Abstract

Characterizing fossils and quantifying paleoenvironmental proxies at a detailed scale is a significant challenge. Three-dimensional tomographic reconstructions are becoming increasingly common, and new imaging approaches, such as synchrotron-based fast X-ray scanning and full-field multispectral imaging, now provide the means to (1) describe fossil morphology at a very fine scale, (2) decipher long-term alteration processes, and (3) better identify conservation requirements. These developments have opened new research avenues for the study of complex heterogeneous materials at the intersection between paleontology, biology, mineralogy, geochemistry, conservation science, materials science, and analytical instrumentation.

INTRODUCTION

Fossils have been carefully documented since the time of the Classical Greek philosophers, in particular by Aristotle and his disciples in the 4th century BC. Fossils consist of highly heterogeneous materials. Such heterogeneity results from their original complexity and the short-to-long-term changes they experienced from their death, their deposition, their entire postdeposition history to their discovery, storage, and display (i.e. taphonomy, geological history, and conservation). Ever since the early microscopic observations of petrified wood by Robert Hooke (1635–1703) in 1665, cutting-edge imaging approaches have been used to investigate the fine-scale character of fossils. The past ten years have witnessed a great increase in imaging studies of specimens. Synchrotron radiation (SR) 3-D X-ray microcomputed tomography (μCT) (together termed SR-μCT) can image a fossils' internal morphology in unprecedented detail and so help constrain phylogenetic relationships (Tafforeau et al. 2006). phylogenetic relationships (Tafforeau et al. 2006). In parallel, a suite of spectro-imaging methods have been developed for fine-scale chemical and structural characterization.

These spectro-imaging approaches rely on interactions between light and matter. In addition to absorbing it, atoms scatter an incident X-ray beam (Box 1). Microcomputed tomography relies on reconstructing electron density, or the complex refractive index, from radiographic images that have been collected using hard X-rays. Furthermore, 2-D (and occasionally 3-D) images can be collected over a vast energy range from the hard X-ray to the infrared. Two-dimensional images are formed either by specimens being fast scanned in front of a macro- to micro- (typically a few μm or below) photon beam or by collecting images of the specimen–beam interaction on an area detector. In general, X-ray absorption spectroscopy (XAS), X-ray fluorescence (XRF) and photoluminescence (PL) techniques provide element and species information, derived from the excitation of electrons in atomic core levels, from molecular orbitals, and from the band structure (in the case of metals and semiconductors). Data from vibrational spectral imaging and X-ray scattering or X-ray diffraction (XRD) provides information on the structure, bonding, supramolecular organization, and texture of materials. Here, we review the benefits and limits of these new imaging approaches and see what information can (or cannot) be gleaned from the complementary investigation of fossil morphology, biogeochemical processes, paleoenvironmental conditions, and conservation strategies.

FINE ANATOMICAL DESCRIPTION

Paleontology, as a discipline, was established in the early 19th century following the work of George Cuvier (1769–1832) on comparative anatomy. Descriptions of fossil anatomy are still the main tool for deciphering the evolutionary history (phylogeny) of extinct organisms. Historically, paleontologists have used a variety of photographic techniques to document and enhance the contrast of fossil morphologies, but many fossil forms remain difficult or impossible to describe using conventional imaging methods. Recent developments in synchrotron-based techniques now allow paleontologists to study specimens in a way that provides a detailed view of a fossil's anatomy with no, or only limited, sample preparation.

Based on local contrasts in the absorption of X-rays in heterogeneous materials, noninvasive μCT provides unprecedented access to internal or histological structures of fragile materials, including entire matrix-encased specimens, such as insects embedded in amber. Synchroton radiation μCT phase contrast capabilities allow for finer discrimination of anatomical features than would be possible for laboratory or medical scanners (Tafforeau et al. 2006). Phase contrast brings improved edge detection from the differential refraction of the planar X-ray wave front at internal material boundaries: one can discriminate between similarly absorbing materials far better than before (Bertrand et al. 2012). For instance, μCT has provided insights into the embryology of the earliest animals that lived on Earth more than 500 million years ago (Donoghue et al. 2006); and it has revealed details on a complete gill skeleton of a 325 Ma shark-like fossil, challenging the view that sharks are primitive jawed vertebrates (Pradel et al. 2014; Fig. 1). Micro-CT has also allowed scientists do nondestructive, multiscale, 3-D virtual paleohistology on hominoid teeth (Tafforeau and Smith 2008). Paleohistology has also benefitted from the development of nanotomography, illustrated by Matzke-Karasz et al. (2014) reporting on preserved organelles in 17 Ma ostracod sperm with 25-nm voxel size (a voxel being a 3-D pixel).

Noninvasive mapping of major- and trace-elements using synchrotron radiation XRF (SR-XRF) enables anatomical details of flat fossils to be defined because X-rays can penetrate materials to a depth of a few fractions of a millimeter (~250 and 300 μm for strontium and yttrium, respectively). Recently, SR-XRF mapping provided evidence that Archaeopteryx feathers are not just impressions but are actually remnant body fossil structures (Bergmann et al. 2010). The technique also allowed observation of the entire skull, vertebrae, and rib insertions of a newly discovered ~95 Ma ray-finned fish from Morocco (Gueriau et al. 2014; Fig. 2). While major- to minor-element maps using SEM–energy dispersive X-ray spectroscopy (EDX) requires between one to three hours to resolve millimeter-scale details at a submicron spatial resolution, maps based on major- to trace-elements from SR-XRF (based on the measurement of integrated intensities in preselected spectral regions) can be collected over centimeter-scale regions at 100 μm resolution with a dwell time of 3 ms per pixel, and so take about 30 seconds per square centimeter. A critical improvement in SEM-EDX is fast raster-scanning (or ‘fly-scan’): by this means one can record the full-range of XRF spectra with a count time of milliseconds per pixel, while synchronously collecting complementary XRD and XAS data.

MINERAL ASSEMBLAGES

Particular taphonomic and burial processes will lead to distinct patterns of preservation. These can include isolated bones in sands, highly flattened articulated specimens in shales, void cavities in calcareous nodules, or insects trapped in amber. Fossils mostly consist of biomineralized remains (the mineralized tissues produced by living organisms during their life), although altered biomolecules can sometimes resist long-term diagenesis (Box 2). These minerals are either altered forms of original biominerals (e.g. bioapatite) or result from the replacement of original organic parts during diagenesis. The fidelity of such replication depends on the reactivity of the replacing minerals and on whether replication occurred via permineralization (mineral precipitation from early infiltration and permeation of tissues by ion-charged water) or via authigenic mineralization (biologically induced mineral precipitation during decay; Briggs 2003). By providing innovative opportunities to assess the spatial distribution and properties of mineral phases, synchrotron-imaging techniques offer crucial information on processes that may have affected paleontological specimens after death.

Diagenetic mineral phases may not precisely replicate original biological structures, which can compromise the identification of fossils. An illustration of this is the reassessment of the Ediacaran Doushantuo fossils by Cunnigham et al. (2012). Among these fossils are the most ancient bilateral organisms on Earth, i.e. organisms without radial symmetry. By combining SR-μCT and electron probe microanalysis (EPMA), the authors reinterpreted these structures as artifacts of deposition caused by diagenetic minerals precipitating in the voids of the decayed tissues. Such a critical reappraisal illustrates the need for precise identification of the mineral assemblage in paleontological specimens.

A number of additional synchrotron-based imaging techniques can be used to map inorganic crystalline phases. For instance, scanning transmission X-ray microscopy (STXM) allows mapping the oxidation state and speciation of a number of elements within mineral phases in complex mixed-valence systems, in the soft X-ray energy range covering the L-edges of K, Ca, Ti–Zn, Ga, Ge, As, Se, Rb and Sr (Bernard et al. 2009). Nowak et al. (2005) combined SR-XRF and XRD to document the spatial distribution of diagenetic minerals within petrified wood. By coupling SR-XRF imaging and X-ray absorption near-edge structure (XANES) spectroscopy on the ~500 Ma Burgess Shale arthropods, Pushie et al. (2014; Fig. 3) identified Cu-enriched phases, mostly in the form of chalcopyrite, which the authors postulate to be the remnants of blood-carrying Cu-hemocyanin, a protein with a role similar to the hemoglobin that is produced by some arthropods and mollusks today.

Laboratory taphonomic experiments combined with synchrotron analyses are increasingly being used to investigate the role played by mineral phases in maintaining the structural and chemical integrity of fossils under diagenetic conditions. For instance, using STXM and micro-Fourier transform infrared spectroscopy (μ-FTIR), Li et al. (2014) found evidence that the calcium phosphates that encrust bacterial cells help in the morphological and chemical preservation of these cells during fossilization. Similarly, Picard et al. (2015) suggested that the presence of Fe-rich phases prevents the degradation of organomineral biostructures, even under pressure and temperature conditions typical of low-grade metamorphism.

BIOGEOCHEMICAL SIGNATURES

A significant complication for interpreting biogeochemical signatures within paleontological specimens is the systematic alteration of biomolecules during the fossilization process (Box 2). Spatially resolved synchrotron techniques now allow documenting the biogeochemical nature of fossils.

Lebon et al. (2011) used SR-FTIR microspectroscopy to map the relative contents of, and the molecular structure of, the different constituents of fossil bones over millimeter-scale areas. Characteristic vibrational frequency maps allowed these authors to correlate the distribution of collagen biomarkers with the degree of preservation of histological structures in fossil bones from distinct ages and contexts.

X-ray absorption spectroscopy (XAS) experiments at the carbon K-edge (i.e. at the energy necessary to excite electrons from the K shell of carbon atoms) can characterize the chemical markers of ancient biomolecules (Fig. 4). Cody et al. (2011) found evidence, via STXM experiments, for the presence of organic molecules that they identified as (partially) preserved chitin–protein complexes within ~310 Ma and ~420 Ma arthropod cuticles. Based on a similar approach, Ehrlich et al. (2013) demonstrated that chitin was preserved within ~500 Ma Burgess Shale sponges. Such molecular preservation is remarkable considering that this shale had been buried to depths of ~10 km. Even more remarkable is the persistence of partially degraded sporopollenin compounds within ~230 Ma lycophyte megaspores that had experienced intense metamorphism at burial depths of about 40 km (Bernard et al. 2007).

X-ray photoelectron emission microspectroscopy (X-PEEM) is a technique comparable to STXM (the electron current emitted from a sample surface being related to X-ray absorbance; Fig. 4). By performing X-PEEM experiments at the carbon K-edge, Boyce et al. (2010) investigated the chemical composition of a partially permineralized ~300 Ma Lycopsid tree. In addition to reporting the signature of ancient biomolecules with no homologue among living plants, they observed a prevalence of aromatic carbon molecules in all tissues investigated resulting from the degradation of original aliphatic compounds during burial-induced processes.

By offering the ability to detect in situ the molecular components within fossils, SR-based imaging techniques led to the conclusion that biogeochemical remnants of ancient biomolecules may be preserved in rocks even after they have experienced intense metamorphic conditions (e.g. Bernard and Papineau 2014). However, the study of biosignatures is not restricted to organics. For instance, based on SR-XRF mapping and XANES spectroscopy at the copper K-edge, Wogelius et al. (2011) proposed a coloration pattern for the ~125 Ma fossil bird Confuciusornis sanctus.

PALEOENVIRONMENTAL INFORMATION

Reconstructing ancient environments and climates requires a range of indirect proxies to be used, including the biological or physical imprints archived in rocks and fossils. For instance, isotopic compositions and trace elemental contents in ocean sediments may provide information on ocean salinity, temperature, and biological productivity; lake sediments, speleothems, and tree rings may record information related to atmospheric temperatures; ice cores are used to estimate paleoatmospheric gas levels. In addition, depending on the extent of diagenetic degradation, isotopes and trace elements in bioapatite from vertebrate hard tissues (bones, teeth, and scales) may provide significant information about paleodiet, paleoclimate, and paleoenvironment.

Although isotopic information cannot be obtained from X-ray methods, high spatial resolution and high signal- to-noise ratio SR-XRF spectroscopy allows semiquantitative mapping the distribution of trace elements at the micrometer scale (Gueriau et al. 2014). For instance, SR-XRF raster-scanning allowed detection of oxidized and reduced As species in 2.72 Ga fossil stromatolites from Western Australia (Sforna et al. 2014). This observation led the authors to propose that As cycling widely occurred on Earth before the oxygenation of the atmosphere, with As(III) oxidation and As(V) reduction being caused by microbes living in permanently anoxic conditions. Using the same technique, Frisia et al. (2012) demonstrated that microbial activity may control/modify P concentration within speleothems, thereby highlighting the caution needed when interpreting climate proxies.

Microscale mapping of strontium over unprecedented sample areas of fossil fish otoliths provided evidence for fluctuations in Sr concentration from the core to the periphery (Cook et al. 2015; Fig. 5). These fluctuations provide essential information on the fish's living environment. Combining synchrotron micro XRF, XAS and XRD, Gueriau et al. (2015) mapped the distribution and speciation of rare earth elements (REEs), in particular cerium, in exceptionally preserved, Cretaceous-aged fossil fishes and shrimps from Morocco. Cerium adopts the +III and +IV oxidation states (most other REEs are purely trivalent with similar reactivity and transport properties), and this led to a depletion/enrichment of Ce compared to other REEs. This ‘cerium anomaly’ is used as a key proxy to assess paleoredox conditions. Therefore, microscale synchrotron mapping and direct probing of the speciation of trace elements open new avenues to better describe and constrain, or even challenge, paleo-environmental proxies.

SPECIMEN CONSERVATION, SAMPLE PREPARATION, AND RADIATION DAMAGE

The extraction, transport, storage, study, and exhibition of paleontological specimens constitute dramatic environmental changes for those specimens. For example, drying, exposure to oxygen, and humidity will lead to mechanical degradation (flaking, cracking) and chemical alteration (oxidation, efflorescence, discoloration). Paleontology benefits from research done by the conservation sciences (e.g. alteration of paints in mural paintings, degradation of stones in historical buildings). For instance, Odin et al. (2015) used sulfur K-edge XANES to follow the degradation of pyritic fossils from the Permian rocks of the Autun Basin (France). In addition to the inorganic sulfide pyrite (FeS2), organic sulfides were found in shale fractions and maceral layers. When exposed to air and moisture, shale-fraction sulfides oxidize to sulfates, whereas maceral layer sulfides remain unaltered. While storage in a cold anoxic environment will slow oxidation processes, it may also induce degradation (as occurs in paint pigments).

For decades, paleontological specimens have been handled without gloves, dipped in ethanol to enhance contrast for observation, or stored in paper, wood, or plastic vessels exposed to numerous sources of contamination. Similarly, the application of consolidants and varnishes significantly impairs the study of fossils as unique scientific archives. Truly noninvasive experiments, such as μCT of whole samples or XRD/XRF macroscanning of compressed fossils, only require the transfer of specimens onto a high-precision rotating stage. Still, specific sample preparations are needed for high-resolution imaging techniques. Only flat specimens can be measured using XRF, XRD, PL, and X-PEEM because the X-rays are extracted from the first tens to hundreds of micrometers only. In contrast, when using STXM, samples have to be transparent to soft X-rays and sample preparation challenges are similar to those required for transmission electron microscopy (TEM).

Embedding samples in organic resins is a common practice to avoid their disintegration during precision sawing, polishing, and handling. However, this procedure leads to organic contamination, which limits the study of any inherent fossil organics. As alternatives, argon-beam milling may be used to sputter the sample surface until nanoscale polishing is obtained and a focused ion beam can facilitate the high-precision extraction of ultrathin sample foils (15 × 5 × 0.1 μm3). These techniques maintain textural integrity even for fragile materials, and prevent shrinkage and deformation of microscale to nanoscale pores. Milling at relatively low ion currents at the final stages of sample preparation efficiently minimizes artifacts due to implantation, amorphization, or the redeposition of sputtered material.

Investigations using ultraviolet (UV) and X-ray light may, of course, lead to radiation-induced changes in a specimen. Even in the absence of visual damage, irradiation may still lead to atomic and molecular changes, resulting in immediate or delayed alteration that can bias future analyses. Organics are highly sensitive to X-rays: amorphization of polysaccharides has been reported at moderate irradiation doses, and racemization of amino acids occurs on the timescale of minutes under high X-ray fluxes (Moini 2014). This duration significantly exceeds counting times in XRF and XRD, but it is comparable to that used for XAS or μCT. Contrasting results were reported regarding the decay of DNA amplification potential in archaeological bones upon X-ray irradiation. Formation of color centers during μCT experiments leads to the blackening of fossil tooth enamel. It can be reversed visually by bleaching under UV light, although occasional darkening may remain (Richards 2012).

Damage formation is a complex phenomenon that is influenced by many parameters such as dose, flux, sample composition and structure, atmosphere composition, etc. (Bertrand et al. 2015). Long exposure favors the appearance of radiation-induced side effects, while high sample absorbance (at low energy or close to absorption edges, and for high electron density) may strongly alter the sample's surface. Decreasing exposure duration (fast shutters, flyscan), optimizing flux, characterizing material behavior under irradiation, and ensuring a proper curation of analytical data are essential measures to mitigate radiation effects on unique or valuable specimens.

CONCLUDING REMARKS

The various synchrotron methods discussed here are well adapted to the sort of multiscale heterogeneous materials that must be studied as is. Capturing information at various successive length scales (100 nm, 1 μm, 10 μm, 100 μm) is crucial for deciphering an animal's taxonomic status and for correlating high-resolution morphologic (evolutionary) information with paleoenvironmental and taphonomic markers.

A potential consequence of this new and exciting field is that it will become harder to get synchrotron beam time. There are currently ~60 synchrotron light sources operating worldwide, sixteen of which entered operation in the past 15 years. Compact X-ray sources are expected to provide increased capacities in the future. Novel synchrotron sources are today being considered or developed to increase X-ray tunability for chemical speciation analysis, to increase the flux rate for micro- or submicroscale imaging, and to more efficiently couple the different types of analytical equipment.

Synchrotron imaging techniques, used in conjunction with the more classic paleontological techniques, therefore, provide a rich source of morphological and chemical clues to the life of the past, and are a means of gathering data on past environments and climates. It can reasonably be expected that the increasing use of synchrotron approaches in paleontology, together with continual technological and methodological improvements, will lead to important new discoveries in the coming years.

ACKNOWLEDGMENTS

The IPANEMA platform benefitted from a CPER grant (MESR, Région Île-de-France). Support from the Research Infrastructures activity IPERION CH (EU Horizon2020, GA n° 654028), the ERC project PaleoNanoLife (PI: F. Robert), the PATRIMA LabEx and within the MNHN/IPANEMA agreement is acknowledged. We warmly thank Dr Alan Pradel (Muséum national d'Histoire naturelle, Paris), Dr Jake Pushie, Prof. Brian Pratt (University of Saskatchewan, Canada), and Phil Cook (IPANEMA) for providing original files for Figs1, 3, and 5, respectively.