Genomics of invasion of the ladybug Harmonia axyridis – GANDHI

WP1 – Extend genomic resource: high quality annotated assembly, linkage and LD maps information

Hi-C sequencing
Medium-density linkage maps and LD maps
Annotation using gene prediction bioinformatics tools
Encapsulation of the genomic resources in a user friendly browser

WP2 - Characterize the overall genome response during invasions

Development of new ABC-Random Forest algorithms and associated statistical solutions
Implementation of the software “DIYABC-Random Forest ” and of the R-package abcrf
Development and exploration of ƒ-statistics estimation and admixture graph construction with Pool-Seq or allele count data to infer population and invasion history
Application on the NGS population data obtained from wild H. axyridis populations
Characterization/comparison of the patterns of Transposable Elements (TE) in native and invasive populations
New statistical developments in BAYPASS
pGWAS on invasive/non-invasive populations and life-history traits
Attempt to determine functional gene networks
Functional validation of the candidate genes

WP3 - Genomic determinants of traits potentially involved in successful invasions

Whole-genome sequencing of invasive and native Red-nSpots individuals
Mapping, genotyping and phasing
Statistical treatments using REHH (and more methods)
RNAseq on hybrids raised at three temperatures
RNAi for factors involved in plasticity
ChIPseq and FAIREseq
Fine characterization of elytral phenotypes (colour morphs) at different temperatures of larval development by means of image analysis methods based on artificial intelligence (AI) algorithms
Task 3.2. Female body mass
New statistical developments in the BayPass and HMML methods for Pool-Seq time series data
pGWAS to identify candidate genes
Attempt to determine functional gene networks
Functional testing of candidate genes (RNAi)
Task 3.3. Time at first reproduction
New statistical developments in the BAYPASS and HMML methods for Pool-Seq time series data
NGS data production (Pool-seq Shotgun) of the selected population samples
pGWAS to identify candidate genes
Attempt to determine functional gene networks
Functional testing of candidate genes (RNAi)

Main results obtained so far

- To optimally address the questions associated to the three aforementioned objectives, we have substantially extended our H. axyridis genomic resources by producing a large number of NGS data of various types) and have explored different bioinformatics methodologies to optimally extract full genome genetic variation from resequencing data.
- We have developed (machine learning) Random Forest algorithms to perform scenario selection and parameter estimates of interest from simulated genetic datasets, as well as tools to evaluate the accuracies of these estimates. We found that ABC-RF approaches are particularly well suited to the analysis of large NGS datasets in order to discriminate complex invasion routes and estimate parameters of interest such as the intensity of bottlenecks at the time of introductions, or the rates of genetic mixing between several source populations. The application on the genomic data generated for the study of the global invasion of H. axyridis will be processed in the coming months.
- We have developed new methods implemented in the CBGP program Baypass to (i) more effectively measure correlations between population allele frequencies and phenotypic or environmental or historical status (e.g. invasive versus native) covariates of the study populations, and (ii) treat altogether Individual and Pool sequencing NGS data. We have started to apply this new version of Baypass on the genomic data generated for the study of the global invasion of H. axyridis and on the genomic data generated for the study of two experimental selection experiments processed on H. axyridis.
- Using RNAseq analyses on appropriate crossings, we found that the main genetic determinant of variation in coloration patterns of H. axyridis elytra, the pannier gene, is also strongly involved in the molecular mechanisms underlying thermal plasticity of coloration in H. axyridis. More precisely, we have shown that only the expression level of the non-melanic pannier allele is strongly temperature responsive.

Future prospects and additional research actions

We found it pertinent to add three research actions in line with the objective 2 (research actions 1/ and 2) and objective 3 (research action 3/) of the project:

Research action 1: The development and exploration of ƒ-statistics estimation and admixture graph construction with Pool-Seq or allele count data to infer population and invasion history (I.e. route of invasion), and this independently from prior knowledge regarding time of introductions.

Research action 2: We have started characterizing/comparing the patterns of Transposable Elements (TE) in native and invasive populations of H. axyridis. Mixture graph approaches are proving to be complementary to ABC Random Forest approaches and have been successfully applied to genomic data generated for the study of the global invasion of D. suzukii and are ready to be applied to genomic data generated for the study of the global invasion of H. axyridis. Transposable elements (TEs) are DNA sequences capable of replicating themselves within genomes independently of the host cell DNA. These sequences are virtually present in all eukaryotic genomes and because of their ability to mobilize in the genomes they are thought to be very important regulatory elements in the genome, especially in the context of biological invasions in which rapid contemporary evolution might be needed. This is the first time that such genomic elements are studied in H. axyridis (and more generally in Coccinellids).

Research action 3: Regarding the study of genomic determinants of the plasticity found for at least some color forms, we observed a significant decrease in the expression level of the melanin allele at high developmental temperatures, although this decrease was much less than that observed for the non-melanic allele, this observation is surprising since melanic morphs of H. axyridis do not show phenotypic thermal plasticity, at least when observed by eye. These molecular results have recently led us to envisage a third new research actions in line with the objective 3 of the project: the fine characterization of elytral phenotypes (colour morphs) at different temperatures of larval development by means of image analysis methods based on artificial intelligence (AI) algorithms. This type of analysis is likely to reveal differences in black coloration intensity that are not visible to the eye. These methodological developments are being carried out in close collaboration with a start-up specializing in AI methods, BionomeeX.

Publications
Collin FD et al.. Extending Approximate Bayesian Computation with Supervised Machine Learning to infer demographic history from genetic polymorphisms using DIYABC Random Forest. Molecular Ecology Resources, Wiley/Blackwell, 2021, 21 (8), pp.2598-2613. ?10.1111/1755-0998.13413?. ?hal-03229207?
Gautier M et al. ƒ-statistics estimation and admixture graph construction with Pool-Seq or allele count data using the R package poolfstat. Molecular Ecology Resources, Wiley/Blackwell, 2021, 22,1394-1416. ?10.1111/1755-0998.13557?. ?hal-03481066?
Oral communications
Estoup A (2021) The genetics of invasive species: the arlequin ladybird Harmonia axyridis as a model species ; journées du réseau EFOR (Réseau d’Etudes Fonctionnelles chez les ORganismes modèles) - 10-11th of May 2021.
Estoup A (2022) Goodness of fit using standard ABC and ABC Random Forest. Meeting of the «Population Genetics/Genomics« group, Montpellier, CBGP, 7 June 2022.
Estoup A, Prud’homme B (2022) Characterisation of phenotypic plasticity of color forms in Harmonia axyridis: a step forward using deep learning based imaging analysis based on deep learning machine learning. Bionomeex, Montpellier, 5 août 2022.
Dissemination actions
Interview and writing assistance (in 2021) for TV/press on the Asian ladybird as an invasive species: france3-regions.francetvinfo.fr/grand-est/faut-il-se-mefier-coccinelles-asiatiques-1652776.html .
Interview + Special issue of Mariane newspaper - Natural insecticide, ecological disaster : the Asian ladybird, this ally turned invader www.marianne.net/societe/sciences-et-bioethique/insecticide-naturel-catastrophe-ecologique-la-coccinelle-asiatique-cette-alliee-devenue-envahissante. Published on 24/07/2022.
Computer packages
Marin et al. (2022) R Package: abcrf (Approximate Bayesian Computation via Random Forests). Version 1.9 (2022), Available on the CRAN.
Collin F-D et al. (2021) DIYABC Random Forest v1.0. This software integrates two functionalities in a user-friendly interface: the simulation, within custom evolutionary scenarios, of different types of molecular data (microsatellites, DNA sequences or SNPs) and Random Forest treatments, including statistical tools to assess the power and accuracy of inferences on model choice and parameter estimation - Available at diyabc.github.io.
Others
Collin F-D et al. (2020) USER MANUAL for DIYABC Random Forest v1.0 - Pp 61 – Available on diyabc.github.io/rf/Diyabc%20Random%20Forest%20User%20Manual%2010-07-20_submitted_with_MER_ms.pdf
Estoup A. Scientific advisor for an ARTE FRANCE documentary film on insects (including some invasive ones). Correction of the scenario, supply of biological material, presence on some filming sites for breeding and handling of insects, and participation in editing) - Director Jennifer Aranda (ARTE FRANCE). Release end of 2022 - beginning of 2023. Expertise mobilised throughout 2021-2022.

Biological invasions are defined as the successful establishment of species outside their native range. They are a component of global change and induce substantial impact on invaded communities and ecosystems as well as considerable economic loss. While the ecological effects of invasions are well documented, the traits, the evolutionary forces and the genomic responses to selective factors that endow populations with a higher propensity to invade a new environment are largely unknown. The harlequin ladybug Harmonia axyridis, an emblematic insect species to study color polymorphism, has emerged as a model species to study biological invasions. In this proposal, building on our previous work and capitalizing on the genomic resources as well as on the innovative statistical methods that we have recently developed and that we plan to further develop (e.g. ABC Random Forest and BayPass ), we will characterize the molecular footprints of adaptations that promoted the successful worldwide invasion of H. axyridis. Using complementary approaches (cf. genomics, experimental and functional analyses), we will focus on three main objectives. (1) we will consolidate and extend our H. axyridis genomic resources by generating a high-quality annotated genome assembly and a medium-density linkage map; (2) we will explore the genome response of H. axyridis during invasion by refining the routes of invasion and by conducting genome-wide association study to find the genes involved in adaptive processes during invasion; and (3) we will characterize experimentally the genetic architecture of three traits potentially involved in successful invasions by H. axyridis: coloration plasticity and variation of adult elytra, female body mass and age at first reproduction. In a general perspective, understanding the genomic response to novel invaded environments will help us to predict the conditions under which invasiveness can be enhanced or suppressed. Importantly, our methodological approaches can be transposed to any other species characterized by non-equilibrium demographic conditions. Beyond its contributions to fundamental population genomics and evolutionary biology, our project can also support innovating management methods of invasive species, including genetic-based control methods.