Results and Discussion

The method of composite imaging was adopted successfully from palaeobotany (Bomfleur et al. 2007) for documenting fossils from Solnhofen-type lithographic limestones (Haug et al. 2008). As the specimen is documented at high resolution under a microscope, small details, not always resolved via macrophotography, can be visualised, and information from small specimens becomes available (Haug et al. 2008). Because the specimen is usually larger than the field of view, several overlapping images are taken to document the complete specimen in the x- and y-axis in order not to omit any possible details of interest. For moving the specimen in the x- and y-axis, placing the specimen on a movable platform (present in every microscope) is advantageous, as the later process of combining the images is much faster if one need not rotate the images, but only move them. Additional to moving the specimen in the x- and y-axis, it is necessary also to take several images in the z-axis, i.e., to produce an image stack, as under the high resolution the flattened fossils of the lithographic limestones also show substantial relief and, therefore, are not completely sharp in one focal plane.

Stacks of images can easily be produced by hand using a typical mounted camera (Haug et al. 2009), but it is much faster with an automated stage, where one only has to define the upper and lower border, while the microscope automatically takes images in defined distance steps in z-axis within this defined range. An alternative for accelerating the process of producing stacks without having an automated stage is using a camera that is capable to take one image after a defined time. We used a DCM 500 ocular camera, which was set to take an image every two seconds, while the focus was shifted after each image by hand. Compared to the automated stage this is much more labour-intensive, but compared to taking each image by hand using a usual mounted camera (see Haug et al. 2009) the method with a time setting is much faster. The resulting fused images have a very high depth of field and are fully in focus.

The fused images then have to be combined in the x- and y-axis. Adobe Photoshop provides an automated tool for this purpose called photomerge. This tool is more reliable in newer versions (CS3) compared to older ones, but still is not able to faithfully combine many images produced for one specimen. Nevertheless, it can be used to combine images to stripes of up to about a dozen images. These stripes then have to be combined further by hand. Alternatively, the whole process can be done by hand also in Photoshop or freely available alternatives, e.g., GIMP.

The result is more or less a virtual specimen, which can be studied in detail later. Also detail images for publications can be extracted directly from the resulting composite image. This method is, therefore, extremely useful for material on loan, especially if obtained from private collections, or for providing direct virtual access, i.e., via databases.

Under normal light, i.e., using a stereomicroscope, it is important to provide homogenous lighting. A ring lamp proved to be the best choice for this purpose. Normal lighting has an advantage over the fluorescence methods because it gives a better impression of 3D structures. Specimens with a high relief are best documented under these settings (cf. Figure 4.1); such specimens are usually too large to be put under the fluorescence microscope anyway. For smaller specimens the use of fluorescence for contrast enhancement is very important. Small details as, for example, the flagellimeres of the antennulae, can only be seen under fluorescence (Haug et al. 2008) (Figure 1.1–2).

We were pleased to discover that specimens that were known to have an indistinct or weak fluorescence under UV, e.g., specimens from the Zandt lagerstätte close to Solnhofen and from Lebanon, show fluorescence under green light (Figure 2.1, Figure 3). The specimens glow orange when exposed to green light of 546 nm wavelength. The contrast is slightly weaker than in specimens demonstrating UV fluorescence. When UV fluorescence can be applied, the matrix appears black, whereas in specimens exhibiting green-orange fluorescence the matrix appears only dark grey (Figure 2.1, Figure 3). Nevertheless, it enhances the possibilities to spot tiny details, and especially for fossils from Lebanon, the enormous difference between normal light and green-orange fluorescence is apparent. For the specimen of ?Sculda sp. the exopod of the uropod is seen under normal light only as a simple stripe, whereas under green-orange fluorescence the teeth on the outer margin become visible (Figure 2). An advantage of using green-orange fluorescence is that, in contrast to UV light, minute dust particles always present on fossils do not glow under green light. In cases when the specimen exhibits fluorescence both under UV light and green light, the use of green light may be preferable for this reason.

Combination of fluorescence images of the part and counterpart can easily be applied, as the contrast to the matrix allows the virtual extraction of the substance of the counterpart of the specimen and the addition of this substance to the part. In the case of the Sculda spinosa specimen mainly the shield and the exopod of the left uropod become much more complete compared to the single images (Figure 1.4–6).

The combination of different composite images further enhances the visualisation of all information from a specimen. As already stated, normal light settings provide a better impression of 3D structures, whereas UV fluorescence makes small structures visible. The combined image of the small specimens probably representing a juvenile of Cancrinos claviger compared to the exclusively normal light image and the exclusively UV-fluorescence images demonstrates that the combined image apparently mediates the best impression of the whole information of the specimen (Figure 1.1–3).

Confocal laser scanning microscopy has become a very important tool for inferring morphologies in Recent animals (Zupo and Buttino 2001; Buttino et al. 2003; Michels 2007) and has been successfully applied to fossil material (Chi et al. 2006; Chen et al. 2007). The antennulae of the ?Sculda sp. specimen were documented under the cLSM. The available cLSM could not perform UV light, therefore only orange-green-fluorescing specimens could be tested. The second candidate, the specimen of Sculda pennata from Zandt, is very flat, as is typical for fossils from this area, and contains no 3D information. Thus, the ?Sculda sp. was the only suitable specimen for applying this method. Most structures were simply too large and only some annuli of the antennulae could be documented. The three-dimensional information of these structures is rather limited, thus the result does not provide significant additional information (Figure 4.5).

A simple way for documenting 3D information is stereo imaging. We further processed these stereo images with programs (MeeSoft, Structure from Motion (SFM), see above) that are freely available and easily applicable. SFM is again a step towards producing "virtual specimens." The first advantage is that the depth impression can be varied. When comparing the original stereo image to that of the SFM-model in the same position, the stereo image of the SFM-model appears to have a much deeper impression. This can be used to improve stereo images, which had been documented under too small an angle (Figure 4.2, 4.4). It is, in principle, possible to rotate the calculated 3D model. When the original image is rendered directly onto the surface of the model, the impression is quite satisfying (Figure 4.3). When looking closer, smaller program-based distortions can be recognised (Figure 4.3). Thus, the algorithm appears to be as yet imperfect, but is adequate for specimens like those documented here. As the method is simple to apply using a stereomicroscope with a mounted camera and as the software is freely available, this method has the potential to become widespread for producing fast and cheap "virtual specimens."

CT scanning has become a popular method for investigating fossils, in some cases with very impressive results (Donoghue et al. 2006; Tafforeau et al. 2006). The CT scans of our fossil did not produce satisfying results. This was mainly due to the low resolution. Although we were allowed to use one of the most modern medical CT scanners worldwide, the resolution was not high enough to resolve the small structures of interest. As the specimen is embedded in a very large slab (Figure 4.6), it was not possible to place it into a µCT-scanner, which can resolve structures of less than 1 µm in size. Such scanners have provided interesting information on fossil specimens from related Lagerstätten of the Crato Formation (Grimaldi and Engel 2005). The results at least show the potential to extract 3D information from lithographic limestone fossils with the aid of CT scans, as the fossil itself can clearly be distinguished from the matrix, for example one of the legs is recognizable (Figure 4.7–8, marked by arrow). CT-scanners with higher resolutions facilitating the investigation of such large specimens will be tested in the future.

EDX analysis was performed mainly as a by-product. The original idea was to test whether the method of enhancing the contrast of fossils with SEM using back-scattered electrons described by Orr et al. (2002) for organically preserved fossil ostracods could be applied to fossils from the lithographic limestones. The result was not successful.

EDX analysis showed significant differences in element composition between the fossils and the matrix. While both the matrices and the fossils contained oxygen, carbon and calcium, both fossils, but not their matrices, contained phosphorus, about 6% in the Solnhofen specimen and about 14% in the Lebanese specimen (Figure 5). This interesting initial result has to be investigated further. It is not clear whether this finding can explain the fluorescence capacities of the fossils, especially as the ?Sculda sp. shows orange-green fluorescence, while the undetermined caridean shrimp shows UV fluorescence. Tischlinger (2002) already pointed to the fluorescence capacities of calcium phosphate, but he referred to bones and did not discuss the composition of fossil arthropod cuticle.