Looks can be deceiving: the deceptive milkcaps (Lactifluus, Russulaceae) exhibit low morphological variance but harbour high genetic diversity

The ectomycorrhizal genus Lactifluus is known to contain many species complexes, consisting of morphologically very similar species, which can be considered cryptic or pseudocryptic. In this paper, a thorough molecular study is performed of the clade around Lactifluus deceptivus (originally described by Peck from North America) or the deceptive milkcaps. Even though most collections were identified as L. deceptivus, the clade is shown to contain at least 15 species, distributed across Asia and America, indicating that the L. deceptivus clade represents a species complex. These species are morphologically very similar and are characterized by a tomentose pileus with thin-walled hyphae and a velvety stipe with thick-walled hyphae. An ITS1 sequence was obtained through Illumina sequencing for the lectotype of L. deceptivus, dating from 1885, revealing which clade represents the true L. deceptivus. In addition, it is shown that three other described species also belong to the L. deceptivus clade: L. arcuatus, L. caeruleitinctus and L. mordax, and molecularly confirmed that L. tomentoso-marginatus represents a synonym of L. deceptivus. Furthermore, two new Neotropical species are described: Lactifluus hallingii and L. domingensis.


INTRODUCTION
Lactifluus is a genus of ectomycorrhizal fungi which has its main distribution in the tropics. More than in Lactarius, the genus in which Lactifluus representatives used to be included (Buyck et al. 2008), it is known for its high molecular diversity, with many species complexes occurring throughout the genus (Stubbe 2012; Van de Putte 2012;De Crop 2016). These species complexes consist of several closely related species, with limited morphological variability, making them hard to distinguish from each other.
A good example of such species complexes can be found in Lactifluus section Albati. Species in this section are characterized by large white basidiocarps, a velutinous cap, an acrid taste of the context, the presence of macrocystidia and a pileipellis consisting of hyphae, without isodiametric cells. This section was first thought to contain only temperate representatives, with species known from Europe and North America. More recently, some Asian species belonging to this section were discovered, expanding its known distribution across the Northern hemisphere (Wen and Ying 2005;Le et al. 2007b). type specimens (Montoya and Bandala 2005). However, molecular data suggested that specimens identified as L. deceptivus represented several species, and a detailed molecular study is imperative to resolve the species boundaries in this complex. Therefore, a phylogeny based on multiple loci of the L. deceptivus clade was constructed in this paper, including samples from different biogeographic regions.

Sampling
Samples were included from sampling expeditions to the United States of America (2005), Canada (2007), Vietnam (2011), Dominican Republic (2016 and Panama (2018). In addition, loans were requested from the NY, NYS, FLAS and MICH herbaria. Type specimens of described species that (could) belong to L. sect. Albati were requested, but of these the type of L. caeruleitinctus was in too poor condition for sequencing. For L. deceptivus, no type was designated in the original description. However, Hesler and Smith (1979) indirectly designated a lectotype, and this collection, in addition to five other collections mentioned in the studied material were requested.

Morphological analysis
Macroscopic characters were observed from fresh material with colour codes referring to Kornerup and Wanscher (1978). Microscopic characters were observed from dried material. Basidiospores were mounted in Melzer's reagent and hymenium, pileipellis and stipitipellis were studied in Congo red in L4. The basidiospore measurements (i.e. length, width and Q = quotient of length and width) are given as [Av a -2 × SD a ]-Av a -Av b -[Av b + 2 × SD b ], in which Av a = lowest mean value for the measured specimens, Av b = greatest mean value, SD a/b = standard deviation of the measurements with the lowest and greatest mean value respectively (number of spores measured per specimen = 20). Basidiospores were measured in side view without ornamentation. Measurements of basidia, cystidia and marginal cells are given as [Av-2 × SD]-Av-[Av + 2 × SD], based on minimum 47 measurements per species. Measurements of basidia do not include sterigmata. Line drawings of spores were made based on stacked photographs (Nikon Eclipse Ni-U, stacking software: Extended Depth of Field, Nikon Nis Elements module) 5000× magnified, those of other elements and sections were made with the aid of a drawing tube at magnification 1500× (Olympus cx31 microscope).

Molecular analysis
DNA was extracted from fresh material preserved in CTAB (Cetyl trimethylammonium bromide) using the CTAB extraction described in Nuytinck and Verbeken (2003). A modified CTAB protocol (Tel-Zur et al. 1999;mod. by Agentschap Plantentuin Meise) was used for dried collections. PCR amplification protocols follow Le et al. (2007a). Two nuclear markers were amplified: (1) the internal transcribed spacer region of ribosomal DNA (ITS), comprising the ITS1 and ITS2 spacer regions and the ribosomal gene 5.8S, using primers ITS1F and ITS4, in addition to primers ITS5, ITS2 and 58SF for problematic material (White et al. 1990;Gardes and Bruns 1993;Tedersoo et al. 2013), and (2) the region between the conserved domains 6 and 7 of the second largest subunit of the RNA polymerase II (RPB2), using primers bRPB2-6F and fRPB2-7cR (Liu et al. 1999;Matheny 2005).
PCR products were sequenced using an automated ABI 3730 XL capillary sequencer (Life Technology) at Macrogen. Forward and reverse sequences were assembled into contigs and edited with Sequencher v5.0 (Gene Codes Corporation, Ann Arbor, MI, U.S.A.) or BioloMICS (BioAware SA NV).
For the 134-year-old type specimen of L. deceptivus, Illumina MiSeq sequencing was chosen as an alternative to conventional Sanger sequencing to overcome the high fragmentation of the ancient DNA and the occurrence of non-target DNA from fungal contaminants. Also, type specimens of L. mordax and L. tomentoso-marginatus were sequenced with this method. Amplification of the ITS1 region was carried out using a two-step PCR process. In the first PCR, universal ITS1-F/ITS2 primers extended with Nextera™ tails (Illumina) were used following the same settings as detailed in Le et al. (2007a). After a DNA quantity and quality check, the PCR product was cleaned up with the NucleoMag NGS Clean-up and Size Select kit (Machery-Nagel). In the second PCR, a Nextera™ XT label (Illumina) was added to the amplicon under the following conditions: 3 μL of template DNA, 1 μL of each primer (10 pmol/μL), and 15 μL of Master Mix for a final volume of 20 μL. Amplification conditions were: 95°C for 10 min, 8 cycles of 30 s at 95°C, 60 s at 55°C and 30 s at 72°C, followed by 7 min at 72°C. After quantification and clean-up, the sample was sent to BaseClear (Leiden, the Netherlands) for paired-end sequencing using the Illumina MiSeq technology (2 × 300 bp) amongst a batch of other amplicons with different Nextera™ labels.
The Illumina sequence reads were processed using the Naturalis Galaxy v.19.01 instance. The reads were demultiplexed on their unique tag to isolate the reads from specific specimens. R1 and R2 reads from pairedend sequencing were merged with FLASH (Magoc and Salzberg, 2011) with the minimum overlap size set at 100 bp. Reads shorter than 250 bp or with more than 8 consecutive N's or a Phred score lower than 28 were discarded and primers were trimmed with Cutadapt (Martin 2011). After a quality control step with   Table 1). Metadata of collections in the L. deceptivus complex are given in Table 2. Sequences were aligned online in the multiple sequence alignment program MAFFT v7 (Katoh and Toh 2008), using the E-INS-I strategy. Trailing ends were trimmed and the alignment was manually edited where needed in Mega 6 (Tamura et al. 2013). The ITS+LSU alignment was partitioned into partial 18S, ITS1, 5.8S, ITS2 and partial 28S. The RPB2 alignment was partitioned into the intron and the first, second and third codon positions of the exon. PartitionFinder was used to find the appropriate partitioning scheme (Lanfear et al. 2017). Maximum likelihood (ML) analyses, using RAxML v8.0.24 (Stamatakis 2014), were combined with the Rapid Bootstrapping algorithm with 1000 replicates under the GTRCAT option (Stamatakis et al. 2008). There was no supported conflict between the separate gene trees, so they were concatenated. The concatenated alignment can be obtained from the first author or TreeBASE (ID

RESULTS
Illumina Miseq sequencing was used to sequence the type specimen of L. deceptivus collected in 1885. After library sequencing, merging read pairs and quality control steps for this type specimen, 727 ITS1 sequences were retained that were clustered in 10 zero-radius OTU's. Only one of these belonged to the genus Lactifluus, and this sequence was used in further phylogenetic analysis. Other OTU's represent contaminants, the most abundant being Penicillium. For the type specimens of L. mordax and L. tomentoso-marginatus, 125-271 sequences were retained, clustered in 4-6 zero-radius OTU's, which each contained one sequence related to L. deceptivus s. lat.
In total, sequences were obtained for 47 collections belonging to the L. deceptivus complex. Of these collections, 36 had been identified based on morphology as L. deceptivus (most other collections were not identified to species level). The phylogeny shows that these collections represent at least 15 species (Fig. 1). Most species originate from the Nearctic, but also two Indomalayan and three Neotropical species were found. Only four species were described thusfar, so an additional 11 new species were uncovered by the molecular analysis. Diagnosis: Differs from clade 5 in the slightly lower average Q of the basidiospores, the slightly larger basidia, the slightly longer cystidia, the Hispaniolan distribution and the association with Pinus occidentalis.
Description: Pileus 35-100 mm diam, compact, at first convex and umbilicate, becoming flat, with depressed centre that becomes deeply depressed with maturity; surface at first bald, but soon roughening, becoming torn into patches or scales, dry to sticky at the centre; margin at first involute and clothed with a soft or cottony tomentum, then spreading or becoming elevated and more or less fibrillose, with a hand-lens loosely arranged brown pigmented hairs are visible; surface whitish at first, soon discolouring tan and brownish. Stipe 17-50 × 10-20 mm, rather short (compared to the diameter of the cap), usually tapering to base; surface dry, irregularly rugged, slightly velvety towards the base, whitish, discolouring brownish or tan; context solid. Lamellae adnate or slightly decurrent, with lamellulae of various length, rather broad, to 5 mm wide, subdistant, occasionally forking, whitish or pale yellowish, becoming cream coloured and finally dingy tan, usually staining and discolouring brownish; edge entire and concolourous. Context firm, white, unchanging when cut; smell pungent at maturity; taste strongly acrid. LATEX white; unchanging; often staining brownish. Chemical reactions: Cap surface reddish with KOH. was found, no Fagaceae occur. There are two species of pines on the island: the endemic Pinus occidentalis (in the central Cordillera) and the introduced P. caribaea (in the northern Cordillera). Lactifluus domingensis has been found exclusively with the endemic P. occidentalis and never with the introduced P. caribaea. This association with P. occidentalis, and therefore Hispaniolan distribution, is probably a unique feature of the species. So far, no other Lactifluus species are known from this island. Aside from its distinctive distribution, L. domingensis bears great resemblance to the other species of the Lactifluus deceptivus species complex, both macro-and microscopically. The phylogeny shows that L. domingensis is most closely related to a single collection from Canada (clade 6) and a clade with a northern North American distribution: clade 5 ( Fig. 1) (although unsupported, but strongly supported in the separate ITS phylogeny).
Distribution: Known from Costa Rica, Panama, and Colombia.
Remarks: As Lactifluus domingensis, L. hallingii bears great resemblance to some of the other species from the L. deceptivus species complex, both macro-and microscopically. In the phylogeny (Fig. 1), we find that L. hallingii is most closely related to a clade with a North American distribution: clade 2 (Fig. 1). Clade 2 differs from L. hallingii in the smaller basidia (30. 5-42.5-54 × 7.5-11.5-16 μm) and the shorter macrocystidia (29-52-75 × 5-9-13.5 μm). Compared to L. hallingii, the species described from Florida are relatively easily distinguishable: L. arcuatus has distinctly smaller spores (4-6 μm long), and L. caeruleitinctus has blue tinges in the stipe which are lacking in L. hallingii. Lactifluus domingensis shares the Neotropical distribution, but that species was found with the endemic Pinus occidentalis on the island of Hispaniola. For microscopic differences between these two species, see the remarks under L. domingensis. Remarks: Based on the original description, this species could fit morphologically in L. sect. Albati because of the white basidiome, the cottony inrolled margin, and the acrid latex. The placement in this section is confirmed molecularly by the position of the type sequence in the phylogeny (Fig. 1). ( Remarks: Based on Murrill's notes on the type specimen, this species could fit morphologically in L. sect. Albati because of the white basidiome, the acrid taste, and the strongly unpleasant smell while drying. The placement of a collection identified as L. caeruleitinctus in L. sect. Albati (Fig. 1) further supports the placement of this species in L. sect. Albati.

Lactifluus caeruleitinctus
Lactifluus mordax (Thiers) Delgat, comb. nov. Remarks: Based on the original description, this species could fit morphologically in L. sect. Albati because of the matted-tomentose margin, the very acrid latex, and the cuticular structure of the pileipellis. The placement in this section is confirmed molecularly by the position of the type sequence in the phylogeny (Fig. 1).

DISCUSSION
A first attempt to find out which clade represents L. deceptivus was made by sequencing several collections from the studied material from Hesler & Smith's description (Hesler and Smith 1979): collections Schaffer 5895, Ammirati 2392, Smith 84,511, and Weber 4277. However, our analysis inferred that these collections represent three different species: clades 2, 3, and 5 ( Fig. 1), thus the interpretation of L. deceptivus remained elusive.
In the phylogeny, there are several clades that have a relatively close distribution to where L. deceptivus was described (i.e. New York State): clades 2, 3, 5, and 12. The only microscopic character mentioned in the original description (Peck 1885) is the length of the spores (8.9-12.7 μm), so spores were measured for these four clades. Clade 12 has significantly smaller spores (5.9-7-8 × 4.6-5.3-6.1 μm), but the other three clades have similar spore sizes (clade 2: 9.6-10.6-10.9-11.9 × 7.5-8.4-8.5-9.5 μm; clade 3: 9.4-10.6-11.8 × 7.8-8.5-9.3 μm; clade 5: 9.9-10.7-10.8-11.9 × 7.4-8-8.6 μm), so all three clades were considered possible candidates to represent the true L. deceptivus. Lactifluus deceptivus was described by Peck (1885), and Hesler and Smith (1979) designated a lectotype (Peck s.n., NYS-F-000959). Samples this old have both time and conservation related DNA damage, besides exogenous DNA contamination, that makes nucleic acid extraction and amplification challenging (Forin et al. 2018). Therefore, Illumina Miseq sequencing was chosen as an alternative to the conventional Sanger sequencing to overcome these problems for the type specimen of L. deceptivus. As expected, due to both the old age and the lack of precautions during the manipulation of specimens throughout the herbarium's long life, contaminants are present in this specimen, and a total of 10 zero-radius ITS1 OTU's were recovered from the sample. One Lactifluus sequence was picked up by the analysis. Phylogenetic analysis shows that this sequence belongs to clade 3, which was indeed considered a possible candidate based on distribution and spore measurements, revealing that this clade represents the true L. deceptivus.
Three other described species were found to belong in the complex: L. arcuatus and L. caeruleitinctus, described from Florida by Murrill (1939Murrill ( , 1941; and L. mordax, described from Texas by Thiers (Thiers 1957). Descriptions of Murrill's species are rather concise, but both species have at least a clear character that sets them apart from L. deceptivus: L. arcuatus has distinctly smaller spores, and L. caeruleitinctus displays blue tinges in the stipe. Lactifluus mordax is can be macroscopically distinguished from L. deceptivus by the pileus colour, which is not white but yellow to cream, and microscopically by the smaller spores (7.5-9 × 6-7 μm) (Hesler and Smith 1979). A sequence was obtained for the holotypes of L. arcuatus and L. mordax, as well as a collection identified as L. caeruleitinctus, which shows that these species belong to a subclade (clades 9-15) of the L. deceptivus complex which is dominated by species known only from Florida (Fig. 1). Since L. deceptivus is situated in the other subclade (clades 1-8), these species are relatively more distantly related to L. deceptivus, which could explain why they are more easily distinguishable from it. Lactifluus tomentoso-marginatus was previously synonymised with L. deceptivus based on a detailed morphological study (Montoya and Bandala 2005), and the phylogeny confirms this synonymy by the position of the holotype in the same clade as the type of L. deceptivus (Fig. 1). However, other collections studied in the paper of Montoya and Bandala, originating from Mexico and initially identified as L. tomentosomarginatus, were also considered to belong to the same species by the authors based on the morphological study. A sequence was obtained for one of these collections, and the phylogeny shows that it represents a distinct clade from L. deceptivus (clade 8, Fig. 1), further demonstrating the difficulty of morphologically delimiting species in this complex.
The Lactifluus deceptivus complex previously exclusively contained species described from the Nearctic. It was shown that species from this complex also occur in Indo-Malaya and the Neotropics, and two new Neotropical species are described. Lactifluus domingensis was found in the Dominican Republic, on the island of Hispaniola, while L. hallingii was found on the mainland, distributed across Costa Rica, Panama and Colombia. For most of the other clades, welldocumented collections are lacking due to the previous perception that L. deceptivus represented just a single, easy to recognize species. In addition, many of the clades contain only one or two collections, so to further unravel this complex there is a need for more well-documented collections. Hopefully, L. deceptivus s. lat. Will be collected and described in more detail now that it is known to represent several morphologically similar species.
Since species in this complex resemble each other so strongly, they can be considered pseudocryptic species.