EVOLUTIONARY GENOMIC ANALYSIS OF THE CHARCOAL ROT FUNGUS MACROPHOMINA PHASEOLINA FOR IMPROVED DISEASE MANAGEMENT UNDER CLIMATE CHANGE By Viviana Ortiz Londoño A DISSERTATION Submitted to Michigan State University in partial ful�llment of the requirements for the degree of Plant Pathology – Doctor of Philosophy Ecology, Evolutionary Biology and Behavior – Dual Major ���� ABSTRACT Global agricultural production is threatened by several diseases caused by fungal pathogens. Recently increased e�orts to characterize genomic diversity in fungal pathogens and the availability of large-scale ecological datasets o�er new opportunities for understanding pathogen adaptation. The twin lenses of population genomics and adaptive evolution are powerful frameworks to interpret this data because of characteristics of fungal pathogens in agroecosystems that allow for their rapid evolution. The environ- ment, biotic and abiotic, is a major driver for the evolution of plant pathogens and greatly in�uences disease outcomes. Macrophomina phaseolina causes charcoal rot in many important economic and sub- sistence crops worldwide. Charcoal rot signi�cantly reduces yield and seed quality of soybean and dry bean and has been recognized as a warm climate-driven disease of increasing concern for crop production under global climate change. Therefore, this dissertation investigated the genetic structure and adaptive potential of M. phaseolina to understand how this pathogen responds to hosts, fungicides, and climate and how to best manage and predict charcoal rot disease. To this end, I �rst characterized the genetic diversity and genotype-environment associations in M. phaseolina �lling in fundamental knowledge of population structure and shedding light on climate adap- tation. Population genomic analyses of �� M. phaseolina isolates from soybean and dry bean across the continental US, Puerto Rico, and Colombia revealed geographic structure and diversi�cation associated to climate. Phylogenomic and clustering approaches di�erentiated isolates into two main clades of the US and Colombian-Puerto Rican origins and �ve divergent genetic clusters within these clades. I identi�ed a predominantly clonal structure in the US and a semi-clonal structure in Colombia and Puerto Rico. Lim- ited genetic di�erentiation between isolates of soybean and dry bean origins was observed. Estimations of the independent contributions of neutral population structure, space, and climate to genetic variation, revealed that climate signi�cantly contributes to genetic variation between genetic clusters. Genotype- environment associations implicated several genomic regions in M. phaseolina adaptation to climate and the loci signi�cantly associated with multivariate climate were found near to genes related to fungal stress responses. Information on the e�cacy of newer fungicides chemistries for charcoal rot management is lack- ing. Therefore, I characterized the in-vitro fungicide sensitivity of M. phaseolina to three major chemi- cal classes of single-site fungicides, succinate dehydrogenase inhibitors (SDHI; boscalid) dicarboximides (iprodione) and demethylation inhibitors (DMI; prothioconazole). This study found no isolates in the US, Colombia or Puerto Rico that were insensitive to any of the fungicides tested. Isolates were most sensitive to prothioconazole indicating its potential use for charcoal rot management. Next, mutations in the fungicides target protein genes were investigated. No mutations that associated to levels of sen- sitivity to boscalid, iprodione and prothioconazole were found among our isolate collection. Finally, a preliminary ecoclimatic suitability model was developed and used to project the climatic suitability of M. phaseolina at a global scale. Importantly, this model predicted areas of high climatic suitability which may be at increased risk of disease. Results from this dissertation work inform and improve charcoal rot management strategies through better understanding of M. phaseolina genetic structure and adaptive potential, in-vitro e�cacy of single- site fungicides and potential disease outcomes under a changing climate. Additionally, this research is expected to contribute to applied issues surrounding plant disease risk prediction, and more broadly pre- dicting short-term evolution of M. phaseolina across climates. Ultimately, this research will lead to better understanding of disease outcomes and more e�cient management of plant pathogens considering adap- tive responses under a changing climate. Copyright by VIVIANA ORTIZ LONDOÑO ���� I dedicate this dissertation to my grandmother Nora Giraldo Castillo and to my mother Fanny Londoño Giraldo. The most loving and strong women in my life. Grandma, when your children did not have access to education, you became their teacher. You taught them to read, write and do math, all while juggling the hard work of a farm in the middle of the Colombian Andes. I will forever admire your courage and how you always have a smile, even through the most di�cult times. Mom, your love and determination are always with me. You were so determined to keep studying that you went far away from home so you could attend high school. Against all odds, you became the �rst graduate in the family. You taught me empathy and to stand up for what I believe in. Abue and mami, you are my greatest source of inspiration, forever. v ACKNOWLEDGEMENTS First, I would like to thank my advisor Dr. Martin Chilvers for guiding me, encouraging me to keep going and above all for always being a source of con�dence. I feel grateful remembering that after each meeting with you I was left reassured that I could make it to the end of this journey. I am thankful to my committee members Dr. Gregory Bonito, Dr. Monique Sakalidis and Dr. Gideon Bradburd for providing expertise and advice throughout my PhD. I am also very grateful for the support of all my lab members, past and present. Thank you all for your help and advice during my PhD. Especially Dr. Alejandro Rojas, Dr. Zachary Noel and Dr. Austin McCoy whom I’m very grateful for their help while I was getting started with my research and throughout the years. I would also like to thank the scientists at Universidad del Valle and CIAT who mentored me during my undergraduate studies and helped me prepare for graduate school. I am specially grateful to Dr. Soroush Parsa, thank you for taking the time to mentor me and for your support and encouragement to pursue a PhD. I would like to thank the United Soybean Board, Michigan Soybean Committee, the North Central Soybean Research Program, Project GREEEN, and scholarships that enabled me to conduct my PhD research. Lastly, I would like to thank my friends and family. My parents, my sister and my niece for your unconditional love and support. To my grandparents, aunts and uncles, I am forever grateful for your love. Since I can remember you have supported me in every way possible. Lastly, my husband, thanks for your unconditional love and for always being there for me. I could not have done this without you. You all have been a constant source of inspiration for me and I would not have become the person that I am today without you. vi TABLE OF CONTENTS CHAPTER 1 AN EVOLUTIONARY GENOMICS PERSPECTIVE OF ADAPTATION IN PLANT PATHOGENS ...................................................................................................................................1 BIBLIOGRAPHY .............................................................................................................................18 CHAPTER 2 POPULATION GENOMIC ANALYSIS REVEALS GEOGRAPHIC STRUCTURE AND CLIMATIC DIVERSIFICATION FOR MACROPHOMINA PHASEOLINA ISOLATED FROM SOYBEAN AND DRY BEAN ACROSS THE US, PUERTO RICO, AND COLOMBIA .........................................................................................................23 BIBLIOGRAPHY .............................................................................................................................60 APPENDIX ........................................................................................................................................68 CHAPTER 3 SENSITIVITY TO SINGLE-SITE FUNGICIDES IN MACROPHOMINA PHASEOLINA POPULATIONS FROM SOYBEAN AND DRY BEAN........................................83 BIBLIOGRAPHY ...........................................................................................................................104 APPENDIX .....................................................................................................................................109 CHAPTER 4 ECOCLIMATIC SUITABILITY AND ADAPTIVE GENOMICS IN MACROPHOMINA PHASEOLINA, THE CHARCOAL ROT PATHOGEN ............................125 BIBLIOGRAPHY ...........................................................................................................................137 APPENDIX .....................................................................................................................................141 CHAPTER 5 CONCLUDING STATEMENT ......................................................................................145 vii CHAPTER � AN EVOLUTIONARY GENOMICS PERSPECTIVE OF ADAPTATION IN PLANT PATHOGENS � Natural selection is a powerful force for the evolution of living organisms. From humans to microor- ganisms, adaptations that overcome challenges in the environment are fundamental for the success of populations in diverse environments. Unraveling the genetic basis underlying adaptive phenotypes has for years fascinated scientists and hence has been the focus of many studies. In host-pathogen systems, adaptation is driven mainly by selection pressures imposed by their interaction itself and by the abiotic environment. Most notably, in agroecosytems, such abiotic environmental pressures are related to cli- mate and pesticides use (Stukenbrock et al., ����). Other evolutionary forces also play an important role in the adaptation of organisms, among them migration is often considered because it can signi�cantly af- fect adaptation processes. In this chapter, I discuss the implications of each of these factors on pathogen adaptation to plant host and the abiotic environment in agricultural ecosystems and the methodological approaches to study local adaptation. This introductory chapter provides an overview of the population genetics and community ecology concepts that constitute a foundation for the analysis and experiments in this thesis. Additionally, it provides a perspective on the use of genomic data in suitability models of plant pathogens under a changing climate. �.� Adaptation in host-pathogen systems Adaptation is considered a central topic of ecological genetics. Adaptation can be de�ned as the evolu- tionary process by which a population becomes better able to live and reproduce in its habitat (Dobzhan- sky, ����). Likewise, an adaptive trait has been de�ned as “a phenotypic trait that has evolved to help an organism deal with something in its environment” (Conner et al., ����). Adaptation is caused exclusively by natural selection; however, the remaining evolutionary forces, mutation, genetic drift and migration can either accelerate or slow down the development of adaptations (Conner et al., ����). Ecological fac- tors, biotic and abiotic, are important drivers of natural selection. These selection drivers vary in the space and such ecological heterogeneity results in populations adapted to the local biotic and abiotic conditions (Kawecki and Ebert, ����). This led me to the concept of local adaptation. In a strict sense, a popula- tion is locally adapted when it has a higher relative �tness in their local environment (habitat) relative to any other population introduced to that site (Kawecki and Ebert, ����). However, local adaptation can be indicated by genetic and phenotypic variation along ecological gradients or contrasting habitats � (Savolainen et al., ����). For example, there is genetically-based variation in growth rates in response to temperature and latitudinal gradients in fungi, including pathogens (Ellison et al., ����; Lendenmann et al., ����). �.�.� Coevolution in the host-pathogen system Plants and pathogens, as biological systems, evolve in response to adaptation in the respective partner. Co- evolutionary interactions, and especially antagonistic ones, impose strong selection on both partners. In this sense, pathogens can act as drivers of natural selection on their hosts, while hosts also impose strong selection on the pathogens by defensive mechanisms (Kawecki and Ebert, ����; Croll and Mcdonald, ����). However, inherent characteristics of pathogens such as large e�ective population sizes, high mu- tation rates and short generation times, provide pathogens with strategic advantages to evolve faster than their hosts (Croll and Mcdonald, ����). Thus, pathogens are hypothesized to have some advantages in the co-evolutionary race and it is widely accepted that pathogen populations become locally adapted to the local pool of host genotypes (Croll and Mcdonald, ����). In agricultural ecosystems, particularly in crop – fungal pathogens interactions, two main considerations will be discussed, the speci�c characteristics of fungi and those of agricultural ecosystems that makes them a unique system to study local adaptation. Key fungal characteristics, such as high reproductive potential and extraordinary capacity to disperse and survive, makes fungal pathogens ubiquitous organisms and particularly competent when adapting to new environments (Croll and Mcdonald, ����). Global agricultural production is threatened by sev- eral diseases caused by fungal pathogens representing the most important cause of crop yield losses, along with diseases caused by oomycetes (Fisher et al., ����). The crop genetic homogeneity present at the �eld level contribute to the devastating e�ects seen in agricultural ecosystems, mainly by favoring local adapta- tion processes in pathogens. Additionally, agro-ecosystems are managed in quite similar ways even when separated in space and time. For example, because of similarity in practices such as fertilization, irriga- tion, tillage and pesticide applications, combined with the planting of genetically uniform monocultures, some crop �elds show remarkably similar environments on all continents, di�ering mainly according to the local climate. � �.�.� Host specialization in plant pathogens Plant pathogenic fungi in agricultural ecosystems, are considered to be adapted to the local host geno- types, thus they constitute excellent models for identifying the genetic basis of local adaptation (Croll and Mcdonald, ����). This adaptation is largely explained by gene for gene interactions (Jones and Dangl, ����). Several genes have been identi�ed in di�erent fungal pathogens involved in virulence and pathogenicity on their hosts. These pathogens include host-specialized fungi attacking important crops, such as rice, wheat, barley, rye and maize (e.g. Pyricularia oryzae, Zymoseptoria tritici, Parastagonospora nodorum, Puccinia spp., Blumeria spp., Ustilago spp.) (Croll and Mcdonald, ����). The gene-for-gene hypothesis �rst proposed by Flor (Flor, ����) designates that avirulence genes in the pathogen are matched by resistance genes in the host. The direct or indirect interaction between the gene products triggers host defense responses that can prevent or reduce the growth of pathogens. Therefore, pathogens on a resistant host are under strong selection and tend to undergo mutation or deletion in the avirulence gene to evolve higher virulence on the host. Due to this highly speci�c interaction between avirulence and resistance gene products, avirulence genes are expected to play an important role in local adaptation processes in agricultural ecosystems (Croll and Mcdonald, ����). Similarly, to the adaptation processes driven by pathogen on their hosts, management practices including the use of resistant host germplasm, a�ects adaptation processes in pathogens. These practices result in the worldwide distribu- tion of genetically similar or identical crops, thus, selection operating on the local pathogen population can lead to occurrence of the same virulence mutations independently even in the absence of gene �ow among the corresponding pathogen populations (Conner et al., ����; Croll and Mcdonald, ����). �.�.�.� Host jumps In plant pathology, a host jump is broadly de�ned as the process by which a pathogen infects a new previ- ously una�ected host species. In some cases, this process is considered a host jump when the new host is genetically distant (i.e., taxonomically distant, from another class or order) from the original host. In con- trast to host shifts, in which the new host is closely related to the old host (Stukenbrock and McDonald, ����). Common scenarios that favor host jumps in agroecosystems include wild plant species growing nearby �eld crops, the introduction of new crops into natural ecosystems and the worldwide movement � of infected plant material (Stukenbrock and McDonald, ����). A pathogen host jump can be exempli�ed by Pyricularia species on wheat and wild grasses. The wheat blast pathogen P. graminis-tritici likely emerged from the Pyricularia population infecting the wild grass Urochloa or other Brazilian grasses approximately �� years ago (Grünwald et al., ����). Multiple host jumps occurred in the Irish potato famine pathogen Phytophthora infestans and related species, between plant hosts belonging to four di�erent families (Ra�aele et al., ����). These were favored because these pathogens originated in central Mexico (Goss et al., ����) which is considered a center of diversity for the genus Solanum (Stukenbrock and McDonald, ����)(Grünwald et al., ����). Comparative genomics approaches can detect genomic signatures of a host jump (Grünwald et al., ����), which are often considered signatures of e�ector evolution (Dong et al., ����). After a host jump, the pathogen is expected to adapt to the new host, leading to host specialization (Ra�aele et al., ����) and often to the emergence of a new pathogen species (Dong et al., ����). Accordingly, a recent host jump may be detected by comparing the genomes of pathogens from host species that represent new and old hosts. The genomes will be very similar except for speci�c changes in the genomic region that enabled the infection of the new host. These are rapidly evolving genomic regions, repeat-rich and usually containing a lot of e�ector genes. Thus, some e�ector genes may be lost because they are not useful anymore in the new host while other e�ector genes will accumulate mutations that will improve or expand the e�ector action in the new host (Dong et al., ����). Such gene loss has been seen after a host jump, in the fungus Melanopsichium pennsylvanicum (Sharma et al., ����). Greater rate of copy number variation of e�ector genes has been observed among P. infestans and related species. Signatures of adaptive evolution identi�ed as having dN/dS ratios >� (indicative of positive selection) were detected in e�ector genes of Phytophthora clade �c species (Ra�aele et al., ����; Dong et al., ����). Importantly, host jumps were proposed as a crucial mechanism for macroevolutionary persistence of host-specialized �lamentous pathogens by Ra�aele and Kamoun (����), who described the “jump or die” model in which the survival of a pathogen over long evolutionary timescales depends on the frequency of host jumps. Under this model, host jumps serve as accelerators of e�ector adaptation and lead to pathogen diversi�cation. Therefore, pathogens with more adaptable genomes, such as those with two- � speed genomes, are more likely to survive as hosts become fully resistant or extinct (Ra�aele and Kamoun, ����). �.�.� E�ect of haploid vs diploid genome on populations Ploidy, the number of chromosome sets in an organisms, greatly in�uences di�erent evolutionary aspects of populations such as the ability of organisms to mask deleterious mutations, the accumulation of dele- terious mutations and the rates of adaptation (Gerstein and Otto, ����). In general, diploid organisms have another layer of genetic variation compared to haploid organisms. Particularly, heterozygosity al- lows the occurrence of modes of gene actions, which is how genotype a�ects the phenotype. Additivity and dominance are di�erent modes of gene actions that in�uence �tness, for example in complete domi- nance a dominant allele can mask the e�ect of the recessive allele in a heterozygous organism (Conner et al., ����). Similarly, overdominance (heterozygote advantage) occurs when the heterozygous genotype has higher �tness than both homozygous genotypes. This may change how an organism responds to its environment and under a given condition may, at least, temporarily increase �tness of diploid heterozy- gous organisms (Gerstein and Otto, ����). Thus, overdominance maintains genetic variation in natural populations, and so in this way heterozygosity prevents the accumulation of deleterious mutation in the genome (Conner et al., ����). The long-term impact of deleterious mutations on the mean �tness of a population depends almost entirely on the genome-wide deleterious mutation rate and not on the selective disadvantage of the mu- tations (Gerstein and Otto, ����; Haldane, ����). Haploids will have the lowest mutation rate (and lower mutation load). This is because the equilibrium mean �tness of a population is reduced by approximately cU (the “mutation load”), where c is the ploidy level and U is the mutation rate per haploid genome. Therefore, haploids will have higher �tness than diploids, despite that deleterious mutations are masked to some degree in diploids (Gerstein and Otto, ����). The e�ect of ploidy in the rates of adaptation of populations have been investigated using experimen- tal evolution in Saccharomyces cerevisiae (Otto and Gerstein, ����; Gerstein and Otto, ����; Gerstein et al., ����; Sharp et al., ����). Adaptation of an organism to a novel environment depends on the rate in which bene�cial mutations are acquired and spread through the population (Todd et al., ����). The � rate of adaptation is a�ected by the rate of appearance and �xation of bene�cial mutations, the �tness e�ect of these mutations, the dominance of mutant alleles, and e�ective population sizes (Gerstein and Otto, ����; Todd et al., ����). This is still an area of ongoing investigation, but in general experiments have showed that large asexual haploid populations of S. cerevisiae were able to adapt faster than diploids. However, in small populations, haploids and diploids adapted at approximately the same speed, and the advantage of haploidy disappeared (Gerstein and Otto, ����). In a recent mutation-accumulation experiment conducted by (Sharp et al., ����) using S. cerevisiae, revealed that haploids were more prone to single-nucleotide mutations (SNMs) and mitochondrial mu- tations, whereas in diploids larger structural changes were more common (Sharp et al., ����). �.�.� E�ect of sexual vs asexual reproduction on populations Sexual reproduction is known to greatly a�ect the population structure of organisms and it is a determi- nant in the evolution of organisms. Plant pathogens, especially fungi with clonal and mixed reproductive systems and highly dynamic genomes constitute remarkable organisms to study the e�ect of sexual repro- duction on di�erent evolutionary aspects. This has led some to consider fungal plant pathogens as model organisms in evolutionary biology, and even as proposed models for investigating cancer cell evolution (Möller and Stukenbrock, ����). Sexual reproduction is crucial to eukaryotic evolution mainly because it can increase genetic diversity and eliminate deleterious mutations (Ni et al., ����). Recombination be- tween loci can occur during meiosis, which creates new combinations of alleles at these loci (Conner et al., ����). These allele combinations may advantageous under certain ecological conditions, thus allow- ing rapid adaptability to new environments. Rapid �xation of advantageous mutations, is also enabled by sexual reproduction by increasing the e�cacy of natural selection (Möller and Stukenbrock, ����). On the contrary, long-term advantages of clonal reproduction include the maintenance of co-adapted allele combinations in the population, and that �t genotypes can be rapidly propagated (Möller and Stukenbrock, ����). A short-term advantage of clonal reproduction is the ability to rapidly propagate while expending less energy (Ni et al., ����) which may play an important role in the development of epidemics. However, clonal populations are sometimes considered “evolutionary impaired” because of their inability to recombine advantageous mutations that may occur independently (Möller and Stuken- � brock, ����). Moreover, deleterious mutations may accumulate in the genome of clonal organisms in an irreversible manner, a process termed “Muller’s ratchet”. Another factor to consider is that asexual species usually have a lower e�ective population size (Ne) than sexually reproducing species, as o�spring are fundamentally copies of their parents. Thus, the e�ect of genetic drift is relatively greater compared to populations with large Ne (Möller and Stukenbrock, ����). Therefore, the smaller the Ne the stronger the selection has to be to counteract the e�ects of genetic drift (Conner et al., ����), which may weaken local adaptation processes. Yet, many clonal species and many fungal pathogens considered to reproduce asexually are common and successful. Approximately one �fth of described fungi are thought to be asexual and clonal (Taylor et al., ����). A possible explanation for the success of asexual fungi is the “two-speed genome” model proposed for fungi and oomycetes. In this model, genomes have a bipartite architecture with e�ectors genes being associated with compartments enriched in repetitive sequences and transposable elements (Dong et al., ����), this suggest that high mutation rates in these genome compartments support adaptive evolution by e�ector innovation (Möller and Stukenbrock, ����). Other explanations to consider are the occurrence of cryptic sex and recombination as unisexual mating, in which meiotic basidiospores are produced from the fusion of mitotically produced nuclei; and parasexual reproduction, in which there is exchange of genetic material between fused hyphae or cells without meiosis. Notably, asexual reproduction has evolved independently many times from sexually reproducing an- cestors in ascomycete fungi (Taylor et al., ����). This has led to speculate that clonal population structures in some pathogens, such as in Verticillium dahliae, have arisen at least partially because selection imposed by agroecosystems (Milgroom et al., ����). �.�.� The abiotic environment as a driver of natural selection �.�.�.� Climate adaptation Climate �uctuation and particularly temperature are important abiotic factors leading to local adaptation on fungal plant pathogens (Savolainen et al., ����; Croll and Mcdonald, ����). Models of climate change for the coming decades predict increases in global temperature, atmospheric CO�, ozone and changes in humidity, rainfall and severe weather (Fisher et al., ����). This is expected to increase the environmental � heterogeneity that already is present across di�erent agricultural systems in di�erent regions of the world. This environmental heterogeneity, acts on genetically di�erent organisms within a population, initially by causing �tness di�erences among phenotypically di�erent populations and over time mutation and recombination generate populations adapted to the local environment (Fisher et al., ����; Savolainen et al., ����). Thermal adaptation has been researched in several fungal species, including the model fungi Neu- rospora crassa (Ellison et al., ����) and the powdery mildew pathosystems Plantago lanceolata– Podosphaera plantaginis (Laine, ����). Temperature had a profound impact on the trajectory of evolu- tion of N. crassa as well as in the co-evolution in the powdery mildew system. In the powdery mildew pathosystem, host and fungal populations were sampled across a natural thermal gradient, and a local vs. foreign experiment was conducted. Host cross-inoculations were conducted using sympatric and allopatric (i.e., local vs. foreign) pathogen populations at three temperatures (i.e., home vs. away environ- ments) using detached leaves in a common garden laboratory environment. Local adaptation patterns di�ered according to temperature. Pathogen populations from the coolest environment had signi�cantly higher �tness on the sympatric host at the coolest tested temperature, but had lower �tness than allopatric pathogen populations at higher temperatures (Laine, ����; Croll and Mcdonald, ����). �.�.� Migration e�ects on pathogen adaptation The outcome of whether populations become adapted or not depend on the balance between selection and migration i.e., the levels of gene �ow among populations and the strength of selection (Savolainen et al., ����; Croll and Mcdonald, ����). Local adaptation can be hindered for certain conditions, for ex- ample by migration rates and recolonization of populations by foreign genotypes. In the context of local adaptation, a high �tness in the local environment also implies a lower �tness in a foreign environment (Savolainen et al., ����), and thus local adaptation occurs only if the e�ect of migration does not over- whelm the e�ect of local selection. Local adaptation may be disfavored by both high and low migration rates. Generally high migration rates overwhelm locally adapted genotypes leading to maladaptation, and low migration rates disfavor local adaptation mainly due to the limited genetic variation that the local population harbor, leaving limited input for selection to act on. Moreover, the unevenness in mi- � gration rates of the pathogen vs. the host also impact local adaptation processes. Pathogens are expected to become more rapidly locally adapted if they have higher migration rates than their hosts (Croll and Mcdonald, ����). �.�.� Fungicide resistance evolution in plant pathogens Fungicides play a key role in crop protection. Modern fungicides function primarily by disrupting partic- ular molecular processes and targeting speci�c proteins, and therefore are often referred to as ‘single-site’ fungicides (Brent and Hollomon, ����). In contrast, older multi-site fungicides act as general inhibitors a�ecting many cellular targets (Brent and Hollomon, ����). The continued use of fungicides may even- tually lead to the appearance of resistant pathogen populations. This phenomenon is called ‘acquired resistance’ (Brent and Hollomon, ����). Fungal pathogens with rapid reproductive rates and large population sizes are particularly prone to develop fungicide resistance (Lucas et al., ����). Although several resistance mechanisms are known, the most common one is an alteration of the target site of the fungicide. In single-site fungicides, a single gene mutation can disrupt the target site function and confer resistance or reduced sensitivity (Brent and Hollomon, ����). In situations in which resistance develops, it can be seen as a qualitative or a quantita- tive change. In quantitative resistance, the pathogen population shifts gradually towards resistance over time (Brent and Hollomon, ����). While in qualitative resistance, a bimodal distribution with sensitive and resistant subpopulations is expected. In both cases, there is positive selection for resistant individuals, ultimately leading to resistance in the population if management strategies to limit pathogen exposure are not implemented (Lucas et al., ����). �.�.� Using genomic data to detect population structure and adaptation in plant pathogens Population structure can be de�ned as a systematic di�erence in allele frequencies between subpopula- tions in a population due to di�erent ancestry (Turchin et al., ����). Population di�erentiation occurs when subpopulations are not completely interbreeding and any of the evolutionary forces (mutation, se- lection, drift, migration) change the allele frequencies within the subpopulation. In other words, when individuals within subpopulations are more closely related than individuals between subpopulations. Approaches to detect population structure include clustering methods. In clustering methods, in- �� dividuals are assigned to populations often by estimating ancestry coe�cients or using dimensionality- reduction approaches. Commonly used dimensionality-reduction approaches are principal component analysis (PCA) and discriminant analysis of principal components (DAPC). PCA is a form of multivari- ate analysis, which involves looking at multiple independent variables simultaneously to understand their contributions to the dependent variable (Abdi and Williams, ����). PCA is used in identifying popula- tion structure to infer the possible number of populations (clusters) without prior knowledge, thus it can be useful to �nd hidden population structure. PCA is commonly used to convert genetic data into a re- duced number of non-correlated variables, called principal components, which summarize the variation between samples closely related individuals can be seen as clusters. DAPC is particularly useful in organ- isms with clonal reproduction, such as many fungi. DAPC di�ers from PCA approaches, in which it does require a priori de�ned populations and maximizes the variance between populations, by partition- ing the total variance into between-population and within-population components (Thibaut Jombart, Sébastien Devillard). Model-based clustering approaches use a broad set of algorithms to characterize population struc- ture. Commonly, these algorithms di�er in the demographic model adopted, the statistical framework (frequentist or Bayesian), in whether selection is included in the model, among other aspects. Their main advantages are that they may be applied to a wide range of data sets and systems and that most of these methods do not need a priori delineation of populations. The main disadvantage is that they often rely on model assumptions. If the assumed model does not re�ect the true model, these approaches may lead to false positives or to the incorrect identi�cation of clusters. New approaches have been developed to overcome some of these limitations, such as models that incorporate a spatial component (Bradburd et al., ����), and PCA-based models (Josephs et al., ����). Genetic variation is the input for selection to act and drive adaptation processes. Genomic divergence can be inferred from polymorphisms and �xed di�erences within and between species. Approaches to in- fer adaptation processes, rely either on population genetic analyses including reverse ecology approaches, quantitative trait mapping or association studies. Each of these approaches has strengths and limitations and a combination of di�erent strategies would be more informative about adaptive natural selection �� than using just one of them. Local adaptation critically depends on selectable genetic variation within local populations. Furthermore, the probability for local adaptation to evolve depends on the genetic architecture of a trait. Phenotypic traits governed by a simple genetic architecture are likely to be more rapidly selected than complex traits (Croll and Mcdonald, ����). Similarly, loci with large e�ects should be favored to contribute to local adaptation as selection acts more rapidly on loci of large e�ects than small e�ects (Croll and Mcdonald, ����). Strategies to identify loci involved in local adaptation are discussed. �.�.�.� Quantitative trait loci mapping The outcome of host–pathogen interactions is thought to be governed largely by gene-for-gene inter- actions. However, recent studies showed that virulence can be governed also by quantitative trait loci (QTL) and that many abiotic factors contribute to the outcome of the interaction (Lendenmann et al., ����; Croll and Mcdonald, ����; Lendenmann et al., ����b; Lendenmann et al., ����a). Quantitative trait loci mapping is based on the joint analysis of phenotype and genotype. QTL anal- ysis uses a progeny of crosses between a pair of parental lines (pedigree) segregating for a speci�c trait, to �nd association between genotypes and phenotypes. QTL mapping is a powerful approach, however present some limitations. To uncover more variation many crosses and a large sample size are needed. QTL approaches can be time consuming since the progeny needs to be genotyped and phenotyped. Fur- thermore, extended linkage disequilibrium (LD) is often observed in the progeny, hindering the accurate location of the QTL. A QTL approach was used to investigate thermal adaptation in the fungal pathogen Zymoseptoria tritici (Lendenmann et al., ����b). They identi�ed four QTL associated with temperature sensitivity, containing six candidate genes including a PBS�, encoding a mitogen-activated protein kinase associated with low temperature tolerance in Saccharomyces cerevisiae. This study demonstrate a QTL approach can be successfully used in fungi, however, the need of progeny implies that QTL mapping can be applied only to sexual fungi. �.�.�.� Association mapping approaches Association mapping studies are also based on phenotype- genotype associations. However, in contrast to QTL mapping, diverse panels of organisms can be used instead of using progeny populations derived form a parental cross. Advantages of association mapping approaches include that the LD is expected �� to be lower than in pedigree-based studies, and multiple di�erent traits can be studied simultaneously. Thus, QTLs can be found in a more accurate way. However, the use of a diversity panel implies the need of correction for population structure. The rates of false positives and false negatives is high and mixed models and correction for multiple hypothesis testing are needed to distinguish real associations from spurious ones. In general, for association mapping studies associations are not necessarily causal and further validation is needed. �.�.�.� Genotype–environment associations and redundancy analysis Genotype–environment association (GEA) methods can be used to identify adaptive loci by correlating genetic data and environmental variables (Lasky et al., ����; Forester et al., ����). Multivariate methods in GEA have recently gained attention because their applications to the analysis of large genomic datasets. The multivariate nature of these methods allows the simultaneous analysis of thousands of loci (Forester et al., ����). One of the most common multivariate approaches used in GEA is redundancy analysis (RDA). RDA is a constrained ordination method that have been used for years in community ecology to examine community composition in relation to environmental variables (Legendre and Legendre, ����; Forester et al., ����). In GEA approaches, RDA can be used to disentangle the e�ects of climatic factors in shaping genetic variation, by modeling sets of molecular markers (e.g. SNPs) as responses to a func- tion of combinations of environmental predictors. RDA has been found to perform better than univari- ate methods in identifying weak, multilocus selection suggestive of polygenic adaptation (Forester et al., ����). Partial RDA models, in which the e�ects of covariables can be removed, have been used to account for underlying population structure in the identi�cation of loci associated with environmental factors in plant and animal systems (Lasky et al., ����; Forester et al., ����; Xuereb et al., ����; Gibson and Moyle, ����; Capblancq and Forester, ����) �.�.�.� Population genetics and reverse ecology Local adaptation has been investigated using population genetics with both forward and reverse ecology approaches. Population genetic analyses are based on FS T (Wright �xation index) and linkage disequi- librium (LD) methods to detect candidate loci for local adaptation in the absence of phenotypic traits. These methods can detect outlier loci with an excessive amount of genetic di�erentiation among pop- �� ulations (i.e., FS T outlier analyses; De Mita et al. ����). The basis is that local selection will exacerbate genetic di�erentiation at loci under selection compared to the genomic background. Reverse ecology (Li et al., ����), is coined because the analogy with reverse genetics and implies that prior knowledge about an ecological trait is not necessary, instead �rst �nding the genetic targets of selection and going back to identify the phenotypic di�erences or the adaptive phenotype. Reverse ecology is especially important to investigate organisms, such as microbes, which are challenging to identify adaptive phenotypes. This will preclude the utilization of associating studies such Genome Wide Association Studies. Moreover, another challenge is exempli�ed in fungi, speci�cally, asexual fungi in which the development of popu- lations to study a speci�c trait is not possible. Thus, approaches such as QTL analysis are not feasible. Reverse ecology may help overcome these challenges by investigating patterns of genetic diversity within and between populations. Ellison et al., (����) implemented a reverse ecology approach to investigate tem- perature adaptation in the model fungus N. crassa by using three di�erent population genetics metrics (FS T , Tajima’s D, and Dxy ). They identi�ed regions of genomic divergence, which are those showing low within-population polymorphism and high between-population divergence, and genes associated with response to cold temperature within those regions. However, among the three metrics used, Ellison et al. found that out of a total of �� regions showing signi�cant signatures of positive selection, only two were identi�ed by all three metrics. This suggests a high proportion of false positives. In fact, it is known that FS T outliers can be seen for reasons other than local adaptation such as deleterious alleles, species-wide selective sweeps and cryptic hybrid zones. Other aspect to consider is that regions identi�ed using reverse ecology constitute just candidate loci of local adaptation, and further functional analysis needs to be done to conclusively identify causal genes. �.� Macrophomina phaseolina the causal agent of charcoal rot Macrophomina phaseolina is a seed- and soil-borne fungal pathogen infecting more than ��� host species (Batista, Lopes and Alves, ����). M. phaseolina is haploid, reproduces asexually, and overwinters in soil and crop residue as microsclerotia. Microsclerotia are melanized structures that serve as the primary in- oculum to initiate infection in subsequent seasons (Gupta, Sharma and Ramteke, ����; Islam et al., ����). Pycnidia have been observed on host plant tissues(Knox-Davies, ����; Dhingra and Sinclair, ����; Mihail �� and Taylor, ����; Ma et al., ����; Gupta et al., ����). Although conidial suspensions have been used to experimentally inoculate soybean plants, suggesting pycnidia may provide inoculum for secondary infec- tion in the �eld, their epidemiological signi�cance has yet to be fully de�ned (Ma et al., ����; Gupta et al., ����). Depending on environmental conditions, M. phaseolina survives as microsclerotia in soil for up to �� years (Short et al. ����; Baird et al. ����), and for up to � years as microsclerotia in symptomatic seeds or as mycelium in asymptomatic seeds (Hartman et al. ����). One of the �rst descriptions of M. phaseolina was made in ���� by Halsted causing disease on sweet potato and the fungus was named Rhizoctonia bataticola (Halsted, ����). Later this fungus was described by Tassi (����) who named the fungus as Macrophomina phaseolina as it is retained today. In ����, Ashby proposed the name Macrophomina phaseoli (Maubl.). Ashby associated the microsclerotia and conidial stage by observing the structures on seedlings of multiple crops. The name Macrophomina phaseoli was changed to Macrophomina phaseolina (Tassi) G. Goidanich, by Goidanich in ����. By ����, there was controversy among researchers over the use of the name, but genera Macrophomina and Macrophoma were used to refer to the pycnidial stage and Rhizoctonia to the sclerotial state. In ����, Von Arx introduced the name Tiarosporella phaseolina (Tassi) van der Aa and reduced the genus Macrophomina to a synonym of Tiarosporella Höhn. However, this has largely been ignored by the plant pathological and mycological community (Crous et al., ����). In ����, Crous et al., in a com- prehensive phylogenetic study of ��� members of the family Botryosphaeriacea using ribosomal DNA sequences, separated the genera Macrophomina and Tiarosporella, retaining the genus Macrophomina and the name Macrophomina phaseolina. The type species of M. phaseolina was originally described from Phaseolus spp. collected in Italy (Sarr et al., ����). Soybean is one of the most economically important crops worldwide, contributing with more than half of the world’s total oilseed production (Boerma et al., ����; Wilson, ����). Seed oil and protein content makes soybean a valuable source not only for food and feed utilization but also for the indus- trial production of biofuels (Boerma et al., ����). Many diseases threaten global soybean production, including charcoal rot, caused by M. phaseolina. Charcoal rot severely a�ects soybean yield under high temperatures and drought conditions (Mengistu et al., ����). Tropical and subtropical areas, including �� the southern US, have been the most a�ected. However, charcoal rot disease in soybean is now a consis- tent threat to soybean production in southern and northern US regions (Bradley et al., ����). Although it is not clear which factors may be driving outbreaks in these regions, climatic changing conditions and resistance overcoming due to pathogen genetic divergence may be involved in the broadening of the ge- ographical range of charcoal rot disease. To date, complete resistance to charcoal rot in soybean is not known and cultural practices and fungicide seed treatments do not provide consistent control to char- coal rot in soybean (Paris et al., ����; Mengistu et al., ����; Gillen et al., ����). The con�uence of these factors, makes imperative to investigate the genetic basis for adaptation in M. phaseolina. �.� Conclusions and dissertation overview Approaches to study patterns of genetic diversity and adaptation in plant pathogens, as well as their main limitations, were discussed. One of those limitations is the di�culty to distinguish between natural se- lection and demographic processes. Thus, it is important to carefully consider the experimental design and approaches in light of the biology and epidemiology of the organism under study. Fungal pathogens can reach very high population sizes in a single plant and clonal reproduction and mixed reproduction systems are commonly observed. Furthermore, complex population dynamics and genome architecture are hallmarks of many fungal plant pathogens. Most computational tools used in population genetics are based on models developed for sexual or- ganisms (Kamvar et al., ����). Populations that reproduce clonally may violate some of the assumptions underlying the population genetic theory. Moreover, the most widely used model is the Hardy-Weinberg model which assumes diploid, sexual organisms, besides no selection, no mutation, no migration, no drift and random mating between sexes (Hahn, ����). An important assumption that is violated in clonal organisms is the random association between alle- les at di�erent loci. In several approaches, this assumption allows the prediction of genotype frequencies from the allele frequencies at each locus (Milgroom, ����). In clonal organisms associations among alle- les at several loci are nonrandom and the entire genome may be e�ectively linked (Anderson and Kohn, ����). Therefore, with clonal organisms the of use clone-corrected unlinked data is appropriate to avoid bias in diversity estimations due to duplicated genotypes (Kamvar et al., ����; Milgroom, ����). �� Although, approaches based on genetic diversity metrics are often employed to identify signatures of adaptation in plant pathogens, population genomics and ordination techniques such as redundancy analysis have the potential to accommodate the intrinsic characteristics of fungal pathogens and begin dis- entangling the e�ects of selection of those of other evolutionary forces. Such methodological approaches in conjunction with population genomics analyses, constitute powerful tools to identify patterns of ge- nomic diversity and adaptive potential of fungal pathogens. The focus of this dissertation is to improve our understanding of M. phaseolina population struc- ture, adaptation to host and climate and its application to local management practices through using the frameworks and tools of population genomics and community ecology. Additional objectives of this re- search are to characterize the sensitivity of M. phaseolina to fungicides currently used in crop production and provide a preliminary climatic suitability model for the monitoring and prediction of disease risk. �� BIBLIOGRAPHY Abdi H, Williams LJ (����) Principal component analysis. Wiley Interdiscip Rev Comput Stat �: ���–��� Anderson JB, Kohn LM (����) Clonality in soilborne, plant-pathogenic fungi. Annu Rev Phytopathol ��: ���–��� Baird RE, Watson CE, Scruggs M (����) Relative Longevity of Macrophomina phaseolina and As- sociated Mycobiota on Residual Soybean Roots in Soil. Plant Dis ��: ���–��� Batista E, Lopes A, Alves A (����) What do we know about botryosphaeriaceae? An overview of a worldwide cured dataset. Forests. doi: ��.����/f�������� Boerma HR, Specht JE, Wilcox JR (����) World distribution and trade of soybean. Soybeans: Im- provement, production, and uses. American Society of Agronomy, Crop Science Society of America, and Soil Science Society of America, pp �–�� Bradburd GS, Coop GM, Ralph PL (����) Inferring Continuous and Discrete Population Genetic. Genetics ���: ��–�� Bradley CA, Allen TW, Sisson AJ, Bergstrom GC, Bissonnette KM, Bond J, Byamukama E, Chilvers MI, Collins AA, Damicone JP, et al (����) Soybean Yield Loss Estimates Due to Diseases in the United States and Ontario, Canada, from ���� to ����. Plant Health Prog ��: ���–��� Branco S, Bi K, Liao H-L, Gladieux P, Badouin H, Ellison CE, Nguyen NH, Vilgalys R, Peay KG, Taylor JW, et al (����) Continental-level population di�erentiation and environmental adaptation in the mushroom Suillus brevipes. Mol Ecol. doi: ��.����/mec.����� Brent KJ, Hollomon DW (����) Fungicide resistance in crop pathogens: how can it be managed? �nd, revised edition. Capblancq T, Forester BR (����) Redundancy analysis: A Swiss Army Knife for landscape genomics. Methods Ecol Evol ��: ����–���� Conner JK, Hartl DL, others (����) A primer of ecological genetics. Sinauer Associates Incorporated Croll D, Mcdonald BA (����) The genetic basis of local adaptation for pathogenic fungi in agricul- tural ecosystems. Mol Ecol. doi: ��.����/mec.����� Crous PW, Slippers B, Wing�eld MJ, Rheeder J, Marasas WFO, Philips AJL, Alves A, Burgess T, Barber P, Groenewald JZ (����) Phylogenetic lineages in the Botryosphaeriaceae. Stud Mycol ��: ���–�� Dhingra OD, Sinclair JB (����) Biology and pathology of Macrophomina phaseolina. Universidade Federal de Vicosa., Minas Gerais Dobzhansky T (����) On Some Fundamental Concepts of Darwinian Biology. Evol Biol. Springer US, Boston, MA, pp �–�� �� Dong S, Ra�aele S, Kamoun S (����) The two-speed genomes of �lamentous pathogens: Waltz with plants. Curr Opin Genet Dev ��: ��–�� Ellison CE, Hall C, Kowbel D, Welch J, Brem RB, Glass NL, Taylor JW (����) Population genomics and local adaptation in wild isolates of a model microbial eukaryote. Proc Natl Acad Sci U S A ���: ����–� Fisher MC, Henk DanielA, Briggs CJ, Brownstein JS, Mado� LC, McCraw SL, Gurr SJ (����) Emerg- ing fungal threats to animal, plant and ecosystem health. Nature ���: ���–��� Flor HH (����) Current status of the gene-for-gene concept. Annu Rev Phytopathol �: ���–��� Forester BR, Lasky JR, Wagner HH, Urban DL (����) Comparing methods for detecting multilocus adaptation with multivariate genotype–environment associations. Mol Ecol ��: ����–���� Gerstein AC, Cleathero LA, Mandegar MA, Otto SP (����) Haploids adapt faster than diploids across a range of environments. J Evol Biol ��: ���–��� Gerstein AC, Otto SP (����) Ploidy and the causes of genomic evolution. Journal of Heredity ���: ���–��� Gibson MJS, Moyle LC (����) Regional di�erences in the abiotic environment contribute to ge- nomic divergence within a wild tomato species. Mol Ecol ��: ����–���� Gillen AM, Paris RL, Arelli PR, Smith JR, Mengistu A (����) Registration of soybean germplasm line DB��-����� with high yield potential and resistance to soybean cyst nematode. J Plant Regist ��: �� Goss EM, Tabima JF, Cooke DEL, Restrepo S, Fry WE, Forbes GA, Fieland VJ, Cardenas M, Grun- wald NJ (����) The Irish potato famine pathogen Phytophthora infestans originated in central Mexico rather than the Andes. Proceedings of the National Academy of Sciences ���: ����–���� Grünwald NJ, McDonald BA, Milgroom MG (����) Population Genomics of Fungal and Oomycete Pathogens. Annu Rev Phytopathol ��: ���–��� Gupta GK, Sharma SK, Ramteke R (����) Biology, Epidemiology and Management of the Pathogenic Fungus Macrophomina phaseolina (Tassi) Goid with Special Reference to Charcoal Rot of Soybean (Glycine max (L.) Merrill). Journal of Phytopathology ���: ���–��� Hartman GL, Rupe JC, Sikora EJ, Domier LL, Davis JA, Ste�ey KL (����) Part I . Infectious Diseases. Compendium of Soybean Diseases and Pests, Fifth Edition. pp ��–��� Islam M, Haque M, Islam M, Emdad E, Halim A, Hossen QM, Hossain M, Ahmed B, Rahim S, Rahman M, et al (����) Tools to kill: Genome of one of the most destructive plant pathogenic fungi Macrophomina phaseolina. BMC Genomics ��: ��� Jones JDG, Dangl JL (����) The plant immune system. Nature Reviews ���: ���–��� Josephs EB, Berg JJ, Ross-Ibarra J, Coop G (����) Detecting Adaptive Di�erentiation in Structured Populations with Genomic Data and Common Gardens. Genetics ���: ���–���� �� Kamvar ZN, Brooks JC, Grünwald NJ (����) Novel R tools for analysis of genome-wide population genetic data with emphasis on clonality. Front Genet �: ��� Kawecki TJ, Ebert D (����) Conceptual issues in local adaptation. Ecol Lett �: ����–���� Knox-Davies PS (����) Pycnidium Production By Macrophomina. S Afr J Agric Sci �: ���–��� Laine AL (����) Detecting local adaptation in a natural plant-pathogen metapopulation: a labora- tory vs. �eld transplant approach. J Evol Biol ��: ����–���� Lasky JR, Des Marais DL, McKay JK, Richards JH, Juenger TE, Keitt TH (����) Characterizing genomic variation of Arabidopsis thaliana: The roles of geography and climate. Mol Ecol ��: ����–���� Lasky JR, Upadhyaya HD, Ramu P, Deshpande S, Hash CT, Bonnette J, Juenger TE, Hyma K, Acharya C, Mitchell SE, et al (����) Genome-environment associations in sorghum landraces predict adaptive traits. Sci Adv �: �–�� Legendre P, Legendre L (����) Canonical analysis. Developments in Environmental Modelling. pp ���–��� Lendenmann MH, Croll D, Palma-Guerrero J, Stewart EL, McDonald BA (����a) QTL mapping of temperature sensitivity reveals candidate genes for thermal adaptation and growth morphology in the plant pathogenic fungus Zymoseptoria tritici. Heredity (Edinb) ���: �–�� Lendenmann MH, Croll D, Palma-Guerrero J, Stewart EL, McDonald BA (����b) QTL mapping of temperature sensitivity reveals candidate genes for thermal adaptation and growth morphology in the plant pathogenic fungus Zymoseptoria tritici. Heredity (Edinb) ���: ���–��� Lendenmann MH, Croll D, Stewart EL, McDonald BA (����) Quantitative trait locus mapping of melanization in the plant pathogenic fungus Zymoseptoria tritici. G� (Bethesda) �: ����–�� Li YF, Costello JC, Holloway AK, Hahn MW (����) “Reverse Ecology” and the power of population genomics. Evolution (N Y) ��: ����–���� Lucas JA, Hawkins NJ, Fraaije BA (����) The Evolution of Fungicide Resistance. Adv Appl Micro- biol ��: ��–�� Ma J, Hill CB, Hartman GL (����) Production of Macrophomina phaseolina Conidia by Multiple Soybean Isolates in Culture. Plant Dis ��: ����–���� Mengistu A, Arelli PA, Bond JP, Shannon GJ, Wrather AJ, Rupe JB, Chen P, Little CR, Canaday CH, Newman MA, et al (����) Evaluation of soybean genotypes for resistance to charcoal rot. Plant Health Prog. doi: ��.����/PHP-����-����-��-RS Mihail JD, Taylor SJ (����) Interpreting variability among isolates of Macrophomina phaseolina in pathogenicity, pycnidium production, and chlorate utilization. Canadian Journal of Botany ��: ����–���� Milgroom MG (����) Recombination and the multilocus structure of fungal populations. Annu Rev Phytopathol ��: ���–��� �� Milgroom MG (����) Population biology of plant pathogens: genetics, ecology, and evolution. APS Press, The American Phytopathological Society St. Paul, MN, USA Milgroom MG, Jimenez-Gasco M del M, Olivares-Garcia C, Drott MT, Jimenez-Diaz RM (����) Re- combination between clonal lineages of the asexual fungus Verticillium dahliae detected by genotyping by sequencing. PLoS One �: e������ Möller M, Stukenbrock EH (����) Evolution and genome architecture in fungal plant pathogens. Nat Rev Microbiol ��: ���–��� Ni M, Feretzaki M, Sun S, Wang X, Heitman J (����) Sex in Fungi. Annu Rev Genet ��: ���–��� Otto SP, Gerstein AC (����) The evolution of haploidy and diploidy. Current Biology ��: R����–R���� Paris RL, Mengistu A, Tyler JM, Smith JR (����) Registration of soybean germplasm line DT��–���� with moderate resistance to charcoal rot. Crop Sci ��: ���� Ra�aele S, Farrer RA, Cano LM, Studholme DJ, MacLean D, Thines M, Jiang RHY, Zody MC, Kunjeti SG, Donofrio NM, et al (����) Genome evolution following host jumps in the irish potato famine pathogen lineage. Science (����) ���: ����–���� Ra�aele S, Kamoun S (����) Genome evolution in �lamentous plant pathogens: why bigger can be better. Nat Rev Microbiol ��: ���–��� Sarr MP, Ndiaye M, Groenewald JZ, Crous PW (����) Genetic diversity in Macrophomina phase- olina, the causal agent of charcoal rot. Phytopathol Mediterr ��: ������ Savolainen O, Lascoux M, Merilä J (����) Ecological genomics of local adaptation. Nat Rev Genet ��: ���–��� Savolainen O, Pyhäjärvi T, Knürr T, Pyhaijiirvi T, Kniirr T (����) Gene Flow and Local Adaptation in Trees. Source: Annual Review of Ecology, Evolution, and Systematics ��: ���–��� Sharma R, Mishra B, Runge F, Thines M (����) Gene Loss rather than gene gain is associated with a host jump from monocots to dicots in the smut fungus Melanopsichium pennsylvanicum. Genome Biol Evol �: ����–���� Sharp NP, Sandell L, James CG, Otto SP (����) The genome-wide rate and spectrum of spontaneous mutations di�er between haploid and diploid yeast. Proc Natl Acad Sci U S A ���: E����–E���� Short G. E., Wyllie T. D., Bristow P. R. (����) Survival of Macrophomina phaseolina in soil and residue of soybean. Phytopathology ��: ��–�� Stukenbrock EH, Bataillon T, Dutheil JY, Hansen TT, Li R, Zala M, McDonald BA, Wang J, Schierup MH (����) The making of a new pathogen: Insights from comparative population genomics of the domesticated wheat pathogen Mycosphaerella graminicola and its wild sister species. Genome Res ��: ����–���� Stukenbrock EH, McDonald BA (����) The Origins of Plant Pathogens in Agro-Ecosystems. Annu Rev Phytopathol ��: ��–��� �� Taylor JW, Jacobson DJ, Fisher MC (����) The evolution of asexual fungi: Reproduction, Speciation and Classi�cation. Annu Rev Phytopathol ��: ���–��� Thibaut Jombart, Sébastien Devillard FB Discriminant analysis of principal components: a new method for the analysis of genetically structured populations. doi: ��.����/journal.pcbi.������� Todd RT, Forche A, Selmecki A (����) Ploidy Variation in Fungi: Polyploidy, Aneuploidy, and Genome Evolution. The Fungal Kingdom. NIH Public Access, pp ���–�� Turchin MC, Chiang CWK, Palmer CD, Sankararaman S, Reich D, Investigation G, Giant T, Hirschhorn JN (����) Evidence of widespread selection on standing variation in Europe at height-associated SNPs. Nat Genet ��: ����–���� Wilson R (����) Soybean: market driven research needs. Genetics and genomics of soybean Xuereb A, Kimber CM, Curtis JMR, Bernatchez L, Fortin MJ (����) Putatively adaptive genetic variation in the giant California sea cucumber (Parastichopus californicus) as revealed by environmental association analysis of restriction-site associated DNA sequencing data. Mol Ecol ��: ����–���� �� CHAPTER � POPULATION GENOMIC ANALYSIS REVEALS GEOGRAPHIC STRUCTURE AND CLIMATIC DIVERSIFICATION FOR MACROPHOMINA PHASEOLINA ISOLATED FROM SOYBEAN AND DRY BEAN ACROSS THE US, PUERTO RICO, AND COLOMBIA �� �.� Abstract Macrophomina phaseolina causes the important disease charcoal rot, which signi�cantly reduces yield and seed quality of soybean and dry bean. Although charcoal rot has been recognized as a warm climate- driven disease of increasing concern under global climate change, knowledge regarding population ge- netics and climatic variables contributing to the genetic diversity of M. phaseolina remains limited. This study conducted genome sequencing for �� M. phaseolina isolates from soybean and dry bean across the continental US, Puerto Rico, and Colombia. Inference on the population structure revealed that the isolates exhibited a discrete genetic clustering at the continental level and a continuous genetic dif- ferentiation regionally. Almost all isolates from the US grouped in a clade with a predominantly clonal genetic structure, while most Puerto Rican and Colombian isolates from dry bean were assigned to a separate cluster with higher genetic diversity. Consistently, climate signi�cantly contributed to genomic variation at a continental level with temperature seasonality and precipitation of warmest quarter having the greatest impact. The loci signi�cantly associated with multivariate climate were found closely to the genes related to fungal stress responses, including transmembrane transport, glycoside hydrolase activity and a heat-shock protein, which may mediate climatic adaptation for M. phaseolina. On the other hand, limited genome-wide di�erentiation among populations by hosts was observed. These �ndings highlight the importance of population genetics and identify candidate genes of M. phaseolina that can be used to elucidate the molecular mechanisms that underly climatic adaptation to the changing climate. �.� Introduction Delineating pathogen populations and identifying the factors shaping the patterns of genetic diversity within and among populations allow for inferences about their biology and evolutionary potential. Plant pathogens are often genetically structured in di�erent agricultural landscapes as a result of geographic and environmental di�erences (Gladieux et al., ����; McDonald and Stukenbrock, ����). Among di�erent environments, agroecosystems provide remarkable conditions for rapid adaptation of plant-pathogenic fungi. The abiotic and biotic factors such as genetic crop uniformity of monocultures, the prevalent occurrence of human-mediated migration (Wing�eld et al., ����; Crous et al., ����), and intrinsic char- acteristics of fungi such as their mode of reproduction (McDonald and Stukenbrock, ����) are known �� to be strong drivers of genomic divergence and adaptation in plant pathogenic fungi (Stukenbrock et al., ����; Savolainen, Lascoux and Merilä, ����; Croll and Mcdonald, ����). However, characterizing how se- lective pressures of abiotic and biotic factors contribute to population genetics of plant-pathogenic fungi remains challenging. Macrophomina phaseolina is a seed- and soil-borne fungal pathogen that infects more than ��� host species (Batista, Lopes and Alves, ����), and causes damping o� and charcoal rot in many important economic and subsistence crops worldwide, including soybean (Glycine max) and dry bean (Phaseolus vulgaris) (Dhingra and Sinclair, ����). During host infection, M. phaseolina invades the xylem preventing water uptake, causing wilting and premature plant death with senesced leaves remaining attached to the petioles (Mengistu et al., ����; Romero Luna et al., ����). These symptoms can develop rapidly causing extensive yield loss and grain or seed quality reduction (Smith and Carvil, ����). Charcoal rot of soybean ranked �th out of �� pests and pathogens causing global yield losses higher than �% (Savary et al., ����), with the potential for yield reductions within individual �elds of up to ��% (Wrather et al., ����). In the US, charcoal rot ranked among the top seven most destructive diseases with economic losses totaling ��� billion dollars from ���� to ���� (Allen et al., ����). Disease is favored by hot and dry conditions (Dhingra and Sinclair, ����), with colonization in the soybean and dry bean tap root and lower stem being greatest under high temperatures (��ºC – ��ºC) and low precipitation (Dhingra and Sinclair, ����; Meyer and Sinclair, ����; Kendig, Rupe and Scott, ����; Mengistu, Arelli, et al., ����; Mengistu, Smith, et al., ����; Reznikov et al., ����). Macrophomina phaseolina is haploid, reproduces asexually, and overwinters in soil and crop residue as abundant, melanized microsclerotia that serve as the primary inoculum to initiate infection in subse- quent seasons (Gupta, Sharma and Ramteke, ����; Islam et al., ����). Pycnidia are occasionally produced on soybean and other host plants, however, their epidemiological signi�cance has yet to be fully de�ned (Knox-Davies, ����; Dhingra and Sinclair, ����; Mihail and Taylor, ����; Ma et al., ����; Gupta et al., ����). Depending on environmental conditions, M. phaseolina may survive as microsclerotia in soil for up to �� years (Short et al. ����; Baird et al. ����), and for up to � years as microsclerotia in symptomatic seeds or as mycelium in asymptomatic seeds (Hartman et al. ����). To date, no clonal lineages or patho- �� types have been identi�ed for M. phaseolina, despite reports of within-species variation in morphology and pathogenicity (Dhingra and Sinclair, ����, ����; Sexton, Hughes and Wise, ����). Population ge- netic studies based on microsatellite markers of isolates representing di�erent geographic regions and hosts across the US have found moderate to high genetic diversity and mixed evidence of population structure by host or geography. Although considerable e�orts have been focused on ascertaining host specialization, it is generally concluded that there is no strong evidence of this speci�city, in which iso- lates from one plant species can often cause disease in other plant species (G Su et al., ����; Zveibil et al., ����; Romero Luna et al., ����). Nevertheless, genetic similarity of isolates according to host and US regions and some degree of host preference have been noted (G. Su et al., ����; Jana, Sharma and Singh, ����; Baird et al., ����; Saleh et al., ����; Arias et al., ����). Notably, a group of M. phaseolina isolates obtained from strawberry in California were found to form a species-speci�c cluster, exhibiting strong host preference for strawberry over other hosts around California (Koike et al., ����; A. K. Burkhardt et al., ����). Studying population genetics using statistical methods that leverage genomic, geographic and en- vironmental data can account for continuous and discrete genetic variation and provide insights into the genetic basis underlying environmental adaptation (Hoban et al., ����; Bontrager and Angert, ����; Bradburd, Coop and Ralph, ����b). These approaches may be used to identify environmental factors driving selection and provide an understanding of how and why pathogen populations vary across space. Population genomics and genotype-environment associations have been applied in numerous studies to resolve the basis of rapid adaptation and identify candidate adaptive loci associated with environmental variation (Lasky et al., ����; Forester et al., ����; Xuereb et al., ����; Gibson and Moyle, ����; Capblancq and Forester, ����). However, characterizing population structure and unravelling the e�ects of contin- uous or discrete processes on the genetic di�erentiation remains challenging for many plant-pathogenic fungi. A major challenge arises because continuous geographic di�erentiation (e.g. isolation by distance or climatic variation along a gradient) can be confounded with discrete processes such as admixture and long-distance migration (human-mediated migration) which are commonly observed in plant pathogens �� (Wing�eld et al., ����; Crous et al., ����; Tabima et al., ����; LeBlanc, Cubeta and Crouch, ����). In ad- dition, collinearity between spatial and environmental variables makes it di�cult to elucidate to what ex- tent geographic and environmental di�erences may be contributing to genetic di�erentiation. To address these issues, multivariate statistical methods, speci�cally redundancy analysis (RDA), have been increas- ingly used to disentangle the e�ects of environmental factors in shaping genetic variation. RDA is a type of constrained ordination in which a set of SNPs are modeled as responses in a function of combinations of environmental predictors. Because of its ability to evaluate many loci simultaneously, RDA has been found to be superior to traditional mixed-models associations methods in identifying weak, multilocus selection (Forester et al., ����), suggestive of polygenic adaptation. Furthermore, partial RDA models, in which covariables can be included, has been used to account for underlying population structure in the identi�cation of loci associated with environmental factors for climate adaptation in a variety of systems including plant and animal species (Lasky et al., ����; Forester et al., ����; Xuereb et al., ����; Gibson and Moyle, ����; Capblancq and Forester, ����). Climate �uctuation and temperature in particular, are important abiotic factors leading to local adap- tation of plant-associated fungi (Savolainen et al., ����; Croll and Mcdonald, ����), especially in species occupying spatially and climatically heterogeneous environments (Ellison et al., ����; Branco et al., ����, ����; Fitzpatrick and Keller, ����). M. phaseolina is recognized for its di�erent ecological roles as an en- dophyte, saprotroph, and latent or opportunistic pathogen with broad geographic distribution (Dhingra and Sinclair, ����; Slippers and Wing�eld, ����; Slippers and Boissin, ����; Parsa et al., ����; Crous et al., ����). Worldwide diseases caused by M. phaseolina have re-emerged in recent decades, with outbreaks occurring mostly in tropical and subtropical regions but in temperate regions as well (Leyva-Mir et al., ����; Casano et al., ����; Koehler and Shew, ����; Meena et al., ����; Nishad et al., ����; Tančić Živanov et al., ����; Wang et al., ����). In the US, charcoal rot of soybean has been primarily an issue in south- ern states. However, more recently charcoal rot has been reported in northern states such as Wisconsin, New York, Minnesota, and Michigan (Bradley et al., ����; Brown, ����; Cummings and Bergstrom, ����; Elaraby, ����; Hughes, ����; Yang and Navi, ����). Although many factors may in�uence disease incidence, greater disease and yield losses have been observed in years with high temperature and low soil �� moisture (Bradley and Allen, ����; Allen et al., ����). When comparing isolates from the northern and southern US states, a recent study concluded that M. phaseolina isolates were regionally adapted (Sexton, Hughes and Wise, ����). Investigations in the context of species within Botryosphaeriaceae suggest that geographical distribution and host a�nity dynamics in M. phaseolina are strongly in�uenced by climate due to its broad host range and ecologically diverse roles (Slippers and Wing�eld, ����; Batista, Lopes and Alves, ����). These factors, together with future extreme rainfall and temperature predicted in the climatic change models (IPCC, ����), make it critical to better understand the genetic structure and cli- matic factors as potential selection agents of M. phaseolina. The broad geographic distribution and population dynamics of M. phaseolina suggest that popula- tions in the continental US, Puerto Rico and Colombia might have been in�uenced by a complex envi- ronmental and agricultural landscape and may be structured and di�erentially adapted at a continental or regional level. However, understanding of the population structure of M. phaseolina has remained lim- ited. In the present study, the �rst aim was to better understand the genetic structure in M. phaseolina populations isolated from soybean and dry bean across the US, Puerto Rico and Colombia using genome- wide single nucleotide polymorphisms (SNPs). Speci�cally, the contribution of discrete vs. continuous genetic di�erentiation was assessed and the hypotheses tested were M. phaseolina populations di�erenti- ated (i) between geography and (ii) between host within the US, using conventional and spatially explicit population structure analyses. The second aim was to investigate whether climatic variables contribute to patterns of adaptive genetic variation in M. phaseolina. Using RDA, the hypotheses tested were (i) spe- ci�c climatic variables contribute to genetic variation, (ii) climatic variables independently contribute to patterns of genetic variation when accounting for underlying spatial and population structure, and (iii) loci in strong association with multivariate climate can be identi�ed and have roles in driving local adaptation to climate. �.� Results �.�.� Whole-genome sequencing for �� M. phaseolina isolates Whole-genome sequences were generated for �� M. phaseolina isolates collected across the US, Puerto Rico, and Colombia, including �� soybean isolates, �� dry bean isolates, two strawberry isolates, and one �� Ethiopian mustard isolate (Fig �.�; Supplementary Table A.�). Sequence coverage varied across individual isolates from �X to ��X, across ��% of the M. phaseolina reference genome (JGI Mycocosm, MPI-SDFR- AT-���� v�.�). A total of �.� million SNPs were identi�ed across all isolates, and a mean read depth (DP) of ��X was obtained for all SNPs after �ltering. Most SNPs had a mapping quality (MQ) value equal to �� (��%) and SNPs with MQ values < �� were removed. The distribution of missing data across the isolates and across the variants was even, with most individuals representing similar missing data (� – �.���%), and all variants containing missing data were removed. The �nal data set contained ��,��� high-quality biallelic SNPs in all isolates, and the data set was retained for all analyses. Figure �.� Geographic location of the �� Macrophomina phaseolina isolates overlaid on temperature and precipitation variables. (A) Isolate collection sites overlaid on temperature seasonality (standard devia- tion; ºC). Temperature seasonality contributed the most to explaining patterns of spatial genetic varia- tion using redundancy analysis (RDA). (B) Isolates overlain on precipitation warmest quarter (mm). US, Puerto Rico and Colombia are outlined in black. �.�.� Phylogenomics di�erentiated �� isolates into two main clades of the US and Colombian- Puerto Rican origins To infer the genetic similarity in M. phaseolina isolates across the continental US, Colombia and Puerto Rico, a maximum-likelihood (ML) phylogenetic tree based on the ��,��� SNPs was constructed. Five genetic clusters were identi�ed across the US (n=�), Colombia and Puerto Rico (n=�). Furthermore, a pattern of hierarchical structure di�erentiating the US and Colombian-Puerto Rican isolates was ob- served. The ML tree provided strong support (���% bootstrap) for two main clades, hereafter referred to �� as US and COLPR, and �ve well-supported clades within the main clades (Fig. �.�A). The US isolates M��-�� and M��-�� from California, and TN��� from Louisiana clustered in the COLPR clade, while the Colombian isolates Mph-��, Mph-��, and Mph-�� in the US clade (Fig. �.�A). Other than these six isolates, all isolates from the US were placed in the US clade, and all isolates from Colombia and Puerto Rico were grouped in the COLPR clade. There were three subclades (US�A, US�B and US�) within the US clade and two subclades (COLPR� and COLPR�) within the COLPR clade. The PCA clustered isolates in �ve distinct groups in agreement with phylogenetic analysis, with little evidence of within group di�erentiation (Fig. �.�B). The �rst PC explains most of the variance (��.�%) and separates out isolates in the US clade from the isolates in the COLPR clade, while the second PC explains ��.�% of the variance dividing isolates into the �ve groups in the phylogenetic analysis (Fig. �.�B). An exception was isolate MP���, which in the PCA was grouped in US�B instead of US�A. Since the phylogenetic and PCA clustering revealed essentially the same hier- archical groupings, they were named genetic clusters US�A, US�B, US�, COLPR� and COLPR�. US�A isolates represented the predominant group in the US, with most isolates collected in the East North Central and Central regions in the states of Michigan (��), followed by Wisconsin (��), Indiana (�), Tennessee (�) and Kentucky (�). Cluster US�B was represented by isolates from Mississippi (�) and South Carolina (�). US� isolates represented the second largest group in the US and were mostly collected in the West North Central [Minnesota (�), South Dakota (�)] and South [Texas (�) and Georgia (�)] regions. Also, within this cluster were isolates from Wisconsin (�), Michigan (�), and Kentucky (�). On the other hand, the COLPR� cluster grouped most isolates from Colombia (��) and Puerto Rico (�) while COLPR� grouped isolates from Colombia (�), one isolate from Puerto Rico, and three isolates from the US. No evidence of population structure by states was found, which indicated that states do not represent genetic groups and M. phaseolina is genetically structured at a broader subcontinental regional extent. A ML phylogeny rooted with the M. phaseolina reference genome was reconstructed using the set of high-quality SNPs. The M. phaseolina reference genome was considered as a suitable outgroup based on its European and Arabidopsis thaliana origin. The phylogenetic reconstruction with the reference genome as a root revealed the COLPR� clade as an outgroup to all other clades, while the US clades were �� Figure �.� Population structure of Macrophomina phaseolina in the US, Colombia and Puerto Rico reveals �ve genetic clusters in a pattern of hierarchical structure. (A) Maximum-likelihood phylogeny reconstructed using ��,��� high-quality SNPs. Bootstrap support values over �� are shown at nodes. Bootstrapping converged after ��� replicates. Colored tips represent the genetic cluster for each isolate as de�ned by principal components analysis. The two main clades, US and COLPR, are highlighted by rectangular shading. The country of collection for each isolate is denoted by colored squares at the right bar. (B) Scatterplot from a principal component analysis based on the two �rst PCs (the eigenvec- tors of the SNP dataset) for all isolates. Points are colored by membership in the �ve genetic clusters. Isolate names include states/municipalities codes: CA: California, CAU: Cauca, GA: Georgia, IN: Indi- ana, ISA: Isabela, JD: Juana Diaz, KY: Kentucky, LA: Louisiana, MAG: Magdalena, MI: Michigan, MN: Minnesota, MS: Mississippi, SC: South Carolina, SD: South Dakota, TN: Tennessee, TOL: Tolima, TX: Texas, VAC: Valle del Cauca, WI: Wisconsin. Country codes: US: United States, COL: Colombia and PR: Puerto Rico. reconstructed as terminal clades (Supplementary Fig. A.�). The topology of the rooted ML phylogeny indicated the COLPR clades as more diverse than the major US terminal clades (US�A and US�). This higher diversity in COLPR clades was indicated by longer average branch length than in the US clades, representing a higher average number of substitutions per site. Di�erences in diversity can also be inferred from the PCA clustering. In PC space, �� isolates in US�A and �� isolates in US� genetic clusters clustered e�ectively on top of each other, while isolates in US�B, COLPR� and COLPR�, although projected near �� each other, clustered distinctively more dispersed (Fig. �.�B). The placement of COLPR genetic clusters and their higher diversity as compared to US genetic clusters indicates them as potential sources to the US clusters. To test the relatedness of M. phaseolina isolates from soybean and dry bean in US, the host infor- mation was mapped to the ML tree (Supplementary Fig. A.�A). Generally, isolates that shared a com- mon host did not cluster within genetic clusters in the US. Isolates collected from soybean and dry bean grouped together in the two larger US genetic clusters (US�A and US�; Supplementary Fig. A.�A). This lack of structure was further supported in a PCA showing overlapping ellipses representing ��% of the isolates from each of the hosts (Supplementary Fig. A.�). �.�.� Spatial population structure de�nes discrete population structure in M. phaseolina be- tween the US and Colombia-Puerto Rico and continuous substructure between genetic clusters within US and COLPR clades To infer the number of distinct genetic groups in M. phaseolina while accounting for continuous ge- ographic di�erentiation, spatial analysis of population structure was conducted using a Bayesian (con- Struct) and a model-free matrix factorization (TESS�) framework. Spatial analysis of population struc- ture incorporates geographic distance in the estimation of ancestry coe�cients (the proportion of indi- vidual isolate’s genome originating from the ancestral genetic group, K). The genetic structure of the �� isolates was explained better by a spatial model of admixture between discrete genetic groups, where isola- tion by distance was accounted for rather than the non-spatial model. This was indicated by the increase in predictive accuracy in the conStruct spatial models for all tested values of K (referred hereafter as layers in conStruct framework; Supplementary Fig. A.�B). This suggests that isolation by distance or climatic gradients likely play a role in shaping patterns of genetic variation in the sampled isolates. Spatial population structure description using TESS� returned the greatest decrease in root mean- squared errors at K=� (�.���, from �.��� at K=� to �.��� at K=�; Fig. �.�D) and detected the US and COLPR clades. At K=�, TESS� spatial estimation strongly assigned ��% of isolates to a single ancestral population (ancestry proportion Q > �.�; Fig. �.�A). All isolates in the US clade, except for the three isolates collected in Colombia, were identi�ed as being derived from a single ancestral population (repre- sented by blue; Fig. �A, bottom). Likewise, all COLPR isolates are estimated to have a majority compo- �� nent of ancestry from a single source population (represented by orange; Fig. �.�A, bottom) including the three isolates collected in the US (M��-�� and M��-�� from California, and TN��� from Louisiana). The three isolates collected in Colombia grouping in the US clade (Mph-��, Mph-�� and Mph-��) were identi�ed as admixed (i.e., to have ancestry from more than one population instead of drawing ancestry mostly [Q > �.�] from a single ancestral population) between the two ancestral groups (Fig. �.�A, bot- tom) as well as the two isolates (IN��-�-� from Indiana and Mph-�� from Colombia) placed outside the supported clusters in the ML tree and PCA. At K=�, further substructure was detected that generally re- �ect the genetic clusters within the US and COLPR clades; except that an ancestral population for US�B isolates was not inferred (Fig. �.�B). The decrease in root mean-squared errors at K=� (�.��; from �.�� at K=� to �.�� at K=�; Fig. �.�D) was the second largest value after that at K =�, re�ecting the hierarchical structure observed in previous analyses. However, although isolates in each genetic cluster (except US�B) were inferred as drawing the most ancestry from their own ancestral population, only ��% of isolates had an ancestry proportion (Q) > �.�� to a single ancestral population (Fig. �.�B, bottom), demonstrating weaker assignments than those at K = �. Consistently, the results from conStruct spatial model with K =� returned the greatest increase in predictive accuracy and primarily partitioned the isolates in two main groups mostly in line with US and COLPR clades (Supplementary Fig. A.�A). Based on cross-validation results, the predictive accuracy increased with increasing values of K (Supplementary Fig. A.�B), however additional layers beyond K = � contribute little to total covariance (Supplementary Fig. A.�C). Therefore, supporting two discrete ancestral populations while population substructure can be explained by continuous genetic di�eren- tiation. Taken together conStruct and TESS� results supported two discrete genetic groups for the US and COLPR main clades and suggested that most isolates within US and COLPR clades can be better described to have ancestry mainly from each single ancestral population. It may therefore be reason- able that the evolutionary processes leading to divergence between genetic clusters within the US (US�A, US�B, US�) and COLPR (COLPR� COLPR�) clades were associated to isolation by distance or climatic di�erences rather than di�erent discrete ancestry. �� Figure �.� Spatial population structure de�nes discrete population structure in M. phaseolina between the US and Colombia-Puerto Rico and continuous substructure between genetic clusters. (A) Isolate membership to ancestral populations identi�ed with TESS� using K = � and (B) K = �. Top: Isolate collection sites overlaid on individual membership, each color representing a population. Each point represents an isolate, points are colored by their assignment to genetic clusters as identi�ed in principal component analysis to show agreement between the methods. Bottom: Ancestry proportions (Q) of all isolates. Isolates identi�ed as admixed (Mph-��, Mph-��, Mph-��, Mph�� and IN���-�) are labeled and indicated with dots. (C) Scatterplot from a principal component analysis for all isolates (from Fig. �). (D) Values of the TESS� cross-validation criterion (root mean-squared errors, RMSE) as a function of the number of ancestral populations (K = � to K= �). �.�.� Genetic diversity and di�erentiation between the US and COLPR clades and genetic clusters of M. phaseolina To examine genome-wide diversity of M. phaseolina within and among clades and genetic clusters, we estimated gene diversity (He) and median pairwise genetic distance for each of the clades and genetic clus- ters. Pairwise genetic distance showed that COLPR isolates had greater genetic distances among isolates than those in the US clade, with a gene diversity (He) signi�cantly higher in the COLPR clade (�.���) than the US clade (�.���; Table �.�) (Hs.test, P = �.���). Among clusters, the COLPR� cluster has the highest genetic diversity, considering both gene diversity and pairwise genetic distance, followed by COLPR�, US�B, US�, and the US�A cluster has the lowest values (Table �.�). The higher genetic distance among isolates in the US�B cluster as compared to other US clusters, likely re�ects that the cluster is only represented by �ve isolates of which two were collected in Mississippi, two in Colombia and one in South �� Carolina. Table �.� Summary statistics for genetic diversity of Macrophomina phaseolina clades and genetic clusters. N is number of isolates (sample size); MLG is number of observed multilocus genotypes; eMLG is the number of expected MLG at a sample size of �� for clades and � for genetic clusters based on rarefaction. MLL is number of observed multilocus lineages by population using a bitwise cuto� distance of �.����; CF is clonal fraction (� - (MLL/N). Clone corrected values are shown and indicated by asterisks for indices of genotypic diversity: Shannon-Wiener Index (H*), Stoddart and Taylor’s Index (G*), Simpson’s index (lambda*) and evenness (E�*). Clade, Genetic Gene Median pairwise N MLG eMLG MLL eMLL CF H* G* lambda* E�* Cluster diversity (He) genetic distance US �� �.��� �.��� �� ��.�� �� ��.� �.�� �.�� �.�� � �.��� US�A �� �.���� �.����� �� �.�� � �.�� �.�� �.��� �.�� �.��� �.��� US�B � �.��� �.�� � � � � �.� �.��� �.�� �.�� �.��� US� �� �.��� �.����� �� � � �.�� �.�� �.� �.�� �.��� �.�� COLPR �� �.��� �.��� �� �� �� �� �.� �.�� ��.� �.��� �.��� COLPR� �� �.��� �.��� �� � � �.�� �.� �.��� �.�� �.��� �.��� COLPR� � �.��� �.��� � � � �.�� �.�� �.��� �.�� �.��� �.��� Note: Summary statistics were calculated using the clone-corrected data at �� MLGs. To evaluate genotypic diversity both in terms of genotypic richness (the number of observed geno- types) and evenness of distribution of genotypes, the number of multilocus genotypes (MLG) was calcu- lated for each clade and genetic cluster. A MLG was de�ned as a unique combination of SNPs. Given the large number of ��,��� SNPs and genotyping error rate from NGS data, it is unlikely that a true clone will be represented by an MLG. Thus, to better represent clones, closely related genotypes were collapsed into multilocus lineages (MLLs) based on a Prevosti’s genetic distance threshold of �.���� (� SNPs). Of the �� isolates, �� had unique genotypes (MLGs) corresponding to �� MLLs (Table �.�). eMLG and eMLL are the number of expected MLGs and MLLs based on rarefaction at the lowest common sample size between clades and genetic clusters and were used to allow comparisons across them given their unequal sample sizes. Genotypic richness was highest in the COLPR clade (�� eMLLs) as compared to the US clade (��.� eMLLs). Among genetic clusters, the COLPR� cluster had the highest number of eMLLs, followed by US�, US�B, COLPR� and US�A. This indicates genotypic richness is highest in COLPR� and lowest in the US�A genetic cluster, in which more than ��% of the isolates were clonal (Table �, CF). Although, lower genotypic richness is inferred in COLPR� and US�B as compared to the gene diversity pattern, this may be due to their low sample size. Evenness and the corrected Shannon-Wiener’s index, �� Stoddart and Taylor’s index and Simpson’s Index, were all highest in the COLPR clade than in the US clade and followed the same pattern among genetic clusters as with genotypic richness (Table �.�). Finally, there were no shared MLGs or MLLs among genetic clusters. Similarly, between countries, signi�cantly higher gene diversity in Colombia (�.���) compared with the US (�.���) (Hs.test, P = �.���). Gene diversity in Puerto Rico (�.���) was intermediate and not sig- ni�cantly di�erent from the US (Hs.test, P = �.���) or Colombia (Hs.test, P = �.���). Pairwise genetic distances, corrected genotypic diversity indices and evenness calculated for each country follow the same pattern of gene diversity (Supplementary Table A.�). To infer migration among countries by tracking genotype �ow, MLLs shared among countries were identi�ed. In total three MLLs were shared among countries. The MLL with one isolate from Colombia (Mph-�) and one from Puerto Rico (UPR-Mph- JD�) clustering in COLPR�, the MLL with one isolate from Puerto Rico (UPR-Mph-ISA�) and one from Louisiana (TN���) clustering in COLPR�, and the MLL with one isolate from Colombia (Mph- ��) and �� isolates from US clustering in US�A (Supplementary Fig. A.�). In addition, all populations clustering approaches indicated that Colombian isolates Mph-�� and Mph-�� are the most closely related to the US isolates clustered in US�B, and Californian isolates M��-�� and M��-�� are the most closely re- lated to Colombian isolates clustering in COLPR�. The rooted ML tree indicated isolate Mph-�� (from Colombia) as an outgroup to US clusters and discriminatory analysis of principal components (DAPC) clustered this isolate along with IN��-�-� (from Indiana) with US�B isolates (Supplementary Fig. A.�). Overall, migration between Colombia, Puerto Rico and US is a likely scenario. To test the hypothe- sis that genetic clusters of M. phaseolina are di�erentiated, we used hierarchical analysis of molecular variance (AMOVA) and Nei’s GST (an FS T -analogous genetic di�erentiation measure applicable to hap- loids). Populations were signi�cantly di�erentiated among clades, genetic clusters, as well as within ge- netic clusters (P < �.���; Supplementary Table A.�). AMOVA revealed that most of the total genetic vari- ance was partitioned among US and COLPR clades (��%) and among genetic clusters (��%), and only ��% within genetic clusters. Consistently, very high genetic di�erentiation was found between US and COLPR clades (GST = �.��) and among genetic clusters (GST = �.�� – �.��; Table �.�). The COLPR� (GST = �.��-�.��) and US�B (GST = �.��-�.��) clusters had the lowest GST when compared with any �� other cluster. Di�erentiation was lowest between COLPR� – COLPR� (GST = �.��) clusters, and US�A – US�B (GST = �.��) and highest between COLPR� – US�A (GST = �.��), COLPR� – US� (GST = �.��) and US�A – US� (GST = �.��). Table �.� Population di�erentiation using Nei’s GST pairwise genetic dissimilarity between genetic clus- ters identi�ed in Macrophomina phaseolina. Genetic Cluster US�A US�B US� COLPR� COLPR� US�B �.�� US� �.�� �.�� COLPR� �.�� �.�� �.�� COLPR� �.�� �.�� �.�� �.�� All other pairwise comparisons had similar intermediate levels of genetic di�erentiation when com- pared to any other genetic cluster (GST = �.��-�.��). The high values of GST in all pairwise compar- isons suggest very high di�erentiation and little migration between genetic clusters. However, US�A – US� GST estimation, which is notably high, was limited in power due to the low levels of gene diversity (Hexp) within these genetic clusters. Across the ��,��� loci, there were only �� and ��� polymorphic loci within US�A and US� clusters, respectively. Thus, low gene diversity (Hexp) in US�A and US� subpopu- lations likely resulted in overestimation of GST in pairwise comparisons of US�A and US� with all other clusters. �.�.� M. phaseolina is predominantly clonal in the US and semi-clonal to mostly-clonal in Colombia and Puerto Rico The predominantly star-like topology with little reticulation, in the Neighbor-Net network analysis, is consistent with a clonally reproducing population (Fig. �.�A). The standardized index of association (IA ) (Brown et al. ����) was used to estimate the degree of clonality for each of the M. phaseolina main pop- ulations (US and COLPR clades). The observed IA distributions for each population were compared to IA distributions for simulated populations with no linkage, ��%, ��%, ��% and ���% linkage. A predom- inantly clonal mode of reproduction was inferred in the US and COLPR populations of M. phaseolina. The simulated distributions and the di�erent populations were signi�cantly di�erent from each other (analysis of variance ANOVA df = �, F = �����, P < �.���). The distribution of the standardized IA for �� the US population fell within the ��% to ���% range of the linkage simulation (Fig �.�B). This indicates a mostly clonal mode of reproduction with little potential for recombination. The distribution of the standardized IA for the COLPR population fell within the �� to ��% range of the linkage simulation, in- dicating semi-clonal to mostly clonal reproduction in COLPR clades (Fig. �.�B). To further investigate the extent to which populations reproduce clonally, the linkage disequilibrium (LD) decay, as measured by the squared correlation coe�cient (r� ) was calculated across pairs of loci for each of the clades. LD extends across a much larger distance in the US clade than in the COLPR clade, decaying over the �rst thousand base pairs, while in the COLPR clade LD decayed over the �rst hundreds of bases. LD half- decay distance, calculated as the average physical distance over which r� decays to half of its initial value was ���� bp for US clade and ��� bp for COLPR clade (Fig. �.�C). This indicates a high level of linkage occurs over larger regions of the genome in the US clade versus the COLPR clade. Importantly, although this may provide evidence for less clonal reproduction and higher recombination rates in the COLPR population, interpretation of standardized IA and LD decay as associated with the frequency of recombination should be done with caution. It is possible that higher LD values did not re�ect greater recombination; instead, it may be a�ected by lower sample size in COLPR and lower diversity in the US clade. �.�.� Climate contributes to SNP variation between M. phaseolina genetic clusters To test the hypothesis that climate variation contributes to genetic variation across M. phaseolina ge- netic clusters a redundancy analysis (RDA) was employed. Four climatic variables were identi�ed as sig- ni�cantly predictive of genetic variation using the simple RDA model with forward variable selection. Temperature seasonality (TSsd) was the strongest predictor, explaining ��% of the variation, followed by precipitation of warmest quarter (Pwq), precipitation seasonality (PScv) and mean temperature of warmest quarter (mTwq) (Table �). Importantly, the climatic variables included in the RDA model were selected by their biological signi�cance and to avoid collinearity with other climatic variables and thus represent a subset of the variables possibly contributing to climate variation. The correlation of these variables with the �rst two RDA axes suggests their di�erential contribution to SNP variation among ge- netic clusters (Fig. �.�). Spatial structure, represented as distance-based Moran’s eigenvectors maps (db- �� Figure �.� Macrophomina phaseolina population structure is potentially driven by clonal expansions and rapid divergence. (A) A reticulating phylogenetic network. Neighbor Net method was used to depict con�icting phylogenetic signal. (B) Estimates of linkage disequilibrium for Macrophomina phaseolina clades based on observed and simulated distributions of the standardized index of association (IA ). Each boxplot represents the observed distribution of IA for one of the clades of M. phaseolina, compared with the distribution of IA values for simulated populations with no linkage and ��, ��, ��, and ���% linkage. The letters above each boxplot represent groupings based on Tukey’s HSD test . (C) Linkage disequi- librium (LD) decay for predicted populations of M. phaseolina, as measured by the squared correlation coe�cient (r� ) for all pairs of SNPs calculated over �� bp windows shown for each population. The dot- ted black lines give the r� decay to half its initial value (r� = �.�� and �.�� in US and COLPR clades, respectively) and the vertical lines indicate the LD half- decay distance for each clade. MEM), was used to identify climatic variables that are structured in space and to account for the e�ect of space in variance partitioning of total genomic variation. A total of three spatial variables were identi- �ed (dbMEM�-�; Supplementary Fig. A.�). Notably, when accounting for spatial structure (dbMEM�-� variables), only Pwq, mTwq and precipitation of driest quarter (Pdq) were signi�cant and accounted for �% of SNP variation across isolates as determined with forward selection (Table �), indicating collinearity �� between TSsd, PScv and space (i.e., spatially structured TSsd and PScv variation). To identify the spatial variables signi�cantly contributing to genomic variation forward selection was used. Of the three spatial variables, only dbMEM� was signi�cant explaining �% of the genomic variation and described broad-scale spatial structure (Supplementary Fig. A.�) Figure �.� Genotype-environment association analyses support the contribution of climate variables to patterns of divergence among Macrophomina phaseolina populations across the US, Colombia, and Puerto Rico. Biplot of all isolates scores for the �rst two RDA axes using (A) Simple RDA (uncondi- tioned) and (B) Partial RDA (conditioned on neutral population structure). Points are colored to show agreement with genetic clusters identi�ed in the PCA (inset). Top and right axes (blue) indicate the cor- relation of each climate variable with RDA axes � and �, respectively. Table �.� Climatic variables signi�cantly contributing to SNP variation as determined by forward variable selection with simple RDA (redundancy analysis) and partial RDA conditioned on space. Simple RDA (unconditioned) Partial RDA (conditioned on space) Variable R� Cum R� Cum R� adj F-value p-value Variable Cum R� adj AIC F-value p-value TSsd �.�� �.�� �.�� ��.�� �.���*** Pwq �.�� ���.�� �.�� �.���** Pwq �.�� �.�� �.�� �.�� �.���*** mTwq �.�� ���.�� �.�� �.���** PScv �.�� �.�� �.�� �.�� �.���*** Pdq �.�� ���.�� �.�� �.���** mTwq �.�� �.�� �.�� �.�� �.���** ***p �.���, **p �.�� Partial redundancy analysis (pRDA) was used to estimate the partial contribution of each set of ex- planatory variables (e.g., climate) while removing the e�ect of the remaining variable sets (e.g. neutral �� population structure and space). Variance partitioning with pRDA revealed that climate (TSsd, Pwq, PScv and mTwq identi�ed by forward selection), neutral population structure (isolate PC scores for the �rst three axes of a PCA using intergenic SNPs) and space (dbMEM� variable identi�ed by forward se- lection) together signi�cantly explained ��% of the total SNP variance. Nearly half of this variance was uniquely attributable to neutral genetic structure (��%), climate (�%), or space (�%), while the other half of the SNP variation was explained jointly between the three sets of variables (Table �.�). The e�ect of climate alone was highly signi�cant and explained �% of the total genetic variance after removing the e�ects of neutral population structure and space (Table �.�). These results support the hypothesis that climate signi�cantly contributes to genetic variation and importantly, suggests that migration, drift, and potentially additional demographic and spatially structured processes (e.g isolation by distance), repre- sented by neutral population structure, play a major role in shaping genomic variation in M. phaseolina. Moreover, the large fraction of variation common to climate, population structure and space, emphasizes the importance of accounting for confounded e�ects in genotype-environment associations, particularly when inferring causal associations. Table �.� Contribution of climate, neutral population structure and space to SNP variation as deter- mined by variance paritioning with partial RDA (redundancy analysis). Inertia Proportion of Proportion of Partial RDA model R� p-value (variance) explainable variance total variance Full model: G ⇠clim. + sp. + struct. ���.� �.��� �.���*** �.�� �.�� Pure climate: G ⇠clim. | (sp. + struct.) ��.� �.��� �.���*** �.�� �.�� Pure structure: G ⇠struct. | (clim. + sp.) ���.� �.��� �.���*** �.�� �.�� Pure space: G ⇠sp. | (clim. + struct.) �.� �.��� �.���*** �.�� �.�� Confounded climate/structure/space ���.� �.�� �.�� Total unexplained ���.� �.�� Total inertia ����.� �.�� ***p �.��� Note: Climate variables are temperature seasonality (TSsd), precipitation of warmest quarter (Pwq), precipitation seasonality (PScv) and mean temperature of warmest quarter (mTwq) as identifed with forward selection. �.�.� Genotype-environment associations identify candidate SNPs for climatic adaptation To identify loci that are potentially involved in local adaptation to climatic conditions, SNPs strongly associated with climatic variables were identi�ed using RDA with and without accounting for population �� structure. Neutral population structure was used as it uniquely contributed the most to genetic variation. The RDA models, whether accounting for population structure (partial RDA) or not (simple RDA), were globally signi�cant (p < �.���) and the �rst three RDA axes explained most of the genomic variation associated with climate. The candidate adaptive loci were identi�ed based on extreme SNPs loadings, ±� or ±� SD from the mean, on each of the �rst three axes (Forester et al., ����). In the partial RDA models, in which the ef- fects of population structure were removed, �� unlinked SNPs (when using the LD-�ltered set and ±� SD from the mean; Supplementary Table A.�) and �� SNPs (using all SNPs and ±� SD from the mean; Supplementary Table A.�) strongly associated with climatic variables were identi�ed along the �rst three RDA axes. Of these SNPs, �� and �� (outliers in Fig. �.�) were identi�ed in the �rst RDA axis when using the LD-�ltered set or all SNPs, respectively, and �� (��%) in both partial models. The strongest associations include SNPs with predicted e�ects in the membrane-associated ������-ankyrin, the ������- Ksh� and the ������-protoporphyrinogen oxidase proteins. Structural modeling of the ������-ankyrin protein revealed that ��� residues (��% of the sequence) was modelled with ���% homology con�dence to the transient receptor potential (TRP) NOMPC (No mechanoreceptor potential C) mechanotrans- duction channel protein in Drosophila melanogaster (chain C, highest scoring template; PDB ID: �VKQ; data not shown). Other SNPs with top associations are located within or in physical proximity to genes related to transmembrane transport, glycoside hydrolase activity, DNA binding and the gene encoding the �����-heat shock protein (Table �.�; Supplementary Table A.�). Because population structure could not be fully disentangled from climate, as revealed in variance partitioning, the candidate loci obtained with population structure correction represent a conservative set subjected to a reduction in the detection of SNPs truly associated with climate. In the simple RDA model, without correcting for population structure, �� candidate unlinked SNPs were identi�ed (Supple- mentary Table A.�). Only two SNPs were identi�ed by both partial RDA and simple RDA models using unlinked SNPs (Supplementary Fig. A.�). This is in line with the high level of collinearity observed be- tween genetic, space and climate (Table �.�), and highlights the importance of accounting for confounded e�ects when identifying candidate loci under selection with genotype-environment associations. �� Figure �.� Manhattan plot of partial RDA scores. Values of squared SNP loadings for the �rst RDA axis conditioning on neutral population structure. (A) Fifteen outlier SNPs identi�ed using using ��,��� unlinked SNPs and ±� SD from the mean and (B) Twenty-�ve using all ��,��� SNPs and ±� SD from the mean. Table �.� Candidate SNPs and gene models along the �rst RDA axis, after accounting for neutral popu- lation structure using the LD-�ltered set of ��,��� SNPs. RDA� Climate SNP Distance from Mycocosm gene Mycocosm SNP position Correlation InterPro/KOG Desc KOG Class/Putative function loading variable category locus (bp) location protein ID sca�old_��:������ -�.��� TSsd �.�� Intergenic ���� sca�old_��:������-������ ������ Ankyrin repeat Cell wall/membrane/envelope biogenesis Involved in the early part of the secretory sca�old_��:������ -�.��� TSsd �.�� Intergenic ���� sca�old_��:������-������ ������ Ksh�(Protein kish) pathway Protoporphyrinogen Coenzyme transport and metabolism/Heme sca�old_�:������� �.��� TSsd �.�� Missense � sca�old_�:�������-������� ������ oxidase biosynthesis sca�old_�:������� -�.��� TSsd �.�� Intergenic ���� sca�old_�:�������-������� ������ None Unknown sca�old_�:������� -�.��� TSsd �.�� Intergenic ��� sca�old_�:�������-������� ������ None Unknown Transcription factor DNA binding/Zinc ion binding (Zn(II)�Cys� sca�old_��:������ -�.��� TSsd �.�� Intergenic ��� sca�old_��:������-������ ����� domain, fungi transcription factor-related) Unknown/Putative transcription sca�old_�:������� -�.��� TSsd �.�� Synonymous � sca�old_�:�������-������� ����� GXWXG domain factor Cmr� homolog AMP-dependent sca�old_��:����� -�.��� TSsd �.�� Synonymous � sca�old_��:�����-����� ������ Lipid transport and metabolism synthetase/ligase sca�old_��:������ -�.��� TSsd �.�� Intergenic ���� sca�old_��:������-������ ������ DUF���� family Unknown Glycoside hydrolase, sca�old_��:������ -�.��� TSsd �.�� Intergenic ���� sca�old_��:������-������ ������ Carbohydrate transport and metabolism family � Allergen V�/Tpx-�-related, sca�old_�:����� �.��� TSsd �.�� Intergenic ���� sca�old_�:�����-����� ������ Unknown conserved site sca�old_�:������� �.��� TSsd �.�� Intergenic ����� sca�old_�:�������-������� ������ Ribonuclease T�-like RNA processing and modi�cation sca�old_�:������� �.��� TSsd �.�� Intergenic ��� sca�old_�:�������-������� ������ Thioesterase superfamily Unknown Cytochrome P���, E-class, sca�old_��:������ �.��� mTwq �.�� Synonymous � sca�old_��:������-������ ������ Lipid transport and metabolism group I Flavin-containing Secondary metabolites biosynthesis, sca�old_��:������ �.��� mTwq �.�� Intergenic ���� sca�old_��:������-������ ������ monooxygenase transport and catabolism �.� Discussion In this study, we describe the population structure of M. phaseolina in the continental US, Puerto Rico and Colombia collected from soybean and dry bean �elds and the contributions of climatic factors to pat- terns of genomic diversity among populations. We found that �ve distinct genetic clusters of M. phase- olina evolved across the US, Colombia and Puerto Rico and evidence suggests migration between genetic clusters and countries. To date, population genetic studies in M. phaseolina have performed their anal- yses at the resolution of microsatellites molecular markers and have provided important information on genetic diversity, host and geographic associations in the US (Baird et al., ����; Arias et al., ����; Koike et �� al., ����). However, no population-level genomic studies have been conducted to investigate population structure in this widespread pathogen. Here, to our knowledge, we present the �rst population genomics study to investigate population dynamics and the role of climate in shaping patterns of genomic varia- tion in M. phaseolina at a continental and regional scale. This study uses population genomics data to identify multiple strongly di�erentiated genetic lineages in the US and demonstrated novel population structure in Colombia and Puerto Rico, which previously remained unstudied. Furthermore, our re- sults highlight the importance of within-species genetic variation in understanding pathogens adaptive response to a changing climate and o�ers new insight with respect to the functional roles of genomic regions potentially underlying adaptation to climate. Notably, this research provides a practical frame- work for genotype–environment associations studies in M. phaseolina and other plant pathogens with complex evolutionary and demographic histories. The in�uence of the low number of loci on limiting inferences about M. phaseolina population struc- ture is emphasized by recent studies that used microsatellites markers (Baird et al., ����; Arias et al., ����). These studies identi�ed genetic groups in the US; however, the genetic groups did not represent lin- eages (i.e., genetic groups and supported phylogenetic clades). Using population genomics, we provided strong evidence for �ve distinct genetic clusters of M. phaseolina and revealed that genomic variation in this globally distributed pathogen was consistent with a population hierarchically structured at a broad subcontinental regional extent. Two genetically di�erentiated M. phaseolina populations at the US and Colombian-Puerto Rican geographical level (US and COLPR clades) and �ve distinct genetic clusters representing �ner population structure within each of these clades were identi�ed. These genetic clus- ters, except for US�B, represent strongly supported phylogenetic clades and monophyletic groups, and likely represent di�erent evolutionary lineages of M. phaseolina. This distinction is important because the identi�cation of lineages allows the inference of ecological and evolutionary processes in a population- speci�c manner and underscores the potential for local adaptation in M. phaseolina populations. Our results provide support for regional clustering within the US and a lack of strong grouping at a state level, also observed in previous studies based on microsatellite data (Baird et al., ����; Arias et al., ����). The US�A cluster, found in the East North Central and Central region, expands previous studies �� con�rming that isolates collected from soybean in these regions represent a largely homogeneous popu- lation (Arias et al., ����). This is supported by low gene diversity and pairwise genetic distances found in the US�A genetic cluster in agreement with low diversity detected with microsatellite markers in soybean isolates collected mostly in Tennessee and Missouri (Arias et al., ����) and midwestern states (group III; Baird et al., ����). The US� genetic cluster found in West North Central and South US regions grouping isolates from Minnesota, South Dakota, Texas, and Georgia is partially consistent with Baird et al. study. Isolates from these states along with isolates from North Dakota represent the majority of a subcluster of group I in Baird et al. Like in the US clusters, grouping at broad geographic regions was observed in COLPR� and COLPR� clusters. Both COLPR� and COLPR� clusters grouped isolates from locations across Colombia and Puerto Rico. In COLPR�, isolates from California and Louisiana grouped closely to isolates from Colombia and Puerto Rico. Although the small sample size from these states (only two isolates collected from strawberry in California and one isolate from soybean in Louisiana) demands that this grouping be reassessed once more isolates are included from these states and hosts in future studies. The clustering of isolates from widespread geographic regions observed in COLPR�, as well as in US�A and US�B clusters, suggests a role for migration in structuring M. phaseolina populations. These results better align our understanding of M. phaseolina population structure with a metapopulation model, that predicts regional persistence of populations while local populations are unstable and connected by some level of migration (Hanski, ����; Milgroom, ����). The metapopulation dynamics view expands the interpretation of past M. phaseolina population structure studies while providing a conceptual basis for the design of future studies. The presence of multiple distinct genetic clusters in the US and higher genetic diversity in COLPR clusters led us to inquire about whether Colombia and Puerto Rico may serve as potential source pop- ulations for US populations. In the rooted ML phylogeny, the reconstruction of COLPR clusters as outgroups to US clusters support this hypothesis. Furthermore, across all analyses we found indications that US�B may serve as a sink population for Colombia and Puerto Rico populations. The US�B genetic cluster grouped isolates from Mississippi and South Carolina along with two Colombian isolates and was the most genetically diverse of the US clusters. Further, US�B was positioned centrally in PCA space, �� basal to US�A cluster in the rooted ML phylogeny and was less di�erentiated, along with COLPR�, from all other clusters based on GST values. Finally, in DAPC analysis, US�B isolates clustered with IN��-�-� and Mph-�� isolates, which are reconstructed intermediate between US and COLPR clades in the rooted ML phylogeny and as admixed in spatial population structure analyses. Although, the high diversity in US�B may be re�ective of the grouping of comparatively few isolates from di�erent geographic regions in this cluster. However, when all data are considered, it suggests the US�B cluster geographic region as a potential route of introduction of isolates from Colombia or Puerto Rico to the US. More isolates from the US and other countries would need to be included in future studies to test this hypothesis. The discrete population structure observed between US and COLPR clades, provides compelling evidence for isolates in each clade drawing ancestry from di�erent ancestral populations. A plausible ex- planation, supported by our results, for this di�erent ancestry would be a demographic event such as a rare long-distance migration (e.g. introduction event) from the COLPR clusters, leading to a recent bottle- neck in the US populations. The high probability assignments observed in US clusters may be consistent with the expected strong recent genetic drift in bottlenecked populations (Lawson, van Dorp and Falush, ����). In this scenario, we speculate that the diversity in US clusters represent a subset of the diversity of the COLPR genotypes found in Colombia and Puerto Rico. At the �ner genetic cluster population structure, isolation by distance provided a potential explanation for the continuous genetic di�erentia- tion in spatial population structure analyses. Although, isolation by distance patterns may be observed as part of a variety of underlying biological processes and demographic scenarios (Sexton, Hangartner and Ho�mann, ����; Milgroom, ����), it is possible that these patterns re�ect a scenario of restricted disper- sal in the context of divergence following clonal expansions in the US genetic clusters. For example, both US�A and US� genetic clusters are found in Michigan, Wisconsin, and Kentucky, supporting dispersal of isolates among these states. However, high population di�erentiation indicated by high GST values between genetic clusters, suggest substantial restriction to gene �ow. Given the soilborne nature of M. phaseolina and limited natural dispersal ability but high potential for anthropogenic mediated dispersal, restricted events of dispersal associated to seed, plant material or farm equipment at limited distances rel- ative to the geographic range of the genetic clusters, seems a likely occurrence (Baird et al., ����). Similar �� isolation by distance patterns has been observed in other soilborne fungal and oomycete pathogens with restricted long-distance dispersal (Grünwald and Hoheisel, ����; Milgroom et al., ����). Diversity was found to be further reduced in US�A genetic cluster as compared to all other clusters. Low diversity and high di�erentiation are signatures of genetic drift but also selection. If reduced diversity in the US�A genetic cluster was consistent with a clonal expansion following a bottleneck, the divergence and marked low diversity could re�ect both genetic drift and selection. Genetic drift is expected to have substantial e�ects on pathogen populations, because migrations resulting in founder e�ects and reduced population sizes associated with pathogens survival in soil (Milgroom, ����). Additionally, we speculate that climatic conditions, particularly strong �uctuations in temperature in the northern US, could im- pose strong selection on M. phaseolina populations in this region. Overall, we believe the genomic signals of discrete and continuous structure that di�erentiate M. phaseolina populations could be re�ective of a complex demographic and evolutionary history. Therefore, alternative demographic scenarios, includ- ing one of multiple independent introductions, should be considered in future studies ideally applying demographic modelling with a broad geographic and temporal distribution of isolates. Across all analyses we found support for Colombia and Puerto Rico as potential sources for US M. phaseolina populations. Genetic diversity between countries also supported this hypothesis. Whereas Colombian isolates were signi�cantly more diverse than US isolates, diversity in Puerto Rico was inter- mediate and not signi�cantly di�erent from US or Colombia. These �ndings may be consistent with the idea of Middle or South America as putative centers of origin for M. phaseolina and with its intro- duction to North America as part of historical crop migrations. For example, common bean Middle American origin, domestication centers in Middle America and South America (Bitocchi et al., ����) and later movement to the US via the Caribbean, Central and Eastern US (Kelly, ����), makes likely an explanation for M. phaseolina introduction to the US in bean seeds. Pathogen geographic origins have been associated with the centers of diversity of their major crop host. Nonetheless, pathogen origin asso- ciated with their hosts’ wild relatives, have been also observed in some plant pathogens. For example, a P. infestans genetically diverse and sexually reproducing population was found in central Mexico consistent with this pathogen’s origin in a secondary center of potato (Solanum tuberosum) diversity and potentially �� involved in a host jump from native Solanum species (Goss et al., ����). Given M. phaseolina host gener- alist nature, a strict host-pathogen coevolution scenario is not expected (Slippers and Wing�eld, ����), obscuring inferences about its center of origin. In Kansas, isolates collected from wild tallgrass prairie were found more diverse than isolates from maize, soybean and sorghum crops (Saleh et al., ����). This �nding may indicate M. phaseolina presence in the US precedes to the introduction of agriculture or it may be explained by connectivity dynamics between natural and agricultural ecosystems contributing to patterns of diversity in M. phaseolina populations from these ecosystems (Saleh et al., ����). Thus, the origin and evolutionary history of M. phaseolina is likely more ancient and complex than could be tested with the isolates included in this study, and future studies may bene�t from considering the potential involvement of host adaptation from wild hosts. Genotype tracking provided compelling evidence for migration among the US, Colombia, and Puerto Rico. The MLL consisting of the Colombian isolate Mph-�� and several isolates from the US clustering in US�A, along with the high clonality found in this cluster and the signi�cantly high diversity in Colom- bia, makes a Colombian source likely. Similarly, the MLL shared between Colombia and Puerto Rico and the MLL between Puerto Rico and Louisiana support migration between countries. Alternatively, the same MLLs could have been introduced independently to US, Puerto Rico, and Colombia, poten- tially from an ancestral and more diverse population not included in this study. Although this scenario seems less likely, it remains a possibility. Given that besides historical crop migrations, migration as part of international seed exchange is a likely occurrence in M. phaseolina, as in other seedborne species and latent pathogens of the Botryosphaeriaceae family (Sakalidis et al., ����; Crous et al., ����), we believe that M. phaseolina has been spread at least intercontinentally, possibly globally, through seed. However, time, frequency, and directionality of migration between US, Colombia, and Puerto Rico, and the potential for multiple introductions would need to be examined in future studies. Although various population genetic studies in M. phaseolina have found patterns of host associa- tions (Jana, Sharma and Singh, ����; Baird et al., ����; Arias et al., ����; Koike et al., ����; Reznikov et al., ����; A. Burkhardt et al., ����), our results did not �nd that genetic variation is associated with host in the two major US clusters. Soybean and dry bean isolates grouped together in US�A and US� clus- �� ters. Given that most previous studies support some degree of host preference, and genomic evidence for genes uniquely present in the M. phaseolina strawberry genotype further support host preference (A. K. Burkhardt et al., ����), we suspect that our sampling scheme was not enough to capture clear associations to plant host. A clear limitation in our study was that the host origin was confounded with geographic origin, except for Michigan where isolates were sampled from both soybean and dry bean. The grouping independently of host might also re�ect crop rotation and equipment practices implemented in �elds. Additionally, it may re�ect that the sampled hosts are both legumes. Genetic similarity has been found to be greater among isolates collected from the same host than from hosts in di�erent families (G. Su et al., ����; Saleh et al., ����). These results do, nonetheless, have important practical implications for soybean breeding resistance to charcoal rot. In the US�A cluster, the high genetic similarity of isolates collected from soybean and dry bean, may indicate that the use of one or few isolates collected from these crops throughout East North Central and Central US regions may su�ce for resistance screening of soybean breeding material. An important limitation to this assumption is that we use a single reference genome approach to characterize genetic diversity and thus accessory genes and other structural variation poten- tially involved in pathogenesis are not considered (Bertazzoni et al., ����). Importantly, the dry bean diversity in research plots from which Colombian and Puerto Rican iso- lates were collected is a factor likely contributing to their higher genetic diversity as compared to US iso- lates. In research plots, multiple lines are continually evaluated as part of breeding programs, in contrast to commercial �elds in which a single or few varieties are used. This coupled with climatic conditions in Colombia and Puerto Rico that favor year-round inoculum presence in crop residue represent important considerations when interpreting isolate genetic diversity in relation to host origin. The population structure results suggest that M. phaseolina populations lay in-between the clonality- recombination spectrum (Smith et al., ����). Furthermore, our results suggest that this may occur in a population-speci�c manner. On one side of the spectrum, we found M. phaseolina to have a markedly clonal population structure (Milgroom, ����). First, most of the intraspeci�c genetic variation in M. phaseolina is explained by di�erences between clades and genetic clusters, while low genetic variation was observed within genetic clusters. Second, the occurrence of nearly identical genotypes (i.e., MLLs) from �� widespread geographic locations found in M. phaseolina is in line with a markedly clonal population structure (Milgroom, ����). On the other end of the spectrum genotypic diversity, network analyses and measures of linkage among loci provided support for recombination within some of the genetic clusters. High levels of genotypic diversity is one of the characteristics re�ective of recombination in fungal popu- lations (Milgroom, ����). The higher genotypic diversity (eMLLs) in US�B, US�, and COLPR clusters, may be consistent with the occurrence of recombination in these clusters. Network analyses account for recombination by allowing to infer homoplasy caused by recombination. The boxes between isolates within genetic clusters in the network and the PHI test supporting recombination within all clusters ex- cept for US�A, strengthen this hypothesis. The index of association, IA , revealed an overall high degree of linkage among SNP markers, in line with a pathogen that reproduces clonally. However, the observed IA values in the COLPR clade and LD decaying faster in COLPR than in US populations, support the potential occurrence of recombination among isolates within COLPR clusters. Although the problem of smaller sample size in COLPR clusters should be at least partially accounted for by using simulations in IA analysis and clone-corrected data in LD-decay analysis, particularly half-decay LD values should be interpreted with caution and examined in future studies to determine the extent of recombination in M. phaseolina populations. These results are consistent with the population structure model that lays in between the “strictly clonal” and “epidemic” structure proposed by Maynard Smith et al., in which frequent recombination does not occur between isolates in separate branches of an evolutionary tree but it occurs between iso- lates within a given branch (Smith et al., ����). These models have been used to describe the population structure of plant pathogens with mixed modes of reproduction or inferred recombination (Grünwald and Hoheisel, ����; Milgroom et al., ����; Milgroom, ����; Milgroom et al., ����). While little is known about the occurrence of recombination in M. phaseolina, recent studies have started to shed light on potential recombination mechanisms involving parasexuality (Pereira et al., ����) and horizontal gene transfer mediated by giant mobile genetic elements (Gluck-Thaler et al., ����). Whether other poten- tial recombination mechanisms occur, and the frequency of recombination in M. phaseolina remains an important and exciting area of study. �� Partial RDA revealed that nearly half of the SNP variance is confounded between neutral genetic structure, climate, and space. This means that this fraction of the variance cannot be statistically associ- ated to a direct e�ect of any single set of variables. Importantly, the e�ects of population structure and space often cannot be independently disentangled from spatially structured process (e.g IBD) or spatially structured environmental variables (Lasky et al., ����). This study, while highlighting the challenges in assessing genotype-environmental associations, provided an assessment of the fraction of confounded variance and allowed us to start disentangling the e�ects of climate, spatial, and population structure on genomic variation in M. phaseolina populations. The genotype-environment association analyses us- ing partial RDA support our hypothesis that local climatic di�erences contribute to patterns of adaptive divergence among M. phaseolina populations across the US, Colombia, and Puerto Rico. Seasonal varia- tion in temperature and precipitation of warmest quarter, were the primary climatic variables associated with variation of candidate adaptive loci without and after accounting for neutral genetic population structure, respectively. We found SNPs within or in physical proximity to genes with functional annota- tions related to transmembrane transport, glycoside hydrolase activity and DNA binding. In fungi, genes involved in these activities are known to be important in responses to environmental stressors including temperature, water availability, and oxidative stress (Aguilera, Randez-Gil and Prieto, ����; Gasch, ����; Branco et al., ����). Similarly, among the candidates, we found the ������-protoporphyrinogen oxidase protein, involved in heme biosynthesis and the putative small heat shock protein �����-Hsp��. Heme has been shown to regulate several mechanisms during cold-shock in Saccharomyces cerevisiae (Abramova et al., ����) while Hsp�� proteins have been found involved in fungal thermal stress response to both heat and cold (Wu et al., ����; Wang et al., ����). The SNP with the highest correlation with temperature seasonality was located upstream to the ������-ankyrin repeat protein (Table �). We found that M. phaseolina ������-ankyrin protein is a pre- dicted homologous to the TRP NOMPC mechanotransduction channel in Drosophila melanogaster (Jin et al., ����). Ankyrin family proteins link membrane proteins, including ion channels, to microtubules of the cytoskeleton by binding of its ankyrin repeat domain. The ankyrin proteins in the NOMPC channel link a displacement of the cytoskeleton to the channel opening, translating external stimuli into intra- �� cellular signals (Jin et al., ����). Moreover, the TRP� (transient receptor potential �) ion channel from the alga Chlamydomonas reinhardtii, which shares structural homology to the TRP NOMPC channel, was found to act as thermal sensor, with ankyrin proteins mediating the channel opening in response to increased temperature (McGoldrick et al., ����). Although there is no structural or functional charac- terization of the M. phaseolina ������-ankyrin protein, it represents a promising candidate to investigate a potential temperature-related mechanism for environmental stimuli transduction. These �ndings are consistent with the established roles of proteins in environmental stress responses both speci�c to fungi and conserved across the tree of life. Although our results cannot con�rm whether SNPs are the causal mechanism, the candidate genes could be used in future functional studies. Additionally, common gar- den experiments could provide support for local adaptation to climate in M. phaseolina. Overall, our observations point to a scenario in which M. phaseolina, as other plant pathogens with clonal population structures, is structured in a subcontinental regional stable manner in the face of in- stability at local scales in line with the metapopulation dynamics perspective. These results are consistent with a scenario of evolution after migration driven by divergence following clonal expansions. The pres- ence of MLLs across countries underscores the potential for a large in�uence of anthropogenic migration introducing M. phaseolina to new environments. The association of genetic divergence with climatic variables and putatively adaptive functions of the genes with SNPs strongly associated that would hypo- thetically bene�t M. phaseolina in speci�c environments, is consistent with potential selection imposed by speci�c climatic variables. Future studies will be needed to identify the degree to which distinct ge- netic groups re�ect their adaptation to host and climate. Such analyses will bene�t from a global sampling collected from diverse hosts in conjunction with multiple reference genomes sequenced with long-read technologies that will allow further characterization of the role of genomic variation, including structural variation, in M. phaseolina adaptation to host and the climatic environment. This knowledge expands the impact that spatial population genomics and genotype-environment as- sociations can have on our ability to characterize adaptive potential in plant pathogens by identifying candidate genes and presents a preliminary and complementary approach to the forward-genetics and phenotypic characterization approaches. The ability to identify candidate genes at a population speci�c �� level in a clonal pathogen presents an opportunity to evaluate candidate genes in a population speci�c manner, which represents a powerful approach specially in clonal pathogens in which unusually high lev- els of linkage prevent the application of genome scan methods. Additionally, the RDA approach could be applied using candidate adaptive genetic markers to predict pathogens’ “adaptive landscape” represent- ing its adaptive variation for any environment across a geographic range (Capblancq and Forester, ����). As climate and agricultural challenges become more demanding, the characterization of pathogen adap- tation capabilities enabled by population genomics should become increasingly utilized for plant disease risk prediction models specially under adverse future climate scenarios. �.� Materials and methods �.�.� Isolate collection and DNA preparation A total of �� M. phaseolina isolates were obtained from culture collections, as well as roots or lower stems of soybean and dry bean plants in production �elds (Supplementary Table A.�). There were �� isolates collected from soybean across a latitudinal range in �� states, including �� isolates from a previous study (Sexton, Hughes and Wise, ����). Forty isolates were collected from dry bean grown in Michigan, Puerto Rico and Colombia. Isolates from Michigan were collected from ���� to ���� (Jacobs et al., ����). Isolates from Puerto Rico and Colombia were collected from research plots at the University of Puerto Rico and at the International Center for Tropical Agriculture (CIAT). Two strawberry isolates collected from California and one isolate from Ethiopian mustard (Brassica carinata) were included as host outgroups. Cultures were routinely grown on potato dextrose agar (PDA; Acumedia, Lansing, MI) medium. For genomic DNA extraction, four �-mm plugs taken from the edge of the culture were used to in- oculate �� mL of potato dextrose broth amended with chloramphenicol (�� mg/L). The broth was incu- bated for � to � d at room temperature. Mycelia were harvested, lyophilized for �� h and ground using a FastPrep FP��� homogenizer (BIO ��� Savant Instruments, Hobrook, NY). Genomic DNA was ex- tracted from the lyophilized tissue using a modi�ed SDS-based method; brie�y, �� mg of ground mycelia were mixed in lysis bu�er (�% SDS (w/v); ��� mM Tris-HCl, pH �.�; �� mM EDTA, pH �.�) followed by phenol/chloroform DNA extraction. The identity of all isolates was con�rmed by multigene DNA analysis of the Internal Transcribed Spacer regions for the nuclear rDNA operon (ITS), part of the Trans- �� lation Elongation Factor (TEF-�) gene region, and part of the actin (ACT) gene region according to (Sarr et al., ����). Maximum likelihood analysis of the combined sequence alignment placed all the isolates tested in the M. phaseolina cluster. A full heuristic search using the �rst ten most parsimonious trees and the Neighbor-joining tree as starting trees with ��� random sequence additions was performed in PAUP v�.�b�� (Swo�ord ����), to �nd the maximum likelihood tree (Supplementary Fig. A.�). �.�.� Whole genome sequencing and variant calling Genomic libraries were constructed and each of the isolates were whole-genome sequenced to ��X cover- age using a ��� base-pair paired-end strategy on the Illumina HiSeq ���� platform at the Michigan State University Research Technology Support Facility Genomics Core (East Lansing, MI). The libraries were prepared using the Illumina TruSeq Nano DNA Library Preparation Kit HT. The resulting sequences were quality assessed using FastQC (Andrews et al., ����) and cleaned using Cutadapt v�.�� (Martin, ����), with the following parameters: -f fastq, -q ��,��, –trim-n, -m ��, -n �, -a AGATCGGAAGAGCA- CACGTCTGAACTCCAGTCAC, -A AGATCGGAAGAGCGTCGTGTAGGGAAAGAGTGTA- GATCTCGGTGGTCGCCGTATCATT. After initial quality �ltering, the remaining sequences were aligned to the M. phaseolina reference genome (JGI Mycocosm, MPI-SDFR-AT-���� v�.�) using bwa- mem (Heng Li, ����). The isolate used for the M. phaseolina reference genome was collected from natural Arabidopsis thaliana populations in France (Mesny et al., ����). The mapping statistics, genome align- ment rate and genome coverage were assessed with SAMtools �agstat (Li et al., ����). Alignments were sorted and indexed using SAMtools (Li et al., ����). After mapping, duplicate reads were identi�ed using MarkDuplicates and removed during the variant calling step. Single nucleotide polymorphisms (SNPs) of all �� isolates were predicted using the Genome Analysis Toolkit (GATK) v�.� (McKenna et al., ����). Initially, SNPs were called individually with GATK’s Hap- lotypeCaller. GVCF �les were combined, and common SNPs jointly identi�ed using CombineGVCFs and GenotypeGVCFs programs. The later using the -new-qual parameter. The combined vcf �le was quality �ltered using vcfR v�.��.� package (Knaus and Grünwald, ����) in R v�.�.� (R Core Team ����). To be included in the high-quality set, SNPs were �ltered to remove SNPs with a minimum read depth (DP) of <�x and greater that the ��th percentile of each sample DP distribution and exclude SNPs with �� minimum threshold mapping quality (MQ < ��) and minimum allele frequency (MAF < �.��) which corresponds to the allele presence in at least two isolates. Only variants with no missing data were retained, which corresponds to positions with � missing data for all the sequenced isolates. The �nal high-quality dataset was used in all subsequent analysis. The �nal vcf was annotated using SnpE� v�.�c (Cingolani, Platts, et al., ����) and a vcf containing only SNPs in intergenic regions was created using SnpSift v�.�c (Cingolani, Patel, et al., ����). �.�.� Phylogenomics and population genetic structure The population structure was inferred according to the results from both model-based and model-free clustering methods and phylogenetic inference. The phylogenetic tree was inferred from the full set of high-quality SNPs among the �� M. phaseolina isolates in RAxML-NG v�.�� (Kozlov et al., ����). The RAxML analysis was performed using the “-all” option which conducted �� maximum likelihood infer- ences on the original SNP alignment, standard bootstrapping with automatic determination of the num- ber of replicates (Felsenstein’s bootstrap, FBP; MRE-based bootstopping test) and the subsequent max- imum likelihood search. The General-Time-Reversible (GTR) model of nucleotide substitution with GAMMA model of rate heterogeneity and correction for ascertainment bias (GTR+G+ASC_LEWIS) was used. The best-scoring ML tree was used for optimizing all model and branch length parameters and model evaluation. A model-free dimensionality-reduction approach, principal component analysis (PCA), and discriminatory analysis of principal components (DAPC) were also conducted on the full set of SNPs using adegenet package (Jombart, ����; Jombart and Ahmed, ����) in R �.�.� (R Core Team ����). To infer population dynamics and reconstruct a rooted M. phaseolina phylogeny, the M. phase- olina (JGI Mycocosm, MPI-SDFR-AT-���� v�.�) reference genome was used as outgroup taxon. Maxi- mum likelihood analysis was run in RAxML-NG v�.�� using the “-all” option with automatic bootstrap replicates and the GTR+G+ASC_LEWIS substitution model. �.�.� Spatial genetic structure Bayesian clustering of allele frequencies was implemented in conStruct (Bradburd, Coop and Ralph, ����). To assess whether population structure was well described by modelling isolates as admixtures be- tween multiple discrete genetic groups or by both discrete and continuous genetic structure, spatial anal- �� ysis of population structure was conducted using conStruct (Bradburd, Coop and Ralph, ����). Spatial analysis in conStruct accounts for isolation by distance by allowing genetic di�erentiation to increase with geographic distance within discrete genetic groups (layers, K). The data was analyzed treating individual isolates as the unit of analysis, using the spatial models setting K between � and � with ����� iterations, and compared these models using cross-validation with �� replicates. For cross-validation, ��% of loci were used to �t the model and the remaining loci for model evaluation. A geographically constrained least-squares method as implemented in TESS� (Caye et al., ����), was used to estimate ancestry coe�- cients and create interpolation maps based on the coe�cients. TESS� uses a spatially explicit algorithm that can be considered model-free. The algorithm was run using the function “tess�” with K between � and � and �� replicates. �.�.� Population genetic and genotypic diversity For each clade and genetic cluster, gene diversity (Nei, ����) was calculated using the Hs function in the adegenet package (Jombart, ����; Jombart and Ahmed, ����). The median estimates of pairwise genetic distance and genotypic diversity indices were calculated within each clade and genetic cluster using the R package poppr v�.�.� (Kamvar et al. ����). Genotypic diversity was assessed by calculating the number of multilocus genotypes (MLGs). A MLG was de�ned as a unique combination of the ��,��� SNPs. MLGs were collapsed into larger groups called multilocus lineages using the average neighbor algorithm and a Prevosti’s distance threshold of �.���� (bitwise.dist function; Kamvar et al., ����). Rarefaction was used to correct for uneven sample sizes using the R package vegan v�.�-� (Oksanen et al., ����) and obtain the number of expected MLGs and MLLs (eMLG and eMLL) at the lowest common sample size (i.e., �� for clades and � for genetic clusters). Genotypic diversity indices, Shannon-Wiener Index (H*), Stoddart and Taylor’s Index (G*), Simpson’s index (lambda*) and evenness (E�*) (Grünwald et al., ����), were calculated using the R package poppr v�.�.� (diversity_ci function; Kamvar et al., ����) based on the number of MLLs in each clade and genetic cluster and correcting for unequal sample sizes based on rarefaction. The function mlg.crosspop in poppr was used to detect the presence of MLGs occurring across populations. Migration was inferred by tracking MLGs across genetic clusters, referred here as genotype �ow (McDonald and Linde, ����). �� �.�.� Population di�erentiation between genetic clusters and countries The FS T analog, GST (Nei ����, ����) was calculated from clone-corrected data using vcfR (Knaus and Grunwald ����) to infer di�erentiation among genetic clusters. To describe the population dynamics between the US, Puerto Rico and Colombia, the degree of genetic di�erentiation across M. phaseolina samples was measured hierarchically by genetic clusters within clades. Analysis of molecular variance (AMOVA) based on the quasi-Euclidean distance matrix was conducted in poppr v�.�.� (Kamvar et al. ����). AMOVA estimates the number of di�erences summed over loci based on a matrix of distances be- tween individuals and covariance components are used to calculate �xation indices for each hierarchical level, among clades, among genetic clusters and within genetic clusters. Signi�cant di�erences of �xation indices were determined by �,��� random permutations (Grunwald and Hoheisel ����). �.�.� Recombination and clonality To account for potential intraspeci�c recombination among M. phaseolina isolates, a phylogenetic net- work was built using the Neighbor-Net algorithm as implemented in SplitsTree� v�.��.�. The extent of clonality was tested by calculating the proportion of signi�cant linkage between pairs of loci, by comput- ing the standardized index of association (IA , Brown et al. ����) for each of the main populations (US and COLPR) using poppr v�.�.� (Kamvar et al. ����). Linkage disequilibrium is expected in asexual or inbreeding populations and IA values close to zero are expected for outcrossing populations (Burt et al., ����). The observed IA distributions for each population were compared to �ve simulated recombined distributions (�%, ��%, ��%, ��% and ���% linkage) generated among ��, ��� loci and �� samples (corre- sponding to the median population size of the two clusters). The observed and simulated IA values were tested for normality using the Shapiro-Wilk’s normality test and an analysis of variance (ANOVA) was conducted to test for signi�cant di�erences among the distributions. Pairwise comparisons between the IA simulated distributions and for each population were tested for di�erence with Tukey’s HSD test in R. The extent of clonality was correlated to clonal (���%), mostly clonal (��%), semiclonal (��%, ��%) or sexual (�%) modes of reproduction. Linkage disequilibrium (LD) decay rate was estimated using the physical distance over which LD decays to half its initial value, as measured by the squared correlation coe�cient (r� ). The linkage disequilibrium decay was calculated for each clade using the correlation co- �� e�cient (r� ) in TASSEL v� (Bradbury et al., ����) within a window of �� sites among SNPs using the clone-corrected dataset (�� MLGs). The mean r� values, representing the correlation between alleles at two loci within �� bp of physical distance, were then plotted in R �.�.� (R Core Team ����). �.�.� Climatic data For each isolate, the �� standard bioclimatic variables available at the WorldClim� database (Fick Hij- mans, ����) were obtained using ‘getData’ function from raster R package (Hijmans, ����). All variables are the average for the years ���� to ���� and were obtained at a spatial resolution of �.� min ( ��.� km�). We used data at a resolution of �.� min ( ��.� km�), because it corresponds with our sampling design (sin- gle isolate samples rather than populations) being at a �eld or county scale. Coarser resolutions could combine multiple sampling locations into a single spatial grid and �ner resolutions (��-s or <��-s), while this may be important for structuring patterns of genetic variation within populations, these data are less suitable for our sampling design and focus on regional to continental-wide patterns. We reduced the number of climatic variables from �� to �ve to account for collinearity among them (|r| > �.�) and to represent our hypothesis about the most important factors potentially driving selection. Diseases caused by M. phaseolina are more prevalent during hot and dry conditions, therefore temperature and precip- itation variables were included. The selected climatic variables were: BIO�� = Precipitation of Warmest Quarter, BIO�� = Precipitation Seasonality (Coe�cient of Variation), BIO�� = Precipitation of Driest Quarter, BIO�� = Mean Temperature of Warmest Quarter and BIO� = Temperature Seasonality (stan- dard deviation *���). Each bioclimatic variable was scaled, centered, and evaluated for inclusion using forward selection with ��,��� permutations using adespatial R package (Dray et al., ����). To account for underlying spatial structure (autocorrelation) and reduce spurious GEA, distance- based Moran’s eigenvector maps (dbMEM) were generated using sample coordinates in the quickMEM R function (Borcard, Gillet, Legendre, ����). The dbMEMs are a matrix of axes that capture spatial pat- terns from multiple angles rather than just a latitudinal or longitudinal vector. Only signi�cant dbMEM axes were selected using forward selection with �,��� permutations. A simple RDA model and partial RDA model conditioning on space, using only signi�cant dbMEMs, were used to identify the climatic variables signi�cantly contributing to genomic variation and those structured in space. �� �.�.� Variance partitioning and outlier loci identi�cation To identify potentially adaptive loci, associations between genetic data (loci) and climatic variables hy- pothesized to drive selection were evaluated using a multivariate method, redundancy analysis (RDA, as implemented by Forester et al., ����). RDA simultaneously tests multiple loci that covary in response to climatic variables. Partial RDA models were used for variance partitioning and outlier loci identi�cation while correcting for neutral genetic population structure. Variance partitioning analysis was performed with linkage-disequilibrium (LD)-�ltered (r� > �.�) dataset of ��,��� SNPs. The independent contribu- tion of each set of explanatory variables: climate, neutral population structure or space, was assessed while removing the e�ect of the remaining variable sets using partial RDA. In outlier loci identi�cation, using a partial RDA is recommended to reduce the number of false-positive detections particularly in sce- narios of multilocus adaptation when selective agents are unknown (Forester et al., ����). On the other hand, partial RDA can lead to high false-negative detections when variance is confounded between cli- matic variables and neutral population structure (Capblancq and Forester, ����). Candidate adaptive loci were identi�ed using simple and partial RDA models to examine the extent of this issue. A partial RDA model conditioning on neutral population genetic structure was used for candidate outlier SNPs detection. Outlier loci were identi�ed in the three signi�cant constrained axes as the SNPs having load- ings ±� or ±� SD from the mean score of each constrained axis using both the LD-�ltered set of ��,��� SNPs and the full set of ��,��� SNPs, respectively (Forester et al., ����; Lasky et al., ����). A simple RDA model, without correcting for population structure, using the LD-�ltered set of ��,��� SNPs and out- lier loci were identi�ed in the three signi�cant constrained axes as the SNPs having loadings ±� SD from the mean score. Gene annotations for the signi�cant candidate SNPs were used to investigate putative adaptive functions, using the annotated vcf. �� BIBLIOGRAPHY Abramova, N. E. et al. (����) ‘Regulatory mechanisms controlling expression of the DAN/TIR mannoprotein genes during anaerobic remodeling of the cell wall in Saccharomyces cerevisiae’, Genetics, ���(�), pp. ����–����. doi: ��.����/genetics/���.�.����. Aguilera, J., Randez-Gil, F. and Prieto, J. A. (����) ‘Cold response in Saccharomyces cerevisiae: New functions for old mechanisms’, FEMS Microbiology Reviews, ��(�), pp. ���–���. doi: ��.����/j.����- ����.����.�����.x. Allen, T. W. et al. (����) ‘Soybean yield loss estimates due to diseases in the United States and Ontario, Canada, from ���� to ����’, Plant Health Progress, ��(�), pp. ��–��. doi: ��.����/PHP-RS-��-����. Andrews, S. (����) ‘FastQC: A quality control tool for high throughput sequence data’. Available at: http://www.bioinformatics.babraham.ac.uk. Arias, R. S. et al. (����) ‘Discriminating microsatellites from Macrophomina phaseolina and their potential association to biological functions’, Plant Pathology, ��(�), pp. ���–���. doi: ��.����/j.����- ����.����.�����.x. Baird R. E., Watson C. E., Scruggs M (����) Relative Longevity of Macrophomina phaseolina and Associated Mycobiota on Residual Soybean Roots in Soil. Plant Dis ��: ���–��� Baird, R. E. et al. (����) ‘Variability of United States isolates of Macrophomina phaseolina based on simple sequence repeats and cross genus transferability to related genera within botryosphaeriaceae.’, Mycopathologia, ���(�), pp. ���–��. doi: ��.����/s�����-���-����-�. Batista, E., Lopes, A. and Alves, A. (����) ‘What do we know about botryosphaeriaceae? An overview of a worldwide cured dataset’, Forests, ��(�). doi: ��.����/f��������. Bertazzoni, S. et al. (����) ‘Accessories Make the Out�t: Accessory Chromosomes and Other Dis- pensable DNA Regions in Plant-Pathogenic Fungi’, Molecular Plant-Microbe Interactions, ��(�), p. MPMI-��-��-����. doi: ��.����/MPMI-��-��-����-FI. Bitocchi, E. et al. (����) ‘Beans (Phaseolus ssp.) as a model for understanding crop evolution’, Fron- tiers in Plant Science, �(May), pp. �–��. doi: ��.����/fpls.����.�����. Bradburd, G. S., Coop, G. M. and Ralph, P. L. (����) ‘Inferring Continuous and Discrete Population Genetic’, Genetics, ���(September), pp. ��–��. doi: ��.����/genetics.���.������. Bradbury, P. J. et al. (����) ‘TASSEL: software for association mapping of complex traits in diverse samples’, Bioinformatics. Oxford Univ Press, ��(��), pp. ����–����. Bradley, C. and Allen, T. (����) ‘Estimates of Soybean Yield Reductions Caused by Diseases in the United States’, University of Illinois Outreach and Extension. University of Illinois Extension Outreach, ����. Available at: http://extension.cropsciences.illinois.edu. Branco, S. et al. (����) ‘Genetic isolation between two recently diverged populations of a symbiotic fungus’, Molecular Ecology, ��(��), pp. ����–����. doi: ��.����/mec.�����. �� Branco, S. et al. (����) ‘Continental-level population di�erentiation and environmental adaptation in the mushroom Suillus brevipes’, Molecular Ecology. doi: ��.����/mec.�����. Burkhardt, A. et al. (����) ‘Assembly, annotation, and comparison of Macrophomina phaseolina isolates from strawberry and other hosts’, Dryad, Dataset. Available at: https://datadryad.org/stash/dataset/doi:��.����/dryad.zkh�����h (Accessed: �� August ����). Burkhardt, A. K. et al. (����) ‘Assembly, annotation, and comparison of Macrophomina phaseolina isolates from strawberry and other hosts’, BMC Genomics. BioMed Central, ��(�), p. ���. doi: ��.����/s�����-���-����-�. Capblancq, T. and Forester, B. R. (����) ‘Redundancy analysis: A Swiss Army Knife for landscape genomics’, Methods in Ecology and Evolution, ��(��), pp. ����–����. doi: ��.����/����-���X.�����. Casano, S. et al. (����) ‘First report of charcoal rot caused by Macrophomina phaseolina on hemp (Cannabis sativa) varieties cultivated in southern Spain’, Plant Disease, p. ����. doi: ��.����/PDIS-��- ��-����-PDN. Caye, K. et al. (����) ‘TESS�: Fast inference of spatial population structure and genome scans for selection’, Molecular Ecology Resources, ��(�), pp. ���–���. doi: ��.����/����-����.�����. Cingolani, P., Platts, A., et al. (����) ‘A program for annotating and predicting the e�ects of single nucleotide polymorphisms, SnpE�’, Fly, �(�), pp. ��–��. doi: ��.����/�y.�����. Cingolani, P., Patel, V. M., et al. (����) ‘Using Drosophila melanogaster as a model for genotoxic chemical mutational studies with a new program, SnpSift’, Frontiers in Genetics, �(MAR), pp. �–�. doi: ��.����/fgene.����.�����. Croll, D. and Mcdonald, B. A. (����) ‘The genetic basis of local adaptation for pathogenic fungi in agricultural ecosystems’, Molecular Ecology. doi: ��.����/mec.�����. Crous, P. W. et al. (����) ‘Botryosphaeriaceae: Systematics, pathology, and genetics’, Fungal Biology. doi: ��.����/j.funbio.����.��.���. Dhingra, O. D. and Sinclair, J. B. (����) ‘Location of Macrophomina phaseoli on soybean plants related to culture characteristics and virulence’. Edited by F. A. O. of the UN. (Phytopathology), pp. ���–���. Dhingra, O. D. and Sinclair, J. B. (����) ‘Survival of Macrophomina phaseolina Sclerotia in Soil: Ef- fects of Soil Moisture, Carbon:Nitrogen Ratios, Carbon Sources, and Nitrogen Concentrations’, Phy- topathology, ��, pp. ���–���. Dhingra, O. D. and Sinclair, J. B. (����) Biology and pathology of Macrophomina phaseolina. Minas Gerais: Universidade Federal de Vicosa. Dray, S. et al. (����) ‘Package “adespatial”’, Ecological Monographs, pp. �–���. doi: ��.����/��-����.�. Ellison, C. E. et al. (����) ‘Population genomics and local adaptation in wild isolates of a model microbial eukaryote’, Proceedings of the National Academy of Sciences, ���(�), pp. ����–����. doi: ��.����/pnas.����������. �� Fitzpatrick, M. C. and Keller, S. R. (����) ‘Ecological genomics meets community-level modelling of biodiversity: Mapping the genomic landscape of current and future environmental adaptation’, Ecology Letters, ��(�), pp. �–��. doi: ��.����/ele.�����. Forester, B. R. et al. (����) ‘Comparing methods for detecting multilocus adaptation with multivari- ate genotype–environment associations’, Molecular Ecology, ��(�), pp. ����–����. doi: ��.����/mec.�����. Gasch, A. P. (����) ‘Comparative genomics of the environmental stress response in ascomycete fungi’, (July), pp. ���–���. doi: ��.����/yea. Gibson, M. J. S. and Moyle, L. C. (����) ‘Regional di�erences in the abiotic environment contribute to genomic divergence within a wild tomato species’, Molecular Ecology, ��(��), pp. ����–����. doi: ��.����/mec.�����. Gladieux, P. et al. (����) ‘Fungal evolutionary genomics provides insight into the mechanisms of adaptive divergence in eukaryotes’, Molecular Ecology, ��(�), pp. ���–���. doi: ��.����/mec.�����. Gluck-Thaler, E. et al. (����) ‘Giant Starship elements mobilize accessory genes in fungal genomes’, bioRxiv, p. ����.��.��.������. Available at: https://www.biorxiv.org/content/��.����/����.��.��.������v� Goss, E. M. et al. (����) ‘The Irish potato famine pathogen Phytophthora infestans originated in central Mexico rather than the Andes’, Proceedings of the National Academy of Sciences, ���(��), pp. ����–����. doi: ��.����/pnas.����������. Grünwald, N. J. et al. (����) ‘Analysis of genotypic diversity data for populations of microorganisms’, Phytopathology, ��(�), pp. ���–���. doi: ��.����/PHYTO.����.��.�.���. Grünwald, N. J. and Hoheisel, G. A. (����) ‘Hierarchical analysis of diversity, sel�ng, and genetic dif- ferentiation in populations of the oomycete Aphanomyces euteiches’, Phytopathology, ��(��), pp. ����–����. doi: ��.����/PHYTO-��-����. Gupta, G. K., Sharma, S. K. and Ramteke, R. (����) ‘Biology, Epidemiology and Management of the Pathogenic Fungus Macrophomina phaseolina (Tassi) Goid with Special Reference to Charcoal Rot of Soybean (Glycine max (L.) Merrill)’, Journal of Phytopathology, ���(�), pp. ���–���. doi: ��.����/j.����- ����.����.�����.x. Hanski, I. (����) ‘Metapopulation dynamics’, Nature, ���(����), pp. ��–��. doi: ��.����/�����. Heng Li (����) Aligning sequence reads, clone sequences and assembly contigs with BWA-MEM, Preprint Oxford University Press. Available at: https://arxiv.org/pdf/����.����.pdf (Accessed: �� Septem- ber ����). Hijmans, R. (����) ‘raster: Geographic Data Analysis and Modeling’. Available at: https://cran.r- project.org/package=raster. Islam, Md et al. (����) ‘Tools to kill: Genome of one of the most destructive plant pathogenic fungi Macrophomina phaseolina’, BMC Genomics. BMC Genomics, ��(�), p. ���. doi: ��.����/����-����-��- ���. �� Jacobs, J. L. et al. (����) ‘Determining the Soilborne Pathogens Associated with Root Rot Disease Complex of Dry Bean in Michigan’, Plant Health Progress, ��(�), pp. ���–���. doi: ��.����/php-��-��- ����-s. Jana, T., Sharma, T. R. and Singh, N. K. (����) ‘SSR-based detection of genetic variability in the charcoal root rot pathogen Macrophomina phaseolina’, Mycological Research, ���(�), pp. ��–��. doi: ��.����/S����������������. Jin, P. et al. (����) ‘Electron cryo-microscopy structure of the mechanotransduction channel NOMPC’, Nature. Nature Publishing Group, ���(����), pp. ���–���. doi: ��.����/nature�����. Jombart, T. (����) ‘Adegenet: A R package for the multivariate analysis of genetic markers’, Bioin- formatics, ��(��), pp. ����–����. doi: ��.����/bioinformatics/btn���. Jombart, T. and Ahmed, I. (����) ‘adegenet �.�-�: New tools for the analysis of genome-wide SNP data’, Bioinformatics, ��(��), pp. ����–����. doi: ��.����/bioinformatics/btr���. Kelly, J. D. (����) The Story of Bean Breeding, BeanCAP and PBG, (June), pp. �–��. Available at: http://www.css.msu.edu. Kendig, S. R., Rupe, J. C. and Scott, H. D. (����) ‘E�ect of irrigation and soil water stress on den- sities of Macrophomina phaseolina in soil and roots of two soybean cultivars’, Plant Disease, ��(�), pp. ���–���. doi: ��.����/PDIS.����.��.�.���. Knaus, B. J. and Grünwald, N. J. (����) ‘vcfr: a package to manipulate and visualize variant call format data in R’, Molecular Ecology Resources, ��(�), pp. ��–��. doi: ��.����/����-����.�����. Knox-Davies, P. S. (����) ‘Pycnidium production by Macrophomina’, S. Afr. J. Agric. Sci, �, pp. ���–���. Koehler, A. M. and Shew, H. D. (����) ‘First Report of Charcoal Rot of Stevia Caused by Macrophom- ina phaseolina in North Carolina’, Plant Disease. Plant Disease, ���(�), pp. ���–���. doi: ��.����/PDIS- ��-��-����-PDN. Koike, S. T. et al. (����) ‘Status of Macrophomina phaseolina on Strawberry in California and Pre- liminary Characterization of the Pathogen’, International Journal of Fruit Science. Taylor and Francis, pp. �–��. doi: ��.����/��������.����.�������. Kozlov, A. M. et al. (����) ‘RAxML-NG: A fast, scalable and user-friendly tool for maximum likeli- hood phylogenetic inference’, Bioinformatics, ��(��), pp. ����–����. doi: ��.����/bioinformatics/btz���. Lasky, J. R. et al. (����) ‘Characterizing genomic variation of Arabidopsis thaliana: The roles of geography and climate’, Molecular Ecology, ��(��), pp. ����–����. doi: ��.����/j.����-���X.����.�����.x. Lasky, J. R. et al. (����) ‘Genome-environment associations in sorghum landraces predict adaptive traits’, Science Advances, �(�), pp. �–��. doi: ��.����/sciadv.�������. Lawson, D. J., van Dorp, L. and Falush, D. (����) ‘A tutorial on how not to over-interpret STRUC- TURE and ADMIXTURE bar plots’, Nature Communications. Springer US, �(�), pp. �–��. doi: ��.����/s�����-���-�����-�. �� LeBlanc, N., Cubeta, M. A. and Crouch, J. A. (����) ‘Population genomics trace clonal diversi�ca- tion and intercontinental migration of an emerging fungal pathogen of boxwood’, Phytopathology, ���(�), pp. ���–���. doi: ��.����/PHYTO-��-��-����-FI. Leyva-Mir, S. G. et al. (����) ‘First Report of Charcoal Rot of Sugarcane Caused by Macrophomina phaseolina in Mexico’, Plant Disease. Plant Disease, ��(�), pp. ���–���. doi: ��.����/PDIS-��-��-����- PDN. Li, H. et al. (����) ‘The Sequence Alignment/Map format and SAMtools.’, Bioinformatics (Oxford, England). Oxford University Press, ��(��), pp. ����–�. doi: ��.����/bioinformatics/btp���. Ma, J., Hill, C. B. and Hartman, G. L. (����) ‘Production of Macrophomina phaseolina Conidia by Multiple Soybean Isolates in Culture’, Plant Disease. The American Phytopathological Society, ��(�), pp. ����–����. doi: ��.����/PDIS-��-�-����. Martin, M. (����) ‘Cutadapt removes adapter sequences from high-throughput sequencing reads’, EMBnet.journal, ��(�), p. ��. doi: ��.�����/ej.��.�.���. McDonald, B. A. and Stukenbrock, E. H. (����) ‘Rapid emergence of pathogens in agro-ecosystems: global threats to agricultural sustainability and food security’, Philosophical Transactions of the Royal Society B: Biological Sciences. The Royal Society, ���(����), p. ��������. doi: ��.����/rstb.����.����. McGoldrick, L. L. et al. (����) ‘Structure of the thermo-sensitive TRP channel TRP� from the alga Chlamydomonas reinhardtii’, Nature Communications. Springer US, ��(�), pp. �–��. doi: ��.����/s�����-���-�����-�. McKenna, A. et al. (����) ‘The Genome Analysis Toolkit: a MapReduce framework for analyzing next-generation DNA sequencing data.’, Genome research, ��(�), pp. ����–���. doi: ��.����/gr.������.���. Meena, B. R. et al. (����) ‘First report of charcoal rot caused by Macrophomina phaseolina in Basella alba in India’, Plant Disease, p. ����. doi: ��.����/PDIS-��-��-����-PDN. Mengistu, A. et al. (����) ‘Charcoal rot disease assessment of soybean genotypes using a colony- forming unit index’, Crop Science, ��(December), pp. ����–����. doi: ��.����/cropsci����.��.����. Mengistu, A., Arelli, P. A., et al. (����) ‘Evaluation of soybean genotypes for resistance to charcoal rot’, Plant Health Progress. doi: ��.����/PHP-����-����-��-RS. Mengistu, A., Smith, J. R., et al. (����) ‘Seasonal progress of charcoal rot and its impact on soybean productivity’, Plant Disease, ��(�), pp. ����–����. doi: ��.����/PDIS-��-��-����. Mesny, F. et al. (����) ‘Genetic determinants of endophytism in the Arabidopsis root mycobiome’, Nature Communications. Springer US, ��(�), pp. �–��. doi: ��.����/s�����-���-�����-y. Meyer, W. . and Sinclair, J. B. (����) ‘Factors A�ecting Charcoal Rot of Soybean Seedlings’, Phy- topathology, ��, pp. ���–���. doi: DOI: ��.����/Phyto-��-���. Mihail, J. D. and Taylor, S. J. (����) ‘Interpreting variability among isolates of Macrophomina phase- olina in pathogenicity, pycnidium production, and chlorate utilization’, Canadian Journal of Botany, �� ��(��), pp. ����–����. doi: ��.����/b��-���. Milgroom, M. G. (����) ‘Recombination and the multilocus structure of fungal populations’, An- nual Review of Phytopathology. Annual Reviews Inc., ��(�), pp. ���–���. doi: ��.����/annurev.phyto.��.�.���. Milgroom, M. G. (����) Population biology of plant pathogens: genetics, ecology, and evolution. APS Press, The American Phytopathological Society St. Paul, MN, USA. Milgroom, M. G. et al. (����) ‘Clonal expansion and migration of a highly virulent, defoliating lin- eage of Verticillium dahliae’, Phytopathology, ���(�), pp. ����–����. doi: ��.����/PHYTO-��-��-����- R. Nishad, I. et al. (����) ‘First Report of Root Rot of Nepeta cataria Caused by Macrophomina phase- olina in India’, Plant Disease. Plant Disease, ���(��), p. ����. doi: ��.����/PDIS-��-��-����-PDN. Oksanen, J. et al. (����) ‘vegan: Community Ecology Package’, pp. ����–����. doi: ��.����/���-��- ���-����-�-������. Parsa, S. et al. (����) ‘Fungal endophytes in germinated seeds of the common bean, Phaseolus vul- garis’, Fungal Biology. doi: ��.����/j.funbio.����.��.���. Reznikov, S. et al. (����) ‘Disease incidence of charcoal rot (Macrophomina phaseolina) on soybean in north-western Argentina and genetic characteristics of the pathogen’, Canadian Journal of Plant Pathol- ogy. Taylor Francis, ��(�), pp. ���–���. doi: ��.����/��������.����.�������. Romero Luna, M. P. et al. (����) ‘Advancing Our Understanding of Charcoal Rot in Soybeans’, Journal of Integrated Pest Management. Bulletin Texas Agricultural Experimental Station No. ��, �(�), pp. ���–���. doi: ��.����/jipm/pmw���. Sakalidis, M. L. et al. (����) ‘The challenge of understanding the origin, pathways and extent of fun- gal invasions: global populations of the Neofusicoccum parvum-N. ribis species complex Diversity and Distributions’, Diversity and Distributions, ��, pp. ���–���. doi: ��.����/ddi.�����. Saleh, A. A. et al. (����) ‘Relatedness of Macrophomina phaseolina isolates from tallgrass prairie, maize, soybean and sorghum’, Molecular Ecology, ��(�), pp. ��–��. doi: ��.����/j.����-���X.����.�����.x. Sarr, M. P. et al. (����) ‘Genetic diversity in Macrophomina phaseolina, the causal agent of charcoal rot’, Phytopathologia Mediterranea, ��(�), p. ������. doi: ��.�����/Phytopathol. Savary, S. et al. (����) ‘The global burden of pathogens and pests on major food crops’, Nature Ecol- ogy and Evolution, �(�), pp. ���–���. doi: ��.����/s�����-���-����-y. Savolainen, O., Lascoux, M. and Merilä, J. (����) ‘Ecological genomics of local adaptation’, Nature Reviews Genetics. Nature Publishing Group, ��(��), pp. ���–���. doi: ��.����/nrg����. Sexton, J. P., Hangartner, S. B. and Ho�mann, A. A. (����) ‘Genetic isolation by environment or distance: Which pattern of gene �ow is most common?’, Evolution, ��(�), pp. �–��. doi: ��.����/evo.�����. �� Sexton, Z. F., Hughes, T. J. and Wise, K. A. (����) ‘Analyzing isolate variability of Macrophomina phaseolina from a regional perspective’, Crop Protection. Elsevier Ltd, ��, pp. �–��. doi: ��.����/j.cropro. ����.��.���. Short G. E., Wyllie T. D., Bristow P. R. (����) Survival of Macrophomina phaseolina in soil and residue of soybean. Phytopathology ��: ��–�� Slippers, B. and Boissin, E. (����) ‘Phylogenetic lineages in the Botryosphaeriales: a systematic and evolutionary framework’, Studies in Mycology. CBS-KNAW Fungal Biodiversity Centre, ��, pp. ��–��. doi: ��.����/sim����. Slippers, B. and Wing�eld, M. J. (����) ‘Botryosphaeriaceae as endophytes and latent pathogens of woody plants: diversity, ecology and impact’, Fungal Biology Reviews, ��(�–�), pp. ��–���. doi: ��.����/j.fbr.����.��.���. Smith, G. S. and Carvil, O. N. (����) ‘Field screening of commercial and experimental soybean culti- vars for their reaction to Macrophomina phaseolina’, Plant Disease, ��(�), pp. ���–���. doi: ��.����/PDIS. ����.��.�.���. Smith, J. M. et al. (����) ‘How clonal are bacteria?’, Proceedings of the National Academy of Sciences of the United States of America, ��(��), pp. ����–����. doi: ��.����/pnas.��.��.����. Stukenbrock, E. H. et al. (����) ‘The making of a new pathogen: Insights from comparative pop- ulation genomics of the domesticated wheat pathogen Mycosphaerella graminicola and its wild sister species’, Genome Research. Cold Spring Harbor Laboratory Press, ��(��), pp. ����–����. doi: ��.����/gr.������.���. Su, G. et al. (����) ‘Host specialization in the charcoal rot fungus, Macrophomina phaseolina’, Phy- topathology. The American Phytopathological Society, ��(�), pp. ���–���. doi: ��.����/PHYTO.����. ��.�.���. Tabima, J. F. et al. (����) ‘Population genomic analyses reveal connectivity via human-mediated trans- port across Populus plantations in North America and an undescribed sub-population of Sphaerulina musiva’, Molecular Plant-Microbe Interactions. doi: ��.����/mpmi-��-��-����-r. Tančić Živanov, S. et al. (����) ‘First Report of Charcoal Rot on Zebra Plant (Aphelandra squarrosa) Caused by Macrophomina phaseolina’, Plant Disease. Plant Disease, ���(��), p. ����. doi: ��.����/PDIS- ��-��-����-PDN. Wang, L. et al. (����) ‘Haploid Genome Analysis Reveals a Tandem Cluster of Four HSP�� Genes In- volved in the High-Temperature Adaptation of Coriolopsis trogii’, Microbiology Spectrum, �(�), pp. �–��. doi: ��.����/spectrum.�����-��. Wang, Weiwei et al. (����) ‘First Report of Macrophomina phaseolina Causing Stalk Rot of Sacha Inchi ( Plukenetia volubilis ) in China’, Plant Disease. Plant Disease, ���(�), pp. ���–���. doi: ��.����/PDIS-��-��-����-PDN. Wing�eld, M. J. et al. (����) ‘Planted forest health: The need for a global strategy’, Science, ���(����), pp. ���–���. doi: ��.����/science.aac����. �� Wrather, J. A. et al. (����) ‘Soybean disease loss estimates for the top ten soybean-producing countries in ����’, Canadian Journal of Plant Pathology, ��(�), pp. ���–���. doi: ��.����/�����������������. Wu, J. et al. (����) ‘Small heat shock proteins, phylogeny in �lamentous fungi and expression analyses in Aspergillus nidulans’, Gene. Elsevier B.V., ���(�), pp. ���–���. doi: ��.����/j.gene.����.��.���. Xuereb, A. et al. (����) ‘Putatively adaptive genetic variation in the giant California sea cucumber (Parastichopus californicus) as revealed by environmental association analysis of restriction-site associated DNA sequencing data’, Molecular Ecology, ��(��), pp. ����–����. doi: ��.����/mec.�����. Zveibil, A. et al. (����) ‘Survival, host-pathogen interaction, and management of Macrophomina phaseolina on strawberry in Israel’, Plant Disease, ��(�), pp. ���–���. doi: ��.����/PDIS-��-��-����. �� APPENDIX SUPPLEMENTARY TABLES AND FIGURES �� Table A.� Macrophomina phaseolina isolates used in this study. Isolates were collected across the US, Puerto Rico, and Colombia from soybean, dry bean, strawberry, and Ethiopian mustard. State / Genetic Collection Isolate ID Longitude Latitude Region Country Host Municipality Source Department Cluster Year CR_Red_� -��.��� ��.��� East North Central US MN Soybean Lamberton, MN US� Dean Malvick - UMN CR_Red_�B -��.��� ��.��� East North Central US MN Soybean Lamberton, MN US� Dean Malvick - UMN CR_Red_� -��.��� ��.��� East North Central US MN Soybean Lamberton, MN US� Dean Malvick - UMN Dm�� -��.�� ��.��� East North Central US WI Soybean Markesan, WI US�A T. Hughes. Obtained from Kiersten Wise Et�� -��.��� ��.��� East North Central US WI Soybean E. Troy, WI US�A T. Hughes. Obtained from Kiersten Wise Et�� -��.��� ��.��� East North Central US WI Soybean E. Troy, WI US�A T. Hughes. Obtained from Kiersten Wise Et�� -��.��� ��.��� East North Central US WI Soybean E. Troy, WI US� T. Hughes. Obtained from Kiersten Wise Et�� -��.��� ��.��� East North Central US WI Soybean E. Troy, WI US�A T. Hughes. Obtained from Kiersten Wise Et� -��.��� ��.��� East North Central US WI Soybean E. Troy, WI US�A T. Hughes. Obtained from Kiersten Wise IN��_� -��.��� ��.��� Central US IN Soybean Benton, IN US�A Purdue Plant Diag. Lab. Obtained from Kiersten Wise IN��_�_� -��.��� ��.��� Central US IN Soybean Lagrange, IN US�A Purdue Plant Diag. Lab. Obtained from Kiersten Wise IN��_�_� -��.�� ��.��� Central US IN Soybean Vermillion, IN NA Purdue Plant Diag. Lab. Obtained from Kiersten Wise IN��_�_� -��.�� ��.��� Central US IN Soybean Vermillion, IN US�A Purdue Plant Diag. Lab. Obtained from Kiersten Wise IN��_PO_� -��.��� ��.��� Central US IN Soybean Posey, IN US�A Purdue Plant Diag. Lab. Obtained from Kiersten Wise M��_�� -���.��� ��.��� West US CA Strawberry Santa Barbara, CA COLPR� ���� Frank Martin M��_�� -���.��� ��.��� West US CA Strawberry Monterey, CA COLPR� ���� Frank Martin M��_� -��.��� ��.��� East North Central US MI Dry bean Kawkawlin, MI US�A ���� M. Chilvers MISO���_� -��.��� ��.��� East North Central US MI Soybean Lyons, MI US�A ���� M. Chilvers MISO���_� -��.��� ��.��� East North Central US MI Soybean Lyons, MI US�A ���� M. Chilvers MISO���_� -��.��� ��.��� East North Central US MI Soybean Lyons, MI US�A ���� M. Chilvers MISO���_� -��.��� ��.��� East North Central US MI Soybean Lyons, MI US�A ���� M. Chilvers MISO���_� -��.��� ��.��� East North Central US MI Soybean Lyons, MI US�A ���� M. Chilvers MISO���_� -��.��� ��.��� East North Central US MI Soybean Lyons, MI US�A ���� M. Chilvers MISO���_� -��.��� ��.��� East North Central US MI Soybean Lyons, MI US�A ���� M. Chilvers MI_SF_��_�� -��.�� ��.��� East North Central US MI Soybean Westphalia, MI US�A ���� M. Chilvers MI_SF_�_�� -��.��� ��.��� East North Central US MI Soybean Galien, MI US�A ���� M. Chilvers MI_SF_�_�� -��.��� ��.��� East North Central US MI Soybean Allegan County, MI US�A ���� M. Chilvers MI_SF_�_� -��.�� ��.��� East North Central US MI Soybean Westphalia, MI US�A ���� M. Chilvers MP��� -��.�� ��.��� Central US KY Soybean Unknown US� R. Baird. Obtained from Kiersten Wise MP��� -��.�� ��.��� Central US TN Soybean Unknown US�A R. Baird. Obtained from Kiersten Wise MP��� -��.��� ��.��� South US TX Soybean Unknown US� R. Baird. Obtained from Kiersten Wise MP��� -��.� ��.��� Southeast US GA Soybean Unknown US� R. Baird. Obtained from Kiersten Wise MP��� -��.� ��.��� Southeast US GA Soybean Unknown US� R. Baird. Obtained from Kiersten Wise MP��� -��.��� ��.��� Southeast US SC Soybean Unknown US�B R. Baird. Obtained from Kiersten Wise M_��_�� -��.��� ��.��� East North Central US MI Dry bean Merrill, MI US�A ���� M. Chilvers M_��_�� -��.��� ��.��� East North Central US MI Dry bean Merrill, MI US�A ���� M. Chilvers M_��_�� -��.��� ��.�� East North Central US MI Dry bean Bad Axe, MI US�A ���� M. Chilvers M_��_�� -��.��� ��.�� East North Central US MI Dry bean Bad Axe, MI US�A ���� M. Chilvers M_��_�� -��.��� ��.�� East North Central US MI Dry bean Bad Axe, MI US�A ���� M. Chilvers M_��_� -��.��� ��.��� East North Central US MI Dry bean Wheeler, MI US�A ���� M. Chilvers M_��_� -��.��� ��.��� East North Central US MI Dry bean Merrill, MI US�A ���� M. Chilvers M_��_� -��.��� ��.��� East North Central US MI Dry bean Merrill, MI US�A ���� M. Chilvers M_��_� -��.��� ��.��� East North Central US MI Dry bean Merrill, MI US� ���� M. Chilvers M_��_� -��.��� ��.��� East North Central US MI Dry bean Merrill, MI US�A ���� M. Chilvers M_��_� -��.��� ��.��� East North Central US MI Dry bean Merrill, MI US�A ���� M. Chilvers M_��_� -��.��� ��.��� East North Central US MI Dry bean Merrill, MI US�A ���� M. Chilvers M_��_� -��.��� ��.��� East North Central US MI Dry bean Merrill, MI US�A ���� M. Chilvers M_��_� -��.��� ��.��� East North Central US MI Dry bean Merrill, MI US�A ���� M. Chilvers Md�� -��.��� ��.��� East North Central US WI Soybean Muscoda, WI US�A T. Hughes. Obtained from Kiersten Wise Md� -��.��� ��.��� East North Central US WI Soybean Muscoda, WI US�A T. Hughes. Obtained from Kiersten Wise Md� -��.��� ��.��� East North Central US WI Soybean Muscoda, WI US�A T. Hughes. Obtained from Kiersten Wise Md� -��.��� ��.��� East North Central US WI Soybean Muscoda, WI US�A T. Hughes. Obtained from Kiersten Wise Md� -��.��� ��.��� East North Central US WI Soybean Muscoda, WI US�A T. Hughes. Obtained from Kiersten Wise MpSDSU -��.��� ��.��� West North Central US SD Ethiopian mustard Brookings County, SD US� Febina Mathew - SD state U Mph_�� -��.��� �.��� Colombia COL VAC Dry bean Buga, VAC COLPR� ���� Gloria Mosquera - CIAT Mph_�� -��.��� �.��� Colombia COL VAC Dry bean Buga, VAC US�B ���� Gloria Mosquera - CIAT Mph_�� -��.��� �.��� Colombia COL VAC Dry bean Buga, VAC US�B ���� Gloria Mosquera - CIAT Mph_�� -��.��� �.��� Colombia COL CAU Dry bean Santander de Quilichao, CAU COLPR� ���� Gloria Mosquera - CIAT Mph_�� -��.��� �.��� Colombia COL VAC Dry bean Palmira, VAC COLPR� ���� Gloria Mosquera - CIAT Mph_�� -��.��� �.��� Colombia COL VAC Dry bean Palmira, VAC COLPR� ���� Gloria Mosquera - CIAT Mph_�� -��.��� �.��� Colombia COL VAC Dry bean Palmira, VAC COLPR� ���� Gloria Mosquera - CIAT Mph_�� -��.��� �.��� Colombia COL VAC Dry bean Palmira, VAC COLPR� ���� Gloria Mosquera - CIAT Mph_�� -��.��� �.��� Colombia COL VAC Dry bean Palmira, VAC COLPR� ���� Gloria Mosquera - CIAT Mph_�� -��.��� �.��� Colombia COL VAC Dry bean Palmira, VAC NA ���� Gloria Mosquera - CIAT Mph_�� -��.��� �.��� Colombia COL CAU Dry bean Santander de Quilichao, CAU COLPR� ���� Gloria Mosquera - CIAT Mph_�� -��.��� �.��� Colombia COL CAU Dry bean Santander de Quilichao, CAU COLPR� ���� Gloria Mosquera - CIAT Mph_�� -��.��� �.��� Colombia COL CAU Dry bean Santander de Quilichao, CAU COLPR� ���� Gloria Mosquera - CIAT Mph_�� -��.��� �.��� Colombia COL CAU Dry bean Santander de Quilichao, CAU US�A ���� Gloria Mosquera - CIAT Mph_�� -��.��� �.��� Colombia COL TOL Dry bean Armero, TOL COLPR� ���� Gloria Mosquera - CIAT Mph_�� -��.��� ��.��� Colombia COL MAG Dry bean Corpoica, MAG COLPR� ���� Gloria Mosquera - CIAT Mph_�� -��.��� ��.��� Colombia COL MAG Dry bean Corpoica, MAG COLPR� ���� Gloria Mosquera - CIAT Mph_�� -��.��� ��.��� Colombia COL MAG Dry bean Corpoica, MAG COLPR� ���� Gloria Mosquera - CIAT Mph_�� -��.��� ��.��� Colombia COL MAG Dry bean Corpoica, MAG COLPR� ���� Gloria Mosquera - CIAT Mph_� -��.��� �.��� Colombia COL CAU Dry bean Santander de Quilichao, CAU COLPR� ���� Gloria Mosquera - CIAT SAG�_� -��.��� ��.��� East North Central US MI Soybean Saginaw County, MI US�A ���� M. Chilvers TN��� -��.�� ��.��� Central US TN Soybean Unknown US�A A. Mengistu. Obtained from Kiersten Wise TN��� -��.��� ��.��� South US MS Soybean Unknown US�B A. Mengistu. Obtained from Kiersten Wise TN��� -��.��� ��.��� Central US TN Soybean Jackson, TN US�A A. Mengistu. Obtained from Kiersten Wise TN��� -��.�� ��.��� Central US KY Soybean Unknown US�A A. Mengistu. Obtained from Kiersten Wise TN��� -��.��� ��.��� South US MS Soybean Stoneville, Mississippi US�B A. Mengistu. Obtained from Kiersten Wise TN� -��.��� ��.��� Central US TN Soybean Jackson, Tennessee US�A A. Mengistu. Obtained from Kiersten Wise TN��� -��.��� ��.��� South US LA Soybean Unknown COLPR� A. Mengistu. Obtained from Kiersten Wise TN��� -��.��� ��.��� South US TX Soybean Unknown US� A. Mengistu. Obtained from Kiersten Wise TN� -��.��� ��.��� Central US TN Soybean Ames, TN US�A A. Mengistu. Obtained from Kiersten Wise UPR_Mph_ISA� -��.��� ��.��� Puerto Rico PR ISA Dry bean Isabela, PR COLPR� Consuelo Estevez De Jensen - UPR UPR_Mph_ISA� -��.��� ��.��� Puerto Rico PR ISA Dry bean Isabela, PR COLPR� Consuelo Estevez De Jensen - UPR UPR_Mph_JD� -��.��� ��.��� Puerto Rico PR JD Dry bean Juana Diaz, PR COLPR� Consuelo Estevez De Jensen - UPR UPR_Mph_JD� -��.��� ��.��� Puerto Rico PR JD Dry bean Juana Diaz, PR COLPR� Consuelo Estevez De Jensen - UPR UPR_Mph_JD� -��.��� ��.��� Puerto Rico PR JD Dry bean Juana Diaz, PR COLPR� Consuelo Estevez De Jensen - UPR W��_� -��.��� ��.��� East North Central US MI Soybean Hamilton, MI US� ���� M. Chilvers W�� -��.��� ��.��� East North Central US MI Soybean Pewamo, MI US�A ���� M. Chilvers W�� -��.��� ��.��� East North Central US MI Soybean Hamilton, MI US� ���� M. Chilvers W�_� -��.��� ��.��� East North Central US MI Soybean Hamilton, MI US� ���� M. Chilvers W_MISO�_�_� -��.� ��.� East North Central US MI Soybean Hamilton, MI US�A ���� M. Chilvers W_MISO�_�_�� -��.� ��.� East North Central US MI Soybean Berlin, MI US�A ���� M. Chilvers �� Table A.� Hierarchical analysis of molecular variance (AMOVA), partitioning total genetic variance into the following components: between clades, between genetic clusters and within genetic clusters. Clone corrected values are shown. Most of the variance was associated with di�erences between clades and between genetic clusters. Source of variation Variation (%) p-value Phi Between clades (US and COLPR) ��.�� �.��� �.�� Between genetic clusters (US-�A, US-�B, US-�), (COLPR-�, COLPR-�) within clade ��.�� �.��� �.�� Within genetic clusters ��.�� �.��� �.�� �� Table A.� Summary statistics for genetic diversity of Macrophomina phaseolina by country. N is num- ber of isolates (sample size); MLG is number of observed multilocus genotypes; eMLG is the number of expected MLG at a sample size of � based on rarefaction. MLL is number of observed multilocus lin- eages by population using a bitwise cuto� distance of �.����; CF is clonal fraction (� - (MLL/N). Clone corrected values are shown and indicated by asterisks for indices of genotypic diversity: Shannon-Wiener Index (H*), Stoddart and Taylor’s Index (G*), Simpson’s index (lambda*) and evenness (E�*). Gene Median pairwise Country N MLG eMLG MLL eMLL CF H* G* lambda* E�* diversity (He) genetic distance US �� �.��� �.��� �� �.�� �� �.�� �.� �.�� �.�� �.�� �.��� Colombia �� �.��� �.��� �� � �� �.�� �.� �.�� �.� �.��� �.��� Puerto Rico � �.��� �.��� � � � � �.� �.�� �.�� �.�� �.��� �� Table A.� Top �� candidate SNPs along the �rst three RDA axes, after accounting for neutral population structure using the LD-�ltered set of ��,��� SNPs. RDA RDA Climate SnpE� SnpE� distance SNP position Correlation SnpE� SNP category SnpE� annotation locus axis loading variable predicted e�ect from locus (bp) � sca�old_��_������ -�.��� TSsd �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_��_������_������ ���� � sca�old_��_������ -�.��� TSsd �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_��_������_������ ���� � sca�old_�_������� �.��� TSsd �.�� missense_variant MODERATE CDS_sca�old_�_�������_������� � � sca�old_�_������� -�.��� TSsd �.�� upstream_gene_variant MODIFIER START_CODON_sca�old_�_�������_������� ���� � sca�old_�_������� -�.��� TSsd �.�� upstream_gene_variant MODIFIER START_CODON_sca�old_�_�������_������� ��� � sca�old_��_������ -�.��� TSsd �.�� upstream_gene_variant MODIFIER START_CODON_sca�old_��_������_������ ��� � sca�old_�_������� -�.��� TSsd �.�� synonymous_variant LOW CDS_sca�old_�_�������_������� � � sca�old_��_����� -�.��� TSsd �.�� synonymous_variant LOW CDS_sca�old_��_�����_����� � � sca�old_��_������ -�.��� TSsd �.�� upstream_gene_variant MODIFIER CDS_sca�old_��_������_������ ���� � sca�old_��_������ -�.��� TSsd �.�� upstream_gene_variant MODIFIER CDS_sca�old_��_������_������ ���� � sca�old_�_����� �.��� TSsd �.�� upstream_gene_variant MODIFIER START_CODON_sca�old_�_�����_����� ���� � sca�old_�_������� �.��� TSsd �.�� intergenic_region MODIFIER CDS_sca�old_�_�������_�������-START_CODON_sca�old_�_�������_������� � � sca�old_�_������� �.��� TSsd �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_�_�������_������� ��� � sca�old_��_������ �.��� mTwq �.�� synonymous_variant LOW CDS_sca�old_��_������_������ � � sca�old_��_������ �.��� mTwq �.�� downstream_gene_variant MODIFIER CDS_sca�old_��_������_������ ���� � sca�old_�_������� -�.��� TSsd �.�� synonymous_variant LOW CDS_sca�old_�_�������_������� � � sca�old_��_���� -�.��� TSsd �.�� upstream_gene_variant MODIFIER START_CODON_sca�old_��_����_���� ��� � sca�old_��_����� �.��� TSsd �.�� synonymous_variant LOW CDS_sca�old_��_�����_����� � � sca�old_��_������ -�.��� PScv �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_��_������_������ ���� � sca�old_��_������ -�.��� PScv �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_��_������_������ ���� � sca�old_��_������ -�.��� PScv �.�� synonymous_variant LOW CDS_sca�old_��_������_������ � � sca�old_��_������ -�.��� PScv �.�� upstream_gene_variant MODIFIER CDS_sca�old_��_������_������ ���� � sca�old_��_���� -�.��� PScv �.�� upstream_gene_variant MODIFIER START_CODON_sca�old_��_����_���� ���� � sca�old_�_������ �.��� PScv �.�� upstream_gene_variant MODIFIER CDS_sca�old_�_������_������ ���� � sca�old_��_����� -�.��� PScv �.�� upstream_gene_variant MODIFIER START_CODON_sca�old_��_�����_����� ��� � sca�old_��_������ -�.��� PScv �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_��_������_������ ���� � sca�old_�_������ -�.��� PScv �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_�_������_������ ���� � sca�old_��_������ �.��� TSsd �.�� upstream_gene_variant MODIFIER START_CODON_sca�old_��_������_������ ��� � sca�old_�_������� �.��� PScv �.�� missense_variant MODERATE CDS_sca�old_�_�������_������� � � sca�old_��_������ -�.��� TSsd �.�� synonymous_variant LOW CDS_sca�old_��_������_������ � � sca�old_��_������ -�.��� TSsd �.�� synonymous_variant LOW CDS_sca�old_��_������_������ � � sca�old_��_������ -�.��� TSsd �.�� upstream_gene_variant MODIFIER CDS_sca�old_��_������_������ ���� � sca�old_�_������� -�.��� PScv �.�� upstream_gene_variant MODIFIER CDS_sca�old_�_�������_������� ���� � sca�old_��_������ -�.��� mTwq �.�� synonymous_variant LOW CDS_sca�old_��_������_������ � � sca�old_��_������ -�.��� PScv �.�� synonymous_variant LOW CDS_sca�old_��_������_������ � � sca�old_��_������ -�.��� PScv �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_��_������_������ ���� � sca�old_�_������� �.��� TSsd �.�� synonymous_variant LOW CDS_sca�old_�_�������_������� � � sca�old_��_������ -�.��� TSsd �.�� upstream_gene_variant MODIFIER START_CODON_sca�old_��_������_������ ���� � sca�old_��_������ �.��� PScv �.�� upstream_gene_variant MODIFIER CDS_sca�old_��_������_������ ���� � sca�old_��_������ -�.��� TSsd �.�� upstream_gene_variant MODIFIER START_CODON_sca�old_��_������_������ ��� � sca�old_��_������ �.��� TSsd �.�� downstream_gene_variant MODIFIER CDS_sca�old_��_������_������ ���� � sca�old_�_������� �.��� PScv �.�� upstream_gene_variant MODIFIER CDS_sca�old_�_�������_������� ���� � sca�old_��_������ �.��� TSsd �.�� missense_variant MODERATE CDS_sca�old_��_������_������ � � sca�old_�_������� �.��� TSsd �.�� upstream_gene_variant MODIFIER CDS_sca�old_�_�������_������� ���� � sca�old_��_������ �.��� Pwq �.�� upstream_gene_variant MODIFIER CDS_sca�old_��_������_������ ���� � sca�old_��_������ -�.��� TSsd �.�� synonymous_variant LOW CDS_sca�old_��_������_������ � � sca�old_��_������ �.��� TSsd �.�� upstream_gene_variant MODIFIER CDS_sca�old_��_������_������ ���� � sca�old_�_������� -�.��� TSsd �.�� synonymous_variant LOW CDS_sca�old_�_�������_������� � � sca�old_�_������� -�.��� TSsd �.�� synonymous_variant LOW CDS_sca�old_�_�������_������� � �� Table A.� Top �� candidate SNPs along the �rst three RDA axes, after accounting for neutral population structure using the the full set of ��,��� SNPs. RDA RDA Climate SnpE� SNP SnpE� SnpE� distance SNP position Correlation SnpE� annotation locus axis loading variable category predicted e�ect from locus (bp) � sca�old_��_������ -�.��� TSsd �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_��_������_������ ���� � sca�old_��_������ -�.��� TSsd �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_��_������_������ ���� � sca�old_��_������� -�.��� TSsd �.�� upstream_gene_variant MODIFIER CDS_sca�old_��_�������_������� ���� � sca�old_��_������ -�.��� TSsd �.�� synonymous_variant LOW CDS_sca�old_��_������_������ � � sca�old_�_������� -�.��� TSsd �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_�_�������_������� ���� � sca�old_�_������� -�.��� TSsd �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_�_�������_������� ���� � sca�old_�_������� -�.��� TSsd �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_�_�������_������� ���� � sca�old_�_������� -�.��� TSsd �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_�_�������_������� ���� � sca�old_�_������� -�.��� TSsd �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_�_�������_������� ���� � sca�old_�_������� -�.��� TSsd �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_�_�������_������� ���� � sca�old_�_������ -�.��� TSsd �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_�_������_������ ���� � sca�old_��_������ -�.��� TSsd �.�� upstream_gene_variant MODIFIER CDS_sca�old_��_������_������ ���� � sca�old_��_������ �.��� TSsd �.�� missense_variant MODERATE CDS_sca�old_��_������_������ � � sca�old_���_����� �.��� TSsd �.�� synonymous_variant LOW CDS_sca�old_���_�����_����� � � sca�old_���_����� �.��� TSsd �.�� upstream_gene_variant MODIFIER CDS_sca�old_���_�����_����� ���� � sca�old_��_������ -�.��� TSsd �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_��_������_������ ���� � sca�old_��_������ -�.��� TSsd �.�� synonymous_variant LOW CDS_sca�old_��_������_������ � � sca�old_��_������ �.��� TSsd �.�� missense_variant MODERATE CDS_sca�old_��_������_������ � � sca�old_�_������� �.��� TSsd �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_�_�������_������� ���� � sca�old_�_������� -�.��� TSsd �.�� downstream_gene_variant MODIFIER START_CODON_sca�old_�_�������_������� ���� � sca�old_��_������ �.��� TSsd �.�� missense_variant MODERATE CDS_sca�old_��_������_������ � � sca�old_�_������� -�.��� TSsd �.�� synonymous_variant LOW CDS_sca�old_�_�������_������� � � sca�old_�_������� -�.��� TSsd �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_�_�������_������� ���� � sca�old_�_������� -�.��� TSsd �.�� synonymous_variant LOW CDS_sca�old_�_�������_������� � � sca�old_��_������ �.��� mTwq �.�� synonymous_variant LOW CDS_sca�old_��_������_������ � � sca�old_��_���� -�.��� TSsd �.�� upstream_gene_variant MODIFIER START_CODON_sca�old_��_����_���� ��� � sca�old_��_����� -�.��� TSsd �.�� downstream_gene_variant MODIFIER CDS_sca�old_��_�����_����� ���� � sca�old_��_������ -�.��� PScv �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_��_������_������ ���� � sca�old_��_������ -�.��� PScv �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_��_������_������ ���� � sca�old_��_������ -�.��� PScv �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_��_������_������ ���� � sca�old_��_������ -�.��� PScv �.�� synonymous_variant LOW CDS_sca�old_��_������_������ � � sca�old_��_������ -�.��� PScv �.�� upstream_gene_variant MODIFIER CDS_sca�old_��_������_������ ���� � sca�old_��_���� -�.��� PScv �.�� upstream_gene_variant MODIFIER START_CODON_sca�old_��_����_���� ���� � sca�old_��_������ �.��� PScv �.�� synonymous_variant LOW CDS_sca�old_��_������_������ � sca�old_��_������ -�.��� TSsd �.�� downstream_gene_variant MODIFIER CDS_sca�old_��_������_������ ���� � sca�old_�_������ -�.��� TSsd �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_�_������_������ ���� � sca�old_��_������ �.��� TSsd �.�� synonymous_variant LOW CDS_sca�old_��_������_������ � � sca�old_��_������ -�.��� PScv �.�� upstream_gene_variant MODIFIER CDS_sca�old_��_������_������ ���� � sca�old_��_������ -�.��� PScv �.�� synonymous_variant LOW CDS_sca�old_��_������_������ � � sca�old_��_������ -�.��� PScv �.�� synonymous_variant LOW CDS_sca�old_��_������_������ � � sca�old_��_����� -�.��� PScv �.�� synonymous_variant LOW CDS_sca�old_��_�����_����� � � sca�old_�_������ �.��� PScv �.�� upstream_gene_variant MODIFIER CDS_sca�old_�_������_������ ���� � sca�old_��_����� -�.��� PScv �.�� upstream_gene_variant MODIFIER START_CODON_sca�old_��_�����_����� ��� � sca�old_��_������ -�.��� PScv �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_��_������_������ ���� � sca�old_��_������ -�.��� PScv �.�� missense_variant MODERATE CDS_sca�old_��_������_������ � � sca�old_��_������ -�.��� PScv �.�� missense_variant MODERATE CDS_sca�old_��_������_������ � � sca�old_��_������ -�.��� PScv �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_��_������_������ ���� � sca�old_�_������ -�.��� PScv �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_�_������_������ ���� � sca�old_��_������ �.��� TSsd �.�� synonymous_variant LOW CDS_sca�old_��_������_������ � � sca�old_�_������� �.��� PScv �.�� missense_variant MODERATE CDS_sca�old_�_�������_������� � � sca�old_��_������ -�.��� PScv �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_��_������_������ ���� � sca�old_�_������ �.��� TSsd �.�� missense_variant MODERATE CDS_sca�old_�_������_������ � � sca�old_��_������ -�.��� TSsd �.�� upstream_gene_variant MODIFIER CDS_sca�old_��_������_������ ���� � sca�old_�_������� -�.��� PScv �.�� upstream_gene_variant MODIFIER CDS_sca�old_�_�������_������� ���� � sca�old_�_������� �.��� TSsd �.�� synonymous_variant LOW CDS_sca�old_�_�������_������� � � sca�old_��_������� �.��� TSsd �.�� synonymous_variant LOW CDS_sca�old_��_�������_������� � � sca�old_��_������� �.��� TSsd �.�� downstream_gene_variant MODIFIER CDS_sca�old_��_�������_������� ���� � sca�old_��_������ �.��� TSsd �.�� synonymous_variant LOW CDS_sca�old_��_������_������ � � sca�old_��_������ �.��� TSsd �.�� synonymous_variant LOW CDS_sca�old_��_������_������ � � sca�old_��_������ -�.��� TSsd �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_��_������_������ ���� � sca�old_��_������ -�.��� TSsd �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_��_������_������ ���� � sca�old_��_������ -�.��� TSsd �.�� upstream_gene_variant MODIFIER START_CODON_sca�old_��_������_������ ��� � sca�old_��_������ -�.��� TSsd �.�� synonymous_variant LOW CDS_sca�old_��_������_������ � � sca�old_��_������ -�.��� TSsd �.�� upstream_gene_variant MODIFIER START_CODON_sca�old_��_������_������ ��� � sca�old_��_������ �.��� TSsd �.�� synonymous_variant LOW CDS_sca�old_��_������_������ � � sca�old_��_������ �.��� mTwq �.�� synonymous_variant LOW CDS_sca�old_��_������_������ � � sca�old_��_������� -�.��� TSsd �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_��_�������_������� ���� � sca�old_��_������ �.��� mTwq �.�� downstream_gene_variant MODIFIER START_CODON_sca�old_��_������_������ ���� � sca�old_��_������ �.��� mTwq �.�� downstream_gene_variant MODIFIER START_CODON_sca�old_��_������_������ ���� � sca�old_��_����� �.��� mTwq �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_��_�����_����� ���� � sca�old_��_������ -�.��� mTwq �.�� upstream_gene_variant MODIFIER START_CODON_sca�old_��_������_������ ���� � sca�old_��_������ -�.��� mTwq �.�� upstream_gene_variant MODIFIER CDS_sca�old_��_������_������ ���� � sca�old_��_������ �.��� mTwq �.�� missense_variant MODERATE CDS_sca�old_��_������_������ � � sca�old_��_������ �.��� mTwq �.�� downstream_gene_variant MODIFIER START_CODON_sca�old_��_������_������ ���� � sca�old_�_������� -�.��� mTwq �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_�_�������_������� ���� �� Table A.� Candidate SNPs and gene models along the �rst RDA axis, after accounting for neutral pop- ulation structure using the full set of ��,��� SNPs. RDA� Climate SNP Distance from Mycocosm SNP position Correlation Mycocosm gene location InterPro/KOG Desc loading variable category locus (bp) protein ID sca�old_��:������ -�.��� TSsd �.�� Intergenic ���� sca�old_��:������-������ ������ Ankyrin repeat sca�old_��:������ -�.��� TSsd �.�� Intergenic ���� sca�old_��:������-������ ������ Uncharacterized conserved protein sca�old_��:������� -�.��� TSsd �.�� Intergenic �� sca�old_��:�������-������� ������ Glycoside hydrolase, family � sca�old_��:������ -�.��� TSsd �.�� Synonymous � sca�old_��:������-������ ������ None sca�old_�:������� -�.��� TSsd �.�� Intergenic ���� sca�old_�:�������-������� ������ None sca�old_�:������� -�.��� TSsd �.�� Intergenic ���� sca�old_�:�������-������� ������ None sca�old_�:������� -�.��� TSsd �.�� Intergenic ���� sca�old_�:�������-������� ������ None sca�old_�:������� -�.��� TSsd �.�� Intergenic ���� sca�old_�:�������-������� ������ None sca�old_�:������� -�.��� TSsd �.�� Intergenic ���� sca�old_�:�������-������� ������ None sca�old_�:������� -�.��� TSsd �.�� Intergenic ���� sca�old_�:�������-������� ������ None sca�old_�:������ -�.��� TSsd �.�� Intergenic ��� sca�old_�:������-������ ������ None sca�old_��:������ -�.��� TSsd �.�� Intergenic �� sca�old_��:������-������ ������ Flavin-containing monooxygenase sca�old_��:������ �.��� TSsd �.�� Missense � sca�old_��:������-������ ������ Glycoside hydrolase, family � sca�old_���:����� �.��� TSsd �.�� Synonymous � sca�old_���:�����-����� ������ None sca�old_���:����� �.��� TSsd �.�� Intergenic ���� sca�old_���:�����-����� ����� Mg�+ transporter protein, CorA-like sca�old_��:������ -�.��� TSsd �.�� Intergenic ���� sca�old_��:������-������ ������ Glycoside hydrolase, family �� sca�old_��:������ -�.��� TSsd �.�� Synonymous � sca�old_��:������-������ ������ None sca�old_��:������ �.��� TSsd �.�� Missense � sca�old_��:������-������ ������ Glycoside hydrolase, family � sca�old_�:������� �.��� TSsd �.�� Intergenic �� sca�old_�:�������-������� ����� Alpha crystallin/Hsp�� domain sca�old_�:������� -�.��� TSsd �.�� Intergenic ��� sca�old_�:�������-������� ������ None sca�old_��:������ �.��� TSsd �.�� Missense � sca�old_��:������-������ ������ Glycoside hydrolase, family � sca�old_�:������� -�.��� TSsd �.�� Synonymous � sca�old_�:�������-������� ������ None sca�old_�:������� -�.��� TSsd �.�� Intergenic ���� sca�old_�:�������-������� ������ Protein kinase-like domain sca�old_�:������� -�.��� TSsd �.�� Synonymous � sca�old_�:�������-������� ������ None sca�old_��:������ �.��� mTwq �.�� Synonymous � sca�old_��:������-������ ������ Cytochrome P���, E-class, group I �� Table A.� Top �� candidate SNPs along the �rst three RDA axes, without accounting for neutral popu- lation structure using the LD-�ltered set of ��,��� SNPs. RDA RDA Climate SnpE� SnpE� distance SNP position Correlation SnpE� SNP category SnpE� annotation locus axis loading variable predicted e�ect from locus (bp) � sca�old_��_������ �.��� PScv �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_��_������_������ ���� � sca�old_�_������ -�.��� PScv �.�� upstream_gene_variant MODIFIER CDS_sca�old_�_������_������ ���� � sca�old_��_����� -�.��� PScv �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_��_�����_����� ���� � sca�old_��_������ �.��� mTwq �.�� downstream_gene_variant MODIFIER CDS_sca�old_��_������_������ ���� � sca�old_�_������� �.��� PScv �.�� downstream_gene_variant MODIFIER START_CODON_sca�old_�_�������_������� ���� � sca�old_��_������ -�.��� mTwq �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_��_������_������ ���� � sca�old_��_������ -�.��� PScv �.�� synonymous_variant LOW CDS_sca�old_��_������_������ � � sca�old_�_������� -�.��� PScv �.�� upstream_gene_variant MODIFIER START_CODON_sca�old_�_�������_������� ���� � sca�old_��_������ -�.��� mTwq �.�� synonymous_variant LOW CDS_sca�old_��_������_������ � � sca�old_�_������ -�.��� mTwq �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_�_������_������ ���� � sca�old_�_������� -�.��� mTwq �.�� upstream_gene_variant MODIFIER CDS_sca�old_�_�������_������� ���� � sca�old_��_������ -�.��� PScv �.�� synonymous_variant LOW CDS_sca�old_��_������_������ � � sca�old_��_������ -�.��� PScv �.�� intergenic_region MODIFIER CDS_sca�old_��_�����_�����-START_CODON_sca�old_��_������_������ � � sca�old_��_������ -�.��� mTwq �.�� upstream_gene_variant MODIFIER CDS_sca�old_��_������_������ ���� � sca�old_��_������ -�.��� PScv �.�� synonymous_variant LOW CDS_sca�old_��_������_������ � � sca�old_�_������ -�.��� PScv �.�� missense_variant MODERATE CDS_sca�old_�_������_������ � � sca�old_��_������ -�.��� PScv �.�� upstream_gene_variant MODIFIER CDS_sca�old_��_������_������ ���� � sca�old_�_������ �.��� mTwq �.�� upstream_gene_variant MODIFIER START_CODON_sca�old_�_������_������ ��� � sca�old_��_������ �.��� mTwq �.�� upstream_gene_variant MODIFIER START_CODON_sca�old_��_������_������ ��� � sca�old_��_������ �.��� mTwq �.�� synonymous_variant LOW CDS_sca�old_��_������_������ � � sca�old_��_������ �.��� mTwq �.�� missense_variant MODERATE CDS_sca�old_��_������_������ � � sca�old_��_������ �.��� mTwq �.�� upstream_gene_variant MODIFIER START_CODON_sca�old_��_������_������ ���� � sca�old_�_������ �.��� mTwq �.�� synonymous_variant LOW CDS_sca�old_�_������_������ � � sca�old_��_������ �.��� mTwq �.�� upstream_gene_variant MODIFIER CDS_sca�old_��_������_������ ���� � sca�old_��_������ �.��� mTwq �.�� synonymous_variant LOW CDS_sca�old_��_������_������ � � sca�old_��_������ �.��� mTwq �.�� synonymous_variant LOW CDS_sca�old_��_������_������ � � sca�old_��_����� �.��� mTwq �.�� upstream_gene_variant MODIFIER START_CODON_sca�old_��_�����_����� ��� � sca�old_��_������ �.��� mTwq �.�� missense_variant MODERATE CDS_sca�old_��_������_������ � � sca�old_��_������ �.��� mTwq �.�� missense_variant MODERATE CDS_sca�old_��_������_������ � � sca�old_��_������� �.��� mTwq �.�� upstream_gene_variant MODIFIER CDS_sca�old_��_�������_������� ���� � sca�old_��_������ -�.��� mTwq �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_��_������_������ ���� � sca�old_��_������ �.��� mTwq �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_��_������_������ ���� � sca�old_��_����� -�.��� mTwq �.�� splice_region_variant&synonymous_variant LOW CDS_sca�old_��_�����_����� � � sca�old_��_������ �.��� mTwq �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_��_������_������ ���� � sca�old_�_������ �.��� TSsd �.�� synonymous_variant LOW CDS_sca�old_�_������_������ � � sca�old_��_������ �.��� TSsd �.�� upstream_gene_variant MODIFIER CDS_sca�old_��_������_������ ���� � sca�old_��_������ �.��� TSsd �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_��_������_������ ���� � sca�old_��_������ �.��� mTwq �.�� upstream_gene_variant MODIFIER START_CODON_sca�old_��_������_������ ��� � sca�old_�_������ �.��� mTwq �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_�_������_������ ���� � sca�old_��_����� �.��� TSsd �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_��_�����_����� ���� � sca�old_�_������ �.��� TSsd �.�� missense_variant MODERATE CDS_sca�old_�_������_������ � � sca�old_�_����� �.��� mTwq �.�� synonymous_variant LOW CDS_sca�old_�_�����_����� � � sca�old_�_������ �.��� mTwq �.�� synonymous_variant LOW CDS_sca�old_�_������_������ � � sca�old_�_������� �.��� mTwq �.�� upstream_gene_variant MODIFIER START_CODON_sca�old_�_�������_������� ���� � sca�old_��_������ �.��� mTwq �.�� upstream_gene_variant MODIFIER CDS_sca�old_��_������_������ ���� � sca�old_��_������ �.��� TSsd �.�� upstream_gene_variant MODIFIER CDS_sca�old_��_������_������ ���� � sca�old_��_������ �.��� TSsd �.�� synonymous_variant LOW CDS_sca�old_��_������_������ � � sca�old_��_����� �.��� TSsd �.�� upstream_gene_variant MODIFIER START_CODON_sca�old_��_�����_����� ���� � sca�old_��_������ �.��� PScv �.�� upstream_gene_variant MODIFIER CDS_sca�old_��_������_������ �� � sca�old_�_����� �.��� TSsd �.�� upstream_gene_variant MODIFIER CDS_sca�old_�_�����_����� ���� � sca�old_��_������ -�.��� mTwq �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_��_������_������ ���� � sca�old_��_����� �.��� mTwq �.�� missense_variant MODERATE CDS_sca�old_��_�����_����� � � sca�old_��_������ �.��� mTwq �.�� intergenic_region MODIFIER CDS_sca�old_��_������_������-START_CODON_sca�old_��_������_������ � � sca�old_��_������ �.��� mTwq �.�� upstream_gene_variant MODIFIER CDS_sca�old_��_������_������ ���� � sca�old_�_������� �.��� Pwq �.�� upstream_gene_variant MODIFIER START_CODON_sca�old_�_�������_������� ���� � sca�old_��_������ �.��� Pwq �.�� upstream_gene_variant MODIFIER CDS_sca�old_��_������_������ ���� � sca�old_��_������ �.��� Pwq �.�� downstream_gene_variant MODIFIER CDS_sca�old_��_������_������ ���� � sca�old_��_������ �.��� Pwq �.�� missense_variant MODERATE CDS_sca�old_��_������_������ � � sca�old_��_������ �.��� PScv �.�� missense_variant MODERATE CDS_sca�old_��_������_������ � � sca�old_��_������ �.��� mTwq �.�� synonymous_variant LOW CDS_sca�old_��_������_������ � � sca�old_�_������� �.��� mTwq �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_�_�������_������� ���� � sca�old_��_������ �.��� PScv �.�� upstream_gene_variant MODIFIER CDS_sca�old_��_������_������ ���� � sca�old_��_����� �.��� PScv �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_��_�����_����� ���� � sca�old_��_������ �.��� Pwq �.�� synonymous_variant LOW CDS_sca�old_��_������_������ � � sca�old_��_������ �.��� Pwq �.�� upstream_gene_variant MODIFIER CDS_sca�old_��_������_������ ���� � sca�old_�_������ �.��� Pwq �.�� upstream_gene_variant MODIFIER CDS_sca�old_�_������_������ ���� � sca�old_�_������ �.��� mTwq �.�� upstream_gene_variant MODIFIER CDS_sca�old_�_������_������ ���� � sca�old_��_������ -�.��� Pwq �.�� missense_variant MODERATE CDS_sca�old_��_������_������ � � sca�old_��_����� �.��� PScv �.�� upstream_gene_variant MODIFIER CDS_sca�old_��_�����_����� �� � sca�old_��_������ -�.��� Pwq �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_��_������_������ ���� � sca�old_��_������ -�.��� mTwq �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_��_������_������ ���� � sca�old_��_������ �.��� mTwq �.�� synonymous_variant LOW CDS_sca�old_��_������_������ � � sca�old_�_������ -�.��� mTwq �.�� missense_variant MODERATE CDS_sca�old_�_������_������ � � sca�old_��_������ -�.��� mTwq �.�� synonymous_variant LOW CDS_sca�old_��_������_������ � � sca�old_��_������ -�.��� mTwq �.�� synonymous_variant LOW CDS_sca�old_��_������_������ � � sca�old_��_������ -�.��� PScv �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_��_������_������ ���� � sca�old_��_������� �.��� mTwq �.�� upstream_gene_variant MODIFIER CDS_sca�old_��_�������_������� ���� � sca�old_��_������ -�.��� mTwq �.�� upstream_gene_variant MODIFIER CDS_sca�old_��_������_������ ���� � sca�old_��_������ -�.��� mTwq �.�� synonymous_variant LOW CDS_sca�old_��_������_������ � � sca�old_��_������ -�.��� PScv �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_��_������_������ ���� � sca�old_��_������ �.��� mTwq �.�� missense_variant MODERATE CDS_sca�old_��_������_������ � � sca�old_�_������ -�.��� PScv �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_�_������_������ ���� � sca�old_��_������ -�.��� mTwq �.�� synonymous_variant LOW CDS_sca�old_��_������_������ � � sca�old_��_������ �.��� mTwq �.�� missense_variant MODERATE CDS_sca�old_��_������_������ � � sca�old_�_������ -�.��� mTwq �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_�_������_������ ���� � sca�old_��_����� �.��� mTwq �.�� missense_variant MODERATE CDS_sca�old_��_�����_����� � � sca�old_�_������ �.��� mTwq �.�� synonymous_variant LOW CDS_sca�old_�_������_������ � � sca�old_��_����� �.��� mTwq �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_��_�����_����� ���� � sca�old_��_������ -�.��� mTwq �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_��_������_������ ���� � sca�old_��_������ -�.��� mTwq �.�� upstream_gene_variant MODIFIER STOP_CODON_sca�old_��_������_������ ���� � sca�old_��_������ -�.��� mTwq �.�� missense_variant MODERATE CDS_sca�old_��_������_������ � �� Figure A.� (A) Rooted phylogeny reconstructed using the M. phaseolina reference genome as outgroup. Maximum-likelihood phylogeny reconstructed using ��,��� high-quality SNPs. Bootstrap support val- ues over �� are shown at nodes. Bootstrapping converged after ��� replicates. Colored tips represent the genetic cluster for each isolate as de�ned by principal components analysis. Individual isolate names include ANSI/ISO codes for US states, and Colombia and Puerto Rico municipalities: CA: California, CAU: Cauca, GA: Georgia, IN: Indiana, ISA: Isabela, JD: Juana Diaz, KY: Kentucky, LA: Louisiana, MAG: Magdalena, MI: Michigan, MN: Minnesota, MS: Mississippi, SC: South Carolina, SD: South Dakota, TN: Tennessee, TOL: Tolima, TX: Texas, VAC: Valle del Cauca, WI: Wisconsin. ISO country codes: US: United States, COL: Colombia and PR: Puerto Rico. (B) Discriminatory analysis of principal components. Each bar and color indicates the posterior probability membership value per isolate to one of the �ve genetic clusters. �� Figure A.� Principal component analysis (PCA) showing isolate host origin. Scatterplot from a princi- pal component analysis based on the two �rst PCs (the eigenvectors of the ��,��� SNPs) for all isolates. Points are colored by host from which isolates were collected. Overlapping ellipses representing ��% of the isolates from each of the hosts. �� Figure A.� Spatial population structure using conStruct. (A) Maps of admixture proportions estimated for M. phaseolina across the US, Puerto Rico and Colombia using the spatial conStruct model for K = � to K = �. Pies show mean admixture results for individual isolates within their diameter. (B) Cross- validation predictive accuracy values as a function of the number of layers (K = �-�) for the spatial and nonspatial conStruct models. (C) Layer contributions for K = � through �. �� Figure A.� MLLs shared among countries. MLL �: one isolate from Colombia (Mph-�) and one from Puerto Rico (UPR-Mph-JD�) clustering in COLPR�, MLL ��: one isolate from Puerto Rico (UPR-Mph- ISA�) and one from Louisiana (TN���) clustering in COLPR�, and MLL ��: one isolate from Colombia (Mph-��) and �� isolates from US clustering in US�A . The two MLLs for isolates IN���-� and Mph�� are not shown. �� Figure A.� Spatial structure variables identi�ed using distance-based Moran’s eigenvector maps (db- MEMs �-�). The variable dbMEM � identi�ed as signi�cant using forward-variable selection described broad spatial structure. Color and size of the points correspond to the sign (+ or -) and magnitude of the dbMEM variables, respectively. �� Figure A.� Venn diagram showing the overlap between outlier loci identi�ed by both partial RDA (con- strained on neutral population structure) and full RDA (unconstrained) models using unlinked SNPs (LD-�ltered set of ��,��� SNPs). �� Figure A.� Maximum-likelihood phylogeny reconstructed using concatenated sequences of the Internal Transcribed Spacer regions for the nuclear rDNA operon (ITS), part of the Translation Elongation Factor (TEF-�) gene region, and part of the actin (ACT) gene region. �� CHAPTER � SENSITIVITY TO SINGLE-SITE FUNGICIDES IN MACROPHOMINA PHASEOLINA POPULATIONS FROM SOYBEAN AND DRY BEAN �� �.� Abstract Charcoal rot, caused by Macrophomina phaseolina, is a soil- and seedborne disease that a�ects soybean and dry bean production worldwide. Strategies for e�ectively managing charcoal rot are limited, and management has primarily focused on varietal resistance and cultural practices. Fungicide e�cacy studies conducted in past years have focused on older active ingredients and information on the sensitivity of M. phaseolina to newer classes of single-site fungicides is lacking. Although not speci�cally targeting M. phaseolina, single-site fungicides are used in soybean and dry bean production as seed treatments, soil applications, and foliar sprays. The in-vitro sensitivity of �� M. phaseolina isolates collected from soybean and dry bean in the United States, Puerto Rico and Colombia was assessed for three classes of single-site fungicides widely used in soybean and dry bean production. The relative mycelial growth of M. phaseolina isolates challenged against boscalid (SDHI), iprodione (dicarboximide) and prothioconazole (DMI) was used to determine the e�ective concentration to inhibit mycelial growth by ��% (EC�� ). All �� isolates were sensitive to boscalid, iprodione and prothioconazole. Mean EC�� values for boscalid, iprodione, and prothioconazole were �.��, �.�� and �.�� �g ml � respectively. The full-length nucleotide sequences of fungicide target genes were assembled to investigate mutations in all isolates. Mutations found in target genes did not associate with levels of M. phaseolina fungicide sensitivity. �.� Introduction Soil-borne fungal pathogens are a major threat to crops and food security and fungicides are key com- ponents of e�ective disease management to prevent yield loss and ensure high-quality crop production. Since the ����’s fungicide use has increased, partly with the advent of broad-spectrum systemic single- site fungicides such as dicarboximides, sterol biosynthesis inhibitors including demethylation inhibitors (DMIs; azoles), and succinate dehydrogenase inhibitors (SDHIs) (Russell, ����). Changes in cultural practices such as reduced or no-tillage systems, which add complexity to disease dynamics by favoring pathogen inoculum in crop residue, have further contributed to the increased use of fungicides (Oerke, ����; Morton and Staub, ����). Only a few years after the commercial use of fungicides, acquired re- sistance became a signi�cant threat to their e�cacy (Kuck and Russell, ����; Leadbeater et al., ����). Therefore, globally, as well as in the US, monitoring for development of resistance is an important com- �� ponent for the implementation of e�ective disease management strategies (Brent and Hollomon, ����). Charcoal rot disease, caused by the soil- and seed-borne pathogen Macrophomina phaseolina, has been recognized as a threat of increasing importance to soybean (Glycine max) and dry bean (Phaseolus vulgaris) production in the US and worldwide (Dhingra and Sinclair, ����; Wrather et al., ����; Reznikov et al., ����; Jacobs et al., ����; Savary et al., ����; Bradley et al., ����). In the �eld, charcoal rot typically develops at reproductive stages of soybean and dry bean. However, infection may occur at emergence and early in the growing season causing up to ���% incidence of seedling infection � to � weeks after plant- ing causing seedling blight (Hartman et al., ����a; Hartman et al., ����b). Seedling disease is most often reported in tropical regions, however in temperate regions damage to soybean seedlings is also observed, particularly under high temperature and low soil moisture conditions (Meyer and Sinclair, ����; Hart- man et al., ����a). Infection begins, most commonly, with microsclerotia present in soil or plant residue. Microsclerotia germination followed by appressoria development allows host penetration through the root epidermis with subsequent invasion of root and stem tissue. Alternatively, colonization can occur from infected seed. Eventually M. phaseolina colonizes the vascular system leading to wilting, necrosis, and plant death (Hartman et al., ����a). M. phaseolina reproduction in infected plants produces abun- dant microsclerotia which, following plant death and crop harvest, can survive in crop residue and in soil for years (Dhingra and Sinclair, ����). Although M. phaseolina can be a devastating pathogen, it can also colonize plants asymptomatically, and it is recognized as an endophyte and latent pathogen in many plant species (Dhingra and Sinclair, ����; Slippers and Wing�eld, ����; Slippers and Boissin, ����; Parsa et al., ����; Crous et al., ����). Management of charcoal rot in soybean and dry bean relies mostly on host genetic resistance which is limited in both crops and cultural control measures, which may be challenging to implement (Pastor-Corrales et al., ����; Hartman et al., ����a; Coser et al., ����; Romero Luna et al., ����; Ambachew et al., ����). Chemical-control strategies are aimed at reducing microsclero- tia in soil and limiting host colonization (Romero Luna et al., ����). Fungicide seed treatments and soil applications can provide protection by delaying colonization and reducing fungal growth within root, stem and vascular tissue (Bradley, ����). Seed treatments with benomyl (benzimidazole) and carboxin (�rst generation succinate dehydrogenase inhibitor) showed some e�ectiveness in reducing incidence of �� charcoal rot in dry bean seedlings under greenhouse conditions (Abawi and Pastor-Corrales, ����). Sim- ilarly, soybean seed treated with thiophanate methyl + pyraclostrobin protected plant emergence in �eld inoculation experiments (Reznikov et al., ����). Recent studies evaluated the in-vitro sensitivity of M. phaseolina to di�erent fungicide classes using a single isolate (Tonin et al., ����; Chaudhary et al., ����). However, current chemical-control strategies for charcoal rot do not provide consistent e�ective control and information on the e�ectiveness of newer fungicides chemistries using a collection of M. phaseolina isolates is lacking (Reznikov et al., ����; Romero Luna et al., ����; Roth et al., ����). In soybean and dry bean, management strategies commonly include the use of single-site fungicides as seed treatments and foliar applications (Hartman et al., ����b; Lehner et al., ����; Bandara et al., ����; Karavidas et al., ����). Most single-site fungicides target mitochondrial respiration function, the cytoskeleton or ergosterol biosynthesis. The demethylation inhibitors (DMIs) are the most important group of fungicides currently used in crop protection, leading the world fungicide market (Leadbeater et al., ����). DMIs inhibit the C��-demethylation step of ergosterol biosynthesis interfering with mem- brane integrity. The succinate dehydrogenase inhibitors (SDHIs) fungicides target the succinate dehydro- genase (mitochondrial complex II in the electron transfer chain), thereby inhibiting fungal respiration. Dicarboximides cause cell death through interference with osmotic signal transduction pathway via in- appropriate activation of the osmosensing class III histidine kinase (Motoyama et al., ����; Yamaguchi and Fujimura, ����)(Fungicide Resistance Action Committee, FRAC: www.frac.info). DMIs, SDHIs and dicarboximides are considered either medium or medium to high risk for the de- velopment of fungicide resistance (FRAC: www.frac.info) and shifts in fungicide sensitivity have been reported in important crops for these three classes (Brent and Hollomon, ����; Hartman et al., ����b; Leadbeater et al., ����). The most common mechanisms of resistance to DMIs, SDHIs and dicarbox- imides are changes in the amino acids of the target proteins. Single point mutations in Sdh succinate dehydrogenase (Sang and Lee, ����), os� histidine kinase, and cyp�� C��-demethylase genes are known to confer reduced sensitivity to boscalid, iprodione and prothioconazole, respectively, in several fungal pathogens (FRAC: www.frac.info). In addition, cyp�� genes overexpression (Schnabel and Jones, ����; Nikou et al., ����; Wei et al., ����) and promoter insertions have been associated with DMI-reduced �� sensitivity of phytopathogenic fungi. Single-site systemic fungicides are highly e�ective, meaning that most individuals are either killed or inhibited resulting in selection for any resistant individuals (Lucas et al., ����). Factors such as fungi- cide distribution in the plant tissues and dilution to non-lethal doses may lead to the development of resistance not only in the target pathogen but in other fungal pathogens or the plant-associated fungal community (Brent and Hollomon, ����; Chamberlain et al., ����). Additionally, large-scale homoge- neous agricultural systems which often have low crop genetic diversity and can sustain large and rapidly reproducing pathogen populations constitute conducive environments for the evolution of resistance (Brent and Hollomon, ����). Selection for resistance can occur in any environment containing fungicides. The risk of fungicide resistance depends mainly on the fungicide mode of action and speci�city (e.g. multisite vs. single-site), the biological characteristics of the fungi, such as reproduction mode and rate of reproduction, and agro- nomic factors related to appropriate fungicide use (Leadbeater et al., ����). In addition, pathogen demo- graphic history, for example greater inoculum load leading to increases of e�ective population sizes, and the existence of �tness trade-o�s may also play an important role in the development of fungicide resis- tance (McDonald and Stukenbrock, ����; Hawkins and Fraaije, ����). Although evolution of resistance to fungicides has been characterized for many fungal pathogens, there are few studies that assessed the fungicide sensitivity and potential mechanisms of resistance in M. phaseolina. Overall, we consider it likely that M. phaseolina is commonly exposed to fungicides used in soybean and dry bean production and that conditions associated with the use of fungicides to protect crops against economically important fungal pathogens could favor the development of fungicide resistance in M. phaseolina populations either as direct or o�-target e�ect. We therefore hypothesize that selection for re- sistance may occur in the internal tissues of plants or seeds treated with fungicides as well as in crop residue and soil containing fungicide residue. Additionally, we hypothesize that populations in tropical and sub- tropical regions in which environmental conditions could allow for year-round pathogen multiplication and therefore sustain large populations, may be at higher risk of developing resistance. The objectives of this study were i) to investigate boscalid, iprodione and prothioconazole in-vitro sensitivity of �� M. �� phaseolina isolates collected from soybean and dry bean from the US, Colombia, and Puerto Rico, and ii) identify mutations in the Sdh, os� and cyp�� target genes of M. phaseolina isolates and examine their asso- ciation with levels of M. phaseolina in-vitro fungicide sensitivity. For this, we conducted mycelial growth assays and investigated fungicide target genes from M. phaseolina whole-genome sequences. This study provides information for e�ective use of active ingredients in current commercial fungicide formulations and aid in the designing of e�ective disease management strategies. �.� Results �.�.� EC�� determination for �� M. phaseolina isolates The EC�� values of �� M. phaseolina isolates were determined based on mycelial growth on Petri plates amended with di�erent concentrations of boscalid, iprodione and prothioconazole (Table �.�). Isolate mean EC�� values were not di�erent across experiments, with con�dence intervals that overlapped zero, for boscalid (Bos�—Bos�: -�.��� �g ml � ; ��% CI -�.��� — �.���) and iprodione (Ipro�—Ipro�: �.��� �g ml � ; ��% CI -�.��� — �.���) (Table A.�). For prothioconazole, although the CI did not overlap zero (Pro�—Pro�: -�.��� �g ml � ; ��% CI -�.��� — -�.���), mean EC�� di�erence was less than �.�� �g ml � , as for boscalid and iprodione (Table A.�). Overall, mean EC�� di�erences indicate that the EC�� values were consistent across experiments for all fungicides (Figure A.�). The EC�� values for boscalid ranged from �.�� to �.�� �g ml � with mean �.�� �g ml � , though only two isolates (CR-Red-�B and MP���) had an EC�� more than � �g ml � (Table �.�). For iprodione, the EC�� of isolates ranged from �.�� to �.�� �g ml � with mean �.�� �g ml � (Table �.�). Isolates were most sensitive to prothioconazole with EC�� values ranging from �.�� to �.�� �g ml � , with mean of �.�� �g ml � (Table �.� and �.�). These results indicate that M. phaseolina isolates evaluated were sensitive to the three fungicides tested. Isolate sensitivity di�ered across the three fungicides. Isolates were most sensitive to prothioconazole, followed by boscalid and least sensitive to iprodione (Figure �.�). Mean EC�� di�erences were �.�� �g ml � (��% CI �.�� — �.��) and -�.�� �g ml � (��% CI -�.�� — -�.��) for iprodione and prothioconazole as compared to boscalid, respectively, and -�.�� �g ml � (��% CI -�.�� — -�.��) for prothioconazole as compared to iprodione (Table A.�). �� Table �.� Mean EC�� (e�ective concentration to reduce growth by ��%) estimates for �� M. phaseolina isolates determined from mycelial growth assays in Petri plates amended with di�erent concentrations of fungicides boscalid, iprodione or prothioconazole. Isolate Boscalid Iprodione Prothioconazole Standard Standard Standard EC�� EC�� EC�� Error Error Error CR_Red_� �.��� �.��� �.��� �.��� �.��� �.��� CR_Red_�B �.��� �.��� �.��� �.��� �.��� �.��� CR_Red_� �.��� �.��� �.��� �.��� �.��� �.��� Dm�� �.��� �.��� �.��� �.��� �.��� �.��� Et�� �.��� �.��� �.��� �.��� �.��� �.��� Et�� �.��� �.��� �.��� �.��� �.��� �.��� Et�� �.��� �.��� �.��� �.��� �.��� �.��� IN��_�_� �.��� �.��� �.��� �.��� �.��� �.��� IN��_�_� �.��� �.��� �.��� �.��� �.��� �.��� IN��_�_� �.��� �.��� �.��� �.��� �.��� �.��� IN��_PO_� �.��� �.��� �.��� �.��� �.��� �.��� Md� �.��� �.��� �.��� �.��� �.��� �.��� Md� �.��� �.��� �.��� �.��� �.��� �.��� MI-SF �-�� �.��� �.��� �.��� �.��� �.��� �.��� MI-SF ��-�� �.��� �.��� �.��� �.��� �.��� �.��� MI-SF �-�� �.��� �.��� �.��� �.��� �.��� �.��� MI-SF �-� �.��� �.��� �.��� �.��� �.��� �.��� MISO���-� �.��� �.��� �.��� �.��� �.��� �.��� MISO���-� �.��� �.��� �.��� �.��� �.��� �.��� MISO���-� �.��� �.��� �.��� �.��� �.��� �.��� MP��� �.��� �.��� �.��� �.��� �.��� �.��� MP��� �.��� �.��� �.��� �.��� �.��� �.��� MP��� �.��� �.��� �.��� �.��� �.��� �.��� SAG�-� �.��� �.��� �.��� �.��� �.��� �.��� TN��� �.��� �.��� �.��� �.��� �.��� �.��� TN��� �.��� �.��� �.��� �.��� �.��� �.��� TN� �.��� �.��� �.��� �.��� �.��� �.��� TN� �.��� �.��� �.��� �.��� �.��� �.��� TN��� �.��� �.��� �.��� �.��� �.��� �.��� TN��� �.��� �.��� �.��� �.��� �.��� �.��� W-MISO� �-� �.��� �.��� �.��� �.��� �.��� �.��� W-MISO� �-�� �.��� �.��� _ _ �.��� �.��� W�� �.��� �.��� �.��� �.��� �.��� �.��� W�� �.��� �.��� �.��� �.��� �.��� �.��� W�-� �.��� �.��� �.��� �.��� �.��� �.��� �� �.�.� Isolate screening and EC�� prediction using single concentrations Single concentrations of boscalid � �g ml � , iprodione � �g ml � , or prothioconazole �.� �g ml � were used to screen the remaining �� M. phaseolina isolates. EC��(P) values for each isolate were predicted af- ter linear regression models based on RMG at the single concentration (Figure A.�). For all fungicides, EC��(P) values were within the range of the EC�� values estimated for the set of �� M. phaseolina isolates initially tested (Table �.� and A.�). None of the �� isolates were found to have EC��(P) values above the threshold to be categorized as less sensitive (�.��, �.��, and �.�� �g ml � for boscalid, iprodione and pro- thioconazole, respectively) (Figure �.�). Consistent with the EC�� estimations for the initial set, isolates were most sensitive to prothioconazole and least sensitive to iprodione (Figure A.�). Table �.� Mean and range EC�� (e�ective concentration to reduce growth by ��%) estimates and predic- tions for di�erent sets of M. phaseolina isolates determined from mycelial growth assays in Petri plates amended with multiple or single concentrations of boscalid, iprodione or prothioconazole. Set EC�� type Boscalid Iprodione Prothioconazole Mean Mean Mean EC�� range EC�� range EC�� range EC�� EC�� EC�� min max min max min max Multiple concentrations EC�� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �� isolatesa Single concentration EC��(P) �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �� isolatesb Combined and validation EC�� and EC��(P) �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� sets �� isolatesc a EC Estimates for �� M. phaseolina isolates determined from mycelial growth assays in Petri plates �� amended with di�erent concentrations of boscalid, iprodione or prothioconazole. b EC��(P) predictions for �� M. phaseolina isolates determined from mycelial growth assays in Petri plates amended with single concentrations of boscalid � �g ml � , iprodione � �g ml � , or prothioconazole �.� �g ml � . c Combined EC�� and EC��(P) values for �� M. phaseolina isolates. This combined data set consists of multiple con- centration, single concentration and validation data points, including two additional isolates from the validation set. Validation of linear regression models for EC��(P) prediction was conducted by estimating EC�� val- ues and correlating them with predicted EC��(P) values in a validation set of �� isolates (Table A.�). These validation isolates were selected to represent the range of EC�� /EC��(P) values previously determined and included two isolates not tested in any of the previous assays. A signi�cant positive linear relation- ship was observed between estimated EC�� and predicted EC��(P) values for all fungicides (Pearson’s R = �� �.��, �.�� and �.�� for boscalid, iprodione and prothioconazole, respectively; P < �.����) (Figure A.�). �.�.� Combined EC�� and EC��(P) for �� M. phaseolina isolates EC�� and EC��(P) values were combined to analyze the in-vitro fungicide sensitivity of the �� M. phase- olina isolates, including data points and isolates from the validation set. The combined EC�� and EC��(P) distribution for the �� isolates ranged from �.�� to �.�� �g ml � with a mean EC�� of �.�� �g ml � for boscalid. For iprodione, EC�� values ranged from �.�� to �.�� �g ml � with a mean value of �.�� �g ml � . For prothioconazole, EC�� ranged from �.�� to �.��, with mean of �.�� �g ml � (Figure �.�). Isolates were most sensitive to prothioconazole, followed by boscalid and least sensitive to iprodione, as indicated by mean EC�� di�erences between fungicides for the �� isolates (Figure �.�, Table �.�). While no resistant isolates were identi�ed, isolates with EC�� /EC��(P) values above three standard deviations from the mean (�.��, �.��, and �.�� �g ml � for boscalid, iprodione and prothioconazole, respectively) were categorized as less sensitive (Figure �.�, Table A.�). Less sensitive isolates were CR-Red�B and MpSDSU to boscalid; MP��� to iprodione; and MP��� and MP��� to prothioconazole (Table A.�). Table �.� Mean EC�� /EC��(P) di�erences for �� M. phaseolina isolates across fungicides: boscalid, ipro- dione and prothioconazole. Pairwise fungicide comparison Mean EC�� di�erence ��% CIa low high Iprodione minus Boscalid �.��� �.��� �.��� Prothioconazole minus Boscalid -�.��� -�.��� -�.��� Prothioconazole minus Iprodione -�.��� -�.��� -�.��� a ��% con�dence intervals adjusting for asymmetrical resampling distributions using bias-corrected and accelerated bootstrap (BCa bootstrap). To examine whether isolates di�er in their sensitivity to each fungicide by host or genetic related- ness, EC�� /EC��(P) values were examined by isolate soybean or dry bean origin and genetic cluster (as determined in chapter �). No di�erences in sensitivity were found between soybean and dry bean isolates for any of the three fungicides (Figure A.�). Isolates collected from strawberry (M��-�� and M��-��) and Ethiopian mustard (MpSDSU) were not included in this analysis because of the low number of isolates from each of these hosts. When analyzed by genetic cluster, isolates in the US�A, US�B and COLPR� ge- �� netic cluster were found to be, on average, more sensitive to the three fungicides, as compared with isolates in US� and COLPR� genetic clusters (Figure S�, Table A.�). Although, higher mean EC�� values were obtained for isolates in the US� and COLPR� genetic groups, mean EC�� di�erences between genetic clusters were all < �.�, �.� and �.�� �g ml � for boscalid, iprodione and prothioconazole, respectively (Table A.�). All isolates were found to be sensitive to boscalid, iprodione and prothioconazole indicating their potential use for M. phaseolina management. Though in-vitro assays provide an initial assessment, in-vivo and �eld e�cacy testing is necessary to determine whether they provide protection under �eld conditions. �.�.� Mutations in fungicide target genes and associations to isolate sensitivity The predicted sequences of SdhB, cyp��, and os� genes of the �� M. phaseolina isolates were obtained to examine whether mutations were associated with fungicide sensitivity. Species sequence alignment of the translated SdhB, cyp��, and os� genes sequences revealed high amino acid identity among all �� isolates and with the reference sequences. Conservation of the SdhB iron-sulfur subunit was detected across all isolates. The cyp��B sequence was ��� amino acids with the heme-binding domain detected at codons ���-��� including the heme coordinating cysteine at codon ���. The predicted cyp��A sequence was ��� amino acids long. Amino acid sequence alignments of the �� M. phaseolina isolates detected cyp��B mutations in four isolates collected from dry bean in Colombia and two isolates collected from soybean in the US (Table �.�). Similarly, mutations in cyp��A sequence were detected in �� isolates col- lected from dry bean in Colombia or Puerto Rico (Table �.�). SdhB mutations were identi�ed in three dry bean isolates from Colombia and one isolate from Puerto Rico. None of the mutations were found to be associated with reduced fungicide sensitivity. Structural modeling of the target proteins localized all mutations outside of the binding pocket and with low probability of a�ecting binding a�nity of the fungicide molecules. �.� Discussion This study provides information on the in-vitro e�cacy against M. phaseolina of three single-site fungi- cides widely used in crop production worldwide. A total of �� M. phaseolina isolates collected mainly from soybean and dry bean across the US, Colombia and Puerto Rico were characterized for their in-vitro �� Table �.� Mutations in fungicide target genes found in M. phaseolina isolates. Host and geographic ori- gins of isolates are shown. Gene Mutation Isolate Origin cyp��B V��I Mph�� Dry bean Colombia V��I Mph�� Dry bean Colombia V��I Mph�� Dry bean Colombia I��T MP��� Dry bean Colombia H���Y IN���-� Soybean North H���Y MP��� Soybean South cyp��A V���F Mph��, Mph�� Dry bean Colombia Mph��, Mph��, Mph��, Mph��, Mph��, Dry bean Colombia D���Y Mph��, Mph��, Mph�, UPR_ISA�, Dry bean Puerto Rico UPR_JD�, UPR_JD�, UPR_JD� sdhB A��T Mph��, Mph��, Mph�� Dry bean Colombia V���L UPR_ISA� Dry bean Puerto Rico sensitivity to boscalid (SDHI), iprodione (Dicarboximide) and prothioconazole (DMI). This represents the largest assessment of M. phaseolina variation in fungicide sensitivity in these countries. Prior stud- ies characterizing the in-vitro sensitivity of M. phaseolina to di�erent classes of fungicides have focused on older active ingredients or have used a limited number of isolates (Tonin et al., ����; Chaudhary et al., ����). We demonstrated that M. phaseolina isolates from soybean and dry bean were sensitive to boscalid, iprodione and prothioconazole active ingredients when tested in mycelial growth assays. While isolate variability in EC�� values to these fungicides was present, no isolate was insensitive to any of the tested fungicides. Notably, we found signi�cant di�erences in M. phaseolina sensitivity to the three fungi- cides tested. Prothioconazole was the most e�cacious active ingredient in reducing fungal growth (mean EC�� /EC��(P) = �.�� �g ml � ), as compared to boscalid (mean EC�� /EC��(P) = �.�� �g ml � ) and ipro- dione (mean EC�� /EC��(P) = �.�� �g ml � ). In a study of M. phaseolina in Brazil, iprodione inhibited mycelial growth of a soybean isolate (EC�� = �.�� �g ml � ) (Tonin et al., ����). To our knowledge, similar in-vitro studies reporting EC�� results of boscalid, iprodione or prothioconazole have not been conducted for a collection of M. phaseolina isolates. Our study found no resistant isolates across M. phaseolina genetic clusters in the US (US�A, US�B and US�), Colombia and Puerto Rico (COLPR� and COLPR�). Variation between genetic clusters was �� Figure �.� EC�� distribution of �� M. phaseolina isolates collected from soybean and dry bean. Less sen- sitive isolates were designated as the ones with an EC�� value higher than three standard deviations from the mean. observed with a general trend of least sensitivity in US� and COLPR� clusters as compared to US�A, US�B and COLPR� cluster across fungicides. We hypothesized that M. phaseolina isolates may develop fungicide resistance in the context of soybean and dry bean production. Furthermore, we hypothesized that Colombian and Puerto Rican isolates may have higher risk of developing fungicide resistance as com- pared to US isolates. The reason for this was that environmental conditions in tropical locations, such as Colombia and Puerto Rico, allow for year-round permanence of M. phaseolina, and therefore, the poten- tial maintenance of large pathogen populations in soil and crop residue. Additionally, the higher genetic diversity found in M. phaseolina Colombian-Puerto Rican genetic clusters (COLPR� and COLPR�), as compared to US clusters (US�A, US�B and US�) (Ortiz et al., under review) led us to consider these populations may be at greater risk of developing resistance. However, our results did not support this hypothesis. Instead, we found isolates in all genetic clusters were sensitive to the three fungicides tested �� Figure �.� Mean EC�� di�erences of �� M. phaseolina isolates across fungicides: (A) Boscalid-Iprodione (B) Boscalid-Prothioconazole (C) Iprodione-Prothioconazole pairwise comparisons. Isolates were most sensitive to prothioconazole, followed by boscalid and least sensitive to iprodione. and most isolates from Colombia and Puerto Rico (grouped in COLPR� genetic cluster) had similar sensitivities to those in other genetic clusters. This suggests that low selection pressure and/or low in- trinsic M. phaseolina risk for fungicide resistance development. Generally, a single fungicide application is conducted during the growing season in dry bean and soybean, in commercial �elds (Hartman et al., ����b) as well as in experimental plots from where the Colombian and Puerto Rican isolates in this study were collected (Gloria Mosquera and Consuelo Estevez, personal communication). In addition, although population sizes may be high, e�ective population sizes may remain low due to the mostly clonal nature of M. phaseolina. Resistance to SDHIs, DMIs or Dicarboximide has not been reported in M. phaseolina (Sang and Lee, ����). Overall, this study indicates that M. phaseolina has a low risk of developing resistance. A limitation of this study is that isolates from Colombia and Puerto Rico were collected only from dry bean experimental plots. A previous study conducted in a dry bean producing region in Colombia reported that fungicides were applied several times during the growing season (Velasquez et al., ����). Future studies involving isolates collected from commercial �elds in Colombia and Puerto Rico would provide a broader assessment of M. phaseolina fungicide sensitivity in these countries. Information regarding mutations in M. phaseolina Sdh, cyp�� and os� genes is lacking. In this study we report the predicted cyp, sdhB and os� sequences for �� M. phaseolina isolates. Three paralogs of cyp�� gene (cyp��A, cyp��B and cyp��C) have been identi�ed in fungi and although their involvement in DMI sensitivity is well known for several fungal phytopathogens (Schnabel and Jones, ����; Mohd-Assaad et �� Figure �.� Predicted EC��(P) distribution of �� M. phaseolina isolates tested in mycelial growth assays in half-strength PDA plates amended with � mg ml � of boscalid, � mg ml � of iprodione, or �.� mg ml � of prothioconazole. al., ����; Zhang et al., ����; Lestrade et al., ����; Wei et al., ����), the occurrence of cyp��A and cyp��C paralogs is unknown in M. phaseolina. The cyp��A and cyp��B paralogs sequences were present in the �� M. phaseolina isolates tested in this study. cyp��A it is thought to play a major role in reduced sensitivity to DMIs, mainly as a functionally redundant mechanism for ergosterol production when fungi are exposed with DMI fungicides (Fan et al., ����; Liu et al., ����). A total of �� isolates showed mutations occurring in the cyp�� genes and four isolates in the SdhB gene. The cyp��A D���Y mutation was the most frequently identi�ed mutation present in twelve isolates of our collection. Interestingly, these isolates were all collected from dry bean in Colombia or Puerto Rico. None of these point mutations were found to be correlated with lower levels of sensitivity. Reduced sensitivity with high resistance factors (strength of resistance) is often observed with mutations located �� Figure �.� Combined EC�� and EC��(P) distribution of �� M. phaseolina isolates. This combined data set consists of multiple concentration, single concentration and validation data points, including two additional isolates from the validation set. Less sensitive isolates were designated as the ones with an EC�� or EC��(P) value higher than three standard deviations from the mean indicated by the dashed line. in putative azole molecules recognition sites (e.g V���A, Y���F, Y���F, Y���F, A���G, I���V) whereas mutations in highly conserved regions of the cyp�� protein close to the heme binding site such as those at codons ���-��� have been correlated with lower resistance factors (Cools et al., ����; Mullins et al., ����; Mehl et al., ����). The cyp��B H���Y mutation was identi�ed in two isolates collected from soybean in the US. In Cryptococcus gattii, the cyp�� N���D mutation conferred azole resistance (Gast et al., ����). Molecular modelling of speci�c mutations in residues proximal to the binding pocket showed to have di�erential impact on cyp�� protein function depending on whether a single mutation was present or in combination with others. The protein function was impacted mainly by alterations in the binding pocket volume. Furthermore, the e�ect of these mutations on DMI sensitivity was di�erent for certain azole molecules (Cools et al., ����; Mullins et al., ����). �� Figure �.� Mean EC�� /EC��(P) di�erences of �� M. phaseolina isolates across fungicides: (A) Boscalid- Iprodione (B) Boscalid-Prothioconazole (C) Iprodione-Prothioconazole pairwise comparisons. Isolates were most sensitive to prothioconazole, followed by boscalid and least sensitive to iprodione. It has been hypothesized that seed treatments may be useful in protecting soybean plants from disease caused by seedborne M. phaseolina (Hartman et al., ����a; Hartman et al., ����b). However, informa- tion regarding the e�cacy of active ingredients currently used in commercial fungicide formulations in soybean and dry bean is lacking (Romero Luna et al., ����). Although in-vivo and �eld studies would be necessary, our results indicate that formulations with prothioconazole, boscalid or iprodione, may reduce seedling infection originating from infected seeds or inoculum in the soil. In cotton, seed treatments with a commercial formulation of boscalid + pyraclostrobin (Signum) showed e�cacy in preventing seedling infection by M. phaseolina in �eld experiments (Cohen et al., ����). However, this protective e�ect was observed only for �� days while roots were exposed to the fungicide in soil (Cohen et al., ����). Our data on the in-vitro e�cacy of prothioconazole suggest that commercial formulations with this active ingredient may be of particular interest for future in-vivo e�cacy testing in soybean and dry bean. Currently, fungicides labeled to control charcoal rot in di�erent crops are available, although limited (http://www.cdms.net). For instance, a formulation of prothioconazole + �uopyram (Propulse) is la- beled for charcoal rot management in soybean. Prothioconazole (Proline) has been shown to suppress plant colonization by M. phaseolina and improve yield under �eld conditions when used in tolerant soy- bean varieties in inoculated plots as compared to non-inoculated plots (USB report, ����). Future stud- ies can be aimed at testing prothioconazole e�cacy in preventing seedling colonization and charcoal rot disease development as part of integrated management programs incorporating host genetic resistance �� and cultural practices. Furthermore, studies investigating novel e�ective fungicides and monitoring the potential development of resistance to single-site fungicides in M. phaseolina populations would be ben- e�cial to charcoal rot management e�orts. �.� Materials and methods �.�.� Macrophomina phaseolina isolates and whole-genome sequencing A total of �� M. phaseolina isolates collected mostly from soybean and dry bean, for which population genomics analysis was conducted, were also used for fungicide sensitivity analysis (Ortiz et al., under re- view). Species identity of these isolates was con�rmed as described previously by sequencing the Internal Transcribed Spacer regions for the nuclear rDNA operon (ITS), part of the Translation Elongation Fac- tor (TEF-�α) gene region, and part of the actin (ACT) gene region (Sarr et al., ����) (Ortiz et al., under review). Brie�y, this isolate collection included �� isolates collected from soybean across the US and �� isolates collected from dry bean in Michigan, Puerto Rico and Colombia, two isolates from strawberry collected in California and one isolate collected from Ethiopia mustard in the US. Whole genome sequencing and SNP calling was conducted as described in Ortiz et al., under review. Genomic DNA was extracted from lyophilized mycelia using a modi�ed SDS-based method; as described previously (Ortiz et al., under review). Brie�y, hyphal tip cultures grown on potato dextrose agar (PDA) medium were used to produce mycelia on potato dextrose broth. Libraries were prepared using the Illu- mina TruSeq Nano DNA Library Preparation Kit HT and whole-genome sequencing to ��X coverage using a ��� base-pair paired-end strategy on the Illumina HiSeq ���� platform at the Michigan State University Research Technology Support Facility Genomics Core (East Lansing, MI) was conducted. Quality assessment and �ltering were conducted using FastQC (Andrews et al., ����) and Cutadapt v �.�� (Martin, ����). Sequences were aligned to the M. phaseolina reference genome (JGI Mycocosm, MPI-SDFR-AT-���� v�.�) using bwa-mem (Heng Li, ����). Single nucleotide polymorphisms (SNPs) of all isolates were predicted using the Genome Analysis Toolkit (GATK) v�.� (McKenna et al., ����) pipeline (Ortiz et al., under review). The resulting vcf �le was quality �ltered using vcfR v�.��.� package (Knaus and Grünwald, ����) in R v�.�.� (R Core Team ����) (Ortiz et al., under review). �� �.�.� Fungicides Commercial fungicide formulations of the SDHI boscalid (��% A.I., Endura, BASF corporation, Re- search Triangle Park, NC), Dicarboximide iprodione (Chipco ��GT �SC, Bayer, Germany) and DMI prothioconazole (Proline ��� SC, Bayer CropScience, Research Triangle Park, NC) were used. Addi- tional information about these fungicides is presented in Table. Aqueous stock solutions of these fungi- cides were prepared at ���� �g ml � of each respective active ingredient. Serial dilutions from the stock solutions were used to produce �nal concentrations in half-strength PDA media of boscalid (�.�, �, ��, ��� and ��� �g ml � ), iprodione (�.�, �, �.�, � and �� �g ml � ) and prothioconazole (�.��, �.�, �.�, � and �� �g ml � ), except for boscalid highest concentration (��� �g ml � ) for which �.��� g of Endura, per liter of media was used. These concentrations were selected based on preliminary experiments which directed the appropriate fungicide concentrations for �tting a dose-response curve. �.�.� Determination of EC�� values using mycelial growth inhibition assays The sensitivity of �� randomly selected M. phaseolina isolates to boscalid, iprodione and prothioconazole was determined based on EC�� (e�ective concentration to reduce growth by ��%) estimates. EC�� values for each isolate was determined using mycelial growth inhibition assays on fungicide-amended medium. Before each experiment isolates were recovered from -��ºC and grown on potato dextrose agar (PDA; Acumedia, Lansing, MI) in the dark at ��ºC for �� h. Then a mycelial plug from the margin was trans- ferred into a new Petri plate containing PDA and incubated in the dark at ��ºC for �� h. A single �-mm agar plug taken from the edge of the ��-h old culture was placed mycelial side down on the center of non- amended half-strength PDA plates and plates amended with boscalid, iprodione or prothioconazole at concentrations mentioned above. The plates were incubated in the dark at ��ºC for �� h. The diameter of each colony was measured in two perpendicular directions with a digital caliper (Ab- solute Digimatic Caliper, model CD-�” AX, Mitutoyo Corp., Sakado �-Chome, Japan). Two separate experiments and two replicates (Petri plates) per each experiment were performed for each isolate and fungicide concentration. Isolates with data from at least two replicates were included in all subsequent analyses. Percent relative mycelial growth (RMG) at each concentration was calculated as the percentage of inhibition relative to the control without fungicides ((average colony diameter on fungicide amended ��� plates / average colony diameter on non-amended plates) X ���). Absolute EC�� values were calculated using a four parameter log logistic (LL�) dose response model as implemented in R (R Core Team, ����) in the ‘drc’ package (Ritz et al., ����), and following guidelines and work�ow provided by (Noel et al., ����). The LL� model was used as it was the best �tting model for most isolates as determined by AIC criteria. Less sensitive isolates were designated based on the fre- quency distribution of the EC�� values as the ones with an EC�� higher than three standard deviations from the mean (EC�� values > �.��, �.��, and �.���g ml � for boscalid, iprodione and prothioconazole, respectively). To investigate variability across experiments, isolate mean EC�� di�erences between ex- periments were estimated using DABEST (‘data analysis with bootstrap-coupled estimation’) (Ho et al., ����). Isolate mean EC�� di�erences between experiments were all less than �.�� �g ml � for all fungi- cides, therefore experiments were combined in subsequent analyses. �.�.� Selection of single screening fungicide concentration and linear regression models To screen the remaining isolates in a reduced resource-intensive manner, single screening concentrations were determined for boscalid, iprodione and prothioconazole using the EC�� results of �� M. phaseolina isolates. A linear regression analysis between RMG and log-transformed EC�� values of each isolate was performed for the �ve tested concentrations for each fungicide. The fungicide concentration at which the linear regression model returned the highest correlation coe�cient (Pearson’s R) and proportion of explained variance (R� ) values was selected as the screening concentration for each fungicide. These screening concentrations were found to be � �g ml � for boscalid and iprodione, and �.� �g ml � for pro- thioconazole. While for prothioconazole, �.� �g ml � and �.� �g ml � concentrations both had similarly high R and R� values (Pearson’s R=�.�, R� =�.� ), at �.� �g ml � most isolates had an RMG below ��% indicating it may di�erentiate better less sensitive isolates than �.� �g ml � . �.�.� Sensitivities and EC��(P) prediction using single screening fungicide concentrations Sensitivities of each of the remaining �� isolates were estimated based on RMG on half-strength PDA plates amended with boscalid at � �g ml � , iprodione at � �g ml � , or prothioconazole at �.� �g ml � . Media preparation, inoculation and mycelial growth measurements were conducted using the methods for EC�� estimation described above. Two separate experiments and two replicates (Petri plates) per experiment ��� were performed for each isolate and fungicide. The linear regression model equations of boscalid (� �g ml � ), iprodione (� �g ml � ), or prothioconazole (�.� �g ml � ) (Figure A.�) were then used to predict an EC�� value, hereafter EC��(P) , for each isolate, using the function ‘predict’ in R (R Core Team, ����). �.�.� Validation of linear regression models used to predict EC��(P) values A validation set of �� isolates (Table A.�) was used to assess the performance of linear regression models in predicting EC��(P) values using a new data set. These validation isolates were selected from those for which an EC�� or EC��(P) was previously estimated in the multiple and single concentration experiments and to represent the range of these values. Additionally, two isolates not tested in any of the previous experiments were included. For these validation isolates, EC�� values were determined using the �ve fungicide concentrations in mycelial growth inhibition assays as described above. Then, the RMG at the screening concentration for each isolate (boscalid [� �g ml � ], iprodione [� �g ml � ], or prothioconazole [�.� �g ml � ]) was used to predict an EC��(P) value using the linear regression models previously selected. A simple linear regression analysis was used to determine the relationship between estimated EC�� and predicted EC��(P) values for the �� validation isolates. �.�.� Sensitivities of �� M. phaseolina isolates using the combined EC�� /EC��(P) values All previous data sets, this is the EC�� values for the initial �� isolates, the EC��(P) values for the �� isolates in the single concentration experiments and the EC�� values for the validation isolates were combined to report the fungicide sensitivity of the �� M. phaseolina isolates. The combined distribution of EC�� /EC��(P) values was used to categorize “less sensitive” isolates. Isolates were designated as less sensitive as the ones with an EC�� or and EC��(P) value higher than three standard deviations from the mean (EC�� or EC��(P) values > �.��, �.��, and �.�� �g ml � for boscalid, iprodione, and prothioconazole, respectively). �.�.� Sequence and target gene mutation analysis To explore if the sensitivity of M. phaseolina to boscalid (SDHI), iprodione (Dicarboximide) and pro- thioconazole (DMI) was associated with mutations in their target genes, the complete SdhB, cyp��, and os� genes of the �� M. phaseolina isolates were analyzed. Genome-guided de novo assembly of each iso- late was done in Trinity (Grabherr et al., ����) using whole-genome sequencing Illumina reads with the M. phaseolina (JGI Mycocosm, MPI-SDFR-AT-���� v�.�) as reference genome. Although Trinity was ��� developed for RNA-seq data, it was used to take advantage of the genome-guided option for de-novo transcript assembly, because it employs a de Bruijn graph approach (used by several whole-genome as- sembly programs) and can identify transcripts resulting from paralogous genes (Grabherr et al., ����). Published sequences of the target genes SdhB, cyp��, and os� (Shk�) (Duan et al., ����) of Fusar- ium graminearum, Botrytis cinerea, Diplodia corticola and Sclerotinia sclerotiorum were used as query sequences to identify the orthologous sequences in the M. phaseolina reference genome using the pro- gram BLAST. The reference sequence of each target gene was used to retrieve the sequences from the assemblies of the �� isolates, using a home-made script (https://github.com/vivianaortizl/). The amino acid sequences were aligned and analyzed using Geneious software. �.�.� Statistical analysis All data analysis was conducted in R (R Core Team, ����). Linear regression modeling and model evalu- ation were done with ‘lm’ and ‘AICctab’ functions (‘bbmle’ package). Isolates with data from at least two replicates were included for analysis. E�ect sizes and con�dence intervals were estimated using DABEST (Ho et al., ����). Bootstrap con�dence intervals were estimated adjusting for asymmetrical resampling distributions using bias-corrected and accelerated bootstrap (BCa bootstrap)(Efron and Tibshirani, ����) as implemented in ‘dabestr’ R package (Ho et al., ����). ��� BIBLIOGRAPHY Abawi GS, Pastor-Corrales M (����) Seed transmission and e�ect of fungicide seed treatments against Macrophomina phaseolina in dry edible beans. Turrialba ��: ���–��� Ambachew D, Joshua J, Mmbaga MT, Blair MW (����) Sources of resistance to common bacte- rial blight and charcoal rot disease for the production of mesoamerican common beans in the Southern United States. Plants ��: �–�� Bandara AY, Weerasooriya DK, Conley SP, Bradley CA, Allen TW, Esker PD (����) Modeling the relationship between estimated fungicide use and disease-associated yield losses of soybean in the United States I: Foliar fungicides vs foliar diseases. PLoS One ��: �–�� Bradley CA (����) E�ect of fungicide seed treatments on stand establishment, seedling disease, and yield of soybean in North Dakota. Plant Dis ��: ���–��� Bradley CA, Allen TW, Sisson AJ, Bergstrom GC, Bissonnette KM, Bond J, Byamukama E, Chilvers MI, Collins AA, Damicone JP, et al (����) Soybean Yield Loss Estimates Due to Diseases in the United States and Ontario, Canada, from ���� to ����. Plant Heal Prog ��: ���–��� Brent KJ, Hollomon DW (����) Fungicide resistance in crop pathogens: how can it be managed? �nd, revised edition. Chamberlain LA, Whitman T, Ané JM, Diallo T, Gaska JM, Lauer JG, Mourtzinis S, Conley SP (����) Corn-soybean rotation, tillage, and foliar fungicides: Impacts on yield and soil fungi. F Crop Res. doi: ��.����/j.fcr.����.������ Chaudhary DH, Pathak DM, Chaudhary MM (����) In vitro E�cacy of Fungicides against Dry Root Rot (Macrophomina phaseolina) of Soybean. Int J Curr Microbiol App Sci �: ����–���� Cohen R, Elkabetz M, Paris HS, Gur A, Dai N, Rabinovitz O, Freeman S (����) Occurrence of Macrophomina phaseolina in Israel: Challenges for Disease Management and Crop Germplasm En- hancement. Plant Dis ���: ��–�� Cools HJ, Mullins JGL, Fraaije BA, Parker JE, Kelly DE, Lucas JA, Kelly SL (����) Impact of recently emerged sterol ��α-demethylase (CYP��) variants of Mycosphaerella graminicola on azole fungicide sen- sitivity. Appl Environ Microbiol ��: ����–���� Cools HJ, Parker JE, Kelly DE, Lucas JA, Fraaije BA, Kelly SL (����) Heterologous expression of mutated eburicol ��α-demethylase (CYP��) proteins of Mycosphaerella graminicola to assess e�ects on azole fungicide sensitivity and intrinsic protein function. Appl Environ Microbiol ��: ����–���� Coser SM, Reddy RVC, Zhang J, Mueller DS (����) Genetic Architecture of Charcoal Rot (Macrophomina phaseolina) Resistance in Soybean Revealed Using a Diverse Panel. �: �–�� Crous PW, Slippers B, Groenewald JZ, Wing�eld MJ (����) Botryosphaeriaceae: Systematics, pathol- ogy, and genetics. Fungal Biol. doi: ��.����/j.funbio.����.��.��� Dhingra OD, Sinclair JB (����) Survival of Macrophomina phaseolina Sclerotia in Soil: E�ects of Soil ��� Moisture, Carbon:Nitrogen Ratios, Carbon Sources, and Nitrogen Concentrations. Phytopathology ��: ���–��� Dhingra OD, Sinclair JB (����) Biology and pathology of Macrophomina phaseolina. Universidade Federal de Vicosa., Minas Gerais Duan Y, Ge C, Liu S, Wang J, Zhou M (����) A two-component histidine kinase Shk� controls stress response, sclerotial formation and fungicide resistance in Sclerotinia sclerotiorum. Mol Plant Pathol ��: ���–��� Efron B, Tibshirani RJ (����) An Introduction to the Bootstrap. An Introd to Bootstrap. doi: ��.����/���-�-����-����-� Fan J, Urban M, Parker JE, Brewer HC, Kelly SL, Hammond-Kosack KE, Fraaije BA, Liu X, Cools HJ (����) Characterization of the sterol ��α-demethylases of Fusarium graminearum identi�es a novel genus-speci�c CYP�� function. New Phytol ���: ���–��� Gast CE, Basso LR, Bruzual I, Wong B (����) Azole resistance in Cryptococcus gattii from the Paci�c Northwest: Investigation of the role of ERG��. Antimicrob Agents Chemother ��: ����–���� Grabherr MG, Haas BJ, Yassour M, Levin JZ, Thompson DA, Amit I, Adiconis X, Fan L, Raychowd- hury R, Zeng Q, et al (����) Full-length transcriptome assembly from RNA-Seq data without a reference genome. Nat Biotechnol ��: ���–��� Hartman GL, Rupe JC, Sikora EJ, Domier LL, Davis JA, Ste�ey KL (����a) Part I . Infectious Dis- eases. Compend. Soybean Dis. Pests, Fifth Ed. pp ��–��� Hartman GL, Rupe JC, Sikora EJ, Domier LL, Davis JA, Ste�ey KL (����b) PART IV: Soybean Disease and Pest Management Strategies. Compend Soybean Dis Pests, Fifth Ed ���: ���–��� Hawkins NJ, Fraaije BA (����) Fitness Penalties in the Evolution of Fungicide Resistance. Annu Rev Phytopathol ��: ���–��� He QY, Mason AB, Pakdaman R, Chasteen ND, Dixon BK, Tam BM, Nguyen V, MacGillivray RTA, Woodworth RC (����) Mutations at the histidine ��� ligand profoundly alter the spectral and iron- binding properties of human serum transferrin N-lobe. Biochemistry ��: ����–���� Heng Li (����) Aligning sequence reads, clone sequences and assembly contigs with BWA-MEM. Ho J, Tumkaya T, Aryal S, Choi H, Claridge-Chang A (����) Moving beyond P values: data analysis with estimation graphics. Nat Methods ��: ���–��� Jacobs JL, Kelly JD, Wright EM, Varner G, Chilvers MI (����) Determining the Soilborne Pathogens Associated with Root Rot Disease Complex of Dry Bean in Michigan. Plant Heal Prog ��: ���–��� Karavidas I, Ntatsi G, Vougeleka V, Karkanis A, Ntanasi T, Saitanis C, Agathokleous E, Ropokis A, Sabatino L, Tran F, et al (����) Agronomic Practices to Increase the Yield and Quality of Common Bean (Phaseolus vulgaris L.): A Systematic Review. Agronomy. doi: ��.����/agronomy�������� ��� Knaus BJ, Grünwald NJ (����) vcfr: a package to manipulate and visualize variant call format data in R. Mol Ecol Resour ��: ��–�� Kuck K, Russell PE (����) FRAC: Combined resistance risk assessment. Asp Appl Biol ��: � Leadbeater A, Mcgrath MT, Wyenandt CA, Stevenson KL (����) CHAPTER �: An Overview of Fungicide Resistance and Resistance Management: History and Future Trends. Fungic. Resist. North Am. Second Ed. pp �–�� Lehner MS, Del Ponte EM, Gugino BK, Kikkert JR, Pethybridge SJ (����) Sensitivity and e�cacy of boscalid, �uazinam, and thiophanate-methyl for white mold control in snap bean in New York. Plant Dis ���: ����–���� Lestrade PPA, Meis JF, Melchers WJG, Verweij PE (����) Triazole resistance in Aspergillus fumigatus: recent insights and challenges for patient management. Clin Microbiol Infect ��: ���–��� Liu J, Jiang J, Guo X, Qian L, Xu J, Che Z, Chen G, Liu S (����) Sensitivity and resistance risk assessment of Fusarium graminearum from wheat to prothioconazole. Plant Dis ����–���� Lucas JA, Hawkins NJ, Fraaije BA (����) The Evolution of Fungicide Resistance. Adv Appl Micro- biol ��: ��–�� McDonald BA, Stukenbrock EH (����) Rapid emergence of pathogens in agro-ecosystems: global threats to agricultural sustainability and food security. Philos Trans R Soc B Biol Sci ���: �������� McKenna A, Hanna M, Banks E, Sivachenko A, Cibulskis K, Kernytsky A, Garimella K, Altshuler D, Gabriel S, Daly M, et al (����) The Genome Analysis Toolkit: a MapReduce framework for analyzing next-generation DNA sequencing data. Genome Res ��: ����–��� Mehl A, Schmitz H, Stenzel K, Bloomberg J, Lp BC, Carolina N (����) Chapter �. Meyer W., Sinclair JB (����) Factors A�ecting Charcoal Rot of Soybean Seedlings. Phytopathology ��: ���–��� Mohd-Assaad N, McDonald BA, Croll D (����) Multilocus resistance evolution to azole fungicides in fungal plant pathogen populations. Mol Ecol ��: ����–���� Morton V, Staub T (����) A Short History of Fungicides. APSnet Featur Artic �–�� Motoyama T, Kadokura K, Ohira T, Ichiishi A, Fujimura M, Yamaguchi I, Kudo T (����) A two- component histidine kinase of the rice blast fungus is involved in osmotic stress response and fungicide action. Fungal Genet Biol ��: ���–��� Mullins JGL, Parker JE, Cools HJ, Togawa RC, Lucas JA, Fraaije BA, Kelly DE, Kelly SL (����) Molecular modelling of the emergence of azole resistance in Mycosphaerella graminicola. PLoS One. doi: ��.����/journal.pone.������� Nikou D, Malandrakis A, Konstantakaki M, Vontas J, Markoglou A, Ziogas B (����) Molecular char- acterization and detection of overexpressed C-�� alpha-demethylase-based DMI resistance in Cercospora beticola �eld isolates. Pestic Biochem Physiol ��: ��–�� ��� Noel ZA, Wang J, Chilvers MI (����) Signi�cant In�uence of EC �� Estimation by Model Choice and EC �� Type. Plant Dis ���: ���–��� Oerke EC (����) Crop losses to pests. J Agric Sci ���: ��–�� Parsa S, García-Lemos AM, Castillo K, Ortiz V, Becerra López-Lavalle LA, Braun J, Vega FE (����) Fungal endophytes in germinated seeds of the common bean, Phaseolus vulgaris. Fungal Biol. Pastor-Corrales MA, Abawi GS, others (����) Reactions of selected bean accessions to infection by Macrophomina phaseolina. Plant Dis ��: ��–�� R Core Team (����) R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/. Reznikov S, Vellicce GR, González V, de Lisi V, Castagnaro AP, Ploper LD (����) Evaluation of chem- ical and biological seed treatments to control charcoal rot of soybean. J Gen Plant Pathol ��: ���–��� Reznikov S, Vellicce GR, Mengistu A, Arias RS, Gonzalez V, Lisi V De, Garcia MG, Rocha Carla ML, Pardo EM, Castagnaro AP, Ploper LD (����) Disease incidence of charcoal rot (Macrophomina phaseolina) on soybean in north-western Argentina and genetic characteristics of the pathogen. Can J Plant Pathol ��: ���–��� Ritz C, Baty F, Streibig JC, Gerhard D (����) Dose-Response Analysis Using R. PLoS One ��: e������� Romero Luna MP, Mueller D, Mengistu A, Singh AK, Hartman GL, Wise KA (����) Advancing Our Understanding of Charcoal Rot in Soybeans. J Integr Pest Manag �: ���–��� Roth MG, Webster RW, Mueller DS, Chilvers MI, Faske TR, Mathew FM, Bradley CA, Damicone JP, Kabbage M, Smith DL (����) Integrated Management of Important Soybean Pathogens of the United States in Changing Climate. J Integr Pest Manag. doi: ��.����/jipm/pmaa��� Russell PE (����) A century of fungicide evolution. J Agric Sci ���: ��–�� Sang H, Lee HB (����) Molecular mechanisms of succinate dehydrogenase inhibitor resistance in phytopathogenic fungi. Res Plant Dis ��: �–� Savary S, Willocquet L, Pethybridge SJ, Esker P, McRoberts N, Nelson A (����) The global burden of pathogens and pests on major food crops. Nat Ecol Evol �: ���–��� Schnabel G, Jones AL (����) The α-Demethylase (CYP��A�) gene is overexpressed in Venturia inae- qualis strains resistant to myclobutanil. Phytopathology ��: ���–��� Slippers B, Boissin E (����) Phylogenetic lineages in the Botryosphaeriales: a systematic and evolu- tionary framework. Stud Mycol ��: ��–�� Slippers B, Wing�eld MJ (����) Botryosphaeriaceae as endophytes and latent pathogens of woody plants: diversity, ecology and impact. Fungal Biol Rev ��: ��–��� ��� Tonin RFB, Avozani A, Danelli ALD, Reis EM, Zoldan SM, Garcés-Fiallos FR (����) In vitro mycelial sensitivity of Macrophomina phaseolina to fungicides. Pesqui Agropecuária Trop ��: ���–��� Velasquez JG, Prada P, Henry G. (����) Snap bean pests and diseases in Sumapaz, Colombia: their present status and implications. Int. Conf. Sn. Beans Dev. World. Cent. Int. Agric. Trop. Wei LL, Chen WC, Zhao WC, Wang J, Wang BR, Li FJ, Wei M Di, Guo J, Chen CJ, Zheng JQ, et al (����) Mutations and overexpression of CYP�� associated with DMI-resistance in Colletotrichum gloeosporioides from Chili. Plant Dis ���: ���–��� Wrather A, Shannon G, Balardin R, Carregal L, Escobar R, Gupta GK, Ma Z, Morel W, Ploper D, Tenuta A (����) E�ect of Diseases on Soybean Yield in the Top Eight Producing Countries in ����. Plant Heal Prog. doi: ��.����/php-����-����-��-rs Yamaguchi I, Fujimura M (����) Recent topics on action mechanisms of fungicides. J Pestic Sci ��: ��–�� Zak O, Aisen P, Crawley JB, Joannou CL, Patel KJ, Ra�q M, Evans RW (����) Iron Release from Recombinant N-lobe and Mutants of Human Transferrin. Biochemistry ��: �����–����� Zhang C, Diao Y, Wang W, Hao J, Imran M, Duan H, Liu X (����) Assessing the risk for resistance and Elucidating the Genetics of Colletotrichum truncatum that is only sensitive to some DMI fungicides. Front Microbiol �: ���� ��� APPENDIX SUPPLEMENTARY TABLES AND FIGURES ��� Table A.� Mean EC�� di�erences and ��% con�dence intervals for �� M. phaseolina isolates across exper- iments for boscalid, iprodione and prothicoconazole. Two separate experiments and two replicates per experiment were performed for each isolate and fungicide. Experiment � Experiment � Mean EC�� di�erence ��% CIa low high Bos� Bos� -�.��� -�.��� �.��� Ipro� Ipro� �.��� -�.��� �.��� Pro� Pro� -�.��� -�.��� -�.��� a ��% CI adjusting for asymmetrical resampling distributions using bias-corrected and accelerated bootstrap (BCa bootstrap). Experiments are Bos� and Bos� for boscalid, Ipro� and Ipro� for iprodione and Pro� and Pro� for prothioconazole. ��� Table A.� Mean EC�� di�erences for �� M. phaseolina isolates across fungicides: boscalid, iprodione and prothicoconazole. Pairwise fungicide comparison Mean EC�� di�erence ��% CIa low high Iprodione minus Boscalid �.��� �.��� �.��� Prothioconazole minus Boscalid -�.��� -�.��� -�.��� Prothioconazole minus Iprodione -�.��� -�.��� -�.��� a ��% con�dence intervals adjusting for asymmetrical resampling distributions using bias-corrected and accelerated bootstrap (BCa bootstrap). ��� Table A.� Predicted EC��(P) (e�ective concentration to reduce growth by ��%) for �� M. phaseolina iso- lates determined from mycelial growth assays in Petri plates amended with � mg ml � of boscalid, � mg ml � of iprodione, or �.� mg ml � of prothioconazole. Isolate Boscalid Iprodione Prothioconazole EC��(P) ��% CIa EC��(P) ��% CIa EC��(P) ��% CIa low high low high low high Et�� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� Et� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� IN��_� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.���� M_��_� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� M_��_�� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.���� M_��_�� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� M_��_�� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.���� M_��_�� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.���� M_��_� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� M_��_� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� M_��_� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.���� M_��_� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.���� M_��_� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.���� M_��_� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� M��-�� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.� M��-�� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� M��_� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.���� Md�� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� Md� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.���� Md� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.���� MISO���-� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.���� MISO���-� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.���� MISO���-� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.���� MISO���-� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.���� MP��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.���� MP��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.���� MP��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.���� Mph_�� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.���� Mph_�� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.���� Mph_�� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.���� Mph_�� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.���� Mph_�� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.���� Mph_�� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.���� Mph_�� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.���� Mph_�� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.���� Mph_�� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.���� ��� Table A.� (cont’d) Mph_�� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.���� Mph_�� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.���� Mph_�� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.���� Mph_�� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� Mph_� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� Mph_�� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.���� Mph_�� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.� Mph_�� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.���� Mph_�� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.���� MpSDSU �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.���� TN��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.���� TN��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.���� TN��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.���� TN��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.���� UPR-ISA� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� UPR-ISA� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.���� UPR-JD� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.���� UPR-JD� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.���� UPR-JD� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.���� EC��(P) and ��% con�dence intervals were back-transformed. a ��% con�dence intervals adjusting for asymmetrical resampling distributions using bias-corrected and accelerated bootstrap (BCa bootstrap). ��� Table A.� Mean EC�� (e�ective concentration to reduce growth by ��%) estimates and EC��(P) predic- tions for a validation set of �� M. phaseolina isolates determined from mycelial growth assays in Petri plates amended with boscalid, iprodione or prothioconazole. �*Isolate Boscalid Iprodione Prothioconazole �*EC�� ± SEa EC��(P) (��% CI)b �*EC�� ± SEa �*EC��(P) b EC��(P) ��% CIb �*EC�� ± SEa �*EC��(P) b EC��(P) ��% CIb low high low high low high Et�� �.��� ± �.��� �.��� �.��� �.��� �.��� ± �.��� �.��� �.��� �.��� �.��� ± �.��� �.��� �.��� �.��� M_��_� �.��� ± �.��� �.��� �.��� �.��� �.��� ± �.��� �.��� �.��� �.��� �.��� ± �.��� �.��� �.��� �.��� MISO���-� �.��� ± �.��� �.��� �.��� �.��� �.��� ± �.��� �.��� �.��� �.��� �.��� ± �.��� �.��� �.��� �.��� MISO���-� �.��� ± �.��� �.��� �.��� �.��� �.��� ± �.��� �.��� �.��� �.��� �.��� ± �.��� �.��� �.��� �.��� MISO���-� �.��� ± �.��� �.��� �.��� �.��� �.��� ± �.��� �.��� �.��� �.��� �.��� ± �.��� �.��� �.��� �.��� MP��� �.��� ± �.��� �.��� �.��� �.��� �.��� ± �.��� �.��� �.��� �.��� �.��� ± �.��� �.��� �.��� �.��� MP��� _ �.��� �.��� �.��� �.��� ± �.��� �.��� �.��� �.��� �.��� ± �.��� �.��� �.��� �.��� Mph_�� �.��� ± �.��� �.��� �.��� �.��� �.��� ± �.��� �.��� �.��� �.��� �.��� ± �.��� �.��� �.��� �.��� Mph_�� �.��� ± �.��� �.��� �.��� �.��� �.��� ± �.��� �.��� �.��� �.��� �.��� ± �.��� �.��� �.��� �.��� Mph_�� �.��� ± �.��� �.��� �.��� �.��� �.��� ± �.��� �.��� �.��� �.��� �.��� ± �.��� �.��� �.��� �.��� W��-� �.��� ± �.��� �.��� �.��� �.��� �.��� ± �.��� �.��� �.��� �.��� �.��� ± �.��� �.��� �.��� �.��� W�� �.��� ± �.��� �.��� �.��� �.��� �.��� ± �.��� �.��� �.��� �.��� �.��� ± �.��� �.��� �.��� �.��� W�-� �.��� ± �.��� �.��� �.��� �.��� �.��� ± �.��� �.��� �.��� �.��� �.��� ± �.��� �.��� �.��� �.��� ��� Table A.� Mean EC�� /EC��(P) (e�ective concentration to reduce growth by ��%) values for �� M. phase- olina isolates determined from mycelial growth assays in Petri plates amended with boscalid, iprodione, or prothioconazole. Isolate Boscalid Iprodione Prothioconazole Mean Mean Mean EC�� /EC��(P) EC�� /EC��(P) EC�� /EC��(P) CR_Red_� �.��� �.��� �.��� CR_Red_�B �.��� �.��� �.��� CR_Red_� �.��� �.��� �.��� Dm�� �.��� �.��� �.��� Et�� �.��� �.��� �.��� Et�� �.��� �.��� �.��� Et�� �.��� �.��� �.��� Et�� �.��� �.��� �.��� Et� �.��� �.��� �.��� IN��_� �.��� �.��� �.��� IN��_�_� �.��� �.��� �.��� IN��_�_� �.��� �.��� �.��� IN��_�_� �.��� �.��� �.��� IN��_PO_� �.��� �.��� �.��� M_��_� �.��� �.��� �.��� M_��_�� �.��� �.��� �.��� M_��_�� �.��� �.��� �.��� M_��_�� �.��� �.��� �.��� M_��_�� �.��� �.��� �.��� M_��_� �.��� �.��� �.��� M_��_� �.��� �.��� �.��� M_��_� �.��� �.��� �.��� M_��_� �.��� �.��� �.��� M_��_� �.��� �.��� �.��� M_��_� �.��� �.��� �.��� M��-�� �.��� �.��� �.��� M��-�� �.��� �.��� �.��� M��_� �.��� �.��� �.��� Md�� �.��� �.��� �.��� Md� �.��� �.��� �.��� Md� �.��� �.��� �.��� Md� �.��� �.��� �.��� Md� �.��� �.��� �.��� MI-SF �-�� �.��� �.��� �.��� MI-SF ��-�� �.��� �.��� �.��� MI-SF �-�� �.��� �.��� �.��� ��� Table A.� (cont’d) MI-SF �-� �.��� �.��� �.��� MISO���-� �.��� �.��� �.��� MISO���-� �.��� �.��� �.��� MISO���-� �.��� �.��� �.��� MISO���-� �.��� �.��� �.��� MISO���-� �.��� �.��� �.��� MISO���-� �.��� �.��� �.��� MISO���-� �.��� �.��� �.��� MP��� �.��� �.��� �.��� MP��� �.��� �.��� �.��� MP��� �.��� �.��� �.��� MP��� �.��� �.��� �.��� MP��� �.��� �.��� �.��� MP��� �.��� �.��� �.��� Mph_�� �.��� �.��� �.��� Mph_�� �.��� �.��� �.��� Mph_�� �.��� �.��� �.��� Mph_�� �.��� �.��� �.��� Mph_�� �.��� �.��� �.��� Mph_�� �.��� �.��� �.��� Mph_�� �.��� �.��� �.��� Mph_�� �.��� �.��� �.��� Mph_�� �.��� �.��� �.��� Mph_�� �.��� �.��� �.��� Mph_�� �.��� �.��� �.��� Mph_�� �.��� �.��� �.��� Mph_�� �.��� �.��� �.��� Mph_� �.��� �.��� �.��� Mph_�� �.��� �.��� �.��� Mph_�� �.��� �.��� �.��� Mph_�� �.��� �.��� �.��� Mph_�� �.��� �.��� �.��� Mph_�� �.��� �.��� �.��� MpSDSU �.��� �.��� �.��� SAG�-� �.��� �.��� �.��� TN��� �.��� �.��� �.��� TN��� �.��� �.��� �.��� TN��� �.��� �.��� �.��� TN��� �.��� �.��� �.��� TN��� �.��� �.��� �.��� TN��� �.��� �.��� �.��� TN� �.��� �.��� �.��� TN� �.��� �.��� �.��� ��� Table A.� (cont’d) TN��� �.��� �.��� �.��� TN��� �.��� �.��� �.��� UPR-Mph-ISA� �.��� �.��� �.��� UPR-Mph-ISA� �.��� �.��� �.��� UPR-Mph-JD� �.��� �.��� �.��� UPR-Mph-JD� �.��� �.��� �.��� UPR-Mph-JD� �.��� �.��� �.��� W-MISO� �-� �.��� �.��� �.��� W-MISO� �-�� �.��� _ �.��� W��-� �.��� �.��� �.��� W�� �.��� �.��� �.��� W�� �.��� �.��� �.��� W�-� �.��� �.��� �.��� ��� Table A.� Mean EC�� /EC��(P) di�erences for �� M. phaseolina isolates across genetic clusters. Genetic cluster US�A was used a reference group for comparisons. Reference Test Boscalid Iprodione Prothioconazole Mean EC�� di�erencea ��% CIa Mean EC�� di�erence ��% CIa Mean EC�� di�erence ��% CIa low high low high low high US�A US�B �.��� -�.��� �.��� �.��� -�.��� �.��� -�.��� -�.��� -�.��� US�A US� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� US�A COLPR� �.��� �.��� �.��� �.��� �.��� �.��� -�.��� -�.��� -�.��� US�A COLPR� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� �.��� ��� Figure A.� Mean di�erences of �� M. phaseolina isolates across experiments. Two separate experiments and two replicates per experiment were performed for each isolate and fungicide. Experiments are Bos� and Bos� for boscalid, Ipro� and Ipro� for iprodione and Pro� and Pro� for prothioconazole. Isolates with data from at least two replicates were included. ��� Figure A.� Linear regression models and correlation analyses of relative mycelial growth and EC�� values of �� M. phaseolina isolates. Log Absolute EC�� values were used. The line shows the linear regression with ��% con�dence interval shaded. Selected concentrations were � �g ml � for boscalid and iprodione and �.� �g ml � for prothioconazole. ��� Figure A.� Mean EC��(P) di�erences of �� M. phaseolina isolates across fungicides. Iprodione and proth- ioconazole as compared to boscalid. Isolates were most sensitive to prothioconazole, followed by boscalid and least sensitive to iprodione. ��� Figure A.� Correlation of mean EC�� (e�ective concentration to reduce growth by ��%) estimates and EC��(P) predictions for a validation set of �� M. phaseolina isolates determined from mycelial growth assays in Petri plates amended with multiple concentrations boscalid, iprodione or prothioconazole. ��� Figure A.� Mean EC�� /EC��(P) di�erences of M. phaseolina isolates by host. Only isolates collected from soybean or dry bean are shown. ��� Figure A.� Mean EC�� /EC��(P) di�erences of �� M. phaseolina isolates by genetic cluster. Isolate TN��� not shown, since there is no information of genetic cluster membership. ��� CHAPTER � ECOCLIMATIC SUITABILITY AND ADAPTIVE GENOMICS IN MACROPHOMINA PHASEOLINA, THE CHARCOAL ROT PATHOGEN ��� �.� Abstract Globally, charcoal rot caused by the fungal pathogen Macrophomina phaseolina is listed among the top diseases threatening agricultural production. The environment has a profound in�uence on plant dis- eases, however the e�ect of accelerated climate change on disease development is uncertain and host- pathogen system speci�c. We studied the distribution and genomic adaptive potential of M. phaseolina, a major fungal plant pathogen, in relation to climate. We retrieved worldwide occurrences of M. phase- olina to develop an explanatory species distribution model using climatically relevant variables. Georef- erenced occurrences in the global biodiversity information facility (GBIF) database and information of M. phaseolina isolates collected in the US, Colombia and Puerto Rico reported in a previous study were used. Occurrence data and climatic variables were used to identify within species worldwide suitability patterns in M. phaseolina. Candidate adaptive loci associated with climatic variation were used to calcu- late an adaptive index and infer the distribution of adaptive genetic variation in M. phaseolina. A global species distribution bioclimatic model for M. phaseolina identi�ed areas of high climatic suitability for its occurrence that is consistent with all current records. Notable areas of high suitability were projected in the southern US, north-eastern Argentina, eastern Australia, and southern Europe, where outbreaks were recently reported. �.� Introduction Changes in climate are already a�ecting disease incidence in agricultural systems (Altizer et al., ����; Váry et al., ����; Velásquez et al., ����). Very often these e�ects depend on the patterns of climate change and the host-pathogen system. For example, pathogen distribution and crop disease severity are driven to a large extent by particular changing patterns in temperature, rainfall events and humidity (Altizer et al., ����; Sparks et al., ����; Velásquez et al., ����; Yonow et al., ����; Dudney et al., ����). Furthermore, responses to changing climate are intricately tied to organisms potential adaptive mechanisms and intra- speci�c variation in those mechanisms (local adaptation), which in turn are in�uenced by factors such as gene �ow, and phenotypic plasticity (Savolainen et al., ����; Savolainen et al., ����; Croll and Mc- donald, ����; Waldvogel et al., ����). Thus, both environmental and evolutionary potential should be investigated and considered when modeling the distribution of species. ��� Models of climate change for the coming decades predict increases in global temperature, rainfall and severe weather (Fisher et al., ����). This is expected to increase the climatic variation that already is present across di�erent agricultural systems and regions of the world. To predict how pathogens geo- graphic distribution will be altered under future climate changes it is necessary to understand how the cur- rent pathogen distribution depend on climatic factors (Shaw and Osborne, ����). However, the speci�c environmental factors that contribute to the current distributions and disease occurrences have not been characterized and species distribution models (SDMs) have not been developed for most plant pathogens (Ireland and Kriticos, ����). Charcoal rot, caused by the widespread pathogen Macrophomina phaseolina, is listed among the top �� diseases causing soybean yield losses in the US as well as globally (Allen et al., ����; Savary et al., ����; Bradley et al., ����). Diseases caused by M. phaseolina are favored by high temperatures and drought episodes and these conditions are known to play a key role in triggering epidemics (Dhingra and Sinclair, ����; Meyer and Sinclair, ����; Kendig et al., ����; Yang and Navi, ����; Mengistu et al., ����a; Mengistu et al., ����b; Reznikov et al., ����). In the past few years, a surge in �rst reports of diseases caused by M. phaseolina in a variety of crops and countries have been observed, including hemp in southern Spain (Casano et al., ����), tomato from Pakistan (Hyder et al., ����), stevia in North Carolina (Koehler and Shew, ����), sugarcane in China (Wang et al., ����), zebra plant in Serbia (Tančić Živanov et al., ����), catnip in India (Nishad et al., ����), grapevine in the US (Nouri et al., ����), strawberry in Italy (Gerin et al., ����), Malabar spinach in India (Meena et al., ����) among others. Interest in the interaction of climate-charcoal rot have been rising and the associations between charcoal rot and climate have been examined through review studies (Batista et al., ����; Cohen et al., ����). However, predicting the e�ects of climate change on M. phaseolina distribution remains limited and models have not been used to predict the climate suitability of this pathogen. Species distribution modelling is an important tool in ecology and biogeography to investigate species ranges and factors contributing to their distribution (Sutherland, ����; Elith and Leathwick, ����; Ju- roszek and Von Tiedemann, ����). SDMs have been used to predict the distributions of plant pathogens as determined by climate (Burgess et al., ����; Yonow et al., ����) and to assess the risk of disease (Sparks ��� et al., ����) and epidemics (Paini et al., ����). In addition to environmental conditions, evolutionary pro- cesses, including those within-species, are crucial in species response to climate (Jay et al., ����). There- fore, genomic data is increasingly considered in SDMs being of special interest adaptive genetic variation (Waldvogel et al., ����). Adaptive genetic variation can provide insights into climate adaptation mecha- nisms and the potential of rapid local adaptation to occur in the future under climate change. Neverthe- less, SDM approaches often encounter challenges incorporating evolutionary information (Waldvogel et al., ����). Recent developments in genotype-environment associations using redundancy analysis allow insights into patterns of adaptive variation and can be used for the identi�cation of candidate adaptive genomic loci and adaptive indices in widespread non-model species (Steane et al., ����; Capblancq and Forester, ����). These tools have the potential to estimate adaptive indices associated with climatic variation in fungal species in a landscape genomics framework. Indeed, candidate adaptive loci were previously iden- ti�ed in M. phaseolina (Ortiz et al., under review) which can be used to calculate adaptive indices in this pathogen. Adaptive indices provide a measure of the adaptive genetic similarity on the landscape as a function of climatic variables values at each location across the landscape (Steane et al., ����; Capblancq and Forester, ����). This study investigated the e�ect of climatic variables in shaping the distribution of M. phaseolina on a global scale, incorporating evolutionary projections. The objectives of this study were to describe the climatic suitability and calculate an adaptive genetic-based index of M. phaseolina on a global scale. We speci�cally developed an explanatory global distribution bioclimatic model by as- sociating recorded locations of M. phaseolina with climatic variables and projected an adaptive genomic index across the M. phaseolina distribution. �.� Results �.�.� Climatic suitability model A correlative bioclimatic model based on M. phaseolina occurrence data and �ve climatic variables re- lated to temperature and precipitation was developed using BIOCLIM. The model captured areas of cli- matic suitability for M. phaseolina occurrences in every continent, which is consistent with this pathogen records (Batista et al., ����). The model mean AUC obtained via cross-validation with presence/pseudo- ��� absence data was �.�� (Supplementary Figure A.�). A high AUC indicates that locations with high pre- dicted suitability scores tend to be locations of known presence (i.e., true positive rate). While an AUC score of �.� correspond to random predictions. We found useful discriminatory ability of suitable vs. unsuitable areas with our model considering the number of records included in this study and that for presence-background data models the maximum possible AUC is less than � (Phillips et al., ����). The model projected high suitability for localities with low precipitation of driest quarter (BIO��) and pre- cipitation of warmest quarter (BIO��), and high mean temperature of warmest quarter (BIO��) (Figure A.�). These predictions correspond to an expected distribution of higher M. phaseolina suitability at warm and dry regions. In the US, predicted suitable regions were concentrated across locations in south- ern states, including Texas, Oklahoma, Kansas, Arkansas, and Missouri. The highest suitability values in the US were projected in a region of Arizona (southwest US). Although, generally lower than for south- ern regions, regions of high suitability were projected as well in locations in the East and West North Central regions (Figure �.�). In Colombia, regions of intermediate suitability were predicted mainly in the extreme north and east- ern plains of Colombia (Caribbean and Orinoquia regions, respectively). Similar, intermediate suitability values were predicted in Puerto Rico and other islands of the Caribbean. A trend of highly suitable val- ues was observed in southern Europe particularly along coastal regions of Spain, France and Italy, and localities of eastern Europe. Notably, the model predicted a large region of high suitability in the north- east of Argentina, referred as the Plata Plain region, with highest values in areas near Buenos Aires and La Pampa provinces. Reports of M. phaseolina occurrence and disease outbreaks in soybean, canola and strawberry has been recently observed in northern provinces of Argentina (Gaetán et al., ����; Baino et al., ����; Viejobueno et al., ����; Reznikov et al., ����). Likewise, a high suitability is observed in regions of eastern Australia and south-eastern South Africa for which increased charcoal rot incidence has been reported (Hutton et al., ����; Jordaan et al., ����a) (Figure �.�). �.�.� Spatial autocorrelation Precipitation of warmest quarter (BIO��) and precipitation of driest quarter (BIO��) are aggregated with similar precipitation values occurring within approximately ���� km showing a maximum correlation ��� Figure �.� BIOCLIM global climatic suitability model for Macrophomina phaseolina. BIOCLIM algo- rithm and presence-background data records were used. A suitability value of � (green) indicates a location with high suitability and a value of � is given for locations predicted as unsuitable. Climatic variables used as predictors in the model were BIO�� = Precipitation of Warmest Quarter, BIO�� = Precipitation Sea- sonality (Coe�cient of Variation), BIO�� = Precipitation of Driest Quarter, BIO�� = Mean Temperature of Warmest Quarter and BIO� = Temperature Seasonality (standard deviation *���). (r > �.�) (Supplementary �gure A.�). The correlation decreases rapidly at distances greater than approx- imately ���� km between points. At distances ���� km, for most distance classes, the correlation is neg- ative. Most climatic variables show a similar pattern of aggregation at points within ���� km of each other, decreasing near to zero rapidly and shifting to negative correlations at greater distances. An ex- ception is BIO��, mean temperature of warmest quarter, which showed peaks of positive correlation at greater distances (Supplementary �gure A.�). Based on these results, we can reject the null hypothesis that geographic and climatic distances are ��� uncorrelated with p = �.��� for BIO��, BIO�� and BIO�. No signi�cant correlation was found for BIO�� (p=�.���) and for BIO�� (p=�.���). The observed correlations for BIO��, BIO�� and BIO� were r=�.��, �.��, �.�� respectively indicate that points that are closer to each other have more similar climatic values than points that are far from each other. �.� Discussion In this study, we developed a correlative BIOCLIM model for M. phaseolina to project the climate suit- ability of M. phaseolina and identify localities at risk of charcoal rot and other diseases caused by this pathogen at a global scale. Previous studies have reported the current M. phaseolina global distribution and its association to climate at a continental or biome resolution (Batista et al., ����). This model con- stitutes the �rst attempt to predict the distribution of M. phaseolina at a resolution of approximately �� km. Importantly, by using global records we provide an examination of temperature and precipitation variables that are predicted to be highly suitable for the occurrence of M. phaseolina. The current distribution and disease dynamics of M. phaseolina are heavily in�uenced by climatic factors such as high temperature and low soil water availability (Sexton et al., ����; Batista et al., ����; Marquez et al., ����; Cohen et al., ����). The model was consistent with charcoal rot reports in areas with high mean temperature of warmest quarter (BIO��) and low precipitation of driest quarter (BIO��) and precipitation of warmest quarter (BIO��) around the world. In the US, areas projected as most suitable are in states with reported highest soybean yield losses due to charcoal rot (Allen et al., ����; Bradley et al., ����). Although, these reports are highest in the warmest and southernmost states, charcoal rot is a consistent threat to soybean grown in the northern US regions as well (Bradley et al., ����; Roth et al., ����). Our results of M. phaseolina potential distribution indicated by areas of intermediate suitability along the east north central and west north central regions suggest potential for further expansion of charcoal rot occurrences to these regions in the US. Globally, our results projected the north-eastern region of Argentina as one of the largest areas with high suitability. The provinces of Buenos Aires, Tucuman and other northern provinces, have already reported charcoal rot epidemics in soybean, strawberry and canola (Gaetán et al., ����; Baino et al., ����; Viejobueno et al., ����; Reznikov et al., ����). Likewise, the model projected areas in the Eastern Cape ��� and Free State provinces of South Africa as intermediate to highly suitable. Charcoal rot in soybean and sun�ower has been reported a�ecting �elds in Free state province, one of the major producing regions for these crops in South Africa (Jordaan et al., ����b). In Australia, the model is congruent with charcoal rot reports on olives, strawberry and other �eld and horticultural crops grown in eastern regions (Sergeeva et al., ����; Hutton et al., ����; Poudel et al., ����). These observations suggest that our model can predict suitability of M. phaseolina in regions for which data points were not included but with reported presence or disease caused by this pathogen. Thus, we suggest this model may be used as an indicator for the potential risk of disease development. Similar models have used climatic suitability as proxy for disease caused by fungal and oomycete pathogens (Burgess et al., ����; Hernández-Lambraño et al., ����; Yonow et al., ����). Accurate predictions on the e�ects of climate on species occurrences face several challenges (Phillips et al., ����; Franklin, ����). For presence-background models, one of such challenges is that the accu- racy of predictions is highly dependent on methods for background data selection (Phillips et al., ����; Hijmans et al., ����). Model performance as assessed by AUC in presence-background models tend to in- crease with larger spatial extents from which background points are sampled. To address this, we sampled background points within a radius of ��� km from the presence records (VanDerWal et al., ����). This, although appropriate for our data, contributed to the relatively low observed AUC value (Phillips et al., ����). In addition, presence-only and presence-background models using environment-only data have been identi�ed as least accurate as compared to true-absence models in which additional factors related to the biology or epidemiology of organisms are accounted for to environment (Phillips et al., ����). Thus, a limitation of our model is the relatively low number of records used to build the model, as com- pared to climatic suitability models developed at a global scale for other pathogens (Burgess et al., ����; Hernández-Lambraño et al., ����; Yonow et al., ����) and the use of climatic only data. To address the lack of biological data in our model and to provide an estimation of the e�ects of evolutionary processes into M. phaseolina distribution we used a complementary approach to model within-species evolutionary factors by estimating an adaptive index. This index was estimated using pre- viously identi�ed candidate loci for climate adaptation in data set of �� M. phaseolina collected across ��� the US, Colombia, and Puerto Rico (Ortiz et al., under review). This data set encompasses isolates col- lected across a wide range of climates, making it suited for studying within-species adaptation to climate. The adaptive index estimated for the data set of �� M. phaseolina isolates suggest that even in locations where similar temperatures are observed, isolates may di�erentially respond depending on the presence of adaptive loci. Climatic gradients have now been reported in plant species (Steane et al., ����; Capblancq and Forester, ����), however this the �rst time it has been used to predict the adaptive landscape in a plant pathogen. In summary, we provided a �rst species distribution model that serve as a basis for future more comprehensive predictive SDMs. Our model will also be useful for local adaptation that constitute the �rst step towards assessing the adaptive response of this fungal pathogen under climate change. Fur- ther improvements of the model will involve including larger data sets and the use of semi-mechanistic models (e.g., MAXENT) that allow the incorporation of biological parameters (Phillips et al., ����), for example growth rates at di�erent temperatures in fungal plant pathogens. Given the increasing impact of M. phaseolina on agroecosystems globally, the modelling of its distribution o�ers an important pre- liminary tool for monitoring and development of management strategies incorporating eco-evolutionary projections. Further, regional distribution models would provide a better assessment of charcoal rot risk in di�erent crops. From a practical standpoint, of particular interest are crops and locations for which disease assessments data over time is available such as is the case for charcoal rot of soybean in the US (Bradley et al., ����). A major need remains for M. phaseolina and other plant pathogens to examine the incorporation of disease risk assessments into management strategies. �.� Materials and methods �.�.� Study area and distribution data Distribution data was obtained from two sources, records of M. phaseolina occurrences retrieved from the global biodiversity information facility (GBIF) database (GBIF.org) and a dataset on a collection of isolates throughout the US, Puerto Rico, and Colombia for which genomic data is available (Ortiz et al., under review). A total of ��� records for “Macrophomina phaseolina” were obtained from GBIF using R v�.�.� (R Core Team ����). After �ltering for missing data and cleaning for potential georeferenti- ation mistakes, ��� records were maintained. For additional �� records without longitude and latitude ��� information, coordinates based on location description were retrieved using the ‘geocode’ function as implemented in R v�.�.� (R Core Team ����). The longitude and latitude information of a collection of �� isolates of M. phaseolina isolates as well as genomic data and adaptive candidate loci available from a previous population genomics study were used (Ortiz et al., under review). In brief, these isolates were collected mainly from soybean and dry bean in the US, Colombia and Puerto Rico from commercial �elds and experimental stations. The entire data set covered occurrence records originating from plant tissues or soil in every continent, but Antarctica (Figure �.�), consistent with the current reported distri- bution of M. phaseolina (Batista et al., ����). Figure �.� Geographic locations of Macrophomina phaseolina records included in the BIOCLIM model. Records obtained from the global biodiversity information facility (GBIF) are shown in orange circles. Isolate collection sites of �� M. phaseolina isolates collected in the US, Puerto Rico and Colombia are depicted in black. ��� �.�.� Distribution model Five bioclimatic variables previously selected from the �� standard bioclimatic variables (WorldClim v�) as described previously were used (Ortiz et al., under review). In summary, the bioclimatic variables are the average for the years ���� to ���� and were obtained at a resolution of �.� min ( ��.� km� ) which cor- respond with that of the data for the isolate collection, recorded at a �eld to municipality scale. This set of climatic variables was selected based on ecological relevance and after removing correlated variables (|r| > �.�). The selected variables were: BIO�� = Precipitation of Warmest Quarter, BIO�� = Precipitation Sea- sonality (Coe�cient of Variation), BIO�� = Precipitation of Driest Quarter, BIO�� = Mean Temperature of Warmest Quarter and BIO� = Temperature Seasonality (standard deviation *���). The species distribution model (SDM) was built using BIOCLIM as implemented in ‘dismo’ R pack- age (Hijmans et al., ����). We used BIOCLIM algorithm with presence-background data. The algorithm creates percentile distributions for the climatic data values at the locations of species occurrence (“training sites”). The values for each climatic variable are compared to the percentile distribution of the training sites providing a measure of similarity between locations. Since one-tailed percentile distributions are used (��th percentile is treated as equivalent to ��th percentile), the closer to the ��th percentile (the me- dian), the more suitable a location is. Here, we used the ‘dismo’ implementation in which the suitability values are scaled, thus resulting in values between � and �. The value of � is given for a location that would have the median values of the training data for all the variables considered, while � will be given for cells with climatic values outside of the range of the training data for at least one of the variables. The �nal BIOCLIM model was �tted with all presence records from the GBIF cleaned dataset and �� records from the previously published M. phaseolina isolate collection using the �ve selected climatic variables as predictors. Since we used a presence-background species distribution modeling approach, we selected background data for model parameterization (Hijmans and Elith, ����). Background localities were generated at ran- dom within a radius of ��� km from the presence records (VanDerWal et al., ����). The models were assessed and compared according to their discrimination capacity of suitable versus unsuitable areas for M. phaseolina using the area under the receiver operator curve (AUC) in the ‘dismo’ implementation. ��� Two additional classi�cation assessment indices were used (Fielding Bell, ����): sensitivity (true posi- tive rate i.e., the proportion of correctly classi�ed presences) and speci�city (true negative rate, i.e., the proportion of correctly classi�ed absences). We divided the presence data in training and testing sets via cross-validation with k-fold (k=�) data partitioning. The background data was only used for model test- ing and was not partitioned. The mean AUC of the �ve cross-validation runs was reported as well as the maximum of the sum of the sensitivity (true positive rate) and speci�city (true negative rate) (Hijmans et al., ����). �.�.� Spatial autocorrelation Spatial autocorrelation was tested using BIOCLIM in ‘dismo’ R package (Hijmans et al., ����). A subset of �� records out of the �� isolate collection records for which the resolution was at least to the munici- pality level. Similarly, the GBIF records with exact longitude and latitude coordinates as recorded in the GBIF dataset were used (i.e., records that georeferenced using geocode were excluded) for spatial auto- correlation analysis. The associated climatic data values for the �ve variables for each record, as it was retrieved for the SDM analysis, was used. A geographic distance matrix was computed using longitude and latitude coordinates as well as distance matrices for each of the �ve environmental predictors. Correl- ograms for each of the climatic variables were performed using �� distance classes with ���� km distance increments. A Mantel test was run for each of the environmental predictors between the geographic distance matrix and the environmental distance matrix for each climatic predictor to test for signi�cant autocorrelation at each distance class. ��� BIBLIOGRAPHY Allen TW, Bradley CA, Sisson AJ, Byamukama E, Chilvers MI, Coker CM, Collins AA, Damicone JP, Dorrance AE, Dufault NS, et al (����) Soybean yield loss estimates due to diseases in the United States and Ontario, Canada, from ���� to ����. Plant Heal Prog ��: ��–�� Altizer S, Ostfeld RS, Johnson PTJ, Kutz S, Harvell CD (����) Climate Change and Infectious Dis- eases: From Evidence to a Predictive Framework. Science (��- ) ���: ���–��� Baino OM, Salazar SM, Ramallo AC, Kirschbaum DS (����) First Report of Macrophomina phase- olina Causing Strawberry Crown and Root Rot in Northwestern Argentina . Plant Dis ��: ����–���� Batista E, Lopes A, Alves A (����) What do we know about Botryosphaeriaceae? An overview of a worldwide cured dataset. Forests. doi: ��.����/f�������� Bradley CA, Allen TW, Sisson AJ, Bergstrom GC, Bissonnette KM, Bond J, Byamukama E, Chilvers MI, Collins AA, Damicone JP, et al (����) Soybean Yield Loss Estimates Due to Diseases in the United States and Ontario, Canada, from ���� to ����. Plant Heal Prog ��: ���–��� Burgess TI, Scott JK, Mcdougall KL, Stukely MJC, Crane C, Dunstan WA, Brigg F, Andjic V, White D, Rudman T, et al (����) Current and projected global distribution of Phytophthora cinnamomi, one of the world’s worst plant pathogens. Glob Chang Biol ��: ����–���� Capblancq T, Forester BR (����) Redundancy analysis: A Swiss Army Knife for landscape genomics. Methods Ecol Evol ��: ����–���� Casano S, Cotan AH, Delgado MM, García-Tejero IF, Saavedra OG, Puig AA, de los Santos B (����) First report of charcoal rot caused by Macrophomina phaseolina on hemp (Cannabis sativa) varieties cultivated in southern Spain. Plant Dis ���: ���� Cohen R, Elkabetz M, Paris HS, Gur A, Dai N, Rabinovitz O, Freeman S (����) Occurrence of Macrophomina phaseolina in Israel: Challenges for Disease Management and Crop Germplasm En- hancement. Plant Dis ���: ��–�� Croll D, Mcdonald BA (����) The genetic basis of local adaptation for pathogenic fungi in agricul- tural ecosystems. Mol Ecol. doi: ��.����/mec.����� Dhingra OD, Sinclair JB (����) Survival of Macrophomina phaseolina Sclerotia in Soil: E�ects of Soil Moisture, Carbon:Nitrogen Ratios, Carbon Sources, and Nitrogen Concentrations. Phytopathology ��: ���–��� Dudney J, Willing CE, Das AJ, Latimer AM, Nesmith JCB, Battles JJ (����) Nonlinear shifts in infectious rust disease due to climate change. Nat Commun. doi: ��.����/s�����-���-�����-� Elith J, Leathwick JR (����) Species Distribution Models: Ecological Explanation and Prediction Across Space and Time. Annu Rev Ecol Evol Syst ��: ���–��� Fisher MC, Henk DA, Briggs CJ, Brownstein JS, Mado� LC, McCraw SL, Gurr SJ (����) Emerging fungal threats to animal, plant and ecosystem health. Nature ���: ���–��� ��� Franklin J (����) Species distribution models in conservation biogeography: Developments and chal- lenges. Divers Distrib ��: ����–���� Gaetán SA, Fernandez L, Madia M (����) Occurrence of Charcoal Rot Caused by Macrophomina phaseolina on Canola in Argentina. Plant Dis ��: ��� Gerin D, Dongiovanni C, De Miccolis Angelini RM, Pollastro S, Faretra F (����) First Report of Macrophomina phaseolina Causing Crown and Root Rot on Strawberry in Italy. Plant Dis ���: ���� Hernández-Lambraño RE, González-Moreno P, Sánchez-Agudo JÁ (����) Environmental factors associated with the spatial distribution of invasive plant pathogens in the Iberian Peninsula: The case of Phytophthora cinnamomi Rands. For Ecol Manage ���–���: ���–��� Hijmans RJ, Elith J (����) Species Distribution Modeling with R. Ref Modul Life Sci. doi: ��.����/b���-�-��-������-�.�����-� Hijmans RJ, Phillips S, Leathwick J, Maintainer JE (����) Package “dismo” Species Distribution Modeling. Hutton DG, Gomez AO, Mattner SW (����) Macrophomina phaseolina and Its Association with Strawberry Crown Rot in Australia. Int J Fruit Sci ��: ���–��� Hyder S, Gondal AS, Ahmed R, Sahi ST, Rehman A, Hannan A (����) First Report of Charcoal Rot in Tomato Caused by Macrophomina phaseolina (Tassi) Goid. From Pakistan. Plant Dis ���: ���� Ireland KB, Kriticos DJ (����) Why are plant pathogens under-represented in eco-climatic niche modelling? Int J Pest Manag ��: ���–��� Jay F, Manel S, Alvarez N, Durand EY, Thuiller W, Holderegger R, Taberlet P, François O (����) Forecasting changes in population genetic structure of alpine plants in response to global warming. Mol Ecol ��: ����–���� Jordaan E, van der Waals JE, McLaren NW (����a) E�ect of irrigation on charcoal rot severity, yield loss and colonization of soybean and sun�ower. Crop Prot ���: ��–�� Jordaan E, van der Waals JE, McLaren NW (����b) E�ect of irrigation on charcoal rot severity, yield loss and colonization of soybean and sun�ower. Crop Prot ���: ��–�� Juroszek P, Von Tiedemann A (����) Plant pathogens, insect pests and weeds in a changing global climate: A review of approaches, challenges, research gaps, key studies and concepts. J Agric Sci ���: ���–��� Kendig SR, Rupe JC, Scott HD (����) E�ect of irrigation and soil water stress on densities of Macrophomina phaseolina in soil and roots of two soybean cultivars. Plant Dis ��: ���–��� Koehler AM, Shew HD (����) First Report of Charcoal Rot of Stevia Caused by Macrophomina phaseolina in North Carolina. Plant Dis ���: ���–��� Marquez N, Giachero ML, Declerck S, Ducasse DA (����) Macrophomina phaseolina: General Char- acteristics of Pathogenicity and Methods of Control. Front Plant Sci. doi: ��.����/fpls.����.������ ��� Meena BR, Nagendran K, Tripathi AN, Kumari S, Sagar V, Gupta N, Rai AB, Singh B (����) First report of charcoal rot caused by Macrophomina phaseolina in Basella alba in India. Plant Dis ���: ���� Mengistu A, Arelli PA, Bond JP, Shannon GJ, Wrather AJ, Rupe JB, Chen P, Little CR, Canaday CH, Newman MA, et al (����a) Evaluation of soybean genotypes for resistance to charcoal rot. Plant Heal Prog. doi: ��.����/PHP-����-����-��-RS Mengistu A, Smith JR, Ray JD, Bellaloui N (����b) Seasonal progress of charcoal rot and its impact on soybean productivity. Plant Dis ��: ����–���� Meyer W., Sinclair JB (����) Factors A�ecting Charcoal Rot of Soybean Seedlings. Phytopathology ��: ���–��� Nishad I, Srivastava AK, Saroj A, Babu BK, Samad A (����) First Report of Root Rot of Nepeta cataria Caused by Macrophomina phaseolina in India. Plant Dis ���: ���� Nouri MT, Zhuang G, Culumber CM, Trouillas F (����) First Report of Macrophomina phaseolina Causing Trunk and Cordon Canker Disease of Grapevine in the United States. Plant Dis PDIS-��-��- ����-PDN Paini DR, Sheppard AW, Cook DC, De Barro PJ, Worner SP, Thomas MB (����) Global threat to agriculture from invasive species. Proc Natl Acad Sci ���: ����–���� Phillips SJ, Anderson RP, Schapire RE (����) Maximum entropy modeling of species geographic distributions. Ecol Modell ���: ���–��� Poudel B, Shivas RG, Adorada DL, Barbetti MJ, Bithell SL, Kelly LA, Moore N, Sparks AH, Tan YP, Thomas G, et al (����) Hidden diversity of Macrophomina associated with broadacre and horticultural crops in Australia. Eur J Plant Pathol ���: �–�� Reznikov S, Vellicce GR, Mengistu A, Arias RS, Gonzalez V, Lisi V De, Garcia MG, Rocha Carla ML, Pardo EM, Castagnaro AP, Ploper LD (����) Disease incidence of charcoal rot (Macrophomina phaseolina) on soybean in north-western Argentina and genetic characteristics of the pathogen. Can J Plant Pathol ��: ���–��� Roth MG, Webster RW, Mueller DS, Chilvers MI, Faske TR, Mathew FM, Bradley CA, Damicone JP, Kabbage M, Smith DL (����) Integrated Management of Important Soybean Pathogens of the United States in Changing Climate. J Integr Pest Manag. doi: ��.����/jipm/pmaa��� Savary S, Willocquet L, Pethybridge SJ, Esker P, McRoberts N, Nelson A (����) The global burden of pathogens and pests on major food crops. Nat Ecol Evol �: ���–��� Savolainen O, Lascoux M, Merilä J (����) Ecological genomics of local adaptation. Nat Rev Genet ��: ���–��� Savolainen O, Pyhäjärvi T, Knürr T, Pyhaijiirvi T, Kniirr T (����) Gene Flow and Local Adaptation in Trees. Source Annu Rev Ecol Evol Syst ��: ���–��� Sergeeva V, Tesoriero L, Spooner-HartA R, Nair N (����) First report of Macrophomina phaseolina on olives (Olea europaea) in Australia. Australas Plant Pathol ��: ���–��� ��� Sexton ZF, Hughes TJ, Wise KA (����) Analyzing isolate variability of Macrophomina phaseolina from a regional perspective. Crop Prot ��: �–�� Shaw MW, Osborne TM (����) Geographic distribution of plant pathogens in response to climate change. Plant Pathol ��: ��–�� Sparks AH, Forbes GA, Hijmans RJ, Garrett KA (����) Climate change may have limited e�ect on global risk of potato late blight. Glob Chang Biol ��: ����–���� Steane DA, Potts BM, McLean E, Prober SM, Stock WD, Vaillancourt RE, Byrne M (����) Genome- wide scans detect adaptation to aridity in a widespread forest tree species. Mol Ecol ��: ����–���� Sutherland WJ (����) Predicting the ecological consequences of environmental change: A review of the methods. J Appl Ecol ��: ���–��� Tančić Živanov S, Dedić B, Dimitrijević A, Dušanić N, Mikić S, Jocić S, Miladinović D, Miklič V (����) First Report of Charcoal Rot on Zebra Plant (Aphelandra squarrosa) Caused by Macrophomina phaseolina. Plant Dis ���: ���� VanDerWal J, Shoo LP, Graham C, Williams SE (����) Selecting pseudo-absence data for presence- only distribution modeling: How far should you stray from what you know? Ecol Modell ���: ���–��� Váry Z, Mullins E, Mcelwain JC, Doohan FM (����) The severity of wheat diseases increases when plants and pathogens are acclimatized to elevated carbon dioxide. Glob Chang Biol ��: ����–���� Velásquez AC, Castroverde CDM, He SY (����) Plant–Pathogen Warfare under Changing Climate Conditions. Curr Biol ��: R���–R��� Viejobueno J, Ramallo AC, Kirschbaum DS, Baino OM, Salazar SM (����) Severe outbreaks of straw- berry crown and root charcoal rot caused by Macrophomina phaseolina in Tucumán, Argentina. Rev agronómica del noroeste argentino ��: ���–��� Waldvogel A-M, Feldmeyer B, Rolshausen G, Exposito-Alonso M, Rellstab C, Ko�er R, Mock T, Schmid K, Schmitt I, Bataillon T, et al (����) Evolutionary genomics can improve prediction of species’ responses to climate change. Evol Lett �–�� Wang W, Liu Z, Wang W, Song X (����) First Report of Macrophomina phaseolina Causing Stalk Rot of Sacha Inchi ( Plukenetia volubilis ) in China. Plant Dis ���: ���–��� Yang XB, Navi SS (����) First Report of Charcoal Rot Epidemics Caused by Macrophomina phase- olina in Soybean in Iowa. Plant Dis ��: ��� Yonow T, Ramirez-Villegas J, Abadie C, Darnell RE, Ota N, Kriticos DJ (����) Black Sigatoka in bananas: Ecoclimatic suitability and disease pressure assessments. PLoS One ��: e������� ��� APPENDIX SUPPLEMENTARY FIGURES ��� Figure A.� Model performance AUC values for the BIOCLIM model. Area under the receiver operator curve (AUC) values for each of �ve cross-validations runs illustrating discrimination capacity of suitable versus unsuitable areas for M. phaseolina. ��� Figure A.� Suitability values predicted for temperature and precipitation variables values found across locations of the complete data set of M. phaseolina occurrences. Predicted values are suitability values. A suitability value of � indicates a climatic value with predicted high suitability and a value of � is given for climatic values predicted as unsuitable. Climatic variables are BIO�� = Precipitation of Warmest Quar- ter, BIO�� = Precipitation Seasonality (Coe�cient of Variation), BIO�� = Precipitation of Driest Quarter, BIO�� = Mean Temperature of Warmest Quarter and BIO� = Temperature Seasonality (standard devia- tion *���). ��� Figure A.� Mantel’s correlograms of climatic variables used in the BIOCLIM model. Correlation be- tween climatic and geopgraphic distances. Geographic distance classes are de�ned by ���� km incre- ments. Climatic variables are BIO�� = Precipitation of Warmest Quarter, BIO�� = Precipitation Season- ality (Coe�cient of Variation), BIO�� = Precipitation of Driest Quarter, BIO�� = Mean Temperature of Warmest Quarter and BIO� = Temperature Seasonality (standard deviation *���). ��� CHAPTER � CONCLUDING STATEMENT ��� Charcoal rot and diseases caused by Macrophomina phaseolina are a threat to agricultural production a�ecting many important economic and subsistence crops worldwide. Importantly, one of increasing concern under climate change. This research focused on understanding the genetic diversity and evolu- tionary potential of M. phaseolina, to inform and provide tools for improved charcoal rot management strategies. Populations of M. phaseolina in the continental US, Puerto Rico and Colombia collected from soy- bean and dry bean �elds were found to be structured in a hierarchical manner with subcontinental re- gional stability and instability at local scales consistent with a metapopulation dynamics perspective. These results are in line with a scenario of evolution after migration driven by divergence following clonal expansions. Additionally, this research identi�ed the potential for anthropogenic in�uence in the move- ment of M. phaseolina to locations around the world. Climate was found to signi�cantly contribute to genetic divergence in this pathogen and identi�ed candidate genomic regions for adaptation. Putatively adaptive functions associated to these regions may bene�t M. phaseolina in speci�c environments. This knowledge expands the impact that population genomics and genotype-environment associations can have on our ability to characterize adaptive potential in plant pathogens. E�ective chemical-control means are lacking for the management of charcoal rot. Therefore, the ef- �cacy of active ingredients currently used in commercial fungicide formulations in crop production was investigated. Our results on the in-vitro e�cacy of boscalid, iprodione and prothioconazole indicate that formulations with these active ingredients, may reduce M. phaseolina seedling infection originating from infected seeds or inoculum in the soil. Particularly, our results on the in-vitro e�cacy of prothioconazole suggest that commercial formulations with this active ingredient may be of particular interest for charcoal rot management. Information regarding mutations in fungicide target genes was lacking for M. phase- olina. None of the point mutations found in our isolate collection were correlated with levels of fungicide sensitivity. Finally, in this study we developed a bioclimatic model for M. phaseolina to project the climate suitability of M. phaseolina at a global scale and identify localities at risk of charcoal rot and other dis- eases caused by this pathogen. The model projected high suitability for localities with low precipitation of driest quarter and precipitation of warmest quarter, and high mean temperature of warmest quarter. ��� Notably, areas of high suitability were projected in the southern US, north-eastern Argentina, eastern Australia, and southern Europe. These predictions correspond to an expected distribution of higher M. phaseolina suitability at warm and dry regions and with increased disease reports in these regions. �.�.� Future directions Future studies investigating the adaptive potential of M. phaseolina will be needed to identify the degree to which global populations re�ect their adaptation to host and climate. Such studies will bene�t from comprehensive samplings schemes including diverse hosts and climates. In addition, long-read sequenc- ing technologies will allow further characterization of the role of genomic variation, including structural variation, in M. phaseolina adaptation to host and the climatic environment. Our data on the in-vitro e�cacy of prothioconazole suggest that commercial formulations with this active ingredient may be of particular interest for future in-vivo e�cacy testing in soybean and dry bean. Data on the in-vivo e�cacy of prothioconazole in preventing seedling colonization and charcoal rot dis- ease development is needed in order to determine its e�ectiveness in charcoal rot control. Additional charcoal rot management e�orts should be directed at identifying novel e�ective fungicides and moni- toring the potential development of resistance to fungicides use in modern crop production. Given the increasing impact of M. phaseolina on agroecosystems globally, the modelling of its dis- tribution constitutes an important tool for the monitoring and development of management strategies. More comprehensive predictive species distribution models, including ensemble models, should provide a better understanding of the adaptive response of this fungal pathogen under climate change. Further improvements of the model presented in this research, will involve the use of larger data sets and semi- mechanistic models. Similarly, regional distribution models would provide a better assessment of char- coal rot risk for major crop production regions. A major need remains to incorporate disease risk assess- ments and eco-evolutionary projections into charcoal rot management strategies. The characterization of adaptation in plant pathogens enabled by population genomics should become increasingly utilized for plant disease risk prediction models especially under climate change. ���