Linear morphometrics fail to support strong sexual dimorphism in Uintatherium anceps

This study challenges the long-held belief that the extinct mammal *Uintatherium anceps* exhibited strong sexual dimorphism by demonstrating that linear morphometric analyses of its skulls reveal no significant size or feature differences between sexes, unlike the clear dimorphism found in the modern bison control group.

The Gist

Imagine you're looking at a group of ancient, prehistoric giants called Uintatheres. These creatures lived millions of years ago and looked like a bizarre mix of a rhino, a bear, and a dragon. They had massive bodies, and their skulls were topped with three pairs of bony horns and huge, saber-like teeth.

For over a century, paleontologists have looked at these skulls and said, "Aha! These are clearly two different groups: Big, scary males with giant horns and teeth, and smaller, plain females." They assumed that, just like modern lions or deer, the males were built for fighting and the females were built for hiding.

Kevin Mulcahy, the author of this paper, decided to put that old idea to the test. He didn't just look at the bones with his eyes; he used math and statistics to see if the "Male vs. Female" story actually held up.

Here is the breakdown of his investigation, explained simply:

1. The Detective's Toolkit: Measuring the Evidence

Since we can't ask a 50-million-year-old skull, "Are you a boy or a girl?", Kevin had to use a clever trick. He treated the skulls like a giant puzzle.

• The Method: He took 27 different skulls and measured them with a digital ruler (using photos and software). He measured everything from the length of the nose to the height of the horns.
• The Analogy: Imagine you have a box of 27 mystery balls. Some are painted red, some blue, and some are gray. You don't know which color is "Male" or "Female." You measure the diameter of every ball. If the "Males" are truly different, you should see two distinct piles of sizes: a pile of big balls and a pile of small balls. If they are all the same size, it's just one big pile.

2. The Control Group: The Bison Test

Before trusting his results on the ancient giants, Kevin needed to make sure his math worked. So, he grabbed a box of American Bison skulls (modern-day buffalo).

• Why Bison? Everyone knows male bison are huge and have big horns, while females are smaller. It's a known fact.
• The Result: When Kevin ran his math on the bison, it worked perfectly. The computer instantly sorted the skulls into two clear groups: "Big Males" and "Small Females." The math proved that his method could spot sexual differences when they actually existed.

3. The Big Reveal: The Uintatheres Were Surprisingly Similar

Now, Kevin ran the exact same math on the ancient Uintatheres.

• The Result: The computer failed to find two groups.
• The Analogy: Imagine you sort the mystery balls again. Instead of finding a pile of big ones and a pile of small ones, you find that the sizes are all mixed up. A "supposedly male" skull might be the same size as a "supposedly female" skull. The data looked like a single, smooth hill rather than two separate mountains.
• The Conclusion: The skulls of Uintatheres didn't show the dramatic size difference between sexes that scientists had believed for 100 years. The "Big Male" and "Small Female" story was likely a myth.

4. Why Did Everyone Get It Wrong?

If the math says they weren't that different, why did scientists think they were?

• The "Horn" Illusion: Scientists saw a skull with a tiny horn and assumed it was a female. They saw a skull with a giant horn and assumed it was a male. But Kevin's math showed that horn size varied a lot within the same sex, just like how some human men have big muscles and some have small ones, but they are all still men.
• The "Solitary" Theory: The paper suggests these animals might have lived alone in the forest, not in herds. In nature, animals that live in herds often have huge differences between males and females (because males fight for the herd). Animals that live alone often look more similar to each other.

The Takeaway

This paper is like a reality check for paleontology. It tells us that just because an animal looks weird and scary, it doesn't mean the males and females looked totally different.

Kevin's study suggests that Uintatheres were likely much more similar to each other than we thought. While there might have been some small differences, they weren't the dramatic "Hulk vs. Tiny" split that had been the standard story for over a century.

In short: The ancient giants weren't as "gender-bent" as we thought. They were probably a more uniform bunch, and it's time to rewrite the textbooks to reflect that the "Big Male/Small Female" rule doesn't apply to everyone in the animal kingdom.

Technical Summary

1. Problem Statement

Uintatherium anceps, an extinct giant mammal from the Eocene, has long been regarded as a classic example of extreme sexual dimorphism in the fossil record. Since Othniel Charles Marsh's initial description in 1886, paleontologists have hypothesized that males were significantly larger and possessed more prominent cranial features (horns and saber-like canines) than females. However, this hypothesis has never been formally tested with quantitative statistical methods.

• The Challenge: Fossil specimens lack primary sexual characteristics (e.g., genitalia), making sex assignment impossible without relying on morphological assumptions. This creates a circular logic problem where morphological variation is attributed to sex, and sex is inferred from that same variation.
• The Gap: Previous interpretations relied on qualitative assessments or univariate tests that are prone to high Type I and Type II errors, especially with small sample sizes and unknown sex ratios. There is a need for a rigorous, multi-method statistical framework to test the null hypothesis of monomorphism against the alternative of strong dimorphism.

2. Methodology

The author employed a comprehensive suite of statistical approaches on a dataset comprising nearly all known mature skulls of Uintatherium anceps ($n=27$) from the AMNH, YPM, and FMNH. To validate the analytical pipeline, a parallel dataset of American Bison (Bison bison, $n=23$) was analyzed, as bison are a known, strongly dimorphic extant analog with similar large body size and cranial ornamentation.

Data Collection:

• Metrics: Eight linear cranial measurements were taken from 2D lateral/anterior images using ImageJ. These captured overall skull size and the relative development of purportedly dimorphic traits (e.g., horn height, canine length, occipital crest).
• Constraints: Due to the massive size and fragile reconstruction of the fossils, 3D geometric morphometrics (CT scanning) was not feasible; linear morphometrics were used instead.
• Validation: Bland-Altman plots and Pearson correlations confirmed that measurement error (image-based vs. manual) did not covary with specimen size.

Statistical Approaches:

1. Univariate Analysis:
   • Normality & Unimodality: Shapiro-Wilk tests and Hartigan's Dip tests to check for bimodal distributions (indicative of two sexes).
   • Mixture Modeling: Gaussian mixture models (using BIC) to determine the optimal number of underlying distributions (1 vs. 2 components).
2. Multivariate Analysis:
   • Hierarchical Clustering: Dendrograms generated to see if purported males and females cluster separately.
   • Principal Component Analysis (PCA): To visualize morphospace and check for distinct clusters based on sex.
   • K-means Clustering: Partitioning data into two clusters to test if the resulting groups align with previous sex assignments (using Fisher's Exact Test).
   • Cluster Validation: Gap statistics, silhouette width, and total within sum of squares to determine the optimal number of clusters.
3. Effect Size Statistics (Novel Approach):
   • Adapted from Saitta et al., this method fits Gompertz growth curves to the data. Specimens are assigned to "sexes" based on residuals from the growth curve. The effect size is calculated as the proportional difference between the asymptotic trait values of the two groups. This allows for estimating the magnitude of dimorphism without a priori sex knowledge.

3. Key Contributions

• First Quantitative Test: This study provides the first rigorous, quantitative statistical test of the sexual dimorphism hypothesis in Uintatherium.
• Methodological Framework: It demonstrates a robust, reproducible workflow for testing dimorphism in fossil taxa lacking primary sex characteristics, utilizing a combination of univariate, multivariate, and effect size methods.
• Comparative Validation: By applying the exact same methods to Bison bison (a known dimorphic species), the study validates that the statistical pipeline is capable of detecting strong dimorphism when it exists.
• Phylogenetic Implications: It challenges the assumption that large, herbivorous mammals with elaborate cranial ornaments must be strongly dimorphic, offering insights into the ancestral state of sexual dimorphism in early placental mammals.

4. Results

Uintatherium anceps:

• Univariate Tests: None of the eight cranial metrics deviated significantly from normality or unimodality, except for canine length (which showed non-unimodality but lacked bimodal separation).
• Mixture Modeling: The data for 7 of 8 metrics was best fit by a single-component model (monomorphism).
• Clustering (Hierarchical & K-means): Purported males and females (based on previous literature) were distributed randomly across dendrograms and PCA plots. They did not form distinct clusters. Fisher's Exact Test showed no correlation between previous sex assignments and cluster membership ($p=1.0$).
• Effect Size: While traits like canine length and maxillary horn height showed slightly higher effect sizes than body size metrics, the confidence intervals overlapped significantly. Crucially, the residual-based sex assignments did not align with the traditional interpretations of which specimens were male or female.

Bison bison (Control):

• Univariate Tests: Four of eight metrics showed significant deviation from normality/unimodality.
• Clustering: Hierarchical and K-means clustering successfully separated males and females into distinct groups (Fisher's Exact Test $p=0.029$).
• Effect Size: Distinct separation between sexes was observed, with non-overlapping prediction intervals for key traits, confirming the method's ability to detect known dimorphism.

Conclusion: The statistical signal of dimorphism in Uintatherium is indistinguishable from monomorphism or very weak dimorphism, whereas Bison shows a clear, strong signal.

5. Significance

• Refutation of "Extreme" Dimorphism: The study refutes the classical interpretation that Uintatherium anceps exhibits "extreme" sexual dimorphism. While modest variation may exist, the evidence does not support the hypothesis that the species was highly dimorphic.
• Re-evaluation of Fossil Interpretations: It suggests that previous attributions of sex based on skull size and horn prominence were likely incorrect or influenced by collection bias (e.g., favoring larger specimens).
• Paleoecological Implications: The lack of strong dimorphism, combined with the absence of mass death assemblages, suggests Uintatherium may have been solitary rather than living in polygynous herds, challenging previous ecological reconstructions.
• Evolutionary Context: Given the basal phylogenetic position of Dinocerata, these results support the hypothesis that strong male-biased sexual dimorphism was not an ancestral trait for all large mammals, but rather evolved later in specific lineages (like Artiodactyla).
• Future Directions: The paper advocates for the use of linear morphometrics combined with effect size statistics as a standard, reproducible tool for investigating sexual dimorphism in other fossil clades where 3D scanning is not possible.