Compositions of Aspergillus flavus populations determine the extent to which crops become contaminated with aflatoxins. In the current study, influences of diverse crop hosts on competition among A. flavus isolates were quantified with pyrosequencing. Maize, cotton, soyabean and sorghum supported different levels of sporulation, but intraspecific differences in sporulation were not detected on any host. However, hosts differentially influenced competition during infection, allowing greater sporulation by some isolates and increased host tissue invasion by others. Furthermore, competitive interactions during host invasion did not predict isolate success during sporulation. Isolates were similarly competitive on maize and sorghum, the two most closely related hosts. Host-specific influences on intraspecific competition may dictate compositions of A. flavus populations and, as a result, the severity of aflatoxin contamination. Host factors should be considered when designing and implementing aflatoxin management strategies including biocontrol with atoxigenic strains.
Aflatoxins, highly carcinogenic secondary metabolites produced by Aspergillus flavus and closely related fungi, frequently contaminate crops including maize, cottonseed, peanuts and tree nuts (Diener et al., 1987). Aflatoxin levels in food are regulated by most nations (e.g. in the USA levels cannot exceed 20 ppb), resulting in significant economic losses for growers of susceptible crops (Robens & Cardwell, 2003). Where regulations are lacking or inadequately enforced, aflatoxin contamination has significant impacts on public health. In Kenya consumption of highly contaminated maize has resulted in acute aflatoxicosis and death (Probst et al., 2010). Isolates of A. flavus produce variable quantities of aflatoxins from none (atoxigenic) to over 106 ppb (Probst & Cotty, 2012). Individual plants (Bayman & Cotty, 1991), fields (Bayman & Cotty, 1991; Pildain et al., 2004) and regions (Horn & Dorner, 1999; Cardwell & Cotty, 2002) harbour diverse populations of A. flavus, and the compositions of A. flavus communities influence frequencies and severities of aflatoxin contamination events (Cotty et al., 2008). For example, incidence of the highly toxigenic S morphotype of A. flavus is positively correlated with crop aflatoxin levels (Probst et al., 2010). Conversely, proportions of crop-associated fungal communities consisting of atoxigenic A. flavus are negatively correlated with aflatoxin content, and atoxigenic strains are effective biopesticides for reducing contamination (Cotty et al., 2008).
Diversity among and within A. flavus populations is evident in distributions of both phenotypes and genotypes. Two morphotypes designated the L and S strains differ from each other in production of conidia and sclerotia (Cotty, 1989). Morphotypes are further subdivided into vegetative compatibility groups (VCGs) by a heterokaryon incompatibility system; isolates able to form stable heterokaryons are considered members of the same genetic lineage (Bayman & Cotty, 1991; Grubisha & Cotty, 2010). DNA sequence analyses support the notion that members of the same VCG are closely related and genetically distinct from other VCGs (Grubisha & Cotty, 2010). Phenotypic characteristics such as aflatoxin-producing ability and morphology are generally conserved within a VCG (Bayman & Cotty, 1993; Novas & Cabral, 2002; Pildain et al., 2004). Characteristics of A. flavus populations, including frequencies of morphotypes (Cardwell & Cotty, 2002; Probst et al., 2010), average aflatoxin-producing potential (Cotty, 1997; Horn & Dorner, 1999) and pectinase production (Mellon et al., 2007) vary by region. Within regions, quantities and frequencies of morphological and genetic types (Horn & Dorner, 1998; Jaime-Garcia & Cotty, 2010) and proportions of the A. flavus community consisting of aflatoxin-producers (Vaamonde et al., 2003) vary among crop hosts, suggesting host species have differential influences on the structure of A. flavus populations.
In addition to infecting diverse domesticated (Diener et al., 1987) and wild plant species (Boyd & Cotty, 2001), A. flavus is a pathogen of animals including humans and insects (Mellon et al., 2007). Though individual isolates vary in production of hydrolases needed for virulence (Mellon et al., 2007), host origin is not correlated with ability to infect plants or animals (St Leger et al., 1997, 2000). The ability of A. flavus to produce a broad spectrum of degrading enzymes and use a wide variety of hosts and non-living substrates suggests it is an opportunistic pathogen with a non-specific host range (St Leger et al., 1997; Mellon et al., 2007). However, competition during host infection may influence the structure of A. flavus populations. This is supported by variation among crops in the composition of associated A. flavus genotypes and phenotypes (Horn & Dorner, 1998; Vaamonde et al., 2003; Jaime-Garcia & Cotty, 2010). Previously, isolates from 38 VCGs varied in ability to compete during maize kernel infection (Mehl & Cotty, 2010). The competitive abilities of several fungi vary by host genotype (Wille et al., 2002; Zhan et al., 2002), but data on the extent to which competitive interactions among A. flavus genotypes are influenced by host species are lacking.
DNA-based methods for genotype identification and quantification have potential to provide new insights into A. flavus population biology. Dissecting the population dynamics of A. flavus has been limited by the extent to which morphologically indistinct genotypes can be distinguished. Recently, quantitative pyrosequencing assays have been developed to quantify proportions of A. flavus genotypes within fungal populations (Das et al., 2008; Mehl & Cotty, 2010, 2011). The quantitative pyrosequencing method, based on a technology similar to but distinct from whole genome 454 pyrosequencing, uses sequencing through synthesis to provide highly accurate measurements of the frequency of single nucleotide polymorphisms (SNPs) in pools of DNA (Lavebratt & Sengul, 2006). Using this technique on SNPs associated with specific genotypes allows dissection of intraspecific interactions during crop infection (Mehl & Cotty, 2010, 2011). Quantitative pyrosequencing examines the entire ‘population’ of A. flavus within pools of DNA rather than the limited subset of isolates possible with culture-based methods. This results in both a higher level of precision and greater potential to detect rare genotypes. Although quantitative pyrosequencing measures SNP proportions rather than absolute quantities of DNA, it has a greater level of precision and sensitivity for quantification of genotype frequencies than real-time quantitative PCR (Mehl & Cotty, 2010).
The current study sought to use quantitative pyrosequencing to dissect crop-specific influences on intraspecific interactions among A. flavus genotypes during host infection. This study provides the first empirical evidence that specific plant hosts influence outcomes of competitive interactions among A. flavus genotypes and, in so doing, frequencies of specific aflatoxin-producing and related fungi associated with crops. The results are discussed in relation to both the epidemiology of crop aflatoxin contamination and the optimization of aflatoxin biocontrol with atoxigenic strains of A. flavus.
Materials and methods
Eight A. flavus isolates varying in aflatoxin production and competitive ability on maize were used in the current study (Tables 1 and 2). Isolates were characterized and described previously (Mehl & Cotty, 2010). To ensure genotypic diversity among the isolates included, each belonged to a separate VCG. Wildtype isolates from silica gel storage were cultivated on 5/2 agar (5% V8 juice, 2% agar, pH 5·2). VCGs were verified by complementation of nitrate non-utilizing auxotrophs derived from wildtypes with tester pairs (cnx and niaD mutants) corresponding to previously identified VCGs (Cotty, 1994). Single-spore purified wildtype isolates were maintained in 4-mL vials by suspending plugs of 5/2 agar with abundant sporulation in sterile distilled water. Conidial suspensions from water vials were used to centrally seed 5/2 agar. Following incubation (31°C, 7 days), conidia were dislodged from plates with sterile cotton swabs and suspended in sterile distilled water. Turbidity of suspensions was measured using a turbidity meter (Turbidimeter, Orbeco Analytical Systems), and a nephelometric turbidity unit (NTU) versus colony-forming unit (CFU) standard curve was used to calculate the number of conidia (CFU = NTU × 49 937). Conidial concentrations were adjusted as described below.
Table 1. Characteristics of isolates and single nucleotide polymorphisms (SNPs) used to distinguish between Aspergillus flavus isolates in different vegetative compatibility groups (VCGs)
The effects of isolate and host on sporulation and seed aflatoxin concentrations were assessed with a factorial anova. Two independent tests with similar results were combined for analyses (total n = 6). Means followed by the same letter are not significantly different (P >0·05) by Tukey's studentized range test. Letters before the comma denote differences among isolates (columns); letters following the comma denote differences among hosts (rows).
Quantities of conidia averaged across hosts; significant differences were not detected (P = 0·51).
Aflatoxin concentration averaged across hosts; isolates differed significantly (P <0·0001).
Conidia quantity or aflatoxin concentration averaged across isolates; hosts differed significantly for both parameters (P <0·0001).
The success of eight genetically distinct A. flavus isolates, both individually and in co-inoculated pairs, was examined during host invasion and sporulation on seeds of maize (Pioneer hybrid 33B50), cotton (Upland cotton 4554 B2RF), sorghum (Sorghum Partners) and soyabean (IA1023). The crop hosts were selected because all are grown in rotations in regions of Texas where aflatoxin contamination commonly occurs (Jaime-Garcia & Cotty, 2010). Intact, uniformly sized maize, sorghum and soyabean seeds were surface-disinfected by submersion in 80°C water for 45 s. Subsamples of seeds were plated on agar and incubated to confirm embryo viability (i.e. a living host) and lack of contaminating fungi. Seeds (5 g) were placed in 250-mL Erlenmeyer flasks sealed with gas-permeable BugStopper plugs (Whatman). Seed water content was measured with a moisture balance (HB43 Halogen Moisture Analyzer, Mettler-Toledo), and conidial suspensions (5 × 104 conidia of each isolate) in the volume of water required to bring total seed water content to 25% were added to each flask. Flasks were gently shaken by hand (10 s) to evenly coat maize, sorghum and soyabean seeds with inoculum.
Ginned fuzzy cotton seeds retaining the linters were similarly disinfected, but to allow rapid and even wetting, seeds were submerged briefly in 80% ethanol prior to submersion in hot water. The linters (short fibres) coating the seed absorbed water, so cotton seeds were placed in sterile Petri dishes and dried (60°C, 24 h) to reduce moisture content to below 25%. After drying, water content was determined, and 10 μL conidial suspensions containing 2 × 103 conidia of each isolate were applied to each seed. Inoculated cotton seeds (25 seeds, c. 2·5 g) were transferred to Erlenmeyer flasks with the appropriate volume of water to bring total moisture to 25%.
To determine behaviour (i.e. sporulation and aflatoxin production) of A. flavus isolates in the absence of competition, cotton, maize, sorghum and soyabean seeds were inoculated with each of the eight isolates individually. Following incubation (31°C, 7 days), seeds were dried (60°C, 48 h) to stop fungal growth. Conidia were washed from seed surfaces by agitating seeds for 10 s in 20 mL 0·1% Tween-80 followed by agitation in 20 mL distilled water. Washings were combined and filtered through Miracloth (EMD Biosciences), and quantified by turbidity as described above. Conidial suspensions were poured into 50-mL conical tubes and centrifuged (9000 g, 5 min). The supernatant was discarded, and the spore pellet was retained for DNA isolation. Seeds were immediately dried again (60°C, 48 h) following removal of conidia and prior to analysis for aflatoxins. Single-isolate inoculations were replicated three times, and the experiment was performed twice.
To test the influence of host species on competition between A. flavus genotypes, maize, sorghum, cotton and soyabean seeds were subjected to six inoculation treatments. Previously on maize, some isolates were significantly more competitive during seed-invasion than during sporulation (‘colonizers’) and others were more competitive during sporulation than during infection (‘sporulators’) (Table 1; Mehl & Cotty, 2010). Competitive abilities were all measured during simultaneous maize infection by the evaluated isolates and a reference isolate, CG136. The first three inoculation treatments included an isolate highly competitive during maize kernel infection (EB01, WM01 or YV10; Table 1; Mehl & Cotty, 2010) paired with an isolate less competitive during infection (DV901, MR17 and CG136, respectively). A second set of three inoculation treatments sought to examine host influences on isolate competitive ability, both during host invasion and during sporulation on the infected host. To achieve this, the reference isolate (CG136), a ‘colonizer’ isolate (RB04) and a ‘sporulator’ isolate (MN902) were inoculated in all possible pairs on each host (i.e. CG136/MN902, RB04/CG136 and RB04/MN902). Following incubation (31°C, 7 days), seeds were rapidly dried to stop fungal growth (60°C, 48 h) and sporulation was quantified as described above for single-isolate inoculations. Treatments were replicated four times, and each experiment was performed twice.
Aflatoxin extraction and quantification
Seeds from both single- and paired-isolate treatments were analysed for aflatoxin content. Maize kernels, sorghum seeds, cotton seeds and soyabean seeds were pulverized for 10 s in an analytical mill (IKA Works). Aflatoxin B1 was extracted and quantified as described previously (Mehl & Cotty, 2010). Briefly, 1·5 g seed was extracted with 15 mL 85% acetone in a 22-mL glass vial with a Teflon septum. Extracts were separated on thin-layer chromatography plates (TLC) (silica gelG, 250 μm) alongside aflatoxin standards (Supelco) with H2O:MeOH:ether (1:3:96). TLC plates were visualized under 365-nm UV light, and samples negative for aflatoxin B1 were diluted with an equal volume of water and extracted twice with 5 mL methylene chloride. Extracts were combined, dried and resuspended in a volume of methylene chloride that allowed accurate quantification. Aflatoxin B1 was quantified directly on TLC plates by scanning fluorescence densitometry with a TLC Scanner 3 (Camag Scientific).
DNA from paired-isolate treatments (i.e. competition experiments) was analysed with quantitative pyrosequencing to determine isolate proportions from pools of A. flavus DNA. Proportions were quantified separately for mycelia (within seeds) and conidia. Conidial DNA was isolated from the total quantity of conidia washed from seed surfaces. Washed, conidium-free seeds were pulverized in an analytical mill as described above. Fungal DNA isolated from 200 mg subsamples of pulverized seeds was assumed to be of mycelial origin. Extractions were performed using a FastDNA SPIN Kit and the FastPrep Instrument (MP Biomedicals).
PCR and quantitative pyrosequencing
A previously developed quantitative pyrosequencing assay targeting a 233-bp portion of an intergenic region in the aflatoxin biosynthesis gene cluster (−476 to −244 upstream of omtA/aflP) (Mehl & Cotty, 2010) was used to measure percentages of isolate-specific SNPs from mixtures of A. flavus DNA (Table 1). PCR conditions for primer pair CG136-Afl-F/CG136-Afl-R were described previously. The forward primer was biotinylated and HPLC-purified.
Biotinylated PCR amplicons were prepared for pyrosequencing as described previously (Das et al., 2008; Mehl & Cotty, 2010). Amplicons (40 μL) immobilized on streptavidin-coated beads (GE Healthcare, Bio-Sciences AB) were captured on the filter probes of the Vacuum Prep Tool (QIAGEN), washed with 70% ethanol, denatured with 0·2 m NaOH, washed with buffer (10 mm Tris-acetate, pH 7·6), and released into the wells of a pyrosequencing plate containing 40 μL annealing buffer (QIAGEN) and 0·5 μm sequencing primer CG136-Afl-S2. The pyrosequencing plate was heated (90°C, 10 min) then cooled to anneal the sequencing primer to the single-stranded DNA.
Pyrosequencing was performed with a PSQ 96MA pyrosequencer using a PSQ SNP Reagent Kit according to the manufacturer's instructions (QIAGEN). The pyrosequencing reaction produces light intensities proportional to the quantity of nucleotides incorporated at each position on the single-stranded template. Percentages of each nucleotide at isolate-specific SNPs were calculated using the allele quantification option of the psqma v. 2.1 software (QIAGEN).
Experimental design and data analysis
Each experiment was performed in two independent trials with either three or four replicates per trial (see above). Treatments were arranged in a randomized complete block designs. Quantities of aflatoxin B1 and conidia were log-transformed and percentage data were arcsine-transformed for statistical analyses. Effects of host species, inoculation treatment and trial on isolate proportions from mycelia and conidia were determined with an analysis of variance. Unless otherwise noted, trial × treatment interactions were not significant, and data from the two trials were pooled for analyses. Treatment means were separated using Tukey's studentized range test. Detransformed means are reported for aflatoxin and conidia data; untransformed means and standard errors are reported for percentage data. Predicted isolate proportions over multiple generations of sporulation on hosts were calculated assuming the advantage (or disadvantage) conferred to an isolate after one cycle of sporulation remains constant over time (% isolate A = [A0 × (A1/B1)N]/[A0 × (A1/B1)N + A0] where A0 = initial % isolate A, A1 = % A after one generation, B1 = % isolate B after one generation, and N = number of generations). Correlations between dependent variables were performed on treatment means. For each isolate pair, values for the isolate more competitive during maize kernel infection (i.e. comprising greater than 50% of the total A. flavus DNA) were used in analyses. Statistical analyses were performed with sas v. 9.1 (SAS Institute).
Aflatoxin B1 production
When grown individually on each of the hosts, variation in aflatoxin B1 production was detected among isolates (P =0·002) and among host treatments (P <0·0001; Table 2). However, ranking of isolates for aflatoxin production was similar on each host (isolate × host interaction: P =0·85). Although aflatoxin-producing potential of the individual isolates varied (Table 2), differences among isolate pairs in aflatoxin B1 production were not detected in the first competition experiment (P =0·08). In the second competition experiment, aflatoxin B1 production varied among inoculation treatments (P =0·002) with RB04/CG136 producing more aflatoxin B1 (P <0·0001) than the other isolate pairs (Table 3). Maize and sorghum supported production of the highest aflatoxin B1 concentrations and soyabean the lowest (Tables 2 & 3).
Table 3. Aflatoxin B1 production by paired Aspergillus flavus isolates on four host species
Isolate pairs were inoculated and analysed in two separate experiments.
Effects of isolate pair and host species on seed aflatoxin concentrations were determined with a factorial anova. Means followed by the same letter are not significantly different by Tukey's studentized range test (P >0·05). Upper case letters denote differences among isolate pairs (columns); lower case letters denote differences among hosts (rows). Two independent tests with similar results (treatment × trial interaction, P >0·05) were combined for analyses (total n = 8).
25 A, a
18 A, a
1 A, b
0·8 A, b
25 A, a
19 A, a
1 A, b
0·7 A, b
35 A, a
23 A, a
0·6 A, b
0·2 A, b
90 W, w
54 W, wx
22 W, x
0·87 W, y
58 W, w
49 W, w
6 W, w
0·08 X, x
29 X, w
26 W, w
3 W, x
0·68 W, x
Differential host preference among A. flavus genotypes was not detected based on single-isolate inoculations
When grown individually, total numbers of conidia produced by isolates varied by host (P <0·0001), with sorghum and maize supporting the highest levels of overall sporulation followed by soyabean and cotton (Table 2). The eight isolates produced similar quantities of conidia (P =0·53), and there was no host × isolate interaction (P =0·19).
Hosts differentially influenced outcomes of competition
Although innate ability to sporulate on a particular host did not vary among isolates, host species influenced success of competing isolates, both during invasion of host tissues and during sporulation (Figs 1, 2). Percentages of DNA attributable to each isolate in both conidia and mycelia were precisely determined by measuring proportions of isolate-specific SNPs with quantitative pyrosequencing (i.e. SE mycelia: 1–7%; SE conidia: 1–3%). An isolate was considered more competitive than its co-infecting isolate if it comprised significantly more than 50% of the total A. flavus DNA. The three most competitive isolates during maize kernel infection (EB01, WM01 and YV10) in a previous study (Mehl & Cotty, 2010; Table 1) comprised a greater proportion of the tissue-invading mycelia within maize and sorghum seeds than their respective co-inoculated isolates (Fig. 1a–c). However, during invasion of cotton and soyabean seeds, the three isolates highly competitive on maize varied in ability to outcompete co-infecting isolates. Success during competition was not correlated with aflatoxin-producing potential of A. flavus isolates on hosts overall (r2 = 0·002, P = 0·74, n = 48) or on sorghum (r2 = 0·007, P = 0·79, n = 12), maize (r2 = 0·05, P = 0·50, n = 12), cotton (r2 = 0·20, P = 0·14, n = 12) or soyabean seeds (r2 = 0·15, P = 0·21, n = 12).
Isolates differed in ability to compete during sporulation on host surfaces (Figs 1d–f & 2c,d). Predicted changes in isolate proportions during multiple cycles of reproduction are shown in Figure 1g–i. Although isolates EB01, YV10 and WM01 outcompeted co-infecting isolates during tissue invasion, the extent to which these isolates were expected to be successful over time (based on outcomes of competition during sporulation) varied by host.
To test the extent to which isolate × isolate and isolate × host interactions influence outcomes of competition, RB04, MN902 and CG136 were co-inoculated in all possible combinations on each of the hosts (Fig. 2). Influences of host species, co-inoculated isolate genotype, and the interaction between these two factors on isolate proportions from mycelia and conidia varied with the isolate being examined. Proportions of conidia comprised of RB04 and MN902 varied by host (P <0·001), but overall CG136 performed similarly during sporulation on each of the hosts (P = 0·053). However, an interaction between the host and co-infecting isolate influenced percentages of CG136 from conidia (P =0·01). Competitive success of RB04 during sporulation was low on all hosts regardless of the identity of the co-infecting isolate (Fig. 2c; P = 0·65). In contrast, MN902 competed poorly during host invasion and performed worst on cotton whether competing with RB04 or CG136 (Fig. 2b; P =0·89). Maize and sorghum similarly influenced outcomes of A. flavus competition.
In total, six pairs of A. flavus isolates were examined, and in each case, competitiveness during invasion of host tissues differed significantly among at least some of the four hosts but never between sorghum and maize (Figs 1a–c & 2a,b). Overall, isolate behaviour on maize was more similar to that on sorghum than to that on any of the other hosts. Sporulation during competition on maize, sorghum and soyabean were all correlated, but success during invasion of host tissues was correlated only for sorghum and maize (Table 4). Neither sporulation nor invasion of host tissues during competition on cotton seeds was correlated with outcomes of competition on any other host species.
Table 4. Correlations between outcomes of Aspergillus flavus competition on different host species
Outcomes of competition were quantified on each host for eight isolates assigned to six pairs. Coefficients of determination (r2) were calculated using mean percentages of isolate EB01 co-infecting with DV901, WM01 with MR17, RB04 with MN902, and CG136 co-infecting with either YV10, RB04 or MN902 on each of the four hosts (n = 6). Outcomes of competition during invasion of host tissues (mycelia) and sporulation (conidia) were analysed separately.
Competitive interactions during host invasion did not reflect competition during sporulation on hosts
Although both were influenced by host species, competitive advantage during host invasion was not correlated with competitive advantage during sporulation on the four hosts overall (r2 = 0·13, P =0·08). Isolate percentages from mycelia and conidia were correlated (but not equal) on maize (r2 = 0·78, P = 0·02, n = 6) but not on sorghum (r2 = 0·25, P = 0·31, n = 6), cotton (r2 = 0·09, P = 0·55, n = 6), or soyabean (r2 = 0·17, P = 0·42, n = 6). Differences between outcomes of competition during host invasion (percentage mycelia) and sporulation (percentage conidia) are shown in Table 5. Isolate percentages from mycelia and conidia differed significantly for six, four, three and one of the six pairs on sorghum, maize, cotton and soyabean, respectively.
Table 5. Differences in outcomes of competition between pairs of Aspergillus flavus isolates during tissue invasion and during sporulation on four host species
Pairs of isolates inoculated onto viable seeds of the indicated hosts. For each pair, the more competitive isolate during invasion of maize kernels was designated isolate A.
The percentage of total A. flavus DNA from conidia comprising isolate A was subtracted from the percentage of isolate A DNA from seed-invading mycelia. Results of two independent tests were combined for analyses (total n = 8). Significant values greater than zero indicate that isolate A was more competitive during invasion of host tissues than during sporulation on hosts and, conversely, that isolate B was more competitive during sporulation than host invasion. Values significantly different from zero are marked with an asterisk (*).
In a previous study (Mehl & Cotty, 2010) RB04 was moderately competitive during invasion of maize kernels but a poor competitor during sporulation on kernels, whereas MN902 was a moderate to good competitor during sporulation but a poor competitor during host invasion (Table 1). In contrast to other isolate pairs, isolate percentages for the RB04/MN902 pair from mycelia and conidia differed on all four hosts, with MN902 being favoured during sporulation and RB04 being favoured during invasion of host tissues (Table 5; Fig. 2). However, magnitudes of competitive advantage during sporulation (MN902) and tissue invasion (RB04) varied by host (Fig. 2).
Aspergillus flavus populations are genotypically and phenotypically diverse, and severity of aflatoxin contamination is influenced by the specific composition of crop-associated A. flavus populations (Cotty et al., 2008). The current study quantified for the first time the potential for hosts to influence proportions of genotypes within A. flavus populations. Sorghum, maize, cotton and soyabean differentially influenced competitive interactions among genetically distinct A. flavus isolates during host infection. Although they vary in susceptibility to aflatoxin contamination, these four crop species have the potential to contribute to A. flavus population structure in the agroecosystems where they co-occur. Proportions of conidia produced by competing isolates differed among the hosts for each of the six isolate pairs tested. Host influences on sporulation provide a mechanism by which crop rotations can influence compositions of A. flavus populations (Fig. 1g–i) and, thereby, aflatoxin contamination of subsequent crops (Jaime-Garcia & Cotty, 2010). Variation among crops in composition of associated A. flavus populations has been observed previously (Horn & Dorner, 1998; Vaamonde et al., 2003; Jaime-Garcia & Cotty, 2010), but this is the first study to quantitatively demonstrate differential influences of crop hosts on reproduction of specific competing genotypes.
Aspergillus flavus growth, sporulation and aflatoxin production varied among hosts, suggesting that A. flavus as a species is, for example, better adapted to maize than to soyabean. Field-based observations have also suggested a preference of A. flavus for certain hosts (Vaamonde et al., 2003; Jaime-Garcia & Cotty, 2010). When grown individually either on sorghum, maize, cotton or soyabean, differences in sporulation among the eight isolates were not detected, suggesting diverse A. flavus do not differ in innate ability to infect and reproduce on these hosts. Thus, differential host preferences among A. flavus genotypes were not apparent based on single-isolate inoculations, supporting previous conclusions that there is a lack of host specialization in A. flavus (St Leger et al., 2000).
Although differential host adaptation among genotypes of A. flavus was not detected based on single-isolate inoculations, crop hosts are typically infected by genetically diverse mixtures of A. flavus (Bayman & Cotty, 1991), and intraspecific competition during host-tissue invasion and reproduction on/in hosts may shape the adaptive evolution of A. flavus and other pathogens (Read & Taylor, 2001). In the current study, an interaction between host species and A. flavus genotype influenced success during co-infection. Thus, a type of host-selective adaptation was observed in which a specific host favours the success of a specific subset of pathogen genotypes. Behaviour of isolates during competition was more similar between sorghum and maize, the most closely related hosts tested, than between any of the other hosts. This suggests that maize and sorghum are sufficiently similar that fungal adaptations favouring success on one also favour success on the other. Adaptations that favour superior competition on a given host may have arisen through long periods of co-evolution or through discontinuous selection of the most fit fungal genotypes from highly diverse A. flavus populations.
Empirical data from the current study and others (Read & Taylor, 2001; Zhan et al., 2002; Salvaudon et al., 2005) demonstrate that in the context of competition, the relationship between virulence and pathogen reproduction is not straightforward and may be dependent on the specific host and pathogen genotypes involved. Trade-offs between success during host invasion and fungal reproduction, and the influence of host species on these trade-offs, have important implications for the epidemiology of crop aflatoxin contamination. Isolate competition during host tissue invasion directly influences aflatoxin contamination (Mehl & Cotty, 2010), but during epidemic increases in A. flavus populations, success during sporulation on the host influences compositions of crop-associated A. flavus populations. The disease cycle of A. flavus is polycyclic (Diener et al., 1987; Bock et al., 2004), so even small differences in competition among A. flavus genotypes during sporulation on host surfaces will influence population structure (Fig. 1g–i). This process may allow genotypes to increase in frequency during cropping and proportionately influence the epidemiology. For example, if isolates encounter a host in equal quantities but one isolate outcompetes another 69–31% during sporulation (e.g. MN902 paired with RB04 on maize; Fig. 2) that isolate will comprise over 90% of the population after only three cycles of sporulation. However, an isolate that outcompetes another only 56–44% (e.g. YV10 with CG136 on soyabean) still has an advantage and will comprise over 75 and 90% of the population after five and 10 generations, respectively. Such effects influence the average aflatoxin-producing potential of crop infecting populations. As a result, outcomes of competition on one host may influence aflatoxin contamination of other hosts sharing the agroecosystem.
The differential ability of A. flavus genotypes to invade and sporulate on diverse hosts may be a form of niche partitioning that allows multiple genetic types to coexist. If success during invasion of host tissues determined success during reproduction, an isolate competing poorly during host invasion, such as MN902 infecting cotton seeds, would rapidly become rare within the A. flavus population. However, MN902 sporulated as well as RB04 and CG136 on cotton seeds (Fig. 2d), suggesting this genotype would be maintained within populations. Success of a genotype during sporulation leads to increased prevalence when conditions are conducive to multiple cycles of infection and reproduction (Diener et al., 1987). When conditions do not favour dispersal of viable conidia to new hosts, the fungus requires structures capable of long-term survival (Jaime-Garcia & Cotty, 2004). Colonized host tissue may serve this purpose, and under such conditions, isolates more successful at invading and capturing host tissues will have an advantage.
Differential success among co-infecting A. flavus genotypes was most frequently observed on the hosts supporting the most sporulation. Soyabean, the host that supported the least sporulation, had the least influence on competition. With only one exception, percentages of paired isolates infecting soyabean equaled inoculated percentages (i.e. 50%). In contrast, sporulation was relatively high on sorghum and isolate percentages deviated from 50% for all isolate pairs. Overall, in the current study, mechanisms through which A. flavus genotypes exclude competitors provided greater advantages on hosts that supported greater A. flavus reproduction. Thus, not only do hosts supporting poor sporulation contribute little to population increases, they may also have little, if any, influence on shifting proportions of genotypes within the population.
Host-specific influences on A. flavus have important implications for the management of aflatoxin contamination. Aspergillus flavus populations include diverse assemblages of atoxigenic genotypes that may be used for biological control of aflatoxin contamination through competitive displacement of aflatoxin-producers (Atehnkeng et al., 2008; Yin et al., 2009; Probst et al., 2011). The current study demonstrates that the ability of A. flavus isolates to competitively displace other genotypes during infection and reproduction varies by host. Thus, selection of atoxigenic isolates for biocontrol agents should be performed on the target crops. Furthermore, identification of isolates competitive on both target and rotation crops may allow for greater dominance of applied atoxigenic strains over time and long-term reductions in the aflatoxin-producing potential of crop-associated fungal communities. Similarly, inclusion of hosts that favour atoxigenic genotypes in crop rotations may decrease the necessity for continuous reapplication of biocontrol strains. Genomic comparisons of isolates with differential ability to compete on a variety of hosts during both infection and reproduction may provide insight into the genetic basis for adaptation to target hosts and additional criteria for selection of atoxigenics.
We thank all the members of our lab, especially J. Tran and L. Price, for technical assistance. We thank Pioneer, Iowa State University, Sorghum Partners, Paul ‘Paco’ Ollerton and Stoneville Pedigree Seed Company for providing the seeds used in this study. This work was supported by the Agricultural Research Service, U.S. Department of Agriculture CRIS project 5347-42000-019-00D.