Multilocus Phylogeography of the Tuber mesentericum Complex Unearths Three Highly Divergent Cryptic Species

Tuber mesentericum is an edible European black truffle, apparently easy to recognize, but showing a high degree of genetic variability. In this study, we performed an integrative taxonomic assessment of the T. mesentericum complex, combining a multilocus phylogeographic approach with morphological analyses, and including authentic specimens of Vittadini, and Berkeley and Broome. We performed maximum likelihood phylogenetic analyses, based on single and concatenated gene datasets (ITS rDNA, β-tubulin, elongation factor 1-α), and including all available sequences from previous studies. Phylogenetic analyses consistently recovered three reciprocally monophyletic and well-supported clades: clade I, with a wide range across Europe; clade II, specimens collected mainly in the Iberian, Italian, and Balkan peninsulas; and clade III, specimens collected almost exclusively in central Italy. Genetic distance between clades ranged from 10.4% to 13.1% at the ITS region. We also designed new primer pairs specific for each phylogenetic lineage. Morphology of spores, asci, and peridium were investigated on specimens representing the three lineages. Macro- and micromorphological analyses of ascomata revealed only a few, but not diagnostic, differences between the three phylogenetic lineages, thus, confirming that they are morphologically cryptic. By studying authentic specimens of Vittadini, and Berkeley and Broome, it was possible to identify the three clades as T. mesentericum, Tuber bituminatum, and Tuber suave sp. nov., and to designate an epitype for T. mesentericum s.s. and a lectotype for T. bituminatum. Future investigations on volatile organic compound (VOC) composition are needed to define the aroma repertoires in this species complex.


Introduction
Truffles are hypogenous ascomata, mainly formed by fungi in the Ascomycota and Basidiomycota phyla. Those in the genus Tuber (Ascomycota, Pezizales), the so-called true truffles, live in mycorrhizal symbiosis with the roots of many trees, shrubs, and herbaceous plants [1]. Tuber species that produce ascomata with a pleasant smell and marketable size are considered edible and, therefore, of commercial interest. In some European countries, edible truffles have traditionally been considered one of the most appreciated and expensive foods, and their cultivation and consumption are now rapidly spreading worldwide [2].
Since the advent of molecular phylogenetics, many researchers have worked to revise the taxonomy within the genus Tuber [3][4][5][6]. Currently, about 120 Tuber species have been described and molecularly characterized from Asia, Europe, and Central and North America [7]. However, the last revision of the genus estimated from 180 to 220 species, grouped into 11 major clades [8]. A large portion of such species richness is due to the presence of cryptic lineages in many of these clades. Indeed, cryptic species have been found in Tuber anniae W. Colgan & Trappe [9], Tuber borchii Vittad. [10], Tuber brumale Vittad. [11], Tuber excavatum Vittad. [12], Tuber indicum Cooke & Massee [13,14], and Tuber rufum Picco [15]. However, the taxonomy of many of these species complexes has not been resolved.
Phylogenetic studies on T. mesentericum are scarce. Pacioni and Pomponi [22], based on allozyme polymorphisms, recognized four distinct clades within T. mesentericum, one of which was later associated with Tuber bellonae Quélet [23]. Sica et al. [24] analyzed rDNA internal transcribed spacer (ITS) sequences in 126 specimens, mainly from the Campania region (southern Italy), and found 34 distinct haplotypes, but did not reveal any significant morphological differences among the analyzed ascomata. Benucci et al. [25] and Marozzi et al. [26] uncovered three distinct clades within T. mesentericum, using ITS and elongation factor 1-α (EF1α) sequences, respectively. These lineages may well represent cryptic species, one of which is T. mesentericum sensu stricto. A proper taxonomic assessment of the T. mesentericum species complex has not been performed yet, and it is unclear whether phylogenetic lineages are also distinct by any morphological feature, or by what their geographic distributions are.
In this study, we performed a comprehensive phylogeographic assessment of the T. mesentericum species complex, based on multilocus data, and including specimens from all over Europe. The taxonomy of the lineages found was resolved through an integrative approach combining molecular data, morphological characters of peridium and spores, and analysis of voucher specimens of T. mesentericum determined by Vittadini, and the type specimens of T. bituminatum determined by Berkeley and Broome. The new species Tuber suave sp. nov. was also attributed to one of these lineages.

ITS Amplification and Sequencing of Historical Voucher Specimens
Due to the low amount and bad quality of fungal material, the four historical voucher specimens from the Kew herbarium were amplified by a direct PCR approach, using newly designed primer pairs specific for each cryptic lineage, inferred from phylogenetic analyses.
Because of the high degree of DNA degradation of the historical voucher specimens, the lineage-specific primer pairs were targeted to short segments (under 380 bp) of the ITS regions. To this aim, all ITS sequences of the T. mesentericum complex considered in this study were aligned with BioEdit, and the most informative domains of ITS1 and ITS2 regions were identified to design forward and reverse primers, respectively ( Figure S2). Primer selection was performed using Primer Express 3.0 (PE Applied Biosystems, Waltham, MA, USA) and Primer3 v 4.1.0 [34]. The specificity of each primer pair was preliminarily tested on in silico analyses, and then on DNAs extracted from ascomata of Tuber species genetically closely related to T. mesentericum (T. melanosporum, T. brumale, T. aestivum, T. indicum, and T. magnatum). The sequences of the lineage-specific primers designed in this study are reported in Table 2. Table 2. Sequences of the clade-specific primer pairs designed in this study.

Clade
Forward ( Small fragments of dry tissue (<1 mg) were transferred to 0.2 mL PCR tubes and rehydrated with 7 µL of ultrapure water, and then incubated at 65 • C for about 30 min before being submitted to a freeze-thaw process (1 min in liquid nitrogen followed by 1 min at 65 • C, three times) to facilitate tissue disruption and DNA release. Sample lysis was enhanced by using sterile glass micropestles specifically tailored for 0.2 mL tubes.
The PCRs were carried out in 50 µL volume reactions containing 25 µL of BioMix TM (2×), 200 nM of each primer, and 40 µg of BSA. The PCRs for each ITS region were run with the following cycling protocol: initial denaturation at 95 • C for 6 min; 40 cycles at 94 • C for 45 s, 53 • C for 30 s, and 72 • C for 45 s; and final extension at 72 • C for 7 min. The PCR products were purified and sequenced, as described above.

Phylogenetic Analyses
DNA sequences were checked and assembled using Geneious R8 (Biomatters Ltd., Auckland, New Zealand). Tuber melanosporum and T. magnatum were used as outgroup taxa (Table S1). Multiple sequence alignments for the three loci were performed in MAFFT7 [35], using the E-INS-i iterative refinement algorithm.
Maximum likelihood (ML) phylogenetic analyses based on single-gene datasets (βtub, EF-1-α, and ITS), as well as on the concatenated dataset β-tub + EF-1-α + ITS, were performed. The ITS dataset was significantly enlarged by adding all available sequences from previous studies (mainly from Sica et al. [24]) to maximize geographic coverage of samples. Maximum likelihood analyses were performed in raxmlGUI 1.5b2 [36], a graphical front end for RAxML 8.2.1 [37], with 100 independent ML searches, 100 bootstrap replicates, and 1000 rapid bootstrap replicates, and by applying the best-fit partitioning scheme and substitution models selected by PartitionFinder 2 [38]. For the concatenated dataset, a partitioned model with four partitions received the highest score under Akaike's information criterion (ITS1 and ITS2: GTR + I + G; 5.8S: GTR + I + G; β-tub GTR + G; EF-1-α GTR + I + G).

Morphological Analyses
Microscopic characters of spores, asci, and peridium were examined on hand-made sections or squash preparations obtained from 16 specimen vouchers (5-6 for each cryptic lineage, Table 1). Each sample was rehydrated for 10 min in 20% KOH, rinsed with sterile water, and then soaked with 3% KOH, following the procedure described by Leonardi et al. [39]. Observations and measurements were made under a Zeiss AXIO imager2 microscope and images were captured by a Leica DFC320 camera.
Only fully mature spores in which the episporia were clearly distinguishable were considered for the analyses. The color of the spores was determined using the Rayner [40] (R) mycological color chart and the ColorHexa (H) color scale (https://www.colorhexa. com/color-names or https://www.w3schools.com/colors/colors_picker.asp, accessed on 10 November 2021), at 400× magnification with a 5000 • K light source, without a filter. The measurements of the microscopic characters were carried out at 400× or 1000× magnification. The following spore characters were measured (Table 3, Figure S3): L1 and W1, and spore length and width, excluding episporium (cell lumen); L2 and W2, and spore length and width, including episporium; L3 and W3, and spore length and width, including episporium and exosporium (ornamentations); Q, length/width ratio (L1/W1, L2/W2, L3/W3); episporium thickness (spore wall); and exosporium thickness (height of ornamentations). L2 and W2 were also measured from one-, two-, three-, four-, five-, and six-spored asci, when present. In addition, dimensions of the asci, peridial elements (layers and cells of the exo-and endoperidium), and warts were also measured. Measurements were reported in the text as minimum and maximum values, while means ± standard deviations are reported in Table 3. Differences in spore size (L2, W2, and L2/W2) among the cryptic lineages were evaluated by one-way repeated measure ANOVA, with repeated measures using the AoV function (measure~species + Error(ascoma)) in R [41]. Statistical analyses were carried out using log-transformed data to account for the effect of data skewness on the statistical tests.

Phylogenetic Analyses
Phylogenetic analyses based on either single genes or on the concatenated dataset consistently recovered three reciprocally monophyletic clades (Figures 1 and 2) that received maximum support (bootstrap support = 100) in the concatenated analysis: (i) one clade with the widest range across Europe (clade I); (ii) a second clade grouping specimens collected mainly in the Iberian, Italian, and Balkan peninsulas (clade II); and (iii) a third clade including specimens collected almost exclusively in central Italy and Greece (clade III). The ranges of the three clades were found to overlap in different European regions, particularly in western Europe, Italy, and Greece ( Figure 3).     Genetic divergence among the three clades ranged from 0.097 to 0.123 (uncorrected p-distance) and from 0.104 to 0.131 (Kimura-2-parameter distance) at the ITS region; from 0.037 to 0.053 (uncorrected p-distance) and from 0.034 to 0.054 (Kimura-2-parameter distance) at the β-tubulin region; and from 0.026 to 0.034 (uncorrected p-distance) and from 0.027 to 0.032 (Kimura-2-parameter distance) at the EF-1-α region. Intra-clade divergence at the ITS region was 0.0047 for clade I, 0.0074 for clade II, and 0.0032 for clade III, using either p-distance or Kimura-2-parameter estimates.
The ITS sequence of T. mesentericum Vittadini's authentic specimen and that of one specimen of T. bituminatum from the Kew herbarium were clustered into clade I ( Figures  1 and 2). ITS sequences of the other two T. bituminatum specimens were grouped with samples of clade II (Figures 1 and 2). All of these sequences were generated with the species-specific primer pairs designed in this study ( Table 2). The efficiency of these specific primers was also tested on the specimens used for phylogenetic analyses, and on the other 21 additional specimens deposited in the AQUI and MCVE herbaria (Tables 1 and S1).

Ascoma Morphology
Micro-and macromorphological characters of ascomata showed a wide overlap among ascomata of different clades and have not proved useful for their taxonomic identification. Few differences were found, only in the shape of peridial cells constituting the center of warts and in the morphology of spore reticulum. Ascomata in clades I and II showed a pseudoparenchymatous peridium with globose-angular cells, whereas those in clade III showed a pseudoparenchyma composed of elongated cells that radiated out from the base of the warts ( Figures 4C, 5B and 6B). The spores in clades I and III always showed regular and completely formed alveolae ( Figures 4D and 6C,D), whereas a variable percentage of those in clade II had labyrinthine pseudoreticula with incomplete and irregular alveolae ( Figure 5C,D). Genetic divergence among the three clades ranged from 0.097 to 0.123 (uncorrected p-distance) and from 0.104 to 0.131 (Kimura-2-parameter distance) at the ITS region; from 0.037 to 0.053 (uncorrected p-distance) and from 0.034 to 0.054 (Kimura-2-parameter distance) at the β-tubulin region; and from 0.026 to 0.034 (uncorrected p-distance) and from 0.027 to 0.032 (Kimura-2-parameter distance) at the EF-1-α region. Intra-clade divergence at the ITS region was 0.0047 for clade I, 0.0074 for clade II, and 0.0032 for clade III, using either p-distance or Kimura-2-parameter estimates.
The ITS sequence of T. mesentericum Vittadini's authentic specimen and that of one specimen of T. bituminatum from the Kew herbarium were clustered into clade I (Figures 1 and 2). ITS sequences of the other two T. bituminatum specimens were grouped with samples of clade II (Figures 1 and 2). All of these sequences were generated with the species-specific primer pairs designed in this study ( Table 2). The efficiency of these specific primers was also tested on the specimens used for phylogenetic analyses, and on the other 21 additional specimens deposited in the AQUI and MCVE herbaria (Table 1 and Table S1).

Ascoma Morphology
Micro-and macromorphological characters of ascomata showed a wide overlap among ascomata of different clades and have not proved useful for their taxonomic identification. Few differences were found, only in the shape of peridial cells constituting the center of warts and in the morphology of spore reticulum. Ascomata in clades I and II showed a pseudoparenchymatous peridium with globose-angular cells, whereas those in clade III showed a pseudoparenchyma composed of elongated cells that radiated out from the base of the warts ( Figures 4C, 5B and 6B). The spores in clades I and III always showed regular and completely formed alveolae ( Figures 4D and 6C,D), whereas a variable percentage of those in clade II had labyrinthine pseudoreticula with incomplete and irregular alveolae ( Figure 5C,D).       No statistical differences were found in any of the measured characters of the gleba hyphae, asci, or spores of the three clades. Nearly significant differences were found only for L2 (p = 0.051) and L2/W2 (p = 0.062) spore parameters when values were logtransformed before ANOVA ( Figure S4, Table S2).

Discussion
The phylogeny of the T. mesentericum complex has been poorly investigated to date, and only recently has it attracted the interest of mycologists. So far, three studies have investigated the genetic diversity of T. mesentericum by applying a single-gene analysis [24][25][26], but none of them have resolved the taxonomic issues within this species complex. In this study, we performed an integrative taxonomic assessment of the T. mesentericum complex by combining a multilocus phylogeographic approach with in-depth morphological analyses, and including authentic specimens of Vittadini and Berkeley and Broome. This approach allowed us to resolve the main phylogenetic clades found within the T. mesentericum complex into three species and to designate an epitype for T. mesentericum s.s.
In agreement with previous studies by Benucci et al. [25] and Marozzi et al. [26], we found three distinct and well-supported lineages, both in single-locus and multi-locus phylogenetic analyses.
Phylogeographic data generated in this study combined with data from public repositories indicate that members of clade I are spread all over Europe, from Sweden to Italy and from Spain to Bulgaria, and are associated with many hardwood trees, including Quercus spp. (Q. cerris L., Q. robur L.), Fagus sylvatica L., Corylus avellana L., Ostrya carpinifolia Scop., but also with conifers, such as Picea spp. The range of clade II seems to be mainly limited to Mediterranean habitats, and their members show a similar host preference. Clade III is only distributed in central Italy and Greece, although a specimen from GenBank (JQ348414) had "France" as its general indication of origin, which should be verified. To date, ascomata of Clade III have been found only under thermophilic oaks.
Clades I and II seem to be sister species that diverged after clade III. While the average intraclade divergence at the ITS marker was <0.8%, the divergence between these No statistical differences were found in any of the measured characters of the gleba hyphae, asci, or spores of the three clades. Nearly significant differences were found only for L2 (p = 0.051) and L2/W2 (p = 0.062) spore parameters when values were logtransformed before ANOVA ( Figure S4, Table S2).

Discussion
The phylogeny of the T. mesentericum complex has been poorly investigated to date, and only recently has it attracted the interest of mycologists. So far, three studies have investigated the genetic diversity of T. mesentericum by applying a single-gene analysis [24][25][26], but none of them have resolved the taxonomic issues within this species complex. In this study, we performed an integrative taxonomic assessment of the T. mesentericum complex by combining a multilocus phylogeographic approach with in-depth morphological analyses, and including authentic specimens of Vittadini and Berkeley and Broome. This approach allowed us to resolve the main phylogenetic clades found within the T. mesentericum complex into three species and to designate an epitype for T. mesentericum s.s.
In agreement with previous studies by Benucci et al. [25] and Marozzi et al. [26], we found three distinct and well-supported lineages, both in single-locus and multi-locus phylogenetic analyses.
Phylogeographic data generated in this study combined with data from public repositories indicate that members of clade I are spread all over Europe, from Sweden to Italy and from Spain to Bulgaria, and are associated with many hardwood trees, including Quercus spp. (Q. cerris L., Q. robur L.), Fagus sylvatica L., Corylus avellana L., Ostrya carpinifolia Scop., but also with conifers, such as Picea spp. The range of clade II seems to be mainly limited to Mediterranean habitats, and their members show a similar host preference. Clade III is only distributed in central Italy and Greece, although a specimen from GenBank (JQ348414) had "France" as its general indication of origin, which should be verified. To date, ascomata of Clade III have been found only under thermophilic oaks.
Clades I and II seem to be sister species that diverged after clade III. While the average intraclade divergence at the ITS marker was <0.8%, the divergence between these clades was as high as 10% or 13%, values that are well above those observed among the currently accepted Tuber species. For example, interspecific divergence observed between closely related pairs of currently recognized species belonging to Rufum and Melanosporum clades are in the range of 4.7-7.4% [39]. Bonito et al. [42] suggested a threshold for Tuber species delimitation based on an interspecific divergence of ≥4%. However, interspecific divergence below 4% has been reported between closely related species of different clades, such as the Excavatum [12], Puberulum [9], and Gibbosum [43] clades.
Morphological and biometric characters of ascomata examined in this study do not allow an unambiguous identification of the clade to which they belong, thus confirming that these lineages represent cryptic species. Only two of the microscopic characters, the peridium (cell pattern of the peridium cells in the center of the single warts) and spores (reticulum integrity), showed some differences among clades, although they are not diagnostic and, therefore, of little taxonomic utility. ANOVA analyses based on spore sizes showed low differentiation among clades, with only spore length (L2) being marginally significant.
The mismatch between phylogenetic and morphological differentiation seems to be a common feature in the genus Tuber. Bonuso et al. [10] identified two genetically isolated groups in T. borchii, but no distinctive morphological features were found in support of this separation. Ascoma morphology has not been useful in solving the taxonomy of the T. indicum complex, where at least two cryptic species exist, according to different phylogenetic studies [13,14,44]. The morphological distinction between T. brumale and T. cryptobrumale is also challenging; Merényi et al. [11] verified that only a combination of different morphological characters would make it possible to differentiate most (95%) specimens belonging to these two pseudocryptic species. The issue of cryptic species seems even more complex in other Tuber lineages. Five different species complexes were identified by Healy et al. [15] in the Rufum clade, and as many cryptic species were found by Puliga et al. [12] for the European truffle T. excavatum. From a taxonomic point of view, the greatest challenge will be to define and name each cryptic species within these complexes.
The integrative approach used in this study has revealed itself to be particularly fruitful in resolving the taxonomy of the cryptic species complex of T. mesentericum. The genotyping of authentic specimens of Vittadini, and Berkeley and Broome from the Kew herbarium allowed us to identify clade I as T. mesentericum sensu stricto, clade II as T. bituminatum, and clade III as a new species named T. suave, for the pleasant smell that seems to be typical of its ascomata. The type of T. mesentericum, as well as the types of the other Tuber species described by Vittadini [45], are no longer available for genotyping. Vittadini's private herbarium was never deposited in a public institution, and it was lost because of insect infestation [46]. Only a few authentic specimens determined as T. mesentericum by Vittadini are actually available at the herbaria of Kew, Padua, and Uppsala ( Figures 4B and S1). The specimen labelled as T. mesentericum (K(M)190347), and one of the three specimens labelled as T. bituminatum from the Kew herbarium (Figures 4B and S1a), were successfully amplified by using the ITS primer pair newly designed for the members of clade I. On the contrary, the PCR amplifications of the specimens deposited in the Padua and Uppsala herbaria ( Figure S1B,C) were unsuccessful, regardless of the lineage-specific primer pair designed in this study. This was probably due to their poor state of preservation, biological contamination, or the chemicals added to preserve the exsiccata. The iconography of Vittadini [45] (Table III, Figure XIX) constitutes an ambiguous lectotype because it only represents a detail of the gleba, and not the entire ascoma with the warty black peridium and the basal cavity, hence the epithet "mesentericum". Therefore, we designated a specimen collected in Vittadini's study area as an epitype of T. mesentericum s.s. This specimen has an ITS sequence identical (100% identity) to that of the authentic specimen voucher from the Kew herbarium (K(M)190347) ( Figure 4B).
Tuber bituminatum was described by Berkeley and Broome [47], but was later considered a synonym of T. mesentericum or T. aestivum [48][49][50]. Our analyses demonstrated that two out of three specimens indicated by Berkeley and Broome as holotypes of T. bituminatum fall into clade II and, therefore, this species should be revaluated and exclusively attributed to the members of this lineage within the T. mesentericum complex. Recently, Crous et al. [51] described the new species Tuber alcaracense Ant. Rodr. & Morte, having ascomata similar to those of species of the T. mesentericum complex, but with a pleasant odour and lacking a basal cavity. A preliminary phylogenetic comparison of the two ITS sequences available for T. alcaracense (MN810046-7) with the ITS sequences generated in this study indicates that T. alcaracense is nested within the clade of T. bituminatum (clade II, results not shown). Therefore, most likely T. alcaracense is a synonym of T. bituminatum, as the latter name has priority. However, before drawing taxonomic conclusions in this regard, more in-depth morphological and phylogenetic analyses must be carried out.
As we observed for the Kew herbarium, it is possible that many collections labelled as T. mesentericum in private or public herbaria can also include T. bituminatum and/or T. suave specimens, due to the scarcity of morphological traits useful in distinguishing them. For example, the authentic specimen of T. mesentericum preserved in the PD herbarium (Milano 1841, leg. et det. C. Vittadini) for its labyrinthine spore ornamentation might belong to T. bituminatum.

Conclusions
This study clarifies the taxonomy of a commercially important truffle group, and will be useful to support further studies on ecology, cultivation, and foodomics of these species. The specific primer pairs tested in this study can be used for rapid and easy identification of members of the three studied species. Finally, specific investigations on VOC composition are needed to define the aroma repertoires existing in the T. mesentericum complex and to commercially promote these species.