Co-opting evo-devo concepts for new insights into mechanisms of behavioural diversity

Abstract

We propose that insights from the field of evolutionary developmental biology (or 'evo-devo') provide a framework for an integrated understanding of the origins of behavioural diversity and its underlying mechanisms. Towards that goal, in this Commentary, we frame key questions in behavioural evolution in terms of molecular, cellular and network-level properties with a focus on the nervous system. In this way, we highlight how mechanistic properties central to evo-devo analyses - such as weak linkage, versatility, exploratory mechanisms, criticality, degeneracy, redundancy and modularity - affect neural circuit function and hence the range of behavioural variation that can be filtered by selection. We outline why comparative studies of molecular and neural systems throughout ontogeny will provide novel insights into diversity in neural circuits and behaviour.

Keywords: Canalization; Modularity; Neural circuits; Neuroethology; Plasticity; Robustness.

Fig. 1.

Evolution of behavioural phenotypes is shaped by developmental mechanisms. Which suites of behavioural phenotypes are likely or even possible in a population depends, in part, on how genetic variants alter developmental processes. Mutations can impact behaviour via effects on diverse molecular and cellular networks throughout development. Different cellular or molecular mechanisms are depicted as layers in the diagram, with only two of the many cellular or molecular network states during development shown. Vertical arrows in the diagram mark developmental pathways for given genotypes (A or B); the abiotic environment, biotic environment and learning affect these pathways to mediate developmental plasticity in behaviour. Differently coloured vertical arrows, which represent any environmental differences, reflect the alternative developmental trajectories in environments X and Y that emerge from accumulating effects of experience on cellular and molecular networks. Genotype A adopts different probable states (indicated by numbers at each level of phenotype) in the two environments owing to developmental plasticity, potentially leading to two distinct behavioural phenotypes. By contrast, environmental influences on development of genotype B are buffered, leading to the same probable states during development and a single behavioural phenotype. The robustness of developmental processes means that many alleles might have no phenotypic consequences if pathways converge onto the same behavioural outcome (as in environment X, where both genotypes exhibit behavioural phenotype 3). Genetic variants with consequences can produce alternative behavioural phenotypes in a particular environment owing to the properties of exploratory mechanisms, versatility and weak linkage that characterize developmental processes. Arrows between the possible behavioural phenotypes depict likely transitions between behavioural states that can be brought about by mutations affecting developmental processes. In this way, developmental mechanisms can bias the behavioural variants present in a population. These biases thus shape the evolutionary consequences of natural selection and genetic drift (the dotted filter shown at the top of the figure) and hence the range of realized behavioural phenotypes present in a population. Figure adapted from Oster and Alberch (1982).

Fig. 2.

Waddington's landscape (adapted from Waddington, 1957) illustrates how phenotypic plasticity and genetic divergence shift robust developmental processes. Time moves from top to bottom of each figure panel, and deep valleys indicate possible developmental trajectories. Circles represent phenotypes of individuals, with colour indicating genotype. Positions of the large circles after development portray adult phenotypes. (A) Canalization: selection for robustness has shaped developmental processes to produce similar phenotypes despite developmental noise. Typical noise levels encountered by the species at two time points are illustrated by curved arrows. Noise is buffered such that organisms stay on the same trajectory despite small deviations. (B) Developmental plasticity: predictable changes in developmental outcomes as a function of rearing environment and learning require overcoming the buffering systems inherent in development. Exposure to distinct environments or experiences at different times in ontogeny shifts developmental processes reliably onto either the right or left trajectories. These shifts require moving the organism outside the buffering capacity indicated in A (across the ridge in the figure). (C) Genetic divergence: trajectories represent three genotypes developing in a common environment. The effects of genetic variants may be buffered by developmental processes. For example, the mutations distinguishing the red and purple genotypes may reliably shift aspects of development without altering the adult phenotype. In contrast, some genetic variants, such as the blue genotype, reliably alter the phenotype by shifting developmental trajectories.

Fig. 3.

Analogies between transcriptional and cellular processes that confer robustness and plasticity in network output during evolution. (A) Systems such as transcriptional and neural networks can produce similar network output from numerous system configurations. We depict here examples of systems-level evolution and the consequences for outputs (depicted by shape) in a generic sense, as such changes occur at all hierarchical levels of biological organization. Many components may shift during evolution without changing system output, as in derived system 1. Some component shifts might cause quantitative changes in the output, as in derived system 2. Adding or removing a new input to the system may cause qualitative changes in system output, as depicted for system 3. (B) Evolved changes in neural populations can produce similar network output via distinct mechanisms. The ancestral neural circuit involves a presynaptic neuron releasing the green neurotransmitter. When the presynaptic neurons increase in number in derived lineage 1, network output is largely similar because developmental processes reduce the number of synapses from each presynaptic neuron. A novel input from another neural population in derived lineage 2 (purple neurotransmitter) influences network output because the postsynaptic neuron has a receptor that can bind the new neurotransmitter. (C) Evolved changes in regulatory networks can produce similar network output via distinct mechanisms. The ancestral condition has an enhancer element that binds the blue transcription factor. In the evolutionary history of derived lineage 1, the enhancer element mutated, but the new enhancer still binds the same transcription factor moderately well. In derived lineage 2, a novel enhancer site arises by mutation and increases transcription by binding the blue transcription factor. Derived lineage 3 also evolves higher transcription levels, but this time the mutation causes novel regulation by the yellow transcription factor.