The LM3 of the mandibular fragment IGF4351 and the three isolated molars FS1996#Fi97 (LM3), FS1996#Fi98 (LM1), and FS1996#Fi99 (LM2) from Fiume Santo were detailed in 2005 by X-ray synchrotron radiation microtomography (SR-μCT) at the ID 17 Beam Line of the European Synchrotron Radiation Facility, Grenoble (experiment SC1749). The scans were performed with an intensity of 200 mA at energies of 51 and 70 keV (for the isolated specimens and the fragments, respectively), with an integration time of 22.6 ms by projections (1500/180°) of 60 ms. Data were collected by a 2.048 × 2.048 fibre-optical taper charge-coupled device Frelon camera (Mazurier et al., 2006). The reconstructions of the sections were saved in a 32-bit floating-point raw format with a voxel size of 45.5 × 45.5 × 45.7 μm.

Using Amira v.5.3 (Visualization Sciences Group Inc.) and ImageJ (Schneider et al., 2012), a semi-automatic, threshold-based segmentation was carried out following the half-maximum height method (HMH; Spoor et al., 1993) and the region of interest thresholding protocol (ROI-Tb; Fajardo et al., 2002) taking repeated measurements on different slices of the virtual stack (Coleman and Colbert, 2007). The specimens were thus digitally reconstructed in 3D using surface rendering allowing the quantification of the enamel cap, the dentine core volume, and the EDJ surface (Fig. 1).

For the specimens detailed by SR-μCT, the scale-free 2D relative enamel thickness (RET; Martin, 1985) was assessed on sections perpendicular to the cervical plane passing through the metaconid-protoconid dentine horns (Fig. 1A). The scale-free 3D relative enamel thickness (3D RET) was computed following Olejniczak et al., 2008a, Olejniczak et al., 2008c and Olejniczak et al., 2008b; (see also, Kono, 2004 and Macchiarelli et al., 2006). For the LM3 of IGF4351, the worn enamel apex and dentine horn of the protoconid and the enamel apex of the metaconid were virtually reconstructed for 2D RET using as a reference model the outline of the nearly unworn LM3 FS1996#Fi97 (Fig. 1A), thus allowing a more realistic assessment of its RET value (cf. Zanolli et al., 2010). For this latter specimen, we did not calculate 3D RET.

Linear measurements were taken using the software package MPSAK v.2.9 (available in Dean and Wood, 2003). Intra- and inter-observer tests for accuracy of the 2–3D measurements were run by two observers; recorded differences were less than 4%, which is compatible with similar previous tests (e.g., Macchiarelli et al., 2009).

High-resolution silicone moulds of well-preserved enamel surfaces were made on the following specimens: the buccal aspects of the UC and the lingual aspect of the UP4, UM1, UM2 and UM3 of IGF4332; the buccal aspect of the LI2 FS#BC-23, of the LP4 FS#Q16-64 and of the LM2 FS#Q16-59 (Fig. 2). The LM2 of IGF4883 V and the two enamel fragments, FS#BC-nn, were also replicated in this way before proceeding further with sectioning.

The above specimens were replicated using the Coltène President putty and light body wash addition curing silicone impression system. Casts from moulds of these teeth were made in clear epoxy resin and then sputter coated with ∼ 18 nm of gold–palladium to create a reflective surface (Fig. 2). Oblique lighting was then used to illuminate surface perikymata.

The two enamel fragments, FS#BC-nn, were first embedded in clear epoxy resin (Fig. 3A and B). One initial axial cut was made in the longitudinal plane with a diamond-embedded wafering blade and low speed saw (Buehler, Isomet). One face of the two resulting blocks was then polished and fixed with epoxy resin to a 1 mm thick glass slide. A second parallel axial cut was then made to remove a thin section attached to the glass slide. This was then lapped plane-parallel by hand, polished to ∼ 100 μm thickness, washed, dehydrated in alcohol, cleared in xylene and mounted with a styrene-based (DPX) mounting medium for light microscopy.

The LM2 of IGF4883 V (Fig. 3C) was chosen to provide information about both cervical enamel and root dentine growth. A quadrant was cut from the mesiobuccal root (Fig. 3C) and a ground section prepared as described above. All sections were then examined in polarized transmitted light and photomontages constructed using 125–500 magnification factors.

Each of the Oreopithecus sections contained regions of daily enamel cross striations that could be measured. Daily average rates in the cuspal regions of the two enamel fragments, FS#BC-nn, were calculated as follows: cuspal enamel thickness was measured from the lateral aspect of the former dentine horn to the surface enamel; these distances were then divided equally into three regions: inner, middle and outer; measurements were made across six consecutive cross striations and then divided by five to give a daily average value. This procedure was repeated 22 times in the inner, 13 times in the middle, and 20 times in the outer regions of the two sections combined.

Lateral and cervical enamel daily rates were measured in the same fashion within, respectively, the approximate outer mid-third of the crown and the cervical region of IGF4883V.

The high-resolution virtual reconstruction of the Oreopithecus lower molar crowns (Fig. 1) reveals that each morphological feature expressed at the outer enamel surface (OES) is sharply expressed at the EDJ level (contra Olejniczak et al., 2004). Indeed, while the main cusps are externally well differentiated, the short precingulid accessory cusps (i.e., paraconid, centroconid, tuberculum sextum) and the complex crest network (i.e., the paracristid between paraconid and protoconid, the protocristid between protoconid and metaconid, the hypolophid between hypoconid and entoconid, the hypometacristid between metaconid and centroconid, the hypoprotocristid between protoconid and centroconid, and the entolophid between entoconid and hypoconulid) (Butler and Mills, 1959) are more markedly expressed at the EDJ. In fact, compared to the OES signal, sharper topographic information of taxonomic and phylogenetic value is commonly observed in extant and fossil hominids at the EDJ level (Corruccini, 1987, Macchiarelli et al., 2006, Macchiarelli et al., 2009, Olejniczak et al., 2007, Olejniczak et al., 2004, Skinner et al., 2008a, Skinner et al., 2008b, Smith et al., 2009, Smith et al., 2006 and Zanolli and Mazurier, 2013).

When comparing the OES and EDJ in Oreopithecus, Homo, Pan and Gorilla LM1s, the first clearly exhibits a set of unique structural features (Fig. 4). Notably, Oreopithecus shows proportionally higher and more slender dentine horns. In addition, instead of having a concave central basin as in extant hominids, it is high and even slightly convex or bulging due to the development of the oblique crest and of the centroconid, which also rise high at the EDJ.

As a whole, based on our 3D comparative record of extant and fossil hominoid lower molars, the degree of structural complexity observed at the EDJ in Oreopithecus is unique.

In the four Oreopithecus lower molars imaged by SR-μCT, the thickest cuspal enamel is uniformly found on the buccal aspects of the protoconid and metaconid (Table 1). Similar to the pattern reported for all extant and some fossil hominids (Olejniczak et al., 2008d, Smith et al., 2011 and Smith et al., 2012), the Oreopithecus specimens investigated here provide evidence for increasing maximal enamel thickness from the LM1, through the LM2, to the LM3. The lingual enamel thickness of the metaconid is nearly uniform, while there is more variation along the protoconid lingual wall.

Similar to Ouranopithecus macedoniensis (maximum enamel thickness of the LM1 protoconid: 2.04 mm, metaconid: 1.85 mm; Macchiarelli et al., 2009), the Oreopithecus LM1 displays slightly thicker maximum cuspal enamel on the protoconid than on the metaconid. In contrast, the LM2 and LM3 exhibit a more even distribution of the enamel on both cusps (Table 1). This pattern differs from that observed in extant African great apes and Homo, where thicker enamel is generally found on the buccal crown side (Grine, 2005, Kono, 2004 and Macchiarelli et al., 2009). In the buccolingual section passing through the mesial cusps, Oreopithecus maximal radial enamel thickness (1.45 mm) slightly exceeds the measurements available for Pan (1.36 mm) and Gorilla (1.40 mm), but not the modern human figures (1.89 mm) (in Macchiarelli et al., 2009).

In order to adjust for size, we calculated both 2D and 3D indices of relative enamel thickness (RET and 3D RET) (Table 2 and Table 3).

Previous measures of 2D RET in Oreopithecus based on a UM1 physical section (spec. BM-11565) were 13.0 (Smith et al., 2003) and 15.5 (Andrews and Martin, 1991). Based on our results, average 2D RET in Oreopithecus lower molars is 15.8, while average 3D RET is 15.6. These estimates near or slightly exceed those reported for members of the Dryopithecinae subfamily, such as Pierolapithecus catalaunicus, Anoiapithecus brevirostris, Dryopithecus fontani, Hispanopithecus laietanus (details in Table 2). Conversely, much thicker values characterize the macrodont taxa Paranthropus robustus, Ouranopithecus turkae and O. macedoniensis (Table 2 and Table 3). Fossil and extant hominins also tend to display relatively thick enamel (except Neanderthals; Macchiarelli et al., 2007 and Olejniczak et al., 2008a). Fossil and extant pongines range from intermediate-thin, in Pongo abelii, to thick-enamelled molars, in Gigantopithecus blacki, while the African extant great apes Gorilla and Pan range from thin to intermediate-thin enamel (Table 2 and Table 3).

RET values in Oreopithecus appear to show an increase in 2D enamel thickness from M3 to M1, i.e., the reverse of the pattern commonly seen in the comparative taxa considered in this study, while the highest 3D RET value was found in the LM2. However, more data from Oreopithecus individuals with complete posterior tooth rows are needed to verify this as a taxon-specific characteristic.

Mean values for daily enamel cross striation spacing in inner, middle and outer cuspal regions of the two fragments of FS#BC-nn were 4.32 μm, 5.29 μm, and 5.19 μm, respectively (see descriptive statistics in Table 4). Thus, there is a slight gradient of increasing rates of daily enamel formation from inner to outer cuspal enamel. The grand mean of these three cuspal means is 4.93 μm, which was rounded to 5 μm. Thus, the 1100 μm and 1200 μm of cuspal enamel in the two sections was estimated respectively to have taken 1100/5 = 220 days and 1200/5 = 240 days to form.

Following this general survey of cross striation measurements, a zone of enamel 200 μm thick from the EDJ was defined from the cusp to the cervix in both sections of the enamel fragments (FS#BC-nn). While there was a gradient of increasing enamel formation rate from the EDJ to the outer enamel, there was no cuspal-cervical gradient within this inner zone close to the EDJ. The range of cross striation spacings did not change in this zone between the cusp and the cervix. The mean value of 26 measurements of cross striation spacing made within this zone between the cusp and cervix was 3.9 μm (Table 4). Thus, on average, any 200 μm thickness of enamel from the EDJ in these two Oreopithecus sections took 200/3.9 = 51 days to form (compared with mean values of 57 days in the Pan M2 sections and 62 days in the Gorilla M2 sections; see Dean, 1998). This calculation was then used to estimate total enamel formation times and enamel extension rates in these two sections following Dean, 2009 and Dean, 2010 and Guatelli-Steinberg et al. (2012).

Following methods described previously (Dean, 2009, Dean, 2010 and Guatelli-Steinberg et al., 2012), a measurement of 200 μm along a prism length (51 days formation time) was made from the EDJ tip in the cusp at the dentine horn. From this point an accentuated marking, or stria of Retzius, was identified and tracked obliquely back to intersect the EDJ further along from the dentine horn. The distance along the EDJ between the start-point at the dentine horn and the end-point further along the EDJ represents the length of EDJ formed in the same time it takes to form a 200 μm thickness of enamel (51 days). This represents the enamel extension rate that can be expressed in μm/day (see Section 4.5 below). This procedure was repeated from the cusp tip to the enamel cervix and each 51-day prism length cumulated to give the total enamel formation time. In both fragments there were 14 complete 51-day prism lengths so the total enamel formation time was 714 days in each fragment.

Crown formation time is equal to the sum of lateral enamel formation time and cuspal enamel formation time. The sum of cuspal enamel formation times (calculated in Section 4.3 above) and lateral enamel formation times in each of the enamel fragments (220 + 494 = 714 days) and (240 + 474 = 714 days) gives identical crown formation times of 1.96 years. However, the fragments are of unequal length along the EDJ from the cusp to the cervix. One fragment measures 7316 μm and the other 8230 μm.

Each section contained prominent accentuated lines within the enamel with an identical pattern in their relative spacing (a white and a black arrow in Fig. 3A and B denote two accentuated lines). This makes it highly likely that both fragments belonged to the same tooth and may well represent the buccal or lingual aspects of a lower molar (or less likely, a premolar). Given this, it is perhaps not surprising that the enamel formation time estimate for each is the same. However, the taller enamel fragment has the first of these shared accentuated lines 230 μm further away from the EDJ in the cusp than observed in the shorter enamel fragment, suggesting ∼ 46 days (i.e., 230/5 = 46 days) extra enamel formation in the former. It follows that this cusp is likely to have initiated earlier and to be staggered in its formation time with respect to the shorter fragment. Since enamel formation times are estimated to be equal in both fragments, the last-formed enamel may be in the cervix of the shorter fragment, ∼ 6–7 weeks after it completed in the taller fragment. If this is the case, the total time taken to form enamel in this tooth would have been 2.08 years.

In summary, crown formation time in these enamel fragments (FS#BC-nn) was close to 2 years. If total enamel formation time was 714 days in each enamel fragment, then this minus the cuspal formation times (220 and 240 days, respectively) leaves 494 and 474 days to form the lateral enamel of each enamel fragment. With either 5 or 6 days between striae of Retzius and perikymata in Oreopithecus, the expectation is that M2s would express between 80 and 100 perikymata between the cusp tip and cervix. This is explored further in Section 4.6 below.

A summary of crown (cusp-specific) enamel formation times in Oreopithecus compared with those of Pan appears in Table 5. The data for Oreopithecus are entered against each cusp for Pan where the range for almost all cusps on all molars would include our estimate for Oreopithecus of 1.96 years.

Enamel extension rates in the two Oreopithecus ground sections (FS#BC-nn) were calculated for each 51-day increment of time beginning with crown initiation in the cusp tip at the dentine horn. In sequence from the dentine horn, these were initially 32, 14, 15 and 10 μm per day in one cusp and 26, 26, 15, 13, and 10 μm per day in the slightly taller fragment. At the enamel cervix extension rates were 5.5 μm per day in one section and 6.1 μm per day in the other. It was also possible to estimate enamel extension rates in the last part of the cervix of the LM2 belonging to IGF4883 V. Here daily enamel formation rates were slightly slower than in the other two sections (average 3.6 μm/day), so that each 200 μm thickness of enamel took 55.5 days and not 51 days to form. Cervical extension rates were measured twice in succession, 5.1 and 5.3 μm/day, and correspond well to the cervical enamel extension rates in the two enamel fragments.

A summary of maximum extension rates calculated in the buccal cusp tips of molar teeth for several taxa appears in Table 6.

The buccal aspect of the LI2 FS#BC-23 (Fig. 2) was worn at the occlusal edge, but expressed in excess of 120 well-preserved perikymata from the cervix. The buccal aspect of the UC of IGF4332 is fractured at the cusp tip, but it expressed in excess of 140 perikymata that also are well-preserved from the cervix. On the same maxillary specimen, only incomplete counts on the lingual aspects of the UP4 (> 75), UM1 (> 75), and UM2 (> 85) were possible. Near-complete counts were possible on the buccal aspects of the well-preserved LP4 FS#Q16-64 (> 96), and LM2 FS#Q16-59 (> 75) (Fig. 2).

The two sections of the enamel fragments FS#BC-nn contained > 90 and > 80 long-period striae of Retzius, respectively. In each fragment at least 1 mm of enamel was worn away at the cusp tip that might have contained ∼ 10, or more, extra perikymata. Regions of the lateral enamel show locations where 5 daily increments are visible between long-period striae of Retzius. This is not inconsistent with the estimated number of striae within these sections (Fig. 5) taking 494 days and 474 days, respectively, to form and suggests that we are observing a degree of internal consistency within the enamel microstructure of these Oreopithecus enamel fragments.

The section of the LM2 of IGF4883 V cervical enamel, on the other hand, showed clear daily cross striations ∼ 3.6 μm apart on average but with regions slightly greater or smaller than this. Striae of Retzius were spaced ∼ 18–20 μm apart and in places 6 increments were visible between adjacent striae (Fig. 5). Taking the evidence from both individuals studied here, a periodicity of 5 or 6 days between striae of Retzius seems likely in Oreopithecus. The number of striae counted in the two enamel fragments is comparable with the number of perikymata counted in the other Oreopithecus M2s (and P4s) reported above and exceeds the number observed in the UM1 of IGF4332. They support the diagnosis of each fragment belonging to an M2 rather than an M1 and their morphology, with a thick occlusal portion, favours them being M2 fragments rather than P4 fragments.

Fig. 6 shows increase in tooth length along the EDJ continuing along the CDJ (cemento-dentine junction in the root) against tooth formation time. The data for each Oreopithecus enamel fragment (FS#BC-nn) are treated separately. The data for the M2 root growth of IGF4883 V have been added to that of the two crown fragments to produce a combined profile of crown and root growth for each fragment. When the data for Oreopithecus are plotted with the Pan M1 sample (Fig. 6A), they fall just beyond those teeth that achieve the greatest heights in the shortest times; however, when they are plotted with the Pan M2 sample (Fig. 6B), they lie well beyond the range observed for Pan.

Our 3D virtual investigation revealed that the high and complex occlusal topography of the Oreopithecus molars is not only found externally (Butler and Mills, 1959), but is also reflected internally, where the dentine horns are particularly elevated and acute, linked by sharp ridges and accessory features (Zanolli et al., 2010). This unique morphology differs from the lower, more squat cusps and dentine horns enclosing a sub-concave occlusal basin commonly seen in fossil and extant hominoid molars, such as Ouranopithecus (Macchiarelli et al., 2009), Paranthropus and Australopithecus (Skinner et al., 2008a), Homo (Macchiarelli et al., 2006, Skinner et al., 2008b, Zanolli, 2014 and Zanolli and Mazurier, 2013), and in the living great apes (Skinner et al., 2010 and Skinner et al., 2008b).

Assuming that Oreopithecus is closely related to the dryopithecines (Harrison and Rook, 1997 and Moyà-Solà and Köhler, 1997), which almost invariably exhibit a low to only moderately elevated external crown topography (Alba et al., 2013 and Begun, 2002), its tall occlusal reliefs could represent an autapomorphic feature developed under conditions of insular isolation. On the other hand, tall cusps could also represent the retention of a primitive feature, as seen in basal hominoids (Begun, 2002). Ongoing systematic characterization of the inner structural morphology of the dryopithecine molar crowns promises to shed light on this relevant aspect (Fortuny et al., 2014).

Oreopithecus displays thicker molar enamel than Pan and Gorilla (for both RET and 3D RET), but slightly thinner than extant humans. Compared to other fossil hominoids, on average, Oreopithecus shows thicker enamel than the dryopithecines reported so far (Alba et al., 2013). Interestingly, megadont fossil hominoids, such as Ouranopithecus, Paranthropus, Australopithecus and Gigantopithecus, show the relatively and absolutely thickest enamel, suggesting an allometric trend.

Variations in tooth enamel thickness result from a complex interplay between ecology and phylogeny (Horvath et al., 2014, Pampush et al., 2013 and Smith et al., 2012). Traditionally used to infer durophagy in fossil primate taxa, it is now considered as intimately related to dietary abrasiveness and selectively responsive to lifetime dental wear resistance (Pampush et al., 2013). Because of its elevated cusps and sub-bilophodont molar pattern, Oreopithecus has been commonly regarded as a specialized leaf eater (e.g., Begun, 2007, Kay and Ungar, 1997 and Szalay and Delson, 1979; but see Moyà-Solà and Köhler, 1997). However, stable isotopic analyses (δ ¹⁸O, δ ¹³C) suggest that, compared to modern chimpanzees or Sivapithecus, Oreopithecus exploited a wider range of habitats and food items (Nelson and Rook, 2008 and Rook and Nelson, 2009). This is supported by microwear analyses that suggest a closer fit with Papio ursinus (DeMiguel et al., 2014, L’Engle Williams, 2013, Smith and Williams, 2010 and Ungar, 1996). Accordingly, similarly to the variation in molar enamel thickness seen in macaques (Kato et al., 2014), the intermediate-thick (Martin, 1985) enamel of Oreopithecus could reflect wider dietary diversity implying the interaction with a greater range of food material properties than previously thought.

High extension rates in the first-formed cuspal enamel play a large role in determining final cusp height and tooth height (Dean, 2009 and Guatelli-Steinberg et al., 2012). Thereafter, a combination of extension rates and time spent forming enamel contribute to final crown height. Tall cusps and overall crown height (relatively higher in Oreopithecus than in Pan and Gorilla) are achieved with fast initial cuspal extension rates that are maintained into crown formation, as they appear also to be in the Miocene cercopithecoid Victoriapithecus macinnesi (Fig. 7; Dean and Leakey, 2004). The dryopithecine Hispanopithecus laietanus (Alba et al., 2012, Dean and Kelley, 2012 and Kelley et al., 2001), which is more similar to extant and fossil hominins and pongines than to Oreopithecus in molar occlusal morphology, has a slower initial crown growth trajectory but continues growing in height for a long period of time.

Table 6 demonstrates how variable the initial maximum cuspal extension rates are in samples of extant taxa. It also evidences how maximum cuspal extension rates observed in Oreopithecus fall within the gorilla, chimpanzee and modern human ranges, but along with those of Victoriapithecus macinnesi tend towards the higher values for M2s than fossil hominids with lower more squat crown morphologies.

Oreopithecus cuspal enamel overall grew at a fast daily rate (Table 7) and the three molars from Fiume Santo analysed here approach the relative enamel thickness values reported for Afropithecus turkanensis, Sivapithecus parvada and Australopithecus africanus (Mahoney et al., 2007, Olejniczak et al., 2008c and Smith et al., 2003). The overall faster daily rates, especially in the inner cuspal enamel, may be part of the process of achieving this degree of thickness in a given crown formation time.

In the Old World monkey Victoriapithecus macinnesi (Dean and Leakey, 2004), there was a near constant secretion rate of 6 μm/day in the M2 cuspal enamel, which is consistently higher than in Miocene apes or thick-enamelled hominins. It is possible this is the primitive condition we might expect for a stem hominoid, but this is not the case in Oreopithecus, where a slight gradient exists from ∼ 4 to 5 μm/day (Table 7). In Proconsul nyanzae, the cuspal enamel gradient ranged from ∼ 4 to 6 μm/day, with inner rates staying slower for a longer time (Beynon et al., 1998). Thus, the gradient in Proconsul nyanzae was more pronounced than in Oreopithecus.

Based on our evidence, it seems there is no simple relationship between the thickness of cuspal enamel and the gradients of daily formation rates within them (see Table 7).

Very few of the 75 chimpanzees studied by Smith et al. (2007) had periodicities as low as 5 ∼ days. The 5-day periodicity in the two of the three Oreopithecus enamel fragments resembles, therefore, the modal value in macaque monkeys, and that occasionally seen in gibbons (Beynon et al., 1998 and Dirks, 1998). The 5- and 6-day periodicities in the two Oreopithecus individuals reported here may turn out to be close to the modal periodicity in Oreopithecus, in which case it might best be explained with respect to what we know of its body size, where males were perhaps the size of a female chimpanzee, or slightly smaller, and females the size of a male baboon (Jungers, 1987). However, the nature and strength of the relationship between long-period stria periodicity and body size remain unclear.

The comparison with the data available for Pan (Table 5) would suggest the Oreopithecus fragments FS#BC-nn have greater formation times than Pan M1 metaconids. But the small sample of M1 or M2 protoconid formation times in Pan both encompasses the time to form these fragments. Even M3 metaconid formation times in Pan encompass 2 years, and M3 protoconid formation times are only a little more than 2 years at the low end of their range (Table 5). One fragment of FS#BC-nn has the typical concave EDJ of lower buccal or upper palatal cusp EDJs but the other does not (Fig. 3A and 3B). It is possible that the taller later initiating enamel fragment is a lower lingual or upper buccal fragment from the contralateral side of the shorter fragment. If this is the case then, unlike Pan M1s but similar to Pan M2s, both aspects of this Oreopithecus tooth would have had similar total enamel formation times.

The evidence based on the shared internal pattern of accentuated markings, the thick-enamelled occlusal projection in one fragment, as well as identical enamel formation times and the surface perikymata counts within the Oreopithecus sample itself, support these fragments very likely coming from the same M2 crown (but not completely excluding a P4 attribution due to its similar formation times with the M2).

In Oreopithecus, the crown formation times of the two possible M2s fragments were ∼ 1.96 years, close to those reported for Proconsul nyanzae (∼ 2 years; Beynon et al., 1998).

Values for most of the variables described here for Oreopithecus fit well within what is known for Pan. Those at the upper limits of variation known for Pan (the high M2 cuspal extension rates in the crown) can be considered as part of an adaptation to grow a tall crown. The faster rates of enamel formation are harder to interpret, but might simply be an adaptation to grow thick enamel in a short period of time. By way of example, faster rates of enamel formation in Paranthropus boisei than in Pan troglodytes do not appear to be linked with any obvious differences in their crown formation times (Lacruz et al., 2008), but they do contribute to crowns with thick enamel being able to complete their formation within a similar time to those of Pan. Thick enamel in Ouranopithecus macedoniensis (Macchiarelli et al., 2009 and Smith et al., 2004) appears to have formed with a minimal gradient of enamel formation from the EDJ to the surface enamel (Table 7), and so no clear link between enamel gradients and enamel thickness appear to exist among hominoids. These new data for Oreopithecus serve to emphasise that thick enamel may form in many ways.

A larger sample of Oreopithecus molars would demonstrate if faster enamel formation rates are actually associated with shorter than average crown formation times than in Pan. The non-destructive analytical method described here (See Section 4.3) would facilitate studies of larger samples of individuals than in this study and might resolve the question of whether the mean total crown formation times for a sample of Oreopithecus molars were actually less than in Pan. It is not presently possible to know, however, whether a single Oreopithecus specimen is representative of the mean of this taxon. The crown formation time for the individual used in this study (∼ 2 years) falls within the ranges known for Pan M1 and M2 crown formation. However, in addition to individual crown formation times, the initiation times of tooth mineralisation and the degree to which M1, M2 and M3 crown formation periods overlap with one another, which is quite variable in Pan and Pongo (Smith et al., 2007 and Winkler et al., 1991), together best reflect the overall duration of dental development (Dean, 2010). Previously, early molar initiation times were shown to play a key role in the accelerated dental development of Anapithecus hernyaki (Nargolwalla et al., 2005). Information made available from this study concerning the timing of individual molar tooth formation in Oreopithecus would be more informative if it were possible to determine both the degree of overlap in molar formation and the initiation times of individual molar crowns.