Nuclear differentiation patterns in western Eurasian beeches: a review of published isoenzyme and ISSR data

Nuclear data reject the traditional split into the European beech (Fagus sylvatica subsp. sylvatica according to Greuter & Burdet 1981) and the Euxinian-Hyrcanian beeches, which have been collectively treated as Oriental beech (F. sylvatica subsp. orientalis (Lipsky) Greuter & Burdet). The westernmost populations of the Oriental beeches, including the type locality (see below F. orientalis), are genetically close and appear to form an active hybridization zone with the European beeches in the eastern Rhodopes (Papageorgiou & al. 2008, 2014). The populations of Oriental beeches in the wider Caucasus region and in the mountain ranges south of the Caspian Sea are notably distinct. Already the isoenzyme data of Gömöry & Paule (2010) revealed a higher diversity within the Oriental beeches than within their European counterparts, with the westernmost Oriental beeches being closest (in absolute terms) to the genetically homogenous and weakly differentiated European kin. The isoenzyme results correlate with recently produced 5S-IGS high-throughput sequencing data (see below). 5S-IGS data show a clear separation between the Iranian and European beeches, whereas samples representing the western Oriental beech (northeastern Greek F. orientalis) cluster with European beeches (F. sylvatica).

Fig. 2.

Recent nuclear SSR Bayesian clustering results mapped on the Oriental beech subgraph in Gömöry & Paule (2010, fig. 4) and illustrating the here proposed four-species concept for western Eurasian beeches. Hexagons indicate densely sampled population level data for beeches of the Caucasian region (Sękiewicz & al. 2022, k=2–3); circles and squares indicate geographically representative samples from Central Europe and across Oriental beeches (circles: Kurz & al. 2023, fig. S4, k=2–6, re-analysed excluding the Nur Mountains beeches; squares: Budde & al. 2023, k=4). Numbers in square brackets indicate the evolutionary sequence of (most recent) speciation processes and two possible rooting scenarios: [1a]; initial split between Hyrcanian and the remaining western Eurasian spp.; [1b], initial split between the western (European-Anatolian) and eastern (Caucasian-Caspian) spp.; [2] divergence of the Caucasian species Fagus hohenackeriana; [3] divergence of F. sylvatica-orientalis. The bicoloured X indicates potential secondary contact (past or ongoing hybridization).

Hence, the eastern Oriental beeches (Fagus orientalis s.str.) are inferred to be direct sisters of the European beeches (F. sylvatica s.str.), both of which form a lineage that diverged earlier (pre-Pliocene) from the ancestor(s) of the eastern Oriental beech lineage(s) (Gömöry & al. 2018; Cardoni & al. 2022).

Recently generated nuclear SSR data (Fig. 2; Kurz & al. 2023, Budde & al. 2023) have further confirmed that the primary split is not between European and Oriental beeches but between the beeches of Europe and adjacent northwestern Anatolia (Asia Minor), Fagus sylvatica and F. orientalis s.str., and those of the Caucasus and further east. The Caucasian and Caspian beeches can be divided in two genetically isolated entities as well (Gömöry & Paule 2010; Sękiewicz & al. 2022; Budde & al. 2023; Kurz & al. 2023). According to isoenzyme data (Gömöry & Paule 2010), the Crimean occurrence either represents a disjunct relict population of F. sylvatica or an ancient hybrid between western and eastern species (Fig. 2). The origin of the beeches of the Nur Mountains in the Hatay province, south-central Turkey, is a complex taxonomic-systematic puzzle. Their distinct nuclear SSR cluster (Kurz & al. 2023, fig. S4) and isoenzyme profiles indicate a sister relationship with both F. orientalis s.str. and the Caucasian beeches (prominent box in Gömöry & Paule's 2010 fig. 4). We re-analysed the data set of Kurz & al. (2023), excluding the Nur Mountain population but using exactly the same set-up and protocol as detailed in Kurz & al. (2023). The result, annotated against the background of previously known and here documented phylogenetic relationships within western Eurasian beeches, is shown in Fig. 3. The basic structuring is the same as when the Nur Mountain population is included: a west-east gradient with the westernmost (F. sylvatica) and easternmost beeches (newly recognized F. caspica) marking the endpoints. There are two notable exceptions. At k=4, the additional genetic cluster (“Ancient link cluster” in Fig. 3) connects the northern Anatolian and southwestern Georgian beeches (Pontic or Pahar Mountains) with the northern Iranian, which are at least 1000 km air-distance apart, while the beeches of the Greater Caucasus appear genetically homogenous (“Eastern”/ “East Euxinian-Hyrcanian cluster”; mixed profiles when the Hatay populations are included). In the northern Anatolian populations, the Ancient link cluster replaces to a large degree the “West Euxinian-Mediterranean cluster” developing at k=3 characterizing the western F. orientalis (F. orientalis s.str.). In the southwestern Georgian population (easternmost part of the Pontic Mountains), it replaces most of the East Euxinian-Hyrcanian cluster as well. The highest supported k is k=5 (supplementary content, file Genotypification.xlsx), at which the Ancient link and East Euxinian-Hyrcanian clusters are replaced by three clusters sorting from west to east: (1) a “Pontic cluster” connecting the easternmost F. orientalis s.str. and south-westernmost F. hohenackeriana, (2) a “Caucasian cluster” (F. hohenackeriana) and a cluster for F. caspica. At k=6, the Pontic cluster is further subdivided along the here proposed species boundaries, with one cluster dominant in the populations of the western Pontic Mountains (northern Anatolia; F. orientalis s.str.), and the other characterizing the eastern Pontic Mountains (F. hohenackeriana). In the light of the 5S-IGS pools (next section) and the results of Gömöry & Paule (2010), the results obtained for k≥4 not only represent a genetic west-to-east gradient along the Pontic Mountains bridging between the western and eastern F. orientalis and potentially stabilized by young or ongoing gene flow but also point to incomplete lineage sorting of a polymorphic ancestor, with the populations along the Pontic Mountains (F. orientalis in northern Anatolia, F. hohenackeriana in northeastern Turkey/southwestern Georgia) being closer to the common ancestor of the western Eurasian species than their kin in northwestern Anatolia (F. orientalis s.str.) and the Greater Caucasus (F. hohenackeriana).

Fig. 3.

Bayesian clustering analysis of Kurz & al. (2023) data, with Nur Mountains population excluded. Arrows indicate potential phylogenetic relationships indicated by the differential k values: k ≤ 3, primary split into a western and eastern lineage; k = 4, emergence of a new cluster partially replacing the west vs east pattern at lower k and shared between disjunct beech populations of the Pontic and Elburz Mountains; k ≥ 5, differentiation between the eastern species and division between western and eastern Fagus orientalis, the latter clustered with south-easternmost populations of F. hohenackeriana, representing a genetic west-east cline along the Pontic Mountains (see Conclusion and outlook).

Fig. 4.

Pie charts showing the relative abundance of main 5S-IGS sequence variants as defined and discussed by Cardoni & al. (2022). Data for Fagus sylvatica, F. orientalis and F. caspica based on Cardoni & al. (2022; graphic modified after Schulze & Grimm 2022, fig. 2); data for F. hohenackeriana added for this study.

Among the here recognized four beech species of western Eurasia, the Caucasian Fagus hohenackeriana shows the highest intra-species variation and least coherence with a clinal west to east variation (Gömöry & Paule 2010; Sękiewicz & al. 2022) connecting it with the adjacent F. orientalis populations in the Pontic mountain chains of northern Anatolia (Fig. 3). It is also the species with the highest 5S-IGS diversity (see following section) and morphological plasticity (section Taxonomic treatment). Its eastern sibling species, F. caspica represents the most homogenous and genetically isolated of all western Eurasian species; it shows the highest genotypic and phenotypic coherence. Notably, it appears to be genetically most basal (results of k=4; next section); possibly retaining the most ancestral of all western Eurasian gene pools. The two western species are increasingly sorted in their genetic (east to west) and morphological traits, with the European beech, F. sylvatica s.str. representing the evolutionary endpoint of beeches in western Eurasia.

Classification of main 5S-IGS lineages with the Evolutionary Placement Algorithm (EPA)

The EPA treats each representative sequence as a query and tries to find its most probable placement within the reference tree; it returns jPlace formatted files including the likelihood weighting ratios (LWR) for each tested internode of the reference tree.

The 38-tip reference used for the “primary EPA” allows distinguishing between the two main types of Lineage A and Lineage B 5S-IGS variants (“Original A”, “European A”, “Original B”, “European B”), lineages (putatively) confined to Fagus crenata (“Crenata B2” and “Crenata B3”) and the two dominant lineages detected within F. japonica Maxim. (“Japonica I”, representing the sister to the B Lineage shared by F. crenata and the western Eurasian beeches; “Japonica O”, most distinct 5S-IGS sequence variants forming the O Lineage and used to root the tree) with high efficiency. The O Lineage has been detected as low-abundant variants with pseudogenic tendencies across western Eurasia (“Eurasian O” in Fig. 4; = “European O” of Cardoni & al. 2022), in addition to rare to very rare variants that stem from an earlier radiation (“Relict Lineage”). Figure 4 is a graphical summary of the results (the full list can be found in the supplementary content, file EPA_annotation.xlsx).

The newly added data from Georgia complement the results of Cardoni & al. (2022) and confirm both the genetic heterogeneity as well as uniqueness of Fagus hohenackeriana, both in relation to its eastern potential sister species, F. caspica, as well as to its western siblings (F. sylvatica-orientalis). Both eastern species lack the European B variants, a most-evolved sequence type and evolutionary lineage that is diagnostic of the western species pair. Within the Georgian beeches, the sample from the Greater Caucasus differs by the relative abundance of A Lineage and B Lineage variants and the lack of European A variants, which are most abundant in the Lesser Caucasus and eastern Georgian samples. The name European A for this group of 5S-IGS variants refers to the observation by Cardoni & al. (2022) that variants of this lineage have not been found in the Iranian samples (F. caspica; see following paragraphs). The Hyrcanian species, F. caspica, is almost exclusively composed of (mostly not shared; cf. Cardoni & al. 2022) Original A and Original B variants. Original A variants are very rare in the western species pair, F. sylvatica + F. orientalis, but subdominant across the Georgian samples (F. hohenackeriana). The lineages formed by the few and low-abundant variants originally unique to the Iranian samples (“Iranian B2”, “Iranian B3” in Fig. 4, 5) appear to be shared to some degree with the southern and eastern F. hohenackeriana but are absent from the Greater Caucasus sample.

A new reference for in-depth EPA and classification of 5S-IGS variants of Western Eurasian beech species into species-diagnostic and shared general sequence types

The 38-tip reference cannot distinguish subtypes within the main types or any putative or potential new major type private to the Georgian beeches. Hence, in order to assess the possible amount and distinctness of new types in the Georgian samples, we also inferred a ML tree adding all variants with a total abundance of ≥ 25 in the Georgian samples to the 38-tip data matrix (Fig. 5). Based on this tree, the primary EPA results (preceding section), and visual inspection of accordingly selected placeholder sequences, we find that many of the Georgian 5S-IGS variants fall within the main lineages and their major types as defined by Cardoni & al. (2022). However, a remarkable number reflect (1) earlier diversification events than seen in the data analysed by Cardoni & al. (2022): “Hohenackeriana B2” and “Hohenackeriana B3”, addition to the Original B lineage; “Hohenackeriana C”, a new lineage of rare variants preceding or coeval with the split between Lineage B and I (Fig. 5–7). Further, (2) we see increased diversity within the Original A and Original B lineages, and (3) antipodal affinities of the Georgian samples to the western species pair Fagus sylvatica-orientalis on the one side (new type “Hohenackeriana A1” within European A lineage) and the Hyrcanian F. caspica on the other (shared “Eastern A” lineage and “Hohenackeriana A2” within Original A lineage).

Cross-correlation between conspicuous split-LWR patterns and the ML trees inferred by Cardoni & al. (2022) and here (Fig. 5) and visual inspection of representative sequences, in particular with respect to the T-dominated length-polymorphic sequence motif at the 5′ end of the 5S rDNA intergenic spacer, lead us to the recognition of additional general sequence variants with varying taxonomic potential that can be used as references. Therefore, we introduce and provide a new reference matrix and tree with 42 tips that captures not only the major 5S-IGS lineages but serves as a detailed framework for the taxonomic assessment of the 5S-IGS gene pools of any beech sample from western Eurasia included in this and future studies (Fig. 6; NEWICK-, extended PHYLIP- and NEXUS-formatted files included in the SDA). In this new reference matrix and tree, the tips are labelled for the 5S-IGS variant types they represent. The purpose of this is to facilitate the interpretation of the EPA results and LWR ratios reported in the jPlace output file from RAxML. Sequences with substantial similarity to the references and/or showing sequence patterns characteristic for each of the distinguished types, will be placed with high (sum) LWRs at the respective tips or their subtree (accordingly coloured internodes in Fig. 6). Variants related to sister types but with ambiguous affinity to either one or ancestral to them will have differential split LWRs and be linked to the subsequent deeper internodes. Since standard phylogenetic trees poorly handle ancestor-descendant relationships, it is not advisable to include ancestral sequence variants (plesiomorphic sequence types) in a reference tree as they will always increase the ambiguity of the EPA placement (and decrease branch support in general). Examples of the potentially ancestral types identified via consistently split LWR profiles and affinity to deeper internodes (“Ancestral B1”; “Primordial B1”; “Eurasian B”) have been verified visually. Hence, we can ascertain that the typical 5S-IGS variants of these types lack uniquely derived sequence features of the tip types (their descendant lineages): they represent either uniquely derived sibling lineages (rarely) or the ancestral sequence variant pool from which uniquely derived variants have evolved (Ancestral B1 → Western B1 + Hohenackeriana B1b; Primordial B1 → Ancestral B1 and derivatives + Shared B1; Eurasian B → Caspica B2 + most other B-types).

In a final step, we used this emended reference for an additional, “secondary EPA” focussing on variants that were identified as Lineage A or Lineage B variants (Fig. 7). The secondary EPA was performed on all samples of Cardoni & al. (2022) plus the three new Georgian samples of Fagus hohenackeriana; all unique sequence variants with an abundance ≥ 4 were used.

The distinction of the herein proposed four species is straightforward in view of the differential 5S-IGS pools depicted in Fig. 7. The western sister pair, Fagus sylvatica and F. orientalis differ from the eastern species by their dominant Western B2 and Western A types. In contrast to its European sister, the Euxinian F. orientalis shows a higher proportion of cross-species types. The Hohenackeriana B1b type may be indicative of post-divergence introgression from Caucasian beeches or remnant of an ancestral polymorphism that has been lost in F. sylvatica as the most homogenized of all western Eurasian beech species. The Hyrcanian F. caspica combines dominant cross-species types with low-frequent (near-) unique types (Caspica B1, Caspica B2) that otherwise can be found only in F. hohenackeriana. This most heterogenous species is genetically characterized by a mix of B-lineage types that represent the sister lineages of both the western species pair as well as its Hyrcanian counterpart F. caspica. In the A-lineage types, the F. hohenackeriana samples show different proportions of the cross-species Shared A type (≈ F. caspica or replaced by Hohenackeriana-specific types). Figure 8 shows a multifurcating key to identify western Eurasian beech species based on their 5S-IGS pool.

Fig. 5.

Partially collapsed maximum likelihood tree inferred from a matrix including all variants with a total abundance ≥ 25 across the added three Georgian samples (Fagus hohenackeriana), and with the 38-tip set of Cardoni & al. (2022, fig. 5; phylogram reprinted in inlet) included for reference. The five main lineages and major types are annotated and coloured accordingly. Asterisks indicate relabelled general 5S-IGS types based on our emended data basis.

Subgeneric classification of Fagus

Fagus L. subg. Fagus – Type (designated by Green 1929: 189): Fagus sylvatica L., Sp. Pl. 2: 998. 1753.

Molecular diagnosis — The subgenus differs consistently from all species of Fagus subg. Englerianae in any sufficiently divergent nuclear marker sequenced so far (Denk & al. 2002, 2005; Renner & al. 2016; Cardoni & al. 2022). Its ITS variants belong to Lineages II–IV as defined in Denk & al. (2005); the sequenced part of the Crabs Claw (CRC) gene (∼1650 bps) and the 2^nd intron of the Leafy gene (LFY, up to ∼1300 bps) include 14 subgenus-sorted SNPs (CRC pos. 212 [C vs T in F. subg. Englerianae], 714 [C↔T], 904 [C↔A], 1148 [A↔T/Y] and 1329 [T↔G]; LFY pos. 241 [A↔C], 398, 402, 465 [all G↔A], 507 [T↔A], 529 [C↔T], 1031 [T↔G], 1206 [A↔G] and 1246 [T↔A]; according reference matrices are included in the SDA, file RefMatrixCRCLFYSpCons .nex). Additional subgenus-diagnostic SNPs can be found in 21 of the 28 nuclear loci sequenced by Jiang & al. (2022; cf. supplement to Cardoni & al. 2022). Plastomes are divergent but geographically sorted and reflect two independent origins: Lineage I in North America; sibling lineages Lineage IV in East Asia and Lineage V in western Eurasia; see Fig. 1) The exception are populations of the Japanese species F. crenata comprising individuals that may carry near-private haplotypes of Lineage II (supplementary content, file Genotypification.xlsx, sheet PlstmDissim; including information from upcoming complete plastome data; Worth & al. 2021, work in progress).

Morphological diagnosis — Trees; buds sessile; leaves thick-chartaceous, abaxial leaf surface commonly smooth or papillate (Fagus longipetiolata), wax ornamentation on abaxial leaf surface missing or present (F. longipetiolata), size of stomata usually large, small in F. hayatae Palib., F. pashanica C. C. Yang, and F. grandifolia, subsidiary cells of stomata usually actinocytic to cyclocytic, anomocytic in F. grandifolia, leaf margin smooth or serrate; cupule peduncle short to long; pollen usually large, intermediate in F. hayatae, colpi usually short with more or less acute apex, or long and narrow with rectangular apex in F. grandifolia and occasionally in F. longipetiolata.

Species — Twelve: Fagus sylvatica, F. orientalis, F. hohenackeriana, F. caspica sp. nov. in western Eurasia (west to east); F. chienii W. C. Cheng (†?), F. crenata, F. hayatae, F. longipetiolata, F. pashanica, F. lucida in East Asia; F. grandifolia, F. mexicana Martínez in North America.

Remarks — Members of Fagus subg. Fagus can be traced in the fossil record based on their pollen and leaf-anatomical similarities with one or several modern-day species (Denk & Grimm 2009; Renner & al. 2016; Worth & al., work in progress). The oldest fossils representing this modern subgeneric lineage are cupules and leaves from the Eocene-Oligocene boundary, Northeast Asia (Pavlyutkin & al. 2014; see also Denk & Grimm 2009). Any molecular-phylogenetic tree analysis (e.g. Denk & al. 2005; Jiang & al. 2022) relying on sufficiently variable nuclear data will produce a prominent split with high (BS ≥ 70, PP ≥ 0.9) to unambiguous (BS = 100, PP = 1.0) support between the subgenera irrespective of the optimality criterion used for tree-inference. Newly sequenced individuals can be easily placed in either subgenus using e.g. the evolutionary placement algorithm implemented in RAxML 8 and its successor RAxML-ng. In contrast, any plastome data requires in-depth analysis and, in some cases, may fail to elucidate the subgeneric affinity.

Fagus subg. Englerianae Denk & G. W. Grimm, subg. nov.

Type: Fagus engleriana Seemen ex Diels in Bot. Jahrb. Syst. 29: 285. 1900.

Molecular diagnosis — The subgenus differs consistently from all species of Fagus subg. Fagus in any sufficiently divergent nuclear marker sequenced so far (Denk & al. 2002, 2005; Renner & al. 2016; Cardoni & al. 2022). Its polymorphic and notably divergent ITS variants belong to Lineage I as defined in Denk & al. (2005); the sequenced part of the Crabs Claw (CRC) gene and the 2^nd intron of the Leafy gene (LFY) include 14 subgenus-sorted SNPs (see above) in addition to several subgenus-restricted length-polymorphic patterns (see supplementary content, file Genotypification.xlsx, sheets CRC LP-patterns, LFY LP-patterns): an AC tetramer at pos. 1634ff in the CRC reference alignment; a 7 nt-long duplication at pos. 199ff, 45 nt-long deletion at pos. 573ff, and diagnostic oligonucleotide motives at pos. 671–696 and 737–764. Additional subgenus-diagnostic SNPs can be found in 19 of the 28 nuclear loci sequenced by Jiang & al. (2022; cf. supplement to Cardoni & al. 2022, data S5): in P4 at reference alignment (SDA, file RefMatrixJiangEtAlNcLoci.nex) positions 296, 321, 361, 504; P12—pos. 50, 230, 290, 590, 695; P14—pos. 106, 159, 195, 492; P21—pos. 420, 746, 793; P28—pos. 53/54: TC dinucleotide ↔ AT, AC, GC in F. subg. Fagus, pos. 135; P37—pos. 7; P42—pos. 96, 239, 274, 308, 409, 453, 649, 665; P48—pos. 145, 241, 262, 287; P52—71, 86, 90, 165, 243, 272, 407, 455; P54—pos. 82, 313, 616, 683, 703; P69—pos. 85, 130, 204, 251, 267, 406, 475, 542, 656, 737; P72—pos. 40, 91, 222, 306, 334, 387, 412, 415f (TT dinucleotide ↔ CA in F. subg. Fagus), 593; P97—pos. 236, 305, 394, 427, 467, 684 (C↔T/G); P98—pos. 461, 509; F128—pos. 272, 297, 406, 448, 522, 637, 661; F159—pos. 42, 252; F253—pos. 47, 72, 133, 140, 177; F286—pos. 58, 70, 133, 281; F289—pos. 106, 159, 195, 492; subgenus-diagnostic indels and oligonucleotide motives (including subgenus-restricted allelic variation) can be found in P42—GTCTA at pos. 619ff ↔ AG in F. subg. Fagus); P48—pos. 544–546: CGT/TGT vs CTG/CGG; P52—pos. 43–54: CAG-tetramer in F. subg. Englerianae, dimer F. subg. Fagus; P69—TATA at pos. 704–709 vs TAYAAA; F253—GGA at pos. 90ff ↔ GGGA in F. subg. Fagus.

Fig. 6.

42-tip reference tree for assigning 5S-IGS sequences variants and classification of variant groups with respect to the four-species framework introduced here for the western Eurasian beeches. Internode numbers as listed in the jPlace file and its spreadsheet transcript. Any query with (sum) LWR ≥ 0.9 associating it exclusively with likewise coloured internodes can be considered as a variant of this 5S-IGS lineage, some potentially ancestral or relict types (coloured outlines) will be expressed in characteristically split LWR profiles (see Appendix 1 for a detailed walk-through and explanatory examples). The inset gives a qualitative overview about the expected abundance of the distinguished (coloured accordingly) sequence types in the samples representing each species. Taxonomically diagnostic types for discriminating western Eurasian species are highlighted in bold font.

Morphological diagnosis — Multi-stemmed or low-branching trees; buds stipitate; leaves thin-chartaceous, abaxial leaf surface papillate, abaxial leaf surface with conspicuous wax ornamentation, size of stomata small, subsidiary cells of stomata anomocytic, leaf margin usually without teeth; cupule peduncle medium-long to long; pollen small, colpi long and narrow with rectangular apex.

Species — Three: Fagus engleriana, F. japonica, F. multinervis Makai; all of which are restricted to East Asia.

Remarks — The name Fagus subg. Englerianae was proposed by the doctoral thesis of Shen (1992) but has not been effectively published. The pollen of modern-day species of F. subg. Englerianae has to be treated as (sym) plesiomorphic (cf. Denk 2003) while leaf anatomical features allow to trace this subgeneric lineage back to the late Eocene, western Japan (Uemura 2002; see also Denk & Grimm 2009).

Fig. 7.

Pie charts showing the relative abundance of taxonomically relevant and generally shared 5S-IGS sequence types. Note that each of the distinguished sequence types represents a genetic-phylogenetic lineage, i.e. implies past speciation or intra- and inter-specific sorting or mixing processes.

Revision of Fagus in western Eurasia

Fagus sylvatica L., Sp. Pl. 2: 998. 1753 ≡ Castanea fagus Scop., Fl. Carniol., ed. 2, 2: 242. 1771. – Lectotype (designated by Jonsell & Jarvis in Jarvis & al. 1993: 47): 1600–1625, Joachim Burser, Herb. Burser XXII: 92 (UPS [V-175662 Fig. 9]).

= Fagus sylvatica f. moesiaca K. Malý in Ascherson & Graebner, Syn. Mitteleur. Fl. 4: 438. 1911 ≡ Fagus moesiaca (K. Malý) Czeczott in Roczn. Polsk. Towarz. Dendrol. 5: 52. 1933. – Lectotype (designated here): Bulgaria, supra pagum Tvierdica [=Tвъpдицa, Tvrditsa] (inter Sliven et Elena), in monte Cumerna, c. 950 m, fruct., 1927, K. Michoff, det. H. Czeczott (WA [WA00000166143 Fig. 10]); isolectotype: WA [WA00000166142]).

See POWO (2023) for other heterotypic synonyms.

Molecular diagnosis — Lineage IV ITS variants, unspecific. The majority of 5S-IGS variants are specific; B Lineage variants largely surpass A Lineage variants in absolute abundance and number of unique sequence variants (Fig. 4; Cardoni & al. 2022). Within Lineage B, European B types (Western B2 variants) outnumber Original B types (exclusively comprising shared types); relatively rare Lineage A variants (probably more common in Pleistocene relict populations, Cardoni & al. 2022, unpublished data) with a preference for Western A type (a lineage shared exclusively with Fagus orientalis) over Shared A type (Fig. 6, 7). Potentially specific SNPs rare but present in CRC, LFY, and Jiang & al.'s (2022) loci P14, P34, P38, P50, F128, F286 and F289 (see F. orientalis); P38 shows a 7 nt-long deletion not detected in any other Fagus sample; P14 characterized by a notably distinct sister genotype of the F. grandifolia genotype: both types differ by six C↔T transitions from the basic type of F. subg. Fagus (Table 3). Distinct isoenzyme Gömöry & Paule (2010) and nuclear SSR profiles (Kurz & al. 2023 marker set at k>2; Fig. 2, 3; see also Budde & al. 2023). Lineage Vb plastomes (species-level plastid type V-Sy; J. Worth & al., work in progress).

Morphological description — Lamina shape rounded to ovate to elliptic to obovate, usually asymmetric, (20–)40–80(–120) mm long in western populations, (40–)60–120(–145) mm in eastern populations, leaf index [length of leaf/width of leaf) × 100] 150 in western populations, 170 in eastern populations; leaf petiole (2–)5–12(–16) mm long, peaks at values 5, 6 and 10; most frequent base/apex pairs obtuse base and acute apex and acute base and acute apex, in addition, cordate base with acuminate apex and cuneate base with acute apex occur; basal leaf margin entire, wavy (to dentate), apical margin (entire to) wavy to inconspicuously dentate or conspicuously dentate in shade leaves; number of secondary veins (5–)6–10(–12); secondary venation pseudocraspedodromous, semicraspedodromous to craspedodromous in shade leaves, brochidodromous to semicraspedodromous and pseudocraspedodromous in sun leaves; length of stomata (16–)19–29(–32) µm, mean 22.5 µm, subsidiary cells incomplete cyclocytic to cyclocytic or actinocytic, dispersed or in groups; cupule peduncle 4–26(–40) mm, mean value 12 mm, length of cupule 12–26(–29) mm, mean value 20 mm in western populations, (12–)16–28(–32) mm, mean value 22.4 mm in eastern populations, basal cupule appendages reddish-brown, narrow (bud scale homologous) and long woody spines with slender apex, apical appendages long woody spines, often twisted.

Fig. 8.

Multifurcating key to identify the species-affinity of a beech population based on the detected 5S-IGS variants.

Distribution — Europe (Albania, Austria, Belgium, Bosnia and Herzegovina, Bulgaria, Central European Russia [Kaliningrad Oblast], Croatia, Czech Republic, Denmark, France, Germany, Greece, Hungary, Italy, Liechtenstein, Luxembourg, Moldova, Montenegro, Netherlands, North Macedonia, SE Norway [Vest-/Ostføld], Poland, Romania, Serbia, Slovakia, Slovenia, N Spain, S Sweden, Switzerland, W Ukraine, S United Kingdom).

Evolutionary significance — The European beeches represent the sister species of Fagus orientalis (as defined below), the divergence between the sister species has been dated to the late Early Pleistocene by Gömöry & al. (2018), but may have deeper roots (Renner & al. 2016). It is noteworthy that the palaeobotanical record also points to an Early Pleistocene origin of F. sylvatica (Denk & al. 2022). Fagus sylvatica is probably the most recently evolved species within the western Eurasian beech lineage. The currently available molecular data fit with a budding-speciation type process, i.e. the first common ancestor(s) of F. sylvatica evolved from a F. orientalis population that got isolated from the main gene pool and underwent a substantial bottleneck before re-radiating into its modern-day range. Fagus sylvatica plastomes are nearly identical across the entire range of the species stretching from Spain, north and south of the Alps into eastern Europe and the Balkans. Increased genetic variation appears to be restricted to the Apennines (Central Italy; Cardoni & al. 2022) and northwestern Greece and the Rhodopes (Hatziskakis & al. 2009), where it may be para- or sympatric with F. orientalis.

Remarks on nomenclature — Velenovský (1898, 1902) described a new variety of Fagus sylvatica from the surroundings of the village Kozludža (today Suvorovo, Cyвopoвo) in northeastern Bulgaria and from Sliven, eastern Bulgaria, as F. sylvatica var. macrophylla. The name macrophylla had earlier been used by Candolle (1868) and hence cannot be used for this taxon. Based on Velenovský's concept of F. sylvatica var. macrophylla, Maly in Ascherson & Graebner (1911) established the form F. sylvatica forma moesiaca. Later, Czeczott (1933) erected the new species F. moesiaca (Maly) Czeczott and in the protologue cited a number of herbarium vouchers. We could not find these vouchers in any herbarium collection until 2024, when Dr Maja Graniszewska, curator of the herbarium of the University of Warsaw, was able to locate most of the herbarium sheets in Hanna Czeczott's archive material at the University of Warsaw, did some cleaning and repairing to them, and provided us with high-resolution scans.

Fig. 9.

Lectotype of Fagus sylvatica (herbarium UPS-BURSER). – Image reproduced under the terms of the Creative Commons Attribution License (CC BY 4.0  https://creativecommons.org/licenses/by/4.0/), Museum of Evolution, Uppsala University.

Among the syntypes, one collection is from the vicinity of Sliven (K. Michoff s.n. WA00000166143, WA00000166142), and a lectotype and isolectotype were chosen from this collection.

Further remarks — From the surroundings of Tran (Tpън), Pernik Province, western Bulgaria, populations with large cupules (30–35 mm) were reported and ascribed to Fagus moesiaca var. borzae Domin (Jordanov & Kuzmanov 1966). All currently available molecular and morphological data suggest that the eastern European entity F. moesiaca should be included within F. sylvatica (Gömöry & al. 2018). In contrast, POWO (2023) treats F. moesiaca as a synonym of F. ×taurica Popl. (F. orientalis × F. sylvatica), a name that was originally used only for the Crimean beeches. There is so far little data on the Crimean populations. According to the isoenzyme data of Gömöry & Paule (2010), the Crimean F. ×taurica falls genetically within the overall variation of F. sylvatica as well. More recently, Gömöry & al. (2018) found F. ×taurica to have originated from relatively recent Middle Pleistocene contact between Caucasian beech populations and F. sylvatica s.str. Therefore, F. moesiaca should not be synonymized with F. ×taurica.

Representative specimens — E. Bourgeau 692 (P [P06857192, P06857207, P06858360]); J.-B. Mougeot s.n. (P [P01035946]); P. Jovet s.n. (P [P00504675]); H. Bouby 493 (P [P06850962]); C. Hering s.n. (P [P06858357]); K. Domin 284c (PRC [PRC454888 type of Fagus moesiaca (Maly) Domin var. borzae Domin]); J. Madalski 42 (P [P06853549]). — Syntypes of F. moesiaca: P. Černjavski s.n. (WA [WA00000166146, WA00000166147]); H. Czeczott s.n. (WA [WA00000166149]); H. Czeczott s.n. (WA [WA00000166150, WA00000166151]); E. Reimesch (WA [WA00000166152, WA00000166153]). — See Shen (1992) for more representative records.

Fagus orientalis Lipsky in Trudy Imp. S.-Peterburgsk. Bot. Sada 14: 300. 1898 ≡ Fagus sylvatica subsp. orientalis (Lipsky) Greuter & Burdet in Willdenowia 11: 279. 1981. – Lectotype (designated by Yaltırık 1982: 658): Turkey, Iter orientale, Paphlagonia, Vilayet Kastambuli [Kastamonu province], Kure-Nahas [Küre district], in sylvis ad Topschi-Chan, 41°48′21.528″N, 33°42′36.972″E, 9 Sep 1892, P. Sintenis 5113 (LE [LE00011314]; isolectotypes: A [A00033880], G [G00358034-36] [Fig. 11], P [P06812072, P06812074], US [NMNH-00409670 = old no. US 2501956]).

Molecular diagnosis — Lineage IV ITS variants, may include specific variants (cf. Denk & al. 2002) but better sampled data would be needed. Lineage B variants more abundant than Lineage A variants, the latter more frequent and more diverse than in Fagus sylvatica; Lineage A variants either shared exclusively or highly similar to F. sylvatica variants (dominant Western A type), or representing types ancestral within the western Eurasian lineage (Cardoni & al. 2022; Shared A in Fig. 6, 7); Lineage B variants of the European B lineage subdominant and either related to, or occasionally shared with F. sylvatica (Western B2 type); Original B lineage much more diverse than in F. sylvatica comprising sequentially distinct types: Shared B1 and related types found across all western Eurasian beeches, Western B1 types exclusively shared with F. hohenackeriana p.p., other evolved (sequentially distinct) Caucasian types (Hohenackeriana B1a, B1b, B3) rare to very rare but present while absent in F. sylvatica (Fig. 6, 7; Cardoni & al. 2022). Currently no CRC data, and a single LFY accession, differing by a unique T-dominated length-polymorphic sequence motif at pos. 737–764 (supplementary content, file Genotypification.xlsx, sheet LFY LP-patterns). The individual included in Jiang & al. (2022) differs consistently by 26 point mutations from the F. sylvatica samples in the nuclear loci P14 (13), P34 (1), P38 (4), P50 (1), P97 (1), F202 (1) and F289 (5). Distinct isoenzyme (Gömöry & Paule 2010) and nuclear SSR profiles (Kurz & al. 2023, at k=3), the latter involving a west-east gradient (k=2–6, Fig. 3). Lineage V plastomes, subtype yet to be determined.

Morphological description — Lamina shape elliptic to obovate, usually symmetric, (30–)50–120(–170) mm long, leaf index 196; leaf petiole (2–)3–10(–15) mm long; most frequent base/apex pairs “obtuse base and acute or acuminate apex” and “acute base and acuminate apex”, sun leaves with acute base and apex; leaf margin entire or with blunt triangular or sharp teeth (shade leaves); number of secondary veins (5–)7–12(–15); secondary venation pseudocraspedodromous, semicraspedodromous to craspedodromous; length of stomata (13–)18–25(–33) µm, mean 22.5 µm, subsidiary cells incomplete cyclocytic to cyclocytic or actinocytic, dispersed or in groups; cupule peduncle (5–)12–14(–75) mm, mean value 25 mm, length of cupule (10–)18–28(–45) mm, mean value 22.5 mm, basal cupule appendages leaf-like, wintergreen (not turning brown in autumn) or summergreen (turning brown during cupule development), spathulate or petiolate, oblong to elliptic in shape, venation dichotomous (in spathulate leaflets) or brochidodromous, area of leaflets decreasing at higher altitudes, apical appendages woody, spine-like.

Fig. 10.

Lectotype of Fagus moesiaca K. Malý) Czeczott (herbarium WA). – Image reproduced with permission.

Distribution — NE Greece (Thrace), SE Bulgaria, W and N Turkey, S Turkey (Kahramanmaraş, Hatay, Osmanye, ?Adana, Mersin).

Evolutionary significance — Sister species of Fagus sylvatica (see above), geographically and genetically (Kurz & al. 2023, supplement fig. 4) forming the bridge between the eastern species and the beeches of Europe. Morphologically, individuals in lowlands and at mid-elevations are characterized by conspicuous, green, spathulate, leaf-like appendages on the lower parts of the cupule and long cupule peduncles (Fig. 12A, B; Table 4). Similar appendages are also found in the Japanese species F. crenata (F. subg. Fagus) and in the East Asian (China, S Korea) species F. engleriana and F. multinervis (F. subg. Englerianae). From the current still limited data, it can be expected that F. orientalis is genetically richer than its western sister species despite its much smaller overall range and population size. Furthermore, the species appears to be generally closer to the common ancestor of F. sylvatica-orientalis, demonstrated by a higher amount of shared genetic types and stronger morphological affinities to fossil members of its lineage (F. castaneifolia, F. haidingeri) but also, in contrast to F. sylvatica, to the fossil-species F. gussonii (Denk & al. 2002). Fagus gussonii is a Miocene fossil-species of ambiguous phylogenetic affinities and hypothetical vector for past trans-Atlantic gene flow (Cardoni & al. 2022; Schulze & Grimm 2022). The SSR data clustering results of Kurz & al. 2023 (k≥3) indicate that the former range of this species may have been much larger, potentially including the disjunct Nur Mountains populations in the Hatay province, southeastern Turkey (Fig. 2). Both the isoenzyme and SSR clustering patterns are possibly affected by (sub)recent gene flow with the Caucasian beech, F. hohenackeriana (in agreement with occasionally found eastern 5S-IGS variants), between their respective ancestors, or incomplete lineage sorting within the ancestor(s) of F. orientalis(-sylvatica) and F. hohenackeriana. Potential hybrid or contact zones include or have included the Nur Mountains and, more importantly, the Parhar Mountains (western extension of the Pontic Mountains) in the hinterland of the Turkish Black Sea coast east of Zonguldak and into southwestern Georgia (Kurz & al. 2023). More in-depth population-level studies are needed to discern to which degree the higher genetic affinity of F. orientalis with its eastern cousin, F. hohenackeriana, than found in its western sister F. sylvatica, is due to (ongoing) gene flow or a generally lower genetic drift (e.g. because of fewer Pleistocene bottleneck events).

Further remarks — This species is genetically severely understudied in its core range, with most research having focussed on the Bulgarian-Romanian Fagus moesiaca as a putative hybrid or intermediate form between F. sylvatica and F. orientalis. Morphologically, the green, stalked leaflets persisting on the cupule (Table 4) appear to be a most conserved trait, never seen in suggested hybrids outside the known range of the species as shown in Fig. 2. Ongoing research on the Greek side of the eastern Rhodopes has nonetheless revealed mixed stands (individuals lacking or showing green cupule leaflets) of F. sylvatica and F. orientalis that show private 5S-IGS variants in addition to those shared with F. sylvatica or F. orientalis and may be transitional between the sister species (A. Papageorgiou & al., work in progress).

Representative specimens — Bulgaria: P. Frost-Olsen 1151 (P [P06812193, P06853547]). — Turkey: J. Bornmüller & F. Bornmüller (E [E00401534]); G. D. Sag 887A (P [P00043490]); P. H. Davis, M. Coode & F. Yaltırık D. 37622 (E [E00401528]); P. H. Davis 18492 (E [E00401511]); B. Balansa 1141 (P [P06812090], US [NMNH-03400246]); J. Manissadjian 369b (P [P06812088 two specimens, one typical F. orientalis, another with transitional cupule appendages, see below]); Det. I. V. Palibin (P [P06812092]); Nur Mountains: P. H. Davis 16398 (E [E00401507]). — Transitional morphologies to Fagus hohenackeriana: Ordu: P. H. Davis & O. Polunin 24935 (E [E00401509]); Amasya: J. Manissadjian 369b (P [P06812041]); Trabzon: A. Stainton 8408 (E [E00401551]); Nur Mountains: Fannie P. A. Shepard 10308853 (US [NMNH-03400252]). — GBIF entries with photographs verify transitional forms to F. hohenackeriana from the provinces of Ordu and Giresun; very rarely transitional forms occur further west (Bolu): C. Aedo 6175 (B [B 10 1167842], MA [MA688421], PRN [PRN2022-024]).

Fagus hohenackeriana Palib. in Bull. Herb. Boissier, sér. 2, 8: 378. 1908, as “hohenackerana”. – Lectotype (designated here): Azerbaijan, Lesser Caucasus, 1838–1833, R. F. Hohenacker s.n. (G [G00358037_a] [Fig. 13]; isolectotypes: E [E00326761], G [G00358037, G00358037_b], US [NMNH-00409517-000001]).

Molecular diagnosis — ITS variants belonging to Lineage IV, preliminary data indicate Lineage IV ITS variants not shared with Fagus sylvatica and F. orientalis. Most diverse and heterogenous 5S-IGS pool of all species analysed so far (Fig. 5), A Lineage variants can be more abundant than B Lineage variants (Lesser Caucasus and eastern Georgian sample) or vice versa (Greater Caucasus sample, Fig. 7); no European B variants. A Lineage variants (co-)dominated either by private to this species European A types (Hohenackeriana A1, sister lineage of Western A type) and/or the unspecific variants of the Shared A type, additional rare A-lineage types shared with the eastern sibling F. caspica; B Lineage variants sequentially and type-wise diverse (Fig. 6), characteristically including shared (Shared B, Western B) and (near-) exclusive (Hohenackeriana B1a, B1b, B2 and B3) types with shifting abundances between samples, a genotypic feature not found in any of the other species. Distinct isoenzyme (Gömöry & Paule 2010) and nuclear SSR profiles (Kurz & al. 2023), separating F. hohenackeriana and F. caspica from F. sylvatica-orientalis at k=2 and k=3. Fagus hohenackeriana and F. caspica differentiated at higher k and in the densely sampled nuclear SSR data of Sękiewicz & al. (2022; mapped in Fig. 3). [No other nuclear data available.] Lineage V plastomes, subtype not yet determined.

Fig. 11.

Isolectotype of Fagus orientalis Lipsky (herbarium G). – Image © Conservatoire et Jardin botaniques de la Ville de Genève.

Morphological description — Lamina shape elliptic to obovate, usually symmetric, (60–)80–140(–200) mm, leaf index 187; leaf petiole (1–)2–9(–13) mm long; most frequent base/apex pairs “oblong very-base and blunt acute or blunt acuminate apex” and “slightly cordate or nearly oblong very-base and attenuate apex” chiefly on vegetative twigs, “acute base and apex” on fruiting twigs and sun leaves; leaf margin entire or with blunt triangular teeth (shade leaves); number of secondary veins (6–)7–12(–16); secondary venation pseudocraspedodromous, semicraspedodromous to craspedodromous in shade leaves; length of stomata (16–)20–26(–30) µm, mean 23.5 µm, subsidiary cells incomplete cyclocytic to cyclocytic or actinocytic, dispersed or in groups; cupule peduncle 5–38 mm, mean value 19 mm, length of cupule (6–)15–25(–38) mm, basal cupule appendages (1) parallelodromous, membranous brownish scales similar to bud scales, usually densely spaced (Fig. 12C), (2) small sessile leaflets initially green but soon turning brown, spathulate to lanceolate, with dichotomous venation or (3) woody spine-like appendages similar to apical ones, apical appendages woody, spine-like.

Distribution — NE Turkey, Lesser Caucasus (Georgia, Armenia, Azerbaijan), Transcaucasus (Georgia, Azerbaijan), North Caucasus (Russia).

Evolutionary significance — The isoenzyme data of Gömöry & Paule (2010) and the SSR data clustering of Kurz & al. (2023) indicate that the split between Fagus hohenackeriana (+ F. caspica) and F. orientalis predates the split between the latter and its European sister, F. sylvatica. This is corroborated by the 5S-IGS lineages and frequent or (very) rare types shared exclusively with either F. orientalis (sister types Hohenackeriana B1b and Western B1, both sequentially distinct derivates of the commonly shared Original B types), F. sylvatica (Hohenackeriana A1, sister type of Western A) or both (Western A). The available high-resolution genetic data also indicate that F. hohenackeriana has been less isolated from its western cousins than the easternmost F. caspica and differs from the latter by an increased intra-species genetic diversity. Sękiewicz & al. (2022) found a genetic cline between the southern, western and central populations of F. hohenackeriana (eastern Pontic Mountains, Lesser Caucasus, western and central High Caucasus in Georgia) and the eastern High Caucasus populations in northeastern Azerbaijan. This finding correlates with the differential composition of the 5S-IGS pools of the three samples included here (Fig. 6) which sets the sample from the Greater Caucasus (Racha) apart from those of the Lesser Caucasus (Borjomi + Bakuriani) and eastern Georgia (Lagodekhi): in the Racha population, hohenackeriana-specific 5S-IGS B Lineage variants co-occur with variants of a lineage also found in the western F. orientalis but very rare in the other two populations of F. hohenackeriana. In contrast, the latter share some types with their eastern sibling, F. caspica, types not found in F. orientalis or the Racha population. At this point, the genetics would fit with two evolutionary hypotheses about the origin of F. hohenackeriana. It may represent the eastern sister species of F. sylvatica + F. orientalis, sharing a common ancestor with the precursor of F. orientalis (+ later evolved F. sylvatica), spreading across Asia Minor and the Caucasus before Anatolia and the entire eastern Mediterranean region dried out and became more continental. Or, it is the sister species of F. caspica and both species evolved from the Pontic-Hyrcanian populations of the fossil-species F. haidingeri in contrast to F. sylvatica-orientalis that evolved from the Euro-Mediterranean populations of F. haidingeri, with the fossil-species F. gussonii being the second donor. Under both scenarios, the notably high 5S-IGS heterogeneity of F. hohenackeriana could be explained by genetic legacy from further species/populations that thrived north/northeast of the modern F. hohenackeriana before the Pleistocene (north of the Paratethys and its remnant, the Caspian Sea) as well as ongoing speciation processes in the Caucasus and adjacent areas characterized by a strong topographic relief and geographic vicinity of strongly differing niches (cf. Denk & al. 2001, for the ecology and biocenoses of beech forests in Georgia).

Remarks on nomenclature — There has been some nomenclatural confusion surrounding the Caucasian-Hyrcanian beeches. Based on the initial work by Hohenacker (1833, 1838) and the taxonomic treatments by Candolle (1868) and Palibin (1908), Shen (1992) correctly considered Fagus hohenackeriana Palib. (F. sylvatica subsp. hohenackeriana sensu Shen; Caucasus-Hyrcanian region) different from F. sylvatica subsp. orientalis. Since no types had been cited in previous works, but both Candolle and Palibin had referred to material collected by Hohenacker, Shen chose a lectotype for F. hohenackeriana from herbarium G collected from “Azerbaijan-Talysh” Mountains (Shen 1992: p. 166). He further listed isolectotypes from G, LE, and US. In doing so, Shen (1992) confused different collections of F. hohenackeriana, namely material collected earlier from the Lesser Caucasus s.l. (Azerbaijan part of Karabakh Mountains; Hohenacker 1833; including the isotypes selected by Shen) and later from the Talysh Mountains of Azerbaijan (Hohenacker 1838). These collections represent geographically and genetically distinct populations and hence they cannot be lectotypes of a single species (cf. Sękiewicz & al. 2022, fig. 3 AZ_01, Lesser Caucasus, versus HZ_01, HZ_02, Talysh).

Fig. 12.

Cupule morphology of Fagus, western and northern Turkey to Iran. – A: F. orientalis, western Turkey, Kütahya, Simav, 1700 m a.s.l., 16 Jun 1965, M. J. E. Coode & B. M. G. Jones 2668 (E); B: F. orientalis, northern Turkey, Bolu, Abant Gölü, 1400 m a.s.l., T. Denk & G. W. Grimm 2006097 (S); C: F. hohenackeriana, eastern Georgia, Caucasus, Lagodekhi, 19 Jun 1971, E. E. Gogina (E); D: F. caspica, northern Iran, Gilan, between Khalkal and Asalem, 1200–1950 m, 15 Jul 1971, J. Lamond 5136 (E). – Prominent leaf-like appendages and long peduncles are most common in northern Turkish populations and differ markedly from Caucasian and northern Iranian ones. Caucasian populations commonly have bud scale-like or thread-like basal appendages that turn brown early in the summer; Iranian populations have reduced thread-like basal cupule appendages reminiscent of European ones. – Scale bars: A–D = 1 cm.

Fig. 13.

Lectotype of Fagus hohenackeriana Palib. (herbarium G). – Image © Conservatoire et Jardin botaniques de la Ville de Genève.

Fig. 14.

Holotype of Fagus caspica Denk & G. W. Grimm (herbarium W [W20180003068  https://www.jacq.org/detail.php?ID=1383327]). – Image reproduced under the terms of the Creative Commons Attribution License (CC BY 4.0  https://creativecommons.org/licenses/by/4.0/).

Fig. 15.

Close-up of the tree produced by Li & al. (2023), with each tip linked to the according population in their map. Note the scattering of the alleged Fagus chienii (likely mislabelled F. pashanica) samples across the entire hayatae-pashanica subtree.

Hohenacker, a Swiss missionary based in Şuşa (German: Schuscha) in Nagorno-Karabach, started collecting plant specimens for the Esslinger Reiseverein (“Esslinger Travel Society”) in the early 1830s (Wörz 2007). These plant specimens were sent to Germany and distributed among the members of the travel society. A first parcel of dried plant specimens, collected by Hohenacker between 1830 and 1833, contained plants from the environs of Şuşa and the mountains of Nagorno-Karabach (see Hohenacker 1833; Hochstetter & Steudel 1834). These plants were accompanied by labels with the locality information “Caucasus” (isolectotypes of Fagus hohenackeriana at herbarium Genève, G00358037; Edinburgh, E00326761; Smithsonian, US00409517).

During a subsequent collecting trip in the summer and autumn 1834, Hohenacker collected plants from the surroundings of Lankaran and explored the montane regions of Zuvand (“district Suwant”) and Dirig/Dırığ (“district Drych”). He collected Fagus from the environs of the villages Cayrud (“Tschaioru”) and Veri (“Weri”; Hohenacker 1838). In October 1835, Hohenacker collected Fagus in southern Talysh, close to the border of Iran. In the publication arising from his collection trip (Hohenacker 1838), he reported F. sylvatica var. macrophylla occurring “in sylvis montium Talysch prope Lenkoran, Drych, Suwant, Astara” (p. 259). The material collected during this expedition does not have “Caucasus” on the labels, but, for example “in forests in the surroundings of Lenkoran” (e.g. Hohenacker 2229 in herbarium Paris, barcode P06812042).

Hence, we here clarify the origin of the lectotype of Fagus hohenackeriana and refer Fagus populations occurring from northeastern Turkey to the Greater and Lesser Caucasus to this species. In contrast, we refer the populations originating from the Talysh Mountains and the Hyrcanian region as F. caspica (see below).

Further remarks — So far, only the spacers of the nuclear-encoded ribosomal DNA have been sequenced (Denk & al. 2002, 2005; this study). The Caucasian populations are not covered in Jiang & al. (2022) nor in the upcoming study of Worth and co-workers. Armenian and Georgian populations have been included in the study of Paffetti & al. (2007) but their data had not the necessary quality or resolution to identify novel plastid haplotypes within the western Eurasian plastid lineage (Lineage V). Based on our experience with other Caucasian trees (Acer, Quercus), we expect that in-depth nuclear-genetic analyses will reveal further diagnostic traits. Screening of the nuclear loci of Jiang & al. (2022) comprising alleles reflecting the trans-Atlantic link (introgression: genes P14, P21, P54, and F289; cf. Cardoni & al. 2022) may help to decide between the two hypotheses outlined above. By extending the sample of Sękiewicz & al. (2022) to adjacent areas (Crimean, northeastern Turkish and additional Armenian and Iranian populations), it may be possible to further explore the nature of the genetic gradient found in Fagus hohenackeriana and, potentially, reveal ongoing speciation processes between the western (+ Pontic Mountains) and eastern Caucasus.

Representative specimens — Turkey: P. H. Davis & I. Hedge D32347 (E [E00401564]); M. Tong 504 (E [E00401561]); P. H. Davis & J. Dodds 21380 (E [E00401562]); P. Sintenis 1609 (P [P06812039]); ENET 33 (E [E00318857]); [?] 828 (G [G00754880]). — Lesser Caucasus: T. Denk 977154 (US [NMNH-03400173]); J. C. Solomon 20783 (US [NMNH-03470668]); T. Denk 977229 (U [U0251466]). — Trans-Caucasus: T. Denk 896127 (BR [BR0000030519022]); V. Vašák s.n. (BR [BR0000030518865]); T. Denk 977011 (BR [BR0000030520585]); E. E. Gogina 47 (BR [BR0000030519084]); E. E. Gogina s.n. (E [E00401536]); S. Kuthatheladze & I. Mandenova (E [E00401545]); J. Reveal 8715 (P [P06851162]); T. Denk 896179 (P [P06812068]); T. Denk 8965 (P [P06812069]); T. Denk 977057 (P [P06851975]); T. Denk 977 (P [P06851976]); T. Denk 977222 (P [P06851978]); E. Gabrielian 12766 (E [E00401548]); V. Manakyan s.n. (E [E00401555]); P. Smirnow 312 (MW [MW0660524]); M. Barkworth & al. s.n. (NY [NY03476707]); L. N. Cilikina s.n. (MW [MW0660529]); L. N. Cilikina s.n. (MW [MW0660537]); T. Alexeenko 84b (DR [DR061665]). — North Caucasus: B. Marcowicz s.n. (P [PI031307]); B. Marcowicz s.n. (E [E00401556]); B. Marcowicz s.n. (E [E00401557]).

Fagus caspica Denk & G. W. Grimm, sp. nov.

Holotype: Iran, Gilan province, Deylaman to Siahkal, 37°02′29″N, 49°54′37″E, 1350 m a.s.l., 7 Jun 2011, J. Noroozi 2349 (W [W20180003068 Fig. 14  https://www.jacq.org/detail.php?ID=1383327]; isotype: W [W20180003067  https://www.jacq.org/detail.php?ID=1383326]).

Molecular diagnosis — ITS variants belonging to Lineage IV, possibly specific (based on limited individual-level and old sequence data). 5S-IGS variants predominately specific, typically not shared with any other western Eurasian species but least evolved, i.e. relatively close to the putative ancestral sequence variants; B lineage variants slightly more abundant or nearly as abundant as A lineage variants, and equally diverse (Fig. 5; Cardoni & al. 2022), no European A and European B variants; A Lineage variants predominately of the Shared A type, types shared with Fagus hohenackeriana p.p. rare (Eastern A) to very rare (Hohenackeriana A2); B Lineage variants dominated by the (originally specific?), derived Caspica B1 type, and the underived, cross-species shared Ancestral B1/Shared B1 types (Fig. 6, 7), the high abundance of the latter is a genotypic characteristic of F. caspica and to a lesser degree, F. orientalis; no Western B1/Hohenackeriana B1b types. Very distinct isoenzyme (Gömöry & Paule 2010) and homogenous nuclear SSR profiles (Kurz & al. 2023), separating F. caspica from its putative sister species F. hohenackeriana at k=4 (see also Sękiewicz & al. 2022, k=2 vs k=3; mapped in Fig. 3). Lineage Va plastomes (species-level plastid type V-EO). CRC distinct from Fagus sylvatica, differing in at least five alignment patterns, three species-consistent SNPs (pos. 800, 85, 1674 in reference alignment) and two to three other (cf. supplementary content, file Genotypification.xlsx, sheet CRC LP-patterns); A-dominated motif at position 830–859 in CRC near-exclusively composed of A (terminating on G, GG, TG in all other species). No LFY or low-copy nuclear loci data so far. Lineage Va plastomes, differing by 564–642 SNPs from the plastomes of their westernmost cousin, F. sylvatica (Lineage Vb plastomes), about the same level of difference as found in East Asian individuals carrying the same plastome lineage (Japanese Lineage II median: 637 SNPs; East Asian Lineage IV median: 462 SNPs) and about half of the maximum difference recorded so far (1198 SNPs between Lineage I plastome of F. grandifolia from Michigan and a Lineage IV F. crenata plastome; supplementary content, file Genotypification.xlsx, sheet PlstmDissim).

Morphological description — Lamina shape ovate to elliptic, usually asymmetric, (60–)80–120(–140) mm long, leaf index 188; leaf petiole (2–)7–12 mm long; most frequent base/apex pairs “cordate asymmetric or symmetric base and attenuate apex”; basal leaf margin entire to wavy, sometimes with blunt teeth, apical margin commonly with prominent teeth; teeth (1) with long, convex, concave or straight basal side and short, steep apical side, or (2) small, pronounced teeth with slightly convex margin between two consecutive teeth; number of secondary veins (7–)8–14(–16); secondary venation brochidodromous to pseudocraspedodromous basally, semicraspedodromous to craspedodromous apically; length of stomata (16–)20–26(–30) µm, mean 23 µm, subsidiary cells incomplete cyclocytic to actinocytic, with transitions to anomocytic; cupule peduncle (5–)9–25(–40) mm, mean value 17 mm, length of cupule 5–29 mm, mean value 17 mm, basal cupule appendages parallelodromous, membranous, reddish-brown, narrow, similar to bud scales or thread-like (Fig. 12:D), or narrow spathulate brownish leaflets with obtuse, forked or acute apex, apical appendages woody spine-like, sometimes forming clusters.

Distribution — SE Azerbaijan (Talysh), N Iran.

Evolutionary significance — Genetically, the Iranian populations are the leftover of the initial speciation processes within the precursor(s) of all western Eurasian beeches, as reflected by their many private 5S-IGS variants. The difference between their plastomes and those of Fagus sylvatica is double to triple as high as observed between East Asian species carrying Lineage IV plastomes, which includes the plastomes of all Chinese and Taiwanese beeches and the southeastern populations of the Japanese beeches (cf. Worth & al. 2021). Morphologically, the numerous, densely spaced secondary veins, the usually distinctly serrate leaf margin, and the elongate leaf apex resemble fossils from Middle Miocene strata of Austria and Russia (Zetter 1984; Yakubovskaya 1975). Denk (1999a), furthermore, pointed out leaf morphological similarities with North American populations of F. grandifolia, a likely symplesiomorphic pattern.

Etymology — The species name refers to the distribution of the species along the southern shores of the Caspian Sea in Azerbaijan and northern Iran.

Representative specimens — R. Hohenacker 2229 (P [P06812042]); T. Alexeenko (MW [MW0660538]); A. Ghorbani & A. Pirani 1196 (TMRC [TMRC0001196]); K. H. Rechinger 20006 (US [NMNH-03400239]); D. Lyskov & T. Krutenko (MW [MW0754378]); A. A. von Bunge (G [G00754877]); J. Lamond 2966 (E [E00401560]); J. Lamond 5136 (E [E00401553]); P. M. R. Aucher-Eloy 5325 (P [P06812078]); D. Walton 242 (E [E00400275]); D. Walton 243 (E [E00401554]); H. de Hell (P [P06812075]); I. V. Palibin (K [K00832763]).

Sampling gaps

Currently there is only limited molecular data on Turkish, Armenian, Georgian and Ciscaucasian beeches, and no material has been studied from southeastern Azerbaijan. Filling these geographic gaps by genetic data of any kind will greatly advance our understanding of speciation processes within the western Eurasian beeches and provide a model data set for discriminating differentiation patterns triggered by incomplete lineage sorting versus ancient and recent reticulation (repeated and potentially ongoing introgression, hybrid speciation and hybridization in current or past contact zones).

Available wide-sampled plastid barcode markers (mainly from the study of Paffetti & al. 2006) are uninformative regarding the substantial difference between the plastome lineage carried by the easternmost species, Fagus caspica (Lineage Va plastome) and that of F. sylvatica (highly similar to near-identical Lineage Vb plastomes found across the entire range as far as studied; cf. supplementary content, file Genotypification.xlsx). Completely sequenced plastomes are needed from key areas within the distribution area of the three Oriental beech species: F. orientalis from the eastern Rhodopes, northwestern Anatolia, and the western Pontic Mountains (N. Anatolia); F. hohenackeriana from the Greater and Lesser Caucasus and eastern Pontic Mountains (SW Georgia).

Nuclear sequence data are limited as well, ITS data covering all main distribution centres (Denk & al. 2002, 2005; Grimm & al. 2007) do not clearly differentiate between most Eurasian species of F. subg. Fagus. Jiang & al.'s (2022) 28 nuclear loci data set covered two individuals of the genetically ambiguous (discussed below) southeastern Turkish beeches and one individual from the genetically unambiguous core area of F. orientalis. Further nuclear sequences (in particular 5S-IGS) and high-resolution nuclear SSR data are needed with focus on the Pontic Mountain chain of northern Anatolia extending into southwestern Georgia, also with respect to altitudinal morphological gradients and whether they show any correlation with the already established genetic gradients connecting F. orientalis and F. hohenackeriana (Fig. 2, 3).

Another special focus needs to be on the Rhodopes as potentially still active contact zone between Fagus sylvatica and F. orientalis and on F. sylvatica along the Apennine Mountains. The assumed relict population sampled for the studies of Cardoni & al. (2022) and Worth & al. (2021, work in progress) is genetically closer to the Greek beeches (both F. sylvatica and F. orientalis) than those from Central Europe and the Alpine region; and carries the most distinct plastome detected within F. sylvatica so far (differing by 47–95 SNPs from other F. sylvatica plastomes). In contrast, plastomes of the (northern) Spanish and Romanian F. sylvatica (addressed as “F. moesiaca”) are near-identical to each other and those of Central Germany: total of 2–12 SNPs (≤ 1 ‰ of the maximum intrageneric plastid divergence) in the data generated by Ulaszewski & al. (2021), gene bank accession numbers MW566772, MW531753, MW566779, NC_041437.1 and MW566781).

Table 3.

Informative nucleotide patterns for infrageneric groups in Fagus in nuclear gene region P14 (data of Jiang & al. 2019; after Cardoni & al. 2022). Putatively apomorphic, lineage-diagnostic nucleotides in bold font; plesiomorphic, ancestral within genus Fagus nucleotides in normal font. Arrows indicate ongoing replacement of the putative ancestral nucleotide.

1 In case of heterozygous individuals, the haplotypes (parental alleles) can be traced by site ambiguity in line with the sequence variants obtained for (near-)homozygous individuals (details provided in Cardoni & al. 2022, data S5).

2 The P14 haplo- and genotypes of F. subg. Englerianae and F. subg. Fagus are further distinguished by four perfectly sorted SNPs at alignment positions 106, 159, 195 and 492.

3 Of the three Hidalgo individuals included by Jiang & al. (2021), individual 554 showed a sequence identical to the southern haplotype of F. grandifolia, individual 540 was homozygotic for the Pacific gene lineage, and individual 550 a perfect blend of the two others.

Outside western Eurasia, the beeches of North America need to be studied across their entire range with a particular focus on the Mexican (relict?) populations, Fagus mexicana, ideally by a combination of 5S-IGS sequence data and nuclear SSR data. The mutation patterns in the 28-nuclear loci data of Jiang & al. (2022) could be indicative of a budding-type speciation process similar to F. orientalis-sylvatica, with F. mexicana having retained a more heterogenous, ancestral genetic pool than F. grandifolia. A further issue is whether the plastid homogeneity observed within F. grandifolia encompasses that of F. mexicana (little plastid sequence data available) and is paralleled by nuclear homogeneity despite the large area covered. In China, more broad-sampled, community- and population-based combined plastid and nuclear studies are needed to assess the extent of past or recent hybridization between F. longipetiolata, F. lucida and F. pashanica. Based on the currently available data, F. pashanica, a cryptic sister species of the insular Taiwanese relict F. hayatae, is the valid species epithet for all continental beeches traditionally treated as F. hayatae. However, F. hayatae genotypic characteristics may indeed be shared by the so far unstudied south-easternmost Chinese populations of the sister pair (e.g. Long Xi Shan, Fujian). Based on the available and upcoming complete plastome data, it is clear that the Chinese-Taiwanese species cannot compete with their Japanese counterparts regarding overall divergence and geographic-phylogenetic structuring (Worth & el. 2021; work in progress) but our understanding of the decoupling of nuclear-morphological and plastid differentiation between these species is in its infancy.

Finally, broad-sampled 5S-IGS and nuclear SSR data across both Japanese species may help understanding within-species differentiation and its correlation with the extraordinary plastid divergence and sharing exhibited by the Japanese beeches; and to which degree these highly complex patterns are linked to climate (see also Worth & al. 2023).

Wild card taxa in western Eurasia

For some populations with a more or less isolated distribution, additional data would be needed to clarify their outpost, hybrid or relict status. Examples are the central southern Turkish Nur Mountains population and adjacent populations across Kahramanmaraş, Osmaniye, Adana and Mersin and the Crimean beech, which may be the primordial Fagus sylvatica or a hybrid between two or three lineages.

In contrast, Fagus moesiaca, usually considered a hybrid between F. sylvatica and F. orientalis (s.str.) does not show evidence of notable genetic differentiation from F. sylvatica or increased similarity to F. orientalis and probably is a lowland eco-morphotype of F. sylvatica (Gömöry & Paule 2010).

Table 4.

Differentiating morphological characters for the four western Eurasian Fagus species.

Future work: species outside western Eurasia

Cryptic speciation in Fagus subg. Englerianae

The beeches of Ulleung Do, a relatively young volcanic island in the Sea of Japan, 150 km off the east coast of South Korea, are traditionally treated as subspecies or variants of the Chinese Fagus engleriana or the Japanese F. japonica. Morphologically, they are indistinguishable from F. engleriana (Shen, 1992). Complete plastome data (Worth & al. 2021, in progress; cf. supplementary content, file Genotypication.xlxs, sheet PlstmDissim) and the 28 nuclear loci data compiled by Jiang & al. (2022; see also supplement to Cardoni & al. 2022) demonstrated that they not only are a disjunct population of F. engleriana (or F. japonica with a more engleriana-like phenotype) but represent a unique genetic resource warranting recognition as a species on its own, F. multinervis.

Pseudocryptic and/or cryptic speciation in East Asian and North American members of Fagus subg. Fagus

The holotype of Fagus hayatae is from the mountains of Taiwan, individuals with similar morphotypes can be found across southern and Central China, occasionally treated as a subspecies or variant pashanica (Shen 1992). In the nuclear loci data of Jiang & al. (2022), the continental (Chinese) individuals of F. hayatae s.l. are genetically as distinct from the insular one (Taiwanese) as the Turkish F. orientalis individuals are from the individuals representing their European sister F. sylvatica, or the Mexican individuals (subsp./var. mexicana) from their U.S. American individuals (F. grandifolia s.str.) Jiang & al. (2022) discuss the latter as argument to accept F. orientalis and F. mexicana at the species level. However, the authors did not mention at all the corresponding and remarkable finding for the East Asian species showing a tree as a main-text figure where the F. hayatae-pashanica clade is collapsed, thereby masking the notable divergence. Re-analysis of their data (Cardoni & al. 2022) revealed further that the genetic divergence between the continental and insular individuals corresponded to substantially different genotypic composition, with F. pashanica genotypes (involving substantial heterozygosity) being much closer to the other Chinese species but also the Japanese F. crenata than those of F. hayatae s.str. As far as studied, F. pashanica plastome haplotypes fall within a widely shared Lineage IV subclade; in contrast, those of F. hayatae are most distinct within Lineage IV and clearly isolated. Whether this is a case of pseudocryptic or cryptic speciation and whether F. hayatae (s.str.) is restricted to Taiwan cannot be decided at this point with the data at hand. It will require population-level genotypic and phenotypic investigations, most importantly including the southeasternmost populations of F. pashanica. A first relatively well-sampled study, trying to elucidate the genotypic affinities of F. chienii by recollecting beeches at the type locality of the latter species (Li & al. 2023, using a poorly informed subset of the nuclear loci of Jiang & al. 2022), did not have the necessary resolution, taxonomic depth and sample to settle this point but gives a first indication that F. pashanica may be genetically even more heterogenous than seen in the data of Jiang & al. (2022). Fagus hayatae may be the product of a budding-speciation event, a population/group of populations that evolved from a local F. pashanica population and got isolated from the main gene pool of the species, or of a precursor species of F. pashanica and other Chinese species. The genetic uniqueness of F. hayatae would have been enforced by inbreeding, while the continental F. pashanica could have been frequently homogenized and enriched by introgression and through contact with other beech species.

Pending further in-depth studies and more informative plastid data than the trnK/matK region used by MacLachlan & al. (2005, a single accession representing Fagus mexicana differing by a 10 nt-long duplication and two C↔A transversions in the downstream trnK intron from all F. grandifolia; supplementary content, file Genotypification.xlsx, sheet AddHTs_tKmK_LinI), we support the recognition of the disjunct Mexican beech populations, traditionally treated as subspecies or variant of F. grandifolia, as a distinct species: F. mexicana. This is based mainly on the mutation patterns seen in the 28 nuclear loci data of Jiang & al. (2022; see Cardoni & al. 2022) but also first CRC data (Oh & Manos 2008). Genotypically, F. mexicana differs from F. grandifolia s.str. by its generally less-evolved and more polymorphic sequences (higher level of heterozygosity) and the lack of derived mutation patterns shared by F. grandifolia s.str. and the western Eurasian species in several of Jiang & al.'s loci. These loci suggest a past introgression from North American into western Eurasian beeches, possibly in the Miocene (Cardoni & al. 2022; Schulze & Grimm 2022); that these patterns are missing in the Mexican population indicates that gene flow had already stopped or was already limited between the northeastern (F. grandifolia lineage) and the Mexican populations (F. mexicana lineage), when the F. grandifolia lineage came into contact with the western Eurasian lineage. In addition, heterozygous and homozygous patterns diagnostic for F. grandifolia s.str. in the so far sequenced nuclear marker regions involve a relatively high number of mutation patterns that did never reach F. mexicana, the most evolved F. grandifolia genotypes are more distinct from F. mexicana than what we can find in the same locus between commonly accepted and morphologically distinct species in East Asia; this may be analogous to the western Eurasian species complex. Fagus mexicana differs from F. grandifolia by its usually slender leaves with elongated attenuate apex. In addition, F. mexicana has a distinct ecology growing at elevations of 1400–2000 m a.s.l. in the bosque mesófilo de montaña along with several Cenozoic relicts (Liquidambar, tree ferns; Rzedowski 1983). This montane forest is situated above the tropical lowland forest and has a distinct subtropical character. Therefore, the two North American Fagus species represent a case of pseudocryptic speciation.