Dupuis, Julian

Owned by UKY CCS RCD
Jun 09, 2024
4 min read

Julian Dupuis’ Research Group Summary

Research in our group focuses on many facets of insect systematics and evolution, which in the big picture aims to understand the dynamics and mechanisms by which biodiversity arises. Despite this broad context, many of our research questions operate on more microevolutionary scales—the wee tips of our great tree (*web) of life, and often the grey zones between them. To answer our questions, we use tools from integrative taxonomy, genomics (particularly phylogenomics and population genomics), and bioinformatics. Taxonomically, we focus on lepidopteran and dipteran systems, but dabble in groups from across the insects.


Species identification & delimitation. Many of our research projects revolve around a simple question: is it one species or two? While basic in scope, this question is often extremely difficult to answer, both methodologically and conceptually. However, it can have far reaching consequences, especially when dealing with groups of conservation concern or recurrently invading or invasive pests. Species identification and delimitation permeate the taxonomic, population genetic, and phylogenetic research conducted in the lab.


Speciation & hybridization. Hybrid interactions have long provided unique “natural laboratories” for studying speciation, as they highlight cases where reproductive isolation breaks down between diverged species. Genomics is providing new insight into the process of hybridization and speciation, and we have ongoing projects studying historical and contemporary hybridization, the genomic architecture of speciation, and phylogeography in the Papilio machaon group of swallowtail butterflies. Members of this species group readily hybridize in the laboratory and are represented by ancient hybrid lineages (via homoploid hybrid speciation) and modern hybrid swarms in nature, yet our understanding of why hybridization occurs in some areas but not others is limited.


Systematics & phylogenomics. Phylogenetic inference provides a conceptual and analytical backbone for many of our research questions. This extends from directed systematic research questions to developing new methodologies for phylogenomic data generation.


Conservation genetics. The need for accurate species identification and delimitation is amplified in species of conservation concern. Given the sheer diversity of insect systems, the main conservation needs are often identifying and delimiting Evolutionarily Significant Units (ESUs) in the context of existing taxonomy and anthropogenic habitat destruction. We have several ongoing projects using population genomic and phylogenomic approaches to evaluate the evolutionary dynamics of threatened and federally-endangered butterflies and moths.


Applied molecular diagnostics. Recurrently invading pests present unique challenges for pest management, but also opportunities to utilize genomics to understand invasion dynamics and inform regulatory management through pathway analysis. In collaboration with multiple USDA-ARS and USDA-APHIS labs, we have ongoing projects to develop diagnostic tools for use by regulatory agencies. These efforts include leveraging phylogenomic datasets for highly informative markers for species identification, using population genomics to develop pathway analysis tools for source determination of intercepted specimens, and developing bioinformatic tools to facilitate data analysis for these diagnostic tools.



Insect population genomics & phylogenomics

We have various projects on the go using population genomic and phylogenomic methods to study insect systems. Most of these are focused on evolutionary or conservation-related questions, such as: determining the genomic architecture of speciation in hybridizing swallowtail butterflies; evaluating evolutionary significance and phylogeography in federally endangered Californian butterflies; and, differentiating the drivers of speciation in North American buckmoths.


Software:

  • BWA, read mapping
  • Stacks suite, RADseq data processing & SNP calling
  • GATK, SNP calling
  • IQTree, maximum likelihood tree estimation
  • BEAST, Bayesian inference tree estimation and various other heuristics
  • STRUCTURE, individual-based clustering
  • VCFtools, data filtering and manipulation


Collaborators:

  • Felix Sperling, University of Alberta
  • Dan Rubinoff, University of Hawaii
  • Scott Geib, USDA-ARS, Hilo, Hawaii
  • Kelly McGowan, University of Missouri
  • Ric Peigler, University of the Incarnate Word


Applied molecular diagnostics for Tephritidae

Tephritid fruit flies include some of the most economically damaging insect pests in the world, and with increased global trade and movement, incursions of these flies outside of their native distribution is common. Additionally, the taxonomy and systematics of these flies are often unresolved, and morphological and traditional molecular diagnostics (mtDNA) are often insufficient to provide reliable identifications. Extensive surveillance and management efforts are undertaken to prevent the establishment of these invasive flies in the USA, and when specimens are intercepted as part of these surveillance networks, there are two questions of immediate importance: What is it, and where did it come from? As part of a large network of collaborators, we develop genomic-based diagnostic tools to answer these questions, as well as bioinformatic and analytical tools to assist regulatory agencies in their diagnostic responsibilities.


Software:

  • Developed in house, in collaboration with Scott Geib (USDA-ARS):
  • HiMAP: https://github.com/popphylotools/HiMA
  • mvMapper: https://github.com/popphylotools/mvMapper
  • BioPython
  • Pandas
  • Various phylogenetic and population genetic tools


Collaborators:

  • Scott Geib, USDA-ARS
  • Reinaldo Brito, University Sao Carlos
  • Dan Rubinoff, University of Hawaii
  • Norman Barr, USDA-APHIS
  • Raul Ruiz-Arce, USDA-APHIS


Deep learning for butterfly species delimitation

Here, we are building off the use of machine learning/deep learning for species identification (e.g. apps to identify photos of plants or insects) but using these approaches to evaluate characters for species delimitation. Angelwing butterflies are morphologically variable and can be difficult to delimit based on traditionally used characters in field guides. However, the phylogeny of North American species based on genomic markers is relatively straightforward. We are using the genomic identity of these butterflies as an a priori identification and using machine/deep learning to evaluate what morphological characters are best for delimiting these genomic entities.


Software: (Python)

  • PyTorch
  • NumPy
  • Matplotlib
  • Seaborn


Collaborators:

  • Scott Geib, USDA-ARS


Students and Staff:

  • Eric G Chapman, Research Analyst, Added 12/14/2020
  • Oksana Vernygora, Postdoc, Added  12/14/2020
  • Kristie Schmidt, Graduate, Added 12/14/2020
  • Kantima Thongjued, Added 12/14/2020
  • Ryan Lardner, Graduate, Added on LCC, 02/11/2022 
  • Reinaldo Otavio, Graduate, Added on LCC, 01/06/2023 


Center for Computational Sciences