Sexual Dimorphism and Biomechanical Loading in Occipital Bone Morphological Variation

bioarchaeology

skeletal biology

sexual dimorphism

biomechanics

ANOVA

discriminant function analysis

Author

Kara C. Hoover

Published

February 2026

Executive Summary

Problem: Occipital bone research has been biased toward evolutionary and forensic questions, often with unexplained variance. Whether behavioral differences – specifically gender-based divisions of daily activity and variation in subsistence workload – account for some of that unexplained variation is poorly understood. Distinguishing developmentally fixed traits from those subject to adult remodeling is essential for interpreting occipital morphology in both bioarchaeological and forensic contexts.

Approach: I examined metric and nonmetric traits of the occipital bone across nine skeletal collections (n=663) representing hunter-gatherers, horticulturalists, and agriculturalists spanning Florida Archaic through 20th-century modern study collections. Two datasets were analyzed – an original Florida collection (n=98) and a mined dataset of published Plains Arikara and study collection data (n=565) – combined where variables overlapped. Two-way Type II ANOVA tested sex and subsistence as predictors for five outcome variables; discriminant function analysis evaluated foramen magnum area as a sex estimation tool.

Insights: All five outcome variables showed significant sex-based differences. Bicondylar breadth and lambda-inion distance also varied by subsistence practice, suggesting a biomechanical signal in traits subject to adult remodeling. Foramen magnum area, fixed developmentally by age 7, showed sexual dimorphism but no subsistence signal – consistent with its canalized developmental trajectory. Discriminant function analysis using foramen magnum area achieved only 71% sex estimation accuracy, insufficient for forensic use; adding bicondylar breadth as a second predictor reduced rather than improved accuracy, indicating that biomechanical loading introduces noise into condylar metrics when used for sex estimation.

Significance: These findings demonstrate that occipital bone traits show consistent sexual dimorphism but limited standalone forensic utility. Subsistence-based biomechanical activity contributes to variation in condylar and nuchal metrics, complicating their use as pure sex indicators. Developmentally fixed and remodeling-prone traits behave differently and should be interpreted accordingly. Larger datasets with known individual occupation data are needed to fully parse sex-based from activity-based variation in the occipital bone.

Key Findings

All five outcome variables showed significant sexual dimorphism
Bicondylar breadth and lambda-inion distance varied significantly by subsistence practice; foramen magnum area, occipital form index, and nuchal index did not
Horticulturalists and hunter-gatherers both exceeded agriculturalists on bicondylar breadth; the horticulturalist vs. hunter-gatherer difference was negligible (Tukey HSD p=0.95)
Foramen magnum area achieved 71% sex estimation accuracy – above chance but below forensic standards
Adding bicondylar breadth as a second DFA predictor reduced accuracy to 66%, reflecting biomechanical loading as noise in condylar sex estimation

Publication

Hoover, K.C., & Thomas, G.P. (2022). Sexual dimorphism and biomechanical loading in occipital bone morphological variation. American Journal of Human Biology, e23792. DOI: 10.1002/ajhb.23792

Research Questions

Does foramen magnum area vary by biological sex and/or subsistence practice?
Does bicondylar breadth vary by sex and/or subsistence?
Do ectocranial traits (lambda-inion distance, occipital form index, nuchal index) vary by sex and/or subsistence?
How accurately can foramen magnum area predict biological sex, alone and combined with bicondylar breadth?

Research Answers

Developmental Timing Determines Biomechanical Sensitivity

Traits fixed early in development and under stronger genetic control (foramen magnum length and breadth) behave differently from traits influenced by muscle use and adult remodeling (bicondylar breadth, lambda-inion distance, external occipital protuberance depth, nuchal crest presence, nuchal line count, general EOP form). This distinction structured both the predictions and the results. Full two-way ANOVA results for all five outcome variables are in the table below; individual results are discussed in the subsections that follow.

Variable	Dataset	Predictor	F	p	Sig.
Foramen Magnum Area	Original	Subsistence	0.00	.952
		Sex	9.27	.003	**
Foramen Magnum Area	Combined*	Subsistence	2.41	.090	.
		Sex	129.26	<.002	***
Bicondylar Breadth	Original	Subsistence	9.93	.003	**
		Sex	14.12	<.002	***
Bicondylar Breadth	Combined	Subsistence	19.6	<.002	***
		Sex	70.9	<.002	***
Lambda-Inion	Original	Subsistence	5.95	.017	*
		Sex	11.93	.001	***
Occipital Form Index	Original*	Subsistence	0.18	.670
		Sex	29.66	<.002	***
Nuchal Index	Original	Subsistence	0.73	.390
		Sex	23.18	<.002	***

Type II SS two-way ANOVA results. * white.adjust = TRUE used for heteroscedastic models. Significance: . p < .10 | * p < .05 | ** p < .01 | *** p < .001

Foramen Magnum Area: Sex Only

The basicranium is fully developed by age 7 and the foramen magnum does not remodel in adulthood, so any sex-based variation in this region reflects early developmental trajectory rather than biomechanical activity. Foramen magnum area was significantly different between the sexes in both datasets (Original: F=9.27, p=.003; Combined: F=129.26, p<.002), with males averaging approximately 10 cm² larger in the combined dataset. No subsistence effect emerged in either dataset (Original: F=0.00, p=.952; Combined: F=2.41, p=.09), consistent with developmental canalization. Panel A shows the sex difference clearly; panel B shows the absence of subsistence-based separation.

Figure 1A–B. Foramen magnum area by sex (A) and subsistence (B). Significant sex dimorphism; no subsistence effect.

Interpretation: Foramen magnum area is sexually dimorphic but subsistence-blind – exactly what developmental canalization predicts. Its utility in sex estimation is real but limited; the overlap between distributions means it cannot classify sex reliably on its own.

The bivariate relationship between foramen magnum sagittal (FMS) and transverse (FMT) dimensions – the raw measurements combined into the area index – shows the same pattern. Despite statistically significant dimorphism, the D statistic of 0.92 by sex reflects substantial distributional overlap. By subsistence the overlap is even greater (D=0.37), with the three groups nearly indistinguishable.

Figure 2. Foramen magnum dimensions (FMS × FMT) by sex. D=0.92; substantial overlap despite significant dimorphism.

Interpretation: The marginal density plots make the overlap visible in a way that mean differences alone do not. Even the largest standardized difference in this dataset (D=0.92 by sex) leaves the distributions substantially overlapping – the core problem for forensic sex estimation from this trait.

Figure 3. Foramen magnum dimensions (FMS × FMT) by subsistence. D=0.37; groups are largely indistinguishable.

Interpretation: Three subsistence groups with very different activity profiles produce nearly identical foramen magnum dimension distributions. This is the clearest evidence that this trait is not responding to biomechanical loading.

Bicondylar Breadth: Sex and Subsistence Signal

Unlike the foramen magnum, the occipital condyles articulate with the first cervical vertebra and may remodel under biomechanical pressure from head rotation, flexion, and extension across the lifespan. Bicondylar breadth was significantly different by both sex and subsistence in both datasets (see ANOVA table above). Sex effects were robust (Original: F=14.12, p<.002; Combined: F=70.9, p<.002). Subsistence effects were also significant (Original: F=9.93, p=.003; Combined: F=19.6, p<.002). Post-hoc Tukey HSD on the combined dataset confirmed horticulturalists and hunter-gatherers both exceeded agriculturalists in condylar breadth; the horticulturalist vs. hunter-gatherer difference was negligible (p=.95), suggesting a threshold effect rather than a linear gradient across subsistence intensity.

Figure 4A–B. Bicondylar breadth by sex (A) and subsistence (B). Significant effects for both predictors.

Interpretation: The subsistence signal in bicondylar breadth – absent in foramen magnum area – is consistent with biomechanical remodeling of the condylar region. Populations with more physically demanding and gender-differentiated subsistence strategies show broader condylar breadths, suggesting activity-based variation alongside sexual dimorphism.

Lambda-Inion Distance: Sex and Subsistence Signal

Lambda-inion distance measures the length of the nuchal plane from the meeting of the sagittal and lambdoid sutures to the external occipital protuberance. This region serves as the attachment site for the nuchal ligament, which supports the weight of the head and may develop in response to sustained physical activity. Males averaged approximately 4.8 mm longer than females (F=11.93, p=.001). Controlling for sex, subsistence was also significant (F=5.95, p=.017), with hunter-gatherers averaging approximately 3.5 mm longer than agriculturalists. Lambda-inion distance may partly reflect EOP morphology – more robust protuberances alter the nuchal plane – but the subsistence signal is consistent with the biomechanical interpretation.

Figure 5A–B. Lambda-inion distance by sex (A) and subsistence (B). Plot p-values are unadjusted one-way comparisons; see ANOVA table above for two-way model result.

Interpretation: Lambda-inion distance is the only ectocranial metric that shows both a sex effect and a subsistence effect, making it the strongest candidate for activity-based variation in the posterior occipital region. The hunter-gatherer advantage is consistent with more physically intensive and biomechanically demanding activity patterns.

Occipital Form Index and Nuchal Index: Sex Only

The ectocranial posterior of the occipital bone – the region of heavy buttressing and muscle attachment for neck and head movement – showed sexual dimorphism across both composite indexes but no subsistence-based variation. The Occipital Form Index (EOP depth + general form) was significantly higher in males (F=29.66, p<.002), reflecting deeper protuberances. The Nuchal Index (nuchal line count + nuchal crest presence) was also significantly higher in males (F=23.18, p<.002), reflecting greater muscle marking. Neither index differed significantly by subsistence (Form Index: p=.670; Nuchal Index: p=.390).

Figure 6A–B. Occipital form index by sex (A) and subsistence (B). Sex effect only.

Interpretation: The absence of a subsistence signal in the occipital form index is somewhat unexpected given the attachment of neck muscles used in physically demanding activities. The result may reflect the composite nature of the index – combining a continuous metric (EOP depth) with a nominal trait (general form) may obscure subsistence-related variation in either component.

Figure 7A–B. Nuchal index by sex (A) and subsistence (B). Sex effect only.

Interpretation: Nuchal muscle marking is clearly sexually dimorphic but insensitive to subsistence practice in these datasets. The result is consistent with the form index finding and may reflect the broad occupational heterogeneity within the agricultural group – modern study collections and 19th-century military remains represent diverse activity profiles that may wash out any subsistence signal.

Discriminant Function Analysis

Given the significant sexual dimorphism in foramen magnum area, we tested its utility for sex estimation using discriminant function analysis on the combined dataset (n=663). Data were scaled and split 70/30 into training and test sets. Foramen magnum area alone predicted biological sex at 71% accuracy (LDA coefficient=1.108) – above chance but well below the threshold for forensic utility. Adding bicondylar breadth as a second predictor reduced accuracy to 66%, with coefficients of 0.89 and 0.41 respectively. The decline in accuracy when adding bicondylar breadth is consistent with biomechanical loading introducing noise into condylar-based sex estimation – the subsistence signal in bicondylar breadth that makes it biologically interesting is precisely what undermines its forensic utility.

Study Design

Data Source: Two skeletal datasets collected and mined by the authors. Original dataset (n=98): Florida skeletal collections including Windover (8BR246), Hutchinson Island (8MT37), FSU Teaching Collection, and St. Mark’s Military (8WA108). Mined dataset (n=565): published thesis data on Plains Arikara (Larson 39WW2, Leavenworth 39CO9, Mobridge 39WW1) and large modern study collections (Hamann-Todd, Terry Collection).

Data Handling: Three measurement trials for five metric traits using an electronic sliding caliper; two observation trials for three nonmetric traits. Intra-observer agreement confirmed via kappa (irr). Measurement error assessed via Levene’s test across trials (all <0.51% of trait variance). Foramen magnum length and breadth combined as an ellipse area index (pi x L x B). PCA on five ectocranial variables identified three vector clusters supporting two composite indexes: Occipital Form Index (EOP depth + general form) and Nuchal Index (nuchal crest + nuchal line count). Datasets combined where variables overlapped (foramen magnum area and bicondylar breadth) to create combined dataset (n=663).

Site / Collection	Dataset	Subsistence	Date Range	n	F	M
Windover (8BR246)	Original	Hunter-Gatherer	8522–7421 BP	47	24	23
Hutchinson Island (8MT37)	Original	Hunter-Gatherer	2000–1000 BP	16	8	8
FSU Teaching Collection	Original	Agriculture	20th century	22	11	11
St. Mark’s Military (8WA108)	Original	Agriculture	19th century	13	0	13
Larson (39WW2)	Mined	Horticultural	1679–1733	115	57	58
Leavenworth (39CO9)	Mined	Horticultural	1802–1832	30	14	16
Mobridge (39WW1)	Mined	Horticultural	1600–1700	56	31	25
Hamann-Todd Collection	Mined	Agriculture	1912–1938	107	54	53
Terry Collection	Mined	Agriculture	1899–1941	257	129	128

Table 1. Sample composition by site, dataset, subsistence category, and sex.

Analytical Approach:

Intraobserver error – kappa2 and Spearman correlation for nonmetric traits; Levene’s test across measurement trials for metric traits
PCA on five ectocranial variables to identify variable clusters and support index construction
Assumption testing – normality (QQ plots), homogeneity of variance (Levene’s test), outliers (Rosner test) for original and combined datasets
Two-way ANOVA (Type II SS, car::Anova) with sex and subsistence as predictors, preceded by interaction testing (Type III); no interactions detected; heteroscedasticity-corrected models (white.adjust=TRUE) used for Index_Form (original) and Index_FM (combined)
Post-hoc Tukey HSD for combined dataset subsistence comparisons on bicondylar breadth and foramen magnum area
Discriminant function analysis (MASS::lda) for sex estimation using foramen magnum area alone and with bicondylar breadth as a second predictor; 70/30 train/test split

Project Resources

Repository: github.com/kchoover14/subsistence-impacts-occipital-biomechanics

Data: Original skeletal data collected by the authors from Florida collections; mined data from published theses on Plains Arikara and modern study collections. All data files included in repository.

data-occ-original-raw-22dec21.csv – raw original Florida dataset (3 measurement trials)
data-occ-westcott-raw.csv, data-occ-williams-raw.csv – raw mined thesis datasets
data-occ-mined.csv – cleaned mined dataset
revised-preparation-original.csv, revised-preparation-mined.csv, revised-preparation-combined.csv – analysis-ready datasets

Code:

revised-preparation-original.R – wrangling, reliability testing, measurement error, PCA, index creation for original Florida dataset
revised-preparation-mined.R – standardization and merging of mined Arikara and study collection data
revised-exploratory.R – assumption testing (normality, Levene’s, outliers) for original and combined datasets
revised-analysis.R – two-way ANOVA, Tukey HSD, discriminant function analysis, and all figures

Project Artifacts:

ANOVA figures (n=5): revised-plot-fm.png, revised-plot-bcb.png, revised-plot-li.png, revised-plot-form.png, revised-plot-nuchal.png
FM dimension scatter figures (n=2): revised-scatter-fm-sex.png, revised-scatter-fm-subsistence.png
PCA biplot (n=1): revised-pca-biplot.png

Environment:

renv.lock and renv/ – restore with renv::restore()

License:

Tools & Technologies

Languages: R

Tools: None

Expertise

Transferable Expertise: Combining original data collection with published data mining to expand analytical scope while maintaining methodological rigor; applying multiple statistical frameworks to distinguish genetic from behavioral sources of skeletal variation; evaluating the forensic utility and limitations of morphological metrics.