Tiger (Panthera tigris) skeletons, as well as individual bones, skulls, teeth, and claws, are very lucrative commodities on the wildlife black market (Mills and Jackson 1994; Wong and Krishnasamy 2022; Nittu et al. 2023). Demand in Asian countries far outstrips illegal supply, so tiger bones are often counterfeited and interchanged for lion (Panthera leo) bones due to their strong similarity (Williams et al. 2015a, b; EIA 2017; Williams et al. 2017). The ability to distinguish tiger and lion skeletons based on morphological features can be very useful when working with law enforcement in the field. Several deterministic characters of tiger and lion skulls are described in the literature (Blanford 1888; Boule 1906; Pocock 1929; Merriam and Stock 1932; Haltenorth 1936, 1937; Hemmer 1966; Christiansen 2008; Christiansen and Harris 2009). Unfortunately, none of these hallmarks is 100% reliable, and a combination of several is necessary to minimize the possibility of confusion (Christiansen 2008). For fieldwork, it is preferable to use easily recognizable hallmarks that can be inspected even from photographs of the skull, and which can be consulted with a zoological expert remotely. One of the so far neglected determinative structures is the foramen ovale and its orientation on the cranial base. The foramen ovale is an opening in the wing of the os basisphenoidale. In cats, it is found in the space between foramen rotundum and fissura sphenopetrosa. Laterally from the foramen ovale, erected joint surfaces of the jaw joint are bordered from the rear by protrusions processus retroarticulares. Caudally, the foramen ovale is adjacent to the processus muscularis protruding from the bulla tympanica of the temporal bone. Differences in the foramen ovale in tigers and lions are mentioned by Haltenorth (1936, 1937) or Hemmer (1966), but the potential of this hallmark to distinguish between species has not been formally assessed.

The material for this study consisted of 112 dried skulls of big cats (Table 1). Skulls with fully developed permanent dentition were classified as adults, and skulls with fully or partially deciduous dentition as juveniles. Both adults and juveniles were included in the study so that the new determinant feature could be assessed in as wide a range of individuals as possible. Studies dealing with the morphology of tiger and lion skulls often work with skulls of adult animals only (Christiansen 2008). However, to combat illegal trade, it is necessary to identify juvenile skulls as well as adult skulls. Therefore, records of the morphological structure of juvenile skulls were considered as also important.

Table 1 Species, sex, and age of the tested individuals
Full size table

The subspecies by Wilson and Reeder (2005) have been recorded in 36 tiger skulls (19 P.t.altaica, 4 P.t.corbetti, 1 P.t.sondaica, 11 P.t.sumatrae, and 1 P.t.tigris) and in 20 lion skulls (2 P.l.bleyenberghi, 5 P.l.leo, and 13 P.l.persica). The skulls came from several sources: Czech National Museum, zoological gardens of the Czech Republic and Slovakia, University of Veterinary Sciences Brno, Faculty of Science of Charles University in Prague, Czech Environmental Inspectorate, and Customs Administration of the Czech Republic. Six tiger and 6 lion skulls originated from animals in the wild, 42 tiger and 36 lion skulls were from animals bred in captivity, and the origin of 7 tiger and 15 lion skulls was unknown. Whilst interpopulation variation and captivity status are known to influence skull morphology in both species (Zuccarelli 2004; Mazák 2010a, b; Yamaguchi et al. 2013; Hartstone-Rose et al. 2014; Cooper et al. 2022), these factors were not included in the analysis due to limited sample size. The morphology of the skull can also be influenced by the health status of the animal (Shamir et al. 2008; Saragusty et al. 2014), therefore, only skulls without obvious deformities were used in this study.

Two parameters were measured from the skulls using a vernier calliper (Fig. 1):

  1. (1)

    Greatest skull length (GSL): the length from the most anterior part of the rostrum (excluding teeth) to the most posterior point of the skull.

  2. (2)

    Zygomatic breadth (ZB): the greatest distance between the outer margins of the zygomatic arches.

Fig. 1
figure 1

Lion and tiger skulls from ventral view, visualization of the parameters measured: GSL—greatest skull length, ZB—zygomatic breadth, A—the midline, boundary between the left and right half of the skull, B—the direction of the line passing through the foramen ovale, LFOA—the angle of the left foramen ovale (angle between A and B on the left), RFOA—the angle of the right foramen ovale (angle between A and B on the right), SFOA—the angle that the two foramina ovalia make together (LFOA + RFOA). Origin of the skulls presented: left and middle—lion skull Panthera leo bleyenberghi, adult female, collection of the Lesna Zoo, ID Lesna2; right—tiger skull Panthera tigris altaica, adult female, collection of the Czech Nat. Museum, ID 48548

Full size image

Digital photographs in high resolution (6000 × 4000 pixels) were taken of each specimen using a NIKON D5500 digital camera. To evaluate the orientation and appearance of the foramina ovalia, an image of the ventral aspects of each skull was used. For each photograph, the appearance of the prominence of the left (LFO) and right (RFO) foramen ovale was described. The orifice of the foramen ovale could be bordered by a semi-circular margin of the os basisphenoidale (C) and thus completely visible (V), or it could be obscured behind the straight margin of the os basisphenoidale (S) and either completely hidden (H) or partially visible (V). The orientation of the LFO and RFO in all photographs was examined independently by two observers (O1 and O2), each of them taking two measurements (trials a and b) at least 1 week apart. They established the midline “A” (the exact boundary between the left and right halves of the skull) and a line passing in the direction of the foramen ovale projection “B”. The angle between these two lines was measured on both sides of the skull using MB-Ruler 5.4 software (Markus Bader, MB-Softwaresolutions, 2022) and expressed as an integer. Thus, for each skull, four sets (O1a, O1b, O2a, O2b) of the following data were obtained: the angle of the left foramen ovale (LFOA), the angle of the right foramen ovale (RFOA), and the angle that the two foramina ovalia make together (SFOA) (Figs. 1, 2, for data see SI1).

Fig. 2
figure 2

Foramina ovalia (highlighted with circles), detailed view of lion and tiger skulls from the ventral side: a adult lion skull, foramina ovalia directed laterally with their outlets partially hidden behind the straight edge of the os basisphenoidale (Panthera leo, adult female, collection of the Czech National Museum, ID 12082); b juvenile lion skull, foramina ovalia directed laterally with outlets entirely hidden (Panthera leo persica, juvenile female, collection of the Czech National Museum, ID 46396); c adult tiger skull, foramina ovalia facing in rostral direction with their outlets bounded by a semi-circular edge of the os basisphenoidale (Panthera tigris corbetti, adult female, collection of the Czech National Museum, ID 47673); d juvenile tiger skull, foramina ovalia directed laterally with outlets entirely visible (Panthera tigris altaica, juvenile, sex unknown, collection of the Ostrava Zoo, ID 37312)

Full size image

The intra- and inter-observer reliability of measured values of LFOA and RFOA were tested in SPSS 28.0.1.0 (IBM SPSS Inc., 2021) using the intraclass correlation coefficient (ICC) and standard error of measurement (SEM) (Weir 2005). Reliability analysis with a two-way mixed model was applied (Koo and Li 2016) to all skull samples except the skull with ID Sumi (n = 111). This skull was not evaluated by one of the observers due to the quality of the photograph (perceived as of too low quality for evaluation by O2). The results of intra- and inter-observer reliability are presented in Table 2.

Table 2 The results of intra (absolute agreement type analysis, results concerning single measures) and inter (consistency type analysis, results concerning single measures) observer reliability
Full size table

According to Koo and Li (2016), ICCs of O1 and O2 indicate good (0.75–0.90) and excellent (greater than 0.90) intra observer reliability, respectively. The inter-observer reliability varied from moderate (0.50–0.75) in RFOA measurements to good (0.75–0.90) in LFOA measurements. Based on the results, the following analyses were counted separately for all datasets (O1a, O1b, O2a, O2b) to be sure of the reliability of the conclusions. Further results are valid for all four datasets.

Tests of normality, assumptions for linear discriminant analysis (LDA), and decision trees were calculated in SPSS 28.0.1.0 (IBM SPSS Inc., 2021). Further analyses were processed in TIBCO Statistica 14.0.0.15 (TIBCO Software Inc., 2021) and included data from all 112 skulls unless stated otherwise. Based on the results of Shapiro–Wilk, and Kolmogorov–Smirnov tests (Table 3), nonparametric measures and tests were used beside the chi-squared test. Specifically, Spearman correlation coefficients (ϱ), Kendall rank correlation coefficients (τ), and Mann–Whitney U tests were calculated. The datasets did not meet the assumptions for LDA calculation differentiating the species based on SFOA, GSL, and ZB (Box’s Test of Equality of Covariance Matrices: PO1a < 0.001, PO1b < 0.001, PO2a < 0.001, PO2b < 0.001). Instead, decision trees were calculated as an alternative (Feldesman 2002) with one (x = 1) as a minimal number of cases in both parent and child nodes. A significance level α = 0.05 was established for all calculations. All results were rounded to three decimal places except the values in %, which were rounded to one decimal place.

Table 3 Results of normality tests in selected variables (significant results are marked by an asterisk)
Full size table

Lion skulls typically showed foramina ovalia directed laterally (Fig. 2a—adult, Fig. 2b—juvenile). From the ventral view, 89.5% (n = 51) of lion skulls, including all juveniles, showed at least one outlet of foramen ovale fully or partially hidden behind the straight edge of the os basisphenoidale. Both outlets of foramina ovalia were entirely hidden from ventral view in 49.1% of cases (n = 28).

As expected, adult individuals reached significantly higher values of GSL (U = 17.000, Z = 4.942, P < 0.001) and ZB (U = 6.000, Z = 5.142, P < 0.001) than juveniles. On the other hand, no significant difference was found between the foramen ovale angles of adults and juveniles, regardless of whether the LFOAs, RFOAs, or SFOAs were compared except for significantly higher SFOAs of juvenile individuals in the dataset O1b, SI2a. Due to the exception in the dataset O1b, only the skulls of adult individuals were included in further analyses. Skulls of adult males showed significantly higher values of GSL (U = 35.000, Z = −3.438, P < 0.001) and ZB (U = 27.500, Z = −3.723, P < 0.001) when compared to adult females. However, no significant difference was found between the foramen ovale angles of adult males and females, regardless of whether the LFOAs, RFOAs, or SFOAs were compared (P > 0.05 in all cases), SI2a.

No significant correlation was found between any of the foramen ovale angles and GSL or ZB except for a weak negative correlation (Schober et al. 2018) between GSLs and RFOAs in the O1a dataset (ϱ = −0.311, P = 0.038 and τ = −0.216, P = 0.037; P > 0.05 in all other cases), SI2b.

In adult tiger skulls, the foramina ovalia were oriented more rostrally (Fig. 2c), however, values of LFOA and RFOA were much more variable than in lions (Fig. 3). The outlets of the foramina ovalia were in 65.3% of cases bounded by a semi-circular edge of the os basisphenoidale and therefore visible (n = 32), in 14.3% of cases, there were straight edges obscuring the foramina ovalia (n = 7), and in 20.4% of cases, there was a combination of one foramen ovale bounded by the semi-circular edge and the second one bounded by a straight edge (n = 10). In 88.2% of cases, when there was at least one straight edge of the os basisphenoidale, the outlets of both foramina ovalia were still visible (n = 15). In juvenile tigers (Fig. 2d), the foramina ovalia were directed more laterally, wide open, and entirely visible.

Fig. 3
figure 3

Foramina ovalia angles and shapes in all studied skulls: LFOA—the angle of the left foramen ovale, RFOA—the angle of the right foramen ovale, O1—observer 1, O2—observer 2, red—Panthera tigris adult, orange—Panthera tigris juvenile, dark blue—Panthera leo adult, light blue—Panthera leo juvenile, ■—straight edges of the os basisphenoidale, ●—semi-circular edges of the os basisphenoidale, ▲—one foramen ovale with the straight edge and second foramen ovale with the semi-circular edge of the os basisphenoidale, fully colored—foramina ovalia visible, uncolored—foramina ovalia hidden, half colored—one foramen ovale visible and second foramen ovale hidden

Full size image

As in lions, adult tigers reached significantly higher values of GSL (U = 0.000, Z = 3.955, P < 0.001) and ZB (U = 0.000, Z = 3.955, P < 0.001) than juveniles. Juvenile tigers showed significantly higher values of the foramina ovalia angles than adults [P < 0.01 in all cases, (Fig. 3, SI2c)]. Therefore, only the skulls of adult individuals were included in further analyses. In adult skulls, both GSL (U = 74.000, Z = −2.829, P = 0.005) and ZB (U = 90.000, Z = −2.345, P = 0.019) were found to be significantly higher in males than in females, but no significant difference was found between the foramen ovale angles of adult males and females, regardless of whether the LFOAs, RFOAs or SFOAs were compared except the dataset O1b where the males showed significantly bigger angles than females, SI2c.

No correlation was found between any of the foramen ovale angles and GSL or ZB in adult tigers (P > 0.05 in all cases), SI2d.

GSL did not significantly differ between species (all skulls: U = 1454.000, Z = 0.658, P = 0.511; adult skulls only: U = 882.500, Z = 1.661, P = 0.097). However, lions showed significantly higher values of ZB in comparison with tigers when the skulls of adult individuals were considered (U = 816.500, Z = 2.008, P = 0.045). Measurements of the observed angles in photographs revealed a significant difference between the angles (LFOA, RFOA, and SFOA) at which the foramina ovalia emerged in lions and tigers (Table 4).

Table 4 Results of comparing LFOA, RFOA, and SFOA between lions and tigers by the Mann–Whitney U test (significant results are marked by an asterisk)
Full size table

The results showed a significant relationship between the species and the shape of the bordering of foramina ovalia (LFO: χ2 = 47.105, df = 1, P < 0.001; RFO: χ2 = 34.350, df = 1, P < 0.001). Similarly, the relationship between species and visibility of the foramen ovale was significant (LFO: χ2 = 34.688, df = 1, P < 0.001; RFO: χ2 = 32.643, df = 1, P < 0.001). The shapes of the bordering of foramina ovalia and their visibility in connection to the species and age of tested individuals are available in Table 5.

Table 5 The shapes of the bordering of foramina ovalia and their visibility in connection to the species and age of tested individuals
Full size table

No relationship between the visibility of foramen ovale and sex of a lion was found (LFO: χ2 = 0.127, df = 1, P = 0.721.; RFO: χ2 = 0.072, df = 1, P = 0.789). However, a significant relationship was found between the visibility of foramen ovale and age of a lion (LFO: χ2 = 5.134, df = 1, P = 0.023.; RFO: χ2 = 7.118, df = 1, P = 0.008). The visualization of the differences in foramina ovalia appearance including angles in all studied skulls is shown in Fig. 3.

Selected variables (SFOA, GSL, ZB, LFO/RFO edge and visibility) were input as independent variables in the decision trees calculated from the data of all studied individuals except one lion and one tiger skull (n = 110). The overall predictability of the species ranged from 81.1 to 97.3%. The first classification step was always based on the assessment of SFOA which was always accompanied by either ZB (O1a and O2b) or GSL (O1b and O2a). LFO/RFO edge did not appear in the decision tree of any dataset. The visibility of LFO/RFO appeared as a variable improving the accuracy only in the dataset O1a. For details see SI3.

The studied characteristics of foramen ovale appear to be useful hallmarks to help distinguish between the skulls of lions and tigers. All sets of observations reached significant differences between lions and tigers, although one of the observers reached only good level of intra-observer reliability and the inter-observer reliability in the case of RFOA measurements reached only moderate level. On the other hand, foramina ovalia do not have the potential to discriminate between sexes. In tigers, considerable variability was found in the orientation and shape of the foramen ovale, while in lions, this structure was relatively stable.

When using this feature in field practice, it is impossible, nevertheless, to measure the exact angle of the foramen ovale, especially since different observers may measure this angle slightly differently. However, it can be estimated whether the foramina ovalia point in a distinctly rostral (at an angle of less than 50°) or distinctly lateral (at an angle of more than 70°) direction, which our results show, are nearly diagnostic for identifying tiger and lion skulls, respectively.

In general, in the skull of an adult animal, if the foramina ovalia were visible from the ventral view, bordered by a semi-circular edge of the os basisphenoidale and pointing in a rostral direction, it was a tiger skull. No forward-facing foramen ovale was found on any lion skull. If the foramina ovalia were invisible, hidden behind the straight edge of the os basisphenoidale and pointing in a lateral direction, it was a lion skull. In our collection, this description was also fully consistent with juvenile lion skulls. The juvenile tiger skulls, on the other hand, had a foramen ovale structure markedly different from that of adult tiger skulls. The foramina ovalia of the juvenile tigers were directed laterally, but unlike the juvenile lion skulls, they were fully visible in all cases. However, given the small number of juveniles examined, it is impossible to determine whether this difference reflects higher variability found in foramen ovale formation in tigers or the ontogenetic development of this structure. Nevertheless, the age of animal must be considered when evaluating the skulls in the field.

Skulls in which the foramina ovalia could not be unambiguously assigned to the above types were found in both lions and tigers. In adult animals it included mainly skulls where the foramina ovalia were oblique (at an angle between 50° and 70°) or skulls where the foramina ovalia pointed laterally but were not hidden behind the straight edge of the os basisphenoidale. In these situations, it is necessary to evaluate other features that can distinguish tiger and lion skulls (Blanford 1888; Boule 1906; Pocock 1929; Merriam and Stock 1932; Haltenorth 1936, 1937; Hemmer 1966; Christiansen 2008; Christiansen and Harris 2009). This is, of course, advisable in any case to confirm the determination. A complex set of features distinguishable from photographs that can be used to determine tiger and lion skulls in field practice was presented by Formanova et al. (2022).

The foramen ovale and its orientation on the cranial base of tiger and lion skulls can be successfully used to identify the skulls of these two species quickly. Rostrally orientated foramen ovale was detected only in tigers. In contrast, a laterally oriented foramen ovale, which is hidden behind the straight edge of the os basisphenoidale from the ventral view, was characteristic of lion skulls, including those of juveniles. Like the other characters described in the literature, foramen ovale cannot be used to distinguish the skulls of both species with absolute certainty. Combined with other characters, it can serve as one of the most determinative.