Biological relationships and frontal sinus similarity in skeletal remains with known genealogical data

Abstract

Frontal sinus analysis has potential utility for detecting biologically related individuals. However, the methodological approach to its evaluation, as well as its informative value, have been questioned. The aim of this work is to introduce a new approach to evaluating the frontal sinus using the ‘external supraorbital line’ (ESOL) and to determine whether there are sex differences within families in frontal sinus measurements and whether frontal sinus similarity reflects known genetic relationships in both measurements and morphology. We examined the skeletal remains of 41 adult individuals (25 males, 16 females), all members of one family over four generations (19th to 20th centuries), including individuals with very close consanguinity. CT images of skulls were acquired, and both the dimensions and morphology of the frontal sinuses were analyzed using their portions above the ESOL. No significant sex differences were found within families based on frontal sinus dimensions. Significant relationships were found between biological distance and the maximum height and morphology of the frontal sinuses. The greatest degree of similarity was found among closely related individuals. Additionally, in several cases, there was a greater degree of similarity between first cousins or grandparents and their grandchildren than among siblings or parents and their children. Total surface, volume and width are not significant indicators of relatedness. Known genetic relationships are also supported by individual morphological features. Variability within families with very close consanguineous relationships was lower than within families with common degrees of consanguinity, although differences are significant only for some variables.

We introduce a new approach to evaluating the frontal sinus using the “external supraorbital line” (ESOL), to determine whether frontal sinus similarity reflects known genetic relationships in both measurements and morphology. Significant relationships were found between biological distance and the maximum height and morphology of the frontal sinuses.

1. INTRODUCTION

The formation and variability of craniofacial pneumatization is a relevant subject in research, particularly in terms of its evolutionary aspects, function, development issues, medical relevance and use in forensic practice (Butaric and Maddux, 2016; Choi et al., 2018; Godinho and O’Higgins, 2018; Nikolova et al., 2018). In research on pneumatization, increasing attention is being paid to the frontal sinuses. The variability of their patterns (i.e. configuration, shape, anatomical variations, dimensions) is wide (Szilvássy, 1982; Guerram et al., 2014; Lund et al., 2014; Choby et al., 2018; Štoković et al., 2018). Because these sinuses are encased in the solid frontal bone, they are usually preserved in archaeological contexts and we can evaluate them even if the overall preservation of skeletal remains is poor (Cameriere et al., 2008; Goyal et al., 2013). For these reasons, they are of interest to bioarchaeology, forensic anthropology and medicine. There are three main areas of potential application of the frontal sinus in forensic anthropology and bioarchaeology: sex estimation, population affinity estimation and individual identification (Christensen and Hatch, 2018).

Currently, dental and osteological non‐metric traits are preferred for the detection of family relationships in skeletal remains, apart from DNA analysis (Stojanowski and Schillaci, 2006). However, the possible use of frontal sinuses for the detection of biologically related individuals has also been discussed (Leicher, 1928; Asherson, 1963). Accordingly, the genetic component of frontal sinus formation is also a subject of interest. This issue has most often been studied in living families (father, mother, children) and dizygotic and monozygotic twins. Frontal sinus similarities have been found in morphology and dimensions between biologically related individuals (Szilvássy, 1986; Kjaer et al., 2012). It is believed that the degree of paranasal sinus pneumatization is possibly more influenced by genetic factors more than by the development of anatomical variations of some other paranasal structures (Chaiyasate et al., 2007). These types of studies, however, face ethical restrictions in terms of research on living people (Kjaer et al., 2012), especially in terms of doses of X‐rays without medical indication. The findings were then verified on osteological samples, for which these similarities were also found (Szilvássy et al., 1987; Alt et al., 1997; Slavec, 2004). It is possible to identify biological affinity even among skeletons in poor condition, although we cannot use frontal sinuses to identify the exact relationship of the individuals or their exact position in the family tree (Cameriere et al., 2008).

Until the beginning of the 21st century, osteological studies focused mainly on comparing individuals using simple visual description of sinus morphology based on anteroposterior radiographs (Szilvássy et al., 1987; Slavec, 2004). However, this approach could be considered subjective (Asherson, 1963) due to its lack of clear interobserver replicability. Furthermore, dimensions in combination with morphological characteristics have been rarely used (Slavec, 2005). Statistical evaluations as well as medical imaging methods, such as CT and MRI, began only relatively recently to be used in frontal sinus analysis, especially on living people. Each of the recent studies focused on different variables: size based on development categories (Chaiyasate et al., 2007), the morphology scoring system used (Cameriere et al., 2008) and exact dimensions expressed in units (Kjaer et al., 2012). Among osteological samples, the possibilities for study are extremely limited by the availability of genealogically documented material (Carson, 2006; Ricaut et al., 2010; Gavrus‐Ion et al., 2017), its state of preservation and, consequently, the evaluation methodologies available (CT, MRI, statistics).

The question of the influence of frontal sinus pattern inheritance, its informative value, and methodology is still under discussion (Latiff et al., 2009; Ito et al., 2015). All of the above‐mentioned studies dealing with frontal sinus and relatedness assume that genetics affects frontal sinus dimensions and morphology. However, each of them evaluates only a small number of variables; none deals with the issue comprehensively, and they apply different methods. This variability can pose problems for comparisons and transfer of knowledge to other studies (Kjaer et al., 2012). Finally, the application of modern imaging methods to enable a much more exact analysis is still in the early stages (Chaiyasate et al., 2007). With regard to these facts, the aim of this work is to show the relationship between the individual basic variables (i.e. total surface area and volume, maximum height and width, and morphology) of the frontal sinus and biological kinship.

Using a comprehensive approach based on CT images respecting previously used approaches, we aim to answer the following questions in this order:

First, are there any significant differences between males and females within families in frontal sinus measurements? Whereas significant sex differences in frontal sinus size across different populations have been demonstrated (Akhlaghi et al., 2016; Choi et al., 2018; Čechová et al., 2019), the unresolved question remains as to whether there are significant differences within families. Therefore, we seek to verify that we do not have to take the sex of individuals into account when detecting family relationships because sexual differences are not significant among biologically related individuals.

Secondly, are there positive relationships between the degree of frontal sinus morphological and dimensional similarity and the biological distance of biologically related individuals? Can these variables be used for kinship analysis?

Finally, are there any significant differences in the degree of frontal sinus morphological and dimensional similarity between members of families with very close consanguineous relationships and members of parallel‐related families with common degrees of consanguinity?

2. MATERIALS AND METHODS

2.1. Materials

The skulls of 41 adult individuals (25 males and 16 females) aged 20–90 years with known genealogical data were available (Cvrček et al., 2018). They were members of five branches of one family over four generations in the 19th–20th centuries from Bohemia, Czech Republic (family tree: Supporting Information No. 1, Figure S1, generation Nos 4–7). The skeletal remains researched were provided to us based on written permission from direct descendants of the studied individuals. Research on the remains was made possible by the repair of family tombs and coffins and relocation of the remains. After the research concluded, the skeletal remains were re‐buried.

There were two consanguineous marriages and their descendants in the sample. The first instance involved doubly biologically related spouses Nos 7 and 8 (first cousins and second cousins, generation No. 4). From them, six offspring could be evaluated: sons Nos 9, 11 and 13, and daughters Nos 12, 14 and 17 (generation No. 5). The next two generations from daughter No. 17 were also available: male No. 19 (generation No. 6) and his sons, Nos 20 and 21 (generation No. 7). Son No. 13 also had a biologically related wife, No. 37 (generation No. 6).

The second case of individuals with very close consanguinity is represented by male No. 41 and his son, No. 42. The mother of the son is the second cousin of his father (generation Nos 5–6), but her remains were not available. Their second son, No. 43, was a neonate; therefore, he was not included in this study.

Skulls were scanned at the Radiology Department of the Hospital Na Homolce in Prague by computed tomography (CT) using a Somatom Sensation 16 scanner (Siemens, Erlangen, Germany). Acquisition parameters were optimally set with a voxel size 0.49 mm, a slice thickness 0.6 mm and a slice increment 0.3 mm.

2.2. Methods

Surface models of frontal sinuses were segmented from the CT scans in Morphome3cs software (www.morphome3cs.com) (Musilová et al., 2016; Čechová et al., 2019) and were further processed and cleaned in meshlab (Visual Computing Lab, Italian National Research Council), Avizo 7.1 (Visualization Sciences Group, SAS) and geomagic studio 2012 (Raindrop Geomagic GmbH, 3D Systems).

Of the total number of 41 individuals, 10 have large postmortem defects: mainly, the bottom of the sinuses is missing or the orbital roofs are damaged (Nos 5, 11, 12, 14, 19, 23, 28, 32, 39, 40). In eight other individuals, the frontal bone was cut during autopsy (a common means of controlling sinus pathological changes in that period, Nos 6, 9, 19, 21, 24, 38, 43, 44); several cases involved chopping directly into the lower parts of the frontal sinuses and roofs of the orbits (Figure 1). Therefore, whole sinuses could not be evaluated, nor could the supraorbital line (SOL), most often used to evaluate radiographs, be employed (Cameriere et al., 2008; Nikam et al., 2015; Nikolova et al., 2018). With regard to these restrictions of our osteological material, we have modified the approach and used the external supraorbital line (‘ESOL’) instead. All of the meshes were cut along a plane that passes through the upper orbital margins and is parallel to the Frankfurt plane (Borovanský, 1936; Nussen, 2007). A portion superior to this plane was targeted for examination; this part was completely preserved in all individuals. Finally, pictures of frontal sinuses were acquired in the frontal view so that the cutting plane was perpendicular to the computer screen.

Figure 1.

CT scan of a skull with a highlighted frontal sinus and ‘external supraorbital line’ (ESOL, red line), individual No. 17 [Colour figure can be viewed at wileyonlinelibrary.com]

2.2.1. Sex differences

Sex classification was performed in morphome3cs software using fitting support vector machines (SVM) on the combinations of total surface area and volume and maximum height and width of the frontal sinuses. Equations created for the contemporary Czech population were used on this sample (Čechová et al., 2019). The sample on which the equations were based included 45 females and 58 males, aged 21–84 years. SVM carries out classification by finding the hyperplane that maximizes the boundary between the two groups. A radial kernel was employed and the parameters were set depending on the variables to obtain the best results. The SVM algorithm was trained and tested on the whole sample. For more objective results, leave‐one‐out cross‐validation was performed to detect overfitting.

2.2.2. Evaluation of dimensions

For each individual, four basic dimensions were found: total surface area and volume using meshlab, and maximum height and width using Avizo 7.1. For each variable, the degree of dimensional similarity (‘DS’) between two individuals was calculated based on the formula originally designed by Szilvássy (1974) for expression of frontal sinus side asymmetry:

$$\text{DS}=100\times \frac{\mathrm{A}}{\mathrm{B}}$$

where A is an individual with a smaller value and B is an individual with a larger value of the given variable. The higher the value of DS, the greater the degree of similarity between individuals. The resulting DS values of each dimension represent the basic data for further analysis:

• We calculated the logarithm of the coefficient of the relationship (logarithm of biological distance) for each possible pair in the sample (820 combinations). The coefficient (r) was calculated using a tabular approach (Falconer and Mackay, 1996) based on the method of VanRaden (1992). Use of a logarithm (logR) eliminates entirely unrelated individuals (their coefficient of relationship is always zero; thus, the logarithm is not defined) and allows graphical expression of the relationship between the degree of similarity and the biological distance only for biologically related individuals (Cvrček et al., 2018). We used a linear model to test the relationship between the DS of each dimension and the logarithm of the biological distance for all related pairs in the sample (level of significance α = 0.05). For the graphical representation using a linear model, we performed arcsine transformation of the data on the Y‐axis (DS).

• We also used cluster analysis (Murtagh, 1985), which was performed by the complete linkage hierarchical method, where the distance between two clusters was determined as the distance between their most distant individuals. The DS of each pair was converted into a distance such that the greatest similarity indicated the smallest distance and vice versa.

• We compared the degree of similarity between family with common degrees of consanguineous relationships (group A, individuals Nos 1, 2, 3, 4, 5, 6, 38) and parallel‐related family with very close consanguinity (group B, individuals Nos 7, 8, 9, 17, 18, 19, 20). Differences were tested using the Wilcoxon signed‐rank test (level of significance α = 0.05) (Siegel, 1956). The resulting p‐values were adjusted by Bonferroni correction (Bonferroni, 1936). The variation in each group was also compared using box plots.

2.2.3. Morphological evaluation

Visual characters were evaluated from pictures of 3D models in frontal view. Nine features were evaluated based on previous publications (Szilvássy, 1986; Yoshino et al., 1987; Reichs, 1993) (Table 1).

Table 1.

Visual features and their classifications for expression of frontal sinus morphology

Category	Classification	Class number
Total size, Yoshino et al. (1987)	Small (0–6)	1
	Middle (6–12)	2
	Large (12–18)	3
	Very large (> 18)	4
Bilateral asymmetry, Yoshino et al. (1987)	Symmetry (100–80)	1
	Slight asymmetry (80–60)	2
	Moderate asymmetry (60–40)	3
	Strong asymmetry (40–20)	4
	Extreme asymmetry (˂ 20)	5
Superiority, Reichs (1993)	Absent	0
	Left equal to right	1
	Left greater than right	2
	Right greater than left	3
Outline of upper border for left side and right side separately, Yoshino et al. (1987)	Absent	0
	Smooth	1
	Scalloped with 2 arcades	2
	Scalloped with 3 arcades	3
	Scalloped with 4 arcades	4
	Scalloped with > 5 arcades	5
Partial septa, Yoshino et al. (1987)	Absent	0
	Present in the left side	1
	Present in the right side	2
	Present in the both sides	3
Supraorbital cells, Yoshino et al. (1987)	Absent	0
	Present in the left side	1
	Present in the right side	2
	Present in the both sides	3
Shape for left side and right side separately, Szilvássy (1986)	Bean shape	1
	Leaf shape	2
	Fan shape	3
	Pyramid shape	4
Lateral range for left side and right side separately (own category)	Absent	0
	˂ 1/3 border of orbit	1
	˂ 1/2 border of orbit	2
	> 1/2 border of orbit	3
	> lateral border of orbit	4

The degree of morphological similarity (‘MS’) between biologically related pairs of individuals based on class numbers of the above‐described categories was calculated using cosine similarity (Singhal, 2000). Based on these data, we could use a statistical evaluation similar to that used for the metric dimensions (points 1–3).

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy and ethical restrictions. The data that supports the findings of this study are available in the supplementary materials for this article.

3. RESULTS

3.1. Sex differences

Neither combination of variables (total surface area and volume, maximum height and width) provides a clear distinction between males and females. No pattern is observed in the data; thus, there is no reason to separate males and females (Figure 2). Males and females overlap, and some males have much smaller sinus parameters than females (e.g. males Nos 3, 18, 43 vs. females Nos 2, 10, 44, etc.). At the same time, our attempt at classification by sex based on the combination of total surface area and volume showed considerable error. The results after cross‐validation suggested a success rate of only 64.1%, and no females were correctly classified (SVM posterior success rate of the male group = 100.0%, SVM posterior success rate of the female group = 0.0%). A combination of maximum height and width provided a similar result—56.4% of the success rate after cross‐validation—but the SVM posterior success rate of the male group was 100.0%, whereas that of the female group was only 14.3% (n = 2; Figure 2). This result very strongly suggests no discernible sex differences.

Figure 2.

Graphical expression of total surface area (s), volume (v), and maximum height (h) and width (w) data by sex of individuals (F = females, M = males). The right graph shows two correctly classified females

Individuals Nos 12 and 40 were excluded from the assessment due to their incomplete preservation; only total width could be evaluated.

3.2. Dimensional similarity

The resulting values of the degree of similarity between individuals based on frontal sinus total surface area, volume, and maximum height and width are shown in Supporting Information No. 2 (Table S1–S4). Due to incomplete preservation, individuals Nos 12 and 40 were excluded from assessments of total surface area, volume, and maximum height.

There is no statistically significant relationship between the biological distance of biologically related individuals and the degree of frontal sinus similarity based on total surface area (p = .133; Figure S2), volume (p = .060; Figure S3) or width (p = .304; Figure S4). However, based on maximum height, the degree of frontal sinus similarity was statistical significant (p = .031; Figure 3). The greater the degree of biological relationship, the greater the degree of similarity. In any case, this does not always apply. Within closely related individuals, there may be a greater degree of similarity between some cousins (r = .125, logR = −0.90309) than among other close relatives, such as parents and children (r = 0.5, log = −0.30103) or grandparents and grandchildren (r = .25, log = −0.60206). For example, cousins Nos 8 and 31 (DS^h = 85.7) and mother No. 8 and daughter No. 17 (DS^h = 66.7), cousins Nos 1 and 7 (DS^h = 100.0) and male No. 1 and his son No. 3 (DS^h = 44.4), or cousins Nos 1 and 7 (DS^h = 100.0) and male No. 7 and his grandsons Nos 20 and 21 (DS^h = 75.0, 85.7 respectively).

Figure 3.

Relationship of the similarity between biologically related individuals based on the maximum height of the frontal sinuses (DS^h, y‐axis) by log‐relatedness of individuals (x‐axis). The red line represents the regression line. Note: log‐relatedness disregards unrelated individuals [Colour figure can be viewed at wileyonlinelibrary.com]

Figures [Link], [Link], [Link], [Link] show the degrees of similarity of all evaluated individuals using cluster analysis based also on total surface area, volume, and maximum height and width. The results reflect some known genetic relationships.

In total surface area, the closest are father No. 18 and son No. 19, father No. 7 and son No. 9, and mother No. 23 and daughter No. 24 (Figure S5).

In total volume, the closest are grandfather No. 3 and grandson No. 6, grandmother No. 31 and granddaughter No. 37, and brothers Nos 9 and 11 (Figure S6).

In maximum height, the closest are brothers Nos 18 and 22, grandfather No. 32 and grandson No. 36, and siblings Nos 11 and 14 (Figure S7).

Finally, in maximum width, the closest are father No. 18 and son No. 19, grandmother No. 17 and grandson No. 19, and mother No. 34 and son No. 36 and daughter No. 37 (Figure S8).

Furthermore, there is greater variability in the degree of similarity among the parallel family with common degrees of consanguinity (group A) compared with the family with very close consanguineous relationships (group B) in all dimensions (total surface area, volume, and maximum height, width) (Figure 4). Each group contains 17 pairs of individuals. Statistically, there are significant differences only in total surface area (p = .031) and volume (p = .012); there is no significant difference in height (p = .116) or width (p = .380).

Figure 4.

Variability in the degree of similarity between family with common degrees of consanguinity (a) and family with very close consanguineous relationships (b) for frontal sinus total surface area, volume, and maximum height and width

3.3. Morphological similarity

Figures showing the frontal sinus are provided in Supporting Information No. 3 (Figures [Link], [Link], [Link], [Link], [Link]). The figures are divided by family. Based on a simple visual description of sinus morphology, several specific features and several configurations among close relatives have been identified. There is the laterally stretched arcade separated by a wide notch in the right sinus in the inbred brothers Nos 9, 11 and 13 or a specific notch (mirror‐occurring) with pointed peaks in the same brothers, Nos 9 (near medial border of the left sinus) and 11 (near lateral border of the right sinus). Mirror similarity can also be observed in their double biologically related parents Nos 7 and 8 (first cousins and second cousins) in the overall character of their sinuses: the right sinus of male No. 7 and the left sinus of female No. 8 have a fan shape (even a ginkgo shape) with arcade category 5, and the left sinus of male No. 7 and the right sinus of female No. 8 are irregularly lobed, with arcade category 3 (Figure S10). Similarly, we found a distinct lateral extension in male No. 39 (right side) and his son No. 41 (left side). These similarities are shown in Figure 5.

Figure 5.

Examples of various morphological similarities (arrows, clamps) shared by closely related individuals (spouses and first cousins Nos 7 and 8, their sons Nos 9, 11 and 13, and father No. 39 with son No. 41). Shared similarities are marked with the same colors [Colour figure can be viewed at wileyonlinelibrary.com]

Manifestations of the morphological similarities of all evaluated individuals are shown in Supporting Information No. 2 (Table S5). Seven individuals whose patterns could not be completely evaluated had to be excluded from the evaluations due to poor preservation (Nos 11, 12, 19, 32, 39, 40, 45). The difference in the degree of frontal sinus morphological similarity (Table S6) and the biological distance of biologically related individuals (p = .034) is statistically significant; that is, the greater the degree of biological relationship, the greater the degree of morphological similarity (Figure 6). However, as with the maximum height of the frontal sinus, within closely related individuals, there may be a greater degree of similarity between some cousins (r = .125, log = −0.90309) than among other close relatives, such as parents and children (r = .5, logR = −0.30103) or grandparents and grandchildren (r = .25, logR = −0.60206), such as cousins Nos 8 and 31 (MS = 0.86) and mother No. 31 and daughter No. 34 (MS = 0.77), cousins Nos 1 and 7 (MS = 0.89) and male No. 1 and his son No. 3 (MS = 0.88), or cousins Nos 1 and 7 (MS = 0.89) and male No. 7 and his grandsons Nos 20 and 21 (MS = 0.75). Some of these cases are also evident when we talk about their overall morphology in the picture, such as cousins Nos 1, 7 and 39 (Figures S9, S10 and S13), and male No. 1 and his son No. 3 (Figure S9). There may also be a case where completely unrelated individuals are very similar, even more than to their children, such as spouses Nos 1 and 2 and their son No. 3 (Figure S9). The only case of full agreement in frontal sinus morphological pattern is shown in brothers Nos 20 and 21 (Table S5).

Figure 6.

Relationship of the similarity between biologically related individuals based on the morphology of the frontal sinuses (MS, y‐axis) by log‐relatedness of individuals (x‐axis). The red line represents the regression line. Note: log‐relatedness disregards unrelated individuals [Colour figure can be viewed at wileyonlinelibrary.com]

Figure S14 shows the degrees of similarity of all evaluated individuals using cluster analysis based on morphology. This figure reflects only some relationships. The greatest degrees of similarity are between uncle No. 33 and niece No. 44 and between half‐siblings Nos 7 and 31.

There is greater variability in the degrees of similarity among parallel family with common degrees of consanguinity (group A) than family with very close consanguineous relationships (group B) in morphology (whiskers), as shown in Figure 7. Group A contains 17 pairs of individuals, and group B contains 11 pairs of individuals due to the exclusion of individual No. 19. However, there are also outlying values, such as the similarity between mother No. 4 and son No. 5, and grandmother No. 4 and grandson No. 6 (group A) and the similarity between great‐grandfather No. 7 and great‐grandson No. 20 (group B). There is no statistically significant similarity between these groups (p = .890).

Figure 7.

Variability in the degree of similarity between family with common degrees of consanguinity (a) and family with very close consanguineous relationships (b) for frontal sinus morphology

4. DISCUSSION

For this study, we used a new method of frontal sinus demarcation using the ‘external supraorbital line’ (ESOL). This approach better reflects the preservation limits of bioarchaeological and forensic material. Frontal sinuses are protected by the solid frontal bone; for that reason, we can evaluate them even when remains are poorly preserved (Cameriere et al., 2008; Goyal et al., 2013), though some parts of the frontal bone are often exposed to decomposition processes and artificial interventions. In our material, this is particularly true for the thin sinus base, the ethmoidal labyrinth and the orbital roofs. In addition to postmortem damage or deformation (24.4% of individuals), the sinuses are often subjects of investigation for pathological changes during autopsy (19.5% of individuals). This represents a very significant limitation for our work, as almost half of the whole sample is impacted. Thus, past approaches, such as using the ‘supraorbital line’ (SOL) (Christensen, 2004; Cameriere et al., 2008; Nikam et al., 2015; Nikolova et al., 2018) or evaluation of complete sinuses, including the base (Čechová et al., 2019), cannot be used for our purposes. Our proposed approach bypasses these limitations. In addition, the results do not seem to be particularly dependent on the method used and are comparable to methodically different studies if the defined procedure is always followed. This will be discussed below.

4.1. Sex differences

There were no significant differences in any of the studied sinus dimensions between males and females. This suggests that although there are sex differences evident within normal populations, i.e. in randomly selected unrelated individuals (Uthman et al., 2010; Sai Kiran et al., 2014; Benghiac et al., 2015; Michel et al., 2015; Akhlaghi et al., 2016; Hamed et al., 2016; Motawei et al., 2016; Choi et al., 2018), the sex differences are not noticeable within families. We do not have to take into account the sex of individuals when investigating family relationships.

There are significant differences between males and females in the total volume and surface area of the frontal sinuses in randomly selected individuals from the contemporary Czech population, which is genetically similar to our sample (Čechová et al., 2019). At the same time, the success rate of sex prediction was substantially decreased when using contemporary sex standards. The cited study also showed that there are greater variabilities in volume and surface area of the frontal sinuses of Czech males than females. This is also suggested by our results. However, a comparison of linear dimensions is currently unavailable. Further verification will be required for other genealogically documented material.

4.2. Dimensional similarity

According to our results, there are no significant relationships between biological distance and size similarity based on total surface area, volume or width. However, we found a significant relationship with maximum height. According to Kjaer et al. (2012) who evaluated lateral radiographs of 42 pairs of monozygotic twins (18 male and 24 female pairs) from the recent Danish population, significant differences between two individuals were found only within one female twin pair and within two male twin pairs in the maximum height of the frontal sinus. There were no significant differences within twin pairs in other studied cases. Although different approaches were used, we can consider these results to support our study in that kinship affects the similarity of frontal sinus maximum height. At the same time, it turns out that the simple formula originally designed by Szilvássy (1974) for expression of frontal sinus side asymmetry is appropriate for our purposes.

Comparison information from past studies is lacking. Other studies using linear dimensions (especially Szilvássy et al., 1987; Slavec, 2004, 2005) lack numerical expressions of degrees of similarity between related individuals and statistical data processing. Also, comparison of their surface area and volume are missing. This is mainly because the frontal sinuses previously could only be evaluated using X‐ray images (Slavec, 2005; Kjaer et al., 2012). However, according to our results, radiographs seem to be quite sufficient for the detection of related individuals. Although the relationship between the width of the frontal sinus and biological distance has not proved significant, our study suggests that linear dimensions could be more important than surface area and volume. At the same time, unlike modern imaging methods, X‐ray examination makes it easier to evaluate even incomplete remains where it would be meaningless to use surface models.

Considering all of the dimensions examined, only a few documented family relationships were reflected in the cluster analysis; less than shown by non‐metric traits (Cvrček et al., 2018). However, the closest pairs are again the closest relatives, such as father and son, mother and daughter, siblings, and grandparents and grandchildren.

4.3. Morphological similarity

There is a statistically significant relationship between biological distance and morphological similarity in the frontal sinuses. The same conclusions reported for the measured dimensions above apply to the results of the morphological trait cluster analysis. The closest pairs in the cluster analysis are closest relatives (father and son, mother and daughter, siblings, and grandparents and grandchildren).

The scoring systems designed by Szilvássy (1986), Yoshino et al. (1987), Reichs (1993) and other authors (Kim et al., 2013) allow the use of frontal sinus morphology for kinship analysis. We came to the same conclusion in our analysis of non‐metric traits on the same sample with known genealogical data (Cvrček et al., 2018); that is, the greater the degree of biological relationship, the greater the degree of morphological similarity. However, we emphasize that the degree of morphological similarity of biologically related individuals based on the frontal sinuses is less than that using non‐metric traits (Cvrček et al., 2018). In addition, identification of exact genealogical position in the family tree of an individual using frontal sinus analysis is impossible (Cameriere et al., 2008). The same can be seen in skeletal non‐metric traits (Ricaut et al., 2010).

The overall morphology of the sinuses can vary, even if they have the same pattern, such as in two brothers in our sample (Nos 20 and 21). Although we can consider categorical evaluation to be more objective (Cameriere et al., 2008) than visual assessment of overall morphology (Asherson, 1963), in our opinion, it is possible to find a subjective level of similarity even using categorical evaluation. This is especially true for assessing the number of arcades. In some cases, however, it may be difficult to determine which arcades should be counted, and whether the small arcades are large enough and their boundaries are well enough marked. Therefore, the simple verbal description of overall character of the frontal sinuses including specific traits (Szilvássy et al., 1987) is still not negligible for expressing similarity among individuals. For non‐metric traits, statistical evaluation does not allow for the capture of some of the markers that further highlight the biological relationships of individuals (e.g. inbred brothers Nos 9, 11 and 13). We consider the mirror similarity of sinuses as one of the most important markers that can indicate biological affinity. In the past, this phenomenon was observed in twins (Asherson, 1963), making it apparent that such sinus development is genetically influenced. However, in that case, the authors argued that the twins were not similar (Asherson, 1963). Although this conclusion cannot be completely denied, we should always consider that this feature may occur, and we should take this possibility into account. The importance of the perception of possible side asymmetry in trait expression, which serves to verify similarity on opposite sides, has already been demonstrated in non‐metric osteological traits (Cvrček et al., 2018). In this study, this is especially apparent in the case of doubly biologically related spouses Nos 7 and 8, their sons Nos 9 and 11, or in male No. 39 and his son No. 41. With emphasis only on the character and peculiarities of the sinuses, we can reduce the subjectivity of the descriptive approach (Szilvássy et al., 1987). The use of both scoring systems and morphological descriptions should be combined (Slavec, 2004, 2005), as supported by our results.

Combined assessment of dimensions and morphology is useful for identification of biological affinity (Slavec, 2004, 2005). Our results confirm that there are similarities between biologically related individuals in both dimensions and morphology. In addition to combining approaches to the evaluation of sinus morphology, we consider it important to compare sinuses (dimensions, morphology) and non‐metric traits where possible. We have shown the obvious similarity of unrelated spouses Nos 1 and 2 and their son No. 3, who is not as similar to them as is the son of No. 38. Their identity is undoubtedly certain, and non‐metric traits confirm that there is a greater degree of similarity between female No. 2 and her son No. 3 than between female No. 2 and her husband No. 1 (Cvrček et al., 2018).

4.4. The rate of similarity between cousins and the influence of inbreeding on variability

It is not absolutely true that the greater the degree of biological relationship, the greater the degree of dimensional or morphological similarity. This has been confirmed in several previous studies using osteological non‐metric traits (Gemmerich‐Pfister, 1999; Velemínský and Dobisíková, 2005; Cvrček et al., 2018). Within closely related individuals, a greater degree of similarity may exist between first cousins or grandparents and grandchildren than between more closely related parents and children or siblings, as illustrated in the well‐known case of the physical resemblance of first cousins King George V, Tsar Nicholas II and Prince Henry of Prussia (Hall, 2018). Previously, based on a small number of studies and limited materials, it was suggested that this resemblance could be an accident or an error due to the limitations of statistical evaluation (Velemínský and Dobisíková, 2005; Cvrček et al., 2018). However, this phenomenon appears to be common among relatives, and the occurrence of such cases may be relatively constant.

Our results suggest that inbreeding can affect the degree of frontal sinus dimensional and morphological variability by reducing it. Although the differences are mostly not statistically significant, similar to the situation for non‐metric traits (Cvrček et al., 2018), they can clearly be seen when presented graphically. Therefore, we consider it important to combine these approaches to identify such differences.

Focusing on single dimensional variables and morphological features showed that the height and morphology of the frontal sinuses correspond to biological family relationships. However, the introduction of this new approach for the identification of relatives would in practice require that a comparative sample of unrelated individuals from a similar population and period be analyzed as a next step, with a focus on comparing calculated similarity with a sample of biologically related individuals, including questioning how often two people might show a certain level of morphological similarity purely by chance.

5. CONCLUSIONS

There were no significant sex differences between males and females within families based on frontal sinus dimensions. On the other hand, the relationship between the degree of biological relationship and the degree of similarity according to total height and morphology of the frontal sinus has been demonstrated. The greater the degree of biological relationship, the greater the degree of similarity for both variables. At the same time, however, there were cases of greater similarity between first cousins or grandparents and their grandchildren than among much closer relatives such as siblings or parents and their children. Use of total height and morphology for the detection of biological affinity between individuals is possible, but we cannot use frontal sinuses to identify the exact relationships of individuals, or their exact genealogical position in the family tree. Although the differences in the assessed variables of the frontal sinus between a family with a common degree of consanguinity and a family with very close consanguineous relationships are not always significant, the reduced variability where there is a greater degree of consanguinity is evident.

Supporting information

Fig S1

Fig S2

Fig S3

Fig S4

Fig S5

Fig S6

Fig S7

Fig S8

Fig S9

Fig S10

Fig S11

Fig S12

Fig S13

Fig S14

Table S1‐S5

Cvrček J, Rmoutilová R, Čechová M, et al. Biological relationships and frontal sinus similarity in skeletal remains with known genealogical data. J. Anat. 2020;237:798–809. 10.1111/joa.13246