Modularity patterns in mammalian domestication: Assessing developmental hypotheses for diversification

Abstract

The neural crest hypothesis posits that selection for tameness resulted in mild alterations to neural crest cells during embryonic development, which directly or indirectly caused the appearance of traits associated with the "domestication syndrome" (DS). Although representing an appealing unitary explanation for the generation of domestic phenotypes, support for this hypothesis from morphological data and for the validity of the DS remains a topic of debate. This study used the frameworks of morphological integration and modularity to assess patterns that concern the embryonic origin of the skull and issues around the neural crest hypothesis. Geometric morphometric landmarks were used to quantify cranial trait interactions between six pairs of wild and domestic mammals, comprising representatives that express between five and 17 of the traits included in the DS, and examples from each of the pathways by which animals entered into relationships with humans. We predicted the presence of neural crest vs mesoderm modular structure to the cranium, and that elements in the neural crest module would show lower magnitudes of integration and higher disparity in domestic forms compared to wild forms. Our findings support modular structuring based on tissue origin (neural crest, mesoderm) modules, along with low module integration magnitudes for neural crest cell derived cranial elements, suggesting differential capacity for evolutionary response among those elements. Covariation between the neural crest and mesoderm modules accounted for major components of shape variation for most domestic/wild pairs. Contra to our predictions, however, we find domesticates share similar integration magnitudes to their wild progenitors, indicating that higher disparity in domesticates is not associated with magnitude changes to integration among either neural crest or mesoderm derived elements. Differences in integration magnitude among neural crest and mesoderm elements across species suggest that developmental evolution preserves a framework that promotes flexibility under the selection regimes of domestication.

Keywords: Evolvability; morphological disparity; selection; shape variation; skull; tameness.

Figure 1

Schematic illustration of the hypothesis framework implemented in this study. (A) Domesticated animals share, to a varying degree, a suite of traits. These traits are hypothesized to have been caused by disruption in neural crest development (the neural crest hypothesis; Wilkins et al. 2014). (B) The neural crest is responsible for the patterning of some cranial bones, studied here using geometric morphometric landmarks collected on domestic/wild mammals, as seen in dorsal (top), ventral (middle), and lateral (bottom) views. Landmarks were subset into two modules, Neural Crest (NC, blue) and Mesoderm (MD, red), based on tissue origin of cranial elements, after Mischina and Snider (2014). (C) The neural crest hypothesis was evaluated using measures of modularity and integration, under the following predictions: i) module disparity will be greater for domestic versus wild forms, and module disparity for the NC module will be greater than that for the MD module, ii) within‐module integration magnitudes will be lower for domestic forms versus wild forms, and integration magnitudes will be lower for the NC module compared to the MD module, and iii) measures of between module integration will be lower for domestic forms compared to wild forms. Cranial illustrations modified after Balcarcel et al. (2021). Not all landmarks illustrated could be recorded for all species due to species‐specific cranial morphology. Landmark descriptions are provided in Table S1.

Figure 2

Boxplots of morphological disparity for cranial landmarks assigned to the Neural Crest (NC) (blue) and Mesoderm (MD) (red) modules, displaying values for six domestic/wild mammal pairs (A‐B, C‐D, E‐F, G‐H, I‐J, K‐L). Disparity was measured as Procrustes Variance and was corrected by the number of landmarks in each module, to enable direct comparison. Disparity was calculated on Procrustes superimposed landmarks (graphed) and residuals (Table S16).

Figure 3

Boxplots of (A) morphological disparity and (B) within‐module integration pooled for cranial landmarks assigned to the Neural Crest (NC) (blue) and Mesoderm (MD) (red) modules. Values are pooled for six wild and domestic mammal pairs. Procrustes Variance and Eigenvalue dispersion values were corrected for the number of landmarks in each module, to enable direct comparison. Disparity and Integration values were calculated using Procrustes superimposed landmarks (graphed) and residuals (Table S16 and Table S18).

Figure 4

Scatterplots of the relationship between (A) morphological disparity (Procrustes Variance) and between‐module integration (z‐scores for between module PLS), and (B) morphological disparity (Procrustes Variance) and within‐module integration (Eigenvalue dispersion). Values calculated from cranial landmarks assigned to the Neural Crest (NC) (blue) and Mesoderm (MD) (red) modules for six domestic/wild pairs of mammals. Ordinary least squares regression lines show fit for the pooled (MD and NC) relationships for domestic (solid line) and wild (dashed line) forms. Filled ellipses are the domestic form and open ellipses are the wild form. Disparity and Integration values were calculated using Procrustes superimposed landmarks (graphed) and allometry‐corrected residuals (Table S16 and Table S18).