Species Diversity of and Toxin Production by Gibberella fujikuroi Species Complex Strains Isolated from Native Prairie Grasses in Kansas

Abstract

Fusarium species from agricultural crops have been well studied with respect to toxin production and genetic diversity, while similar studies of communities from nonagricultural plants are much more limited. We examined 72 Fusarium isolates from a native North American tallgrass prairie and found that Gibberella intermedia (Fusarium proliferatum), Gibberella moniliformis (Fusarium verticillioides), and Gibberella konza (Fusarium konzum) dominated. Gibberella thapsina (Fusarium thapsinum) and Gibberella subglutinans (Fusarium subglutinans) also were recovered, as were seven isolates that could not be assigned to any previously described species on the basis of either morphological or molecular characters. In general, isolates from the prairie grasses produced the same toxins in quantities similar to those produced by isolates of the same species recovered from agricultural hosts. The G. konza isolates produce little or no fumonisins (up to 120 μg/g by one strain), and variable but generally low to moderate amounts of beauvericin (4 to 320 μg/g) and fusaproliferin (50 to 540 μg/g). Toxicity to Artemia salina larvae within most species was correlated with the concentration of either beauvericin or fusaproliferin produced. Organic isolates from some cultures of G. moniliformis were highly toxic towards A. salina even though they produced little, if any, beauvericin or fusaproliferin. Thus, additional potentially toxigenic compounds may be synthesized by G. moniliformis strains isolated from prairie grasses. The Fusarium community from these grasses appears to contain some species not found in surrounding agricultural communities, including some that probably are undescribed, and could be capable of serving as a reservoir for strains of potential agricultural importance.

The Gibberella fujikuroi (Sawada) Ito in Ito & K. Kimura species complex, which contains species from Fusarium sections Liseola and Elegans, is composed of at least nine reproductively isolated biological species (mating populations) denoted by letters A through I (3, 16, 20, 22, 55). Separate Gibberella species names have been assigned to all but one of these mating populations (47, 55), and numerous additional Fusarium anamorphs within the Liseola and Elegans sections have been delineated on the basis of morphology and/or DNA sequence differences, e.g., those reported by Marasas et al. (28-30), Nirenberg and O'Donnell (40), and Nirenberg et al. (41). Collectively, these Fusarium species have a worldwide distribution and cause disease on a wide variety of agricultural and horticultural plants (for reviews, see references 21 and 22).

These species also produce a diverse array of known mycotoxins and secondary metabolites, including gibberellic acid (5), moniliformin (31), fumonisins (43), fusaric acid (2), beauvericin, and fusaproliferin (34, 44), that can contaminate animal and human feed and foods. There are significant qualitative and quantitative differences in the secondary metabolites produced by individual species, e.g., those reported by Fotso et al. (9), Leslie et al. (23, 25), Logrieco et al. (27), and Rheeder et al. (43). Differences in secondary metabolite production have been suggested as taxonomic characters for distinguishing species in the G. fujikuroi species complex, e.g., reports by Frisvad et al. (11), Moretti et al. (34) and Thrane (51).

In addition to morphological characters and secondary metabolite profiles, other methodologies used to separate and identify species within the G. fujikuroi complex include random amplified polymorphic DNA (1), isozyme polymorphisms (12), and DNA sequence data (32, 42, 49, 50). Recently, amplified fragment length polymorphism (AFLP) markers have been used to assess genetic similarity within and between Fusarium species (30, 55). AFLPs generate relatively complex genetic fingerprints that have been used to construct genetic maps (13, 14) of two Gibberella species and to measure genetic variation within fungal populations (45, 53, 53a). When mapped AFLP markers are used in population genetic analyses, however, linkage disequilibrium analyses of limited chromosomal regions become possible (53, 53a). AFLP fingerprint data are generally of limited use for phylogenetic purposes, however, other than determining if two isolates belong to the same or different species (30, 55).

The present study had two objectives: (i) to determine the species composition and genetic diversity of isolates from the G. fujikuroi species complex associated with nonagricultural grasses in a native North American tallgrass prairie, and (ii) to examine the types and quantity of secondary metabolites produced by strains recovered from nonagricultural hosts. There has been considerable study of Fusarium species diversity and toxin production among isolates collected from major grain crops such as rice, maize, or sorghum, e.g., reports by Desjardins et al. (7) and Leslie et al. (24, 25). However, this study represents the first population-level characterization of a nonagricultural, grasslands Fusarium community to determine whether such a community differs qualitatively or quantitatively from populations associated with the agricultural crops to which these fungi are economically important pathogens on a worldwide basis.

MATERIALS AND METHODS

Strains.

Standard strains from species in the G. fujikuroi species complex were used for mating tests, isozyme comparisons, and AFLP comparisons, including the standard mating population tester strains for mating populations A through I from the Fungal Genetics Stock Center (Department of Microbiology, University of Kansas Medical Center, Kansas City) FGSC 7600 (MATA-1), FGSC 7603 (MATA-2), FGSC 7611 (MATB-1), FGSC 7610 (MATB-2), FGSC 8931 (MATC-1), FGSC 8932 (MATC-2), FGSC 7615 (MATD-1), FGSC 7614 (MATD-2), FGSC 7616 (MATE-1), FGSC 7617 (MATE-2), FGSC 7057 (MATF-1), FGSC 7056 (MATF-2), FGSC 8934 (MATG-1), FGSC 8933 (MATG-2), FGSC 9022 (MATH-1), FGSC 9023 (MATH-2), FGSC 8910 (MATI-1), and FGSC 8911 (MATI-2), and ex-type isolates (9) from 15 additional species that probably belong to this species complex as well (28, 29, 40-42). Isozyme comparisons were made with the mating type testers of mating populations A through G. Sexual crosses were made as previously described (17) on carrot agar with the standard strains from the Fungal Genetics Stock Center as the female parents. Crosses between unidentified sets of isolates were made only between isolates identified as opposite mating type through molecular analyses (15, 50). In these experiments, the unidentified isolates were all tested as both the male and the female parent in each cross combination.

Field collections and fungal isolation.

The Flint Hills region encompasses over 50,000 km² in eastern Kansas, contains the largest remaining area of unplowed tallgrass prairie in North America, and retains many of its native characteristics. Several species of perennial warm-season grasses, e.g., Andropogon gerardii Vitman, Andropogon scoparius Michx., and Sorghastrum nuttans (L.) Nash, dominate vegetation in the Flint Hills, while subdominant species include a diverse mixture of other warm- and cool-season grasses, composites, legumes, and other forbs (10). In October of 1995 in Lyon County, Kansas, and October 1996 in Riley County, Kansas, we collected a few stems and seed heads of A. gerardii and S. nuttans as part of a pilot study to determine whether Fusarium isolates could be isolated from these plants. In October 1997, we conducted a more extensive sampling of A. gerardii, S. nuttans, and A. scoparius from the Konza Prairie Biological Station and Long Term Ecological Research site. The Konza Prairie is a remnant native and unplowed tallgrass prairie site located in the Flint Hills of eastern Kansas just south of Manhattan, Kansas (39°05′ N, 96°35′ W) and managed by Kansas State University for the Nature Conservancy.

We arbitrarily selected 15 plants each of A. gerardii and A. scoparius and 14 plants of S. nuttans across three different experimental watersheds of the Konza Prairie that differ in the frequency in which they are burned. Flowering tillers of each plant were cut at ground level, bagged individually, and catalogued on site. Fungal isolations were made from these materials within 3 days of collection.

For isolation of fungi from plant material, each plant was cut into small sections (2 to 5 cm length) and surface sterilized in 95% ethanol for 2 min. Sterilized sections were rinsed briefly in sterile H₂O and placed on a peptone-PCNB medium (Nash-Snyder medium) semiselective for Fusarium spp. (37). Cultures were incubated under fluorescent lights at 25°C with a 12-h-12-h day-night cycle for 4 to 7 days to allow fungal colonies to grow out onto the medium. Two to five morphologically distinct Fusarium colonies were isolated from each plant and then transferred to complete medium (6) slants. From these colonies, cultures were started from microconidia separated by micromanipulation, and a subculture was frozen for long-term storage at −70°C in 15:85 (vol/vol) glycerol-water.

We cultured isolates for isozyme characterization, extracted soluble proteins, and identified putative mating populations within the G. fujikuroi species complex as described by Huss et al. (12). Specifically, malate dehydrogenase, isocitrate dehydrogenase, and fumarase profiles were compared in a morpholine buffer system, and triose phosphate isomerase profiles were compared in a Soltis-6 buffer system for each sample. Mating population identifications were confirmed, and mating type specificity (MAT-1 or MAT-2) was identified in crosses in which the field isolates served as males and the standard tester strains for these mating populations served as the female parents. Isolates with variant isozyme profiles were crossed with testers from the mating population or set of mating populations with profiles that most closely resembled those of the field isolates. Isolates that were cross-fertile with a mating type tester strain were tested for female fertility by using the field isolate as a female in a cross with the mating type tester as the male parent. All successful crosses were repeated at least once. Mating type idiomorphs of isolates that were not cross-fertile with any of the tester strains for the known mating populations were identified by allele-specific PCR amplification (15, 50).

AFLP analyses and comparisons.

Isolates for DNA extraction were cultured by inoculating approximately 1 ml of a spore suspension (typically 10⁶ to 10⁷ conidia) into 40 ml of liquid complete medium (6). Isolates were grown on a rotary shaker (150 rpm) for 2 days at room temperature (23 to 26°C) and harvested by filtration through milk filters (KenAG, Ashland, Ohio). Mycelia were blotted dry between paper towels, and the dried mycelia were stored at −20°C until DNA extraction.

DNA was extracted with a cetyltrimethylammonium bromide protocol as described by Kerényi et al. (15). AFLPs (52) were generated as described by Zeller et al. (54). We used all buffers and DNA modifying enzymes following either the manufacturer's instructions or standard protocols (46). EcoRI primers used in the final specific PCR amplifications were labeled with [γ-³³P]ATP (Perkin-Elmer, Shelton, Conn.).

The presence or absence of polymorphic AFLP bands ranging in size from 100 to 800 bp in each gel was scored manually, and the data were recorded in a binary format. All polymorphic markers in this size range were scored, even those that were unique to a single individual. Bands appearing at the same mobility in different individuals were assumed to represent the same allele. Each band of differing mobility was treated as a single independent locus with two alleles (present or absent), and unresolved bands or missing data were scored as ambiguous.

Initially, a single AFLP primer pair was used to group isolates that could not be identified to a specific mating population on the basis of either isozyme profile or cross-fertility. Results for each group were compared to profiles from isolates of previously identified Fusarium species. The resulting binary data set was analyzed with the unweighted pair group method with averages (UPGMA) clustering option of PAUP (version 4.10b*; D. L. Swofford, Sinauer Associates, Sunderland, Mass.) to suggest possible species associations for the unidentified isolates. These results were used to reorder and rerun AFLPs with members of potential species groupings placed for direct, side-by-side comparisons. UPGMA similarities were calculated based on these AFLP gels. Final UPGMA genetic distances within and between sets of species were calculated with the Dice coefficient (38) and the CLUSTER option of SAS (version 6.11 for PC; SAS Institute, Cary, N.C.).

In vitro mycotoxin production and toxicity.

Isolates were cultured on 100 g of autoclaved yellow maize kernels (approximately 45% moisture content) in 500-ml Erlenmeyer flasks and inoculated with 2 ml of an aqueous suspension containing approximately 10⁷ conidia/ml. Cultures were incubated at 25°C for 4 weeks. Harvested culture material was dried in a forced draft oven at 60°C for 48 h, finely ground, and stored at 4°C until used. Noninoculated control samples were inoculated with water and treated in the same manner.

Beauvericin and fusaproliferin recovery and measurement.

For beauvericin and fusaproliferin extractions, 5 g of each sample was ground and homogenized (13,500 rpm for 3 min) at room temperature with 50 ml of methanol in an Ultra-Turrax model T25 apparatus (IKA Werke, Staufen, Germany) and then clarified by centrifugation at 2,000 × g for 5 min at 4°C. Samples were filtered through Whatman no. 4 filter paper, and 30 ml of filtrate was collected (corresponding to 3 g of sample) and then evaporated at 35 to 40°C under reduced pressure (Heidolph Instruments, Schwabach, Germany). The raw organic extract was resuspended in 3 ml of methanol (23) and loaded onto a C₁₈ column (Varian, Inc., Palo Alto, Calif.). The prepurification column was washed with 2 ml of methanol, and the eluted sample was dried under vacuum. Finally, the collected residue was resuspended in 1 ml of methanol and filtered through an Acrodisk filter (0.22-μm pore diameter) (Varian, Inc.) before injection of 20 ml into the high-performance liquid chromatograph (HPLC). The recovery efficiencies of this extraction method were 94 and 71% for beauvericin and fusaproliferin, respectively.

The amount of beauvericin and fusaproliferin was determined by HPLC as previously described by Monti et al. (33) and Munkvold et al. (36), respectively. In brief, HPLC analyses were performed using LC-10AD pumps and a diode array detector from Shimadzu (Kyoto, Japan). A Shiseido Capcell Pak C₁₈ column (250 by 4.6 mm; 5 mm) was used with a constant flow of 1.5 ml/min and acetonitrile-H₂O (65:35 [vol/vol]) as the initial eluent. The starting ratio was kept constant for 5 min and then changed linearly to 70% acetonitrile in 10 min. After 1 min at 70% acetonitrile, the mobile phase was shifted back to the starting conditions in 4 min. Fusaproliferin was detected at 261 nm, and beauvericin was detected at 205 nm. Mycotoxins were identified by comparing retention times and UV spectra of samples with those of authentic standards. Further confirmation of identity was obtained by coinjecting pure standards with each sample. Mycotoxins were quantified by comparing peak areas from samples with a calibration curve of standards. The detection limit was 100 ng/g for beauvericin and 25 ng/g for fusaproliferin. All analyses were run in triplicate, and the mean values are reported; standard deviations were <5% of the reported values.

Recovery and measurement of fumonisins.

A 5-g sample of each fungal culture was finely ground and added to 50 ml of a 75:25 methanol-water solution. The samples were homogenized in an Ultra-Turrax homogenizer (13,500 rpm for 3 min at room temperature) and then clarified by centrifugation at 2,000 × g for 5 min at 4°C. Thirty milliliters of the supernatant (corresponding to 3 g of sample) was evaporated at 35 to 40°C under reduced pressure as described above. The residue was resuspended in 5 ml of methanol and dried in a centrifugal evaporator (RC 10.10; JOUAN S.A., St. Herblain, France), at 35 to 40°C. The recovery efficiency of this extraction method was 90 and 68% for fumonisin B₁ (FB₁) and FB₂, respectively.

For fumonisin detection, mass spectrometry (MS) analyses were performed. A bench-top API 100 (Perkin-Elmer Sciex, Shelton, Conn.) single quadrupole mass spectrometer equipped with an atmospheric pressure ionization source and ion spray interface was used. All quantitative results were performed in positive ion mode with the orifice voltage set at 30 V. The acquired data were processed using Multiview and MacQuan software (Perkin-Elmer Sciex). The resolution was set at 0.5 atomic mass unit (measured at half height), and the mass calibration and resolution adjustments on the resolving quadrupole were made in ion spray with 10⁻⁴ M polypropylene glycol solution introduced via a model 22 Harvard infusion pump (Apex Equipment, Hillsborough, N.J.). Initial MS spectra were collected in continuous flow mode by infusion directly to the ion spray probe. A standard mixture of FB₁ and FB₂ was infused at 10 μl/min. Chromatographic runs were performed with Perkin-Elmer LC-200 pumps (Perkin-Elmer Sciex) controlled by a pump controller.

The LC-MS experiments were performed with a Pecosphere Brownlee C₁₈ 4.6- by 33-mm HPLC column (Perkin-Elmer Sciex). Column flow was 0.8 ml/min, with an aqueous eluent of 5 mM ammonium acetate acidified with 1% formic acid containing 80% methanol. Only 30 μl/min was delivered to the ion spray source. Standards of FB₁ and FB₂ were purchased from Sigma (St. Louis, Mo.), and a stock solution was made in methanol (1 mg/ml). Further dilutions were made with a 50% aqueous solution of methanol containing 2 mM ammonium acetate and 0.1% formic acid. FB₁ and FB₂ were quantified at m/z 722 and 706 for FB₁ and FB₂, respectively. The detection limit for the fumonisin standards, determined by using the protonated signal at m/z 722, was equivalent to 0.5 ng/g, and the limit of quantification was equivalent to 1.2 ng/g. Measurements were made in quadruplicate within an experiment, and the data are based on the average of two experiments.

Brine shrimp assays.

Toxicity to brine shrimp (Artemia salina) larvae was determined as previously described (26). Briefly, larvae were exposed to fungal culture extracts in 24-well cell culture plates (30 to 40 larvae per well in 500 μl of 3.3% [wt/vol] marine salt in H₂O). The number of dead larvae was recorded after incubation at 27°C for 24 h. The total number of larvae in each cell was counted after killing the surviving larvae by freezing at −20°C for 12 h. Water-soluble extracts, containing fumonisins, in general had little or no toxicity (0 to 3% mortality) (data not shown) towards the brine shrimp larvae, and data are reported only from extracts made in methanol that usually contained fusaproliferin and/or beauvericin. Tests in each experimental run were performed in quadruplicate. Data are based on the averages of two independent experiments (see Table 1). We examined correlations between toxin production by individual Fusarium species and mortality to A. salina by comparing observed toxin concentrations (in parts per million) to percent mortality observed with the CORR procedure of SAS version 6.12 (SAS Institute).

TABLE 1.

Species identification for and secondary metabolite production by strains of the G. fujikuroi species complex recovered from grasses growing on the Konza Prairie

KSU no.	ITEM no.	Plant hosta	Isozyme patternb	AFLP patternb	Mating type	FB1 (μg/g)	FB2 (μg/g)	FUPe (μg/g)	BEAf (μg/g)	% A. salina mortality
A-10548	3117	Ag	Gm	Gm	MAT-2	590	420	NDg	ND	93
A-10560	3122	Ag	Gm	Gm	MAT-2	1,200	710	ND	9	97
A-10568	3126	Ag	Gm	Gm	MAT-2	760	780	ND	ND	97
A-10577	3131	Ag	Gm	Gm	MAT-2	750	670	ND	ND	68
A-10584	3136	Ag	Gm	Gm	MAT-1c	1,000	700	ND	ND	39
A-10594	3140	Ag	Gm	Gm	MAT-2c	110	130	ND	ND	78
A-10605	3147	As	Gm	Gm	MAT-1	280	64	ND	4	100
A-10621	3152	As	Gm	Gm	MAT-1c	480	150	ND	2	100
A-10631	3158	As	Gm	Gm	MAT-1c	890	200	ND	5	48
A-10636	3161	As	Gm	Gm	MAT-2	280	97	ND	ND	100
A-10685	3181	Sn	Gm	Gm	MAT-2c	500	130	ND	ND	76
A-10691	3184	Sn	Gm	Gm	MAT-1	18	4	ND	ND	100
X-10626	3155	As	Gf	Gf-var	MAT-1	900	170	ND	180	71
X-10677	3178	Sn	Gi-var	Gf-var	MAT-1	ND	ND	ND	30	11
D-08384	3110	Sn	Gi	Gi	MAT-1	430	350	840	310	100
D-08387	3111	Sn	Gi	Gi	MAT-2	37	36	49	110	97
D-08392	3112	Sn	Gi	Gi	MAT-1	780	360	540	230	100
D-08403	3114	Sn	Gi	Gi	MAT-1	3,600	4,200	590	170	100
D-08411	3108	Ag	Gi	Gi	MAT-2	400	110	560	120	48
D-08420	3109	Ag	Gi	Gi	MAT-1	500	570	830	180	100
D-10550	3118	Ag	Gi	Gi	MAT-1	160	230	2,000	450	100
D-10552	3120	Ag	Gi	Gi	MAT-1	6,300	4,000	260	ND	100
D-10557	3121	Ag	Gi	Gi	MAT-2	640	210	500	210	99
D-10563	3124	Ag	Gi	Gi	MAT-1	1,300	530	490	430	100
D-10565	3125	Ag	Gi	Gi	MAT-1	610	700	150	830	100
D-10580	3133	Ag	Gi	Gi	MAT-1	280	78	390	310	100
D-10582	3134	Ag	Gi	Gi	MAT-2	620	760	550	1,400	100
D-10583	3135	Ag	Gi	Gi	MAT-1	1,200	830	200	1,200	100
D-10587	3137	Ag	Gi	Gi	MAT-1c	1,500	840	110	450	100
D-10590	3138	Ag	Gi	Gi	MAT-2	250	120	270	530	100
D-10591	3139	Ag	Gi	Gi	MAT-2c	1,000	620	570	580	100
D-10599	3145	As	Gi	Gi	MAT-1	1,400	360	200	1,300	100
D-10609	3148	As	Gi	Gi	MAT-2c	1,100	370	430	66	72
D-10614	3149	As	Gi	Gi	MAT-2c	1,300	170	510	660	100
D-10616	3150	As	Gi	Gi	MAT-2c	3	1	200	7	28
D-10617	3151	As	Gi	Gi	MAT-2	710	190	280	1,000	100
D-10625	3154	As	Gi	Gi	MAT-2	450	93	180	700	100
D-10627	3156	As	Gi	Gi	MAT-1	150	29	ND	66	12
D-10630	3157	As	Gi	Gi	MAT-2	520	120	51	12	3
D-10647	3165	As	Gi	Gi	MAT-2d	210	67	320	570	100
D-10649	3166	Sn	Gi	Gi	MAT-1	240	74	210	1,200	100
D-10657	3170	Sn	Gi	Gi	MAT-2d	86	27	130	140	26
D-10659	3171	Sn	Gi	Gi	MAT-1c	ND	ND	86	270	77
D-10668	3174	Sn	GflGi	Gi	MAT-2	810	230	560	780	100
D-10670	3175	Sn	Gi	Gi	MAT-1	170	29	170	350	89
D-10675	3176	Sn	Gi	Gi	MAT-2	340	51	630	820	100
D-10694	3185	Sn	Gi	Gi	MAT-1	860	58	45	4	35
D-08374	3107	Ag	Gi-var	Gi	MAT-2	1,500	1,000	450	480	100
D-10544	3115	Ag	Gi-var	Gi	MAT-1	10	23	ND	ND	5
D-10545	3116	Ag	Gi-var	Gi	MAT-1	60	820	150	ND	3
D-10571	3127	Ag	Gi-var	Gi	MAT-2	1,500	360	200	1,000	99
D-10572	3128	Ag	Gi-var	Gi	MAT-2	1,700	1,400	32	170	92
E-10562	3123	Ag	Gs	Gs	MAT-2c	Trace	ND	300	ND	98
E-10646	3164	As	Gs	Gs	MAT-1	ND	ND	190	5	24
E-10688	3182	Sn	Gs	Gs	MAT-1	ND	ND	1,300	10	100
F-10597	3142	Ag	Gt	Gt	MAT-1	6	9	ND	ND	6
I-08373	3106	Ag	Gs-var	Gk	MAT-1	10	12	540	120	79
I-10595	3141	Ag	Gs-var	Gk	MAT-2	17	12	210	160	80
I-10638	3162	As	U	Gk	MAT-2	120	24	ND	5	2
I-10653	3168	Sn	Gs-var	Gk	MAT-1c	ND	ND	250	91	28
I-10663	3173	Sn	Gs	Gk	MAT-1	ND	ND	310	4	17
I-10676	3177	Sn	Gs	Gk	MAT-1	ND	ND	210	230	80
I-10678	3179	Sn	Gs-var	Gk	MAT-2	ND	ND	50	59	32
I-10689	3183	Sn	Gs	Gk	MAT-2	ND	ND	140	89	43
I-10681	3180	Sn	U	Gk	MAT-2	ND	ND	78	320	78
I-10578	3132	Ag	Gs-var	Gk-like	MAT-2d	Trace	Trace	160	650	99
X-10622	3153	As	U	Type α	ND	18	5	ND	80	100
X-10635	3160	As	U	Type α	ND	ND	ND	ND	32	100
X-10661	3172	Sn	U	Type α	ND	ND	ND	ND	190	57
X-10576	3130	Ag	Gm	Type β	ND	9	12	ND	ND	0
X-10634	3159	As	Gm	Type β	ND	ND	ND	ND	10	1
X-10551	3119	Ag	U	Type γ	MAT-1d	13	13	ND	32	13

^aSpecies abbreviations: Ag, A. gerardii; As, A. scoparius; Sn, S. nuttans.

^bSpecies abbreviations: Gf, G. fujikuroi; Gi, G. intermedia; Gk, G. konza; Gm, G. moniliformis; Gs, G. subglutinans; Gt, G. thapsina. U, unique. Patterns that are substantially the same as one species but differ at one or a few bands (usually new) are designated as “var” for variant. Strains with a var designation need not have the same banding pattern.

^cFemale fertile.

^dMating type determined via mating type allele-specific PCR (15, 50).

^eFUP, fusaproliferin.

^fBEA, beauvericin.

^gND, not detected.

RESULTS

We examined 72 isolates that produced microconidia from a total of 241 Fusarium isolates recovered. Of the 52 grass stems sampled, 45 produced at least one isolate that made microconidia and that grouped morphologically within Fusarium sections Liseola or Elegans (39). More than one species from the G. fujikuroi species complex was recovered from 16 of 45 plants, and members of more than two species from the G. fujikuroi species complex were recovered from 5 of 45 plants, indicating that multiple infection of native grasses with more than one species of Fusarium is common. Of the 72 Fusarium isolates examined, 12 belonged to G. moniliformis, 1 to G. thapsina, 3 to G. subglutinans, 9 to G. konza, 40 to either G. fujikuroi or G. intermedia, and 7 could not be assigned readily to any of the described species in the G. fujikuroi species complex (Fusarium section Liseola and associated species). Isolates were characterized (Table 1) with isozymes, AFLP fingerprint profile, and sexual cross-fertility, with mating type determined by using PCR-based methods whenever crosses with known testers failed. Collectively, these characterizations did not always yield the same answer. These data are presented by species.

G. moniliformis (F. verticillioides or G. fujikuroi mating population A).

All 12 isolates identified as G. moniliformis had isozyme profiles similar to those previously reported for this species. All of the isolates were fertile in crosses with the standard mating type tester strains for this species when the field isolates were used as the male parent, and 5 of 12 field isolates also were fertile when used as the female parent. Both mating types were present (five MAT-1, seven MAT-2) among this set of field isolates. AFLP similarity for the 12 isolates in this group was >76%.

All 12 isolates produced FB₁ (range, 18 to 1,200 μg/g; 570 ± 360 μg/g [mean ± standard deviation]) and FB₂ (4 to 780 μg/g; 340 ± 300 μg/g), but only three isolates produced detectable levels of beauvericin (2 to 9 μg/g) and none produced detectable levels of fusaproliferin. Culture extracts varied in toxicity towards brine shrimp, with 39 to 100% (83 ± 22%) mortality among brine shrimp exposed to fungal culture extracts for 24 h. Brine shrimp toxicity and the amount of fumonisins produced were not correlated since these toxins, which are soluble in water and/or highly polar solvents, were localized in the water extracts that were not toxic to the brine shrimp larvae. Neither was there a correlation between the level of toxicity (for some isolates 100%) and the very poor occurrence of the only toxin detected in the extracts tested, beauvericin.

G. thapsina (F. thapsinum or G. fujikuroi mating population F).

The sole isolate of this species had an isozyme profile typical for this species. This isolate was sexually cross-fertile with the standard MATF-2 tester strain when the field isolate was the male parent, but not when it was the female parent. The isolate produced detectable levels of fumonisins but no fusaproliferin or beauvericin, and culture extracts were essentially nontoxic in the brine shrimp assay.

G. subglutinans (F. subglutinans or G. fujikuroi mating population E).

Three isolates of this species were recovered, and all had the isozyme profile that is typical for the species. All three isolates were fertile in crosses with the standard mating type tester strains for this species when the field isolates were used as the male parent, and one of the field isolates (E-10562) also was fertile when used as the female parent. Both mating types were present among the three field isolates. AFLP similarity for the three strains in this group was >83%. None of the three strains produced detectable levels of either FB₁ or FB₂. All three strains produced significant levels of fusaproliferin (190 to 1,300 μg/g), and two of the three strains produced low levels of beauvericin (5 to 10 μg/g). Culture extracts varied in toxicity towards brine shrimp, with one strain resulting in 24% mortality and the other two strains resulting in the death of all or nearly all (98 to 100% mortality) of the brine shrimp. The two more toxic of these isolates also were the two that produced the highest concentrations of fusaproliferin (Table 1).

G. konza (F. konzum or G. fujikuroi mating population I).

Three of the isolates in this group had isozyme profiles that were indistinguishable from that of G. subglutinans, five other isolates had a novel isocitrate dehydrogenase allele with greater mobility than those previously described, and two isolates had isozyme profiles with insufficient resolution to be classified within the framework of that of Huss et al. (12). AFLP similarity among 9 of the 10 isolates in this group was >80% (55). The 10th isolate, X-10578, was considerably less similar (∼53%) to the other nine isolates and could represent a closely related sibling species, rather than being a member of G. konza. The set of three isolates with an isozyme profile similar to that of G. subglutinans were >80% similar, the four isolates with the alternative isocitrate dehydrogenase isozyme profile were >82% similar, and the two isolates with unresolved isozyme profiles were 86% similar. Morphologically, all 10 isolates resembled F. anthophilum and thus could be easily distinguished from G. subglutinans.

All of the isolates except X-10578 were fertile as males with the standard G. konza tester strains. Since the overall genetic similarity among the nine cross-fertile isolates was >80%, these results are consistent with the hypothesis that all of these isolates belong to a single interbreeding population with simple allelic polymorphism at the isocitrate dehydrogenase locus. Two of the isolates (I-10676 and I-10678) came from the same plant but differed in both their AFLP fingerprint type and the MAT idiomorph present.

Three of the nine isolates produced relatively low levels of FB₁ (trace to 120 μg/g) and FB₂ (trace to 24 μg/g). All nine isolates produced beauvericin (4 to 320 μg/g), and eight of nine produced fusaproliferin (50 to 540 μg/g). Culture extracts varied in toxicity towards brine shrimp, with 2 to 80% mortality among brine shrimp exposed to fungal culture extracts for 24 h. The correlation between toxin level and the mortality of brine shrimp larvae was much higher for beauvericin (r = 0.835; P = 0.005) than for fusaproliferin (r = 0.371; P = 0.33). X-10578 produced traces of fumonisins, both beauvericin (650 μg/g) and fusaproliferin (160 μg/g), and was quite toxic to brine shrimp larvae (99% mortality).

G. intermedia (F. proliferatum or G. fujikuroi mating population D) and G. fujikuroi (F. fujikuroi or G. fujikuroi mating population C).

This group was both the largest, 40 isolates, and the most complicated in terms of their genetic relationships to one another and to the standard tester strains. Two of the isolates in this group had the isozyme profile of G. fujikuroi, 33 isolates had the isozyme profile typical of G. intermedia, and 5 isolates had a typical G. intermedia isozyme profile except for a novel, higher-mobility isocitrate dehydrogenase allele.

Crosses with standard tester strains confirmed that all 38 isolates with either the standard or the variant G. intermedia isozyme profile clearly belonged to this species. Female fertility for these 38 isolates was relatively low, with only 6 of 38 field isolates capable of functioning as a female parent in a cross with the standard tester strains. The other two isolates, X-10626 and X-10677, produced unusual results in crosses with the standard mating type tester strains of both G. fujikuroi and G. intermedia. Isolate X-10626 produced perithecia and viable progeny when crossed as a male with the MAT-2 testers from each of these species. Similar crosses with isolate X-10677 also resulted in perithecia being produced with the testers from both species, but viable ascospores were produced only in the crosses with the G. intermedia tester strain. Neither of these field isolates was fertile as the female parent in crosses with the standard testers of either G. intermedia or G. fujikuroi.

AFLP similarity among the 38 isolates with the MP-D or related isozyme phenotype and a standard tester strain for G. intermedia, D-04853, was ≥71%. The five isolates with the variant isozyme type were more similar to each other (82% average) than to the remainder of the G. intermedia cluster (71% average). Isolates X-10626 and X-10667 were 81% similar to each other based on AFLPs. These isolates were not highly similar to the tester strain for either G. fujikuroi (65%) or to the G. intermedia tester or the other G. intermedia isolates from the Konza Prairie (∼50%).

Isolate pairs in this group were recovered from eight different plants. These isolate pairs were D-08384 and D-08387, D-10544 and D-10545, D-10571 and D-10572, D-10582 and D-10583, D-10616 and D-10617, D-10625 and D-10627, D-10657 and D-10659, and D-10668 and D-10670. In all but two of these cases (D-10544 and D-10545; D-10571 and D-10572), the isolates from each pair had different AFLP fingerprint genotypes. In four of the six cases in which the AFLP genotypes differed, the isolates also had different MAT alleles.

All but one of the G. intermedia isolates produced both FB₁ and FB₂, often at levels of ≥1,000 μg/g of total fumonisins (1,400 ± 2,000 μg/g). All but two of the G. intermedia isolates produced fusaproliferin (32 to 2,000 μg/g; 360 ± 360 μg/g), and all but three isolates produced beauvericin (4 to 1,400 μg/g; 450 ± 410 μg/g). Thus, many of the G. intermedia isolates produced significant levels of all four of the measured secondary metabolites. X-10626 produced fumonisins and beauvericin, but X-10677 produced only beauvericin at a relatively low level. Culture extracts from the G. intermedia isolates were generally quite toxic (81% ± 33%), with no brine shrimp surviving a 24-h exposure to extracts from 22 of the 38 isolates. There was a significant positive correlation between beauvericin production and brine shrimp toxicity (r = 0.560; P < 0.001) and a weaker, but still significant, correlation between fusaproliferin production and brine shrimp mortality (r = 0.378; P = 0.02). The X-10626 culture extract was relatively toxic towards the brine shrimp (71% mortality), while the X-10677 culture extract (11% mortality) was not.

Unassigned isolates.

Based on AFLP data and comparisons of these profiles with those from the other species in the G. fujikuroi species complex, the remaining seven unassigned isolates can be placed into three small groups each containing one to three of those isolates. Two of these groups (AFLP types α and β) produce macroconidia that are morphologically similar to Fusarium oxysporum, as well as producing microconidia in false heads on short monophialides and abundant chlamydospores. All three groups differ from one another in their isozyme profiles. The α type has a new MDH allele, the β type has a profile similar to those for G. moniliformis or G. nygamai, and the γ type has a novel combination of previously identified alleles (12). None of the six isolates was cross-fertile with any of the standard tester strains from the nine described mating populations, and in the cases in which both mating types were present in a subset, none of the possible pairwise crosses between members of the same group was cross-fertile. Within an AFLP type, the AFLP similarity was >70%, and the closest similarity to any other species in this study was <40%.

Toxin production and toxicity to brine shrimp also varied in this heterogeneous group. Isolate X-10578, morphologically similar to F. konzum, was highly toxic (99% mortality) and produced high concentrations of both beauvericin and fusaproliferin (Table 1). None of the other six isolates produced high levels of any of the toxins, but at least one isolate of every group made FB₁, FB₂, and/or beauvericin. Isolates from AFLP species type α (X-10622, X-10635, and X-10661) were relatively toxic (57 to 100% mortality) but generally produced little of the tested toxins. The two isolates of AFLP species type β (X-10576 and X-10634) had identical AFLP fingerprint profiles, produced no detectable amounts of the tested toxins, and were nearly nontoxic to the brine shrimp larvae. The single tested isolate of AFLP species type γ (X-10551) produced low concentrations of fumonisins and of beauvericin but was not highly toxic to brine shrimp (13% mortality).

DISCUSSION

Species recovered.

Agricultural ecosystems are characterized by host (crop) populations that are relatively uniform genetically, numerically, and spatially (reviewed by Burdon et al. [4]) and that experience regular, human-induced disturbance. Agricultural ecosystems are unlikely to be in equilibrium, and strong selection for pathogenically specialized and aggressive populations of plant parasites normally is expected (see Lenné and Ortiz [19]). In nonagricultural ecosystems, inter- and intraspecific variabilities generally are expected to be higher, there is greater diversity in both biotic and abiotic environments, and greater spatial and temporal variability occurs in host-plant populations. Unlike agricultural ecosystems, the effects of natural selective processes that operate at or near equilibrium, e.g., migration and genetic drift (reviewed by Finckh and Wolfe [8]), are more likely to be observed. In native grasslands, such as the Konza Prairie, some of the expected changes in Fusarium populations relative to those observed in agriculture might include higher species diversity, changes in species composition relative to those observed from agricultural grasses, higher abundance of species generally viewed as generalists, or alterations in factors, e.g., mycotoxin production, that could be involved in pathogenicity or other host-fungus interactions.

We recovered isolates from six of nine described biological species (mating populations) within the G. fujikuroi species complex from these native prairie grasses. In general, the range of species found in the prairie grasses paralleled that typically recovered from maize or from sorghum crops (21, 24), which are grown in the adjacent area. Isolates belonging to G. moniliformis, G. intermedia, and G. subglutinans can be recovered from maize, while isolates belonging to G. intermedia or G. thapsina can be recovered from sorghum. The only species that we collected that have not been typically reported from either of these two crops is G. konza and the representatives of the three unidentified species. The only species common to maize or sorghum that we did not recover from the Konza Prairie was Fusarium andiyazi (30). Thus, the range of species recovered from the Konza Prairie generally mirrored that seen in the adjoining agricultural areas, although G. konza and the three unidentified species are potential endemic species in this area and as such require additional sampling from other grasslands for testing and verification.

The relative frequency of the species recovered from the Konza Prairie was not the same as that commonly found in the agricultural areas. The Fusarium species most commonly reported from maize and from sorghum are G. moniliformis and G. thapsina, respectively (18, 21, 24). While both of these species have been collected from other crops, they appear to be most common on their respective crop hosts and may represent aggressive, and at least partially specialized, pathogen populations. Although both G. moniliformis and G. thapsina were present on the Konza Prairie (17 and 2% of the isolates, respectively), neither was found at the high (often 70% or greater) frequencies at which they can be recovered from agricultural fields of maize or sorghum. From the Konza Prairie, G. intermedia was the dominant species and was recovered from more than half of the plants examined.

G. intermedia is often viewed as a generalist, as it has been recovered from a broad range of agricultural hosts that includes asparagus, banana, maize, mangos, millet, pine, rice, sorghum, and tobacco (21). When G. intermedia is recovered it often is relatively infrequent (5 to 15% of the total population), especially in the Great Plains region in the United States (21, 24). This species also is not considered generally to show any significant host preference or specialization. The dominance of G. intermedia on the Konza Prairie is consistent with a hypothesis in which a generalist species is expected to be at an advantage in a native ecosystem, even if this species can survive and persist in a more-specialized agricultural ecosystem. If this hypothesis is correct, then G. intermedia, or similar generalist species, should dominate the Fusarium communities found in other native ecosystems.

Collectively, these results challenge the phylogeographic origins of different species within the G. fujikuroi species complex as proposed by O'Donnell et al. (42). G. subglutinans and G. konza, which is closely related to G. subglutinans but which was not included in the original O'Donnell et al. analysis (42), would be in the “American” clade and might be expected to be present in a native grassland in Kansas. The representatives of G. moniliformis and G. thapsina, which are hypothesized to have originated in Africa, and the representatives of G. intermedia, which is hypothesized to have originated in Asia, however, all must have been introduced. The sampled areas of the Konza Prairie, and the overwhelming majority of the site in general, have never been tilled, which means that any introduced species must have drifted in from nearby farming operations. Such an occasional introduction might suffice to explain the presence of G. moniliformis and G. thapsina at the relatively low frequencies observed. However, the dominance of G. intermedia within this population and the presence of G. intermedia strains that can hybridize with other biological species in this species complex (25a, 56) suggest that this species has a long-term presence in this native grassland that is inconsistent with a phylogeographic origin of these species in Asia.

Mycotoxin production.

When we examined patterns of mycotoxin production among species from these prairie isolates, we discovered that toxin production was neither qualitatively nor quantitatively different from isolates of the same species drawn from agricultural settings (21, 22). Isolates of G. moniliformis generally produce fumonisins but undetectable quantities of beauvericin or fusaproliferin, the majority of the isolates of G. intermedia produce detectable quantities of all three of these mycotoxins, and isolates of G. subglutinans generally produce fusaproliferin but little or no fumonisins or beauvericin (25, 35). In general, the isolates of G. intermedia produced as much or more fumonisins as did the isolates of G. moniliformis. Six of the isolates of G. intermedia produced more than 2,000 μg/g of total fumonisins, including D-08403, which produced nearly 8,000 μg/g of total fumonisins and D-10552, which produced >10,000 μg/g, more than the highest level produced by any of the isolates of G. moniliformis in this collection (Table 1). Our findings fit well into general expectations for each of these species and provide additional support that chemotaxonomy (11, 51) may be useful in distinguishing species within the G. fujikuroi species complex.

This report is the first of mycotoxins by the recently described G. konza, which has so far been identified only from Kansas prairie grasses. All of these isolates could produce beauvericin, and eight of nine isolates could produce fusaproliferin at low to moderate levels. Only three of the nine isolates could produce fumonisins, however, and the levels of fumonisins produced were all in the low to moderate range. Qualitatively, this pattern most closely resembles that of G. intermedia, but the levels of the toxins produced by the G. konza isolates were generally less than those produced by the isolates of G. intermedia.

Toxicity to A. salina.

In general, the toxicity, i.e., 50% lethal dose, of strains from species in the G. fujikuroi species complex (26, 35) to A. salina is correlated with the concentrations of beauvericin and/or fusaproliferin in the corresponding organic extracts. For the isolates of G. intermedia examined, toxicity was positively correlated with the observed levels of both fusaproliferin (P = 0.02) and beauvericin (P < 0.001), but it was not correlated with fumonisin production. There are exceptions to this general trend, however. For example, strain D-08411 produced significant levels of beauvericin (120 μg/g) and fusaproliferin (>500 μg/g) but caused only 48% mortality to A. salina (Table 1), even though other isolates producing similar or lower levels of one or both of these compounds were much more toxic. For the isolates of G. konza, toxicity was correlated with beauvericin levels (P < 0.01) but not with either fusaproliferin or fumonisin levels. All of the species except for G. fujikuroi, G. thapsina, and unknown types β and γ contained at least one strain that caused 100% mortality among the brine shrimp.

The data for the G. moniliformis isolates require an alternate explanation, as these isolates make little or no beauvericin or fusaproliferin (Table 1). Thus, the compound responsible for the observed toxicity of these isolates remains to be identified. The results of the present study are consistent with the earlier results of Leslie et al. (23), who found that strains of G. moniliformis that produced neither fumonisins nor moniliformin were still toxic to ducklings. Moniliformin was not evaluated in the present study, although this toxin usually is not produced at high levels by strains of G. moniliformis (9, 48). Thus, it seems likely that at least G. moniliformis is synthesizing additional mycotoxigenic compounds that remain to be identified. Whether the metabolite that is responsible for the toxicity of these strains of G. moniliformis also is responsible for the toxicity of the isolates of unknown type α or the anomalous strains in G. intermedia, e.g., D-08411, also remains to be determined.

In summary, the Fusarium population of the Konza Prairie has a number of similarities, in terms of species present, with those found in the agricultural fields that surround it. It also differs, however, in that there are some apparently unique species present and in the relative frequencies at which the species are recovered. Based on their relative frequency at recovery, we think that G. intermedia and G. konza are native to the prairie and that G. moniliformis and G. thapsina result from introduction from nearby agricultural fields, since the latter species are thought to have host preferences for maize and sorghum, respectively, and were relatively rare in our samples. It also should be possible to evaluate the effect of ecosystem treatments, e.g., burning and grazing, on species frequency and diversity, since comparable areas that are managed in different manners are available within the Konza Prairie Long Term Ecological Research site. In terms of mycotoxin production, the isolates recovered produce toxins at levels similar to other isolates recovered from an agricultural setting. These results are not consistent with a hypothesis that mycotoxins have evolved as a mechanism through which these fungi can more readily colonize an agricultural monoculture. Instead, it suggests that these traits have persisted evolutionarily for an extended period of time due to selection pressures that are not necessarily related to the role these compounds might play in an agricultural host-pathogen interaction and that more potentially mycotoxigenic compounds remain to be identified.