Experimental evidence has also demonstrated that insect herbivores evolve into pests that are more difficult to control when a genetically variable population undergoes selection in the novel pressures of an agricultural environment . For example, strains of the polyphagous spider mite Tetranychus urticae moved from high-quality common bean host plants to low-quality tomato host plants showed a progressive increase in expression of 7.5% of all genes over five generations . This finding indicates that creating more adverse conditions for pests through the development of resistant varieties and pesticide applications could lead to better adaptation among pests in the future. The coevolutionary relationship between crop wild relatives and insect pests in the center of origin predates domestication . In some cases, coevolutionary relationships between wild plants and insects were disrupted by domestication but in other cases there is strong evidence of continued coevolution. For example, in apples, it has been found that autumn color has been maintained by coevolution with overwintering aphids . Red leaf color in autumn is achieved through the active production of anthocyanins. While aphids lay their eggs in the bark rather than on the leaves, blueberry packing boxes red leaf color is an honest signal of the defenses of the tree and its quality as a host. Aphids have lower fitness on trees with red leaf color in the fall, and overall, these trees attract fewer aphids looking for egg-laying sites in the autumn .

The selective pressures of cultivation and storage can also create significant problems for pest management. In the example of bean weevils which feed on both the wild and domesticated seeds of common bean, research has shown that the Horismenus parasitoids that control this pest in the wild have been unable to adapt well to conditions in which large quantities of seeds are stored between planting seasons . Seed storage creates an ideal environment for the bean weevil population to quickly amplify through successive generations while the parasitoids require alternate sources of nutrition such as nectar and pollen to complete their lifecycle .Domestication has greatly altered the interactions of insect pests and their host crop plants. While this topic has been well studied, it is complex and highly specific to the evolutionary history of each crop and insect pest. Understanding the ways in which domestication has altered the genetic control of anti-herbivore defense traits may help with breeding for recovery of these traits . More research is needed to improve understanding of these relationships in agricultural systems, especially on indigenous crops, which have been understudied but may play an important role in food security as climate change creates adverse conditions for current staple crops . While additional research may provide critical insight into the function and evolution of unitary anti-herbivore defense traits, more consideration should also be given to the interaction of multiple defense traits and what, if any, tradeoffs and synergies may occur between them. Integrating the concept of defense syndrome and domestication syndrome could yield interesting insight into overarching patterns of change within crop and insect pest interactions .

Meta-analyses have also provided helpful insight into the ways in which anti-herbivore defense traits have been altered by domestication . With newly developed software and machine learning methods improving the ease and accuracy of such studies, more broad patterns may be revealed .Cyanogenesis is a trait that has evolved multiple times across diverse plant families . The percentage of edible crop plants that are cyanogenic exceeds with statistical significance the percentage of wild plants that are cyanogenic . This fact has given rise to the theory that cyanogenesis was an advantageous trait selected for by early farmers as it protected crops from herbivores but could be removed by post-harvest processing . Knowledge of detoxification methods such as fermenting, leaching, and cooking predate agriculture and may even predate the evolution of modern humans . Cyanogenesis in harvested organs was selected against during domestication in some species like cassava, almond, and Lima bean despite the knowledge of detoxification methods and the trait’s utility as an anti-herbivore defense . This may contribute to higher vulnerability of these and other crops to insect herbivores . Analysis of the genes underlying cyanogenesis using a genome-wide association study of a Mesoamerican diversity panel that includes wild and domesticated accessions could elucidate the mechanism by which this trait has been so reduced or eliminated in the domesticated gene pool. This in turn could improve understanding of the domestication history and aid conservation and plant breeding efforts. Lima beans are one of only a few pulse crops that produce cyanide and are the only species in the genus Phaseolus to do so .

Hydrogen cyanide has been found in wild Lima beans at levels lethalto humans, but it is consistently less abundant or undetectable in the seeds of domesticated lines . Cooking eliminates the cyanide content of mature beans, rendering them safe for human consumption . Cyanide production in the leaves of domesticated Lima beans has been found to be equal to or greater than concentrations in wild Lima beans . Floral tissue was not tested in the previously referenced study, but expression of cyanogenesis in flowers may be important as they are especially vulnerable to economic damage by hemipteran pests like Lygus hesperus The two main cyanogenic glucosides produced by Lima bean are linamarin and lotaustralin . Linamarin is synthesized from valine and lotaustralin from isoleucine . Cyanogenic glucosides are stored in the vacuole of the plant cell and released when the tissue is ruptured . At that point, they are hydrolyzed by linamarase which produces glucose and acetone cyanohydrin. The acetone cyanohydrin then spontaneously or with hydroxynitrile lyase splits into acetone and hydrogen cyanide. Prior research has identified several QTL for volatile cyanide produced in the floral buds, immature pods, and leaves of a biparental recombinant inbred line population of Lima beans derived from an Andean and a Mesoamerican domesticated parents . Additional research has identified the sequence of both the gene that controls production of cyanogenic glucosides and the hydrolyzing enzyme in white clover , a legume relative of Lima bean . During the process of domestication, one or both genes may have been lost – or their expression reduced – resulting in lower cyanogenic capacity of domesticated lines compared to wild lines. Whilethese data provide clues to the control of cyanogenesis in Lima bean, package of blueberries data from more diverse germplasm would strengthen these results. This study examines the relationship between these previously discovered QTL and the findings of a genome-wide association study in a Mesoamerican diversity panel of wild and domesticated Lima beans. This research aims to elucidate how cyanogenesis has been affected by domestication in the Mesoamerican gene pool of Lima beans with special consideration of the cyanogenic capacity of California cultivars.A diversity panel of 363 lines of wild and domesticated Lima beans was selected for the study. Of these, 270 lines were drawn from the diversity panel used for the study of Lima bean domestication , 76 lines were from a diversity panel of Lima beans adapted to growth in the Central Valley of California, and an additional 20 wild lines were collected in Mexico and grown for the study. Due to issues with germination, photoperiod sensitivity, and disruptions from the COVID-19 pandemic, flower phenotypes were collected for only 207 of these lines and pod phenotypes were collected for only 164 of these lines. Plants were grown in a greenhouse setting over the course of three plantings initiated in January 2020, September 2020, and January 2021. Plants were grown under natural lighting conditions with a vertical black plastic curtain blocking artificial light from a neighboring greenhouse. Flowers were collected on the first day they opened as judged by their color: white or purple as opposed to the yellow of the second day. Young pods were collected at approximately 2cm in length. Samples were stored in 96-well plates. Immediately after collection, tissue samples were taken to a -80 °C freezer and stored until processing.

Samples were removed from the freezer and their caps were removed and replaced with a prepared piece of Feigl-Anger paper. The freezing and subsequent thawing of the samples disrupted the cells, freeing the cyanogenic glucosides from the vacuole and allowing the hydrolyzing enzymes to cleave cyanide from the sugar. The liberated cyanide volatilized and rose to meet the Feigl-Anger paper. Once in contact with the Feigl-Anger paper, the cyanide interacted with the chemical treatments to turn the paper blue. The Feigl-Anger paper was changed after 15, 30, 60 and 90 minutes from the start time of the assay to create four distinct exposure windows. This timing was selected to maximize HCN capture and avoid paper saturation based on the results of a time trial that was conducted prior to analysis of actual samples. The Feigl-Anger paper was scanned and analyzed using the readplate2 plugin on ImageJ 1.52q to measure the intensity of blue colored caused by the volatile cyanide . Early observations suggested that the position of a sample on the plate appeared to affect the results. Samples on the outside of the 96-well plate appeared to defrost more quickly than samples in the interior of the 96-well plate. During the time trial, this concern did not arise because fewer samples were included and so the thermal mass of the plate was lower. To correct for a possible influence of plate position, the ‘emmeans’ package in R version 4.2.1 was used to create a linear mixed effects model with variety as a fixed effect and plate, row within plate, and column within plate as random effects . The phenotypes tested in the GWAS were the estimated means of each tissue type measured in each exposure window for each variety.The sequence data for 157 of the lines were downloaded from the NCBI Sequence Read Archive . Genetic data were acquired for the other 93 lines by first extracting DNA from embryonic radical tissue using a DNeasy 96 Plant Kit . The DNA was then sequenced with genotyping-by-sequencing using the restriction enzyme ApeK1. This restriction enzyme was selected to ensure compatibility between this dataset and the preexisting dataset . It should be noted for future studies, however, that the CviAII restriction enzyme produces more evenly spaced markers when conducting GBS on the genus Phaseolus . Sequence data were aligned to the Phaseolus lunatus reference genome , annotated, and filtered using the Next Generation Sequencing Eclipse Plugin version 4.2.0 . Data used for masking SNPs in repetitive regions of the genome were provided by Jorge Duitama . The data was then moved to TASSLE, where it was imputed with the LD-kNNi method . In a test with 1% of the data masked, the error rate of this imputation was 0.023.A STRUCTURE analysis was performed to identify population structure that could cause misleading results in the GWAS. This was done in the command line version of STRUCTURE software 2.3.4 . The genome-wide association study was conducted using the Bayesian-information and Linkage-disequilibrium Iteratively Nested Keyway Model in the Genomic Association and Prediction Integrated Tool Version 3 in R version 4.2.1 . A Q matrix was not included as a covariate variable to control for kinship, because all Andean lines were removed from the sample pool following the STRUCTURE analysis. Testing additional analyses with a Q matrix included will be a step taken prior to publication.It has long been established that Lima beans were domesticated twice, once in Mesoamerica and once on the western slope of the Andes . The primary gene pool of Lima bean, including domesticated lineages and the wild populations from which they were derived, can be optimally divided into five subgroups: Mesoamerican I , Mesoamerican II , Andean I , Andean II , and admixed . Given that the accessions included in this study were a diverse mixture of wild, landrace, and elite breeding lines from throughout the Americas, it was important to check for population structure prior to conducting a genome-wide association study, as the linkage disequilibrium between sites with causaland non-causal alleles in related individuals can lead to artificial associations . As expected, a kinship matrix produced by the ‘GAPIT’ package in R revealed significant population structure within the 363 lines included in this study. Given that most of the available phenotypes came from the Mesoamerican gene pool, removing the Andean lines from the study was the most efficient way to remove biases introduced by this population structure.