A parasitic coevolution since the Miocene revealed by phase-contrast synchrotron X-ray microtomography and the study of natural history collections

The discovery of a new fossil species of the Caribbeo-Mexican genus Proptomaphaginus (Coleoptera, Leiodidae, Cholevinae) from Dominican amber, associated with a new fossil parasitic fungus in the genus Columnomyces (Ascomycota, Laboulbeniales), triggered an investigation of extant species of Proptomaphaginus and revealed the long-enduring parasitic association between these two genera. This effort resulted in the description of the fossil species †Proptomaphaginus alleni sp. nov., and one fossil and two extant species of Columnomyces, selectively associated with species of Proptomaphaginus: †Columnomyces electri sp. nov. associated with the fossil †Proptomaphaginus alleni in Dominican amber, Columnomyces hispaniolensis sp. nov. with the extant Proptomaphaginus hispaniolensis (endemic of Hispaniola), and Columnomyces peckii sp. nov. with the extant Proptomaphaginus puertoricensis (endemic of Puerto Rico). Based on biogeography, our current understanding is that the Caribbean species of Proptomaphaginus and their parasitic species of Columnomyces have coevolved since the Miocene. This is the first occurrence of such a coevolution between a genus of parasitic fungus and a genus of Coleoptera. The phylogenetic relations among Proptomaphaginus species are also addressed based on a parsimony analysis. Fossil specimens were observed by propagation phase-contrast synchrotron X-ray microtomography (PPC-SRμCT) and extant specimens were obtained through the study of preserved dried, pinned insects, attesting for the importance of (i) technological advancement and (ii) natural history collections in the study of microparasitic relationships.


Material and methods
Material. Four specimens of †Proptomaphaginus alleni were examined. Two samples labelled as MP025 (male) and MP033 (female) are preserved at the Staatliches Museum für Naturkunde Stuttgart, Germany (SMNS). Two samples labelled as MP031 (female) and MP032 (male) will be preserved at FH (Farlow Herbarium, Harvard University, Cambridge, Massachusetts) and the Muséum national d'Histoire naturelle, Paris, France (MNHN), respectively. No details are available about amber deposits or collecting conditions. The newly described fossil species of Laboulbeniales, †Columnomycetes electri, is attached to the right metatibia of specimen MP031 (Fig. 1a).
Preliminary screening of specimens of Proptomaphaginus hispaniolensis and Proptomaphaginus puertoricensis in the personal collection of M. Perreau in Paris, France (CMPR) resulted in the discovery of Laboulbeniales thalli on both species. We reached out to Dr. Stewart S. Peck, whose insect collection is preserved at the Canadian Museum of Nature (CMN) in Ottawa. His screening efforts revealed more infected specimens of Proptomaphaginus, which were kindly sent to the second author for study of their associated parasites.
External and internal structures of the Coleoptera were illustrated using both visible light observations and propagation phase-contrast synchrotron X-ray microtomography (PPC-SRμCT) 28 .
Propagation phase contrast synchrotron X-ray microtomography. Microtomographic observations allow virtually dissecting specimens in a non-destructive way 29 and displaying the cellular structure of associated organisms, such as ectoparasitic Laboulbeniales fungi. PPC-SRμCT was performed at the European Synchrotron Radiation Facility (ESRF, Grenoble, France). Scans were performed on the beamline ID19 (for MP031, MP032, MP033) or BM5 (MP025) with a monochromatic X-ray beam at the energy of 20 keV, using a multilayer monochromator. The CCD detector was a FreLoN HD2k (fast read-out low noise) with 2048 × 2048 pixels, coupled to a microscope system with a single crystal YAG:Ce scintillator screen of 25 mm of thickness. The resolution (voxel size) of the scans of MP025, MP031, and MP032 was 0.685 μm. Two pixel sizes were used for MP033, 1.406 μm and 0.703 μm. A continuous rotation was used to blur out undesired details located outside the field of interest (far from the rotation center) to decrease their contribution to the noise of the final reconstructed slices 30 . A special scan was performed on MP031 to resolve the parasitic fungus at 0.212 μm, with a distance sample-detector of 20 mm and with 1500 projections over 180°, also with a continuous rotation. Inhouse software packages present at ESRF were used for tomographic reconstructions. Segmentations were done with Vgstudiomax 2.1 (Volumegraphics, Heidelberg, Germany) on a computer based on an AMD motherboard Magny-Cours (48 cores) and 512 Go of random-access memory. www.nature.com/scientificreports/ In addition to the standard pictures, selected figures are also presented in anaglyph red/cyan to visualize the three-dimensional structures of key morphological details with red/cyan-filtered glasses. They are presented in Fig. 2; numbering is the same as the corresponding flat figures.
Modern specimens imaging. After dissection, genital structures of Coleoptera were treated as follows.
The aedeagus of male specimens was dehydrated in 95% ethanol, before being mounted in Euparal on a microscope slide. Female genitalia were cleared in hot 10% KOH for 10 min, stained with a diluted ethanolic solution of Azoblack 31 , rinsed in demineralized water, and then mounted in dimethyl-hydantoin-formaldehyde (DMHF) on a microscope slide. Visible light photographs were captured by a Spot Insight IN1820 camera attached on a Leica M10 stereomicroscope for Fig. 1a, and by a Keyence VHX5000 microscope with lens VH-Z250T for Figs. 3a-f, 4a-j.   32 . Photomicrographs were taken with an Olympus SC30 camera mounted on an Olympus BH2 bright field compound microscope and viewed and processed using cellSens 1.18 imaging software (Olympus, Tokyo, Japan). Line and stipple drawings were made with PITT artist pens (Faber-Castell, Nürnberg, Germany) based on photomicrographs. All slides are deposited at the Kriebel Herbarium, Purdue University (PUL; West Lafayette, Indiana). Herbarium acronyms are according to Index Herbariorum 33 .
Parsimony phylogenetic analysis of Proptomaphaginus. We performed the phylogenetic analysis of the genus Proptomaphaginus using a matrix comprising nine terminal taxa, with Ptomaphagus as out group and 23 characters of which 17 were parsimony-informative, 7 were external characters, 12 were characters of male genitalia; and 4 were characters of female genitalia.
The reduction of eyes was not included since this is generally associated with an hypogean adaptation, frequent in Ptomaphagini, and is likely homoplasic. The sexual dimorphism of the apex of elytra (Figs. 1d,e) was also excluded because this is a highly variable character in the genera Ptomaphagus, Ptomaphaginus, and Ptomaphaminus.
The matrix was compiled using Paup* 4.0b10 34 . A first analysis was done by exhaustive search of maximum parsimony. All characters were equally weighted and multi-state characters were treated as unordered. Next, a bootstrap analysis based on a heuristic search with 10,000 replicates was performed. Tree visualization and mapping of characters were done in Winclada 1.00.08 35 .

Results
Systematic paleontology of the host.     www.nature.com/scientificreports/ Etymology. Named in recognition of Albert D. Allen, a private beetle enthusiast, who provided the first specimens of this fossil species.
Description. Habitus: Fig. 1a,b. Head with normally developed eyes (Fig. 1f), and with an occipital carina. Punctation aligned in transverse microstrigae. Antennae compact, continuously widened from base to apex, the eighth antennomere strongly transverse (Fig. 1c). Pronotum approximately 1.3 × longer than wide, the widest at base, with dorsal punctation aligned in transverse microstrigae. Elytra elongated, approximately 1.7 × longer than combined width, with a single parasutural longitudinal stria. Apex sexually dimorphic, widely rounded in female (Fig. 1d), truncate in males (Fig. 1e). Flight wings present. Ventral structures: mesoventral process low, narrow and rounded, with some setae on the ventral surface. Epimeron of the mesoventrum transverse. Metaventrum prominent longitudinally on the median line, with a longitudinal median groove. Metaventrahl lateral sutures roughly parallel to the body axis ("ms" in Fig. 1f). Abdomen with six visible segments. Legs: protibiae with a single row of regular spines along the external edge (Fig. 1g), ventral spines are randomly distributed on the ventral surface. Mesotibiae and metatibiae with a comb of regular spines surrounding the insertion of tarsi. Male protarsi weakly dilated (Fig. 1g), female protarsi undilated. Male genital segment with a long and narrow spiculum gastrale, protruding forward beyond the anterior edges of the epipleurites by half of its length. Epipleurites with setae along the posterior edge (Fig. 3k,l). Aedeagus ( Fig. 3g-j) fairly thick, median lobe with two lateral apical prominent expansions ("le" in Fig. 3h) and one central less prominent expansion ("ce" in Fig. 3h).
Lateral sides narrowed at apex in dorsal view (Fig. 3h). Ventral ligulae strongly developed ("lg" in Fig. 3g-i). The internal stylus partly protruding beyond the ventral orifice of the median lobe ("is" in Fig. 3g). Lateral expansions and stylus are also clearly visible in frontal view ( Fig. 3j). At least two lateral setae and one apical seta are present. The median lobe was perhaps compressed laterally in its apical part during diagenesis since the ventral ligulae which generally closes the genital orifice significantly oversteps laterally the lateral margins of the median lobe. Female genital segment indistinct in the two female paratypes, but a helical spermaduct with a wide base is partly visible in specimen MP033 (Fig. 3m). The apical capsule that generally has the shape of a club in Proptomaphaginus is missing, likely resulting from a preservation artifact. Length 2.0-2.3 mm (approximated from the sum of lengths of head, pronotum, and elytra from the scutellum).
Notes. The external morphology is extremely similar to the single known extant species from the island of Hispaniola (Haiti and the Dominican Republic): Proptomaphaginus hispaniolensis. Eyes are normally developed, differing from other extant species that have reduced eyes as an adaptation to a hypogean lifestyle 26 . No reliable external characters are available. We eventually detected differences in the size of antennomeres but cannot exclude the possibility that these are preservation artifacts. However, male genital structures, which are important for species identifications of Leiodidae, give clear characters to distinguish the fossil species from its extant Caribbean relatives. In dorsal view, the apical part of the aedeagus-the only part that is well preserved-is much more narrowed, with the two lateral expansions of the median lobe nearly contiguous ("le" in Fig. 3h) while they are distant in the extant Caribbean species, Proptomaphaginus hispaniolensis ( Fig. 3b), Proptomaphaginus puertoricensis ( Fig. 3d), and Proptomaphaginus apodemus (Fig. 3f). In lateral view, lateral expansions are more sinuous and more prominent ( Fig. 3g versus Fig. 3a,c,e). The internal stylus of the endophallus ("is" in Fig. 3a,g), which is partly protruding outside the median lobe at the apex in Proptomaphaginus hispaniolensis is also visible in †Proptomaphaginus alleni but seems straight rather than curved. The male genital segment is similar in Proptomaphaginus hispaniolensis (Fig. 3j) and †Proptomaphaginus alleni (Fig. 3l), with a very long, thin spiculum gastrale protruding beyond the anterior margin of the epipleurites. Etymology. Referring to the host species, Proptomaphaginus hispaniolensis, to emphasize the presumed strict host specificity and biogeographical patterns that have resulted in these intimate parasite-host associations.  Description. Thallus 254 µm long from foot to tip of appendages, 233-422 µm long from foot to perithecial apex; colored amber-brown; basal cell of the receptacle above the blackened foot, the appendages, and the lower quarter to (maximum) half of cell VI hyaline; the remaining part of cell VI and perithecial venter darkening with age. Receptacle 113-192 × 25-30 µm, multicellular, semicircular in cross section, consisting of 18-36 superposed tiers of flattened cells (except those in the upper 2-3 tiers); 5-13 lowest tiers each consisting of a single cell, broadening upwards; cells above the 5-13 basal-most ones arranged in 2-3 rows; upper 2-3 tiers many-celled, with cells longer than wide. Appendages many, arising by proliferation from upper 3-4 tiers of receptacular cells, septate, slender, reaching up to 56 × 1.5-3.0 µm; some antheridial, some sterile. Cell VI 69-203 × 19-21 μm, between 3.6 and 9.7 × longer than wide, slender, with parallel margins, straight; arising laterally from one cell of a tier (10 to 17), attached to it with a darkened, obtriangular-shaped, foot-like structure. Perithecium 104-136 × 50-63 μm, 2.1 × longer than wide, ovoid to pyriform, broadest at lower third, asymmetrical, with one margin more convex than the other; upper part tapering; apex smooth and blunt, undifferentiated, symmetrical.

Systematic paleontology of the parasite.
Notes. The three extant species of Columnomyces can be distinguished based on morphological evidence. Columnomyces peckii stands out because of its size (Fig. 5f,h); it is generally conspicuously larger compared to the other species. Compared to Columnomyces hispaniolensis, the perithecium of Columnomyces peckii is similar in width but it is generally longer; this is reflected in the length/width ratios:  Table 1. Due to diagenesis (see "Discussion"), appendages and antheridia are missing in the amber-embedded thallus of †Columnomyces electri. It seems that (at least) the upper three of four tiers of the receptacle are missing. In addition, cell VI is detached from the receptacle. Despite these damages, we were able to assign the preserved thallus to the genus Columnomyces based on the pseudoparenchymatous receptable. In addition, †Columnomyces electri can be easily separated from the other species in the genus by several features: the cells of the receptacular lateral rows are longer than wide, while they are flattened in the other species; cell VI is much narrower; the perithecium of †Columnomyces electri is much more elongate, with the length/width ratio (2.7) far exceeding    26 . We extended his analysis to include the newly described fossil species and the two most species-rich Asian genera that belong to the same subtribe of Ptomaphagini, Ptomaphaginina. These genera are Ptomaphaginus and Ptomaphaminus. Ptomaphagus, which belongs to another subtribe of Ptomaphagini (Ptomaphagina), was selected as outgroup. The phylogeny does not aim to resolve the relationships among all Ptomaphagini or Ptomphaginina, rather it aims to make more precise the placement of Proptomaphaginus. We used 23 characters, of which 17 were parsimony-informative. The list of characters and the corresponding data matrix are given in Supplementary File 2. An exhaustive search resulted in three minimal trees. The strict consensus tree is shown in Fig. 7a. A bootstrap analysis based on a heuristic search with 10,000 replicates resulted in the tree shown in Fig. 7b.
Relationships among Proptomaphaginus microps, and Proptomaphaginus reddelli, and the Asian genus Ptomaphaminus are unresolved, resulting in a polytomy. The monophyly of the set of Caribbean species received maximum bootstrap. This suggests that Mexican species and Caribbean species may belong to two different genera. The corresponding apomorphies are (i) the thick ventral ligulae closing the apical orifice of the median lobe (versus thin ligulae for Mexican species and other genera of Ptomaphaginina), (ii) the presence of three apical expansions to the median lobe of the aedeagus (versus only two for Mexican species and other genera of Ptomaphaginina), (iii) the approximately symmetric shape of the aedeagus, (iv) the repartition of the lateral setae of the median lobe of the aedeagus, and (v) the presence of apical setae at the tip of the median lobe of the aedeagus.
The sister relationship of Proptomaphaginus apodemus and Proptomaphaginus darlingtoni is also supported (bootstrap 70%). Their identity as two different species has been extensively discussed, before Peck 36 gave a characterization based on genital characters. The two Hispaniolan species †Proptomaphaginus alleni and Proptomaphaginus hispaniolensis are supported as sister species with a bootstrap of 65%. This clade is retrieved as sister to (Proptomaphaginus puertoricensis + (Proptomaphaginus apodemus + Proptomaphaginus darlingtoni)) with maximum support. Of note is the high support for the sister relationship of (Ptomaphaminus + (Proptomaphaginus microps + Proptomaphaginus reddelli)) and the Caribbean set of Proptomaphaginus species (bootstrap 95%). However, as mentioned earlier, the set (Ptomaphaminus + (Proptomaphaginus microps + Proptomaphaginus reddelli) www.nature.com/scientificreports/ forms an unresolved polytomy. The genus Ptomaphaminus is widespread in southeastern Asia, from China to the Sunda Islands 37,38 , whereas Proptomaphaginus microps and Proptomaphaginus reddelli have a Mexican distribution 26 . We need more characters-sequence data-to resolve this node and to eventually move the Mexican species of Proptomaphaginus into a new genus.

Discussion
Generic association. Currently four species are recognized in the genus Columnomyces-one fossil species and three extant species. The type species of the genus, Columnomyces ptomaphagi R.K. Benj., was described from an unidentified species of Ptomaphagus (Ptomaphagini, Ptomaphagina) collected at Giant City State Park, Illinois 19 . We were unable to localize the host specimens from the type collection in order to make an identification at species level. However, it seems likely that the host is Ptomaphagus brevior Jeannel, 1949. This species has a wide distribution in North America and, more importantly, it is the single species of Ptomaphagus that has been reported from Illinois to date 39  Most thalli of Columnomyces peckii are located on legs, some on elytra. Thalli of Columnomyces hispaniolensis are located on legs and elytra, but also on the pronotum and abdomen (tergites and ventrites). The single fossil thallus of †Columnomyces electri is located on a leg. The type material of Columnomyces ptomaphagi was located on the elytra of the host 19 . It seems that Columnomyces species do not exhibit any specificity for position on the host.
When available, fossils in the forms of organisms entrapped in soft resin and subsequently preserved for millions of years after amber hardening, are used to describe many organisms around the world as well as to better understand evolutionary relationships among extant species 42 . The state of preservation and completeness of specimens in amber are highly variable among specimens 43 . Damages can occur pre-or perimortem, including loss of appendages while a living organism is struggling to be released from the resin; postmortem during the polymerization of resin, including changes of external morphology and body proportions in soft-bodied organisms such as arthropods and fungi; and post-extraction, either slow or rapid changes outside of the original anoxic sediments, as a result of anthropogenic changes or mechanical treatment 44 .
The single thallus of the fungus †Columnomyces electri in the piece of Dominican amber has a dislocated cell VI (a logical consequence of rapid preservation) and it lacks the distal-most tier(s) of the receptacle. This likely represents postmortem distortion-the damages happened during hardening of the amber. As a result of the receptacle missing its upper tier(s) in †Columnomyces electri, we do not know for sure how long the receptacle can be, that is, how many tiers of cells it consists of. The uppers tiers also carry the appendages, which are missing in the fossil thallus. In extant species, the appendage system is composed of many slender, septate branchlets up to 50-56 μm in length. Likewise, Benjamin 19 noted "many broken specimens" among his material of the extant Columnomyces ptomaphagi, including the holotype, which also shows a displaced base of cell VI. It was suggested that the pressure of the cover glass might have caused the detachment of cell VI from the rest of thallus, a process that could be somewhat compared to the hardening of amber.
As a final note, previous studies have also observed damage of Laboulbeniales fungi associated with cholevine beetles. This has been linked to the hypogean lifestyle of these hosts 24,[45][46][47] . Almost all cholevine beetles live underground where they feed on all sorts of decaying organic material-litterfall, rotting fungi, dung, carrion, detritus from vertebrate nests, et cetera 48,49 . Given the fact that many thalli of extant species of Laboulbeniales associated with Cholevinae hosts are damaged for myriad reasons, it comes as no surprise that the fossil species, represented by a single thallus, is damaged.
Biogeography. The distributions of species of the Proptomaphaginus species and their Columnomyces parasites are illustrated in Fig. 8a www.nature.com/scientificreports/ Pacific Ocean from Central America (Mexico) in Asia by migration through the Pangea or Laurasia supercontinent before its fragmentation, all European representatives becoming extinct. If true, this would make Proptomaphaginus a relict of these ancestors. Under this hypothesis and according to the present phylogeny, their direct descendants are likely members of the "microps species group" rather than those of the "apodemus species group". However, we cannot consistently resolve whether the microps species group is more closely related to the apodemus species group or to the Asian genus Ptomaphaminus. The apodemus species group is characterized by several autapomorphies that do not occur in any other group of Ptomaphaginina which suggests that it has independently evolved on the Caribbean islands. Unfortunately, no older fossil of Ptomaphaginina is available for assessing this scenario. For example, all fossils of the Cretaceous Albian amber deposit of Kachin in Myanmar (Cretoptomaphagus microsoma 59 and several other undescribed species) belong to Ptomaphagina, not Ptomaphaginina.

Conclusion
The description of †Proptomaphaginus alleni and the discovery and description of the new fossil fungus attached to the right metatibia of one specimen would not have been achieved without the application of propagation phase-contrast synchrotron X-ray microtomography (PPC-SRμCT). This technology was introduced to reveal the presence of fossils in fuzzy amber 30 and has since been applied to make virtual dissections visualizing internal structures at a resolution as high as 0.7 μm 28,29,60-63 . Here, we successfully applied PPC-SRμCT for obtaining a resolution of ~ 0.2 μm. This allowed resolving and understanding cell structures of the fungus, which was necessary for a reliable generic placement. It should be noted that this study was initiated with the discovery of the fossil insect specimens and the associated fossil fungus. Subsequent study of extant species revealed the two new extant species of Columnomyces. This is a reversed order of discoveries, compared to the more frequent situation where extant species are described prior to their related fossil congeners. What contributed to a great extent to this paper was the study of dried, pinned specimens preserved at CMN and CMPR, which revealed two undescribed extant and host-specific species of Columnomyces. In total, 115 specimens of Proptomaphaginus hispaniolensis were screened, thirteen of which were found with thalli of Columnomyces hispaniolensis (parasite prevalence of 11.3%). Similarly, 136 specimens of Proptomaphaginus puertoricensis were screened, of which seven were parasitized by Columnomyces peckii (5.1%). In recent years, several papers have made a case to increasingly use natural history collections for biodiversity research 6,11,64,65 . Specimens in plant herbaria can be carefully screened for the presence of fungal associates such as downy mildews and rust fungi, leading to identification based on morphology and DNA, descriptions of new species, and studies of host-parasite dynamics 66 . Likewise, museum insect collections are being studied for the presence of fungal ectoparasites, resulting in species descriptions 22,67-70 -as also illustrated in our work, studies of host usage patterns 3 , and increased understanding of geographic distributions of the parasites through time 71 .

Data availability
The data concerning the tomographic-reconstructed slices and segmentations are publicly available at the ESRF online paleontological database (http://paleo .esrf.eu).