Cretaceous ambers are widely distributed in France. Lacroix (1910) carried out a broad survey of the amber-bearing sites, which was recently updated (Nel et al., 2004, Perrichot and Néraudeau, 2014, Perrichot et al., 2007 and Ragazzi et al., 2009), many deposits not being currently accessible. Our observations are based on samples collected from various amber-bearing deposits all over France, ranging in age from Albian to Campanian (Fig. 1, Table 1). According to Nohra et al. (2015), chemical characterization indicates that French Cretaceous ambers are produced by conifers belonging to diverse families.

In most cases, we collected in situ amber samples. We paid particular attention to samples from sites in Charente-Maritime (Archingeay, Cadeuil, Puy-Puy) (Girard, 2010) remarkable for the exceptional preservation of filamentous microstructures.

For comparison, extant samples of Leptothrix were collected from a natural water trickle.

For optical and confocal microscopy, petrographic thin sections of approximately 30 microns standard thickness, covered with a cover-slip, were made.

About 60 thin sections were examined with an optical microscope (Zeiss Axioscope 40 equipped with a photographic device) provided with ×40, ×63, and ×100 immersion objectives. For the images taken at several focus levels, a specific image processing software (Helicon Focus^®) was used.

Application of confocal laser scanning microscope (CLSM) technology in amber provides various advantages complementing traditional optical microscopy (Ascaso et al., 2003). An autofluorescence signal can be emitted by organic molecules preserved in amber revealing poorly distinguishable features. The examination under CLSM of specific planes of depth allows the elimination of any refractive optical illusion. CLSM is also interesting for the three-dimensional (3D) information that may be obtained by acquisition of stacks of closely spaced sections. CLSM techniques were successfully used in the study of microorganisms in amber (Ascaso et al., 2003, Ascaso et al., 2005, Martín-González et al., 2009, Saint Martin et al., 2013, Speranza et al., 2015 and Speranza et al., 2010) or other organic remains such as spider webs (Saint Martin et al., 2014). The CLSM observations were essentially performed on a Leica TCS SP5 microscope (upright and inverted) at the “Université Pierre-et-Marie-Curie”. Some observations were also conducted with the Zeiss LSM700 microscope at the University Aix-Marseille. Samples were observed through 20 × 0.7 NA, 40 × 1.25–0.7 NA and 63 × 1.40 NA oil-immersion objectives. For spectral analysis, the excitation wavelength was 405 nm and the emission was monitored in the 415 to 765 nm range with a band width of 10 nm. The images obtained from confocal microscopy allow viewing of the autofluorescent objects in false colours. 3D reconstruction was obtained by acquisition of stacks of closely spaced confocal optical sections at 0.10 μm to 0.20 μm intervals that guarantee the finest resolution. These images are treated with the open source software Image J (Rasband, 1997–2013) for the optimization of the 3D data.

Examinations with Scanning Electron Microscope (SEM) were performed using three SEM equipments: JEOL JCM-6000 NeoScope mainly for exploration, XL 30 ESEM Philips and Hitachi SU3500 for better magnifications, both of which provide Energy Dispersive Spectrum XRay (EDS) micro-analysis. For this purpose, pieces of amber were broken and the fracture surfaces immediately coated with gold-palladium. This method allows the examination of fresh surfaces with minimal risk of contamination (Girard et al., 2009b). The collected samples of Leptothrix were washed with sterilized water and then vacuum-dried. The dried samples were mounted onto a stub and coated with gold-palladium.

The filamentous networks studied herein occur in different types of amber grains (Fig. 2) as described from other Cretaceous ambers (Girard, 2010 and Speranza et al., 2015). In all cases, they impart an opaque appearance to the invaded portion of the piece of amber. They often constitute thick opaque cortices (up to 1 cm thick) ranging in color from milky white to brown around drop-shaped or nodular grains (Fig. 3A–G). They can also entirely occupy drop-shaped grains or more complex and larger nodules (Fig. 3H–K).

In the case of the presence of cortex at the periphery of the amber grains, the filament network is centripetally oriented. The growth of filaments is therefore oriented from the surface into the core, the density of filaments being higher at the periphery (Fig. 3C, E, G).

In composite nodules, often consisting of several resin flows, the filament network may be partially distorted by the flow. However, it is frequently observed that the filaments cross the resin flow, which demonstrates colonization of a still liquid resin (Fig. 3J–K).

The studied filamentous microorganisms in French Cretaceous amber generally exhibit the consortium of characters previously indicated by the different authors mentioned above, especially regarding the dimensions and shapes of the filaments. Entire filaments can reach up to a few millimeters in length. Light microscope examination revealed that a filament comprises an empty core (lumen) and a cylindrical envelope (coat or “sheath”) that appears like a halo surrounding the lumen. The total diameter varies between 4 and 10 μm, depending on the layers that form the coat/“sheath”. In contrast, the lumen diameter is fairly constant between 1 and 1.5 μm. In detail, the filament can show thickening and constrictions sometimes giving it a puffy appearance. The filaments are characterized by a true dichotomous branching, usually marked by a slight swelling. The branching angles usually range between 50° and 90°. The distance between the branches is quite variable or irregular, depending on the samples and the location in the network.

Despite generally common features, the appearance of the filament network varies optically according to the samples or even within a single amber grain as observed by Breton (2007). The confocal microscopy supports the morphological observations due to the natural fluorescence of some constituents of the preserved filaments and the visualization in space (3D restitution) of filament structures (Fig. 4 and Fig. 5). However, the fluorescence signal can also vary greatly, or even be absent, depending upon the samples. In order to better understand these different cases, we examined numerous samples under SEM. Such examination is an interesting spatial exercise on the theme “to be or not to be” applied to preservation of structures as voids, internal molds, external molds (or imprints) and casts (Fig. 6).

The different methods of observation highlighted the structural elements of the filamentous network, at various scales (Fig. 7). The lumen (lu) clearly and constantly appears as a void (Fig. 7A–K). Around the lumen a sheath-like envelope is constituted (when present) of successive layers (Fig. 7A–E): the first layer (L1) about 5 to 10 μm in diameter, can sometimes be subdivided into several parts only distinguishable in optical and confocal microscopy (L1i: inner layer 1 and L1o: outer layer 1); the second layer corresponds to a thin void (L2) about 0.30–0.5 μm in thickness; the third more external layer (L3) consists of a zone marked by very fine fibrillar elements sometimes delimited outwards by a new thin empty ring (L4). The structural elements can also be marked by imprints. The external surface of L1 and internal surface of L3 shows similar features corresponding to the internal and external L2 structure preserved here only as imprints (Fig. 7F–H). A very important feature, probably decisive, concerns the quasi-constant presence in the L1 layer of radial hair-like ultrastructures of different lengths (hl) (Fig. 7 and Fig. 8), emanating from the lumen and developing around it. It should be noted that these ultrastructures are always clearly visible under SEM appearing as voids and are more rarely highlighted in optical and confocal microscopy. In this case, it is useful to use the information provided by reflection of the laser: the void of the lumen and the hair-like tufts around the lumen structures are generally better evidenced with the reflection mode (Fig. 7 and Fig. 8). The correspondence between radial hair-type ultrastructures (hl) and minute circular voids at the surface of lumen (hli) is generally clearly observable (Fig. 7 and Fig. 8).

Careful examination of several thin sections and amber pieces allows us to distinguish, often in the same network, many cases of preservation and filament development that exhibit either only some of the layers, or the totality of layers described above: •

lumen apparently without envelope or with poorly defined shadow sheath-like envelope (Fig. 3 and Fig. 9);

Optical and SEM investigations revealed the presence of minute bubbles varying in size and shape in relation to the filaments (Fig. 9). The bubbles are distinctly defined by a very thin outline (Fig. 9M, O–S), intersecting all visible structures. They can thus occupy a portion of the lumen and the envelope of hair-like expansions or only the external part of the filament. Although generally globular (subspherical) they can take on an oval shape. In some cases, they follow each other along the lumen and form a funnel alignment (Fig. 9K, M, P, R). Their size ranges from a few microns to tens of microns. Similar bubbles were already described from different ambers and interpreted in different ways: hyphal buds (Speranza et al., 2015 and Waggoner, 1994), liquid bubbles resulting from filaments growing in freshwater and trapped by resin flow (Breton and Tostain, 2005, Girard, 2010, Girard et al., 2009a and Girard et al., 2013b) or emanations from metabolic activity (Beimforde and Schmidt, 2011).

Many studies were devoted to modern Leptothrix, one of the most widespread sheathed iron bacteria known from freshwater environments with special emphasis on the nature and composition of the extracellular tubular sheath that consists of a network of polysachharides and protein-rich fibrils (e.g., Emerson and Ghiorse, 1993 and Ghiorse, 1984). More recent publications (e.g. Furutani et al., 2011a, Furutani et al., 2011b and Ishihara et al., 2013) explored the formation of the sheath in cultures of strains of Leptothrix. These authors wonderfully present schemas of the processes of sheath constructions as following: •

globular secretions from the cell envelope;

Interestingly, our observations under SEM and confocal microscopy fit well with this proposed schema for modern Leptothrix except for the last step of Fe deposition. But perhaps iron and manganese were absent in the resin. For extant Leptothrix these elements derive from the environmental water. So the Cretaceous bacteria inside the resin likely had no or just limited access to Fe/Mn.

We assume that different cases of micro and nanostructures observed in our material may correspond in fact to different states of sheath construction. The sheath of modern sheathed bacteria contains saccharic polymers such as glucose, mannose, galactose, N-acetyl-d-galactosamine and N-acetyl-d-glucosamine detected by fluorescence (Chan et al., 2004 and Furutani et al., 2011b). The presence of a natural fluorescence in our material may be explained by the preservation of some of these components. The mapping of fluorescence signal in our material fits well with the morphological features observed in optical microscopy. When present, the first envelope made of successive coats around the core of filament (lumen) may correspond to the intervening space that housed the thread-type structures of Leptothrix. This intervening space is probably a liquid phase acting in the continuous transfer of bacterial secretions from the cell to the sheath layers (Furutani et al., 2011a and Furutani et al., 2011b). Moreover, according to Ghiorse (1984), these thread-type structures contain acid extracellular polymers. Finally, we can consider that the thread-like structures observed in extant Leptothrix correspond to the radial hair-like structures of our fossil material. All these data argue for the initial presence of different substances and secretions whose possible preservation in amber might explain the autofluorescence signal showed in the present study.

Our observations under SEM and confocal microscopy allow us to augment our comprehension of the development of filaments in liquid resin before its solidification and transformation into amber. Modern sheathed bacteria are able to utilize a large variety of organic compounds as sources of carbon: glucose, galactose, fructose, methanol, ethanol, butanol, glycerol, mannitol, acetate, succinate, malate, gluconate, citrate, etc. (Van Veen et al., 1978). Laboratory experiments and field studies carried out by Beimforde and Schmidt (2011) showed that bacteria and fungi actively colonized liquid exudate of Cycas revoluta and resin of Pinus strobus and sheathed bacteria with sheaths of about 10 microns diameter and a lumen of about 1 micron diameter occur in fresh resin of Pinus ellioti in Florida swamp forest. Cretaceous filamentous bacteria could use resin as a favorable environment for growing. We have to keep in mind that amber was initially a resin that experienced successive steps from a liquid phase to a solid one. At the beginning of their development, filaments found resin a natural favorable environment for growing. With ongoing solidification, resin transformed into a hostile environment that challenged the filaments to develop survival strategies. Sometimes we observed an autofluorescent supra-layer (L3) recovering the first envelope surrounding the lumen (Fig. 7B and C). We suppose that this supra-layer represents a supplementary extracellular polymeric substance layer secreted by bacterium in response to stress conditions in order to react against desiccation (Roberson and Firestone, 1992). Actualistic studies of EPS from bacterial biofilms evidenced the fluorescence of different compounds of EPS, such as proteins (tyrosine and tryptophan) and humic acids (Bhatia et al., 2013 and Chen et al., 2003).

We also observed some variations in the dimensions of sheaths when making comparisons to known extant Leptothrix sheath. This may be explained by the various characteristics liquid resin has. Indeed, microscopic analysis of the formation of sheaths of Leptothrix in culture made by Suzuki et al. (2012) revealed that the thickness of the inner and outer layer of the envelope is influenced by the quality of the growth environment: a nutrient-rich medium activates cell metabolism and favors the formation of a thicker inner sheath wall and thicker outer coat fibrils; the sheath observed in culture medium may be twice as thick as that of the natural environment. Liquid resin may be considered a special culture medium that could induce some modifications in sheath dimensions. Ghiorse (1984) thus noted that the forms of sheaths produced by iron bacteria change under different conditions of culture media.

As already explained above, we did not observe iron or manganese impregnation of the sheaths in any of the studied samples. This might be explained by the state of maturity of sheaths. Rogers and Anderson (1976) showed that the process mediating Fe deposition is independent of the Fe concentration in the media; it appeared to be self-sustaining after certain cellular sheath constituents had been synthesized and deposition was effective at the end of the growth phase of bacterium. We may assume that in the case of filaments occurring in amber, the preserved sheath did not achieve its final state of growth and so was unable to initiate Fe deposition.

As explained above, the same filamentous structure preserved in diverse ambers was interpreted as sheathed bacteria or fungi. These two visions generate two different taphonomic schemas. Speranza et al. (2015) considered the filamentous microorganism preserved in cretaceous amber from Spain to be fungal mycelium and proposed a model of progressive hyphal cell wall degradation. These authors stated that “some previously indicated taxonomic features of this microorganism may actually be fossil diagenetic artefacts”. Their image data (light microscopy and SEM) sustaining the three stages of fungi cell wall degradation are very similar to our own image data of French Cretaceous amber which explains in our opinion different stages of bacterial sheath construction. Of particular interest is the question of “empty radial fibrillar structures” of these authors (Fig. 6A, B in Speranza et al., 2005) quite similar to radial hair-like expansions described in this paper (Fig. 6, Fig. 7 and Fig. 8). In both cases the radial structures develop from the void of the lumen; in both cases there can be observed the imprints of the origin of these radial structures appearing as little circular voids on external imprints. Speranza et al. (2005) consider the “empty radial fibrillar structures” to be matrix produced by degradation of structural polysaccharides from the fungal cell wall skeleton appearing after the solidification of resin since they are not infilled with resin. It is generally known that the major components of fungal cell wall are chitin, glucans and glycoproteins that are cross-linked together to constitute a fibrillar network, most of the chitin being located near to the plasma membrane (Bowman and Free, 2006). Because these components are interwoven to form a tangled network, it seems difficult to imagine in our opinion that their degradation could produce radial structures which will cross the whole cell wall as suggested by Speranza et al. (2005). Moreover, the scale of dimensions of these fibrillar structures is a key to the comprehension of what we observe. In our opinion, these empty fibrillar (or more accurately hair-like) structures, represent the external molds of fine morphological features of the filamentous microorganism. As explained above, they may correspond in fact to the thread-like projections produced by a bacterium in order to start the construction of an envelope. The insertion of the hair-like structures is clearly distinguishable in external mold appearing as little round voids of various diameters (300–500 nm) (Fig. 7C and J). Based on their material, Speranza et al. (2005) curiously argue that the peripheral holes in the external surface of the wall correspond to a loss of amber material but they do not really explain this surprising phenomenon. Moreover, Speranza et al. (2015, Fig. 6D–F) present photos of tubular filaments and their external imprints that are identical to our image data (Fig. 6 and Fig. 7). Speranza et al. (2015) consider these tubular filaments to be hypha and note the irregular eroded appearance of the wall. From our point of view, this cylinder (tube) represents in fact the geometric volume of the bacterium structure. In fact, we can easily observe the continuity of amber mater from the sheath to the surrounding amber. Besides, EDS SEM analysis does not show significant differences between the sheath (the preserved fungal cell wall for Speranza et al., 2015) and the surrounding amber (Fig. 11). It is a peculiar reasoning to invoke the loss of material (amber): if this really is the case, we must question where the lost material has gone.

The irregular external appearance of filaments suggests a close similarity to the outer coat of modern Leptothrix sheath (Fig. 12) that present tightly agglutinated thin fibrils (Furutani et al., 2011b, Hashimoto et al., 2015 and Ishihara et al., 2014). The SEM image data from French amber are also much closer to those presented for the sheathed bacterium L. resinatus by Schmidt and Schäfer (2005) and Beimforde and Schmidt (2011), who interpret the rings of small holes as morphological sheath features.

Generally, the fossil world is revealed by various modes of preservation in rocks: impressions, inner mold, outer mold, casts, all resulting in a set of volume, negatively (holes) or positively (casts, etc.). It is also the case for amber that exhibits the same type of preservation processes. For instance, examination of amber samples under SEM reveals the molding process (Fig. 13) allowing the preservation of very fine structures of various microorganisms such as dinoflagellates (Masure et al., 2013) or actinomycetes (Néraudeau et al., 2016 and Saint Martin et al., 2012). It is this fundamental aspect that strongly separates the two taphonomic stories.

The preservation of the filaments in amber as observed today represents various moments of a complex process involving successive or interactive constraints: •

biological, in relation to the vital functions of living organisms;

An essential taphonomic aspect which is very difficult to assess concerns the subtle interactions between the resin and the filaments developing inside: how to produce potential osmotic exchanges and what is the result; what is the role of resin impregnation on the structures that the microorganisms implement and, conversely, how extracellular productions such as EPS diffuse from microorganism into the resin itself. Deciphering these intimate phenomena would better contribute to understand the autofluorescence signal. The story continues with burial conditions in the amber-bearing sediments and diagenetic processes that affect them. The complexity of all possible interactions during the whole process partially explains the diversity of preservation estimated from different observation techniques. However the scenario proposed here allows the better integration of all these uncertainties, considering that contrary to assumptions of Speranza et al. (2015) the observed morphological features are not diagenetic artifacts, but correspond to fossilized structures. First, there is the growth of mycelium in a liquid medium in which the flow is variable depending on the type of resin production (location, composition), the timing of the beginning of microbial colonization, the external conditions (temperature, humidity) that induce the gradual hardening of the resin. During this time, living cells will start to produce a sheath according to the stages listed above. Relationship and/or exchanges with ambient resin will likely have an impact on the composition of this sheath, which unfortunately is difficult to decipher in the fossilized material.

During the fossilization process of the sheathed bacterium in amber, the original living matter may disappear, thereby leaving voids. In contrast, it was observed that modern Leptothrix cells can leave the sheath when the environmental conditions become unfavorable. Maybe the cells just abandoned the resin-impregnated sheath towards the outside of the resin lump when the resin hardened and moved to nearby fresh resin flows, as suggested by Schmidt and Schäfer (2005). Anyway we observe in amber several kinds of voids: the lumen corresponding to the location of cells, the voids of hair-like structures, the void corresponding to the outer part of the sheath and, when present, the voids of the fibrous structure of the outer layer. The remaining mass around these different voids consists of the original material of the resin that is, it must be repeated, the growth medium for the microorganism. After hardening of the resin, these voids express the frame of the initial morphological architecture. In addition, the missing living matter has also left imprints emphasizing ultrastructural features. This helps to explain the observed continuity of amber material between the different parts of the filament structure and the surrounding amber. Indeed, contrary to a degradation scheme (see Fig. 7 in Speranza et al., 2015), this continuity is everywhere visible and even apparent in the SEM representations proposed by these authors.