Spatial pattern of genetic diversity in field populations of Fusarium incarnatum‐equiseti species complex

Abstract Fusarium is associated with a number of wilt, blight, scab, and rot diseases in a range of economically important staple food crops worldwide. An assessment of the genetic structure and population stratification of Fusarium incarnatum‐equiseti species complex (FIESC) pathogen populations is important to understand the evolutionary potential of such populations in adapting to environmental change. Based on intersimple sequence repeat polymerase chain reaction (ISSR‐PCR), it was found that the pathogen population was structured into three genetic clusters for which genetic differentiation was higher within than among populations. There was high intrapopulation genetic diversity for population 1 (94.63%) which consisted largely of isolates collected from North Trinidad. Populations 2 and 3 had a low level of admixture among the populations based on overall population differentiation. Population 1 accounted for the highest amount of genetic variation (95.82%) followed by populations 2 and 3. Population stratification was reflected in the dendrogram topology, which consisted of three main genetic clusters and which coincided with the outcome of Bayesian and PCoA analyses. The populations were isolated by distance, and Voronoi tessellations indicated physical or structural barriers to gene flow which contributed to restricted admixture between two of three populations. These findings suggest a high evolutionary potential for this FIESC pathogen population, the implications of which directly affect disease management strategies.


| INTRODUC TI ON
Fusarium is among the most studied groups of plant-pathogenic fungi, and member species are associated with a number of wilt, blight, scab, and rot diseases in a range of economically important staple food crops worldwide (Dean et al., 2012). This genus is large and comprises a membership of more than 1,500 species many of which are grouped into 23 defined Fusarium species complexes (Summerell, 2019). Certain Fusarium species, including Fusarium incarnatum-equiseti species complex (FIESC), produce a range of mycotoxins; of notable importance are the trichothecenes, which pose a serious risk to domestic animal and human health (Desjardins, 2006). Some species are also responsible for opportunistic or secondary infections | 9011 RAMPERSAD (fusariosis) in immunocompromised humans in clinical settings (Nelson et al., 1994;Nucci & Anaissie, 2007;O'Donnell et al., 2013;Short et al., 2011). Such trans-kingdom pathogenicity with the proven ability to develop cross-resistance to azole fungicides presents a major public health concern and contributes to the estimated billion-dollar healthcare burden (Ananda-Rajah et al., 2011;Benedict et al., 2019;Brown et al., 2012).  (Medina et al., 2017). New introductions may arise inadvertently through anthropogenic movement via seed and grain trade (Lee et al., 2015). The impact of long-range, aerial dispersal of pathogens on global distribution of plant diseases remains a critical epidemic-associated factor (Brown & Hovmoller, 2002;Schmale & Bergstrom, 2003;Schmale et al., 2006). Palmero et al. (2011) and others reported that spores (ascospore and macroconidia) of Fusarium, with differing levels of pathogenicity, are able to cross the Atlantic carried by winds from the Sahara (Africa) to the Caribbean, Europe, and the Mediterranean (Griffin et al., 2003;Prospero et al., 2005;Prospero & Lamb, 2003). As much as 2 million metric tons of dust are deposited each year (Guerzoni et al., 1997).

Adoption of strategies for integrated management of
Selection pressures work cooperatively to favor the predominance of certain introduced genotypes that are well-adapted to available host(s) and their new environment (Summerell et al., 2007).
Fusarium populations are dynamic in their responses to environmental cues, the consequence of which include increased yield losses, reduction in quality, or changes in the mycotoxins produced (Valverde-Bogantes et al., 2019). An assessment of the genetic structure and population stratification of pathogen populations is important to understanding the evolutionary potential of such populations in adapting to environmental change, the effects of selection pressures, and the relative impact of the various drivers of evolutionary change, for example, genetic drift, mutation, gene flow, isolation with the ultimate selection of those genotypes with increased biological fitness, and subsequent adaptive advantage (McDonald & Linde, 2002). Genetic structure would also reveal signatures of sexual recombination and clonal spread of isolates within and among populations (Heule et al., 2017). A population with highly pathogenic isolates and different trichothecene chemotypes can replace or dominate another (Fulcher et al., 2019;Kelly et al., 2015).
An efficient approach to estimating the genetic structure of pathogen populations is to randomly screen representatives using selectively neutral molecular marker loci on geographically defined samples collected under a hierarchical sampling strategy (Grünwald et al., 2017). ISSR-PCR (Inter Simple Sequence Repeat PCR/single primer amplification reactions (SPAR)/microsatellite-primed PCR (MP-PCR)) relies on amplification of a target region of nucleotide sequences defined by anchored or nonanchored SSR (simple sequence repeats) homologous primers (Zietkiewicz et al., 1994). ISSRgenerated polymorphisms are analyzed as dominant genetic markers and are widely used over other polymorphism generators, for example, RAPD-randomly amplified polymorphic DNA and AFLPamplified fragment length polymorphism, to determine genetic variations due to cost effectiveness, reproducibility, generation of a high level of polymorphism because they are inherently very variable, and a priori knowledge about the target sequences is not required (Ng & Tan, 2015). Further, because ISSRs are ubiquitously distributed across the genome, the entire genome under study is theoretically represented in the resulting multilocus data rather than a few, user-selected gene regions. A number of studies have characterized different Fusarium populations in plants using ISSR genetic markers (Akbar et al., 2018;Altinok et al., 2018;Dinolfo et al., 2010;Hamdi et al., 2019;Mishra et al., 2003Mishra et al., , 2004Nawade et al., 2017;Singh et al., 2019;Thangavelu et al., 2012). However, Lee et al. (2015) re- . Fusarium incarnatumequiseti (FIESC) species complex is among the major fungal pathogens that cause fruit rot disease which leads to high production cost, severe economic losses, and, ultimately, inconsistent production (Ramdial et al., 2016(Ramdial et al., , 2017. The FIESC includes over 30 recognized phylogenetic species (phylo-species) which have been characterized through the steadfast efforts of a number of research groups, worldwide (Aoki et al., 2014;Maryani et al., 2019;O'Donnell et al., 2009O'Donnell et al., , 2010Santos et al., 2019;Villani et al., 2016;Wang et al., 2019).
Correct identification of Fusarium haplotypes is carried out according to published guidelines O'Donnell et al., 2015;O'Donnell et al., 2010)  In view of the severity of the disease in bell pepper in Trinidad, the dispersal of fusaria via Saharan dust storms, poor management of the disease, and inability to reduce pathogen populations through chemical means, there are several biological questions that must be answered through population genetics analyses for example: Are FIESC populations differentiated? Is there gene flow among populations? Are populations clonal, sexual, or mixed? It is hypothesized that FIESC populations in bell pepper fields would be connected by some level of contemporary gene flow partly because seed and seedlings are commonly sourced and the comparatively short distance (km) among fields may not facilitate isolation by distance. As such, the level of population differentiation and genetic diversity may vary according to region, that is, North and South Trinidad but not necessarily at a finer scale, that is, according to field location.
One expectation is that the genetic structure of pathogen populations would be subjected to local extinction-recolonization events as the host is harvested, destroyed, and then re-planted during the growing season. Trans-Atlantic spore movement which although stochastic may be representative of unrestricted gene flow and leads to less differentiated populations. Related to these expectations is the "monoculture effect" where the association between low host species diversity and high disease incidence is a result of high homogeneity in the host population (King & Lively, 2012;Thrall et al., 2003). Based on these hypotheses, the objectives of this study were to (a) estimate the genetic diversity and population structure of FIESC in Trinidad, (b) characterize genetic differentiation within and among populations, (c) determine whether populations were isolated by distance.

| ISSR markers
Sampling sites are indicated in Figure 1, and isolates collected and used in the analysis are outlined in Table S1. Only those individuals that produced a polymorphic banding pattern for the ISSR primers and for which the interpretation of the banding pattern was unequivocal were retained in the final data set. Fragments with the same molecular weight were considered to be the same locus. In total, 335 polymorphic loci were generated from ISSR-PCR using the five ISSR primers selected from a screen of 22 ISSR primers (Table S2).

| Population stratification
Assignment of isolates was carried out by Bayesian clustering im- and 17.16% of the total genetic variation, respectively. The PCoA showed that the isolates from North formed a distinct genetic cluster but there were isolates from the North which were more similar to those in the South genetic cluster than the North. Also, there were isolates from the South that were more similar to isolates from Mayo ( Figure 3).
The total genetic variation was partitioned into within-and among-population components. AMOVAs (Table 1) revealed that F I G U R E 2 (a) Dendrogram inferred from Nei's genetic distance matrix using the heuristic approach of the neighbor-joining algorithm with 1,000 bootstrapped replicates. Numbers are the nodes indicate bootstrap values ≥75%. Colors correspond to STRUCTURE-inferred optimal K clusters. (b) Bar plot of individual Q matrix coefficients for FIESC isolates from assignment tests carried out in STRUCTURE with K = 3 as the optimal number of populations. Each vertical bar represents an individual isolates, and bars are divided based on the probability of assignment of each individual to a given population. (c) Hard clustering membership based on genetic data and geographical distance. Colors correspond to STRUCTURE-inferred optimal K clusters F I G U R E 3 A principal coordinate (PCoA) plot of the first two coordinates calculated using Nei's D genetic distance among FIESC isolates in Trinidad. Colors correspond to STRUCTURE-inferred optimal K clusters the estimated genetic variation in the dataset was largely due to within-population variation (81%) as opposed to among-population variation (19%) at p < 0.001.

| Genetic clusters
Data from the genetic distance matrix were analyzed using the NJ algorithm ( Figure 2). Population stratification was reflected in the dendrogram topology, which consisted of three main genetic clusters and which coincided with the outcome of Bayesian and PCoA analysis. In genetic cluster 2 (red), there were two isolates (23 and 24-denoted by dashed branch lines) that were collected from North but placement occurred with South isolates. Similarly, two isolates (34 and 37-denoted by dashed branch lines) were collected from the North but were more genetically similar to isolates collected from Mayo (blue). These connections were depicted by the gradient lines extending from the corresponding nodes in population 2 (red) and population 3 (blue) to population 1 (green).

| Gene diversity
The overall mean gene diversity (H T ) was 0.1006, and the mean within-population gene diversity (H S ) was 0.0729. Population 1 had the highest gene diversity (h = 0.1202) followed by population 2 (h = 0.0599) and population 3 (h = 0.0386). Demographic variables such as population size appeared to be a factor in the gene diversity indices for populations. Pairwise comparisons of Nei's unbiased measures of genetic identity and genetic distance estimated for three inferred populations are given in Table 2. The overall fixation index G ST , as it relates to nearness to fixation, for the Trinidad population (G ST ) was 0.2755 which indicated that the total genetic variation mainly exists within the population.

| Isolation by distance
The correlation between genetic distance and logarithmic geographic distance was significant (p < 0.01) after 999 permutations based on Mantel's test ( Figure S1). The spatial clustering of Voronoi tessellation model behaved similar to the Pritchard et al. (2000) model from which the Q matrix and structure plot was based. Although both models utilize Bayesian statistical inference, the assumptions of each approach are different. POPS data indicated genetic differentiation by distance with physical and or structural barriers to gene flow among the three populations (Figure 2c).  (Wright, 1931).

| Index of association
In the index of association tests (Weir, 1979), the rBarD value is expected to be zero if populations are freely recombining (sexual reproductive mode) and greater than zero if there is association between alleles (i.e., clonality or asexual reproductive mode). The rBarD statistic is considered to be a more robust measure of association (Agapow & Burt, 2001). There was support for the hypothesis that alleles are linked across loci (p < 0.05) in the three deduced populations. These results suggest significant clonality within each of the three populations with population 3 having the highest I A value and population 1 having the lowest I A value (Table 3).

| D ISCUSS I ON
The aim of this study was to understand and compare the extent and distribution of genetic variation within pathogenic Fusarium populations, the level of population subdivision, and the interconnectivity of these and other evolutionary factors involved in population TA B L E 1 Analysis of variance within-and among deduced populations implemented in POPS), it is possible that a finer degree of substructuring as a result of higher genetic variation in genetic clusters 2 and 3 may be detected with larger sample sizes.
In this study, there was high intrapopulation genetic variation for population 1 (94.63%) which consisted largely of isolates collected from North Trinidad compared to populations 2 and 3, perhaps due to larger sample size. It was reported that for Fusarium poae isolates, genetic variability was explained by differences within rather than between Argentinean and English populations (Dinolfo et al., 2017).
Similarly, in several different studies of Fusarium graminearum populations in Canada, genetic diversity was distributed within populations and not among populations (Miedaner et al., 2001(Miedaner et al., , 2003Mishra et al., 2004). Akbar et al. (2018)  There was a significant correlation between genetic distance and geographical distance based on two models assessed in this study.
Infected seeds and/or host plants would have been distributed to different growing areas through anthropological agricultural activities. Mishra et al.(2004) suggested that highly diverse populations may arise from free movement of the pathogen as wind-borne ascospores and/or via anthropogenic exchange of infected plant material.
Re-colonization by pathotypes during successive growing seasons is related to the geographical proximity to the source population in which a given pathotype was present (Thrall et al., 2003). Rosenberg et al. (2005) noted that there are a number of variables that influence the clustering of individuals using genome-wide markers, for example, sample size, number of loci, number of clusters, assumptions about correlations in allele frequencies across populations, and the geographic distribution of samples. The relationship between genetic and geographic distance should not be an artifact of the sampling scheme, but should represent discontinuity of pairwise genetic distances of two populations on opposite sides of a structural or physical barrier, when compared with pairwise genetic distances of two populations on the same side of the same barrier (Rosenberg et al., 2005). Usually, a minority of individuals that exist in intermediate/neighboring geographic locations can have mixed membership in the main genetic clusters (Rosenberg et al., 2005). Naef and Défago (2006)  than 1 (Nm = 1.3147). According to Wright (1952), if gene flow is greater than 4, the individuals are considered to be part of a single population. Gene flow was identified as an important contributor to maintaining the high genetic diversity in populations of F. graminearum through the dispersal of sexual and asexual propagules (Gale et al., 2002;Mishra et al., 2004;Zeller et al., 2004). It is worth noting that various factors affect gene flow, for example, host availability and climatic events (environmental), reproductive, migratory, and dispersal mechanisms (biological), and modes of reproduction (genetic) (Rogers et al., 1999).

Linkage disequilibrium (LD) throughout the genome is indica-
tive of population history, the breeding system, and the pattern of geographic subdivision in populations (Slatkin, 2008). Pathogens whose survival depends on "mixed" mating/reproduction systems, that is, both sexual and asexual reproduction, tend to have high genetic diversity. The FIESC isolates in Trinidad appeared to be freely recombining as evidenced by the rbarD statistic which was less than zero for each of the three genetic clusters identified and for which population 1 had the highest genetic variation and the highest rbarD value compared to the other two populations. Mishra et al. (2003) reported that index of association data suggested a mixed mode of reproduction for Fusarium culmorum. This finding contradicts the assertion that a sexual stage of F. culmorum has never been observed (Leslie & Summerell, 2006). Sexual recombination was found to be frequent in F. graminearum populations from western Canada which contributed to high withinpopulation genetic diversity (Mishra et al., 2004). Genetic drift, selective forces acting within populations, and population admixture can also cause LD between genetic markers (Remington et al., 2001).
The implications of high evolutionary potential within a given pathogen population directly affect disease management (McDonald & Linde, 2002). High genetic diversity of populations of F. graminearum, Fusarium pseudograminearum, and F. culmorum enabled greater adaptive flexibility of these pathogens exemplified by population shift from F. culmorum to F. graminearum in the Netherlands (Waalwijk et al., 2003), the United Kingdom (Jennings et al., 2004), northern Germany (Miedaner et al., 2008), and in the western provinces of Canada (Clear & Patrick, 2019;Mishra et al., 2004Mishra et al., , 2006. Changes in pathogenicity have been reported in relation to a shift from DON producers to a higher proportion of NIV-producing F. graminearum and F. culmorum in Europe (Waalwijk et al., 2003). Fusarium solani was reported to be an important pathogen of bell pepper fruit in Trinidad (Ramdial & Rampersad, 2010) but has since been largely replaced by the now predominant FIESC with demonstrated higher genetic variation and pathogenicity.

| Isolate collection
Bell pepper fruits that were symptomatic of Fusarium fruit rot (Ramdial et al., 2017) were collected from the main pepper production areas in Trinidad. This involved a country-wide survey that was carried out from 2014 to 2017 (Table 1; Figure 1). In total, 79 FIESC isolates were collected and their identity was confirmed by multilocus sequence comparisons in a separate study (Villafana & Rampersad, 2020). There were no a priori assumptions about the number of populations or represented by these Trinidad isolates.
One isolate (79) was identical to isolate 78 and thus was not included in tests of genetic diversity, AMOVA, and association for which the dataset consisted of 78 isolates.

| ISSR profiling
Total genomic DNA was extracted from single-spore cultures of FIESC isolates grown in potato dextrose broth for 7 days in the dark using the Maxwell-16 ® automated DNA extraction kit in accordance with the manufacturer's instructions (Promega Corporation).
Ten FIESC isolates were used in a preliminary screen to determine those ISSR primers that enabled generation of a high number of polymorphic (>100) and reproducible markers (identically sized bands in two experiments) that could be used to generate polymorphic DNA fragments for all isolates. Twenty-two ISSR primers (Integrated DNA Technologies Inc.) were screened which included an assessment of optimal primer annealing temperature by gradient annealing temperature analysis (Tables S1 and S2). These primers were also used in another study to determine the genetic variability of Colletotrichum species in Trinidad (Rampersad, 2013

| Data analysis
The assignment of each isolate to a particular population or genetic cluster was carried out STRUCTURE v. 2.3.4 (Pritchard et al., 2002) which applies a Bayesian clustering approach using Markov Chain No predefined population was assumed. Bayesian inference enables identification of genetic and geographical discontinuities in mixed populations (François et al., 2006;Guillot et al., 2005). Different localities are plotted as spatial domains, and populations occupy a particular subdomain under the assumption that each subdomain is connected in space using convex polygons; as such, a Voronoi mosaic is produced. To determine the optimal number of populations, K, the deviance information criterion (DIC) was compared to determine the best fit of a set of Bayesian hierarchical models (Spiegelhalter et al., 2002).
GenAlEx version 6.5 (Genetic Analysis in Excel (Peakall & Smouse, 2006)) was used to produce a genetic distance matrix for which analysis of molecular variance (AMOVA) was carried out.
Principal coordinates analysis (PCoA) was also conducted to visualize separation of the isolates into discrete genetic clusters.
An unrooted, neighbor-joining (NJ) dendrogram was constructed to represent the genetic distances among the population using MEGAX (Molecular Evolutionary Genetic Analysis software (Kumar et al., 2018)). Bootstrapping was carried out in PAST 4 (Paleontological statistics (Hammer et al., 2001)). Nei's gene diversity (h) (Nei, 1973), allele frequencies, and pairwise comparisons of genetic diversity for each inferred population were determined using POPGENE version 1.32 (Yeh et al., 1999) with 1,000 replicates. Isolation by distance among populations was evaluated using Mantel tests between log-transformed geographic distance and genetic distance in GenAlEx 6.5. Significance of the correlation was tested assessing the p-value from 999 permutations.

CO N FLI C T O F I NTE R E S T
The author declares no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
Upon acceptance of this manuscript, all genotyping data will be archived and made available in the Dryad data repository (https://doi. org/10.5061/dryad.w6m90 5qpm).