Abstract
Biotic interactions between plants and insects can significantly shape their reciprocal evolution. Seed predation by ants can be viewed as either a mutualistic or antagonistic interaction, depending on the balance between its costs and benefits for the plant species. In Mediterranean agroecosystems, harvester ants specialize in collecting seeds, including those of cultivated cereals. Yet their potential role in shaping crop traits remains largely unexplored. Durum wheat (Triticum turgidum ssp. durum), a key Mediterranean crop domesticated from wild emmer (Triticum. turgidum ssp. dicoccoides), has undergone major morphological changes during domestication, particularly in reproductive organs such as spike and seeds. However, whether interactions with seed predators like harvester ants contributed to the selection specific plant phenotypes remains unknown. In this study, we investigate the genetic architecture underlying durum wheat susceptibility to ant predation, and then test in a separate experiment whether M. barbarus foraging behavior reflects plant genetic variation. Building on observations from a field experiment, we investigated the phenotypic and genetic determinants of seed predation by the harvester ant Messor barbarus across a panel of 180 durum wheat inbred-lines exhibiting high phenotypic and genetic diversity. First, we visually assessed spike predation by M. barbarus on 95 genotypes within ant foraging areas and conducted a genome-wide association study (GWAS) using SNP markers to uncover the genetic basis of harvester ant predation on durum wheat. Based on the identified genomic regions, we examine the functions of candidate genes potentially involved in ant predation. We also designed a cafeteria experiment using 208 spikes from 26 genotypes placed at the entrance of eight nests to validate our genetic findings and investigate how spike and seed traits influence ant preferences. We estimated the heritability of wheat on ant preference using the spike mass loss after 24 and 42 hours and tested whether ant preference is partly genetically determined in wheat. Then, we analysed the selective pressure exerted by ants on spike and seed traits, such as morphological characteristics and protein content, to better understand the selective pressures at play. We detected a significant Quantitative Trait Loci (QTL) on chromosome 2A, whose allelic variation explained 21% of the observed phenotypic variation in predation rates by M. barbarus. This region encompasses a 3.6 Mb segment that includes a chromosomal inversion between wild emmer wheat and durum wheat, and 46 candidate genes, from which one MYB transcription factors that could be involved in cuticle and chemical trait regulation. This QTL was validated by the cafeteria experiment: ants preferentially removed spikes from genotypes carrying the preferred allele detected from the GWAS analysis. Furthermore, ants preferentially started to harvest seeds from shorter spikes, with some ants sometimes bringing the whole spikes into the ant nest. While spike morphology did affect ant choice, we did not find any significant correlation between seed morphology or protein content and predation levels. Our study provides the first evidence that harvester ants exhibit genotype-specific preferences in durum wheat, and that these preferences are linked to a major QTL on chromosome 2A. These findings suggest that ant foraging may be an underappreciated factor in cereal crop evolution, providing newperspectives to better understand interactions between seed predators and crops.