INTRODUCTION

Top Abstract Introduction Materials and methods Results Discussion References

Maize was introduced into East Africa more than 300 years ago and has adapted to diverse conditions of soil, climate, and altitude (15). Maize is the most important cereal crop in Kenya and is used primarily for direct human consumption. In Kenya, about 1.4 million hectares are planted, which yield an estimated 2.8 million tons of grain annually (1). Maize is typically produced by resource-poor smallholder farmers under low-input conditions. Productivity is limited by rainfall and low soil fertility. In addition, an estimated 20 to 40% of the grain is lost nationwide due to pests and diseases (1). Stalk and ear rots caused by a number of fungi not only decrease yields but also have the potential to contaminate grain with mycotoxins that can adversely affect human health (1, 12, 20).

The tropical highlands of western Kenya, bordered on the west by Lake Victoria and on the east by the Great Rift Valley, are a major maize-growing region. Two recent surveys of maize ear rot in western Kenya have found that Fusarium species are the most frequent contaminants (6, 7, 12). Fusarium species from Kenyan maize have been identified by a number of methods, including differences in morphological characters, randomly amplified polymorphic DNA analysis (12), and assignment to mating populations within the Gibberella fujikuroi species complex (6, 7, 12). Overall, the most frequently isolated species was Fusarium moniliforme (synonym, F. verticillioides; G. fujikuroi mating population A), followed by F. graminearum, F. subglutinans (G. fujikuroi mating population E), and other Fusarium species (6, 7, 12). The predominance of F. moniliforme in Kenyan maize is cause for concern because most isolates of this species produce fumonisins, mycotoxins that can cause equine leucoencephalomalacia, porcine pulmonary edema, and experimental liver cancer in rats (13). Furthermore, some studies have associated consumption of maize containing high levels of F. moniliforme and fumonisins with the occurrence of high rates of human esophageal cancer in certain regions of South Africa and China (11, 16, 23, 27).

Fumonisins have been detected in maize and maize-based foods and feeds in North America, South America, Europe, Asia, and South Africa, where extensive survey results have been reported (2, 4, 5, 16, 23, 24, 25, 27). However, there is little information on the occurrence of fumonisins in maize in Kenya or in other countries of sub-Saharan Africa other than South Africa. Comparisons of data from worldwide surveys associate high levels of F. moniliforme infection and fumonisins with drier, warmer climates (27). The relatively warm tropical highlands of western Kenya thus appear to provide suitable conditions for the production of fumonisins in maize. A preliminary survey of good-quality maize from four districts of western Kenya found only low levels of fumonisins (<100 ng/g) in 27 of the 33 samples tested (8, 16). The objective of the present study was a larger and more representative survey of both good-quality and poor-quality maize kernels from smallholder farm storage facilities in nine districts of western Kenya for contamination with Fusarium species and fumonisins.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and methods Results Discussion References

Media. Fusarium species were isolated on peptone-pentachloronitrobenzene agar medium (17), and the isolates were routinely maintained on a modified Czapek-Dox minimal or complete medium (3), while potato dextrose agar was used to culture strains for identification. Carrot agar was used for sexual crosses (9).

Sample collection and isolation of Fusarium species. Shelled maize kernel samples were collected from smallholder farm storage facilities in the districts of Bomet, Bungoma, Kakamega, Kericho, Kisii, Nandi, Siaya, Trans Nzoia, and Vihiga in western Kenya in 1996. The farms were selected randomly from those that had dry maize in their storage facilities (basket granary, cribs, gunny bags, etc.). One or two samples were taken from each of 148 storage facilities: 99 storages were sampled once, and 49 were sampled twice, for a total of 197 samples for fumonisin analysis. Samples from an additional five storage facilities in the Bomet district were analyzed for Fusarium infection but not for fumonisins. The samples comprised good-quality maize (125 samples), with a less-than-34% incidence of visibly diseased kernels, for human consumption and poor-quality maize (72 samples), with a more-than-34% incidence of visibly diseased kernels, for livestock feeding. Most samples were white maize. Sample sizes ranged from 500 g to 1 kg of grain. The proportion of visibly moldy, rotted, or discolored kernels in each of the 197 samples was determined by scoring all kernels in a representative sample of 100 kernels. One sample from each of 153 storage facilities was analyzed for Fusarium infection. Five kernels (each randomly picked from a container) from each sample were surface disinfected by immersion in 1% NaOCl for 30 s, rinsed in sterile distilled water for 20 s, and then transferred to peptone-pentachloronitrobenzene agar medium. These kernels were incubated at 25°C for 5 to 7 days, and one colony per kernel was transferred to potato dextrose agar for identification based on morphology by the system of Nelson et al. (18).

Crossing procedure. Macroconidial morphology, the trait most commonly used to key Fusarium species, is not useful for distinguishing species in section Liseola (teleomorph G. fujikuroi) (10, 18). Members of section Liseola appear to exist as reproductively isolated mating populations. Field isolates can thus be identified by the ability to form fertile perithecia with standard mating population testers. Crosses were made on carrot agar plates (60 by 15 mm) as described by Klittich and Leslie (9) by using standard tester strains (one each of the + and  mating types) of mating populations A (F. moniliforme), B (F. sacchari), C (F. fujikuroi), D (F. proliferatum), E (F. subglutinans), F (F. thapsinum), and G (F. nygamai). Tester strains were kindly supplied by J. F. Leslie, Kansas State University, Manhattan. All crosses were made by using a standard tester strain as the female and the uncharacterized field isolate as the male. Only 2 of the 563 isolates tested in the Liseola section were not fertile, and their identification was based on morphology.

Fumonisin extraction, cleanup, and analysis. Maize kernel samples, 250 to 300 g, were shipped by air express to the National Center for Agricultural Utilization Research, Peoria, Ill. The samples were scored for the presence of visibly moldy and discolored kernels and then stored at 4°C until analysis. The samples were analyzed for fumonisins by high-performance liquid chromatography (HPLC) using standard methods (22, 29). In brief, a 50-g subsample was finely ground in a laboratory mill and thoroughly mixed. Aliquots (5 g) of the ground subsample were extracted with 1:1 acetonitrile-water for 3 h with shaking every 15 min. The extracts were filtered through Whatman 2V filter paper and cleaned up by chromatography on a Bond-Elut strong anion-exchange resin cartridge previously conditioned by the successive passage of methanol (5 ml) and methanol-water (3:1, 5 ml). The cartridge was then washed with methanol-water (3:1, 8 ml), followed by methanol (3 ml), and fumonisins were eluted with 0.5% acetic acid in methanol (14 ml). The eluate was evaporated to dryness under nitrogen and stored at 4°C until analysis. The cleaned extracts were derivatized with ortho-phthalaldehyde immediately before analysis on a Spectra Physics 8700 liquid chromatograph.

	RESULTS

Top Abstract Introduction Materials and methods Results Discussion References

The incidence and geographical distribution of Fusarium species in Kenyan maize are reported in Table 1. F. moniliforme (G. fujikuroi mating population A) was recovered from 60% of the samples overall and was the dominant species in all nine of the districts surveyed. F. graminearum was recovered from 31% of the samples and from eight of the nine districts surveyed. F. solani and F. subglutinans (G. fujikuroi mating population E) were also widespread throughout most districts but were recovered at lower frequencies (18 and 15%, respectively). Other Fusarium species, including F. equiseti and F. oxysporum, were occasionally present.

Fumonisin B₁ levels were above the detection limit of 100 ng/g in 93 (47%) of the 197 maize samples tested (Table 2). The proportion of fumonisin B₁-positive samples ranged from a low of 10% in the Bomet district to highs of 56% in the Vihiga district, 59% in the Kisii district, and 72% in the Bungoma district. The incidence of F. moniliforme in the maize samples from these three districts also was high (71 to 72%), suggesting a trend toward a higher proportion of fumonisin B₁-positive samples in districts with a higher incidence of F. moniliforme. There were several exceptions to this trend, however, including Bomet district samples, which had a relatively high percentage (57%) of F. moniliforme-positive samples but a low percentage (10%) of fumonisin B₁-positive samples (Table 2). The mean fumonisin B₁ levels of the 97 positive samples ranged from 280 ng/g in the Trans Nzoia district to 3,000 ng/g in the Kisii district (Table 2). Fumonisin B₁ levels were above 1,000 ng/g in only 10 samples, 5 of which were collected from farms in the Kisii district. The Kisii samples included one sample (2,100-ng/g fumonisin B₁) of high-quality maize being used for human consumption and four samples (3,600- to 12,000-ng/g fumonisin B₁) of low-quality maize being used for animal feed.

All of the 197 maize samples in this survey were scored for diseased kernels by counting the number of visibly moldy, rotted, or discolored kernels in a representative sample of 100 kernels. To compare fumonisin levels, samples were assigned to four quality grades based on the percentages of diseased kernels. Forty-eight percent of the samples were grade 1 (0 to 25% diseased), 22% were grade 2 (26 to 50% diseased), 9% were grade 3 (51 to 75% diseased), and 21% were grade 4 (76 to 100% diseased). Half of the samples in quality grades 1, 2, and 3 contained fumonisin B₁ at levels above the detection limit of 100 ng/g, but only 3 of the 156 samples in these grades contained fumonisin B₁ at more than 1,000 ng/g. In samples of the poorest quality, grade 4, 17% contained fumonisin B₁ at more than 1,000 ng/g, but 65% contained no detectable fumonisin B₁.

	DISCUSSION

Top Abstract Introduction Materials and methods Results Discussion References

The prevalence of F. moniliforme (G. fujikuroi mating population A) in this survey confirms this fumonisin-producing species as the predominant Fusarium species in Kenyan maize. A field survey for Fusarium in the major maize-growing areas of Kenya in 1993 also found F. moniliforme to be predominant (82% of isolates from maize), followed by F. graminearum (9% of isolates) and F. subglutinans (7% of isolates) (7). Furthermore, in a recent survey of maize grain purchased from market stalls and roadside traders in central and western Kenya, Macdonald and Chapman (12) also reported a high incidence of F. graminearum (9% of the kernels tested) and of "F. moniliforme" (14% of the kernels tested), defined in a broad sense that included several mating populations. In their study, mating population A accounted for 86% and mating population E accounted for 14% of the isolates of "F. moniliforme" from Kenyan maize.

Despite the prevalence of F. moniliforme in maize and the importance of maize as a food staple, there is little information available on the natural occurrence of fumonisins in maize consumed by rural populations in sub-Saharan Africa, with the exception of South Africa. Table 3 compares fumonisin survey data from this study to data from previous surveys of fumonisin levels in African maize. Surveys of maize from rural smallholder farms in the Transkei region of South Africa were conducted in 1985 and 1989 (23). High incidences and levels of fumonisin B₁ were found in both good-quality and moldy maize. Surveys of South African maize grown commercially and for export from 1989 to 1993 found a high incidence of fumonisins but much lower levels than in the maize from smallholder farms in the Transkei region (24, 27). Limited surveys of good-quality maize from hybrids grown in Benin and Zambia in 1992 and from various countries in eastern and southern Africa in 1994 (including one sample from Kenya) also found a high incidence, but low levels, of fumonisins (4, 5). Data from these surveys can be directly compared to data from the present study because all of the fumonisin analyses used the HPLC method of Sydenham et al. (29), and the fumonisin detection limits were similar, either 50 or 100 ng/g.

An overview of the limited data available indicates that fumonisin B₁ levels in maize from smallholder farms in Kenya, with the possible exception of the Kisii district, are generally lower than expected based on the high incidence of F. moniliforme and visibly diseased kernels. These data confirm prior observations of a generally poor correlation between the incidence of F. moniliforme and fumonisin levels in maize collected from smallholder farms in South Africa (16, 23, 24, 27). Another reason for these results is that the maize samples tested are infected with multiple species of Fusarium, all of which cause similar ear and kernel rot symptoms. Thus, kernels exhibiting disease symptoms could be infected with F. moniliforme or F. proliferatum, which do produce fumonisins, with F. subglutinans, which produces little or no fumonisins, or with F. graminearum, F. solani, F. oxysporum, F. equiseti, and other Fusarium species, or with other fungi that do not produce fumonisins (11, 19, 30).

Some studies of fumonisin contamination of maize have indicated that environmental conditions in the area of cultivation play a role in the production of fumonisins in maize (4, 16, 21, 26, 28). It has been observed in the United States that commercial hybrids differ in the tendency to accumulate fumonisins and that hybrids grown outside their adapted range tend to accumulate higher concentrations (26). In the present study, agroecological conditions in the various districts where the maize was cultivated were not determined. The maize genotypes grown in all of the districts except Siaya were primarily of the Hybrid 600 series (Kenya Seed Company, Nairobi) with an identical genetic base. Future studies should investigate the influence of environmental conditions and plant genotypes on fumonisin production in Kenyan maize, and the ability of isolates of F. moniliforme from Kenyan maize to produce fumonisins under controlled conditions in the laboratory and in the field. Furthermore, although the relatively low level of fumonisin contamination of Kenyan maize from smallholder farms is a reassuring finding, some locations yielded maize with unacceptable levels of fumonisins. Our survey data should be useful in estimating the actual exposure to fumonisins of Kenyan populations that depend on maize as their primary source of nutrition. In addition, the widespread presence of F. graminearum and F. subglutinans warrants further surveys for the presence of the mycotoxins, such as deoxynivalenol, zearalenone, moniliformin, and fusaproliferin, that can be produced by these Fusarium species in maize (1, 5, 14, 25).