Measuring incremental line width and appearance in the tooth cementum of recent and archaeological human teeth to identify irregularities: First insights using a standardized protocol

International Journal of Paleopathology 27 (2019) 24–37

Contents lists available at ScienceDirect

International Journal of Paleopathology journal homepage: www.elsevier.com/locate/ijpp

Measuring incremental line width and appearance in the tooth cementum of recent and archaeological human teeth to identify irregularities: First insights using a standardized protocol

T

⁎

Gabriela Mani-Caplazia, , Gerhard Hotza,b, Ursula Wittwer-Backofenc, Werner Vacha,d a

Integrative Prehistory and Archaeological Science, University of Basel, Switzerland Natural History Museum of Basel, Anthropological Collection, Switzerland c Biological Anthropology, University of Freiburg, Germany d Department of Orthopaedics and Trauma Surgery, University Hospital Basel, Switzerland b

A R T I C LE I N FO

A B S T R A C T

Keywords: Tooth Cementum Annulation Dental cementum Stress marker Pregnancy Variability Reproducibility

Objective: Irregular incremental lines (ILs) in the tooth cementum were previously associated with pregnancy and certain diseases. This study aims to identify irregular ILs and assess their patterns and reproducibility. Materials: 24 recent and 32 archaeological teeth from the nineteenth century with known birth history. Methods: Histological sections of tooth roots were microscopically assessed. The width and appearance of 16,605 ILs were measured according to a standardized protocol. Results: Irregular appearing ILs were present in earlier deposited ILs, which correspond to younger years in life. Irregular appearances decreased as the IL number increased, whereas irregular width was spread evenly across all ILs. Within-section reproducibility was relatively high for irregular appearance (intra class correlation close to 0.70 in recent and archaeological teeth) and irregular width (intra class correlation: recent: 0.49; archaeological: 0.58), whereas the across-section reproducibility was moderate. Conclusions: Irregular width and appearance in ILs were identified successfully with within-section reproducibility. The moderate reproducibility across sections needs to be addressed in further studies by more systematic sampling of sections. Significance: The proposed protocol identifies irregularities in a reproducible manner and may suggest that irregular ILs could be used in paleopathology to identify pregnancies and diseases. Limitations: The correlation between the identified irregular ILs and known pregnancies has not been assessed as part of this study. Suggestions for further research: The identified irregular ILs need to be validated by correlating them with known life history data.

1. Introduction Tooth cementum is a mineralized tissue covering the tooth root and shows a continuous appositional growth throughout life, supporting the anchoring of the tooth in the alveolar bone (Schroeder, 2000). The alternating phases of cementum growth in a circannual rhythm result in the formation of incremental lines (ILs) observed in many species. These ILs are visible under transmission light microscopy as pairs of bright and dark layers (Kagerer and Grupe, 2001) (Fig. 1). Each pair is assumed to represent one year of life (Lieberman, 1994; Wedel, 2007). Tooth cementum annulation (TCA) is an established method for determining the age-at-death in humans in recent and archaeological

⁎

teeth (Grosskopf, 1990; Wittwer-Backofen et al., 2004). Counting ILs in the acellular extrinsic fiber cementum (AEFC), the type of cementum most appropriate for TCA, and adding it to the mean tooth eruption age per tooth type (Adler, 1967) provides the TCA age. The formation of the ILs is described to be influenced by endogenous cycles and environment factors related to seasons (Grue and Jensen, 1979; Lieberman, 1994). The brighter and usually wider layer is described to correspond to periods of growth, while the darker and usually thinner layer corresponds to periods of rest or slower growth (Lieberman, 1994). Several studies have assessed the reason for the layered structure and conclude that the alternating bright and dark layers are due to variation in the collagen orientation of Sharpey’s fibers (Lieberman,

Corresponding author at: Integrative Prehistory and Archaeological Science, University of Basel, Spalenring 145, 4055, Basel, Switzerland. E-mail address: [email protected] (G. Mani-Caplazi).

https://doi.org/10.1016/j.ijpp.2019.07.003 Received 3 August 2018; Received in revised form 14 July 2019; Accepted 14 July 2019 1879-9817/ © 2019 Elsevier Inc. All rights reserved.

International Journal of Paleopathology 27 (2019) 24–37

G. Mani-Caplazi, et al.

would provide insight into individual life events. This would provide an invaluable instrument in paleopathology, complementing the reconstruction of life and health conditions of ancient populations. This method could also provide insight into stress loads over the course of life and differences between individuals and populations over periods of time. It would also be relevant in forensic contexts to support the identification of individuals. A method capable of detecting and dating pregnancy-related signals may inform researchers about the number of pregnancies, age at first and last pregnancy and pregnancy intervals; hence, assisting with paleodemographic analyses. On the other hand, irregularities due to mechanical stress as higher strain from chewing (Lieberman, 1993; Grupe et al., 2013) or perhaps random fluctuations in the formation process cannot be etiologically excluded. However, before speculating about wide-ranging use, we first need a reliable method to identify potential signals, which may reflect stimuli. Given that former studies have seen broader and, in terms of contrast, more pronounced ILs in years of pregnancies and certain diseases, a systematic approach to measure the width of ILs and index their appearance to capture irregularities may in the long run provide a method for signal detection. To date, such a systematic assessment of IL characteristics has not been conducted in human teeth. Such an approach must address issues regarding data collection and interpretation. These issues are discussed below and guided the preparation and analysis of this study. Should absolute or relative values of IL width be used to identify irregularities? The quantitative evaluation of IL width (IL-WI) is complicated by variation in tooth cementum band thickness. This variation is due to multiple factors such as the tooth size, tooth type, or variation in mechanical exposure within a tooth (Schroeder, 1986; Bosshardt and Selvig, 1997; Grupe et al., 2013). Hence, this limits direct comparability across sections, teeth, and individuals. Similarly, methodological issues, such as non-perpendicular cutting direction to the ILs, which can occur due to the cone-shaped form of the tooth root, can lead to systematically deviating IL-WIs. To identify relevant deviation in width, it is necessary to transform absolute values into relative values taking into account the local background level (assessable by local average IL-WI) and the local background noise of the IL-WI measurements (assessable by local IL variability). In this way, we obtain measurements invariant under scale transformations, allowing us to perform comparisons between sections, teeth and individuals. Can quality assessment improve the analysis? For the IL-WI measurement, the accuracy of the measurement may depend on the quality of the ILs. Less distinguished ILs may provide less accurate IL-WI measurements, which can lead to misleading results. Former TCA studies focusing on age estimation found that the reliability of results was not affected by the cementum quality in the images (Wittwer-Backofen et al., 2004; Dias et al., 2010). It is, however, unknown whether this is also the case for assessing appearance and width. How to handle subjectivity in assessment of ILs? The assessment of the IL appearance (IL-AP) and quality (IL-QU) is subjective. Also, the measurement of the IL-WI is not fully independent of human judgement. These challenges require a protocol to standardize the indexing of the appearance and quality of ILs and the measurement of their width. Is there a relationship between irregular width and appearance of ILs? The appearance was the primary marker in previous human studies on cementum-based signal detection, with the observation that those ILs with an irregular appearance very often also differ in their width. Understanding whether these markers usually appear together or independently of each other can guide the choice of markers in future studies validating potential signals. Do irregularities appear mainly singular or in a cluster? Stimuli may cause signals present either in single ILs or over several consecutive ILs, depending on their length. The length of periods of irregularities may give us a first clue regarding which stimuli may play a role in generating this type of signal. Is there a need to analyse several sections? Signals presenting stimuli

Fig. 1. Microscopic view (×400 magnification) of the acellular extrinsic fiber cementum (AEFC) of an archaeological canine (Z_1516). The white lines encircled in black show two examples of ILs, the white arrows indicate irregular ILs which might refer to stress periods.

1993), or orientation of hydroxyapatite crystals (Cool et al., 2002). There is growing evidence from recent studies that the alternating light and dark bands correspond to differences in relative mineralization (Lieberman, 1993; Colard et al., 2016; Stock et al., 2017; Mani-Caplazi et al., 2017; Dean et al., 2018). One of the most fascinating properties of ILs is the high regularity of the layered structure, which seems to visually mirror the aging of an individual. However, aging is not a smooth process, and the deviations from regularity have attracted the attention of researchers. Researchers found irregular ILs (Fig. 1) varying in terms of IL width and contrast in humans, and could link them to pregnancies and certain diseases (Kagerer and Grupe, 2001; Künzie and Wittwer-Backofen, 2008; Umiger, 2012), and to reproduction in animals, and cold stress or variation in the seasonal diet (Lieberman, 1993; Cipriano, 2002; Medill et al., 2010). Kagerer and Grupe (2001) found qualitative traces (broader and more translucent ILs) in the tooth cementum, which corresponded exactly with the age women were pregnant or in cases where individuals had a skeletal trauma or diseases such as a renal disease. These are conditions affecting calcium metabolism; hence, calcium level may influence the IL formation. However, the authors could not find qualitative traces of diabetes, liver or thyroid disorders, osteoporosis or malnutrition in the cementum. In some of studies irregular ILs were called stress markers (SMs) (Umiger, 2012). Umiger (2012) described SMs as white lines followed by a particularly dark one, and developed a method to identify potential SMs based on their deviation from the average IL width. During pregnancy, there is a high calcium demand due to the growing foetal skeleton, and this causes major changes in calcium homeostasis (Naylor et al., 2000; Møller et al., 2013) and a high rise of several hormones with direct or indirect effects on calcium and bone metabolism (Kovacs and Kronenberg, 1997). Since tooth cementum is a bone-like tissue consisting mainly of collagen and hydroxyapatite crystals, similar factors and hormones may regulate their function (Saygin et al., 2000; Gonçalves et al., 2005). It can be assumed that those hormones and growth factors, active during pregnancy and affecting bone turnover, may also influence the tooth cementum growth and mineralization. Similarly, diseases or other events, in particular those affecting calcium metabolism, may also leave traces in the cementum. In contrast to bone, cementum is not vascularized and is rarely remodelled during life (Saygin et al., 2000). Hence, irregular ILs may serve as biological records of pregnancies, lactation periods or diseases. Consequently, it can be hypothesized that a method to identify signals 25

International Journal of Paleopathology 27 (2019) 24–37

G. Mani-Caplazi, et al.

are expected to be present across different sections. However, irregularities may not be visible uniformly and, therefore, it may make sense to analyse several sections in order to capture signals, which are not consistently represented by irregularities. Reproducibility analyses of irregularities can inform us about how consistently irregularities appear, and guide how many measurements per section and how many sections are needed to reliably identify signals. Improving identification of signals by pooling information from multiple measurements requires a proper alignment of ILs. This may be challenging, as ILs sometimes bifurcate, interrupt, or merge together, leading to differences in the number of identified ILs in areas of the same tooth. Are there differences between recent and archaeological teeth? Signal identification is relevant both in recent and archaeological context. As we expect, biological processes caused by stimuli lead to irregularities, we expect irregularity-based signal detection to be applicable in both recent and archaeological populations. However, there may be some discrepancies due to differences in life conditions or in post mortem taphonomic processes. The long-term goal of our research is to develop a method capable of identifying biological markers in the human tooth cementum indicative of stimuli. In this paper, we offer a first, preliminary step, towards this goal; namely introducing a measurement procedure minimizing allometric, variability, or methodical effects. This study assesses the feasibility of measuring and indexing ILs following a standardized protocol to identify irregular ILs. In addition, we consider aspects of suitability of IL-WI and IL-AP for signal detection, and compare the results from recent and archaeological teeth.

Table 1 Number of teeth in analysis.

Tooth by type Incisor Archaeological Recent Canine Archaeological Recent Premolar Archaeological Recent Total Archaeological Recent

Total screened

Excluded due to low quality cementum

Remaining teeth for measurement and indexing

(n)

(n)

%

(n)

%

8 28

0 11

0 39.3

8 17

100 60.7

25 5

4 2

16.0 40

21 3

84 60

3 11 80 36 44

0 7 24 4 20

0 63.6 30.0 11.1 45.5

3 4 56 32 24

100 36.4 70.0 88.9 54.5

archaeological teeth) were used in this analysis (Table 1). Of the total number of teeth, 24 were excluded due to insufficient quality of cementum noted as low contrast of ILs, too many post-mortem artefacts, unidentifiable starting points, or unstable or not intact outer margin. This selection process is described in detail in the methods section and in the flow chart (Supplementary Information (SI 2)). 56 teeth from 55 individuals (24 recent teeth and 32 archaeological teeth) remained in the sample and were measured and indexed (Supplementary Information (SI 1)). Of the archaeological group, two teeth from one individual have been measured and indexed. Exclusion of teeth was based on insufficient quality of the cementum, rendering identification of IL width and appearance impossible. This led to a high exclusion rate in the recent tooth group, and a low exclusion rate in the archaeological group. This difference can be explained by the fact that teeth in the recent group were often extracted due to caries and/or periodontal disease, whereas in the archaeological group we analysed teeth which were still present at death. In general, caries and/or periodontal disease may affect the integrity of the cementum (Dias et al., 2010). On the other hand, Wittwer-Backofen et al., 2004 found no significant influence of periodontal disease on the accuracy of TCA age estimation. However, TCA age estimation only requires identification of ILs, and irregularities are not assessed. The low prevalence of low-quality cementum in teeth of skeletons of the Basel Spitalfriedhof collection has been reported previously (Caplazi, 2004). The age at tooth extraction of the individuals in the recent group ranges from 12 to 78 years (mean 50.1), and the age-at-death of individuals in the archaeological group ranges from 25 to 75 years (mean 48.7). The mean age is comparable between the two groups; however, the age distribution is more dispersed in the recent group. The types of teeth analysed in the two groups are not balanced in term of tooth type. In the recent group there are predominantly incisors and in the archaeological group mainly canines.

2. Material Recent teeth were collected from a dental clinic in Germany, which treats patient residing in the region. The teeth are from women who consented to share their childbirth history and to include their extracted teeth in the SM study. In all cases, the tooth extractions were medically required and took place in the clinical setting during the years 2001 to 2004. The records of each patient contain the date of extraction, the reason for extraction, the patient’s date of birth, diseases and self-reported childbirth history, which includes dates of births and miscarriages. The teeth of these individuals were already sectioned as part of a previous study (Künzie and Wittwer-Backofen, 2008). The archaeological teeth belong to the extensively documented Basel Spitalfriedhof collection from the nineteenth century stored at the Natural History Museum Basel, Switzerland (Hotz and Steinke, 2012). These were patients who lived during the early industrialization era and who died in the Basel city hospital between 1845 and 1868. The skeletons were excavated in 1988 and 1989 by the Basel-Stadt Archaeological Research Center. One or more transcribed medical records (stored in the Staatsarchiv Basel-Stadt) are available for each skeleton, depending on the number of hospital stays. Hence the sex, date of birth, age-at-death, medical history and the cause of death are available for each patient. The date of birth and age-at-death were additionally verified (Hotz et al., 2015; Karakostis et al., 2017) and the childbirth history assessed and validated genealogically based on different archives. Further information, such as the geographical, social and family situations for three generations (parents, the selected individual and the next generation) were confirmed genealogically and historically (Hotz and Steinke, 2012; Pavel and Kumpf, 2017). In this study, we include women with a documented and genealogically verified childbirth history (births and no births), along with a few men. The study population is comprised predominantly of females from the lower class (Pavel and Kumpf, 2017). The teeth of these individuals had been previously sectioned and images of the sections and section areas are archived in the digital database. Our study is an observer-blind study, with the childbirth information disclosed in future studies. In total, 80 teeth from 69 individuals (44 recent and 36

3. Methods 3.1. Selection of images with identifiable incremental lines The selection of sections and images with best IL quality was conducted systematically (see flow chart (Supplementary Information (SI 2)). Up to five consecutive cross sections per tooth with a thickness of 60–80 μm were prepared from the apical part of the middle third of the tooth root towards the crown, where AEFC is present according to the technique described by Wittwer-Backofen (2012). These sections were examined by transmission light microscopy (Leica DM RXA) in ×25 to ×400 magnification. Several areas of the sections with good IL presentation were scanned using a digital camera (Leica DC 250) and stored in the image database. Finally, one image of an area of the best IL quality per section (typically three sections) was identified and 26

International Journal of Paleopathology 27 (2019) 24–37

G. Mani-Caplazi, et al.

Fig. 2. ILs were measured three times (IL measurement of column C1, C2, C3) starting at the eruption layer (EL) to the outer cementum border (CB) and were then numbered consecutively.

- The mean IL width, averaging the three IL-WI measurements over the three columns - An overall index of appearance (IAP) as the sum over the three IL-AP measurements over the three columns - An overall index of quality (IQU) as the sum over the three IL-QU measurements over the three columns - Local residuals based on mean IL-WI and corresponding peaks - The number of 3-SD, 2-SD and 1-SD peaks at the three columns

transmitted to a large-scale monitor for measurement and indexing. 3.2. Measurement of incremental line width and indexing of the incremental line appearance and quality IL-WIs (each consisting of a bright and dark line) were measured at three selected columns (column C1, C2, C3) starting from the eruption layer, usually a broader bright line at the inner cementum border, until the last IL at the outer border of the cementum using the IM1000 software from Leica. Corresponding ILs were numbered consecutively starting at 1 (Fig. 2). Details regarding the measurement and indexing of the ILs can be found in Supplementary Information (SI 3). The appearance and quality of each measured IL were assessed and indexed defined by three levels (0=low, 1=medium, 2=high) of appearance and quality (Supplementary Information (SI 3)). Appearance takes brightness and contrast into account, whereas Quality focuses on the ability to identify the IL and to assess its width. While an IL number refers to the same IL across the three columns of a section, alignment was not possible across sections.

As some phenomena may be related to the fact that we are close to the last IL at the outer cementum border, we consider referring to the original order of the ILs a reverse numbering with the last IL numbered as -1. When considering the relation to the IL number, we restrict the analyses to the first/last 45 ILs to ensure a sufficient sample size per IL number. 3.4. Statistical methods to assess the variation To access the reproducibility within a section and within a tooth across sections, we utilize intra class correlation (ICC) coefficients. The ICC is the expected correlation when, for each single IL, we randomly draw one pair out of all available pairs of measurements for each single IL (with random order within each pair). We also use Cohen's kappa coefficient to assess the agreement between two categorical variables.

3.3. Pre-processing: Definition of relative incremental line width values and summary measures IL-WI measurements were transformed into relative IL-WI values as follows: At the level of each single IL-WI measurement, we defined local residuals to catch local deviation of the width in comparison to neighbouring values. These are defined as z-scores, comparing the actual measured width with the local mean and the local SD, following the formula: z = (measured width – local mean)/(local SD) The local mean and the local SD are based on 12 measurements (6 to both sides), above and below the current values, not including the neighbouring ILs. This is visualized in Fig. 3 and in the schematic graphic in the Supplementary Information (SI 4). No explicit boundary corrections are used, i.e. just the available values are used. Based on the local residuals, we also define local peaks. A local peak is present if the z-score is above a threshold x, that is, if the measured width is at least x times the local SD above the local mean. Three choices of x are considered: 3, 2 and 1. The peaks are referred to as 3-SD peaks, 2-SD peaks and 1-SD peaks, respectively. Local residuals were not defined for one individual contributing only 5 ILs. At the level of a section, the following summary measures were defined for each IL:

4. Results 4.1. Description of material In our investigation we included 156 sections from 24 recent and 32 archaeological teeth. The number of ILs varied from tooth to tooth. The minimum, median and maximum occurrences of ILs in recent teeth were 4, 38, and 66, respectively, and in archaeological teeth 14, 32.5 and 68. 4.2. Overall distribution of IL Width (irregular width), Appearances (irregular appearance), Quality, local residuals and peaks Overall, we could perform measurements for 16,605 ILs. The IL width measurements show a nearly perfect log-normal distribution, with a median of 2.96, a 10% percentile of 1.81, and a 90% percentile of 4.27 (Supplementary Information (SI 5)). The distributions of the variables considered at the column level are depicted in Table 2, stratified by recent and archaeological teeth. High Appearances could be 27

International Journal of Paleopathology 27 (2019) 24–37

G. Mani-Caplazi, et al.

Fig. 3. Calculation of the local mean is based on 12 IL measurements, 6 ILs above and below the current IL with a gap of 1 IL on both sides. This is demonstrated here for IL 26 and IL 33. Table 2 Characteristics of the variables considered at the column level.

Width

Appearance

Quality

Local residuals

mean median p10 p90 low medium high low medium high < 1SD 1-2SD 2-3SD > 3SD

Table 3 Characteristics of the variables considered at the section level.

Recent N=7206

Archaeological N=9399

2.99 2.96 1.97 4.27 6069 (84.2) 814 (11.3) 323 (4.5) 2942 (40.8) 3517 (48.8) 747 (10.4) 5752 (80.3) 849 (11.9) 312 (4.4) 251 (3.5)

2.94 2.72 1.70 4.42 7390 (78.6) 1481 (15.8) 528 (5.6) 3728 (39.7) 3911 (41.6) 1760 (18.7) 7546 (80.3) 1012 (10.8) 429 (4.6) 412 (4.4)

Mean width

Appearance Index Quality Index Number of 2-SD peaks in the 3 columns

Local residuals

observed in about 5% of all ILs, and medium Appearances in 14%, and both were more frequent in archaeological teeth. About 40% of the ILs were of low quality and 15% of high quality, with distinctly more high quality ILs in archaeological teeth. 4% of the ILs show a 3-SD peak and an additional 4.5% a 2-SD peak, with slightly higher frequencies in archaeological teeth. Both frequencies are distinctly larger than then values of 0.1% and 2.2%, which we expect in the case of a normal distribution of the local residuals reflecting pure measurement error. This does not hold with respect to the additional number of 1-SD peaks with an expected frequency of 13% and an observed of 11%. After aggregating the data at the section level, 5535 ILs could be analysed. The distributions of the variables considered at the section level are reported in Table 3. We can again observe a similar distribution of the (mean) width in recent and archaeological teeth. The differences in Appearance and Quality seen at the column level between recent and archaeological teeth translate into corresponding differences for the indices. The differences in peak frequency between recent and archaeological teeth became more pronounced by aggregating over the columns.

mean median p10 p90 ≤2 3-4 ≥5 ≤2 3-4 ≥5 0 1 2 3 < 1SD 1-2SD 2-3SD > 3SD

Recent N= 2402

Archaeological N= 3133

2.99 2.85 1.97 4.16 2161 (90.0) 156 (6.5) 85 (3.5) 1383 (57.6) 822 (34.2) 197 (8.2) 1978 (82.8) 292 (12.2) 83 (3.5) 35 (1.5) 1914 (80.2) 290 (12.1) 107 (4.5) 77 (3.2)

2.94 2.83 1.87 4.19 2687 (85.8) 309 (9.9) 137 (4.4) 1654 (52.8) 962 (30.7) 517 (16.5) 2605 (83.2) 302 (9.6) 137 (4.4) 89 (2.8) 2506 (80.0) 323 (10.3) 142 (4.5) 162 (5.2)

width, we can see in both groups a sharp decline over the first 4 ILs (Fig. 4a). For recent teeth, this is followed by a stable phase around an IL-WI of 3 μm until IL 18, followed by a decline in thickness, whereas archaeological teeth show a stable phase at IL-WI 2.8 μm until IL 18, with an increase afterwards up to a thickness above 3 μm. The relation of the Appearance Index to the IL number shows that Appearances tend to become less frequent with increasing age, in particular close to the cementum border (Fig. 4b). However, in recent teeth, Appearances are rare in the first ILs, whereas in archaeological teeth Appearances are more frequent in the first ILs. Quality increases up to IL 10, decreases afterwards, and is best at around IL 7 to 25 (Fig. 4c). The first ILs immediately after the cementum-dentin junction are of particularly low quality, and when considering the reverse order, it becomes obvious that quality decreases close to the tooth cementum border toward the periodontal ligament, and that the last six ILs are clearly of lower quality (Fig. 4i). There is a tendency for the archaeological teeth to be of better quality over the whole range of IL numbers. Peaks are identified frequently within the first 5 ILs, which probably reflects the fact that due to the sharp decline of the width, it is hard to determine local peaks in a reliable manner, and that we cannot distinguish peaks from noise (Fig. 4d–f). Around IL 5 the minimum number of peaks is reached,

4.3. Relation of Width, Appearance, Quality, local residuals and peaks to the IL number Fig. 4 reports the association of several variables at the section level to the IL number (see also Supplementary Information (SI 6)). For 28

International Journal of Paleopathology 27 (2019) 24–37

G. Mani-Caplazi, et al.

Fig. 4. Average values of mean IL width, Appearance, and Quality Indices, as well as the number of 1-, 2-, and 3-SD peaks dependent on the (reverse) IL number. Standard errors are shown as one-sided bars in light colour.

4.4. Relation of local residuals to Appearance

and after IL 5, we observe a slightly increasing number of peaks, in particular when we consider 2-SD and 3-SD peaks. Close to the cementum border, we observe a rather sharp increase in peak-frequency, particularly in recent teeth (Fig. 4j–l).

In Fig. 5. we report the relation of local residuals to Appearance (see also Supplementary Information (SI 7)). We observe a relationship both at the column and section level: The more pronounced the Appearance, the larger the local residuals. However, the trend is not very 29

International Journal of Paleopathology 27 (2019) 24–37

G. Mani-Caplazi, et al.

Fig. 5. Distribution of local residuals dependent on Appearance, visualized as violin plots, with Spearman correlation shown. The analyses are based on all ILs except the first four ILs in order to avoid the influence of potentially unreliable peaks. The residuals are truncated at +/- 5.

4.5. Clustering of peaks, Appearance and Quality

pronounced. For example, when a level 2 Appearance is observed at an IL in a single column there is no 1-SD peak at this IL with 60% probability. Even if we observe a level 2 Appearance in all three columns, the probability of finding no 1-SD, based on the mean width at the section level, is above 40%. On the other hand, if we do not find any Appearances in all three columns, the probability to observe a 2-SD peak at the section level is less than 5%– a rate distinctly lower than the overall frequency of 9.5%. The relationship is of similar magnitude in recent and archaeological teeth.

First, we evaluated autocorrelation, i.e. the correlation between a measurement and the next, the second next, etc., along one column (this distance is called lag). These were analysed with respect to local residuals, Appearance size, and Quality (see Supplementary Information (SI 8)). For the local residuals, we observe a clear correlation at lag 1; that is, a tendency that neighbouring values are correlated. The correlation moves to negative values for larger lags, which is a simple consequence of the definition of local residuals. The correlation at lag 2 increases when compared to correlations at higher lags, so we have a weak tendency for a correlation of local residuals that are 2 30

International Journal of Paleopathology 27 (2019) 24–37

G. Mani-Caplazi, et al.

Fig. 6. Tendency of three types of signals to cluster within a column. Reported is the distribution of cluster length over all IL numbers (IL-nr) and stratified by four groups of IL numbers. Colour indicates the cluster length: Blue: across 1 IL, red: 2 ILs, green: 3 ILs, orange: 4 ILs etc. The horizontal line indicates the chance probability of clusters of length 1. It is determined by permuting the IL numbers randomly within each column 50 times.

observed and random chance level of clustering is roughly the same, suggesting that clustering occurs predominantly for high Appearance. At the section level (Supplementary Information (SI 10)) we observe clustering for 2-SD peaks far above chance levels, whereas for Appearance this is less pronounced.

ILs apart. There appears to be no difference between recent and archaeological teeth, or for dependence on the IL number. For Appearance, we observe a correlation at lag 1. The fact that the correlation is less pronounced probably reflects that Appearance is discrete for values 0, 1, and 2. Quality behaves differently, with correlations appearing at higher lags. So, Quality is changing more slowly within a column. Results at the section level, when considering local residuals based on the mean width and Appearance or Quality Index, are very similar (Supplementary Information (SI 9)). Secondly, we evaluate the tendency of signals to cluster. We evaluate how often we have 1, 2, 3 or more signals consecutively in a row, and call this a cluster of length 1, 2, 3, etc. We consider this for three types of signals at the column level: 2-SD peaks, high Appearance and medium or high Appearance. The results are shown in Fig. 6. We observe that about 70%–80% of the peak clusters are length 1, and peak clusters of length 3 or more are rare (this may be a consequence of the definition of the local residuals with respect to excluding only +/-1 neighbours). The number of clusters of length 2 or more is much larger than expected by random chance. There is no change with increasing IL number, but peaks tend to cluster to a higher degree in archaeological teeth. With respect to high Appearance, we observe a less pronounced clustering, but again far above that expected by random chance. When we consider high or medium Appearance, the difference between the

4.6. Reproducibility of visually identified local peaks and Appearances: three case studies Before reporting on reproducibility across columns and sections, it was important to assess IL-WI growth curves in order to visually identify irregularities. Three teeth were chosen for evaluation (see Supplementary Information (SI 11)). Here we noted some important properties. Within three columns of a single section, we observed similarities in the pattern of Appearance and local peaks, but a lower degree of similarity was observed across the three sections. 4.7. Between-column reproducibility of Appearance, Quality, local residuals and peaks Fig. 7 shows the reproducibility of Appearance, local residuals, 2-SD peaks and Quality at the column level. Between-column reproducibility of Appearance is good, with a kappa value above 0.55 and ICC close to 31

International Journal of Paleopathology 27 (2019) 24–37

G. Mani-Caplazi, et al.

Fig. 7. Intraclass correlations (ICC) and kappa values dependent on the IL number at the column level. For Appearance and Quality, both ICCs and kappa values are shown. The IL numbers were grouped in consecutive intervals of length 3 before the computations were performed. The overall ICCs and kappa values are shown at the bottom of each figure.

decreases with IL number in recent teeth. Evaluating reproducibility in dependence on reversed IL numbers does not provide further insights (Supplementary Information (SI 12)).

0.7 in both recent and archaeological teeth. Associations between reproducibility of Appearance and increasing IL numbers were not observed. Reproducibility of local residuals is less with an ICC of 0.49 for recent teeth and 0.58 for archaeological teeth. This leads to kappa values, with respect to 2-SD peaks, of only 0.28 for recent teeth and 0.43 for archaeological teeth. In both samples, we observe decreasing reproducibility with increasing IL number. Quality needs not be reproducible across different columns of a section, as it may reflect a very local property. However, Quality seems to not vary much across columns, as indicated by an ICC above 0.55 and a kappa value above 0.4. The uniformity of quality across columns is less pronounced and

4.8. Between-section reproducibility of summary indices Due to the lack of alignment between different sections of a tooth, we focus here on properties not requiring an alignment. Consequently, we consider two measures of Appearance and the local residuals: The average over the life of the individual (the life load), and the slope of a regression compared to the IL number (the life trend). In addition, 32

International Journal of Paleopathology 27 (2019) 24–37

G. Mani-Caplazi, et al.

based on the mean IL width; 3) An IL with an Appearance Index of 5 or more, i.e., at least 2 columns with high Appearance and at least moderate Appearance in the third column. We allow for these signals to appear in other sections of the same tooth at a slightly weaker strength: A 2-SD peak Index of at least 2, a 2-SD peak, or an Appearance Index of 4 or more. Fig. 8 shows the frequency of finding a weaker signal exactly +/- d ILs above or below (left side) or within +/- d ILs above or below (right side). We can observe that for each of the three signals, the corresponding signal at the same IL number appears in around 20% of the sections, but distinctly above chance level. Local peaks are observed above chance level, at +/- 1 IL, and Appearance up to +/- 3 ILs, suggesting some reproducibility, but also illustrating problems with alignment. To quantify reproducibility, we evaluate the maximum difference between the frequencies in order to observe a signal with a range of +/- d ILs, and the corresponding chance level of d over all values. We observe a value of about 11% for local peaks and about 22% for Appearance, i.e. only 11% of the local peaks and 22% of Appearance is reproduced at different sections of the same tooth, if we take into account that a signal may also appear by chance.

Table 4 Reproducibility of load and trend, based on local residuals and Appearance, at the section and column level.*. Local residual quantity

period

section level

column level

load load load trend trend trend

all youth adult all youth adult

p10 −0.02 −0.19 −0.14 −0.58 −9.52 −0.45

p90 0.46 1.16 0.36 0.20 1.47 1.27

ICC 0.186 0.218 0.424 0.349 0.264 0.089

p10 −0.03 −0.18 −0.12 −0.47 −7.62 −0.47

p90 0.35 0.92 0.28 0.17 1.74 0.92

ICC 0.529 0.536 0.557 0.486 0.635 0.653

Appearance load load load trend trend trend

all youth adult all youth adult

0.37 0.00 0.26 −0.84 −4.19 −0.98

1.32 1.78 1.18 0.12 2.40 0.44

0.460 0.547 0.617 0.766 0.189 0.314

0.10 0.00 0.08 −0.29 −1.43 −0.32

0.44 0.64 0.41 0.06 0.85 0.15

0.783 0.785 0.777 0.863 0.739 0.598

* The trend is expressed as a change over 10 IL numbers. Within a section/ column, the load is only computed if at least 3 measurements are available, and the trend is computed if at least 5 measurements are available.

4.10. The impact of Quality As mentioned in the introduction, the quality of the ILs varies in respect to detection of Appearance and measuring the width. We have assessed the quality in order to determine its impact on these two processes. Hence, we now consider the distribution of Appearance and local residuals at the column level in relation to the quality. In Fig. 9, it is shown that the observer of this study (G.M.) was reluctant to detect Appearance in cases of low quality, and noted that a distinct difference between medium and high quality can be observed. The relationship between quality and detecting peaks is less pronounced, but we observe a tendency towards more peaks in cases of low quality and, within archaeological teeth, fewer peaks with high quality. Low quality may also lead to an increased variability in measuring IL width, but only to a moderate degree. As these results indicate, Quality may have an impact on the reliability of detected Appearance and peaks. Next, we evaluate the impact of Quality on the reproducibility and degree of association between Appearance and local residuals. In Table 5, we observe no systematic effect of Quality on reproducibility, but a higher association between Appearance and local residuals with increasing quality.

calculations were made for each individual for a period defined as “youth” (noted by IL numbers up to 18 minus the eruption age), and “adult”, defined by the remaining IL numbers, since understanding overall stress load or decreasing or increasing stress loads during life will be of interest to researchers. To allow a comparison between the column and section levels, we define the measures at the column level. For Appearance, we examined the original three level index at the column level and the summary index IAP at the section level. Since the number of sections is limited, we do not consider these analyses separately in recent and archaeological teeth. The results for the local residuals and Appearance are summarized in Table 4. With respect to the local residuals, we observe that the overall load shows variation at the section level comparable to the column level, and it is reproducible at the section level only to a moderate degree when compared to the reproducibility at the column level. The same appears for the trends, with the exception that the trend over the whole adult period is not reproducible across sections. Appearance, for both load and trend, show similar variation at the section and column level, especially if we take into account that the possible range at the section level is three times the possible range at the column level. The overall load is reproducible at the section level, but still less reproducible than at the column level. Also, trends over the whole life show a high reproducibility at the section level, whereas this is less the case when we consider the youth and adult period separately. In summary, the degree of reproducibility at the section level is worse than at the column level, even though the information at the section level has been summarized over three columns. Again, we observe better reproducibility for Appearance than for local residuals.

5. Discussion In this study, the feasibility of measuring the width of ILs and indexing their appearance in order to identify growth irregularities was assessed. Based on observed IL growth trends and irregular patterns, the potential and challenges of interpreting irregular ILs, will be discussed. In addition, we provide suggestions for using irregularities in future studies, aiming to investigate their potential as signals for stimuli. 5.1. Incremental line growth trends and irregularities over the course of life Broader and less defined ILs showing a sharp decline in width were present at the inner cementum border. This may indicate differences in the growth rate in the early phase of cementum apposition, which begins around the time of tooth eruption (Grue and Jensen, 1979; Bosshardt and Selvig, 1997; Cho and Garant, 2000), compared to the later more stable phase, which lasts until between 20 and 30 years old. The recent and archaeological teeth showed different tendencies, with recent teeth displaying decreased width (a similar age effect was also observed in animals (Carrel, 1994; Medill et al., 2009)), and archaeological teeth displaying increased width. There was a clear association between decreased quality of ILs and increasing age. ILs had a less clear structure and were more difficult to evaluate in advanced ages. Similar observations were made in TCA studies, and were explained by age-

4.9. Between section reproducibility of signals Due to a lack of alignment between the sections of a tooth, the evaluation of the between-section reproducibility of single peaks and Appearance is not straight forward. We must recognize that a signal in one section may appear in another section at a slightly different IL number. Hence, our basic approach is to consider the frequency at which a similar signal occurs in another section of a tooth at the same IL number or at an IL number d positions below or above. Our starting point is given by three types of distinct signals: 1) An IL with a 3-SD peak Index of at least 2, i.e., a 3-SD peak in at least two of the three columns; 2) An IL with a 3-SD peak, i.e., a local residual above 3 SD 33

International Journal of Paleopathology 27 (2019) 24–37

G. Mani-Caplazi, et al.

Fig. 8. Frequency of finding signals exactly +/- d ILs above or below (left side) or within +/- d ILs above or below (right side). The red line indicates expected frequencies at chance level, which are determined by randomly permuting the sections used for comparison 50 times.

34

International Journal of Paleopathology 27 (2019) 24–37

G. Mani-Caplazi, et al.

with irregular widths, based on relative width values (peaks), were spread over the course of life, with only a slight decrease associated with age, except at the cementum start and towards the outer border of the cementum where more peaks were present. This may indicate that ILs tend to be both broader and more variable at the cementum start and end. It is unlikely that this finding is related to stimuli. 5.2. Relationship of Width and Appearance and clustering of signals Irregular ILs in humans have been described as being broad and differently mineralized (Kagerer and Grupe, 2001; Künzie and WittwerBackofen, 2008). Analysing the relationship between these two characteristics revealed that ILs with an irregular appearance tended to be broader. These markers, however, can often be present independently. We observed that irregular ILs typically occur singularly, but in about 10–20% of all instances, they occur over two consecutive ILs, but rarely over more. Considering that an IL represents one year (Lieberman, 1994; Wedel, 2007), it may be concluded that stimuli causing irregularities in IL formation may last more than one year. However, without a deeper understanding of the processes responsible for IL formation, it is not possible to link clusters to a specific length of a stimulus. 5.3. Reproducibility of signals within and across sections Irregular appearances were reproducible within a section, whereas irregularity in the width were only moderately reproducible. The reproducibility of irregular width across sections was rather low. Irregular appearance, however, were moderately reproducible. The alignment issue may contribute to the reduced reproducibility, as identified potential signals may not be present in the same IL numbers across sections. 5.4. Quality labelling

Fig. 9. The distribution of Appearance and local residuals at the column level in relation to Quality (QU).

Two thirds of ILs were labelled “good quality”, but decreased with age. However, during the time period when pregnancies are most likely to occur, ILs were typically of good quality. Quality had a strong impact on detecting Appearance and some impact on measuring width. Since ILs that are not easily identifiable are more difficult to evaluate, subsequent assessments are more prone to errors. Including only good quality ILs in our datasets provided slightly more pronounced results, but the findings were similar to those including all IL measurements. It is therefore questionable whether systematic quality assessment is worth the effort.

Table 5 Intra class correlation (ICC) of Appearance and local residuals at the column level in dependence of the quality.

Appearance

Local residuals

Appearance and local residuals

Quality

Recent ICC

Archaeological ICC

0 1 2 0 1 2

0.72 0.68 0.65 0.47 0.46 0.41 Correlation 0.09 0.22 0.24

0.70 0.68 0.74 0.54 0.60 0.65 Correlation 0.20 0.26 0.28

0 1 2

5.5. Differences between recent and archaeological teeth Overall, the archaeological teeth showed more irregular ILs, particularly during adolescence. The archaeological sample in our study included teeth predominantly from women from the lower class (Pavel, 2016) who may have been particularly burdened by demanding life conditions. In the nineteenth century, the quality of life in European cities was deteriorating considerably due to rapid urbanization and increasing industrialisation (Spielvogel, 2012; Pavel, 2016). Post-mortem alteration of the cementum in archaeological teeth must be taken into consideration. However, we did not notice diagenetic changes in the archaeological sample. Authors have reported taphonomic influences leading to lower TCA performance (Roksandic et al., 2009) and band-like structures visible at multiple locations (Stutz, 2002). We found IL irregularities during the early phase of cementum apposition near the inner cementum border and at younger years. Irregularities were most commonly present across one or two ILs. These observations are consistent with the assumption that biological processes, rather than taphonomic ones, created the irregularities. The presence of better quality ILs in archaeological teeth may, however,

related influences on cementum formation (Klevezal et al., 2006; Dias et al., 2010; Willershausen et al., 2012). The last ILs, at the outer cementum border, tended to be broader and of lower quality. This feature was observed in all age classes and may be due to ILs that have not finished their development in contrast to earlier formed ILs. Newly deposited cementum consists primarily of uncalcified collagen, which gets mineralized later (Wittwer-Backofen et al., 2004). Evaluating irregularities, we found that ILs with an irregular appearance are present in younger years and decline with increasing age. Hence, the question arises: if there are more stimuli affecting the IL growth and mineralisation in younger years, perhaps this is connected to reproduction? In particular, the archaeological teeth showed this tendency. Perhaps this group was exposed to more stimuli leading to fluctuation in the IL formation, reflecting unstable life conditions. ILs 35

International Journal of Paleopathology 27 (2019) 24–37

G. Mani-Caplazi, et al.

stabilizing signals across sections. Our results suggest that irregularities in IL formation can serve as signals reflecting biological response to stimuli. For paleopathology, this might allow researchers to address issues of heterogeneous frailty, gain insight into individuals’ life courses in the past, and explore paleodemographic differences between populations.

point to some diagenetic influences leading to clearer presentation of ILs, but may also be related to the differences in selection of the teeth. Overall, the results indicate that there are no substantial differences between recent and archaeological teeth that impact signal detection. 5.6. Considerations for signal identification

Acknowledgements

Both Appearance and Width are promising markers for signal identification. Irregularities indicated by both markers turned out to be reproducible. In particular, Appearance showed high reproducibility within a section and demonstrated a distribution over the course of life suggesting that biological processes were present. The two markers do not necessarily occur together and, hence, an irregular appearance should be considered even if the IL is of normal width. Both markers, in particular Width, showed a lower reproducibility across sections, challenging their usefulness in signal detection. Since the correct alignment of ILs across sections is challenging, it remains unclear to what degree the measurement of several sections can contribute to signal recognition. An alternative approach may be to explore how external factors, like mechanical exposure, affects the development of ILs, and then to select a priori the most informative sections. Our results showed a high variability at the start and end of the cementum. A better understanding of the cause of these trends may facilitate the identification of irregularities. Overall, the results suggest that Appearance may play a relevant role in signal identification. The same holds for Width, but this variable may represent a more challenging marker. Further investigation is needed to clarify the true potential of both markers in signal detection. The quality of ILs had no major impact on the results of this study and, thus, integrating a quality assessment into future studies does not appear to be necessary.

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. We thank the team from the department of Biological Anthropology, University of Freiburg i. Br. for the collection of the recent teeth, tooth section preparation, providing the images of the archaeological teeth sections, and the use their equipment. We also thank Professor Dr. Jörg Schibler from the Department of Integrative Prehistory and Archaeological Science of the University of Basel for the research support. Additionally, we would like to thank the Citizen Science Project BBS “Bürgerforschungsprojekt Basel-Spitalfriedhof” (https://www. ipna.unibas.ch/bbs), for their volunteer work on basic research concerning historical sources, and especially the team of genealogists who have collected and validated the background information of the patients: Marie-Louise Gamma, Diana Gysin, Odette Haas, Ludwig Huber and Marina Zulauf (coordinating genealogist). We also thank Verena Fiebig-Ebneter, in charge of the patient-database “Bürgerspital 18401870″; the Natural History Museum of Basel: Basil Thüring, Dr. Loic Costeur and Tandra Fairbanks-Freund for the English revision; the Staatsarchiv Basel-Stadt: Esther Baur, Hermann Wichers and team and the Archäologische Bodenforschung Basel-Stadt: Guido Lassau, Norbert Spichtig and team.

5.7. Limitations of the study

Appendix A. Supplementary data

We are aware of the following limitations of our study: 1) The recent and archaeological groups include tooth types recommended to use for TCA (Wittwer-Backofen, 2012), however they differ in the percentage of teeth per tooth type (Table 1), which may explain some differences between the two samples. Unfortunately, our sample size was too small to directly investigate the influence of tooth type. 2) Several images were already available from a previous study of the archaeological teeth. Hence, there may be some unknown and undetected differences between the two groups of images. 3) The ILs have only been assessed once by one observer based on the standardised procedure; consequently, no inter- or intra observer variability could be assessed. 4) Measurements between different columns and the assessment of Appearance and Width could not be conducted blind, which may have contributed to high reproducibility within a section.

Supplementary material related to this article can be found, in the online version, at doi:https://doi.org/10.1016/j.ijpp.2019.07.003. References Adler, P., 1967. Zahn-, Mund- und Kieferheilkunde im Kindesalter. In: Harndt, E., Weyers, H. (Eds.), Die Chronologie der Gebissentwicklung. Die Quintessenz. Berlin, pp. 38–74. Bosshardt, D., Selvig, K., 1997. Dental cementum: the dynamic tissue covering of the root. Periodontol 2000 (13), 41–75. Caplazi, G., 2004. Eine Untersuchung über die Auswirkungen von Tuberkulose auf Anlagerungsfrequenz und Beschaffenheit der Zementringe des menschlichen Zahnes. Bull. Soc. Suisse d’Anthrop. 10, 35–83. Carrel, W.K., 1994. Reproductive history of female black bears from dental cementum. Int. Conf. Bear Res. Manage 9, 205–212. Cho, M.I., Garant, P.R., 2000. Development and general structure of the periodontium. Periodontol 2000 (24), 9–27. Cipriano, A., 2002. Cold stress in captive great apes recorded in incremental lines of dental cementum. Folia Primatol. 73, 21–31. Colard, T., Falgayrac, G., Bertrand, B., Naji, S., Devos, O., Balsack, C., Delannoy, Y., Penel, G., 2016. New insights on the composition and the structure of the acellular extrinsic fiber cementum by raman analysis. PLoS One 11, e0167316. Cool, S.M., Forwood, M.R., Campbell, P., Bennett, M.B., 2002. Comparisons between bone and cementum compositions and the possible basis for their layered appearances. Bone 30, 386–392. Dean, C., Le Cabec, A., Spiers, K., Zhang, Y., Garrevoet, J., 2018. Incremental distribution of strontium and zinc in great ape and fossil hominin cementum using synchrotron Xray fluorescence mapping. J. R. Soc. Interface 15, 20170626. Dias, P.E., Beaini, T.L., Melani, R.F., 2010. Age estimation from dental cementum incremental lines and periodontal disease. J. Forensic Odontostomatol. 28, 13–21. Gonçalves, P.F., Sallum, E.A., Casati, M.Z., Toledo, S.D., Junior, F.H., 2005. Dental cementum reviewed: development, structure, composition, regeneration and potential functions. Braz. J. Oral Sci. 4, 651–658. Grosskopf, B., 1990. Individual age determination using growth rings in the cementum of buried human teeth. Z. Rechtsmed. 103, 351–359. Grue, H., Jensen, B., 1979. Review of the formation of incremental lines in tooth cementum of terrestrial mammals. Dan. Rev. Game Biol. 11, 1–48. Grupe, G., Lippitsch, A., McGlynn, G., 2013. Functional Morphology of the Human Dentition and Its Probable Influence on Tooth Cementum Thickness and Incremental Line Count Annual Meeting - American Association of Physical Anthropology, Columbus, Ohio.

6. Conclusion In this study, a standardized protocol to measure the width and the appearance of ILs in human tooth cementum from two samples of recent and archaeological teeth was evaluated in order to identify irregularities. The procedure turned out to be feasible, and irregularities in width and appearance were identified with good reproducibility within sections, but lower reproducibility across sections. Clear patterns over the course of life of individuals were observed, with more pronounced Appearance occurring in younger years, especially in the archaeological sample. This may suggest the presence of stimuli affecting IL formation, reflecting stressful life conditions during these years. IL Appearance was successfully reproducible, whereas Width appeared more variable. These two irregularities often occurred singularly, but sometimes across two ILs. They tended to appear together, but could be observed alone and thus should be considered independently. We recommend using our protocol in future studies requiring identification of irregularities across ILs, with attention paid to reproducing irregularities across sections and 36

International Journal of Paleopathology 27 (2019) 24–37

G. Mani-Caplazi, et al.

Naylor, K.E., Iqbal, P., Fledelius, C., Fraser, R.B., Eastell, R., 2000. The effect of pregnancy on bone density and bone turnover. J. Bone Miner. Res. 15, 129–137. Pavel, M., 2016. Die Krankengeschichten des Bürgerspitals Basel ein Spiegel der Zeit? Eine quantitative-sozialhistorische Untersuchung von 1843 bis 1868 (unpublished Masterthesis, University Basel). Pavel, M., Kumpf, M., 2017. Krank und Arm? Die Krankenakten des Bürgerspitals als Spiegel der Zeit. Sozio-ökonomische und epidemiologische Analysen aufgrund von 19`000 Patientendaten (1843 bis 1868) des Bürgerspitals Basel. B.Z.G.A. 117. pp. 144–157. Roksandic, M., Vlak, D., Schillaci, M.A., Voicu, D., 2009. Technical note: applicability of tooth cementum annulation to an archaeological population. Am. J. Phys. Anthropol. 140, 583–588. Saygin, N.E., Giannobile, W.V., Somerman, M.J., 2000. Molecular and cell biology of cementum. Periodontol 2000 (24), 73–98. Schroeder, H.E., 1986. Types of cementum and their distribution. In: Schroeder, H.E. (Ed.), The Periodontium. Handbook of Microscopic Anatomy V/5. Springer Verlag, Berlin, Heidelberg, New York, Tokyo, pp. 26–32. Schroeder, H.E., 2000. Zementogenesis und Wurzelzement. In: Schroeder, H.E. (Ed.), Orale Strukturbiologie. Georg Thieme Verlag. Stuttgart, pp. 144–166. Spielvogel, J.J., 2012. Western civilization, since 1300. In: Spielvogel, J.J. (Ed.), The Industrial Revolution and Its Impact on the European Society. Wadsworth, Boston, pp. 607–635. Stock, S.R., Finney, L.A., Telser, A., Maxey, E., Vogt, S., Okasinski, J.S., 2017. Cementum structure in Beluga whale teeth. Acta Biomater. 48, 289–299. Stutz, A.J., 2002. Polarizing microscopy identification of chemical diagenesis in archaeological cementum. J. Archaeol. Sci. 29, 1327–1347. Umiger, E., 2012. Qualitative und quantitative Stressmarker-Detektion im Zahnzement altersbekannter Tuberkulosepatienten des 19. Jahrhunderts. Bachelorarbeit Universität Freiburg i.Br. Wedel, V., 2007. Determination of season at death using dental cementum increment analysis. J. Forensic Sci. 52, 1334–1337. Willershausen, I., Försch, M., Willershausen, B., 2012. Possibilities of dental age assessment in permanent teeth: a review. Dentistry. S1 1. Wittwer-Backofen, U., Gampe, J., Vaupel, J., 2004. Tooth cementum annulation for age estimation: results from a large known-age validation study. Am. J. Phys. Anthropol. 123, 119–129. Wittwer-Backofen, U., 2012. Age estimation using tooth cementum annulation. Methods Mol. Biol. 915, 129–143.

Hotz, G., Steinke, H., 2012. Knochen, Skelette, Krankengeschichten. Spitalfriedhof und Spitalarchiv - zwei sich ergänzende Quellen. B.Z.G.A. 112, 105–138. Hotz, G., Zulauf-Semmler, M., Fiebig-Ebneter, V., Schumacher, B., Meyer, L., Gysin, D., Gamma, M.-L., Haas, O., Blatter, S., Blatter, R., Gianola, A., 2015. Neue Quellen zur Skelettserie Basel-Spitalfriedhof – Grundlagen interdisziplinärer Zusammenarbeit: Das Bürgerforschungsprojekt Basel-Spitalfriedhof. Bull. Soc. Suisse d’Anthrop. 21, 9–13. Kagerer, P., Grupe, G., 2001. Age-at-death diagnosis and determination of life-history parameters by incremental lines in human dental cementum as an identification aid. Forensic Sci. Int. 118, 75–82. Karakostis, F.A., Hotz, G., Wahl, J., Scherf, H., Harvati, K., 2017. Occupational manual activity is reflected on the patterns among hand entheses. Am. J. Phys. Anthropol. 164, 30–40. Klevezal, G.A., Kirillova, I.V., Shishlina, N.I., Sokolov, A.A., (Selkova), Trunova, Yu, E., 2006. Growth layers in tooth dentin and cementum: problems and perspectives of their use in the study of fossil and subfossil mammal remains including humans. Documenta Archaeolbiologiae. 4, 113–123. Kovacs, C.S., Kronenberg, H.M., 1997. Maternal-fetal calcium and bone metabolism during pregnancy, puerperium, and lactation. Endocr. Rev. 18, 832–872. Künzie, M., Wittwer-Backofen, U., 2008. Stress markers in tooth cementum caused by pregnancy. Annual Meeting – American Association of Physical Anthropologists. Lieberman, D.E., 1993. Life history variables preserved in dental cementum microstructure. Science 261, 1162–1164. Lieberman, D.E., 1994. The biological basis for seasonal increments in dental cementum and their application to archaeological research. J. Archaeol. Sci. 21, 525–539. Mani-Caplazi, G., Schulz, G., Deyhle, H., Hotz, G., Vach, W., Wittwer-Backofen, U., Müller, B., 2017. Imaging of the human tooth cementum ultrastructure of archeological teeth, using hard x-ray microtomography to determine age-at-death and stress periods. Proc. SPIE 10391, Developments in X-Ray Tomography XI, 103911C. Medill, S., Derocher, A.E., Stirling, I., Lunn, N., Moses, R.A., 2009. Estimating cementum width in polar bears: identifying sources of variation and error. J. Mammal. 90, 1256–1264. Medill, S., Derocher, A.E., Stirling, I., Lunn, N., 2010. Reconstructing the reproductive history of female polar bears using cementum patterns of premolar teeth. Polar Biol. 33, 115–124. Møller, U.K., Streym, S., Mosekilde, L., Heickendorff, L., Flyvbjerg, A., Frystyk, J., Jensen, L.T., Rejnmark, L., 2013. Changes in calcitropic hormones, bone markers and insulinlike growth factor I (IGF-I) during pregnancy and postpartum: a controlled cohort study. Osteoporos. Int. 24, 1307–1320.

37