Distribution of active ingredients of a commercial aflatoxin biocontrol product in naturally occurring fungal communities across Kenya

Distributions of the active ingredients of the aflatoxin biocontrol product Aflasafe KE01 across the most vulnerable portions of Kenya were determined with 17 SSR loci. The four active ingredient haplotypes were all among the top 10 most frequent and each belonged to both a distinct vegetative compatibility group and a distinct clonal lineage containing greater than 25 haplotypes. The data provide baselines for assessing long‐term influences of biocontrol applications in highly vulnerable production areas of Kenya.


Introduction
Several species in Aspergillus section Flavi produce aflatoxins (highly toxic, carcinogenic fungal metabolites) during infection of host crops. Aflatoxin contaminates 25% of world food crops, and 4.5 billion people in developing countries, especially in Africa and Asia, are chronically exposed to aflatoxins through consumption of staple crops (Williams et al., 2004). Maize is the primary staple in Kenya, a major 'hotspot' for aflatoxin  with exposure frequently documented as aflatoxin B1-lysine adduct in serum of children (Gong et al., 2012) and aflatoxin M 1 in mother's milk . Indeed, Kenya repeatedly experienced epidemics of lethal aflatoxicosis since 1981 (Liu and Wu, 2010;WHO, 2011) and nearly 317 cases of acute aflatoxicosis with over 125 deaths occurred in 2004 in the Coastal and Eastern provinces of Kenya (Azziz-Baumgartner et al., 2005). In a 2010, human serum aflatoxin levels in Kenyans were among the highest ever recorded (Obura, 2013) and a cross-sectional serosurvey documented a widespread urgent need to implement interventions to minimize aflatoxin exposure (Yard et al., 2013).
The Centers for Disease Control and Prevention (CDC) and the Kenya Ministry of Public Health and Sanitation focused on efforts to reduce aflatoxin exposure . Biocontrol of aflatoxin-producing fungi is a promising method for reducing crop contamination (Probst et al., 2011). Atoxigenic strains of A. flavus are used to alter natural fungal communities with nonaflatoxin-producing (atoxigenic) A. flavus so that aflatoxin producers are less common and the average aflatoxin-producing potentials of fungal communities are reduced (Cotty et al., 2008;Bandyopadhyay et al., 2016). This change in the fungal community results in greatly reduced aflatoxin concentrations in crops. Atoxigenic strain-based biocontrol has been successfully used in the United States for over two decades in a variety of crops, such as cotton (Cotty, 1994a,b;Cotty et al., 2007), groundnut , pistachio (Doster et al., 2014) and maize . Influences of atoxigenic biocontrol products extend beyond the season of treatment as a result of survival of the atoxigenic active ingredients in soils Bandyopadhyay et al., 2016). Recently, biocontrol technology has been adapted to reduce aflatoxin in maize and groundnut in Kenya and several other African nations through a partnership between the International Institute of Tropical Agriculture (IITA), the USDA-Agricultural Research Service (USDA-ARS) and several National Agricultural Research Services including the Kenya Agricultural and Livestock Research Organization (KALRO) (Bandyopadhyay and Cotty, 2013;Bandyopadhyay et al., 2016).
Atoxigenic A. flavus-based aflatoxin biocontrol products directed at various regions in Africa are named Aflasafe TM with a country-specific suffix. Each Aflasafe product contains atoxigenic strain active ingredients endemic to the target nation. Aflasafe KE01 TM (hitherto called KE01) is registered with the Pest Control Products Board (PCPB) of Kenya as a Kenya-specific aflatoxin biocontrol product. This product contains four distinct atoxigenic A. flavus Lmorphotype active ingredients (C6E, E63I, R7H and R7K; Bandyopadhyay and Cotty, 2013;Bandyopadhyay et al., 2016;Moral et al., 2020). These active ingredients were isolated from maize produced in Kenya from 2004 to 2006 with the intent of selecting effective fungi well adapted to maize agroecosystems in Kenya (Probst et al., 2011). However, natural distributions of these selected genotypes in Kenyan agroecosystems beyond the grain from which they were isolated are unknown. Data on distribution of KE01 active ingredients prior to initiation of commercial treatments are necessary to quantify extents of long-term, cumulative benefits in treated regions of Kenya.
Natural populations of A. flavus are composed of hundreds of vegetative compatibility groups (VCGs) (Bayman and Cotty, 1991;Cotty et al., 1994). Vegetative compatibility groups are defined by a heterokaryon incompatibility system that allows anastomosis and gene flow among closely related genotypes but restricts anastomosis among dissimilar individuals Cotty, 2010, 2015). Aspergillus flavus VCGs evolve as clonal lineages (Papa, 1986;Leslie, 1993;Ehrlich et al., 2007;Grubisha and Cotty, 2010). Vegetative compatibility analyses (VCA) are used to track and characterize A. flavus active ingredients (Bayman and Cotty, 1993;Cotty, 1994a,b;Atehnkeng et al., 2014;Agbetiameh et al., 2020;Senghor et al., 2020). However, VCA are laborious culture-based techniques, and each VCG must be assayed independently with the workload multiplying by the number examined. Vegetative compatibility analyses also require maintenance of living, functional tester mutants for each VCG (Bayman and Cotty, 1993;Horn et al., 1996;Probst et al., 2011), and these mutants must be shared among laboratories wishing to identify or monitor active ingredients. In this regard, DNA technologies can be advantageous (Shenge et al., 2019).
In the present study, 17-SSR loci distributed across all eight chromosomes of A. flavus were used to determine distributions of the four active ingredients of KE01 in soils under maize cultivation in seven counties across Kenya. An objective of this study was to determine relationships of the active ingredient SSR haplotypes to VCA used to monitor the active ingredients in Kenya. In addition, this work sought to characterize natural frequencies and distributions across fungal communities in Kenyan agroecosystems with high incidences of aflatoxin contamination prior to initiation of aflatoxin management programmes based on the biocontrol product. During the process, genetic lineages containing each of the active ingredients were characterized.

Results
Diversity of potential aflatoxin-producing Aspergillus section Flavi A total of 7189 Aspergillus section Flavi fungal isolates were recovered from 629 maize field soils during the , 2020) from a high aflatoxin-producing unnamed species with S morphology associated with lethal aflatoxin contamination of maize produced in Kenya (Probst et al., 2007;Probst et al., 2012) and 986 A. parasiticus ( Table 1). Each of these species was detected in all ten of the sampled agricultural areas across Kenya. However, LAF and the L morphotype of A. flavus were the most common. Similar frequencies of LAF occurred in all areas except Area-8 (Tana River County), where LAF was rarely detected (Table 1).

SSR amplification, genotyping and clone correction
A total of 2744 isolates were clearly amplified at 17-SSR loci and included in this analysis (Table 2). SSR amplification across these isolates was free of PCR artefacts and only generated single peaks in the expected size range. There were no missing data and no null alleles. The number of alleles per locus was high and ranged from 9 to 72. High allelic diversity resulted in high genotypic diversity, with 2140 unique SSR haplotypes (Table 2). Isolates were culled to only allow a single isolate with any given SSR haplotype from each soil. After this clone correction, 2408 isolates remained for analysing frequencies, distribution, adaptation and lineage relationships of the four KE01 active ingredients within naturally occurring fungal populations in Kenyan soil.

Relationships between SSR haplotype and VCG of KE01 isolates
The four KE01 active ingredients (C6E, E63I, R7H and R7K) were separated by both VCG analysis and SSR typing ( Table 3). Haplotypes of 24 soil isolates matched the R7K haplotype, and nitmutants of each of these isolates complemented an auxotrophic tester mutant of R7K but not any of the six tester mutants developed for the other three KE01 active ingredients. Complementation of an R7K tester mutant indicates membership in the R7K VCG. In a similar manner, all isolates matching the haplotypes of C6E (n = 15), E63I (n = 3) and R7H (n = 10) were shown to belong to the same VCG as the respective active ingredient. In addition, isolates differing from C6E (n = 5), E63I (n = 13) and R7H (n = 3) at only  c.
Number of isolates identified as Aspergillus flavus L morphotype.

d.
Number of isolates belonging to the Kenyan S morphotype referred to as the Lethal Aflatoxicosis Fungus (LAF; Kachapulula et al., 2017;Singh et al., 2020) which has been associated with contamination events that led to deaths in Kenya (Probst et al., 2007 (Table 3). In no case did any of the active ingredients react with a tester mutant developed from a different active ingredient.

Distributions of KE01 haplotypes
In total, 118 haplotypes occurred in 2 to 23 soil samples (data not shown). The haplotypes of each of the four KE01 active ingredients were among the first ten most frequently occurring haplotypes (Table 4). These haplotypes occurred in multiple areas and both seasons. The haplotype of active ingredient R7K (H-1462) was detected in 3.8% (23 total) of soils and was the most widespread haplotype (occurring in eight of ten study areas) in both long rains and short rains seasons (Table 4). Haplotypes belonging to two other active ingredients (C6E and R7H) were found in 2.4% and 1.8% of the soils, respectively, and were both found in five areas and both seasons (Table 4). H-1019, the haplotype of the fourth KE01 (E63I), was detected in only 3 soils (0.6%) from two areas. However, H-1019 differs from H-1017 at only a single locus and at that locus by only a single SSR repeat. H-1017 belongs to the same VCG as H-1019, and H-1017 was the fourth most commonly detected haplotype and occurred in 13 soils from multiple areas and both seasons (Tables 3 and 4).

Distribution of closely related lineages of Aflasafe TM KE01
The eBurst analysis identified 246 groups (clonal lineages) of closely related haplotypes based on a user-defined criteria (differing at no more than 3 of 17 loci and > 80% genetic similarity). The four KE01 isolates were each included in the top ten largest clonal groups containing 26 to 45 haplotypes distributed across multiple areas (7 to 9 areas; Fig. 2). However, no member of any of the clonal groups of the four KE01 isolates was found in Area-8 (Fig. 2). The clonal groups of R7K, C6E, R7H and E63I were found in 7.2%, 4.1%, 4.5% and 5.1% of the soil samples respectively (Fig. 2). In total, clonal lineages of the active ingredients occurred within 21% of examined soils.

Discussion
The current results agree with previous observations that the A. flavus L morphotype is widely distributed in agricultural soils across Africa and including Kenya (Donner et al., 2009;Probst et al., 2014; Table 1). Populations of S-morphology fungi in Africa and North America are distinct (Cotty and Cardwell, 1999). Although the A. flavus S morphotype is common in the USA (Singh et al., 2020), it either very rare or does not occur in Kenya (Probst et al., 2007. However, there are common S-morphology fungi in Kenya, sometimes referred to as LAF (LAF = Lethal Aflatoxicosis Fungus; Kachapulula et al., 2017;Singh et al., 2020) that are phylogenetically distinct from A. flavus but still consistently produce high concentrations of aflatoxins in maize and were associated with lethal aflatoxicoses occurring in Kenya from 2004 to 2006 (Probst et al., , 2014. The LAF is common in regions with histories of lethal aflatoxicosis, including Machakos, Makueni and Kitui counties (Probst et al., 2007(Probst et al., , 2010). In the current study, LAF was only infrequent in Area-8 (Tana River County; Table 1) which is primarily the Tana River Irrigation Scheme near the town of Bura. Another high aflatoxin-producing species, A. parasiticus, composed more than 10% of the aflatoxin producers in four areas but was also not detected in Area-8. A. parasiticus and LAF may not be a. Area locations are illustrated in Fig. 1 well adapted to the agroecology of Area-8, the only area where maize is produced primarily under furrow irrigation. Factors dictating geographic distributions of LAF and A. parasiticus need further investigation. Where these fungi occur, applications of KE01 are expected to alter their frequencies downwards. The current data provide baselines with which to assess extents to which such alterations occur across Kenya. The observed high allelic diversity (Table 2) resulted in a d over 2000 haplotypes across the study populations. This indicates an exceptionally large population in Kenya soils because SSR allelic and haplotype diversity is a function of population size, stability and evolution . High diversity of the A. flavus population in Kenya soil can also be measured with VCG diversity Cotty, 1991, 1993;Atehnkeng et al., 2016;Ortega-Beltran and Cotty, 2018). A. flavus VCGs in natural populations can evolve independently for long periods (Ehrlich et al., 2007;Grubisha and Cotty, 2010), and each VCG can be considered an individual genetic lineage delineated by multiple heterokaryon incompatibility loci (Leslie, 1993). Multiple VCGs may occur within a single seed or pinch of soil (Bayman and Cotty, 1991;Mehl and Cotty, 2013). Diversity of A. flavus VCGs in soil populations is high in established agricultural settings in warm climates (Bayman and Cotty, 1991;Horn, 2003), including Kenya (Probst et al., 2011). It is expected that there are fewer VCGs than SSR haplotypes because each VCG typically contains multiple, closely related SSR haplotypes (Grubisha and Cotty, 2010;Ortega-Beltran and Cotty, 2018).
Aspergillus flavus is primarily recognized as an asexual fungal species. Under clonal reproduction, identical haplotypes displayed by isolates from two or more soils (repeated haplotypes) are expected. Although a large number of unique haplotypes were detected because of the exceptionally large and ancient population of A. flavus in Kenya soils , only 118 haplotypes occurred in two or more soils (data not shown). This relatively low number of repeated haplotypes may be a result of how difficult it is to thoroughly sample such a diverse population. Each of the four active ingredients of the biocontrol product has a distinct SSR haplotype and a distinct VCG. Results from the SSR assay were in complete agreement with VCG analyses used to track other atoxigenic A. flavus and, as such, this assay is also a reliable tool (Cotty, 1994a,b;Atehnkeng et al., 2014;Ortega-Beltran and Cotty, 2018;Agbetiameh et al., 2019). Vegetative compatibility groups are clonal lineages (Papa, 1986;Leslie, 1993), and linkage equilibrium of SSR loci occurs among isolates within the same VCG, possibly indicating an active parasexual cycle, but not among isolates in different VCGs (Grubisha and Cotty, 2010;Grubisha and Cotty, 2015; Ortega-Beltran  . Over long periods, mutation and clonal evolution may cause diversity at SSR loci within VCGs . This is evident in the current data set with three of the active ingredients. The VCGs of those active ingredients include one to three haplotypes different at one SSR locus (Table 3). Indeed, the fourth most common haplotype belongs to the VCG of active ingredient R7K, again suggesting better adaptation to Kenya's agroecosystems than most VCGs not included in KE01. Soil collections for this study were undertaken 6-7 years after harvest of the maize from which the active ingredients of the biocontrol product were collected (Probst et al., 2011) indicating persistence of these clonal lineages over long periods. The high frequencies, wide distributions (Table 4) and long persistence of the KE01 haplotypes and their close relatives (clonal lineages) across the study areas suggest that these genotypes are well adapted to Kenyan agroecosystems. Although haplotypes of the KE01 active ingredients are relatively common, and natural presence of these atoxigenic genotypes likely positively influence crop contamination (Cotty et al., 2008), the background levels of these genotypes is far below what would be required to prevent unhealthy levels of aflatoxins. Therefore, biocontrol treatments with KE01, which are expected to result in over 80% frequencies of the active ingredients on the crop , will likely provide benefit. The current work provides reference points from which to judge both singleseason and long-term efficacies across portions of Kenya most in need of aflatoxin mitigation. Area-1 c Area-2 Area-3 Area-4 Area-5 Area-6 Area-7 Area-8 Area-9 Area-10 S1 d S2 e S1 S2 S1 S2 S1 S2 S2 S1 S2 S1 S2 S1 S2 S2 H-1462 (R7K) a 2 3 3 3 3 1 1 4 0 Haplotype of one of the four active ingredients of Aflasafe KE01, a registered biopesticide for aflatoxin management in Kenya. H = haplotype; unique haplotypes were numbered H0001 to H2710. Haplotypes were identified based on 17-multilocus SSR loci. Only the ten most common haplotypes across both seasons and all areas are included. Numbers in columns indicate the number of isolates with the indicated haplotype in each area (Area 1 to 10) and season (S1 & S2). b. N = total number of isolates overlapped with same haplotypes. c. Locations Area 1 through Area 10 are indicated in Fig. 1. d. S1, Long Rain Season (May 2012). e. S2, Short Rain Season (November 2012). Certain VCGs and SSR haplotypes of A. flavus are naturally widely distributed. For instance, several VCGs of A. flavus isolated from agricultural soils occur across a large section of the United States (Horn andDorner, 1998, 1999;Ehrlich et al., 2007;Cotty, 2010, 2015;Ogunbayo et al., 2013) and others occur across Africa (Ogunbayo et al., 2013). Dispersal of A. flavus is consistent with the production of large quantities of airborne conidia (Bock et al., 2004) and association with both insects (Stephenson and Russell, 1974) and human transported crop materials (Garber and Cotty, 2014;Shenge et al., 2019). Detection of the KE01 active ingredients in all areas except Area-8 (Tana River County) of the Tana River Irrigation Scheme near Bura is most likely due to either physical barriers to dispersal (i.e. stretches of arid land) or requirements for adaptive success differing among areas. Different requirements for niche residence in Area-8 may cause high frequencies of locally restricted haplotypes . In some cases, production areas with distinct agroecologies or aflatoxin aetiology may warrant the development of targeted biocontrol agents.
In conclusion, high genotypic diversity in the A. flavus population resident in Kenya is attributable to the size (very large) and age (ancient) of that population . The wide distribution and persistence of SSR haplotypes closely related to the four active ingredients of KE01, especially 6-7 years after initial isolation, indicate that these fungal genotypes and genetic lineages have been growing in and adapting to Kenyan agroecosystems for long periods. Thus, the KE01 active ingredients can be expected to perform and compete well against resident aflatoxin producers. The results of the current study suggest that these active ingredients are excellent selections for a Kenya-specific aflatoxin biocontrol product for which long-term efficacy, persistence and cumulative benefits are desired.

Sample collection
Soils from agricultural fields cropped to maize were sampled in seven counties (Embu, Machakos, Makueni, Kitui, Tana, Homabay and Migori) located across the regions of Kenya most vulnerable to aflatoxin contamination, as previously described . Sampling was performed during both the long rains (late April to Early June) and the short rains (November to December) seasons of 2012 (Table 1). The two seasons are separated by approximately six months. Five counties (Tana River, Makueni, Kitui, Machakos and Embu) were sampled just prior to the long rains season (May 2012), and six counties (Embu, Machakos, Makueni, Kitui, Homabay and Migori) were sampled just prior to the short season rains (November 2012) (Table 1). Fields were geo-referenced prior to sampling. Samples were collected from 629 fields across ten agricultural areas (Fig. 1), where 390 fields were sampled during the long rains season and 239 fields during the short season rains ( Fig. 1 and Table 1) following a previously described method (Cotty, 1997). Briefly, in each field, a 30 to 40 m transect was sampled at 3-to 4-m intervals. At each point, four small scoops of soil (1 to 3 g each) were collected to a depth of 2 cm and combined with samples from other points. These pooled samples were homogenized and stored dry at room temperature. All samples were shipped to the USDA-ARS laboratory at the School of Plant Sciences of the University of Arizona, Tucson, under a permit from USDA-APHIS (permit to move live plant pests, noxious weeds, and soil). The active ingredients of KE01 (C6E, E63I, R7H and R7K) belong to the L morphotype and were originally isolated in the USDA-ARS laboratory at the University of Arizona from maize produced in Kenya during the lethal aflatoxicosis outbreaks of 2004 through 2006 (Probst et al., 2011).

Isolation of Aspergillus section Flavi fungi
Soil samples were subjected to the dilution plate technique on the semi-selective medium modified Rose Bengal agar (M-RB) (Cotty, 1994a,b). Briefly, 5 to 10 g of soil was suspended in 250 ml sterile distilled water with 0.1% TWEEN-80 solution by stirring for 20 min at 200 rpm. Aliquots of the resulting suspension were spread on three M-RB plates (200 µl per plate). When necessary, adjustments to aliquot volume and soil quantity were made to obtain no more than 12 Aspergillus section Flavi colonies per plate. After incubation at 31°C for three days, Aspergillus section Flavi colonies were subcultured onto 5/2 agar [(5% V-8 juice (Campbell Soup Company, Camden, NJ), 20 g l À1 bacto-agar, pH 6.0] at 31°C (Cotty, 1989). A total of 10 to 15 Aspergillus section Flavi isolates were cultured from each soil sample, and specific fungal species or morphotypes were identified on the basis of colony characteristics and spore morphology. All isolates belonging to the A. flavus L morphotype (Table 1) were included for further analyses. Cultures were stored as 3-mm plugs of 5-day-old cultures on 5/2 agar in sterile distilled water at 4°C for further use. For long-term storage, select cultures were stored on silica gel. The active ingredients (C6E, E63I, R7H and R7K) of KE01 were cultured from silica storage maintained at the Tucson location. One piece of silica was subcultured on 5/2 agar as previously (Cotty, 1989). Sporulating cultures (5-7 days old) were stored as 3-mm plugs on 5/2 agar in sterile distilled water at 4°C until further use.

DNA extraction and SSR amplification
DNA was extracted from active ingredients (C6E, E63I, R7H and R7K) of KE01, and all A. flavus L-morphotype isolates recovered from soil samples (Table 1). A previously described protocol was followed for DNA extraction (Callicott and Cotty, 2015). Briefly, isolates were plated onto 5/2 agar (Cotty, 1989) and incubated at 31°C for 8-10 days. Conidia were then suspended in 500 µl lysis buffer and incubated with agitation in a Thermomixer 5436. DNA was ethanol precipitated, dissolved in sterile water, quantified with a spectrophotometer (Nanodrop, Wilmington, DE) and stored at À20°C until use.
Variability was determined at 17 SSR loci distributed across all eight A. flavus chromosomes, as previously described . Briefly, five multiplex PCR panels optimized for high-throughput SSR amplification and genotyping. Multiplex PCRs were carried out using 0.08 µmol l À1 of each primer, 1X AccuStart II PCR SuperMix (Quanta Biosciences, Gaithersburg, MD, USA) and 5 ng of genomic DNA in a final reaction volume of 10 µl. PCR conditions were as follows: 94°C for 1 min, 26 cycles of 94°C for 30 s, 57°C for 90 s, 72°C for 30 s and 30 min at 60°C. PCR amplicons were sized at the University of Arizona Core Genomics Laboratory with the LIZ500 standard and an ABI 3730 analyzer (Applied Biosystems, Waltham, Massachusetts, USA). Allele sizes were called with GENEMARKER version 2.6 (SoftGenetics LLC, State College, Pennsylvania, USA).

SSR genotyping and clone correction
Allelic data from 17-SSR markers were combined for each of the study isolates, and haplotypes (genotypes) were determined using the program HAPLOTYPE-ANA-LYSIS V 1.04 (Eliades and Eliades, 2009). In order to construct a clone-corrected data set for each sample, each haplotype was only included in the dataset once for each soil sample (Table 2). Clone-corrected data sets were used as the basis for further analysis.

Analysis of vegetative compatibility group (VCG)
In order to test whether isolates with an SSR haplotype matching one of the active ingredients of KE01 belonged to that active ingredient's VCG, isolates were typed using VCA based on complementation of nitrate non-utilizing auxotrophs (Bayman and Cotty, 1991). All isolates with haplotypes either matching an active ingredient of KE01 or closely related to an active ingredient's haplotype (differing at only one SSR locus) were typed with VCA. Briefly, isolate auxotrophs (nitmutants) were generated for each isolate on chlorate containing media (Cotty, 1994a,b). Complementary tester auxotrophic mutants (one cnxand one niaD -) from each of the four active ingredients of KE01 were previously developed (Probst et al., 2011). A nitrate auxotroph (nit -) from each isolate was paired with both tester auxotrophs. Isolate auxotrophs that complemented one or both tester auxotrophs were classified as belonging to the VCG defined by that tester pair.

Analysis of genotypic diversity and distributions
Haplotypic diversity, frequencies and distributions were estimated for both cropping seasons (long and short rains) for the year 2012, both individually and together. Area-wide haplotypic diversity, including the number of haplotypes and frequencies of repeated haplotypes displayed by two or more isolates, was measured using the program haplotype-analysis V 1.04. Frequencies of the most common haplotypes across the study areas were statistically tested with analysis of variance (ANOVA) based on the general linear model procedure in sas (version 9.2; SAS Institute, Cary, NC). Mean separations were performed on data from tests with statistically significant differences (P = 0.05) using Tukey's honestly significant difference test (Pagano and Gauvreau, 2000).

Analysis of closely related lineages of the biocontrol product
Closely related lineages (closely related haplotypes are presumed to have evolved through clonal reproduction and mutation) of the active ingredients of KE01 were identified based on principles used for determining clonal lineages of the asexually reproducing bacteria with the program eburst v3 (http://eburst.mlst.net/) (Feil et al., 2004). A user-defined group definition was set to include those groups that shared identical genotypes for at least 14 of the 17 SSR loci.