FIB-SEM Study of Archaeological Human Petrous Bones: 3D Structures and Diagenesis

Abstract

The petrous bone generally preserves ancient DNA better than other fossil bones. One reason for this is that the inner layer of the petrous bone of pigs and humans contains about three times as many osteocytes as other bones, and hence more DNA. A FIB-SEM study of modern pig petrous bones showed that the 3D structure of the thin inner layer is typical of woven bone that forms in the fetus, whereas the thicker outer layer has a lamellar structure. The lamellar structure is common in mammalian bones. Here we study human petrous bones that are about 2500 years old, obtained from three Phoenician sites in Sicily, Italy. A detailed FIB-SEM study of two of these bones, one well preserved and the other poorly preserved, shows that the 3D bone type structure of the human petrous inner layer is woven bone, and the outer layer is lamellar bone. These are the same bone type structures found in pig petrous bones. Furthermore, by comparing nine differently preserved petrous bones from the same archaeological region and age, we show that their collagen contents vary widely, implying that organic material can be significantly altered during diagenesis. The mineral crystals are better preserved and hence less crystalline in the inner layers compared to the outer layers. We therefore infer that the best-preserved DNA in fossil petrous bones should be found in the thin inner layers immediately adjacent to the otic cavity where much more DNA is initially present and the mineral phase tends to be better preserved.

1. Introduction

The petrous bone, also referred to as the inner ear bone, is an unusually dense bone housing the otic capsule [1,2]. Unlike most other bones of the mammalian skeleton, the petrous bone is not subject to direct loading forces. The major function of the petrous bone is to protect the inner ear, and the petrous bone is also thought to participate in sound reception [3]. Fossil petrous bones of mammals, including humans, were found empirically to preserve DNA better than all other tested bones [4]. Petrous bones are therefore a primary source of information for the field of paleogenetics as determined by the analysis of ancient DNA (aDNA) [5,6]. Petrous bone aDNA has contributed significantly to our understanding of human evolution.

Both left and right petrous bones are located on the inner side of the temporal bone (Scheme S1). The main function of the petrous bone is to provide protection for hearing and balance organs, in the cochlea and vestibule, respectively [3]. The cochlea is located at the base of the petrous bone and is connected through the ossicles to the tympanic membrane (eardrum). The otic chamber is enclosed by a thin layer of primary bone with extremely high mineral density (labyrinthine capsule) [2]. Ibrahim et al. [7] referred to the innermost layer as the high mineral density bone (HMD), which in pigs and humans contains about three times as many osteocytes as other bones [8,9]. This layer is formed during the development of the fetus and has a woven bone structure [3,7]. The bone layer that surrounds the inner fetal bone is called the periosteal layer [2]. In modern porcine samples it was shown that the periosteal layer bone is composed of lamellar bone with a plywood structural motif [7]. Ibrahim et al. [7] referred to this layer as the low mineral density bone (LMD). The periosteal layer contains twice as many osteocyte lacunae as the cortical bone from the pig femur [9].

An important property of the petrous bone is that it preserves DNA better than other bones [4,5,10]. This is in part responsible for the enormous progress in the last decade of characterizing ancient animal genomes, including those of hominids. This preservation phenomenon was mostly attributed to the high density of the petrous bone [4,11,12]. Structural studies of modern pig petrous bones do indeed show that the high mineral density, especially of the inner petrous bone, reduces porosity, which in turn may protect the DNA [7,13]. Furthermore the fact that this bone contains two to three times as many osteocytes compared to other bones, means that it initially has two to three times more DNA [8,9]. This too is a factor that contributes to the preferential DNA preservation.

The preservation of the petrous bone in archaeological sites is in itself a fascinating phenomenon. In an archaeological ritual site at Whitton Hill in Northumberland, for example, petrous bones were preserved in 30 cremation pits [14]. It was shown that the petrous bone can withstand high temperatures and is resilient to chemical processes [15]. Furthermore, radiocarbon dating has also been carried out using petrous bones [16]. Little, however, is known about changes in the microstructure of the petrous bone after it is buried (diagenesis).

The questions that we address in this study are as follows: 1. Does the human petrous bone also have an inner layer composed of woven bone and an outer layer of lamellar bone, as was observed using FIB-SEM in the pig petrous bone? 2. Can we detect diagenetic processes that have altered the 3D structure of bone types and the composition of archaeological human petrous bone at the nanometer resolution? To address these two different questions, we analyzed a collection of archaeological human petrous bones from three Phoenician (2500–3000 years old) sites in Italy that were also being radiocarbon dated in our laboratory.

2. Materials and Methods

2.1. Materials

Archaeological human petrous bones (2600–2000 years old) were obtained from 3 different archaeological collections from Sicily (Italy) (Table 1) in order to carry out radiocarbon dating and aDNA studies. From 46 petrous bones, based on morphological preservation 9 were selected for this study and from these, two samples were chosen for detailed 3D structural imaging.

2.2. Methods

2.2.1. Collagen Extraction

This protocol is used for preparing pure collagen for radiocarbon determinations [17]. The outer surfaces were mechanically cleaned. Between 0.5 to 3 grams of bone were crushed and powdered. The bone mineral was dissolved with 0.5 N HCl solution. Distilled water was added and the solution centrifuged at 3000 rpm for 3 min to retrieve the insoluble organic fraction. This fraction was treated with 0.1 N NaOH for 30 min, and the suspension was then centrifuged at 3000 rpm for 5 min. The pellet was washed with distilled water and centrifuged again at 3000 rpm for 5 min. Finally, the pellet was treated with 0.5 N HCl for 5 min and then rinsed with Nanopure water three times (pH = 3).

The suspension was then transferred to glass tubes and then gelatized overnight at 80 °C in a heating block. The next day, pre-cleaned Eezi-filters™ were inserted in the sample tubes and the solution that passed through the filter was transferred to the upper part of a pre-cleaned ultrafilter (Vivaspin 15, 30kD MWC-cutoff; Vivaproducts, Inc., Littleton, MA, USA) [18]. The ultrafilter was then centrifuged at 3000 rpm for 20 min, and the concentrated gelatin that remained in the upper part of the ultrafilter meniscus was transferred to conical Pyrex tubes. A few drops of water were added to the top part of the ultrafilter in order to recover more gelatin. The Pyrex tubes were then covered with aluminum foil with a few pin holes and freeze dried for at least 48 h. The fluffy white powder was weighed and then transferred to glass vials and filled with argon.

2.2.2. Fourier Transform Infrared (FTIR) Spectroscopy

A cut section from each petrous bone was mechanically cleaned with a scalpel, and then a few milligrams of bone was extracted from the cut surface with a scalpel directly into a freshly cleaned agate mortar. Each petrous bone was sampled at two locations; the outer layer bone sample was located at some distance from the otic chamber and the inner layer bone sample was located adjacent to the semi-circular canals. Following [7], these outer and inner layer samples are also referred to as lower mineral density (LMD) and higher mineral density bone (HMD) samples, respectively. The sample was ground and then a few milligrams of spectra grade KBr were added and mixed with the sample, and then pressed (Pike Handpress; Pike Technologies, Madison, WI, USA) into a pellet. The infrared spectrum was obtained using a Nicolet iS5 spectrometer (Thermo Fisher Scientific, Waltham, MA, USA) and the splitting factor (SF) was calculated following [19]. The full width at half maximum (FWHM) of the main absorption peak was measured directly off the spectrum.

2.2.3. Focused Ion Beam–Scanning Electron Microscopy (FIB-SEM)

We acquired 4 FIB-SEM image stacks: one each from the inner and outer layers of sample 10610 and one each from the inner and outer layers of sample 10619. Before FIB-SEM acquisition each polished and sputter coated (10–20 nm iridium) sample was imaged using the Phenom SEM to assign locations for further analysis. The specific location for obtaining each image stack was based on an analysis of the ESB images, paying particular attention to the absence of bone cavities, such as cracks and lacunae. Furthermore, the areas chosen showed aligned patterns in the surface ESB images of the outer layer, and towards the bone surrounding the semicircular canals for the inner layer. The direction of image acquisition, i.e., the direction of milling, was as perpendicular as possible to these lineations.

Two petrous bone samples were selected for structural analysis. A section of bone containing at least one semi-circular canal and a natural forming surface was embedded in a polymer (Epofix) and the surface of interest was ground and polished. The surface was imaged using a backscattered electron detector in the Phenom SEM (Thermo Fisher Scientific) to identify areas for FIB-SEM imaging. Polished sections were sputter coated with iridium (thickness 20 nm) and then analyzed in the FIB-SEM (Crossbeam 550, Zeiss, Oberkochen, Germany). After locating the region of interest in the SEM, the stage holding the sample was tilted to 54° in order to deposit a protective layer of platinum (0.5 µm thickness) over the area of interest (GIS parameters: 30 kV:700 pA, dose factor = 1). The sample was maintained at the same tilt for all subsequent steps. A trench was milled using the FIB (30 kV:35 nA, dose factor = 10). Smoothing the trench was carried out with 30 kV:700 pA and dose factor = 10. When the imaging surface was smooth the SEM parameters were optimized for dual channel acquisition using the following detectors: electron selective backscatter detector (ESB) and secondary electron detector (Inlens). SEM imaging parameters were as follows: voxel size = 9.8 nm, store resolution 1024 × 768 pixels, scan speed (dwell time) = 2, line average for noise reduction N = 140. In all scans, we maintained an isometric pixel size for x, y, and z for milling thickness. In some cases, after a long exposure, charging started to affect the Inlens image quality. In this case the scan speed was reduced to 1 and the number of lines increased to 200. This resulted in a 14 s reduction in scan cycle time and the increase in line averaging for noise reduction compensated the quicker dwell time and provided good resolution images.

2.2.4. Image Processing and Directionality Analysis

Raw data images were loaded onto Dragonfly 3D image processing software (V.2022.2). All stacks of images were aligned using the SIFT protocol for feature-based slice alignment using the built-in plugin in the software. InLens images were subjected to contrast enhancement using CLAHE and for ESB images filtering was carried out using the median filter to obtain uniform thresholding segmentation. This was achieved for all datasets to produce a binary image of the pores [20]. In our previous work on fresh porcine petrous bones, we used the term less-mineralized collagen fibrils to refer to the nanoscopic pores surrounding collagen fibril bundles [7,9]. Although we are using the same terminology in this work, for the sake of coherence, we have to note that this term also includes pores that are the result of bone diagenesis, as is the case with the archeological samples used in this study especially in the badly preserved petrous bone samples. The directionality of the less-mineralized collagen fibrils was assessed using the anisotropy approach. In brief, the binary images obtained from ESB images were segmented using the threshold segmentation option and then each segmented images stack was either divided into two or three sub-stacks (to reduce computation time). The segmented 3D sub-stacks were then analyzed using the eigenvalue approach to produce 3D presentation on the directionality of the less-mineralized collagen fibrils after defining the thickness and clusters diameter of the less-mineralized fibrils. The 3D representation shows the preferred directionality using vectors assigned by different colors. The color code for the vector direction is always encoded in the 3D axis indicator. For more detailed information please refer to protocol described in [7].

2.2.5. Volumes of the Less Mineralized Collagen Fibrils

From each image stack 5 random blocks with identical dimensions (2 × 2 × 2 µm³) were selected for porosity calculations. After segmenting the pores we obtained the percentage of the segmented material from each block and averaged that for each sample to show the difference in porosity between differently preserved bones.

3. Results

The two major aims of this study are different, namely to determine the 3D bone structural types of the inner and outer petrous bones of humans and compare them to the known 3D structural types in pigs, and secondly to obtain insights into possible diagenetic changes that occur in these petrous bones over time. To achieve both of these aims using FIB-SEM, we need to choose one fossil Phoenician petrous bone that is well preserved and will reveal the overall 3D structural types even though it is 2600–2000 years old, and another Phoenician petrous bone that has clearly undergone diagenesis at least at the level of collagen loss that can be compared to the well-preserved bone.

3.1. Determining of the Preservation States

We therefore first surveyed the preservation states of the mineral and matrix (mainly collagen) phases of all 46 petrous bones. From these we chose nine petrous bones for a more detailed study that show a range of preservation states. The properties we measured in these nine samples are listed in Table 1.

Table 1. The properties of the 9 petrous bones from which 2 were chosen for an in-depth 3D structural analysis using FIB-SEM (sample numbers marked with an asterisk). SF—infrared spectrum SF (reflects mineral preservation) [19]. The samples are ordered by the site and for the same site by the Gelatin %. Gelatin (wt.%) is determined after removal of the mineral and then solubilization of the HCl insoluble collagen and filtration to remove minerals. This reflects the preservation of the collagenous matrix. C%: amount of carbon in the gelatin. C/N ratio: carbon/nitrogen ratio reflects how pure the collagen is. pMC: percent of modern carbon 14 reflects the age of the specimen; the lower the pMC, the older the sample. Note that the inner layers of the petrous bone are very thin (less than 2 mm) compared to the outer layers (more than 5 mm). Thus, the values measured in Table 1 mainly reflect the outer layer values. Note that sample 11158 has a higher pMC value compared to other samples as it is an outlier and has a much later date around 1000 AD.

Sample No.	Site	SF	Gelatin	C %	C/N	pMC ± 1σ
(RTD)			wt.%
10610*	Birgi	2.73	13.2	44.26	2.80	78.7 ± 0.33
11184	Birgi	3.42	8.7	43.58	2.74	76.0 ± 0.38
10611	Birgi	2.76	6.8	43.62	2.76	72.9 ± 0.24
11198	Birgi	3.43	2.7	43.25	2.77	73.4 ± 0.22
11199	Birgi	3.30	2.4	46.73	2.78	73.0 ± 0.22
11183	Birgi	3.69	1.3	43.24	2.77	77.6 ± 0.41
11158	Selinunte	3.97	10.2	43.58	2.76	88.5 ± 0.34
10619*	Selinunte	4.16	6.0	43.94	2.78	75.9 ± 0.33
11187	Lilibeo	3.48	1.9	43.3	2.78	77.0 ± 0.20

Table 1. The properties of the 9 petrous bones from which 2 were chosen for an in-depth 3D structural analysis using FIB-SEM (sample numbers marked with an asterisk). SF—infrared spectrum SF (reflects mineral preservation) [19]. The samples are ordered by the site and for the same site by the Gelatin %. Gelatin (wt.%) is determined after removal of the mineral and then solubilization of the HCl insoluble collagen and filtration to remove minerals. This reflects the preservation of the collagenous matrix. C%: amount of carbon in the gelatin. C/N ratio: carbon/nitrogen ratio reflects how pure the collagen is. pMC: percent of modern carbon 14 reflects the age of the specimen; the lower the pMC, the older the sample. Note that the inner layers of the petrous bone are very thin (less than 2 mm) compared to the outer layers (more than 5 mm). Thus, the values measured in Table 1 mainly reflect the outer layer values. Note that sample 11158 has a higher pMC value compared to other samples as it is an outlier and has a much later date around 1000 AD.

Sample No.	Site	SF	Gelatin	C %	C/N	pMC ± 1σ
(RTD)			wt.%
10610*	Birgi	2.73	13.2	44.26	2.80	78.7 ± 0.33
11184	Birgi	3.42	8.7	43.58	2.74	76.0 ± 0.38
10611	Birgi	2.76	6.8	43.62	2.76	72.9 ± 0.24
11198	Birgi	3.43	2.7	43.25	2.77	73.4 ± 0.22
11199	Birgi	3.30	2.4	46.73	2.78	73.0 ± 0.22
11183	Birgi	3.69	1.3	43.24	2.77	77.6 ± 0.41
11158	Selinunte	3.97	10.2	43.58	2.76	88.5 ± 0.34
10619*	Selinunte	4.16	6.0	43.94	2.78	75.9 ± 0.33
11187	Lilibeo	3.48	1.9	43.3	2.78	77.0 ± 0.20

As we intend studying the inner and outer layers of the chosen petrous bones separately using the FIB-SEM, we also made a detailed analysis of the SFs of the two layers separately (Figure 1). Following Asscher et al. [21,22], we plot the SF values against the full width at half maximum of the major peak of the carbonated hydroxyapatite mineral phase (FWHM).

In the plot of the SF versus the FWHM of the main mineral peak (Figure 1), increasingly poor preservation of the mineral phase will result in higher SFs and narrowing main peaks, i.e., leading to smaller FWHM. Figure 1 shows that the mineral phases of the inner and outer petrous layers in these fossil petrous bones are similar. The fact that the highest SFs and the lowest FWHM values are from all eight outer layers, shows that there is a tendency for the outer layers to be less well preserved than the inner layers. This may be attributed to the unusually high density of the inner layers compared to the outer layers [2]. It may also indicate that the outer layers provide protection for the inner layers. The two bones chosen for detailed structural analyses are also marked in Figure 1. In the well-preserved bone (10610), the inner and outer layers have similar SFs, whereas the SFs of the poorly preserved bone (10619) are very different between layers.

Volumes of the less-mineralized collagen fibrils can be calculated from the FIB-SEM stacks. The results are shown in

Table 2. Average volumes of the less-mineralized collagen fibrils calculated from the FIB-SEM stacks after applying thresholding segmentation using the median filter.

Sample	Petrous Layer	Average (vol.%)	Std. Dev.
10610	Outer	1.0	0.2
	Inner	0.7	0.2
10619	Outer	7.3	3.2
	Inner	4.3	0.8

. In the well-preserved bone (10610) the volumes are similar in the outer and the inner layers. In contrast, in the poorly preserved bone (10619), the volumes of these less-mineralized fibrils are much larger. As the calculation is based on thresholding the images, the values do not differentiate between pores and less-mineralized fibrils (both are black in the images). Hence, the significant increase in the volume implies that bone (mineral and collagen) was lost from both the outer and inner layers.

3.2. Petrous Bone Structure Using FIB-SEM

We chose two out of the nine petrous bones identified in Table 1 for detailed structural analyses using FIB-SEM. For the well-preserved petrous bone, we chose sample 10610 as it has the highest amount of preserved collagen and the atomic order of the mineral phase as determined by the infrared SF is the same as modern bone (2.6–2.8) (see Table S1). The C% and C/N ratio are also consistent with well-preserved collagen. For the poorly preserved petrous bone we chose 10619 as the collagen content is half that of the best-preserved bone and it has the highest SF. A higher SF implies that the mineral crystals have undergone more recrystallization. We did not choose the bones with really low levels of collagen for fear that the high-resolution images obtained in the FIB-SEM would be full of voids.

3.2.1. Outer Petrous Bone Layers

The characteristic D-banding of the collagen cannot be identified, but there appears to be two different directions of lineations (Figure 2). In order to decipher the 3D structure, we used the Eigenvector approach applied by [7]. This is based on the relatively easily recognizable less-mineralized collagen fibrils, which in the ESB image are dark lineations or round structures with diameters 50–70 nm when seen in cross section. These less-mineralized collagen fibrils are located on the periphery of the mineralized collagen fibril bundles [23]. Bone diagenesis might well change the properties of these nanoscopic lineations and this became apparent in this study. Identifying the less-mineralized fibrils/pores and determining their orientations will reveal the overall orientations of the collagen fibril bundles. The Eigenvector directionality approach identifies round-shaped dark objects from 50 to 70 nm in diameter, and then searches for the same objects in the adjacent image stack volume in clusters with the user-defined diameter. In this way, the directions of these less-mineralized collagen fibrils which have diameters in the 50–70 nm range and 1–1.5 µm clusters are determined [7]. For the archaeological samples used in this study we used the same strategy for the well-preserved sample; however, adjusting the protocol was needed for the badly preserved bone as the diameter changed from 50–70 to 100–130 nm for the badly preserved bone and cluster diameter increased to 3 µm. In the wellpreserved sample (10610) two predominant orientations can be observed in the outer layer as illustrated in Figure 2A. Using the Eigenvector approach, these two different directions can be seen in Figure 2B. When viewed from top-down, layers of red and blue-green vectors are also seen (Figure 2C). This shows that the poorly mineralized collagen fibrils and thus collagen bundles have an organizational motif similar to that of lamellar bone.

FIB-SEM analysis of the outer petrous bone layer from Sample 10619, namely the poorly preserved sample, does not show an easily identifiable 3D organization, neither in the InLens images nor the ESB images. An example of one such surface is shown in Figure 3A. Figure 3B shows the result of an Eigenvector analysis of the outer layer of sample 10619. The presence of only two colors (green and blue) shows that in this plane, there are two prominent collagen fibril bundle directions, and that these arrays of aligned collagen fibril bundles, are present in two distinct layers (Figure 3B). This layered organization can also be seen in Figure S1 and Movie S1.

3.2.2. Inner Petrous Bone Layers

The FIB-SEM images from the inner petrous bone layers (adjacent to the otic chamber) reveal a much more varied structure (Figure 4 and Figure 5) than that observed in the outer layers, with no lineations that show preferred orientations (Figure 2 and Figure 3). The inner layer bone shows bundles of collagen fibrils oriented in many different orientations in the same image, as evidenced by the presence of at least three different colors of the Eigenvectors (Figure 4, Figure 5 and Figure S2) and no clear shifting in fibril directions as in the case of the outer petrous bone layers. Thus, the analyses of the inner layers of both samples 10610 and 10619 show randomly oriented fibril bundles with no layered structure, which is similar to the structure of woven bone.

4. Discussion

Our detailed analyses of two archaeological human petrous bones that are about 2500 years old, show that the outer and less mineralized layer of the petrous bone has a lamellar structure with a plywood-like structural motif. The collagen fibrils in the inner layer immediately surrounding the otic chamber have a random organization. This structure is similar to woven bone. Based on this analysis we conclude that the two different structural bone types in the petrous bone of humans are the same as those identified by FIB-SEM in the modern pig, namely lamellar bone in the outer layer and woven bone in the inner layer [7]. We can therefore also surmise that the structural reasons that Ibrahim et al. [7,9] proposed for explaining the preferential preservation of ancient DNA based on the pig petrous bone, are also relevant to human bones.

As fossil petrous bones are so widely used in order to extract preserved DNA, understanding as much as possible about the changes that occur in these bones after burial (diagenesis), is important. Part of these changes presumably also involve loss of DNA, as it is well established that not all petrous bones preserve DNA, and those that do have DNA usually preserve different amounts of DNA [11,24].

Here, we characterized different aspects of diagenesis of petrous bones from several sites in Italy that range in age from 2500 to 3000 years BP. The diagenetic processes of petrous bones may well differ from most other bones, because petrous bones are unusually dense [2]. We estimate that in the pig petrous bone, the outer layer is about 75 weight percent mineral, and in the inner layer the density is even higher [7]. This is compared to the range of 60–70 weight percent mineral for most bones [25].

During life and after burial, the very small plate-shaped crystals of bone undergo a sintering process; namely, some crystals grow larger at the expense of other crystals. This can be monitored using the so-called SF (SF) as determined by FTIR [19,26]. The SFs of the nine Phoenician bones analyzed range from 2.7, which is the SF of modern bone, to 4.2 (Table 1). The SFs of modern pig petrous bone inner and outer layers are 2.7–3.1 and 2.8–3.0, respectively. From this we can conclude that despite the high density of petrous bones, the crystals in the modern petrous bone are the same or possibly slightly more crystallized compared to most other bones. In fossil bones, the crystals in the petrous bones can undergo sintering. When comparing the outer layers between poorly preserved and well-preserved samples, the poorly preserved outer layer is more porous compared to the well-preserved bone, this may well be due to diagenesis. We do not know if diagenesis can alter bone matrix organization in addition to collagen loss. We also noted that the denser inner petrous bone tends to undergo less sintering than the outer less dense bone (Figure 1). So, density and structure do affect mineral diagenesis.

Collagen is the major component of the petrous bone matrix. Here, we measured the percent gelatin in these fossil bones, which is the insoluble matrix fraction that remains after removal of the mineral, and then solubilized by mild heating followed by filtration. We measured a range of gelatin in the nine fossil bones from 1.3 to 13.2 weight percent (Table 1). The high mineral content of these bones does not protect the collagen from undergoing diagenesis. Note that the inner and outer layers of modern pig bone yield 9.1 and 10.3 weight percent gelatin. The low value compared to the best-preserved fossil bone is probably due to the fact that being better preserved, the collagen is more difficult to solubilize into gelatin. Clearly, some of these petrous bones have lost up to 90% of their collagen contents. We also note that the %C and C/N ratios of the gelatin do not change as a function of the amount of gelatin preserved (Table 1). This indicates that during diagenesis of these Phoenician petrous bones the entire matrix is lost, namely collagen and non-collagenous proteins. This is not as has been observed in other cases of bone diagenesis, where the non-collagenous proteins are better preserved than the collagen [27].

The analyses of the two bones studied in detail by FIB SEM, showed that the better-preserved bone maintained the same volume of less-mineralized material in the inner and outer layers, whereas the less-mineralized component of the less well-preserved bone increased significantly (Table 2). We do not know the reasons for this.

Fernandes et al. showed that separation of bone powder based on density has an influence on the amount of endogenous DNA extracted in terms of quantity and degree of contamination [28]. This indicates that the denser petrous bone has more preserved aDNA. The inner layers of modern pig and human petrous bones have up to three times per unit volume more osteocytes than other bones and hence more DNA [8,9]. Pinhasi et al. have shown that the inner layer does contribute more aDNA than the outer layer in fossil petrous bones [5]. So, the correlation between density and DNA preservation could be that the denser inner petrous bone initially contains more DNA, and not necessarily because the denser bone protects DNA better than the less dense bone. If this is indeed the explanation, then it would be worthwhile to specifically sample the inner layer of the fossil petrous bones for optimal aDNA extraction.

5. Conclusions

The 3D structure of the human petrous bone comprises an inner layer of woven bone, and an outer layer of lamellar bone, as was reported for modern pig petrous bone.

Even though the petrous bone preserves DNA relatively well, we show that in these 2500-year-old archaeological bones, the collagen has been degraded and that the mineral phase is generally better preserved in the inner petrous bone layer compared to the outer layer. The outer layer of the petrous bone is less dense than the inner layer and thus may be more subject to diagenetic alterations than the inner layer. Collagen contents varied widely between the nine analyzed archaeological human petrous bones, implying significant alterations to the organic material during diagenesis. In the inner layers the mineral crystals are better preserved and hence less crystalline compared to the outer layers, implying that the inner layers are ideal locations for aDNA sampling.