand Cedex, France), A. de Miguel (Department of Botany, University of Navarra, Pamplona, Spain), B. Granetti (Department of Plant Biology and Biotechnology, University of Perugia, Italy) and G. Pacioni (Department of Environmental Sciences, University of L’Aquila, Italy), as well as to Comunità Montana Monti Martani e del Serano, Comunità Montana Valnerina Norcia, and Mr D. Manna for kindly providing us with T. melanosporum samples.

The authors thank Mrs Helen Sullivan Sini for editing the manuscript.