Journal of Human Evolution 63 (2012) 475e486
Contents lists available at SciVerse ScienceDirect
Journal of Human Evolution journal homepage: www.elsevier.com/locate/jhevol
Enamel extension rate patterns in modern human teeth: Two approaches designed to establish an integrated comparative context for fossil primates Debbie Guatelli-Steinberg a, *, Bruce A. Floyd b, M. Christopher Dean c, Donald J. Reid d a
Department of Anthropology, The Ohio State University, 174 West 18th Ave., Columbus, OH 43210, USA Department of Anthropology, University of Auckland, Private Bag 92019, Auckland, New Zealand c Department of Cell and Developmental Biology, University College London, Gower Street, London WC1E 6BT, United Kingdom d Department of Oral Biology, Newcastle University, Newcastle upon Tyne NE2 4BW, United Kingdom b
a r t i c l e i n f o
a b s t r a c t
Article history: Received 16 September 2011 Accepted 1 May 2012 Available online 28 June 2012
Enamel extension rates (EERs), the rates at which ameloblasts differentiate, determine how fast tooth crowns grow in height. Studies of fossil primate (including hominin) enamel microstructure usually focus on species differences in enamel formation time, but they have also begun to address species-level variation in enamel extension rates. To improve our ability to compare EERs among primate species, a better understanding how EERs vary within species is necessary. Using a large and diverse modern human histological sample, we find that initial EERs and patterns of EER change along the enameldentine junction (EDJ) vary in relation to EDJ length. We also find that enamel formation time varies in relation to EDJ length, but that it does so independently of initial EERs. These results suggest that EDJ length variation within a species sample can affect interspecific comparisons not only of EERs but also of enamel formation times. Additionally, these results lend within-species support to the hypothesis, based on comparisons among hominin species, that EERs and crown formation times can vary independently (Dean, 2009). In a second approach, we analyzed EER changes specifically in the lateral enamel of two modern human population samples as these changes relate to the distribution of perikymata. As surface manifestations of internal enamel growth increments, perikymata provide a valuable source of information about enamel growth in fossils. We find that EER declines in the lateral enamel are associated with an increase in perikymata density from first to last-formed lateral enamel. Moreover, variation in the extent of EER decline among individuals is associated with variation in the distribution of perikymata along their enamel surfaces. These latter findings suggest that the distribution of perikymata on the enamel surface provides information about rates of EER decline in lateral enamel, at least in modern humans. Ó 2012 Elsevier Ltd. All rights reserved.
Keywords: Dental development Enamel Extension rate Hominin Teeth
Research on fossil primate, including hominin, enamel microstructure to date has primarily focused on estimating (or calculating) the duration of enamel formation, insofar as it can be used to reconstruct the pace of fossil species’ life histories (e.g., Bromage and Dean, 1985; Beynon and Wood, 1987; Dean, 1987, 2006, 2010; Dean et al., 2001; Ramirez-Rozzi and Bermudez de Castro, 2004; Guatelli-Steinberg et al., 2005; Macchiarelli et al., 2006; Smith et al., 2007a, 2007b, 2010). Related to the question of species variation in enamel formation time is the question of species variation in rates of enamel extension, that is, the rates at which
* Corresponding author. E-mail addresses:
[email protected] (D. Guatelli-Steinberg), b.floyd@ auckland.ac.nz (B.A. Floyd),
[email protected] (M.C. Dean), donald.reid14@ gmail.com (D.J. Reid). 0047-2484/$ e see front matter Ó 2012 Elsevier Ltd. All rights reserved. doi:10.1016/j.jhevol.2012.05.006
ameloblasts (enamel forming cells) differentiate along the presumptive enamel-dentine junction or EDJ (Shellis, 1984). Rates of enamel extension determine how fast tooth crowns grow in height, as they extend the length of the EDJ from cusp to cervix (Dean, 2009). Dean (2009) has noted that fewer studies of fossil primates have focused on enamel extension rates, even though these rates appear to differ among primate species, both extinct and extant. Rapid rates of enamel extension relative to modern humans might be expected for fossil hominin species with particularly short crown formation times, such as Paranthropus robustus. Yet, Dean (2009) showed that for three fossil hominin specimens, one each of Paranthropus robustus, Homo erectus, and Homo neanderthalensis, enamel extension rates (EERs) measured at regular intervals along the EDJ fell within modern human ranges. Therefore, for these hominin species, differences in overall enamel formation time did
476
D. Guatelli-Steinberg et al. / Journal of Human Evolution 63 (2012) 475e486
not appear to be produced by differences in enamel extension rates. Both fossil hominins and modern humans exhibit what appears to be an exponential decline in EERs from cusp to cervix, such that much of a tooth’s crown height is formed relatively quickly (Dean, 2009). On the other hand, Smith et al. (2010) presented data suggesting that faster average enamel extension rates as compared to modern humans underlie shorter enamel formation times in most Neandertal tooth types. Thus, the relationship between enamel extension rates and enamel formation time across fossil hominin species, and across fossil primates generally, is currently unresolved. To more clearly evaluate inter-specific differences in patterns of EER change along the EDJ it is important to determine how these patterns vary within a species. Hence, here we focus on EER variation within modern humans for whom large histological samples are available. Beginning with an expanded version of Dean’s (2009) modern human data set, we first ask whether patterns of EER change from cusp to cervix are related to EDJ lengths, as Shellis (1984) contended. Comparing EER graphs of molar and anterior teeth as well as EER graphs of buccal and lingual cusps of the same tooth type, Shellis (1984) suggested that initial rates were higher and the decline in extension rates more gradual, when EDJ lengths were longer. Shellis’s (1984) results, however, did not clearly demonstrate that EDJ lengths were associated with faster rates of enamel extension independently of tooth type. In addition, his study did not assess the potential contribution of extended crown formation time to EDJ length. Here, we investigate the degree to which EERs, independent of tooth type, as well as enamel formation time, are related to the total length of the EDJ within modern human teeth. In doing so, we aim not only to determine if patterns of EER change in modern humans along the EDJ are related to EDJ length but also to clarify relationships among EDJ lengths, enamel extension rates, and enamel formation times. As detailed in our discussion, our analysis provides a context for further evaluating the relationship between enamel formation time and EERs across fossil primate species and allows us to assess the importance of EDJ length as a source of within-species variation in both. Also in this study, using a new data set collected for this purpose, we investigate how EERs change along the lateral enamel of modern human tooth crowns and how they relate to the distribution of perikymata on the enamel surface. Lateral enamel, enamel on the side of teeth, is covered by perikymata, which are surface manifestations of internal enamel growth increments known as striae of Retzius. As such, perikymata provide an important source of information about enamel growth in fossil hominins, whose teeth often cannot be sectioned. In anterior teeth, lateral enamel formation time makes up the majority of enamel formation time (Aiello and Dean, 1990; Hillson, 1996), even though lateral enamel itself may correspond to only about half of EDJ length in some teeth (Dean, 2009). It is well known that perikymata become compressed as they approach the cervix in modern human teeth, generally to a greater degree than they do in the teeth of several hominin species or groups, including those of P. robustus (Dean and Reid, 2001), H. erectus (Dean and Reid, 2001), H. neanderthalensis (RamirezRozzi and Bermudez de Castro, 2004; Guatelli-Steinberg et al., 2007; Guatelli-Steinberg and Reid, 2008; Reid et al., 2008) and the Qafzeh hominins (Guatelli-Steinberg and Reid, 2010). Indeed, the distribution of perikymata on Neandertal teeth differs from that of diverse modern human populations whose perikymata distributions show far more similarity to one another (Guatelli-Steinberg et al., 2007, 2008; Reid et al., 2008). Modern humans have a greater percentage of their total perikymata packed into the cervical regions of their crowns than do those of Neandertals or H. erectus (Dean et al., 2001), on which perikymata tend to be more uniformly distributed.
Although these differences in perikymata distribution between modern humans and various fossil hominin samples have been documented, the enamel growth processes which underlie them remain obscure. It has been suggested that the compression of perikymata toward the cervix in modern human teeth might reflect a slowing of EERs along the EDJ (Ramirez-Rozzi and Bermudez de Castro, 2004). Boyde (1964) first noted that the angulation and relative length of incremental lines to the EDJ reflects the rate at which ameloblasts cover the crown surface. Developing this idea, Shellis (1984) proposed a method for quantifying the extension rate using the inclination of the prism direction to the EDJ together with the angle subtended by incremental lines to the EDJ. These incremental lines represent the former position of the ameloblast sheet during enamel formation. Shellis (1984) used measurements of daily cross striations within 30 mm of the EDJ to put a time scale in days to the rate at which ameloblasts differentiated along the EDJ. He then made successive measurements along the EDJ, from the cusp to the cervix, using 30 mm lengths along prism paths. EERs are therefore clearly related to both the angle that a stria makes with the EDJ and to the orientation of prisms to the EDJ that reflects the direction of ameloblast movement (Shellis, 1984). While the interrelated geometry of these angles and the daily rate of enamel formation at the EDJ would be expected to affect the spacing of perikymata, these values may not actually be preserved at the outer enamel surface because they can change within the thickness of the forming enamel (Hillson and Bond, 1997; Dean and Shellis, 1998). Enamel growth occurs not only “downward” through extension along the EDJ, but also “outward” from the EDJ, as ameloblasts secrete the full thickness of the enamel. Fast daily secretion rates in outer enamel can drive prisms, and with them stria contours, inferiorly, changing the angle at which they crop-out as perikymata at the outer enamel surface (Dean and Shellis, 1998). This phenomenon, together with the surface contour of a tooth (Hillson and Bond, 1997), has the potential to dissociate EERs from perikymata spacing. Given uncertainty about the role of extension rates in producing the compression of perikymata in modern humans, here we address the following three questions: “Do EERs in modern humans decline from first-formed to last-formed lateral enamel? Does the pattern of EER change along the EDJ in lateral enamel depend on tooth type or population? If EERs do decline, do individuals with greater declines also exhibit greater increases in perikymata numbers from first-formed to last-formed lateral enamel?” The answers to these questions, although they do not reveal the causes of perikymata distribution differences between fossil hominins and modern humans, illuminate the causes of variation in perikymata distribution within modern humans and thus have potentially important implications for interpretations of fossil material. Materials and methods Samples and measurements Two data series are used here. The first consists of EERs of modern humans calculated (by M.C. Dean) from cusp to cervix from 15 anterior tooth and 28 molar crown ground sections. The sample of ground sections is broken down by tooth type and population of origin (where known) in Table 1. For each tooth type, each section represents a different individual. The procedure for calculating enamel extension rates, detailed in Dean (2009, 2012), begins with the choice of sections with clearly visible cross striations close to the EDJ. Cross striations are counted along the direction of an enamel prism over a distance of 200 mm in at least three regions of the crown: mid-occlusal, mid-lateral, and mid-cervical, approximating the mid-point of each third of the total length of the EDJ. Across a large sample of modern human teeth of all tooth types and
D. Guatelli-Steinberg et al. / Journal of Human Evolution 63 (2012) 475e486 Table 1 First data set of ground sections of 15 anterior teeth and 28 molar crowns. Tooth type
N
Groups
Number of 70-day enamel growth units per tooth
Maxillary central incisor
3
21 to 26
Maxillary canine Maxillary 1st molar Maxillary 2nd molar Maxillary 3rd molar Mandibular central incisor Mandibular lateral incisor Mandibular canine
5 5 3 2 4 1 2
Chinese & Southern African & UK origina Southern African Southern African Southern African Southern African Southern African UK origin Chinese and UK origin Southern African Southern African Southern African and UK origin
Mandibular 1st molar Mandibular 2nd molar Mandibular 3rd molar
10 3 5
16 10 13 12 12
to 21 to 13 to 15 to 14 to 18 24 26 to 27 12 to 15 12 to 16 8 to 16
a UK origin refers to teeth from Spitalfields Cemetery or from present-day Newcastle-upon-Tyne.
of diverse geographic origins, the continuous cumulative cross striation count across a 200 mm thickness of enamel from the EDJ varies between 68 and 92 days (Dean, 2009, 2012). The greater times in this range are, however, due not to slower enamel secretion rates, but to greater degrees of enamel decussation near gnarled cuspal enamel. In the teeth chosen for this study, the average daily rate of enamel secretion within 200 mm of the EDJ was 2.85 mm per day. No correction was made for prism decussation since all prism lengths used (beyond the first prism length in each tooth) were cervical to the heavily decussating occlusal enamel. Thus, for teeth included in this study, the average time taken to form each 200 mm unit was 70 days, with minimal variation (0.443 days (SD) in 428 of 430 enamel units). However, some tooth types, especially M3s and M2s do show a slowing in the cervical region. In these cases daily rates or the percentage reduction in stria counts within each 200 mm segment were used to calculate the thickness of enamel formed in 70 days. With these exceptions, a 200 mm prism length from the EDJ was used in all calculations.
Figure 1. Method by which enamel extension rates were calculated (by M.C. Dean) for first data set, illustrated in a high power image of lateral enamel at the EDJ in a human lower lateral incisor. A 200 mm line is drawn from the EDJ along the direction of the enamel prisms (line AC). In this tooth, the average time taken to form a 200 mm thickness of enamel from the EDJ (A to C) was w70 days. At point C, a second line is drawn, parallel to (or along) a stria of Retzius (or accentuated line), back to a lower position on the EDJ (point B). Striae of Retzius mark out the enamel forming front at any given time. Thus, the distance along the EDJ between points A and B was also formed in w 70 days. The extension rate, the rate at which a cohort of new ameloblasts differentiates and covers this distance (A to B), is calculated by dividing this distance by 70 days.
477
This method is illustrated in Figs. 1 and 2. Fig. 1 is a high power image of lateral enamel at the EDJ in a human lower lateral incisor. The extension rate is the time taken for a cohort of new ameloblasts to differentiate and cover a length along the EDJ between point A and point B. This is the same time that it takes an ameloblast to secrete enamel between point A and point C. In this tooth the average time taken to form a 200 mm thickness of enamel from the EDJ (A to C) was w70 days. Fig. 2 is a low power image of the ground section of the same lower lateral incisor shown in Fig. 1. The central image is a higher power image of the cuspal and upper portion of the lateral enamel of another incisor from the same individual. Within the enamel, the orientation of both accentuated striae and regular striae of Retzius can be seen, although it is not possible to distinguish one type of stria from another unless they are associated with perikymata troughs at the surface of the lateral enamel. (Unlike regular striae of Retzius, accentuated striae are more pronounced, e.g. either thicker or darker, do not form at the regular time intervals, and are not necessarily associated with perikymata troughs). Nonetheless, both kinds of striae represent former positions of the mineralizing enamel surface. Their orientation or angulation to the EDJ reflects the rate of extension of new ameloblasts along the crown during tooth development. From a start point at the dentine horn (equivalent to point A in Fig. 1) a 200 mm line was drawn along the direction of the prism path (equivalent to point C in Fig. 1). From this position, a line drawn parallel to or directly along a coincident stria of Retzius or accentuated line was traced back to an end point at the EDJ (equivalent to point B in Fig. 1). In the central image of Fig. 2 this procedure has been repeated four times using the end point at the EDJ as the next start point in the sequence. Extension rates are high here. Only the cuspal portion of this tooth is shown but, by way of example, the same procedure was repeated 25 times between the dentine horn and the enamel cervix in this tooth (25 70 days ¼ 1750 days or 4.8 years enamel formation time). The image on the right in Fig. 2 is a higher power image of the cervical third of the enamel crown of SK 63 (a lower canine attributed to P. robustus from Swartkrans South Africa). This tooth provides a clear example of the methods used in this study and of how EERs reduce towards the cervix. Again, the orientation of the striae to the EDJ is clear but when working close to the EDJ, exactly which are regular striae of Retzius close to the EDJ and which are accentuated striae is not so clear. In this image, only the last nine sets of measurements along the EDJ (from 11 to 19) are shown, but they illustrate a slowing of the EER, which then increases again in the last three increments measured. The average daily rate of enamel formation within a 200 mm distance from the EDJ in SK 63 is 3.2 mm/day and so it takes on average 62 days to form this thickness of enamel (yielding 19 62 days ¼ 1178 days or 3.23 years enamel formation time from dentine horn to enamel cervix). This same procedure as outlined above was repeated along the EDJ of each of the 15 anterior teeth and 28 molar teeth. Enamel extension rates were calculated for each 70 day interval, or “unit” of growth. As illustrated above, teeth can take different lengths of time to form, the number of units over which enamel extension rates were calculated can differ by tooth (Table 1). While it was possible to calculate extension rates in the most cervical portions of the EDJ for smaller time intervals of enamel formation, the final unit used in this study was the last whole 70 day increment. A second data set was used to address questions relating EERs to the distribution of perikymata (Pk) over the enamel surface of anterior teeth. Data were gathered from ground sections of at least five teeth per anterior tooth type (UI1, UI2, UC, LI1, LI2, and LC) for each population group, one from the United Kingdom (Newcastleupon-Tyne) and the other from southern Africa. This sample is
478
D. Guatelli-Steinberg et al. / Journal of Human Evolution 63 (2012) 475e486
Figure 2. Additional detail about the methods used to calculate enamel extension rates in the first data set. At left is a low power image of the ground section of the same lower lateral incisor shown in Fig. 1. The central image is a higher power image of the cuspal and upper portion of the lateral enamel of another incisor from the same individual, showing how extension rates were calculated beginning at the dentine horn. The image on the right is a higher power image of the cervical third of the enamel crown of SK 63 (attributed to Paranthropus robustus from Swartkrans South Africa), showing how extension rates primarily decrease during this phase of enamel formation, but then start to increase again during the last three measurement intervals. See text for a more complete discussion.
broken down by tooth type and population in Table 2. Again, for each tooth type, each section represents a different individual. In this data set, crown heights were divided into deciles in which Pk were counted. Then, average extension rates were calculated (by D.J. Reid) over the area of the EDJ corresponding to each decile. The purpose of calculating extension rates in this manner was to be able to analyze changes in extension rates along the EDJ in direct relation to changes in the distribution of Pk along the enamel surface. Thus, unlike data set one, extension rates in data set two were calculated not over equivalent time intervals, but over equivalent divisions of lateral enamel. First, the periodicity of striae of Retzius was determined for each enamel section. Periodicities, while ranging from 6 to 12 days across modern human individuals (Reid and Dean, 2006), are constant in all of the permanent teeth of a given person (FitzGerald, 1998). Pk within Table 2 Ground sections per tooth type and population group from 2nd data set. Population group
N
Tooth type
Range of periodicities (days)
UK origina Southern African UK origin Southern African UK origin Southern African UK origin Southern African UK origin Southern African UK origin Southern African
5 7 7 5 10 9 7 5 5 5 5 5
Maxillary central incisor Maxillary central incisor Maxillary lateral incisor Maxillary lateral incisor Maxillary canine Maxillary canine Mandibular central incisor Mandibular central incisor Mandibular lateral incisor Mandibular lateral incisor Mandibular canine Mandibular canine
8e10 8e10 8e11 10e11 7e11 7e11 7e8 8e10 8e9 9e10 7e10 7e11
a
UK origin refers to teeth from present-day Newcastle-upon-Tyne.
each decile were traced back along striae of Retzius to the EDJ. Thus, the number of Pk and striae of Retzius per decile are identical. The EDJ was measured between the striae of Retzius giving rise to the first and last Pk of each decile (Fig. 3). Average enamel extension rates corresponding to each decile were calculated by dividing the EDJ length by the time taken to form this length, which is the product of striae (or Pk) number and periodicity for that individual. Thus, in this second data set, for each tooth type there were 10 regions of the EDJ length over which average extension rates were calculated, each corresponding to a decile of crown height in which Pk were counted. As shown in Fig. 3, the average EER for a decile of crown height is directly proportional to EDJ length and inversely proportional to the product of periodicity (PD) and the number of Pk within the decile. Thus, there is an inherent relationship between EERs and the number of Pk per decile, but, across individuals, EDJ length and periodicity also affect the average enamel extension rate corresponding to a decile. For a single tooth, because periodicity is constant, average EERs per decile depend only on EDJ length and Pk number. Therefore, across the deciles of a single tooth, as Pk numbers increase from cusp to cervix by some factor, EERs will decrease only if EDJ lengths do not increase by this same, or a greater, factor. EERs will remain the same if EDJ lengths increase by the same factor. EERs will increase if EDJ lengths increase by a greater factor. Data analysis To analyze patterns of EER change statistically, univariate multiple regression analysis and repeated measures analyses were implemented in SYSTAT v10. Repeated measures analyses are an
D. Guatelli-Steinberg et al. / Journal of Human Evolution 63 (2012) 475e486
479
Figure 3. Method by which average extension rates were calculated (by D.J. Reid) in second data set. At left is a low power image of an upper central incisor. The horizontal lines divide the lateral enamel into deciles of crown height. The thick dark line in the enamel represents the boundary between cuspal and lateral enamel. The higher power image at right from another tooth shows particularly clear striae of Retzius associated with perikymata troughs at the tooth surface. Perikymata which formed the borders of each decile (marked by white triangles) were traced back along associated striae of Retzius to the EDJ. To obtain the average EER, the length of the EDJ corresponding to the decile was then divided by the product of the number of perikymata within the decile and the perikymata periodicity of the individual.
appropriate method for evaluating trends in serial measurements of EDJ length or serial values of extension rates within ground sections. They permit an analysis of variance of differences across equivalent 70-day units of growth within ground sections (first data set) or across equivalent deciles of lateral enamel (second data set) as they are influenced by predictors such as tooth type, population group, or periodicity. Orthogonal polynomial contrasts
Table 3 Multivariate repeated measures analysis of change in per unit EDJ length across the first 10 units of the first data set. “Unit” refers to the 70 day units of enamel growth over which EDJ lengths were measured.
permit independent linear, quadratic, cubic or k higher order tests of trends within each ground section (k being equal to one less than the number of repeated measures in a model) (Wilkinson et al., 1994; Sokal and Rohlf, 1995). For example, a linear test evaluates whether, within ground sections, serial values of EDJ lengths or extension rates fall along a least squares regression line with a significant non-zero slope. The 2nd and 3rd order polynomials evaluate whether serial values fall along quadratic or cubic curves. Overall trends in EDJ lengths or in enamel extension rates within ground sections, as well as interactions between serial values and one or more predictors, can be evaluated in this way. The same repeated measures analyses also
Within subjects Source
SS
df
MS
F
P
H-F
Unit 58962400.00 9 6551380.34 277.39 0.000 0.000 Unit by Tooth 2763210.24 36 76755.84 3.25 0.000 0.001 Unit by Jaw 65612.71 9 7290.30 0.31 0.972 0.790 Unit by Tooth by Jaw 1421655.98 36 39490.44 1.67 0.012 0.100 Error 6376965.65 270 23618.39 Huynh-Feldt Epsilon : 0.288 Single Degree of Freedom Polynomial Contrasts Test of Order 1 (Linear)
Source
SS
df
MS
Unit 32181500.00 1 32181500.00 Unit by Tooth 1142346.89 4 285586.72 Unit by Jaw 26969.15 1 26969.15 Unit by Tooth 528792.31 4 132198.08 by Jaw Error 1445808.82 30 48193.63 Significant Higher Order Polynomial Contrasts (P 0.05) 2 (Quadratic) Unit 15668100.00 1 15668100.00 2 (Quadratic) Unit by Tooth 532748.52 4 133187.13 3 (Cubic) Unit 6800463.83 1 6800463.83 3 (Cubic) Unit by Tooth 567401.56 4 141850.39 4 Unit 3179371.61 1 3179371.61 5 Unit 979969.27 1 979969.27 6 Unit 126584.94 1 126584.94 7 Unit 25817.27 1 25817.27
F
P
667.76 5.93 0.56 2.74
0.000 0.001 0.460 0.047
315.32 2.68 174.20 3.63 110.35 41.48 10.32 4.70
0.000 0.051 0.000 0.016 0.000 0.000 0.003 0.038
Figure 4. Adjusted least square means (SE) from repeated measures analysis over first 10 units of growth along the EDJ from the first data set. The primary difference is the reduced extent of change per-unit of enamel growth across units 2 through 5 of anterior teeth.
480
D. Guatelli-Steinberg et al. / Journal of Human Evolution 63 (2012) 475e486
other (r ¼ 0.029, p ¼ 1.00), we used a least squares regression model to evaluate the extent to which early EERs and total enamel formation time account for variation in total EDJ lengths when tooth type is statistically controlled. Using the second data set, we first investigated whether EERs in modern humans decline across first to last-formed deciles of lateral enamel, and if the presence or nature of these declines is associated with particular tooth types or population groups. Changes in EERs across deciles were assessed in a repeated measures analysis, statistically controlling for tooth type, population group, and the associated interaction terms. The second data set was then used to investigate whether individuals with greater declines in EERs across deciles also exhibit greater increases in perikymata numbers across deciles. Analyzing this relationship directly is statistically problematic because EERs per decile are mathematically related to the number of perikymata per decile. Instead, we used an indirect approach that analyzes the component parts of the enamel extension rate: EDJ length, Pk number, and periodicity. As noted earlier, within individual teeth, because periodicity is constant, if increases in EDJ lengths per decile do not keep pace with increases in Pk numbers per decile, then extension rates are declining. To examine variation in extension rate declines across individuals, we analyzed how periodicity affects changes in EDJ lengths in relation to changes in Pk numbers across deciles. To do so, EDJ lengths and Pk numbers were first transformed into Z-scores, and the difference between EDJ Z-score and Pk Z-score per decile was calculated. A repeated measures analysis was then performed, with per-decile differences in EDJ length Z-score and Pk number Z-score as the criterion variable and periodicity as the primary predictor, with tooth type and population group included as covariates. This approach hinges on the known relationship between periodicity and perikymata distribution. Reid and Ferrell (2006) showed that lower periodicities are associated with greater increases in Pk numbers across deciles. As a preliminary step we verified this relationship within our second data set (r ¼ 0.64; p 0.0005) and found that it was similar for both population samples. We therefore predicted that differences between EDJ Zscore and Pk Z-scores across deciles would be most pronounced for individuals with low periodicities, and hence in those who also exhibit the greatest increases in Pk number across deciles.
Figure 5. Mean EDJ lengths per unit (SE) for ten mandibular first molars grouped into those with short and long total EDJ lengths (the median total EDJ length was used to divide first molars into short vs. long-EDJ length categories).
permit assessment of mean differences in the total sum of serial values between groups; however, as these results were not the focus of this investigation, they are not reported. Assumptions of multivariate normality and equivalence of variance were checked using plots of residuals against predicted values for each model. The extent of violations of compound symmetry for models were evaluated by comparing initial p-values with more conservative HuynheFeldt (HeF) adjusted p-values, though emphasis was given to results of orthogonal polynomial contrasts as described above, because they do not depend upon assumptions of compound symmetry (Wilkinson et al., 1994). Using the first data set, a repeated measures analysis was performed to examine patterns of change in per-unit EDJ length across the first 10 units of enamel growth. Here, an enamel growth unit is defined as a 70-day interval over which EDJ lengths were measured. Per-unit EDJ lengths were the serial criterion variables; tooth type, jaw, and their interaction term were predictor variables. Differences in absolute tooth size affected the number of units for a given tooth. Thus, the first 10 units were chosen to maximize the number of serial intervals analyzed while minimizing reduction in n-value. EDJ lengths were also plotted in the best represented tooth type in this sample, LM1 (n ¼ 10), to graphically assess the patterns of change in per-unit EDJ lengths within a particular tooth type. We then investigated Shellis’s suggestion that higher extension rates in the first-formed enamel characterize teeth with longer EDJ lengths. To do so, as a preliminary step, we calculated Pearson correlation coefficients between enamel extension rates in the first formed enamel unit, total enamel formation times and total EDJ lengths. Because EERs in unit 1 and total enamel formation time were independent of each
Results Data set one Table 3 reports the results of the repeated measures analysis examining patterns of change in per-unit EDJ length across the first 10 units (each unit is a 70-day interval over which EDJ lengths were measured). There was a significant within-subjects decline in EDJ lengths across serial units (p < 0.0005) as well as a significant unit
Table 4 Least squares regression showing variation in enamel formation time, early EERsa, and tooth type accounting for variation in total EDJ length. Criterion Variable: Total EDJ Length N: 43 Multiple R: 0.968 Squared multiple R: 0.937 Analysis of Variance Source
Sum-of-Squares
df
Mean-Square
F-ratio
Total Formation Time EER Units 1 & 2 Tooth Typec Error
22101400.00 17442800.00 5630586.37 10801100.00
1 1 4 36
22101400.00 17442800.00 1407646.59 300031.76
73.66 58.14 4.69
P 0.000 0.000 0.004
% Variance
b
73.7 16.6 3.4
a When EERs of just Unit 1 were used in model, results were similar with about 5.5% less variance in total EDJ length accounted for by early EERs (11.1% vs. 16.6% in the current model). b Percentage of variance estimated from step-wise regression analysis. c Tooth type consolidated by jaw.
D. Guatelli-Steinberg et al. / Journal of Human Evolution 63 (2012) 475e486 Table 5 Multivariate repeated measures analysis of per-decile change in EERs across deciles of lateral enamel. Within subjects Source
SS
Decile of Lateral Enamel 715.85 Decile by Population 18.46 Decile by Tooth 95.63 Decile by Pop by Tooth 150.04 Error 1101.67 Huynh-Feldt Epsilon : 0.645
df
MS
F
P
H-F
9 9 45 45 567
79.54 2.05 2.13 3.33 1.94
40.94 1.06 1.09 1.72
0.000 0.394 0.317 0.003
0.000 0.388 0.341 0.013
MS
F
P
176.46 0.17 0.83 2.20
0.000 0.678 0.531 0.065
27.67 8.41 6.55 4.17
0.000 0.005 0.013 0.045
Single Degree of Freedom Polynomial Contrasts Test of Order 1 (Linear)
Source
SS
df
Decile 587.56 1 587.56 Decile by Population 0.58 1 0.58 Decile by Tooth 13.87 5 2.77 Decile by Pop by Tooth 36.68 5 7.34 Error 209.77 63 3.33 Significant Higher Order Polynomial Contrasts (P 0.05) 3 (Cubic) Decile 94.91 1 94.91 5 Decile 10.39 1 10.39 6 Decile 8.29 1 8.29 6 Decile by Population 5.28 1 5.28
by tooth interaction effect (p ¼ 0.001). Trends across the 10 units were similar for maxillary and mandibular teeth. About 55% of the variance in decline in EDJ lengths per unit was accounted for by a linear trend, with the remaining variance accounted for by progressively smaller percentages associated with significant quadratic, cubic and higher order polynomial contrasts. The unit by tooth interaction was primarily a linear contrast (p ¼ 0.001), though there was also a significant cubic contrast (p ¼ 0.016) and nearsignificant quadratic contrast (p ¼ 0.051). These contrasts reflect the slower reductions in per-unit EDJ length of anterior teeth as compared to molars (see Fig. 4). Given unequal variance in the first unit compared to others, this analysis was repeated with only units two through ten included. Outcomes were similar and are therefore not reported. To evaluate whether the pattern of EDJ length change that differentiates molars and anterior teeth is mimicked within tooth types, EDJ lengths were plotted in the best represented tooth type in this sample, the LM1 (n ¼ 10). The goal was to determine if LM1s with relatively long EDJ lengths have higher initial EDJ lengths, with smaller reductions in extension rate in adjacent units than molars with relatively shorter overall EDJ lengths. Fig. 5 shows the pattern of EDJ length change across LM1s with relatively long EDJ lengths
481
Table 6 Influence of periodicity within a repeated measures analysis of per-decile change in EDJ length relative to change in perikymata number controlling for population group and tooth type. Within Subjects Source
SS
Decile of Lateral Enamel 73.47 Decile by Population 3.01 Decile by Tooth 18.01 Decile by Periodicity 20.26 Error 194.49 Huynh-Feldt Epsilon : 0.944
df
MS
F
P
H-F
9 9 45 36 576
8.16 0.33 0.40 0.56 0.34
24.18 0.99 1.19 1.67
0.000 0.447 0.195 0.010
0.000 0.445 0.201 0.012
Single Degree of Freedom Polynomial Contrasts Test of Order 1 (Linear)
Source
SS
df
MS
Decile 41.11 1 41.11 Decile by Population 0.00 1 0.00 Decile by Tooth 1.60 5 0.32 Decile by Periodicity 4.78 4 1.19 Error 27.74 64 0.43 Significant Higher Order Polynomial Contrasts (P 0.05) 2 (Quadratic) Decile 9.31 1 9.31 3 (Cubic) Decile 17.58 1 17.58 4 Decile by Tooth 3.23 5 0.65 5 Decile 1.56 1 1.56 5 Decile by Tooth 2.68 5 0.54 8 Decile 1.70 1 1.70 8 Decile by Periodicity 3.79 4 0.95 9 Decile by Periodicity 3.14 4 0.78
F
P
94.86 0.00 0.74 2.76
0.000 0.949 0.599 0.035
27.35 25.86 2.38 7.81 2.69 6.16 3.44 3.63
0.000 0.000 0.048 0.007 0.029 0.016 0.013 0.010
(those with EDJ lengths greater than the median value) as compared with the pattern of EDJ length change across LM1s with relatively short EDJ lengths (those with EDJ lengths lower than the median value). This graph suggests that the patterns that distinguish anterior teeth from molars (as shown in Fig. 4) are, in part, related to their longer average EDJ lengths and do not simply reflect differences in shape. To investigate the extent to which initial EERs, total enamel formation time, and tooth type predict variation in EDJ length, we performed a regression analysis with each of these predictors and their interactions. Initial EERs are represented by the first two “units”, that is, the first two 70-day intervals. As none of the interactions were statistically significant, the regression was then performed without the interactions, and the results are reported in Table 4. The squared multiple R for the total model is 0.937, suggesting that this model explains a large portion of the variance in EDJ length. The analysis of variance shows that total enamel
Figure 6. a & b: Changes in enamel extension rate and perikymata per decile of lateral enamel by population group (second data set). Declines in extension rates plotted by decile in this manner appear slightly steeper than they would if EERs had been plotted in equivalent time intervals.
482
D. Guatelli-Steinberg et al. / Journal of Human Evolution 63 (2012) 475e486
Figure 7. a & b: Unadjusted (for tooth type and population) least square mean estimates of EDJ Z-score minus Pk Z-score for each decile of lateral enamel (with distance weighted smoothing) distinguishing among ground sections with different periodicities (7a) or as means (SE) per decile for each periodicity adjusted for tooth type and population group (7b) (second data set).
formation time is the strongest predictor, explaining 73.7% of the variance (p < 0.0005). Early EERs account for about 16.6% of the additional variance in total EDJ length. Tooth type accounts for a smaller percentage of variance (3.4%) but is statistically significant (p ¼ 0.004). The by-tooth-type contrast was primarily between anterior teeth (canines and first incisors) and molars. Data set two Table 5 reports the results of the repeated measures analysis of change in EERs across deciles of lateral enamel using tooth type and population group (southern African or United Kingdom in origin) as predictor variables. Within subjects, there is a statistically significant, primarily linear, decline in extension rates across deciles (p < 0.0005). Although there is no significant interaction between decile of lateral enamel and either population (p ¼ 0.39) or tooth type (p ¼ 0.34), there appears to be a significant interaction among the three variables combined (p ¼ 0.013). Plots of extension rate and perikymata numbers across deciles are given in Fig. 6a and b, respectively. Data are separated by
Figure 8. Distance-weighted least square smoothing shows greater increases (in number of perikymata across deciles of lateral enamel) among teeth with lower periodicities (second data set).
population, but are consolidated for tooth type for simplicity. As can be seen in these figures, while perikymata numbers increase across deciles, extension rates decline. However, it is important to note that the two curves are not inverses of each other. Thus, the pattern of change of perikymata numbers per decile appears to be related to, but not completely determined by, the pattern of change in extension rates per decile. In addition, it is important to note that plotting enamel extension rates by deciles, which divides time into longer intervals as the cervix is approached, slightly exaggerates the rate of decline that would have been obtained had extension rates been plotted in equivalent time intervals as they were in the first data set. For there to be linear declines in extension rates in the lateral enamel, increases in EDJ length across deciles must not be keeping pace with increases in Pk numbers. This can be seen in a repeated measures analysis conducted on the difference between EDJ Zscore and Pk Z-score, with periodicity as the primary predictor and population and tooth-type as covariates. As shown in Table 6, as
Figure 9. Change in mean EERs per decile (SE) illustrates greater decreases in extension rates across deciles of lateral enamel in teeth with higher percentages of perikymata in their cervical halves. Teeth were divided into three groups based on the percentage of perikymata in their cervical halves. Those in the top third were compared to those in the bottom third. (Top third, n ¼ 26, >66.6%; Bottom third, n ¼ 26, 64.0%). Differences in the time intervals within deciles upon which EERs for the two groups (top and bottom thirds) were computed do not make a substantive difference in the outcomes.
D. Guatelli-Steinberg et al. / Journal of Human Evolution 63 (2012) 475e486
483
Figure 10. Cumulative EDJ lengths by age for molars of Miocene, Pliocene and Pleistocene fossil primates, Gorilla, Pan, and modern H. sapiens. Human M1s selected represent the two with the longest and shortest EDJ lengths in the sample. Taxa in legend are ordered according to descending maximum crown height; fossil teeth in order top to bottom in legend are as follows: Gigantopithecus blacki e M3, H. neanderthalensis e M1 (La Chaise BD-J4-C9), H. erectus e M1 (S7-37), Victoriapithecus macinnesi e M2 (KNM-MB 19841), Australopithecus anamensis e M1 (KNM ER-30748), Proconsul nyanzae e M1 (HT34/93), Hispanopithecus laietanus M1 (1781), V. macinnesi e M2 (KNM-MB-27844), H. laietanus e M2 (1794), Proconsul heseloni e M2 (KNM-RU 7290).
well as in Fig. 7a and b, EDJ Z-scores do not keep pace with PK Zscores across deciles (p < 0.0005) as evidenced by the increase in the difference between Z-scores in moving between deciles. The significant first order polynomial test indicates that this trend is primarily linear (p < 0.0005). Trends are not significantly different across deciles for population group or tooth type (p ¼ 0.447 and 0.195 respectively), but changes across deciles vary by periodicity (p ¼ 0.012). (The first order polynomial test of decile by periodicity suggests that these changes are partially linear (p ¼ 0.035), though significant interactions also occur for polynomial tests of order eight (p ¼ 0.013) and nine (p ¼ 0.010)). EDJ Z-score falls relative to Pk Z-score as a function of declining periodicity whether or not adjustments are made for tooth type and population group (Fig. 7a and b). For teeth with lower periodicities, EDJ Z-scores become negative relative to Pk Z-scores
beginning in about the sixth decile, on average. However, at higher periodicities, negative values generally do not begin until the ninth or tenth decile. As noted before, like Reid and Ferrell (2006), we also found that the lower the periodicity, the steeper the increase in Pk numbers across deciles. Fig. 8 illustrates how the shapes of the perikymata curves vary by periodicity. As the cervix is approached, more perikymata become packed into cervical deciles, but this increase is most pronounced at lower periodicities. Fig. 9 shows that the 26 ground sections (approximately one third of the total sample) with the highest percentages of perikymata in their cervical halves (>66.6%) have consistently higher EERs over their first five deciles than do the 26 ground sections with the lowest percentages (64.0%). Numbers of teeth in these two categories do not differ significantly by population (Fishers exact, p ¼ 0.58) or tooth type (X2 ¼ 0.37, df ¼ 2, p ¼ 0.83). These analyses strongly suggest that individuals who have a higher proportion of their total perikymata packed in to their cervical regions have lower periodicities and more rapid EERs in the first several deciles. In addition, it is especially in individuals with lower periodicities that changes in EDJ lengths do not keep pace with increases in Pk numbers in later deciles. Together, these analyses indicate that whether from southern African or northern Europe, individuals with higher percentages of perikymata in the cervical regions of their teeth tend to have steeper declines in extension rates across deciles of lateral enamel. Discussion and conclusions
Figure 11. Polynomial regression plot for EDJ length against crown formation time (CFT) in a sample combining the modern human teeth described in this study with 54 mandibular teeth of Pan troglodytes. R2 ¼ 0.907; p < 0.0001. Y ¼ 5123.257 e 371.222 * X þ 364.202 *X2. Y ¼ EDJ length, X ¼ CFT. In this comparative context, rising EERs at the cervix of very tall incisors and canines probably account for the non-linear relationship between EDJ length and CFT.
This investigation began with an assessment of EER changes from cusp to cervix. To Shellis (1984) it appeared that the initial rate of enamel extension and subsequent rate of decline in EERs was related to total EDJ length in permanent teeth. Shellis based this conclusion on graphs showing different patterns of EER decline in anterior teeth (with long EDJ lengths) vs. molars (with shorter EDJ lengths) and in cusps with different EDJ lengths in the same tooth type. Our analysis is consistent with Shellis’s findings, demonstrating a statistically significant association between tooth type and the pattern of EER change. Anterior teeth tend to start with higher initial EERs and decline more gradually than do molars. However, we also examined the pattern of EER change within the
484
D. Guatelli-Steinberg et al. / Journal of Human Evolution 63 (2012) 475e486
best-represented tooth type in this sample, the LM1. Lower first molars with longer EDJ lengths start with higher initial rates and decline to lower rates more gradually than do those with shorter total EDJ lengths. Thus, the pattern of EER change that differentiates molars from anterior teeth is similar to the pattern that differentiates short-EDJ vs. long-EDJ lower first molars, supporting Shellis’s contention that the main explanatory factor is total EDJ length. The regression analysis of EDJ length as a function of total enamel formation time, initial EERs (in units 1 and 2), and tooth type also supports the relationship between EDJ length and initial EERs, independently of tooth type. In this analysis, with tooth type and enamel formation time statistically controlled, initial EERs accounted for a statistically significant portion of the variance in EDJ length (16.6%). Yet, we found that total enamel formation time was the greatest predictor of variation in EDJ length (73.7%). Combined, initial EERs and enamel formation time explained 90.3% of the variance in EDJ length. Thus, these findings suggest that within modern human tooth types, taller crowns are associated with faster rates of enamel extension early in crown development and extended periods of enamel formation when EERs are relatively low. Because EERs reflect crown growth rates and enamel formation periods reflect crown growth periods, taller human crowns therefore appear to be produced by elevating rates of growth early in crown development and by prolonging the period of growth when growth rates are low. In this study, total enamel formation time and the EER of unit 1 were found to be independent of each other (see preliminary analysis in Materials and Methods), with high initial enamel extension rates and extended crown formation times both associated with longer EDJ lengths. The independence of initial EERs and crown formation time is consistent with the fact that enamel extension rates do not determine when crown formation ceases or how long it takes. This is entirely controlled by the signals that switch off ameloblast differentiation at the cervix (Tummers and Thesleff, 2003). This independence is also consistent with Dean (2009) finding that the crown formation time of the P. robustus lower canine SK 63, which is shorter than that of modern humans, is not associated with faster rates of enamel extension (as calculated at regular intervals along the EDJ). Indeed, the initial rate of enamel extension in SK 63 falls within the lower end of the modern human range (Dean, 2009). Taken together, these findings point to the independence of EERs and crown formation time at the molecular level (Tummers and Thesleff, 2003), within modern human tooth types (this study), and possibly also across species (Dean, 2009). Although the comparative data for changing EERs and cumulative growth in crown height are sparse (with rarely more than one example from one tooth type per fossil species published), enough published data now exist to strengthen the case that enamel formation times and EERs across fossil primate species vary independently (Fig. 10). Among Miocene and Plio-Pleistocene fossil primates, high EERs in the first formed cuspal enamel of Victoriapithecus macinnesi (Dean and Leakey, 2004) and Gigantopithecus blacki (Dean and Schrenk, 2003) are both comparable to those known for Gorilla molars (Dean, 2010) but the two fossil species have very different enamel formation times (2.5 years respectively). Both, like Gorilla, however, achieve relatively tall crown heights early on in their development. Other examples (Proconsul heseloni, Proconsul nyanzae (Beynon et al., 1998), Hispanopithecus laietanus (Dean and Kelley, 2012), and Australopithecus anamensis (Dean, 2010)) have cuspal EERs that reduce quite quickly to values typical of lateral enamel and tend as a consequence to have squat low crowns but again with contrasting enamel formation times. Clearly, taller cusps may result from a short period of time when EERs remain high but high EERs may have little impact on final crown height if they reduce quickly.
Enough data for Pan have now been published to explore the relationship between EDJ length and crown formation time in a comparative context. Fig. 11 combines the data for modern humans described in this study with published data (Dean, 2010; Dean and Kelley, 2012) for 54 mandibular teeth of Pan collected using exactly the same methods. All molar and anterior tooth types are represented in this combined sample. As demonstrated for the human sample described here, Fig. 11 shows that a strong relationship appears to exist between EDJ length and crown formation time in this broader comparative context. The relationship, however, becomes non-linear when teeth with longer crown formation times are included. All teeth with crown formation times greater than five years in this plot are tall Pan canines for which perikymata per mm have been shown to reduce in number at the cervix (Dean and Reid, 2001). The relationship between EERs and perikymata spacing at the surface (demonstrated in this study) appears now to explain how a greater EDJ length is achieved for a given crown formation time when EERs begin to rise again towards the cervix of tall anterior teeth. Within the tooth types of any particular fossil primate species, the relative contribution of crown formation time and EERs to variation in EDJ length is not currently known. However, the previously published data shown in Fig. 10 suggests that it is not just within Homo sapiens that longer EDJ lengths are in part produced by faster rates of crown growth, as reflected in faster EERs. In Fig. 10, growth curves are plotted for two teeth each of H. sapiens, H. laietanus, and V. macinnesi. For each of these species, the tooth with the longer EDJ length appears to start with a higher initial rate and declines more gradually. Often only small samples of fossil primate teeth are available for histological comparison of their EERs or crown formation times with modern humans (e.g., Dean, 2009; Smith et al., 2010). As EDJ lengths are related to crown heights, specimens close to a species’ average crown height would provide the most species-typical data on EERs and/or crown formation times. Certainly, it is not always possible to choose teeth with representative crown heights for studies of fossil enamel growth. However, it is important to bear in mind that EER and crown formation time data are both likely to be sensitive to within-species variation in crown height. Such withinspecies variation is of sufficient magnitude to complicate attempts to compare EERs and crown formation time across species. For example, as shown in Fig. 10, the V. macinnesi tooth with the longer EDJ length has an enamel growth curve much more like that of G. blacki than does the V. macinnesi tooth with the shorter EDJ length. As an additional example, the larger H. laietanus tooth falls along the growth curve of the smaller modern human tooth, while the smaller H. laietanus tooth falls well below it. Independence of EERs from crown formation time would help to resolve an apparent contradiction in recent studies of Neandertal dental development. Smith et al. (2010) suggested that faster whole crown average EERs in Neandertals underlie their generally shorter crown formation times. Dean (2009), however, found that serial EER measurements in the La Chaise Neandertal lower first molar (BD-J4-C9) all fell within modern human ranges (although with relatively high values in the cusp and low values in the lateral enamel). Whole crown average extension rates are calculated by dividing total EDJ length by total crown formation time (Smith et al., 2010) and are therefore related to crown formation time by definition, while serial EER measures along the EDJ are not. As noted, across primate taxa, fast EERs along the EDJ do not necessarily result in teeth with short crown formation periods (Fig. 10), most likely because the end of crown formation is under independent molecular control through the processes that terminate ameloblast differentiation (Tummers and Thesleff, 2003). It follows that Neandertals might well have shorter than average modern
D. Guatelli-Steinberg et al. / Journal of Human Evolution 63 (2012) 475e486
human crown formation times and serial EERs within the modern human range. With crown formation times at the low end of the modern human range, the tail of the exponentially declining EER curve would be truncated, driving up Neandertal whole crown average EERs. Cuspal EERs at the high end of the modern human range, if those of La Chaise BD-J4-C9 can be taken as representative, would also tend to drive up Neandertal whole crown average EERs. The present study suggests that high cuspal EERs would be expected if Neandertals have taller molar crowns (i.e., longer EDJs) than do modern human comparative samples. Nevertheless, to determine whether Neandertals are in general similar to modern humans in their EER curves (i.e., not just in the case of La Chaise BDJ4-C9) will require an analysis of a sample of teeth that are representative of the range of Neandertal crown heights. In the present study, we also found that there are statistically significant declines in EERs across deciles of lateral enamel, though they are far less pronounced than the declines in EERs from cusp to cervix. Two previous studies on EER variation along the EDJ (Shellis, 1984; Dean, 2009) indicated that EERs remain low and fairly constant during what appears to be relatively large portions of lateral enamel formation time. However, in neither of these studies was enamel formation divided between cuspal and lateral enamel formation, so whether EERs remained constant throughout the whole of lateral enamel formation time was unclear. Our findings demonstrate that declines in EERs are not limited to the earliest periods of enamel extension; statistically significant, though much more gradual declines are common in lateral enamel as well. Our second data set, in which extension rates were averaged within deciles, however, does not pick up the slight increase in extension rates close to the cervix that is evident in the first data set (Fig. 2; Fig. 5), in which extension rates were calculated in smaller time intervals. In this study, central tendencies in the rate of decline across lateral enamel deciles were not related to population or tooth type. They were, however, statistically significantly related to periodicity. Across deciles, especially in teeth with lower periodicities, the number of perikymata per decile increases at a faster rate than does EDJ length, reflecting declining extension rates. In teeth with higher periodicities, the difference in rates of change in perikymata numbers and EDJ lengths across deciles is not as great, reflecting a less rapid decline in extension rates. As these relationships hold for both southern Africans and people from the United Kingdom, these findings suggest that there is a common cause for variation in perikymata distribution in modern humans. The same cannot be determined for fossil hominins, for whom it remains possible that other factors might be affecting the distribution of perikymata. As noted, such factors include variation in secretion rates in the outer enamel and variation in the surface contour of teeth. These factors would be expected to differ more between modern humans and fossil hominins than among modern human populations. Nevertheless, by integrating previous studies with the present study, it is possible to further evaluate potential causes of the difference in perikymata distribution between Neandertals and modern humans. Differences in enamel surface curvature between Neandertals and modern humans do not appear to be sufficient to explain the perikymata distribution differences between these species (Guatelli-Steinberg et al., 2007). Furthermore, it appears that Neandertals and modern humans may not differ in enamel secretion rates (Macchiarelli et al., 2006; Smith et al., 2010). It becomes more probable, therefore, that differences in the rate of EER decline in Neandertals and modern humans are responsible for their different perikymata distributions. However, because EERs in the lateral enamel of both modern humans and the La Chaise Neandertal (BD-J4-C9) are all low relative to cuspal values, such potential EER differences must represent only a subtle growth
485
difference and may be more marked in some tooth types than in others. A better understanding of space constraints within the crypts of different tooth types during lateral enamel formation might reveal why some teeth in the jaws grow more slowly in height at the cervix than others, and why modern humans appear to differ from other fossil taxa in their cervical EERs. In conclusion, this study demonstrates that there are exponential declines in EERs in modern humans, but that initial EERs and patterns of decline (abrupt or more gradual) are related to the length of the EDJ. Furthermore, this study suggests that crown height variation within modern human tooth types results from both variation in crown formation time and variation in enamel extension rates. Taller human teeth grow both longer and faster. These findings echo the developmental mechanisms producing the sexually dimorphic canines of some non-human anthropoid primates, for whom male-female differences in crown height are primarily achieved through differences in crown formation time, with a lesser role for differences in rates of growth (Swindler, 1985; Schwartz et al., 1999, 2001; Schwartz and Dean, 2001; Leigh et al., 2005; Guatelli-Steinberg et al., 2009). In examining other primate taxa for which previously published data exists, we find that EDJ lengths appear to be related to initial EERs and crown formation time to varying degrees. These relationships suggest that studies of EERs and crown formation times across primate fossil taxa, for which there are often only a small number of available teeth, can be affected by within-taxon variation in EDJ length. Finally, this study also indicates that there are declines, though far less pronounced, in the extension rates of lateral enamel. Variation in these declines appears to be reflected in variation in perikymata distribution. These results also may indicate that the perikymata distribution difference between Neandertals and modern humans is related to a difference in the pattern by which EERs decline in the lateral enamel in these species. It is important to bear in mind, however, that species differences in surface curvature and enamel secretion rates can also influence species differences in perikymata distribution. Acknowledgments We thank the editor and reviewers for their thoughtful suggestions which helped us to improve this manuscript. References Aiello, L., Dean, C., 1990. An Introduction to Human Evolutionary Anatomy. Academic Press, London. Beynon, A.D., Wood, B.A., 1987. Patterns and rates of enamel growth in the molar teeth of early hominids. Nature 326, 493e496. Beynon, A.D., Dean, M.C., Leakey, M.G., Reid, D.J., Walker, A., 1998. Comparative dental development and microstructure of Proconsul teeth from Rusinga Island, Kenya. J. Hum. Evol. 35, 163e209. Boyde, A., 1964. The Structure and Development of Mammalian Enamel. PhD thesis, University of London, London, UK. Bromage, T.G., Dean, M.C., 1985. Re-evaluation of the age at death of immature fossil hominids. Nature 317, 525e527. Dean, M.C., 1987. The dental development status of six East African juvenile fossil hominids. J. Hum. Evol. 16, 197e213. Dean, M.C., 2006. Tooth microstructure tracks the pace of human life-history evolution. Proc. R. Soc. B 273, 2799e2808. Dean, M.C., 2009. Extension rates and growth in tooth height of modern humans and fossil hominin canines and molars. In: Koppe, T., Alt, K.W. (Eds.), Comparative Dental Morphology. Front. Oral. Biol. Basel, vol. 13. Karger, pp. 68e73. Dean, M.C., 2010. Retrieving chronological age from dental remains of early fossil hominins to reconstruct human growth in the past. Phil. Trans. R. Soc. B 365, 3397e3410. Dean, M.C., 2012. A histological method that can be used to estimate the time taken to form the crown of a permanent tooth. In: Bell, L.S. (Ed.), Forensic Microscopy for Skeletal Tissues: Methods and Protocols. Methods in Molecular Biology, vol. 915. Springer Science and Business Media, LLC.
486
D. Guatelli-Steinberg et al. / Journal of Human Evolution 63 (2012) 475e486
Dean, M.C., Kelley, J., 2012. Comparative dental development in Hispanopithecus laietanus and Pan troglodytes. J. Hum. Evol. 62, 174e178. Dean, M.C., Leakey, M.G., 2004. Enamel and dentine development and the life history profile of Victoriapithecus macinnesi from Maboko Island, Kenya. Ann. Anat. 186, 405e412. Dean, M.C., Reid, D.J., 2001. Perikymata spacing and distribution on hominid anterior teeth. Am. J. Phys. Anthropol. 116, 209e215. Dean, M.C., Schrenk, F., 2003. Enamel thickness and development in a third permanent molar of Gigantopithecus blacki. J. Hum. Evol. 45, 381e387. Dean, M.C., Shellis, R.P., 1998. Observations on stria morphology in the lateral enamel of Pongo, Hylobates, and Proconsul teeth. J. Hum. Evol. 35, 401e410. Dean, C., Leakey, M.G., Reid, D.J., Shrenk, F., Schwartz, G.T., Stringer, C., Walker, A., 2001. Growth processes in teeth distinguish modern humans from Homo erectus and earlier hominins. Nature 414, 628e631. FitzGerald, C.M., 1998. Do enamel microstructures have regular time dependency? J. Hum. Evol. 35, 371e386. Guatelli-Steinberg, D., Reid, D.J., 2008. What molars contribute to an emerging understanding of lateral enamel formation in Neandertals vs. modern humans. J. Hum. Evol. 54, 236e250. Guatelli-Steinberg, D., Reid, D.J., 2010. Brief communication: the distribution of perikymata on Qafzeh anterior teeth. Am. J. Phys. Anthropol. 141, 152e157. Guatelli-Steinberg, D., Reid, D.J., Bishop, T., 2007. Did the lateral enamel of Neandertals grow differently from that of modern humans? J. Hum. Evol. 52, 72e84. Guatelli-Steinberg, D., Reid, D.J., Bishop, T.A., Larsen, C.S., 2005. Anterior tooth growth periods in Neandertals were comparable to those of modern humans. Proc. Natl. Acad. Sci. 102, 14197e14202. Guatelli-Steinberg, D., Ferrell, R.J., Spence, J., Talabere, T., Hubbard, A., Schmidt, S., 2009. Sex differences in anthropoid mandibular canine lateral enamel formation. Am. J. Phys. Anthropol. 140, 216e233. Hillson, S., 1996. Dental Anthropology. Cambridge Univ. Press, Cambridge. Hillson, S., Bond, S., 1997. Relationship of enamel hypoplasia to the pattern of tooth crown growth: a discussion. Am. J. Phys. Anthropol. 104, 89e103. Leigh, S.R., Setchell, J.M., Buchanan, L.S., 2005. Ontogenetic bases of canine dimorphism in anthropoid primates. Am. J. Phys. Anthropol. 127, 296e311. Macchiarelli, R., Bondioli, L., Debéhath, A., Mazurier, A., Tournepiche, J.-F., Birch, W., Dean, C., 2006. How Neandertal molar teeth grew. Nature 444, 748e751. Ramirez-Rozzi, F.V., Bermudez de Castro, J.M., 2004. Surprisingly rapid growth in Neanderthals. Nature 428, 936e939.
Reid, D.J., Dean, M.C., 2006. Variation in modern human enamel formation times. J. Hum. Evol. 50, 329e346. Reid, D.J., Ferrell, R.J., 2006. The relationship between number of striae of Retzius and their periodicity in imbricational enamel formation. J. Hum. Evol. 50, 195e202. Reid, D.J., Guatelli-Steinberg, D., Walton, P., 2008. Variation in modern human premolar enamel formation times: implications for Neandertals. J. Hum. Evol. 54, 225e235. Schwartz, G.T., Dean, C., 2001. Ontogeny of canine dimorphism in extant hominoids. Am. J. Phys. Anthropol. 115, 269e283. Schwartz, G.T., Reid, D.J., Dean, C., 2001. Developmental aspects of sexual dimorphism in hominoid canines. Int. J. Primatol. 22, 837e860. Schwartz, G.T., Reid, D.J., Dean, M.C., Chandrasekera, M.S., 1999. Aspects of tooth crown development in common chimpanzees (Pan troglodytes) with a note on the possible role of sexual dimorphism in canine growth. In: Mayhall, J.T., Heikkinen, T. (Eds.), Proceedings of the 11th International Symposium on Dental Morphology. University of Oulu Press, Oulu, Finland, pp. 323e337. Shellis, R.P., 1984. Variations in growth of the enamel crown in human teeth and a possible relationship between growth and enamel structure. Archs. Oral Biol. 29, 697e705. Smith, T.M., Tafforeau, P., Reid, D.J., Grün, R., Eggins, S., Boutakiout, M., Hublin, J.-J., 2007a. Earliest evidence of modern human life history in North African early Homo sapiens. Proc. Natl. Acad. Sci. 104, 6128e6133. Smith, T.M., Toussaint, M., Reid, D.J., Olejniczak, A.J., Hublin, J.-J., 2007b. Rapid dental development in a Middle Paleolithic Belgian Neanderthal. Proc. Natl. Acad. Sci. 104, 20220e20225. Smith, T.M., Tafforeau, P., Reid, D.J., Poueceh, J., Lazzari, V., Zermeno, J.P., GuatelliSteinberg, D., Olejniczak, A.J., Hoffman, A., Radov ci c, J., Makaremi, M., Toussaint, M., Stringer, C., Hublin, J.-J., 2010. Dental evidence for ontogenetic differences between modern humans and Neanderthals. Proc. Natl. Acad. Sci. 107, 20923e20928. Sokal, R.R., Rohlf, F.J., 1995. Biometry: The Principles and Practice of Statistics in Biological Research, third ed. W. H. Freeman and Company, New York. Swindler, D.R., 1985. Nonhuman primate dental development and its relationship to human dental development. In: Watts, E.S. (Ed.), Nonhuman Primate Models for Human Growth and Development. Alan R. Liss, New York, pp. 67e94. Tummers, M., Thesleff, I., 2003. Root or Crown: a developmental choice orchestrated by the differential regulation of the epithelial stem cell niche in the tooth of two rodents. Development 130, 1449e1457. Wilkinson, L., Hill, M.A., Howe, P., Birkenbeuel, G., Beck, J., Liu, J., 1994. SYSTAT for DOS: Advanced Applications, Version, sixth ed. SYSTAT, Inc., Evanston, IL.
