Fusarium Species from Nepalese Rice and Production of Mycotoxins and Gibberellic Acid by Selected Species

Abstract

Infection of cereal grains with Fusarium species can cause contamination with mycotoxins that affect human and animal health. To determine the potential for mycotoxin contamination, we isolated Fusarium species from samples of rice seeds that were collected in 1997 on farms in the foothills of the Nepal Himalaya. The predominant Fusarium species in surface-disinfested seeds with husks were species of the Gibberella fujikuroi complex, including G. fujikuroi mating population A (anamorph, Fusarium verticillioides), G. fujikuroi mating population C (anamorph, Fusarium fujikuroi), and G. fujikuroi mating population D (anamorph, Fusarium proliferatum). The widespread occurrence of mating population D suggests that its role in the complex symptoms of bakanae disease of rice may be significant. Other common species were Gibberella zeae (anamorph, Fusarium graminearum) and Fusarium semitectum, with Fusarium acuminatum, Fusarium anguioides, Fusarium avenaceum, Fusarium chlamydosporum, Fusarium equiseti, and Fusarium oxysporum occasionally present. Strains of mating population C produced beauvericin, moniliformin, and gibberellic acid, but little or no fumonisin, whereas strains of mating population D produced beauvericin, fumonisin, and, usually, moniliformin, but no gibberellic acid. Some strains of G. zeae produced the 8-ketotrichothecene nivalenol, whereas others produced deoxynivalenol. Despite the occurrence of fumonisin-producing strains of mating population D, and of 8-ketotrichothecene-producing strains of G. zeae, Nepalese rice showed no detectable contamination with these mycotoxins. Effective traditional practices for grain drying and storage may prevent contamination of Nepalese rice with Fusarium mycotoxins.

Rice (Oryza sativa L.) is a major food crop in the foothills of the Nepal Himalaya. In the subtropical zone between altitudes of 1,000 and 2,000 m, rice is usually grown as a monsoon season crop from June to October. Rice may have been domesticated from wild species in Nepal; rice cultivation is described in Nepalese texts dating to 2,800 B. C. E. (3). Traditional farmers in the Himalayan foothills grow diverse local varieties of rice. Improved rice varieties also have been introduced into the foothills, and some are grown extensively, especially in the agriculturally developed Kathmandu valley and adjoining regions. At harvest, rice is threshed by traditional methods that separate the grain from stems and panicals, but not from husks. To control storage diseases, grain is sun dried to decrease its moisture content and is left in the husk for long-term storage.

Both local and improved varieties of rice are affected by a number of fungal diseases, including bakanae disease, which is widespread in the foothills (17, 19). Bakanae disease is caused by one or more Fusarium species and is a complex of disease symptoms including seedling blight, root and crown rot, stunting, and the classic symptoms of etiolation and abnormal elongation induced by fungal production of gibberellin hormones (27, 33, 35). In Khumal-4, a variety introduced for its high resistance to blast, grain yield losses of up to 40% due to bakanae disease have been recorded (20). Bakanae disease is primarily seed borne, and high levels of seed infection with Fusarium species have been found in samples of Khumal-4 from affected fields (19).

Although bakanae disease was first described more than 100 years ago in Japan, it is still not clear which Fusarium species are associated with the varied symptoms of the disease. Early research in Japan identified the pathogen as “Fusarium moniliforme” in a broad sense (30); however, this taxon comprises a number of distinct species, now collectively termed the Gibberella fujikuroi species complex. Formation of the Gibberella sexual stage can distinguish mating populations, or biological species, within this group (9, 13–15). Three mating populations of the G. fujikuroi complex have been associated with bakanae-diseased rice. Mating population C (MP-C) (anamorph, Fusarium fujikuroi) (28) was first identified in 1977 among strains from rice from Taiwan (9). More recently, mating population A (MP-A) (anamorph, Fusarium verticillioides [synonym, F. moniliforme]) and mating population D (MP-D) (anamorph, Fusarium proliferatum) have been isolated from rice from Asia, and MP-D has been isolated from rice from Africa, Australia, and the United States (2, 7, 34). Thus, more than one species of Fusarium may be able to infect rice and cause symptoms of bakanae disease.

Infection of cereal grains with Fusarium species can cause contamination with mycotoxins that affect human and animal health (22, 24). Species within the G. fujikuroi complex differ in production of mycotoxins, including beauvericin, fumonisins, and moniliformin (15, 16). We previously found that strains of G. fujikuroi MP-D from rice from several geographical areas produced fumonisins, but that strains of MP-C from rice from Taiwan did not (7). Contamination of grain with fumonisins has been associated with human esophageal cancer in South Africa (24). Fumonisins also cause leucoencephalomalacia in horses, lung edema in swine, and cancer in experimental animals (24). Mycotoxin-producing species of Fusarium isolated from rice in Asia include Gibberella zeae (anamorph, Fusarium graminearum) which causes rice ear scab in China, India, and Japan (30) and produces the 8-ketotrichothecene mycotoxins nivalenol and deoxynivalenol (22). Contamination of grain with trichothecenes has been associated with hemorrhagic syndromes in animals and with human disease epidemics in Eastern Europe and Japan (22).

Despite the importance of rice as a food grain and the reported high frequency of bakanae disease, there have been no surveys of Fusarium mycotoxins in Nepalese rice. Our objectives in this study were to determine the incidence of Fusarium species in representative samples of bakanae-resistant and -susceptible rice from the same geographical area and to determine the natural occurrence of fumonisin B₁, nivalenol, and deoxynivalenol in rice grain used for human consumption in Nepal. For strain characterization, traits included morphology, mating population tests, production of the mycotoxins beauvericin, fumonisins, moniliformin, and trichothecenes, and production of gibberellic acid. This study is the first report of mycotoxigenic Fusarium strains and the first survey of Fusarium mycotoxins in Nepalese rice.

MATERIALS AND METHODS

Rice seed sample collection.

Rice seeds with husks, approximately 0.5 kg per sample, were collected in February and March 1997 from smallholder farms and from Nepal Agricultural Research Stations in the foothill region of central Nepal (Lamjung district) and in the Kathmandu valley and adjoining region of east-central Nepal (Bhaktapur, Dolakha, Kathmandu, Kavre, and Lalitpur districts) (Table 1, footnote a).

TABLE 1.

Fusarium species isolated from Nepalese rice samples

Fungal speciesb	No. of samples infected with the indicated speciesac			No. of strains identified from each group
	Group I	Group II	Group III	Group I	Group II	Group III
Total no. of samples	18	8	22	67	100	42
G. fujikuroi species complex	15	8	11	36	24	15
G. fujikuroi MP-A	0	0	2	0	0	3
G. fujikuroi MP-C	6	0	0	7	0	0
G. fujikuroi MP-D	5	7	6	6	13	8
G. fujikuroi, nonfertile	13	7	4	23	11	4
G. zeae	3	7	7	14	34	9
F. graminearum, nonfertile	1	1	3	1	1	3
F. semitectum	4	7	5	13	36	9
F. equiseti	1	3	2	1	3	3
F. chlamydosporum	1	1	0	1	1	0
F. oxysporum	0	0	2	0	0	2
F. avenaceum	1	0	0	1	0	0
F. acuminatum	0	1	0	0	1	0
F. anguioides	0	0	1	0	0	1
None (clean sample)	2	0	8

^aGroup I comprised 18 samples of the bakanae-susceptible, improved variety Khumal-4 from the Kathmandu valley and the following adjoining regions: Kavre district (four samples), Dolakha district (six samples), and Lalitpur district (eight samples). Group II comprised eight samples of bakanae-resistant varieties from the Kathmandu valley and the following adjoining regions: two samples each of improved varieties Chainung-242 and Taichung-176 from Bhaktapur district, one sample of local variety Sankharika from Lalitpur district, and one sample each of improved varieties IR51672 and Masuli from Kavre district. Bakanae susceptibility or resistance rating was based on field performance. Group III comprised 22 samples of local varieties from Lamjung district, whose bakanae susceptibility or resistance was unknown. 

^bSurface-disinfested seeds with husks (50 seeds from each sample), were placed on a selective medium. For all samples except 14 Lamjung samples (Group III) that showed no infection, an additional 50 to 75 seeds were tested. For 29 of the 38 infected seed samples, every Fusarium colony was identified to species. For the nine most highly infected seed samples, colonies were placed into groups by morphological criteria, and representative colonies were identified to species. 

^cNumbers in columns sometimes exceed the total numbers of samples due to infection of some samples with multiple Fusarium species. 

Isolation and identification of Fusarium strains.

There were three different seed treatments, with 50 seeds per treatment per sample unless indicated otherwise: seeds with husks, untreated; seeds with husks, surface disinfested by placing in 1% NaOCl for 1 min and rinsing twice in sterile water; and seeds with husks removed, then surface disinfested. Seeds were placed on the surface of a plate of Fusarium-selective agar medium containing pentachloronitrobenzene (25). Seeds were incubated for 5 to 7 days, then one colony per seed was reisolated from a single spore and identified by morphology when grown in carnation leaf agar and potato dextrose agar (25). Strains identified as F. graminearum were tested for production of the G. zeae sexual stage on carrot agar (4). Seed infection levels were compared by single-factor analysis of variance (SAS Institute Inc., Cary, N.C.). Representative strains of the species collected are available from the Fusarium Research Center, The Pennsylvania State University, University Park, including accession numbers M-8496 to M-8534 for strains of the G. fujikuroi species complex.

Seventy-five strains identified by morphology to the G. fujikuroi species complex were crossed as males with female fertile tester strains of G. fujikuroi MP-A, MP-C, and MP-D. Strains were not systematically tested for female fertility. The strains were tested for male fertility once on carrot agar (11) and once on slants of 1% water agar with pieces of autoclaved rice straw. Both of these substrates have been reported to be favorable for crossing MP-C, but carrot agar was usually more effective than rice straw in this study (9, 34). Mating tests were incubated in an alternating 12-h, 25°C light/20°C dark schedule. Strains were scored as fertile if they produced ascospores in one or more tests. Selected nonfertile strains were tested further on carrot agar with tester strains of MP-F (Fusarium thapsinum) (12), and with various female fertile field strains and ascospore progeny of MP-C and MP-D that were identified in the initial mating population tests. The MP-A tester strains were M-3125 and M-3120, the MP-C tester strains were M-6883 and M-6884, the MP-D tester strains were M-6992 and M-6993, and the MP-F tester strains were M-6563 and M-6564 (11, 13, 15). Strains preceded with an M refer to the accession numbers at the Fusarium Research Center. Female fertile strains HKM35, HKM41, and HKM28 identified in this study have been deposited with the Fungal Genetics Stock Center, University of Kansas Medical Center, Kansas City, Kans., as accession numbers 8381, 8382, and 8383, respectively.

Chemical analysis of fungal cultures.

The production of beauvericin, fumonisins, gibberellic acid, moniliformin, and trichothecenes was assessed in cultures grown on autoclaved rice. In brief, 30 g of polished rice and 9 ml of distilled water were autoclaved in a 250-ml Erlenmyer flask. Inoculum was prepared by adding 7 ml of sterile water to a petri dish (diameter, 10 cm) containing a 10- to 14-day-old culture of the test strain growing on V-8 juice agar (32). Each flask of rice medium was inoculated with 5 ml of spore suspension, mixed well, and incubated for 28 days at 25°C in the dark. For chemical analyses, the entire contents of each flask of culture material was extracted with 150 ml of acetonitrile-water (1:1) with occasional shaking. The extract was filtered through Whatman no. 2V filter paper and was stored at 5°C until used. The levels beauvericin, fumonisins, gibberellic acid, and moniliformin were determined directly from aliquots of culture extract without further sample clean up. Results are given as micrograms of mycotoxin per gram of rice, based on the starting dry weight of polished rice.

Fumonisins were determined by high-performance liquid chromatography (HPLC) of orthophthalaldehyde derivatives as reported previously (24). The detection limit for fumonisin B₁ in culture material was approximately 1 μg/g.

Beauvericin was determined by HPLC-mass spectrometry (HPLC/MS) with electrospray ionization. Typically, 10 μl of culture extract was diluted to 1 ml with acetonitrile-water (1:1). Ten microliters of this working solution was injected onto the HPLC column (5 μm Intersil octyldecyl silane (ODS)-3 [column dimensions, 15 cm by 3 mm]; MetaChem Technologies, Torrance, Calif.) and was eluted with methanol-water (95:5) at 0.3 ml/min. The effluent from the column was introduced to the mass spectrometer through the electrospray ionization interface, and beauvericin was detected by comparing retention time and mass spectrum with an authentic standard. Quantitation was based on response for the sodiated molecular iron at m/z 806. When diluted extracts were negative for beauvericin, a 10-μl aliquot of the undiluted extract was injected. The detection limit for beauvericin in culture material was approximately 1 μg/g.

Cultures were analyzed for the presence of gibberellic acid (GA₃), the end product of the fungal gibberellin biosynthetic pathway, by using the same extract, HPLC/MS system, and column as described above for beauvericin, except that the column was eluted with methanol-water-glacial acetic acid (40:60:0.3). Detection and quantitation were based on retention time and spectrum in comparison with authentic gibberellic acid. The detection limit for gibberellic acid was approximately 2 μg/g.

Moniliformin was determined by HPLC by using a 250- by 3-mm Intersil ODS-2 column (MetaChem Technologies) and a UV diode array detector. The solvent system was acetonitrile-water (13:87) with the addition of 10 ml of an ion pair solution consisting of 1.1 M KH₂PO₄–tetrabutyl-ammonium hydroxide (2:1) per liter. The retention time for moniliformin was approximately 10 min. The level of moniliformin in the extract was calculated from chromatograms of absorbance at 230 nm by comparing the retention time and area of the peak to a moniliformin standard peak. The full UV spectrum confirmed the presence of moniliformin at levels greater than 10 μg/g in culture material, amounts below that level were reported as not detected.

Trichothecenes were determined by HPLC/MS by using the column, detector, and flow rate described above for beauvericin. The trichothecenes were detected in the negative ion mode. The column was solvent programmed from methanol-water-glacial acetic acid (1:99:0.01) to methanol-water-glacial acetic acid (33:67:0.3) in 10 min. Nivalenol and deoxynivalenol were identified by retention time compared with injection of authentic standard solutions. Quantitation was based on response of the (M − H + HOAc)⁻ adduct ion (the molecular ion minus hydrogen plus acetic acid). The detection limits for nivalenol and deoxynivalenol were 1 μg/g.

Mycotoxin analysis of rice seed samples.

Selected rice seed samples (10 g) were ground and extracted with acetonitrile-water and analyzed for nivalenol and deoxynivalenol by HPLC/MS as described above. For fumonisin analyses, which were conducted at the Nepal Agricultural Research Council, 100 g of each of the 48 rice seed samples were ground to a powder of similar consistency in a coffee grinding mill or on a grinding stone. Two 50-g aliquots of each sample were mixed with 200 ml of tap water in a 1-liter erlenmeyer flask or in a glass jar of similar volume. The samples were extracted at room temperature for 16 to 20 h with occasional mixing then filtered through a fast filter paper such as Whatman no. 1. Filtrates were assayed immediately by enzyme-linked immunosorbent assay (ELISA) without further dilution, and aliquots of samples were frozen for reanalysis. Each of the two sample aliquots was analyzed in duplicate. Toxin recovery using this extraction method was determined periodically during the survey by extracting a naturally contaminated standard whose level of fumonisins had been previously determined by HPLC.

Monoclonal antibodies from clone P2A5-3-F3 mouse ascites fluid and fumonisin-horseradish peroxidase prepared from fumonisin B₁ were used in a direct ELISA as previously described (21). Fumonisin concentration was determined from standard curves of fumonisin B₁ diluted (in an extract of low fumonisin maize, <0.2 μg/g) equivalent to 16, 4, 1, 0.3, and 0 μg/g. The detection limit of the ELISA was 1 μg of fumonisin per g of rice sample.

RESULTS

Occurrence of Fusarium species in rice seed.

All 48 rice samples in this study (Table 1) were collected from storage as seeds with husks. All untreated samples of seeds with husks and 60% of the individual untreated seeds with husks were infested with Fusarium species. Recovery of Fusarium species from untreated seeds from Lamjung district was low due to infestation with fast-growing saprophytic fungi, especially Trichoderma species. Seed treatment by surface disinfestation with hypochlorite removed saprophytic organisms found on the seed surface and decreased the overall incidence of Fusarium species to 70% of samples tested and 8% of seeds tested, but allowed better recovery of species that were primary invaders. Dissection to completely remove the husk, and often the attached germ, further decreased the incidence of Fusarium species to 40% of samples tested and 3% of seeds tested, indicating that endosperm is relatively free of internal infection.

Surface-disinfested seeds with husks were used to determine the incidence and distribution of Fusarium species in Nepalese rice (Table 1). Fusarium strains were recovered from 79% of the 48 samples. Species of the G. fujikuroi complex were recovered from 70%, F. graminearum from 40%, and Fusarium semitectum (synonym, Fusarium pallidoroseum) from 30%, and Fusarium equiseti from 12% of the 48 samples. Five other Fusarium species were occasionally recovered: Fusarium chlamydosporum, Fusarium oxysporum, Fusarium avenaceum, Fusarium acuminatum, and Fusarium anguioides. Although the incidence of sample infection with Fusarium species was high, seed infection levels were 10% or lower in 37 of the 48 samples tested. The level of seed infection with Fusarium species ranged from 0 to 70% with a mean of 14% in sample group I, from 6 to 49% with a mean of 16% in sample group II, and from 0 to 7% with a mean of 1% in sample group III. The level of seed infection with species of the G. fujikuroi complex ranged from 0 to 70% with a mean of 10% in sample group I, from 1 to 11% with a mean of 4% in sample group II, and from 0 to 2% in sample group III. Because of variation among samples within each group and the small sample size, none of the seed infection levels were significantly different between sample groups, except that the frequency of Fusarium infection was significantly (P < 0.005) lower in sample group III than in sample groups I and II.

Characterization of G. fujikuroi from rice seed.

Seventy-five strains from rice were identified as members of the G. fujikuroi species complex based on standard morphological traits, including abundant single-celled microconidia in long chains and in false heads, long, delicate macroconidia; and the absence of chlamydospores (25). The 75 strains were characterized for mating population by crossing to tester strains of MP-A, MP-C, and MP-D. Three fertile strains of MP-A were identified in samples from Lamjung, but this species was not found in samples from the Kathmandu valley and adjoining districts. Six strains that were identified by morphology as F. verticillioides and were not fertile with MP-A, MP-C, or MP-D also were not fertile with tester strains of MP-F.

In crosses with MP-C tester strains, four fertile strains were identified. Three additional fertile strains of MP-C were identified by crosses between one of these four fertile strains, HKM34, and 10 nonfertile strains. Two strains were MATC-1 and five were MATC-2. Most of the MP-C crosses were of low fertility with few mature perithecia per plate. Only crosses of strain HKM34 as a male to strains M-6884 and HKM41 as females were highly fertile, with abundant perithecia and ascospores. Fertile strains of MP-C were isolated only from the bakanae-susceptible cultivar Khumal 4, sample group I (Table 1).

In crosses with MP-D tester strains, 18 fertile strains were identified. Nine additional fertile strains of MP-D were identified in crosses between 50 nonfertile strains and a female fertile ascospore progeny of a cross between strains M-6992 and HKM55. Seventeen strains were MATD-1 and 10 strains were MATD-2. Eleven fertile strains were crossed with each other to test for female fertility; only strain HKM28 was female fertile. Most of the MP-D crosses were highly fertile, with dozens to hundreds of mature perithecia per plate. Viability of a random sample of 50 ascospore progeny from each of eight different crosses averaged 62 ± 18%. Fertile strains of MP-D were isolated from all three rice sample groups (Table 1).

Metabolite profiles of the seven strains of MP-C, 15 of the 27 strains of MP-D, and seven of the 38 nonfertile strains of the G. fujikuroi species complex were determined (Table 2). Strains of MP-D and nonfertile strains were selected to represent 22 different rice samples. Strains of MP-C produced beauvericin, moniliformin, and gibberellic acid, but little or no fumonisin. In contrast, none of the 15 strains of MP-D produced gibberellic acid, but all produced beauvericin and fumonisin, and 11 of 15 strains produced moniliformin. Nonfertile strain HKM38 (Table 2, strain 16) produced none of the tested metabolites. The remaining six nonfertile strains produced beauvericin, moniliformin, and gibberellic acid, but little or no fumonisin—a metabolite profile very similar to MP-C. Both fertile strains of MP-C and nonfertile, gibberellic acid-producing strains were recovered from sample group I, but only nonfertile, gibberellic acid-producing strains were recovered from sample groups II and III.

TABLE 2.

Mating populations and metabolite profiles of strains of the G. fujikuroi species complex from Nepalese rice

Group and strain no.	Mating population and mating type	Production (μg/g) of metabolites in rice cultureb
		GA3	FUM	MON	BEA
Rice sample group Ia
1	MATC-2	390	ND	3,100	650
2c	MATC-2	1,590	ND	2,020	>1,000
3	MATC-2	510	7	3,350	>1,000
4c	MATC-1	440	ND	3,460	>1,000
5	MATC-1	450	ND	4,340	>1,000
6	MATC-2	570	ND	520	>1,000
7c	MATC-2	240	3	1,340	>1,000
8	NFd	610	ND	200	>1,000
9	NF	330	3	3,970	>1,000
10	NF	110	ND	3,020	450
11c	NF	190	ND	2,060	>1,000
12	MATD-2	ND	520	760	>1,000
13	MATD-1	ND	340	ND	>1,000
14	MATD-1	ND	9	200	760
15	MATD-2	ND	800	310	>1,000
Rice sample group II
16c	NF	ND	ND	ND	ND
17	NF	350	ND	3,380	>1,000
18	MATD-1	ND	1,140	ND	>1,000
19c	MATD-1	ND	1,570	1,700	>1,000
20	MATD-1	ND	620	ND	200
21	MATD-1	ND	390	200	>1,000
22c	MATD-1	ND	2,980	1,470	>1,000
23	MATD-1	ND	370	80	>1,000
24	MATD-1	ND	240	560	>1,000
Rice sample group III
25	NF	120	ND	140	6
26	MATD-2	ND	730	170	>1,000
27	MATD-1	ND	20	430	570
28	MATD-2	ND	810	150	570
29	MATD-2	ND	400	ND	>1,000
30	MATD-2	ND	2,200	3,690	>1,000

^aRice sample groups as in Table 1. 

^bND, not detected. Detection limits for gibberellic acid (GA₃), fumonisin B₁ (FUM), moniliformin (MON), and beauvericin (BEA) were 2, 1, 10, and 1 μg/g, respectively. 

^cMetabolite data are from one culture of each strain, except seven strains which were grown twice and averaged. 

^dNF, not fertile. 

Characterization of G. zeae from rice seed.

Sixty-two strains were identified as F. graminearum based on morphology. Fifty-seven of these strains were self fertile and produced perithecia of the G. zeae sexual stage in culture. The five nonfertile strains were confirmed as F. graminearum by PCR amplification with species-specific primers (27). G. zeae was recovered from all three rice sample groups. Six representative strains of G. zeae were tested for their ability to produce trichothecenes. Two strains produced nivalenol with little or no deoxynivalenol, three strains produced deoxynivalenol with no nivalenol, and one strain produced trace levels (approximately 0.5 μg/g) of both toxins.

Natural occurrence of fumonisins, nivalenol, and deoxynivalenol in rice seed.

All 48 rice samples were analyzed for fumonisins by competitive immunoassay by using an antibody that detects fumonisin B₁, fumonisin B₂, and fumonisin B₃. Fumonisins were below the detection limit of 1 μg/g in all samples tested. Eight rice samples that were infected with G. zeae were analyzed for nivalenol and deoxynivalenol, both of which were below the detection limit of 1 μg/g in all samples tested.

DISCUSSION

Rice seed samples collected in Nepal were frequently infected with Fusarium species, but seed infection levels were generally less than 10% after treatment by surface disinfestation. One of the most frequently isolated species was G. zeae, which is well established as the causal agent of rice ear blight in China, India, and Japan (30). G. zeae strains from rice and other cereal grains in Nepal (6), in common with strains from Japan, can be divided into two approximately equal groups: nivalenol producers and deoxynivalenol producers. In other regions of the world, such as North America and Europe, nivalenol-producing strains of G. zeae are usually rare (22). Other species found in several of the samples include F. semitectum and F. equiseti, both of which are generally weak pathogens common to subtropical areas and have previously been isolated from rice in Nepal (17) and in other Asian countries (5, 8, 30, 35). Other species recovered in low numbers were the common cereal pathogens F. acuminatum and F. avenaceum, and F. chlamydosporum and F. oxysporum, which have previously been isolated from rice from Asia (5, 30). One rice sample from Lamjung district yielded the rare species F. anguioides, which has recently been neotypified based on strains collected from soil in China (26). The last five Fusarium species identified are new records on rice from Nepal.

In the present study, the most frequent Fusarium isolates from Nepalese rice were species of the G. fujikuroi complex. G. fujikuroi MP-A was represented by three strains isolated from local rice varieties from Lamjung district. This mating population is commonly associated with Nepalese maize (2, 6), but there have been few reports of its isolation from rice (15, 34). A cropping pattern of rice after maize on small farms in the foothills may facilitate movement of this pathogen from maize to rice. Twenty-eight strains of MP-A isolated from maize from Lamjung district produced the high levels of fumonisins that are typical of this species (6).

G. fujikuroi MP-C, the classic, gibberellin-producing, bakanae pathogen (anamorph, F. fujikuroi), was represented by seven strains isolated from the bakanae-susceptible variety Khumal-4. The ability of these Nepalese strains of MP-C to produce beauvericin, gibberellic acid, and moniliformin is consistent with previous reports of secondary metabolite production in this species (7, 21, 23). The present survey also identified a number of apparently nonfertile strains that share the unique secondary metabolite profile of MP-C, producing beauvericin, gibberellic acid, and moniliformin, but little or no fumonisin. Additional genetic analyses will be necessary to establish the relationship of these gibberellic acid-producing nonfertile strains to MP-C. In a previous study, 14 strains that produced high levels of gibberellic acid, whether fertile or nonfertile, were assigned to MP-C by DNA marker analysis (34). Poor fertility may indicate variation within the species, or may simply mean that optimal conditions for laboratory crosses of MP-C have not been identified.

G. fujikuroi MP-D was isolated from 40% of the Nepalese rice samples and from all three sample groups. Analysis of 15 strains of MP-D confirmed previous reports that this species does not produce high levels of gibberellins, but does produce beauvericin, fumonisins, and moniliformin (7, 16, 21, 23). The relationship between MP-C and MP-D from rice and the role of these species and their metabolites in bakanae disease remain unclear. Genetic and molecular phylogenetic analyses indicate a close relationship between MP-C from rice and MP-D from a wide range of host plants (7, 29, 34).

Nepalese rice showed no contamination with fumonisins, nivalenol, or deoxynivalenol above the detection limit of 1 μg/g, despite the occurrence of strains of G. fujikuroi MP-D and of G. zeae that could produce high levels of these mycotoxins. Limited surveys of rice grain in the United States have detected fumonisins in some samples (1, 31). The low level of mycotoxin contamination of Nepalese rice from smallholder farms may be due to traditional postharvest practices such as sun drying to lower moisture content and winnowing to remove seeds that are lighter in weight due to poor grain fill or disease. Both sun drying and winnowing reduced levels of seed infection with Magnaporthe grisea (anamorph, Pyricularia oryzae) (19) and are likely to similarly reduce levels of seed infection with Fusarium. For long-term storage, rice is usually left in the husk, which helps protect the grain from insects and fungi. In farmhouses, rice is stored on rafters and in attics to allow for air circulation which helps keep the grain dry.

In conclusion, 11 Fusarium species were isolated from Nepalese rice. The fumonisin-producing species G. fujikuroi MP-D and the trichothecene-producing species G. zeae were widespread, but rice samples were not contaminated with these mycotoxins. The gibberellin-producing species G. fujikuroi MP-C is commonly associated with bakanae disease of rice and was present in seeds of the bakanae-susceptible variety Khumal-4 grown in Nepal. A wide range of rice samples also were infected with gibberellin-nonproducing strains of G. fujikuroi MP-D and with nonfertile, gibberellin-producing strains of the G. fujikuroi species complex. Further analyses of phenotypic and genetic variation within and between these populations of G. fujikuroi, and pathogenicity tests in particular, will be necessary to resolve their genetic relationship and relative importance in producing the complex symptoms of bakanae disease.