Sexual Size Dimorphism Correlates With the Number of Androgen Response Elements in Mammals, But Only in Small-Bodied Species

Abstract

Sexual size dimorphism is common throughout the animal kingdom, but its evolution and development remain difficult to explain given most of the genome is shared between males and females. Sex-biased regulation of genes via sex hormone signaling offers an intuitive mechanism by which males and females could develop different body sizes. One prediction of this hypothesis is that the magnitude of sexual size dimorphism scales with the number of androgen response elements or estrogen response elements, the DNA motifs to which sex hormone receptors bind. Here, we test this hypothesis using 268 mammalian species with full genome assemblies and annotations. We find that in the two smallest-bodied lineages (Chiroptera and Rodentia), sexual size dimorphism increases (male-larger) as the number of androgen response elements in a genome increases. In fact, myomorph rodents-which are especially small-bodied with high sexual size dimorphism-show an explosion of androgen receptor elements in their genomes. In contrast, the three large-bodied lineages (orders Carnivora, Cetartiodactyla, and Primates) do not show this relationship, instead following Rensch's Rule, or the observation that sexual size dimorphism increases with overall body size. One hypothesis to unify these observations is that small-bodied organisms like bats and rodents tend to reach peak reproductive fitness quickly and are more reliant on hormonal signaling to achieve sexual size dimorphism over relatively short time periods. Our study uncovers a previously unappreciated relationship between sexual size dimorphism, body size, and hormone signaling that likely varies in ways related to life history.

Keywords: androgen signaling; evolution; genetics; hormone response elements; sexual size dimorphism.

Fig. 1.

The correlations between SSD and total ARE counts among orders. ARE counts 50 kb upstream/downstream of a TSS, natural-log-transformed. Each point on the plot is a species, and the slope of the line is derived from phylogenetically controlled linear models (see methods).

Fig. 2.

The five orders represented by at least 20 species (number in parentheses). Relationship between average body mass, SSD, ARE counts, and ERE counts. For SSD, the vertical dashed line indicates SSD = 0, species to the right are male-larger species to the left are female-larger. For all other phenotypes, the vertical dashed lines indicate phylogenetic mean estimated at the root of the mammalian tree. Rodentia has been divided into two groups for visualization: the second row of orange dots are myomorph rodents, showing their dramatic increase in AREs near genes. Small triangles indicate the respective Rodentia group's means. ARE and ERE counts 50 kb upstream/downstream of a TSS, natural-log-transformed. The most recent common ancestor in this tree occurred 72 million years ago. The most recent common ancestor for each order occurred between 49 and 57 million years ago.

Fig. 3.

Gene-centric analyses of the linear model SSD ∼ log(HRE) + log(body_size). Each point is a gene represented by at least ten rodent species that show variability in HRE counts within 50 kb of TSS. The x-axis indicates the effect of HRE count on SSD. The y-axis indicates the −log₁₀ × P-value. For AREs, a large majority of genes, regardless of statistical significance, have a positive influence on SSD. This pattern contrasts with EREs, where the majority of genes have a negative impact on SSD.