The first epiphytic macrolichen and its implication to the interacting with Mesozoic forest ecosystem


 Lichens are well known as pioneer organisms or stress-tolerant extremophiles playing a core role in the early formation of terrestrial ecosystems, of which epiphytic lichens make a distinct contribution to the water-cycle and nutrient cycling in forest ecosystem. But due to the scarcity of relevant fossil records, the evolutionary history of epiphytic lichens is poorly documented. Herein, based on the new material of Daohugouthallus ciliiferus, we demonstrated that the hitherto oldest macrolichen inhabited a gymnosperm branch, representing the first unambiguous Jurassic epiphytic lichen. Combing the fossil and extant macrolichen representatives, we performed the geometric morphometric analysis and comprehensive comparison to infer the systematic status of this rare Jurassic macrolichen. The results declared that D. ciliiferus cannot be assigned to any known macrolichen lineages for its elder age and particular habits, and therefore a new family, Daohugouthallaceae was proposed. This work updated the current knowledge to the historical evolution of epiphytic lichens, implying the macrolichens may have diversified much earlier than the generally accepted K–Pg boundary. In addition, our new finding also provided direct evidence for tracing the continuing joint development of epiphytic lichens and forest ecosystem since the Jurassic of 165 Mya.


Introduction
Lichens are a stable symbiosis composed of fungi and algae and/or cyanobacteria; they also include a diverse microbiome (Spribille et al., 2016;Lücking and Nelsen, 2018;Hawksworth and Grube, 2020). Lichens are components of mostly terrestrial ecosystems from the polar regions to the tropics (Lumbsch and Rikkinen, 2017), growing on all kinds of substrates, including bark, rock, leaves and soil (Belnap et al., 2001;Nash, 2008). Lichens play important roles in ecosystem function, including weathering of rock and accelerating formation of soil (Lindsay, 1978;Chen et al., 2000), xing carbon and nitrogen from the atmosphere (Wu et al., 2011), and as food source for animals (including humans, Cornelissen et al., 2007). Due to their sensitivity to environmental changes, lichens also are widely used as bioindicator of air pollution and environmental health.
The evolutionary history of lichen-forming fungi is poorly understood, due to the sparse fossil record, and has been primarily reconstructed based on molecular dating analyses (Lücking et  . Although these approaches proposed a framework to illustrate how the lichen symbiosis may have evolved, the fossil evidence is indispensable in testing and supplementing the current understandings especially when the earlier fossil was discovered. To date, only about 190 fossils are accepted to represent genuine lichens, and a few are considered ambiguous, i.e. potentially representing lichens (Lücking and Nelsen, 2018). The earliest accepted lichen are two crustose lichens from Devonian fossils, i.e. Cyanolichenomycites devonicus and Chlorolichenomycites salopensis (419-411 Mya), which were inferred to be saxicolous or terricolous (Lücking and Nelsen, 2018). The other earlier lichen is from the Lower Cretaceous, i.e. Honeggeriella complexa (ca. 133 Mya) that was suggested as a squamulose or foliose lichen, although no larger pieces showing its architecture are preserved (Matsunaga et al., 2013;Honegger et al., 2013). Other than this fossil, lichens with foliose and fruticose thalli, the principal forms of macrolichens, have no unambiguous fossil record before Cretaceous-Paleogene (K-Pg) boundary 65 Mya (Lücking and Nelsen, 2018), which represents one of the most dramatic turnover events in the fossil record (Renne et al., 2013). Diverse macrolichen fossils including foliose and fruticose Parmeliaceae from Eocene Baltic amber (38-44 Mya) seem to corroborate the fact (Kaasalainen et al., 2017). Actually, some presumed Mesozoic macrolichens were also mentioned from the Keuper formation (230-200 Mya) (Ziegler, 2001), but this work had not received much attention and was overlooked by lichenologists due to the absence of voucher information (Lücking and Nelsen, 2018). Until recently, a study with the convincing evidence demonstrated the occurrence of macrolichen, Daohugouthallus ciliiferus from the Middle Jurassic (ca. 165 Mya), unequivocally provided an opportunity to catch a glimpse of the earlier evolution of macrolichens (Fang et al., 2020).
Considering the diverse ecological interactions with environments especially for the epiphytic macrolichens that were most focused concerning their distinct contributions to balancing the water-cycle and nutrient cycling in forest ecosystem dynamics (Ellis, 2012), the evolution of macrolichens is of particular interest.
Given that macrolichens have evolved in convergent fashion in multiple, unrelated lineages in Basidio-and Ascomycota, it is vital to clarify the position of Daohugouthallus ciliiferus in reconstructing and understanding the evolution of macrolichens. Unfortunately, diagnostic features, such as hamathecium, ascus, and ascospore structure, are not known from this fossil, which renders its exact classi cation challenging. However, techniques such as automated image recognition have allowed to at least analyze morphometric features in a way that allow a quantitative approach to hypothesis testing when placing macrofossils. In the present study, we therefore provide an updated morphological assessment of new material of Daohugouthallus ciliiferus, corroborating it as the rst Jurassic epiphytic macrolichen. And moreover, we employed imagebased, geometric morphometric analysis to compare the fossil with a range of extant macrolichens. In parallel, we used the large molecular clock analysis by Nelsen et al. (2020), which due to the comprehensive sampling offers a much broader framework than other molecular clock studies including lichen formers (e.g. James et al., 2006;Lutzoni et al., 2018;Kraichak et al., 2018), to reassess stem and crown node ages for major clades of macrolichen formers in the Lecanoromycetes, in comparison to the age of the fossil. As a result, we propose a new family, Daohugouthallaceae, for this fossil, and reveal the Jurassic lichens with particular association with the contemporaneous gymnosperm plants, possibly play an important role in the early Jurassic forest ecosystem.

Geometric morphometric analysis
The geometric morphometric analysis of 140 images (Fig. S1) resulted in cumulative values for all the principal components were listed in Table S1. The cumulative eigenvalues for the main axes (principal components) with the cumulative variance of the rst four principal components amounting to 61.1% (Table S1) Fig. 1), the plot combining the rst two principal components (cumulative variance 42.4) showed that the fossil Daohugouthallus ciliiferus, in group 2, appeared morphologically closest to foliose Parmeliaceae, in group 3, including the genera Hypotrachyna and Hypogymnia and two foliose Parmeliaceae fossils (Kaasalainen et al., 2017;Lücking and Nelsen, 2018). However, placement of the fossil within the family Parmeliaceae is not possible, as the inferred stem age of Parmeliaceae is much younger than the age of the fossil (Kraichak et al., 2018;Nelsen et al., 2020). Therefore, a new family, Daohugouthallaceae is proposed, which is tentatively placed in the order Lecanorales given the close morphological similarity to Parmeliaceae.

Molecular clock assessment
We used the detailed molecular clock tree provided by Nelsen et al. (2020) to illustrate inferred ages for selected family-level clades in the Lecanoromycetes that include macrolichens (Fig. 2). Most of these families have stem node ages relative to the macrolichen genera contained therein substantially younger than 100 Mya, including Caliciaceae, Cladoniaceae, Pannariaceae, Parmeliaceae, Physciaceae, Ramalinaceae, Sphaerophoraceae, Stereocaulaceae and Teloschistaceae. The stem node ages of a few families including macrolichens were reconstructed as between 150 and 100 Mya, including Baeomycetaceae, Coccocarpiaceae, Collemataceae, Pannariaceae, Peltigeraceae, and Umbilicariaceae. However, all these families have a crown node age signi cantly younger than 165 Mya, the age of Daohugouthallaceae (Fig. 3). The only macrolichen family older than the fossil is Icmadophilaceae, with an inferred crown node age of approximately 200 Mya (Figs. 2-3). However, members of this family differ strongly in morphology from Daohugouthallaceae (Fig. 3), and its substrate ecology is also different, its members preferring terrestrial such as acid soil or peat (Rambold et al., 1993). In addition, the macrolichen genera within Icmadophilaceae distinctly diversi ed after the K-Pg boundary: Siphula approximately 48 Mya and Thamnolia about 16 Mya (Fig. S2). If these estimates are correct, the Jurassic macrolichen cannot be included in any extant family containing macrolichens, and so the occurrence of Daohugouthallaceae in the Jurassic may re ect a scenario of early macrolichen evolution long before the diversi cation of extant lineages of epiphytic macrolichens. Remarks: Daohugouthallus ciliiferus was rst published almost a decade ago (Wang et al., 2010), and its lichen a nity was recently corroborated based on anatomical characters from scanning electron microscopy (Fang et al., 2020). Due to limitations of the available material, the systematic relationships of D. ciliiferus have not been explored in more detail up to the present. The newly available material allowed to better assess the phenotypic characters of D. ciliiferus, including morphology and anatomy, furthermore, to gather new information about its substrate ecology as the most direct evidence of this oldest macrolichen.

Phylogenetic placement of Daohugouthallaceae
The new family based on the fossil lichen Daohugouthallus ciliiferus is tentatively placed in Lecanorales, although sexual reproductive structures are missing that would allow to test this placement. Ascomata, asci and ascospores are crucial to assess systematic a nities within Lecanoromycetes (Hafellner, 1994) and have ever been demonstrated in fungal fossils as old as 400 Mya (Taylor et al., 2005). Apothecia-like structures seem to be present in the fossil (Fig. 4C), but no structures interpretable as asci and ascospores were detectable. In fact, the thin nature of compression fossils like Daohugouthallus ciliiferus (Fig. 4H)  presumably extracted and ampli ed from fossils as old as 250 Mya (Cano et al., 1993;Fish et al., 2002), but these ndings have been challenged and considered artifactual (Pääbo et al., 2004). Successful DNA extraction from permineralized or compression fossils as old as Daohugouthallus ciliiferus seems impossible and so this is not an avenue that could be followed to clarify the systematic a nities of this and other lichen or fungal fossils.
In lieu of sexual reproductive structure evidence to clarify the potential a nities of Daohugouthallaceae, our geometric morphometric analysis seems a suitable alternative to provide at least a hypothesis based on quantitative data. This approach, based on homologous landmarks or structural outlines (Rohlf and Marcus, 1993), was here apparently used for the rst time in this context but has been widely used in entomology (Bai et al., 2014;Ren et al., 2017). However, it requires a careful approach to data assessment (Fox et al., 2020). The CVA plots (Fig. 1) based on the comparison with homologous landmarks of 59 extant macrolichens and two Parmeliaceae fossils showed that Daohugouthallaceae are most similar to foliose Parmeliaceae (Lecanorales). Given the much younger stem age of the latter (and other related families such as Physciaceae), the introduction of a new, monogeneric family for this fossil therefore seems justi ed.
Too thin and incomplete save status of the new fossil material led to intact and strati ed thallus structures unavailable, but fortunately, Energy Dispersive X-Ray Spectroscopy (EDX/EDS) helped to distinguish the lichen fossil areas, including fungal hyphae and photobiont cells, from the rock areas including rock particles in this study, which showed they obviously differed in the main elements contained and lichen a nity of the new fossil are further con rmed (Table S2;  macrolichens are also known for their ability to explore additional niche spaces (Huang et al., 2019), and gymnosperms such as conifers could have provided suitable new substrata helping the macrolichens away from the constraints of substrate surface as common in microlichens, promoting the better growth and spreading of this kind of macrolichen, thus laying an important foundation for the advance of the earlier evolution of epiphytic macrolichens, especially during the life recovery period just after the end-Triassic mass extinction. The presence of a fossil macrolichen such as Daohugouthallus ciliiferus is therefore not surprising and we expect that more macrolichen morphotypes may be found in this and other formations of the Mesozoic, hopefully expanding the evidence for the diversi cation of macrolichen lineage well before the most recent mass extinction event, the K-Pg boundary.
Interactions of Daohugouthallaceae and Mesozoic forest ecosystem Among the new specimens of Daohugouthallus ciliiferus collected from the type locality, one thallus was found attached to the branch of an unidenti ed cone-bearing gymnosperm fossil. The Daohugou paleoenvironment has been analyzed to be a gymnosperm-dominated forest vegetation (Ren and Krzeminski, 2002;Zhang et al., 2006), and Daohugouthallus ciliiferus has been reconstructed to be epiphytic on gymnosperms due to its association with a small seed cone (Wang et al., 2010), but in the original study the two fossils were not directly connected. Our new specimen (Figs. 4A, E) clearly shows the thallus growing directly on a thin branch of a gymnosperm with an associated cone (Fig. 4D), possibly representing a conifer, suggesting that gymnosperms may have served as substrate for epiphytic macrolichens already in the Jurassic.

Experimental methods
The lichen fossils were examined and photographed using an Olympus SZX7 Stereomicroscope attached to a Mshot MD50 digital camera system. For selected fossil we made cross sections using a stonecutter, one piece was embedded in EXAKT Technovit 7200 one-component resin, then cut using an EXAKT 300CP cutting system. The thin sections were grinded and polished to the thickness of about 20 µm using an EXAKT 400CS variable speed grinding system with P500 and P4000 abrasive papers; one piece was sputter-coated with gold particles using Ion Sputter E-1045 (HITACHI). SEM images were recorded using a scanning electron microscope (Hitachi SU8010) with a secondary electron detector operated at 5.0 kV; one piece was analyzed with a Zeiss MA EVO25 scanning electron microscope under a high vacuum mode by using an accelerating voltage of 20kV. Energy Dispersive X-Ray Spectroscopy (EDX/EDS) spectra were obtained with an Oxford X-act detector. The working distance was kept between 8-10 mm. Acquisition time was set up to 60 seconds for each EDS spectrum. Plates were composed in Adobe Photoshop. Most lab work was performed at the Institute of Microbiology, except the stonecutter was operated at the Institute of Geology and Geophyscis, and the fossil thin slicing and EDX were taken at the Institute of Vertebrate Paleontology and Paleoanthropology. All the above three Institutes are in Beijing and subordinate to Chinese Academy of Sciences. Geometric morphometrics: For this purpose, 140 images (Fig. S1) of 59 representative extant macrolichen species were selected from 12 families and 6 orders of Lecanoromycetes (Table S3), including specimens deposited in HMAS-L, photos provided by Robert Lücking, and pictures downloaded from the CNALH (Consortium of North American Herbaria) Image Library https://lichenportal.org/cnalh/imagelib/ and the Hypogymnia Media Gallery http://hypogymnia.myspecies.info/gallery, among which 13 species had more than 2 samples and images, 15 species had only one sample each but more than 2 images, and 31 species had one image each, together with two images of accepted Parmeliaceae fossils (Kaasalainen et al., 2017), and 25 sub-images cut from the images of Daohugouthallus ciliiferus fossil. The sampling number of images in this study comprehensively considered the quality requirement for the geometric morphometric analysis, representativeness and availability of the discernable topology of thallus lobes or branches. The whole image set was divided into ve groups according to lobes types: microfoliose group, the Daohugouthallus ciliiferus fossil group, a long branches group, a wide-lobed group, and a fruticose group (Table S4). The selected images were twodimensional graphs with two views of the front or back of the thallus where the branch tips were clearly recognizable. To orientate the images in the same direction, they were adjusted so that the end of the branches faced right. Images were named in a uni ed format: growth type-order-family-genus-species (sample number) except for the two selected reference fossil images only corresponding to family name.
The external forms were represented by one curve extracted from the end of branches or lobes and the curve was resampled into 60 semi-landmarks by length (Fig. S4). The starting point of the curve was selected as a point on the upper edge of the lobe or branch near the center or substrate, and after describing the outline of the whole lobe or branch, the end point returning to the lower edge near the starting point. The curves and semi-landmarks were digitized using TPS-DIG 2.05 (Rohlf, 2006). To merge all semi-landmarks into the same data le to produce the data set for morphological analysis, the data le was opened as text le to convert the semi-landmarks to landmarks, by deleting the line with the curve number and point number and replacing the landmark number by the point number (Tong et al., 2021).
MORPHO J 1.06a (Klingenberg, 2011) was used for subsequent analysis of the data set. Through Procrustes analysis, the morphological data of all test features were placed in the same dimensional vector space to screen out physical factors such as size. Principal component analysis (PCA) and geometric modeling of the mathematical space formed by PC axis were used to coordinate the shape changes of the entire dataset. We then selected the data set to generate a covariance matrix. In this context, the rst two principal components corresponding to the highest cumulative variance represent the best variation pattern of test shape. The relationships among different morphological groups were then visualized through canonical variate analysis (CVA).