Epitypification of the Central African Cantharellusdensifolius and C.luteopunctatus allows for the recognition of two additional species

Abstract Cantharellusdensifolius and C.luteopunctatus are epitypified on the basis of recently collected specimens from the Central African rain forest that correspond in every way to their respective original descriptions. Sequences obtained from these new collections demonstrate that both epitypes represent distinct species that belong in different subclades of Cantharellussubg.Rubrinus. Previously, the name C.densifolius has been consistently misapplied to more or less similar species from the African woodland area, including C.densilamellatussp. nov. which is described here, In addition, C.tomentosoidessp. nov., a rain forest species that is easily confused with C.densifolius, is described.


Introduction
Tropical African Cantharellus species ("chanterelles") have been well-documented compared to other tropical regions. Nonetheless, there is a great need for sequence data to provide the foundation for unambiguous species concepts. This is due to the highly variable and potentially deceptive macromorphologies, compounded by the limited interspecific micromorphological variation among Cantharellus species (Buyck et al. 2014, 2016e, Olariaga et al. 2015. Despite this need for sequence data, Cantharellus has been difficult to work with from a molecular standpoint. Cantharellus ribosomal genes have high rates of molecular evolution (Moncalvo et al. 2006) and Cantharellus species often have an unusually long ITS sequence, ranging from around 900 to 2200 base pairs. The ITS barcode locus is consequently difficult to obtain for chanterelles (Schoch et al. 2012). This is especially true for old type specimens, due to their degraded DNA and resulting difficulties in extraction, and the frequent failures in the annealing of fungal primers designed to amplify the ITS locus or part of it.
While phylogenetic understanding of Cantharellus in Europe and North America has recently improved (Buyck et al. 2016c,d,e,f;Olariaga et al. 2015Olariaga et al. , 2016, the continuing lack of sequence data for Old World Cantharellus has helped to perpetuate taxonomic confusion regarding species delimitation and infrageneric relationships (Buyck et al. 2013(Buyck et al. , 2014. For Africa, some Cantharellus species from Madagascar and the African mainland have been circumscribed by single or multilocus molecular phylogenies , Buyck et al. 2014, 2016a, 2017, De Kesel et al. 2016). However, species recognition for the majority of chanterelles from the Guineo-Congolian rain forest is still based on morphological descriptions published over half a century ago (Heinemann 1958(Heinemann , 1959(Heinemann , 1966.
Many of the older species names for African chanterelles have been misapplied to morphologically similar specimens gathered in dense, closed-canopy rain forest versus the surrounding seasonal woodlands, or in open woodlands of neighboring Madagascar (Buyck et al. 2016g). As type specimen DNA of these earliest described rain forest chanterelles appears completely degraded (fide De Kesel et al. 2016), epitypification with sequencing of newly collected specimens is the most efficient way for unambiguous species delimitation. Until recently, new, well-documented specimens of rain forest chanterelles have not been available for sequencing. Thanks to renewed collecting efforts for Cantharellus in the African rain forest, the limits of species bearing these older names can be assessed, and the epitypification process has begun (Buyck et al. 2016a,b,g, De Kesel et al. 2016. Here we epitypify Cantharellus densifolius Heinem. based on recent collections made ~400 km from the type locality but in the same forest habitat. The chosen epitype, which is in perfect agreement with the original description, clearly demonstrates that the name has been misapplied to different species for decades. The new collections constitute the first records for C. densifolius since this species was collected by Mme. Goossens-Fontana in 1929 and later described by Heinemann (1958). In this paper we also epitypify Cantharellus luteopunctatus Heinem, previously considered a yellowish color-variant of C. densifolius (Eyssartier 2001), but shown here to be a morphologically well-defined, independent species. Additionally, two new species previously confused with C. densifolius are described.

Collecting and macromorphology
Basidiomata were collected in the Central African Republic (RCA) during dry conditions in early May 2016 in pure Gilbertiodendron dewevrei stands of the Dzangha-Sangha Forest Reserve. In Cameroon, basidiomata were collected during the Aug.-Nov. rainy seasons of 2014, 2016, and 2017 from the Dja Biosphere Reserve, Northwest Sector near the village of Somalomo, Upper Dja River Basin, within a two km radius of a base camp located at 3°21'29.8"N, 12°43'46.9"E, 650 m a.s.l., in forests dominated by G. dewevrei. Photographs and descriptions of macromorphological features were made from fresh material in the field. Colors were compared with color plates from Kornerup and Wanscher (1978) and are cited in parentheses. Collections were dried with a self-made drier (RCA) or silica gel (Cameroon). Epitype material and additional specimens are deposited in PC, Museum national d'histoire naturelle, Paris, and for the Cameroon collections also in the following herbaria: YA, Cameroon National Herbarium; HSC, Humboldt State University.

Micromorphology
Microscopic observations and measurements were made from ammoniacal Congo red mounts after a short pretreatment in a 10% aqueous KOH solution to improve tissue dissociation and matrix dissolution. Original drawings for all elements of the hymenium and pileipellis were made at a magnification of 2400× with the aid of a camera lucida. Measurements of basidiospores cite length, width and length/width ratio (Q) in this format: (minimum-) mean minus standard deviation -mean value -mean plus standard deviation (− maximum measured); basidiospore size statistics are based on 20 basidiospores measured per specimen.

Taxon sampling and phylogenetic analyses
Genomic DNA isolation, amplification and sequencing for the transcription elongation factor 1-alpha (tef-1) of the new Cantharellus collections were obtained as described in Buyck et al. (2014). The tef-1 sequence data from other taxa were obtained from our previous publications (Buyck et al. 2014(Buyck et al. , 2016aDas et al. 2018). Sequences were assembled and corrected with the software package Sequencher 3.0 (Gene Codes Corp., USA). Alignment of tef-1 was performed manually in MacClade 4.05 (Maddison and Maddison 2002). Searches for the optimal tree and branch robustness were conducted with the program PhyML (Guindon and Gascuel 2003), under a GTR nucleotide substitution model, with the search starting from a distance-based tree and with the proportion of invariable sites, gamma shape parameter and number of substitution categories estimated during the search. Three independent runs were conducted to check for convergence toward the same likelihood value. Branch support was estimated based on 500 bootstrap replicates (ML-bs) and was considered significant when ≥ 70% (Mason-Gamer and Kellogg 1996;Alfaro et al. 2003).

Results
Seven new sequences were produced for this study (five collections of C. densifolius, one of C. tomentosoides, and one of C. luteopunctatus). The alignment used for phylogenetic analyses included sequences of 90 Cantharellus specimens and one of Craterellus tubaeformis used for outgroup. The full alignment length was 864 base pairs. After exclusion of three spliceosomal introns, the remaining 629 characters were used for the analyses.
Spores hyaline, 5.6-7 × 3.7-4.5 μm, shortly ellipsoid, thin-walled, not amyloid; apiculus small. Basidia slender, 37-48 × 6-8 μm, probably 6-spored. Hymenophoral trama pseudoparenchymatic, slightly bilateral. Pseudoparenchyma very compact. Pileipellis squamulae composed of easily detaching cells that are irregularly cylindrical, often undulating, thick-walled with a very thick yellow wall in ammonium solution; the terminal cells obtusely rounded. Clamp connections rare." Description of the epitype. Basidiomata solitary or in small groups. Pileus medium-sized to rather large and up to 100 mm diam., 1-2 mm thick at mid-radius, yet firm and leathery; margin undulating, irregularly waving to strongly lobed, smooth; surface layer remaining more or less continuous in the center, then disrupting toward the margin with expansion of the pileus and forming dark, more or less concentrically arranged squamules or fibres; observed under a hand lens these can be appressed and flat, or forming a woolly-cottony mass of suberect fibers, pale brown (5AB3) to warm chocolate brown or dark brown (5EF7-8, 5F4-8, 5D5-8, 5C5-6) when young, but rapidly tinged with ochraceous orange as a consequence of the exposure of the underlying pileus tissue and the yellowing tendency of the context. Hymenophore composed of very crowded (>30/mm) gill folds, which are very low (<1 mm) and thick, not interveined, often transversely fissuring over their entire height, repeatedly forking, strongly decurrent, off-white when young, then darkening to the color of coffee with copious milk, moderately to strongly yellowing upon handling. Stipe 40-60 × 4-5 mm, widening toward the hymenophore and there up to 8(-17) mm wide; surface smooth, whitish, pale brown just beneath the hymenophore. Context whitish, thin and leathery, fibrous in the stipe, faintly to strongly yellowing upon injury or handling, occasionally turning rusty brown. Odor faint. Taste mild. Spore print off-white.
The epitype specimen selected here perfectly agrees with the original description of C. densifolius. Indeed, Heinemann (l.c.) described it as a medium-sized species with an infundibuliform, ochre-orange and finely squamulose pileus measuring ca. 80 mm diam. and ending in an irregular but stretched margin, with strongly decurrent, crowded, frequently forking and very low gill folds (< 1 mm high) with blunt edges, lacking any intervenation. Heinemann cited shortly ellipsoid basidiospores of near identical size, more precisely given by Eyssartier (2001), as important microscopic features, along with the pileipellis composed of easily disintegrating, very thick-walled hyphal extremities that are sinuous-undulating in outline (compare Heinemann 1958 fig. 45B with our Fig. 3c).
The typical features appear to be quite constant across all specimens of C. densifolius examined here, including both the size and shape of basidiospores (Table 1), as well as the undulate, thick-walled, often apically tapered or constricted hyphal extremities (although less so in Buyck 16.137). The ochre-orange color of the pileus described for the holotype was also present in the epitype (Fig. 2a, c) but across collections examined here pileus color ranged from ochraceous yellow over orange to pale brown and even dark chocolate brown, but never to bright lemon yellow as described for C. luteopunctatus. This is consistent with the highly variable color within many other Cantharellus species (Olariaga et al. 2015). For C. densifolius, the general color of the pileus also depends on the degree of yellowing of the context underneath the disrupted surface tomentum, which can vary between or within individual basidiomata.
The form of the hyphal extremities composing the pileal tomentum is very similar to that of various other squamulose species in subg. Rubrinus sect. Isabellinus Eyssart. & Buyck, in particular those of the African C. tanzanicus Buyck & V. Hofst. (Buyck et al. 2013) and the Malagasy C. eucalyptorum Buyck & V. Hofst. and C. tricolor Buyck & V. Hofst., the latter two species being associates of introduced eucalypts . The differences in habitat and basidiospore size allow differentiation of C. densifolius from these species.
Cantharellus densifolius has repeatedly been reported from the surrounding woodland area in Africa (e.g. Heinemann 1966, Buyck 1994, Buyck et al. 2000 Table 1. Comparison of basidiospore measurements for the discussed species.
Basidiospores short-ellipsoid to ellipsoid, (5.4-)5.7-6.04-6.4(-7.1) × (3.9-)4.0-4.29-4.6(-5.0) μm, Q = (1.24)1.29-1.41-1.53(-1.79), smooth. Basidia quite short, mostly 30-40(-50) × 6-7 μm, (5-)6-spored. Cystidia none. Subhymenium cells mostly hardly wider than the basidium base, but locally more inflated parts make it somewhat intermediate between distinctly pseudoparenchymatous and filamentous. Pileipellis composed periclinal thin-walled hyphae of variable diameter, but most ca. 10 μm wide, that locally emit anticlinal tufts of short-septate chains of more or less inflated cells, with the largest cells in these chains distinctly zebroid incrusted and the more terminal cells distinctly thick-walled (up to 1 μm thick); terminal cells 30-60 μm long, mostly (6-)10-15 μm wide, subcylindric or clavulate to lageniform, with obtusely rounded to attenuated tips, never remarkably undulate or irregular in outline. Discussion. Cantharellus luteopunctatus has long been considered as "uncomfortably close" to C. densifolius. Eyssartier (2001) found no significant microscopic differences between their holotypes, and suggested that C. luteopunctatus was likely a more yellowish color form of the latter species. For several decades following its original description C. luteopunctatus was not discussed in the literature, until recently by Eyi N'dong et al. (2011) who accepted it as an independent species. Both C. luteopunctatus and C. densifolius were also maintained as independent entities in a recent identification key to all African chanterelles (De Kesel et al. 2016). Our choice of epitype has been based on the specimen with the highest degree of similarity with the original macro-and microscopic description of C. luteopunctatus, and the original watercolor showing a species with a similar stature, color, and laterally compressed stipe (Fig. 4c).
Close reading reveals some differences between the original descriptions for C. luteopunctatus and C. densifolius. Apart from the difference in pileus color, the second most important difference, also noted by Eyi N'dong et al. (2011), concerns the configuration of the hymenophore. Gill folds were originally described for C. luteopunctatus as "irregularly forking and with many interstitial anastomosing veins", versus those of C. densifolius which were "1-4 times forking, not interveined" (Heinemann 1958). Although the presence and frequency of anastomoses between gill folds can be highly variable among and within Cantharellus species, it remains nevertheless a very informative feature to characterize those species that appear always to be on one side of the continuum (De Kesel et al. 2016). In this particular case, the macromorphologies of recent collections suggest that this feature is consistent across, and different between, specimens of C. luteopunctatus and C. densifolius.
Our collections also demonstrate a difference in hymenophore color between the species, something that is not evident from Heinemann's descriptions. Heinemann described C. luteopunctatus as having a yellow hymenophore, but does not indicate the color for the hymenophore of C. densifolius, although the original watercolor clearly shows it to be more or less ochraceous (see Fig. 2c here, and Heinemann 1959, Plate XXVI, fig. 6). Our collections confirm this ochraceous to dirty isabelline color of the hymenophore of C. densifolius, even when still relatively young, whereas the hymenophore color is more variable in C. luteopunctatus due to the translucent context above. The hymenophore of C. luteopunctatus is pure white when young, but it may also have pinkish tinges when the pileus surface is still densely covered by pinkish brown squamae, and then becomes more yellowish (which is the color mentioned in the original description) with maturation due to the yellowing context and absence of squamae over the expanded pileus margin. As in C. densifolius, the yellowing can be of variable intensity; for instance, the exposed context of TH 10442 is more strongly chrome yellow than that of the epitype (Fig. 5a).
Other considerable differences between C. luteopunctatus and C. densifolius include the surface structures of the pileus and stipe. In C. luteopunctatus, distinct central pileal squamae are erect, flesh brown to pinkish brown, and strongly separated and paler toward the margin. The pinkish color of the squamae was also mentioned in the original description of Heinemann (1958). In contrast, the pileus surface of C. densifolius is a continuous tomentum that lacks a pinkish flesh color and is woolly-fibrous, before breaking up concentrically in appressed fragments. Furthermore, in C. luteopunctatus the upper stipe surface has the same squamae as the pileus center, whereas in C. densifolius the stipe surface is smooth (compare Figs 2,4,5).
Micromorphologically, the basidiospores are nearly identical in both species (Table 1), but the pileipellis differs dramatically. In C. luteopunctatus the pileipellis is composed of fascicles of thin-to slightly thick-walled hyphae (corresponding to the erect squamae) that are recognizable on the background of more or less parallel, thin-walled hyphae of the interstitial surface, whereas in C. densifolius, the thicker-walled hyphae are not organized in fascicles. Moreover, in the latter species the distal cells of these thick-walled hyphae are much more undulate-sinuous in outline and narrower than those of C. luteopunctatus.
A final remark concerns the edibility of these Central African chanterelles: In Cameroon C. luteopunctatus basidiomata are mild-flavored and consumed by the indigenous Baka, while C. densifolius slowly develops a very strong bitterness and is not consumed by the Baka (T. Henkel, pers. obs.). While the bitter taste was also noted in the original description (Heinemann 1958), the first author did not detect bitterness for C. densifolius specimens from the Central African Republic. Diagnosis. Cantharellus tomentosoides is similar to C. densifolius in its low, blunt and crowded gill folds, overall yellowish brown color, identical basidiospores, and same habitat, but differs in its mostly smaller basidioma size, pileus surface texture, slightly more yellowish olive hymenophore color, and less thick-walled, less sinuous pileipellis extremities.
Holotype   Description. Basidiomata in small clusters, up to 40 mm high. Pileus 20-30 mm diam., thin and leathery, wavy with inrolled margin, young entirely hirsute-rugose, remaining lacerate-fibrillose to cottony in the center, elsewhere slightly rugose but lacking well-defined appressed scales, overall pale grayish brown with dark brown center, very early on becoming narrowly but strongly depressed centrally. Stipe slender, 6 mm diam., 20-30 mm high, rapidly elongating while pileus is still small, paler to concolorous with pileus margin, occasionally white at base; interior distinctly fistulose. Hymenophore composed of very crowded (up to 40/cm), low but comparatively thick and blunt gill folds, these 1 mm high, repeatedly forking, frequently fissuring over their full height, yellow with brownish tint, transitioning to warm egg-yolk yellow near extreme margin. Context leathery, whitish in the pileus, almost concolorous with the stipe surface, yellowing slowly. Odor agreeable, typical. Taste mild. Spore print not obtained.
Cantharellus tomentosoides resembles C. densifolius in its similarly crowded gill folds, overall yellowish brown coloration and identical basidiospores, but differs in its mostly smaller size, different texture of pileus surface, slightly different color of hymenophore, and thinner-walled hyphal extremities at the pileus surface. Additionally, these two species are phylogenetically distinct (Fig. 1). Diagnosis. Cantharellus densilamellatus resembles C. densifolius in its overall yellowish to orange-brown color, but differs in its thinner and comparatively well-developed gill folds with less blunt edges, smaller size, nearly thin-walled, less undulate hyphal extremities at the pileus surface, more elongate basidiospores, and its seasonal woodland habitat.
Discussion. Cantharellus densifolius has long been the only available name for yellowish brown, clampless chanterelles in Africa with a squamulose pileus and crowded gill folds. Indeed, the holotype collection of C. densilamellatus reported here was initially identified as C. densifolius in Buyck et al. (2000) and, in the absence of any reliable concept for Heinemann's species, was even maintained as C. densifolius in the multigene Cantharellus phylogeny of Buyck et al. (2014). Although more than one woodland species might have been referred to as 'C. densifolius', C. densilamellatus as described here is undoubtedly one of the more common and widespread inhabitants of the Zambezian miombo woodlands. It differs from the true C. densifolius not only in its habitat preference, but also in its more elongated basidiospores, which are very similar to those of C. tomentosus Eyssart. & Buyck (another, but much less common, woodland species with crowded gills and much darker pileus surface and hymenophore -see Buyck 1994 [as 'nyarympu', its Kirundi vernacular name] and Buyck et al. 2000). Cantharellus densilamellatus further differs micromorphologically from C. densifolius in its more regular, less undulate and thinner-walled hyphal extremities of the pileus surface (Fig. 10). While the phylogenetic analysis presented here (Fig. 1) shows a close relationship of C. densilamellatus with C. luteopunctatus, the latter species differs in its pinkish, erect squamae, bright yellow pileus color, and initially white and strongly anastomosing gill folds.

Conclusion
Phylogenetic analysis including the newly obtained sequence data demonstrated that C. densifolius and C. luteopunctatus, here epityped, belong in the same subgenus but in different subclades. Additionally, the name C. densifolius has been consistently misapplied to at least one, and possibly several, similar taxa from the African woodland area (De Kesel pers. comm., Buyck 1994, Buyck et al. 2000, Härkönen et al. 2015. One of these woodland species described here, C. densilamellatus, is very different morphologically from, but phylogenetically sister to, C. luteopunctatus. Cantharellus tomentosoides is a new species that is morphologically similar to, but phylogenetically distinct from, C. densifolius and is from the same local habitat in the G. dewevrei rain forest. These results, along with the continuing discovery of new, morphologically unique African chanterelles, emphasize the importance of Africa as a global diversity hotspot for Cantharellus (Buyck 2016, De Kesel et al. 2016, Buyck et al. 2017.