1. Automated analysis of naturalistic social behaviors in cichlid fishes
We develop automated approaches to automatically record and analyze behaviors of Lake Malawi cichlids. Cichlid fishes offer a number of key advantages over both model and non-model organisms of evolutionary research for the study of speciation. Few, if any, vertebrates offer such a powerful opportunity for comparative approaches; cichlids are one of the most diverse and species-rich families of vertebrates, accounting for ~5% of all vertebrate species. Speciation in cichlid fishes involved extremely high levels of phenotypic divergence, including changes in body shape and jaw structures, feeding behaviors related to prey in their ecological niches, and social and parenting behaviors related to mate choice and reproduction. Their genetic diversity is low; for example, across cichlid species in Lake Malawi, nucleotide diversity between species is only 2x higher than between human individuals. Additionally, many of these species can interbreed, providing the opportunity to use genetic approaches to study causality of associated changes.
In order to enable experimental approaches such as QTL mapping, single-cell transcriptomics, and optogenetic manipulations in behaving animals, we design cheap and automated setups to characterize specific social behaviors in naturalistic environments. We use a combination of Raspberry Pi microprocessors, HD cameras, and depth sensors to collect data during 10 days of bower building, where male cichlids build castles made of sand to attract female mates. A variety of computer vision and machine learning approaches are used to characterize the bower shape and identify the hundred of thousands of sand manipulation events that occur during bower construction.
We are interested in expanding these approaches to improve overall behavioral characterization and assay additional types of social behaviors such as mating behaviors, aggressive behaviors, and parental behaviors.
Example projects: Adaptation of machine learning networks to track individual fish in complex environments. Simulate fish to use as visual input for a variety of social behaviors. Utilize these tools for identifying genes and neurons responsible for specific behaviors.
Key papers: Johnson et al, 2020, Long et al, 2020
2. Identification of causal variants that influence feeding and fitness
As organisms encounter new environments, feeding strategies and metabolic networks can be out of balance with new potential dietary sources. Exposure to new diets can also have medical consequences, such as increased amounts of sugar and fats in European and American diets are associated with increased levels of obesity. Over evolutionary timescales, it is likely that humans will adapt to these new diets. If we could predict the adaptive alleles to this diet, we could potentially design drugs or therapies to mimic their effects. Fundamental understanding of feeding evolution could provide the foundational basis for such an endeavor.
We use C. elegans nematodes as a model to identify the genetic variation that are responsible for changes in feeding behaviors. These genes are most often conserved in mammals including humans, suggesting our work should create insights that will translate to human health. Our approaches include experimental evolution, behavioral and feeding assays, optogenetic manipulation, and genome-editing.
Example projects: Identification of genes responsible for differences in feeding rates in different wild strains of C. elegans. Directed evolution to evolve changes in feeding behaviors in response to different diets.
Example publications: Greene et al, 2016, Zhao et al, 2018, Xu et al, 2020, Zhao et al, 2020
3. The role of structural variants in the evolution of behavior
Genetic variation is the substrate for evolution to act on. From the generation of new species on the Galapagos Islands to metastatic tumors that affect millions of Americans a year, phenotypic changes are only possible because of genetic variation between individuals and cells. One important class of genetic variation are large structural variants. These include simple structural changes such as large insertions/deletions, inversions, translocations, and duplications. More complicated rearrangements of DNA are also possible, such as seen in cancer cells and in human evolution. Structural variants are especially interesting from an evolutionary and health perspective due to the large impact they can have on gene function and expression. They new substrates for gene birth (through duplications), create chimeric fusions between different genes (through translocations), or suppress recombination to allow the evolution of supergenes (through large inversions).
Despite their importance, large structural variants are very difficult to identify using short-read sequencing technologies. We are taking advantage of new optical DNA mapping technologies to identify large structural variants previously hidden to resequencing approaches. We are interested in understanding how these variants contribute to behavioral change.
Example projects: Determination of the new mutation rate for structural variants. Identification of their role in speciation in cichlid fishes. Determine the genetic and molecular mechanisms by which they influence behavior.
Key Publications: Large et al, 2016, York et al, 2018, Zhao et al, 2020, Patil et al, 2021
Key Collaborators
Each of these projects have benefited from multidisciplinary expertise provided by a number of different laboratories. Our collaborators include the following research groups:
Andersen Lab (Northwestern University Molecular Biosciences)
Bargmann Lab (Rockefeller University Neural Circuits and Behavior)
Lu Lab (Georgia Tech Chemical and Biomolecular Engineering)
Streelman Lab (Georgia Tech Biological Sciences)