Weak genetic signal for phenotypic integration implicates developmental processes as major regulators of trait covariation

Abstract

Phenotypic integration is an important metric that describes the degree of covariation among traits in a population, and is hypothesized to arise due to selection for shared functional processes. Our ability to identify the genetic and/or developmental underpinnings of integration is marred by temporally overlapping cell-, tissue- and structure-level processes that serve to continually 'overwrite' the structure of covariation among traits through ontogeny. Here, we examine whether traits that are integrated at the phenotypic level also exhibit a shared genetic basis (e.g. pleiotropy). We micro-CT scanned two hard tissue traits, and two soft tissue traits (mandible, pectoral girdle, atrium and ventricle, respectively) from an F₅ hybrid population of Lake Malawi cichlids, and used geometric morphometrics to extract 3D shape information from each trait. Given the large degree of asymmetric variation that may reflect developmental instability, we separated symmetric from asymmetric components of shape variation. We then performed quantitative trait loci (QTL) analysis to determine the degree of genetic overlap between shapes. While we found ubiquitous associations among traits at the phenotypic level, except for a handful of notable exceptions, our QTL analysis revealed few overlapping genetic regions. Taken together, this indicates developmental interactions can play a large role in determining the degree of phenotypic integration among traits, and likely obfuscate the genotype to phenotype map, limiting our ability to gain a comprehensive picture of the genetic contributors responsible for phenotypic divergence.

Keywords: bone; development; geometric morphometrics; heart; phenotypic integration.

Figure 1.

The phenotype is shape by a multitude of genetic and developmental processes. a, Few interactions among developmental processes impact the eventual phenotype leading to a clear picture of how genotype patterns phenotype. b, Developmental interactions occurring among tissues obscure our ability to fully trace the underlying genetic basis resulting in a phenotype that mostly reflects tissue-level processes. c, Developmental interactions occurring among structures can obscure our ability to understand both the genetic basis for a given trait, and the cellular and tissue level processes that may have impacted the shape of each structure. Gray arrows refer to interactions between levels that will go undetected due to the hierarchical nature of developmental processes.

Figure 2.

Positioning of the mandible, pectoral girdle, atrium, and ventricle in the cichlid head. a, Schematic demonstrating the full compliment of muscles and bones in the head. b, Schematic of the cichlid head with the suspensorium removed to clearly reveal the heart, mandible, and pectoral girdle. c, μCT scan of the head region including models of the heart reconstructed from contrast enhanced images. MAN, mandible (dark green); PGL, pectoral girdle (dark blue); HRT, heart, Atr, atrium (orange); Ven, ventricle (red); Den, dentary; Art, articular; P-T, post-temporal; SC, supra-cleitrum; Cle, cleithrum; Sca, scapula; Cor, coronoid; Rad, radials; FRs, fin rays.

Figure 3.

Geometric morphometric schematics for the placement of all landmarks. a, mandible; b, pectoral girdle; c, ventricle; d, atrium. Fixed landmarks, red circles; semi-landmark curves, blue circles; patch landmarks, yellow circles.

Figure 4.

Symmetric configuration principal component morphospaces for all hybrid structures. A, atrium; b, ventricle; c, mandible; d, pectoral girdle. Hybrids, black circles; LF, red circles; TRC, blue circles.

Figure 5.

Two-block partial least squares analysis to assess association between ventricle and atrium for each landmark configuration. a, original landmark configuration; b, symmetric component of shape variation; c, asymmetric components of shape variation. Hybrids, black circles; LF, red circles; TRC, blue circles.

Figure 6.

Two-block partial least squares analysis to assess association between heart and bony tissues using the symmetric component of shape variation. A, association between atrium and mandible; b, association between ventricle and mandible; c, association between atrium and the pectoral girdle; d, association between ventricle and the pectoral girdle. Hybrids, black circles; LF, red circles; TRC, blue circles.

Figure 7.

QTL map for all structures and all components of variation. Lines reflect significant QTL regions at the 95% level for the original configurations (dark gray), symmetric components of variation (red), and asymmetric components of variation (blue). Letters on top of the lines indicate structure identity: Atrium (A), ventricle (V), mandible (M), pectoral girdle (P).

Figure 8.

Fine maps depicting significant QTL markers across a linkage group and the degree of overlap among those QTL markers. Each line reflects a significant marker, and the opacity of a line indicates significance with solid lines representing highly significant markers, while translucent lines reflect less significant markers. Black lines reflect overlapping markers between traits with their opacity determined by their shared significance values. a, comparing the symmetric component of variation between ventricle and mandible. b, comparing the symmetric component of variation between atrium and pectoral girdle. c, comparing the asymmetric component of variation between pectoral girdle and mandible.