MicroCT as a Useful Tool for Analysing the 3D Structure of Lichens and Quantifying Internal Cephalodia in Lobaria pulmonaria

: High-resolution X-ray computer tomography (microCT) is a well-established technique to analyse three-dimensional microstructures in 3D non-destructive imaging. The non-destructive three-dimensional analysis of lichens is interesting for many reasons. The examination of hidden structural characteristics can, e.g., provide information on internal structural features (form and distribution of fungal-supporting tissue/hypha), gas-ﬁlled spaces within the thallus (important for gas exchange and, thus, physiological processes), or yield information on the symbiont composition within the lichen, e.g., the localisation and amount of additional cyanobacteria in cephalodia. Here, we present the possibilities and current limitations for applying conventional laboratory-based high-resolution X-ray computer tomography to analyse lichens. MicroCT allows the virtual 3D reconstruction of a sample from 2D X-ray projections and is helpful for the non-destructive analysis of structural characters or the symbiont composition of lichens. By means of a quantitative 3D image analysis, the volume of internal cephalodia is determined for Lobaria pulmonaria and the external cephalodia of Peltigera leucophlebia . Nevertheless, the need for higher-resolution tomography for more detailed studies is emphasised. Particular challenges are the large sizes of datasets to be analysed and the high variability of the lichen microstructures.


Introduction
The availability of non-destructive 3D examination technologies and increases in personal computer hardware and software enables various 3D graphical applications, such as surface visualisations of biological specimens. This approach indirectly recalled an old morphological method back to life, namely the investigation of soft part anatomy by a light microscopical serial section analysis [1]. The correlative examination of a small biological specimen offers many advantages in analysing, visualising, and validating results. The topographical correlation of high-resolution areas to the morphology of the whole specimen provides significant benefits for understanding the overall organisation of the sample [2].
Lichens comprise a large morphological diversity and appear as one organism, but in fact, they represent a symbiosis of at least two primary partners, a fungus (mycobiont) and an alga (photobiont), and many additional microorganisms of still unknown function [3][4][5][6][7]. In most cases, the fungal partner forms the major part and outer side of the lichen thallus and is often judged to be the predominant symbiont [8]. It positions the photoautotrophic algal partner in such a way to ensure optimal illumination and an increasing water supply while still facilitating a gas exchange, especially in the highly developed foliose and fruticose lichens. The mycobiont produces lichen secondary substances to protect the photobiont against damage caused by high irradiation or UV light ( [8][9][10][11] and references therein). The non-invasive determination of the internal morphology and anatomy of these fascinating life forms will help to understand their organisation and functioning better. Species recognition has important implications for assessing the diversity and ecological and biogeographical patterns [12]. Based on phylogenetic and population studies, many cases have shown that numerous distinct lineages may be hidden under a single species name [12][13][14][15][16][17][18]. In these cases, microCT may be a beneficial tool, providing a complementation to other morphological examinations. It allows the visualisation and analysis of internal lichen microstructures, which could not be observed using routine light microscopy alone or can only be seen partially (with the need of preparing cross-sections).
In this proof of principle study, we intend to demonstrate the workflow for examining external and internal structures of lichens using datasets obtained from microCT. We used six samples of lichens from different genera and various growth forms: Bacidia rubella (Hoffm.) A. Massal., Evernia divaricata (L.) Ach., Hypogymnia physodes (L.) Nyl., Lobaria pulmonaria (L.) Hoffm., Peltigera leucophlebia (Nyl.) Gyeln., and Xanthoria parietina (L.) Th. Fr. The two lichens Lobaria and Peltigera have a green alga as the primary photobiont and incorporate nitrogen-fixing cyanobacteria into gall-like structures, the cephalodia. These are located either on the thallus surface (external cephalodia) or in its interior (internal cephalodia) [19]. Cephalodia have an essential ecological function: incorporated cyanobacteria fix atmospheric nitrogen, leading to a relatively high organic nitrogen content and, thus, enabling the colonisation of extremely oligotrophic habitats [9,20]. Consequently, these lichens contribute significant amounts of organic nitrogen to these ecosystems [21][22][23]. Lobaria pulmonaria is a tripartite lichen that contains the green alga Symbiochloris reticulata (Tschermak-Woess) Škaloud, Friedl, A. Beck & Dal Grande as the primary photobiont and Nostoc sp. in the internal cephalodia [24][25][26][27]. However, the structure and amount of these internal cephalodia are, so far, unknown. For this reason, we selected the internal and external cephalodia of L. pulmonaria and P. leucophlebia, respectively, for further analysis and quantification in this proof of principle study. The volumes of the segmented cephalodia of L. pulmonaria and P. leucophlebia relative to the size of the whole thallus were measured for the first time. Moreover, the shape and size of the internal cephalodia of L. pulmonaria were examined.

Taxon Sampling
We used six samples of lichens from different genera representing various growth forms: crustose, foliose, and fruticose (Table 1).

CT Acquisition
Lichen samples (Table 1) were stained with 1% iodine in 100% ethanol for approximately 24 h to enhance the contrast [28]. Dried lichen thalli were then directly glued with contact cement to a glass holder or put into a gelatin capsule glued to the glass holder. CT acquisition was performed with a Phoenix Nanotom M CT scanner (GE Measurement and Control, Wunstorf, Germany) using a voltage of 80 kV and a current of 325 mA while using a tungsten ("standard") target. Projection images (1700-1800) were taken during a 360 • rotation at a total duration ranging from 71 to 90 min. Voxel size ranged from 1.2 to 1.6 µm. The 16-bit dataset generated by reconstruction (with phoenix datos|x 2.2 software) was cropped and converted to 8 bit at a histogram adjustment prior to the other graphical processes.

Image Processing Using Amira Software
Data examination and analysis was performed using AMIRA ® software v. 6.4 (Thermo Fisher Scientific, Hillsboro, OR, USA). It is a powerful, multifaceted software platform for visualising, manipulating, and understanding research data from many image modalities, including CT techniques [29]. The tools we applied are indicated in italics with a capital letter.

Visualisation of Lichen Structure Performed by Volume Rendering and Clipping Tools
We used OrthoSlice and Volume Rendering (Volren) modules to visualise the lichen image stack. OrthoSlices were combined with a specifically adapted Grayscale Colormap with standard grey values and a transparency function where the opacity can be set between 0 and 255. The adjustment of transparency allowed for the observation of internal lichen structures. The Volren module used for volume rendering provided a 3D visualisation of the whole sample dataset. The Volren module provided several options for adjustment: shaded and classical texture-based Volume Rendering (VRT), Maximum Intensity Projection (MIP), and Digitally Reconstructed Radiograph (DRR) (for more details, see reference [29]). To find the most appropriate one, we tested all the techniques and finally selected the option of VRT. Its virtual lighting effects enabled a better visualisation of fine details, which fit for our lichen objects best. The Specular lighting effect was enabled (Figures 1-4) as the most suitable for better visualisation of the thalli surface.
We applied the Clipping tool to examine the internal structures of the lichen thalli. We could quickly explore the data images by screening through multiple orthographic sections, which provided us with easy access and observation of the internal parts of lichen thalli. After Clipping, the Volren image (3D) was complemented with OrthoSlices (2D) for a better visualisation and examination of the internal surfaces ( Figure 3).

Segmentation
Segmentation (separation of parts of the lichen tissue) was the most time-consuming step. It was mainly performed following the procedures outlined by Ruthensteiner [1], including a description and explanation of the Segmentation tools. A further detailed description of the applied techniques can be found in reference [29].
As an initial step of Segmentation, we applied Thresholding to separate the object (lichen tissue) from the background (air) by classifying the voxels into the foreground and background voxels based on their intensities or voxel values [29].
Segmentation of the internal and external lichen structures was performed manually (subjective decision), since we could not acquire an automated workflow. When trying automatic tools, some parts of interest were not included, or vice versa; the false attributes were labelled as well. Thus, at manual Segmentation, approximately every 10th section was labelled and followed by Interpolation to facilitate the segmentation process by obtaining relatively smooth material surfaces right away [1]. When necessary, we carried out additional correction steps by labelling missed parts followed by an Interpolation step until the complete labelling of the selected structure was achieved.

Measurements of Segmented Structures Performed by Label Analysis
After Segmentation, we applied Label Analysis to measure the volume of the whole lichen sample and the segmented cephalodia. Label Analysis computes a group of measurements on each component (Material) of the input label field dataset and calculates the resulting values [29], such as Volume 3D: Volume of the component (mm 3 ). After Label Analysis, we calculated the cephalodia volume relative to the whole thalli volume.

Morphological Observation
The microCT volumes allowed the detailed observation and examination of both the external and internal components of lichen thalli. The majority of lichens, includ-ing many crustose species, develop internally stratified thalli: upper cortex, photobiont layer, medulla, and lower cortex [19]. MicroCT resolved the overall structure of all the lichens investigated in this study ( Figures 1A,C,E, 2A,C,E and 3). Figure 3 shows virtual sections of the stratified foliose thalli of Xanthoria parietina and Hypogymnia physodes differentiated in the upper cortex, medulla, and lower cortex. For Xanthoria parietina, the structure of the fruiting bodies (apothecia) is discernible in Figures 1A,B and 3A. In Hypogymnia physodes, the typical hollow part between the medulla and lower cortex is prominent ( Figure 1C,D and Figure 3B). Virtual sections through the foliose Lobaria pulmonaria ( Figure 1E,F and Figure 3C) and Peltigera leucophlebia (Figures 2A,B and 3D) reveal that the lower surfaces of these species are covered by hyphae forming a dense, feltlike, woolly mat, the tomentum. A cross-section through the thallus of the fruticose Evernia divaricata shows the characteristic loosely woven medulla ( Figures 2C,D and 3F). Figures 2E,F and 3F illustrate a section through the thallus and apothecium of the crustose lichen Bacidia rubella.

Segmentation Results and Label Analysis
After the Segmentation process, we applied Label Analysis to assess the volume of the thalli samples and segmented cephalodia. Subsequently, we calculated the volume of all cephalodia relative to the whole thalli volume.
The L. pulmonaria sample had a total volume of 2.091 mm 3 ( Table 2) and nine cephalodia of volumes ranging from 0.000046 to 0.004525 mm 3 . The total cephalodia volume of 0.153 mm 3 corresponds to 0.73% of the whole thallus volume. The size of the P. leucophlebia sample had a total volume of 0.200 mm 3 (Table 3) and four cephalodia of volumes from 0.000201 to 0.000873 mm 3 . The total cephalodia volume of 0.0019 mm 3 corresponds to 0.97% of the whole thallus volume. Table 3. Label Analysis of the cephalodia of Peltigera leucophlebia. Volume 3D: Volume of the object (10 −3 mm 3 ).

Pros and Cons in the Visualisation of Lichens Using MicroCT
Previous studies with microCT applications mainly focused on animal or biomedical data, e.g., references [30][31][32][33][34]. For lichens, examination using X-ray tomography was done very rarely-namely, to estimate the relative mass contribution of lichen symbionts in Stereocaulon sorediiferum Hue [35]. MicroCT analysis is a beneficial tool in detailed morphological investigations. For visual examinations, a variety of settings provided by Amira software allow changing or tweaking the appearance of different components of lichen thalli. MicroCT can also be combined with a light microscopic technique [31], complementing the examination of morphological characteristics.
Most importantly, microCT imaging allows the non-destructive exploration of both the external and the internal structures of lichen thalli simultaneously. To this end, we used Volume Rendering (Volren) by applying the Grey Colormap, VRT, and Specular lighting effects. Moreover, the adjustment of transparency ( Figure 4C,D), selective imaging of individual components using Segmentation (Figure 4), or clipping off parts of surfaces (Figures 1-3) help to explore the internal lichen structures.
Consequently, a microCT analysis improved the understanding of the structural and functional features of lichens. Separation of the main morphological components depicted in Figures 1-3 (viz., cortex, medulla, and cephalodia) was possible but only related to the fungal component. Thus, the resolution was insufficient to visualise the algal layer with the current setup and depict the algal cells, which appeared as a continuum with the upper cortex. Other approaches, such as staining and embedding (in resin) or applying a higher resolution to smaller thallus areas at microCT acquisition need to be tested for a more detailed examination of the algal cells. These techniques may lead to an enhanced resolution of the microstructure.
Some lichens are very hardy organisms and can even survive conditions in outer space [36]. It is therefore likely that they can withstand the X-ray treatment in the CT instrument and resume growing afterwards. If this proves right, it will pave the way for developmental studies of 3D visualisation of one and the same lichen thallus at different points in time. Lichen developmental studies would greatly benefit from such investigations.
The microCT technique may also advance the knowledge on trapped particles or lichen secondary metabolites-often deposited in crystals-in the thallus. Given the voxel size of 1.25-1.65, particles/crystals larger than 4 µm (2.5× voxel size) can routinely be detected. The MicroCT analysis of these components will be useful for ecological studies investigating the effect of the lichen secondary metabolites, e.g., in herbivory, or the amount of substrate erosion by particle uptake.

Segmentation and Statistical Analysis of External and Internal Cephalodia
Most cephalodia are morphologically simple globose, sacculate, lobate, or coralloid structures that differ from the rest of the vegetative thallus by their photosynthetic partner and, thus, coloration [19]. The number of cephalodia is ecologically important, because the incorporated cyanobacteria bind atmospheric nitrogen, resulting in a com-paratively high organic nitrogen content of the lichen thallus and, in consequence, contribute significant amounts of organic nitrogen to forests and, especially, to oligotrophic ecosystems [21][22][23]. Cornejo and Scheidegger [25] studied the ontogeny of cephalodia structures in L. pulmonaria, indicating the active incorporation of the cyanobacteria into the lichen thallus. The application of the Segmentation tool allowed us to examine formerly hardly accessible internal cephalodia. Thus, using a microCT analysis, we depicted the form of the internal cephalodia of L. pulmonaria for the first time. Our study could clearly demonstrate that L. pulmonaria is characterised by having ± globose-shaped internal cephalodia ( Figure 4B-D). Moreover, we surveyed the external cephalodia of P. leucophlebia as convex and oblong with narrowed ends (so-called peltate cephalodia), which corresponds to previous studies [37].
To further demonstrate the power of the microCT analysis in lichenology, we disclosed the number and distribution of cephalodia within/on the surface of the lichen sample. This was followed by measuring the volume of all cephalodia and their relation to the whole lichen sample (see above). We could clearly demonstrate that the internal cephalodia of L. pulmonaria have a volume ranging between 0.046 and 4.5 × 10 −3 mm 3 . In comparison, volumes of the external P. leucophlebia cephalodia are smaller and range between 0.2 and 0.87 × 10 −3 mm 3 . These data considerably complement our knowledge of the morphological and quantitative characteristics of these lichens.
So far, we have performed a manual Segmentation of cephalodia-the most timeconsuming step in whole-image processing. Future studies may include automatised (machine learning-based) approaches, which may significantly improve the workflow in terms of efficiency and accuracy. This will allow for the analysis of larger quantities of lichen thalli, facilitating the comparison of lichens grown under different ecological regimes and allowing to investigate the algal layer.