Human Biology Open Access Pre-Prints

Transcription

1 Wayne State University Huan Biology Open Access Pre-Prints WSU Press -- New Approaches to Juvenile Age Estiation in Forensics: Application of Transition Analysis via the Shackelford et al. Method to a Diverse Modern Subadult Saple Kelly R. Kanikar Michigan State University, kellykanikar@gail.co Nicholas P. Herrann Texas State University, San Marcos, nph@txstate.edu Aber M. Pleons Michigan State University, pleonsa@su.edu Recoended Citation Kanikar, Kelly R.; Herrann, Nicholas P.; and Pleons, Aber M., "New Approaches to Juvenile Age Estiation in Forensics: Application of Transition Analysis via the Shackelford et al. Method to a Diverse Modern Subadult Saple" (). Huan Biology Open Access Pre-Prints.. This Open Access Preprint is brought to you for free and open access by the WSU Press at DigitalCoons@WayneState. It has been accepted for inclusion in Huan Biology Open Access Pre-Prints by an authorized adinistrator of DigitalCoons@WayneState.

2 New Approaches to Juvenile Age Estiation in Forensics: Application of Transition Analysis via the Shackelford et al. Method to a Diverse Modern Subadult Saple Kelly R. Kanikar, * Nicholas P. Herrann, and Aber M. Pleons Departent of Anthropology, Michigan State University, East Lansing, Michigan, USA. Departent of Anthropology, Texas State University, San Marcos, Texas, USA. *Correspondence to: Kelly R. Kanikar, Departent of Anthropology, Michigan State University, Auditoriu Drive, East Lansing, MI USA. E-ail: kanikar@su.edu. Short Title: Transition Analysis in Dental Age Estiation KEY WORDS: FORENSIC ANTHROPOLOGY, DENTAL DEVELOPMENT, RADIOGRAPHS.

3 Abstract Dental developent is one of the ost widely utilized and accurate ethods available for estiating age in subadult skeletal reains. The tiing of tooth growth and developent is regulated by genetics and less affected by external factors, allowing reliable estiates of chronological age. Traditional ethodology focused on coparing tooth developental scores to corresponding age charts. Using the Moorrees, Fanning, and Hunt developental scores, Shackelford and colleagues ebed the dental developent ethod in a statistical fraework based on transition analysis. They generated nuerical paraeters underlining each stage and age-at-death distribution and applied the to fossil hoinins and Neanderthals with liited application to odern huans. We use this sae ethod on a subadult test saple (n=), representing odern individuals that ay becoe part of the forensic record. We assess the probability coverage of the Shackelford et al. ethod derived fro MFH standards as it applies to all available dentition. Results indicate proise as the age range at and confidence levels include the chronological age of alost every individual tested. The axiu likelihood age estiates (MLE) underestiate age by. to. years for individuals aged -, and greater than. years fro to years, as previously shown. In an attept to refine the ethod, we adjusted the nuerical paraeters underlying the stages for developing teeth based on a cobined odern reference saple (n=) and tested these revised paraeters using the sae test saple. The estiated ages fro the odified ethod differ fro the original Shackelford et al. ethodology by underestiating age to a lesser degree. The odified ethod does include ean age-at-attainent values for earlier stages of several teeth allowing for the calculation of ore narrow confidence intervals. While this study highlights areas of future research in refining

4 dental developental aging by transition analysis, it also deonstrates that the Shackelford et al. ethod is applicable and accurate when aging odern subadults in forensic work.

5 Age estiation is an essential coponent in establishing a biological profile of a set of unidentified huan skeletal reains. This inforation is used to narrow down the nuber of potential anteorte coparisons when aking an identification. In its ost basic for, age estiation is based on predictable patterns of growth and developent or degeneration of bony features and/or dentition. In subadults, dentition-based ethods are preferred for estiating age because of their high degree of accuracy and reliability. The tiing and sequence of tooth growth and developent is heavily regulated by genetics and inially ipacted by environental or cultural factors, as is the case with other skeletal age indicators (Ubelaker ; Moorrees et al. a, b; Scheuer and Black ). Because of their strict sequence in growth and developent, dental age estiation ethods are highly reliable, with estiates as narrow as onths to greater than three years in either direction depending on the ethod used (Reppien et al. ; Liversidge ; Phillips and van Wyk Kotze ). Further, dental developent is largely applicable and can be used across populations. Traditional dentition-based age estiation ethods for subadults have focused on coparisons of crown and root developent (Moorrees et al. a; Deirjian et al. ) and coparisons of erupted teeth to dental charts and atlases (Schour and Massler ; Gustafson and Koch ; Ubelaker ; Kahl and Schwarze ; AlQahtani et al. ). Methods that assess the degree of enael and root foration have proven to be superior to dental eruption patterns as eruption patterns are affected by various factors including tooth loss and available space in the dental arcade (Shackelford et al. ). A popular ethod exaining dental developent was created by Moorrees et al. (a, b). They developed graphical representations of dental developent phases throughout the subadult life stage based on a longitudinal study of subadult dental radiographs. Unfortunately, the nuerical paraeters

6 associated with the study saple are not available, liiting their assessent capabilities (Shackelford et al. ). Nonetheless, researchers believed the ethod to be valuable, and have adapted the ethod to provide nuerical and statistical data associated with the phases developed by Moorrees et al. (referred to herein as the MFH ethod ) (a, b) (Phillips and van Wyk Kotze ; Liversidge ). A review of the literature deonstrating the process in refining and odifying the MFH ethod is discussed in Shackelford et al. () and will not be reiterated here. Shackelford et al. () expanded the MFH ethod through the application of transition analysis to developental phases. This ethod, referred to here as the SSK ethod, was developed to estiate age in odern, archaeological, and early hoinin fossil groups. Shackelford et al. () calculated age at death paraeters through digitization of the graphics in the original Moorrees et al. (a) publication. Because the SSK ethod was developed for early hoinin saples, and inially tested on odern individuals, its perforance reliability is unknown for a large saple of forensic casework. The SSK ethod provides axiu likelihood point age estiates (MLE) and age ranges expressed as confidence intervals (CIs) at the and levels, satisfying the Daubert requireents for forensic evidence (Christensen and Crowder ). Iportantly, transition analysis allows for age to be estiated without the need of inforative priors, reducing the ipact of age iicry, a coon issue in age estiation ethods in forensic casework (Milner and Boldsen ). Harris () argues that the MFH ethod allows for lower observer error and higher accuracy. This, in cobination with transition analysis, akes the ethod ideal for forensic casework.

7 The current study is twofold. First, it ais to validate the SSK ethod on forensically significant subadult skeletal reains. Second, it tests the accuracy of the estiates using the original transition paraeters derived fro Moorrees et al. (a) by Shackelford et al. ) against a recalculated age of transition structure based on a ore recent subadult saple. Here, we use a odified version of the SSK ethod to assess dental age through MLE and CIs in a U.S. forensic saple with known ages. We then generate new age paraeters using forensically significant speciens of known age individuals fro London and South Africa, which are then substituted into the original SSK code to reflect variation in odern dental developent. Lastly, we use a odern U.S. saple to evaluate the odified ethod using the newly calculated ean age-at-attainent paraeters. The purpose of this research is to validate the use of transition analysis in odern subadult dental aging ethods and explore refineent of age estiation paraeters in subadult aging ethods using dental developent fro forensically significant saples. Materials and Methods Three different saples of known age individuals and their associated tooth scores were used to address the research questions. Two saples were cobined and used as reference aterial for recalculating age paraeters, while the third saple was used for testing. The Reference Saple. The reference dataset, (n=) is derived fro two, large, known-age saples of odern subadults fro South Africa (Phillips and van Wyk Kotze ) and London, England (Liversidge ) (see Table ). The South African saple is derived fro two different sources of radiographic aterial taken in the late s to early s. The first

8 source is coposed of pantoographic radiographs fro the archival records of the Dental Faculty of the University of Western Cape fro ixed ancestry children and Xhosa children, a Bantu population. Individuals of ixed ancestry represent individuals with various ancestral groups fro slaves, indigenous Khoisan, and European descent (Phillips and van Wyk Kotze ). The second source includes an Indian saple and a Zulu subsaple fro two orthodontic offices in Durban Kwa-Zulu Natal. Ages in the South African saple range fro to years. Each tooth in the dental arcade was previously scored following the Moorrees, Fanning, and Hunt (a) ethodologies. The London saple is coposed of panoraic dental radiographs taken at the Institute of Dentistry, Bart s and the London School of Medicine and Dentistry in London, England. The patients range in age fro. to. years old and are coposed of ales and feales fro White and Bangladeshi ethnic groups. No scan dates were provided in the original publication (Liversidge ). Mandibular teeth on the left side were previously scored in the London saple following MFH ethod with the addition of a crypt stage described in Liversidge (). The England dataset was reconciled to atch the original Moorrees et al. (a) scores prior to analysis. The raw tooth scores were used fro both datasets to create our reference saple. The Test Saple. A test saple was created fro a subset of radiographic data (n=; N=,) collected fro the Pediatric Radiology Interactive Atlas (Patricia) databank (Ousley et al. ). The Patricia databank is a forensic saple coposed of non-standard radiographic iages taken during autopsy or physical exaination of subadults that died in the U.S. after January,. We aied to collect forty individuals fro each age group but were liited by two criteria (see Table and Figure ). Radiographic iages were chosen based on two query

9 variables: iage quality and age. Only iages that corresponded to an iage quality of very good or good were collected for individuals aged to. This saple ay not be ideal, but because of its nature, it represents the type of data coonly encountered by forensic practitioners in casework as any edical exainer s and coroner s offices do not have access to advanced iaging technology. Dental developent was assessed fro visible dentition in each radiograph following Moorrees et al. s (a) original publication. The SSK ethod estiates age by assessing dental developent scores via the statistical software, R (R Core Tea ). Dental developent phase data is via a data.frae in R that requires a nuerical score or NA for the following dentition: dc, d, d, UI, UI, LI, LI, C, PM, PM, M, M, and M. Available and clearly visible teeth were scored for every individual. If a tooth was absent or not easily visible, it was assigned a value of NA. Anterior dentition was frequently unobservable due to the lateral radiographs depicting the incisors and canines as stacked and difficult to distinguish. All individuals who had only one tooth scored, or the full suite of dentition scored as Ac (apex closed, ) for all teeth were reoved fro subsequent analysis, as TA analysis requires at least one tooth to still be developing in order to provide the upper range estiate. Elain and Liversidge () note that alnutrition doesn t significantly ipact the tiing and developent of dentition. Therefore, the use of Patricia, a forensic saple where cause of death was unknown, was deeed appropriate for use in this study. We first calculated the coverage of the reference saple within its age liits. This test easures the perforance of the saple within the age bounds (Liversidge ), by assessing the relationship between the chronological age and estiated age of the reference saple. Acceptable coverage eans that of the saple should be captured within the calculated age

10 range (have actual ages within the range), while the reaining of the saple should be split equally above and below the range (Konigsberg et al. ). Coverage was assessed coparing the calculated MLE values to the age cohort based on chronological age. The scores for each individual in the Patricia saple were first run through the original SSK ethod code in R Studio (see Konigsberg s website ), then a odified version. Our odified version, called tooth.test (see Suppleent A) is a function that loops a large dataset through the get.age function and copiles each output in a single.csv file. This function has two iportant aspects. One aspect displays a line at the MLE, and another set of lines reflecting the within plus between-tooth variance values and the within-tooth variance value in the associated age estiation graphic (Figure ). The other aspect sets the high value of the age estiate based on tooth scores of the teeth present in the data entry sheet. Values returned were the high age estiate value (hi), the ean natural log conception-corrected age (u), the withintooth variance, the between-tooth variance, and the lower and upper liit of integration on a straight scale. We calculated the MLE, the upper range and the lower range at the,, and CI using u. Testing the SSK Method. In the second part of this study, the original age paraeters fro the SSK ethod were replaced with the newly generated age paraeters, and the Patricia test saple was run through the tooth.test loop function again. The MLE ages and CIs fro the new age paraeters were calculated and copared to the unodified ethod. Recalculation of the Age-at-Death Paraeters

11 Based on the MLE scores fro the reference saple, the age paraeters of the SSK ethod were recalculated to reflect dental developent in a ore forensically significant population. First, the scoring syste for each tooth was optiized following the Lagrange ultiplier test described in Konigsberg et al. (). With this test, outliers for each tooth at each stage were identified and reoved. Next, the three separate data tables (MFH, MFH, and MUS) that infor the get.age function were recalculated using the reference saple. A discussion on the ethods used to copile these tables are beyond the scope of this paper and can be found in Shackelford et al. (). Results The general project outline was to assess the SSK ethod, recalculate the underlying paraeters, and copare the odified ethod to the original. The results are structured to reflect that order. In general, the original SSK ethod perfored well for estiating age in subadults between and years old in the Patricia dataset. The original ethod underestiated age by less than one year for individuals aged to years. Once individuals reached age, underestiation increased to to years. After age, underestiation increased to + years (Table ). At age, ages were underestiated were by. years. Coverage values for the original ethods at the, and CI are displayed graphically in Figure a. Between the ages of and, thirteen individuals in the test saple did not produce enough inforation to calculate a between-tooth variance value, which is necessary to calculate CI bands. The underlying paraeters in the MFH, MFH, and MUS tables were recalculated for each stage and tooth (see Suppleent B). Values for dc, d, d, UI, UI, and early stages of developent in C and M were suppleented with Shackelford et al. s () original data due

12 to underrepresentation in the reference saple. Results fro the Lagrange test are listed in Table. The optiization test did indicate that scores for P and M in feales, ight benefit fro reevaluation or collapse of scoring stages. All other stages were optiized once the outliers were reoved. The test saple under the odified paraeters produced an MLE that was closer to : ratio with chronological age than the SSK paraeters. Table shows the percentage of individuals whose chronological age fell within the calculated age range (CI band). The odified paraeters narrowed the CI bands, which soeties excluded chronological age fro the estiated range. These excluded individuals were typically less than +/- year outside of the cohort s age range. Coparison of the Original and Modified SSK Methods Because the reference saple did not include individuals under years old, we excluded individuals younger than years fro the Pearson test. Correlation between the MLEs and chronological age on individuals older than years of age returned a value of. for the original and the odified SSK ethods. Despite a high correlation with age, coparisons of average differences between MLEs and chronological age by cohort were different across the two ethods (Table ). The odified ethod underestiated age to a lesser degree than the original SSK ethod (Figure ). Further, the revised ethod generated a narrower age range fro CI calculation (Figure b). Interestingly, under the paraeters of the odified ethod, CI bands were generated for ore of the test saple for ages and, indicating better perforance in estiating the variance than the original ethod (Table ).

13 Discussion The goals of this research were to ) validate the original SSK ethod for use in forensic casework and ) test the original paraeters against recalculated age-of-attainent paraeters in a odern subadult saple to deterine if the ethod could be further refined. Overall, the SSK ethod perfors well when estiating age, especially in individuals younger than years old. After years of age, the ethod begins to slightly underestiate age, a trend that increases to + years after years of age. Constrained by the Patricia test saple, estiates of individuals in their teenage years ay not be accurately capturing variation, as ore than one third of the saple is outside the bounds of our reference saple. The Patricia saple ay not be the ost suitable for evaluating a ethod s perforance; however, it is realistic and represents real-world scenarios. In our test of the SSK ethod, several cohorts had differences between the chronological age and MLE of - years or less. The largest average difference between estiated MLE and chronological age was for the.-. cohort, with an average difference of -. years. Because of our sall test saple size for -year-olds, this could represent delayed developent in the second and third olars, which is not unusual as third olar foration is ore variable between the sexes (Mincer et al. ) and populations (Prieto et al. ). Underestiation of age using the MFH score syste is consistent with previous studies (Liversidge ; Phillips and van Wyk Kotze ). The SSK ethod is based on Moorrees et al. s original study and graphs, which, when reevaluated (Šešelj et al. ), indicate discrepancies in crown and root developent ages in the original publication (Moorrees et al. a), which ay explain soe of the underestiation. The Recalculated Method

14 Results fro our odified version of the SSK ethod indicate that there is a difference in age estiation. The MLE values reported were closer to the : MLE to chronological age ratio under the new paraeters. Additionally, three changes were apparent when coparing CI band values. First, the odified ethod narrowed the CI band estiates, which soeties excluded the actual age if the age was underestiated. This occurred ore often in the,, and - year-old cohorts, and likely reflects saple size. It is necessary to address this in future research, as too narrow age range estiates can be detriental to forensic investigations, excluding the target individual fro analysis. Second, the odified ethod also calculated CIs for individuals that were not calculated in the original ethod. This iproveent is reflected in Table where an increased nuber of individuals had CI bands for the odified ethod in early cohorts. Lastly, another difference between the two ethods was in the ethod estiation paraeters. The recalculation of the age-at-attainent paraeters refined soe of the values in the SSK ethod, including the age-at-attainent values for earlier developental stages (Cr., Cr., and Cr.c) in the lower peranent incisors (LI and LI). This refineent allowed for the calculation of a between-tooth variance value, which was not calculable under the original SSK paraeters for certain individuals with tooth scores ranging fro to for LI and LI (see Figure a and b). Further, the Cr.c and Ri values were reexained and refined for LI and LI, allowing for further refineent of MLE estiates. In the original SSK ethod, the age-of-attainent values for Cr.c and Ri were the sae for all four peranent incisors. Although distinguishing these two stages is difficult because of their siilarity in expression, the optiization test indicated that stages did not need to be collapsed for these teeth. The optiization test in this study suggested that P and M for feales would benefit fro reevaluation of the scoring stages. We did not

15 investigate the possible collapsing of stages here and note that this ay contribute to inaccurate estiates in age when these teeth are present. Although this study provided valuable results, there are three potential liitations that relate to sapling. First, the Patricia saple is representative of radiographs frequently encountered in forensic casework in the United States; the iages are not standardized and ay not be of the best quality, which can hinder observation and scoring of teeth. The lateral radiographic iages in Patricia were taken at autopsy, where dentition was likely not a priary focus of the iage. Anterior teeth appeared crowded and overlapping in the radiographs, aking the difficult to score. Additionally, it was difficult to distinguish between d, d, and M in very young individuals with early developental scores. Misidentification of teeth could contribute to errors in age estiation. One potential reedy to this issue is to use the plot.teeth function within the SSK ethod package to assess the nored likelihood developent sequence of each tooth. If a particular tooth is not in alignent with the suite of teeth in the graphic, it could suggest isidentification of a tooth, and call for reexaination of the radiograph. However, it is not unusual to find individuals that have accelerated or decelerated growth rates of a particular part of a dental sequence. Shackelford et al. () noticed differential growth on scores for the Roc de Marsal fossil (Bayle et al. ), and three individuals fro Anderson et al. s () saple. In instances such as this, we advise a reexaination of the tooth or teeth in question, but we caution the observer against changing the score purely to fit it within the bounds of the other scores. Finally, this study evaluated the aging through ean age-at-attainent paraeters. Lastly, this study evaluated the age at which individuals transition fro one stage into the next on an aggregate level. In order to understand individual variation within transitions,

16 longitudinal data fro a series of radiographs on the sae individual over soe interval of tie is required. One issue observed in this study was the frequent underestiation of age for M when copared to other teeth within an individual. When reviewing the plots, we noted that M frequently produced age ranges slightly younger than other teeth observed within an individual, particularly those over the age of. This issue will be addressed in future research as M will likely be an iportant assessent in forensic casework because of radiographic liitations and retention in skeletal reains. In casework, the practitioner ay be liited to lateral cranial radiographs rather than dental radiographs, aking M an easily defined and clearly visible landark for scoring enael and root developent. The authors relied heavily on M in this study, which was liited to lateral cranial radiographs, with M being the ost frequently scored tooth (. scored) for the U.S. odern saple, followed by M at.. Second, there is a tendency to lose single-rooted dentition postorte, while the two and three-rooted olars are ore coonly preserved in occlusion. Thus, it will be iportant to accurately estiate age when liited to posterior dentition. A final observation worth noting is that this research suggests possible secular change in dental developent, which Šešelj et al. () report for root developent. This contrasts with Liversidge and Sith s () conclusions that dental developent exhibits insignificant levels of secular change in saples with birth years fro the s to the early s. Application of this ethod to archaeological and undocuented historical saples ay provide slightly inaccurate estiates. Secular change will be an iportant coponent to explore in future studies in order to ake this ethod applicable across anthropological research.

17 Perspectives Our study confired that the odified version of the SSK ethod perfors better when estiating age on odern juveniles, specifically individuals aged between birth and years. Future research will attept to iprove upon age estiation through a larger saple collection that includes ore individuals in their teenage years and individuals younger than years. Additional considerations will exaine the ethod s perforance by sex and ancestry. Further refineent of the early developental ean age-at-attainent values for the incisors and a reassessent of all developental stages that the reference saple failed to cover in this study would be beneficial to test and iprove accuracy in classification. Additional research will focus on exploration into the type, nuber, and cobination of teeth used in age estiation odels. Given that forensic anthropologists are often given radiographs or skeletal cases with issing dentition, assessing the usefulness of specific, anchor teeth in calculating accurate estiates is iportant. Lastly, we hope to iprove the accuracy of this ethod on odern subadults and increase its user-ability in hopes of attracting practitioners to use this reliable age estiation ethod in practice. Acknowledgents The authors would like to thank B Algee-Hewitt, C Hughes and J Ki for the invitation to participate in this special volue of Huan Biology; HM Liversidge and VM Phillips for the use of previously collected data; LW Konigsberg for statistical insight; EA Verlinden for help with saple organization; JT Hefner; A Sionetto; and three anonyous reviewers for insightful coents that strengthened this anuscript.

18 Received February ; revision accepted for publication May.

19 Literature Cited AlQahtani, S. J., M. P. Hector, and H. M. Liversidge.. Brief counication: The London Atlas of huan tooth developent and eruption. A. J. Phys. Anthropol. :. Anderson, D. L., G. W. Thopson, and F. Popovich.. Age of attainent of ineralization stages of the peranent dentition. J. Forensic Sci. :. Bayle, P., J. Braga, A. Mazurier et al.. Brief counication: High resolution assessent of the dental developental pattern and characterization of tooth tissue proportions in the late Upper Paleolithic child fro La Madeleine, France. A. J. Phys. Anthropol. :. Christensen, A. M., and C. M. Crowder.. Evidentiary standards for forensic anthropology. J. Forensic Sci. :,,. Deirjian, A., H. Goldstein, and J. M. Tanner.. A new syste of dental age assessent. Hu. Biol. :. Elain, F., and H. M. Liversidge.. Malnutrition has no effect on the tiing of huan tooth foration. PLoS One :e. Gustafson, G., and G. Koch.. Age estiation up to years of age based on dental developent. Odontol. Revy. :. Harris, E. F.. Mineralization of the andibular third olar: A study of Aerican blacks and whites. A. J. Phys. Anthropol. :. Kahl, B., and C. W. Schwarze.. Updating of the dentition tables of I. Schour and M. Massler of. Fortshr. Kieferorthop. :. Konigsberg, L. W., S. R. Frankenberg, and H. M. Liversidge.. Optial trait scoring for age estiation. A. J. Phys. Anthropol. :.

20 Konigsberg, L. W., N. P. Herrann, D. J. Wescott et al.. Estiation and evidence in forensic anthropology: Age-at-death. J. Forensic Sci. :. Liversidge, H. M.. Tiing of huan third olar foration. Ann. Hu. Biol. :. Liversidge, H. M.. Peranent tooth foration as a ethod of estiating age. In Coparative Dental Morphology, T. Koppe, G. Meyer, K. W. Alt et al. eds. Basel, Switzerland: Karger Publishers,. Liversidge, H. M.. Siilarity in dental aturation in two ethnic groups of London children. Ann. Hu. Biol. :. Liversidge, H. M.. Controversies in age estiation fro developing teeth. Ann. Hu. Biol. :. Liversidge, H. M., and H. Sith.. Nolla s longitudinal dental study revisited. A. J. Phys. Anthropol. :. Milner, G. R., and J. L. Boldsen.. Transition analysis: A validation study with known age odern Aerican skeletons. A. J. Phys. Anthropol. :. Mincer, H. H., E. F. Harris, and H. E. Berryan.. The A.B.F.O. study of the third olar developent and its use as an estiator of chronological age. J. Forensic Sci. :. Moorrees, C. F., E. A. Fanning, and E. E. Hunt. a. Age variation of foration stages for ten peranent teeth. J. Dent. Res. :,,. Moorrees, C. F, E. A. Fanning, and E. E. Hunt. b. Foration and resorption of three deciduous teeth in children. A. J. Phys. Anthropol. :. Ousley, S. D., K. Frazee, and K. E. Stull.. A radiographic database for estiating biological paraeters in odern subadults. Washington, D.C.: National Institute of Justice. Award No.: -DN-BX-K.

21 Phillips, V. M., and T. J. van Wyk Kotze.. Dental age-related tables for children of various ethnic groups in South Africa. J. Forensic Odontostoatol. :. Prieto, J. L., E. Barberia, R. Ortega et al.. Evaluation of chronological age based on third olar developent in the Spanish population. Int. J. Legal Med. :. R Core Tea.. R: A language and environent for statistical coputing. Vienna: R Foundation for Statistical Coputing. Retrieved fro Reppien, K., B. Sejrsen, and N. Lynnerup.. Evaluation of post-orte estiated dental age versus real age: A retrospective -year survey. Forensic. Sci. Int. S S. Scheuer, L., and S. Black.. The Juvenile Skeleton. London: Elsevier Acadeic Press. Schour, L., and M. Massler.. The developent of the huan dentition. J. A. Dent. Assoc. :,,. Šešelj, M., L. W. Konigsberg, and R. J. Sherwood.. Tiing of developent of the peranent andibular dentition: New reference values fro the Fels Longitudinal Study radiographic database. Proceedings fro the th Annual Meeting of the Aerican Association of Physical Anthropology, Austin, TX. Shackelford L. L., A. E. Stinespring Harris, and L. W. Konigsberg.. Estiating distribution of probable age-at-death fro dental reains of iature huan fossils. A. J. Phys. Anthropol. :. Ubelaker, D. H.. Huan Skeletal Reains. Chicago: Aldine. Ubelaker, D. H.. The estiation of age at death fro iature huan bone. In Age Markers in the Huan Skeleton, M. Y. Iscan, ed. Springfield, IL: Charles C Thoas,.

22 Table. Age Structure of the Reference and Test Saple Reference Saple Test Saple Age Cohort South (Year) London Total: Patricia African Totals:

23 Table. Average Differences (by age cohort) between Chronological Age and Age Estiates Age Cohort (years) Original SSK (years) Modified SSK (years) *.* Negative values: underestiation of age. Positive values: overestiation of age. *The -year-old cohort only had one individual and reflects the difference between chronological age and the age estiate.

24 Table. Probability Values fro the Optiization Test Feales Males Tooth (outliers (outliers (all data) (all data) (n) reoved) (n) reoved) noral log noral log noral log noral log LI......E-.. LI C P.....E-... P M.....E-.E-.. M M Bolded values are significant at the p=. level; n=nuber of teeth used.

25 Table. Percent of Individuals That Fall within the Generated Age Range at the,, and CI and Percent of Individuals That Do Not Produce a CI Range Age Cohort (years) Total (n) Original SSK (CI bands) no band Modified SSK (CI bands) no band

26 Suppleent A tooth.test=function () { tooth_ages <- c() for (i in :nrow(tooth.scores)) { # vector output <-tooth.scores[i,] [is.na()]<- if ([,] > ) {h=} else if ([,] > ) {h=} else if ([,] > ) {h=} else h=. if ([,] > ) {h=} # set to ([,] > ) {h=} for the recalculated MFH and MUS atrices else if ([,] > ) {h=} # set to ([,] > ) {h=} for the recalculated MFH and MUS atrices else if ([,] > ) {h=} # set to ([,] > ) {h=} for the recalculated MFH and MUS atrices else h=h if ([,] > ) {h=} else if ([,] > ) {h=} else if ([,] > ) {h=} else h=h if (([,] < ) & ([,] < ) & ([,] < ) & ([,] < )) {h=} else h=h odel <- get.age(i,hi=h,def.int=.) scores_i <- cbind(odel$lab,h,odel$u,odel$within,odel$between,odel$p.seq) # add vector to a datafrae age_i <- data.frae(scores_i) tooth_ages <- rbind(tooth_ages,age_i) } write.table(tooth_ages, file="pat_original_new_tooth_results.csv",sep=",",col.naes=c("lab","hi","u","within","betw een","p.seq"),row.naes=false) } return(data.frae(tooth_ages))

27 Table B. The Recalculated MFH Table Suppleent B x Sex Toot WithoutMea WithMea Stage LSD LSD Mean USD USD h n n M c Cco NA NA NA M c Coc M c Cr/ M c Cr/ M c Crc M c Ri M c R/ M c R/ M c R/ M c Rc M c A/ M c Ac M c Res/ M c Res/ M c Res/ M c Exf M Coc NA NA NA M Cr/ M Cr/ M Crc M Ri M Rcleft M R/ M R/ M R/ M Rc M A/ M Ac M Res/ M Res/

28 M M d M d Res/ Res/ Res/ Res/ M d M d Exf M Cco NA NA NA M Coc M Cr/ M Cr/ M Crc x Sex Toot WithoutMea WithMea Stage LSD LSD Mean USD USD h n n M Ri M Rcleft M R/ M R/ M R/ M Rc M A/ M Ac M M M M d Res/ Res/ Res/ Res/ Res/ Res/ M d..... M d.... M d Exf... F c Coc NA F c Cr/ F c Cr/ F c Crc F c Ri

29 F c R/ F c R/ F c R/ F c Rc F c A/ F c Ac Res/ F c Res/ F c Res/. F c F c Exf..... F Cr/ NA F Cr/ F Crc F Ri F Rcleft F R/ F R/ F R/ F Rc F A/ F Ac F x Sex Toot h F F F d F d Res/ Stage LSD LSD Mean USD USD Res/ Res/ Res/ Res/ Res/ WithoutMea n WithMea n F d..... F d Exf.... F Cr/ F Cr/ F Crc F Ri

30 F Rcleft F R/ F R/ F R/ F Rc F A/ F Ac Res/ F F Res/ F F F F d d d Res/ Res/ Res/ Res/ F d Exf M UI Crc M UI R/ M UI R/ M UI R/ M UI R/ M UI Rc M UI Crc M UI R/ M UI R/ M UI R/ M UI R/.... M UI Rc

31 x M LI Cr/ M LI Cr/ M LI Crc M LI Ri M LI R/ M LI R/ Sex Toot h Stage LSD LSD Mean USD USD WithoutMea n WithMea n M LI R/ M LI R/ M LI Rc M LI A/ M LI Ac M LI Cr/ M LI Cr/ M LI Crc M LI Ri M LI R/ M LI R/ M LI R/ M LI R/ M LI R/ M LI Rc M LI A/

32 M LI Ac M C Ci M C Cco M C Coc M C Cr/ M C Cr/ M C Crc M C Ri M C R/ M C R/ M C R/... M C Rc. M C A/ M C Ac M PM Ci M PM Cco M PM Coc M PM Cr/ M PM Cr/ M PM Crc M PM Ri M PM R/ M PM R/

33 x M PM R/... M PM Rc. M PM A/ Sex Toot h M PM Ac Stage LSD LSD Mean USD USD WithoutMea n WithMea n.. M PM Ci M PM Cco M PM Coc M PM Cr/ M PM Cr/ M PM Crc M PM Ri M PM R/ M PM R/... M PM R/. M PM Rc. M PM A/ M PM Ac.. M M Cco NA M M Coc M M Cr/ M M Cr/ M M Crc

34 M M Ri M M Rcl M M M M M M M M M M M M R/ R/ R/ Rc A/ Ac M Md R/ M Md R/ M Md R/ M Md Rc M Md A/ M Md Ac M M Ci M M Cco M M Coc M M Cr/ M M Cr/ M M Crc M M Ri M M Rcl M M R/

35 x Sex M M M M M Toot h M M M M M Stage LSD LSD Mean USD USD R/.. R/ Rc A/ Ac M Md R/... M Md R/.. M Md R/ M Md Rc M Md A/ M Md Ac WithoutMea n WithMea n M M Ci M M Cco... M M Coc.. M M Cr/. M M Cr/. M M Crc M M Ri M M cleft M M M M M M R/ R/ R/

36 x M M M M M M Rc A/ Ac M Md R/ M Md R/ M Md R/ M Md Rc M Md A/ M Md Ac F UI Crc F UI R/ F UI R/ F UI R/ F UI R/ F UI Rc.... F UI A/ F UI Cr/ F UI Crc F UI R/ F UI R/ Sex Toot h Stage LSD LSD Mean USD USD WithoutMea n WithMea n F UI R/ F UI R/

37 F UI Rc... F UI A/ F LI Cr/ F LI Cr/ F LI Crc F LI Ri F LI R/ F LI R/ F LI R/ F LI Rc F LI A/ F LI Ac F LI Cr/ F LI Cr/ F LI Crc F LI Ri F LI R/ F LI R/ F LI R/ F LI R/ F LI R/ F LI Rc F LI A/

38 x F LI Ac F C Ci F C Cco F C Coc F C Cr/ F C Cr/ F C Crc F C Ri F C R/ F C R/ F C R/.... F C Rc.. F C A/. F C Ac F PM Ci F PM Cco F PM Coc Sex Toot h Stage LSD LSD Mean USD USD WithoutMea n WithMea n F PM Cr/ F PM Cr/ F PM Crc F PM Ri F PM R/

39 F PM R/.... F PM R/... F PM Rc. F PM A/ F PM Ac F PM Ci F PM Cco F PM Coc F PM Cr/ F PM Cr/ F PM Crc F PM Ri F PM R/ F PM R/... F PM R/. F PM Rc F PM A/ F PM Ac F M Cco F M Coc F M Cr/ F M Cr/ F M Crc

40 x F M Ri F M cleft F F F F F F M M M M M M R/ R/ R/ Rc A/ Ac F Md R/ F Md R/ F Md R/ F Md Rc F Md A/ F Md Ac.. Sex Toot h... Stage LSD LSD Mean USD USD.. WithoutMea n WithMea n F M Ci F M Cco F M Coc F M Cr/ F M Cr/ F M Crc F M Ri F M cleft

41 F F F F F F M M M M M M R/... R/.. R/ Rc A/ Ac F Md R/... F Md R/.. F Md R/ F Md Rc F Md A/ F Md Ac F M Ci.... F M Cco... F M Coc.. F M Cr/. F M Cr/. F M Crc F M Ri F M Rcl F F F M M M R/ R/ R/

42 M..... F Rc.. M..... F A/.. M..... F Ac F Md R/ F Md R/ F Md R/ F Md Rc F Md A/ F Md Ac.. Bolded values: recalculated by the authors. Plain text values: fro the Shackelford et al. () publication. When iporting into R, oit the first colun to calculate the withean and withoutean values.

43 Table B. Recalculated Log-Corrected Mean Age-at-Attainent Values (MFH table) for Each Tooth and Developental Stage dc d d UI UI LI LI C P P M M M..... C.i C.co C.oc Cr. Cr Cr.c R.i Cl.i R R R R.c A A.c Res..... Res. Res... Bolded values: recalculated by the authors. Plain text values: fro the Shackelford et al. () publication

44 Table B. Recalculated Mean Log Conception-Corrected Ages in Years for Teeth within Specific Developental Stages dc d d UI UI LI LI C P P M M M C.i C.co C.oc Cr Cr Cr.c R.i Cl.i NA R R R R.c A A.c Res Res Res Bolded values: recalculated by the authors. Plain text values: fro the Shackelford et al. () publication.

45 This table replaced the MUS table.

46 Figure.

47 Figure.

48 Figure A.

49 Figure B.

50 Figure.

51 Figure A.

52 Figure B.

53 Figure Captions Figure. Distribution of the reference saple. Figure. Graphical output fro the get.age function with odifications by the second author. The solid line in the center of the density is the MLE; the dotted lines represent the within and within+between tooth variance. Figures A and B. Graphical representations of age estiates and confidence intervals using the unodified (a) and odified (b) versions of the SSK ethod paraeters. MLE ages are represented by black hatch arks, which are shown within low and high bounds of the,, and CI bands. Figure. Plot of MLEs against known chronological ages for the test saple for both ethods. Dotted line: loess fit of the data under the original SSK ethod. Solid line: loess fit of the data using the odified ethod. Diagonal line: a : line between MLE and chronological age. Figures A and B. Figure a and b: The plot.teeth plots for PAT_ using the original (a) and the odified (b) SSK ethods. The plot.teeth graphic exhibits a ore coplete developental score due to the addition of ages-of-attainent for earlier stages.