Factors associated with the incidence of Aspergillus section Flavi and aflatoxin contamination of peanuts in the Busia and Homa bay districts of western Kenya

Authors


E-mail: hkn3@psu.edu

Abstract

The aim of this study was to identify agronomic, ecological and sociocultural factors that could be modified to reduce the risk of aflatoxin contamination of peanuts from western Kenya. Presence of fungi within section Flavi of the genus Aspergillus and levels of total aflatoxin were determined for 436 peanut samples from the Busia and Homa bay districts. A total of 1458 cultures of Aspergillus flavus or A. parasiticus isolated from the samples were assayed for production of aflatoxin B1, B2, G1 and G2. Associations among the incidences of fungal species, incidences of samples with ≥10 μg kg−1 aflatoxin, production of specific aflatoxin types and various agronomic, ecological and sociocultural factors were modelled with chi-squared and logistic regression methods. The predominant species were A. flavus L-strain (78% incidence), A. flavus S-strain (68%) and A. niger (65%). Occurrence of A. caelatus, A. alliaceus and A. tamarii in Kenya was also documented. Samples from the Busia district were three times (odds ratio = 3·01) as likely to contain ≥10 μg kg−1 of total aflatoxin as were samples from the Homa bay district, while samples containing A. flavus S-strain were 96% more likely to exceed this threshold compared with samples from which this fungus was not isolated. Grading, planting improved cultivars and membership of a producer marketing group were negatively associated with the incidence of A. flavus, while crop rotation was negatively correlated with the incidence of B aflatoxins. These sociocultural factors can be modified to reduce the risk of peanut contamination with aflatoxin.

Introduction

Fungal contamination is a major safety concern for many crops including peanuts cultivated in eastern and southern Africa (Siame & Nawa, 2008). Common fungal contaminants of peanuts include Aspergillus, Penicillium, Rhizopus and Fusarium species (Gachomo et al., 2004; Youssef et al., 2008). Most of these fungi produce mycotoxins such as aflatoxin, ochratoxin, zearalenone and cyclopiazonic acid, of which aflatoxin has received the most attention due to its associated health concerns. Not surprisingly, studies addressing fungal contamination of foods in Kenya have focused on aflatoxin contamination of maize, following outbreaks of aflatoxicosis in the eastern parts of the country (Lewis et al., 2005; Probst et al., 2007; Okioma, 2008), and the significance of maize as the national staple.

Aflatoxins are produced mainly by A. flavus, A. parasiticus and to a lesser extent, A. nomius (Dorner, 2002; Vaamonde et al., 2003), which are within section Flavi of the genus Aspergillus (Klich, 2002; Horn, 2007). As many as 16 aflatoxin types have been identified but the most common and naturally occurring are aflatoxins B1, B2, G1 and G2, of which B1 is the most common and potent (Stoloff et al., 1991). Aspergillus flavus and A. parasiticus are ubiquitous fungi that can colonize many crops including maize, peanuts and tree nuts (Hill et al., 1985; Diener et al., 1987; Gachomo et al., 2004). Aspergillus flavus, which is considered the main source of aflatoxin in most commodities, produces the B aflatoxins while A. parasiticus produces both the B and G aflatoxins. Even though not commonly isolated, A. nomius, which has morphological characteristics similar to those of A. flavus, produces both aflatoxin B and G (Vaamonde et al., 2003).

Aspergillus flavus has two distinctive morphological types, the S-strain that is characterized by colonies with sparse mycelia on agar media and production of numerous sclerotia with a diameter of <400 μm, and the L-strain that produces fewer but larger sclerotia (Horn & Dorner, 1998; Cotty & Cardwell, 1999). Recent studies (Probst et al., 2007; Okioma, 2008) have implicated A. flavus S-strains in the production of high levels of aflatoxin in maize in the drought-prone parts of eastern Kenya, which resulted in lethal outbreaks of aflatoxicosis that claimed the lives of over 150 people between 2004 and 2006. Probst et al. (2010) characterized the community structure of Aspergillus section Flavi in maize and soil samples from the aflatoxicosis hotspots and confirmed the S-strain to be the predominant strain, and that a large percentage of the L-strain population was atoxigenic. A recent study of the prevalence of aflatoxin contamination of peanuts from western Kenya showed that peanut samples from the wet, more humid Busia district were more likely to be contaminated than those from the drier Homa bay district (Mutegi et al., 2009), although the composition of fungal species contaminating peanuts and their toxin production profiles was not characterized. Thus, it is not known whether A. flavus S-strain is the predominant contaminant and source of aflatoxin in peanuts, as is the case in maize. Moreover, while A. flavus is still the main source of aflatoxin in peanuts, colonization of peanuts by A. parasiticus is more common than in other crops (Diener et al., 1987), and production of the B aflatoxin types can vary widely among A. flavus isolates from different locations and substrates (Horn & Dorner, 1998, 1999; Probst et al., 2010). There is a need, therefore, to identify the Aspergillus species responsible for the aflatoxin detected in peanut samples in western Kenya.

Several environmental, agronomic and sociocultural factors or practices during pre- and post-harvest handling of peanuts may predispose peanuts to colonization by aflatoxin-producing fungi. For example, contamination has been found to be widespread where peanuts are grown under rain-fed conditions compared to those grown under irrigation (Reddy et al., 2003). End-of-season drought stress and elevated soil temperatures, which are common in sub-Saharan Africa, promote aflatoxin contamination (Rachaputi et al., 2002; Bankole et al., 2006), while differential cultivar susceptibility has also been documented (Reddy et al., 2003; Mutegi et al., 2009). Infestation of peanut pods by pests, and mechanical damage during harvesting, provide avenues for colonization of the nuts by the aflatoxin-producing fungi (Waliyar et al., 2008). Additionally, other mycotoxins such as sterigmatocystin, ochratoxin and zearalenone have been isolated alongside aflatoxins in peanuts (Youssef et al., 2008), suggesting concomitant colonization by other fungi that can affect the aflatoxin production (Horn & Dorner, 1998). Because the options available to small-scale farmers for mitigating aflatoxin contamination of peanuts are limited to agronomic and sociocultural strategies, identifying those factors associated with the incidence of peanut contamination by aflatoxin-producing fungi is essential to developing aflatoxin management strategies.

This study is a follow-up to an earlier study (Mutegi et al., 2009) that identified agronomic, agro-ecological and sociocultural factors associated with the incidence of contamination of peanuts in western Kenya with aflatoxin. Therefore, the objectives of this study were to: (i) determine the incidence of Aspergillus section Flavi fungi, (ii) identify factors associated with the incidence of various fungi, (iii) study associations between the incidence of aflatoxin-producing fungi and total levels of aflatoxin, and (iv) identify the factors associated with the incidence of specific aflatoxin types among A. flavus and A. parasiticus.

Materials and methods

Sampling

Samples and information relating to each sample were gathered through a household survey conducted in western Kenya in 2006. Details of the sampling methodology are published elsewhere (Mutegi et al., 2009). Information was obtained through personal interviews using a pre-tested questionnaire that was developed after conducting focus group discussions involving 40 and 44 participants from the Busia and Homa bay districts, respectively. A 1-kg sample was obtained from each surveyed household, and stored in a coldroom until it was processed. All of the samples were analysed for total aflatoxin content with an indirect competitive ELISA as described previously (Mutegi et al., 2009). Of the 769 samples obtained, 436 samples (252 from Busia and 184 from Homa bay) were randomly selected and assayed for the presence of Aspergillus section Flavi, A. niger, Rhizopus spp. and Penicillium spp. as described below. Isolates of A. flavus and A. parasiticus from each sample were also assayed for production of the four main aflatoxin types: B1, B2, G1 and G2. Ten replicate plates of each sample were used during isolation, and up to 35 A. flavus and A. parasiticus colonies on each plate were screened for production of the aflatoxins as described below.

Isolation and identification of Aspergillus species

Isolation of Aspergillus section Flavi was carried out using modified dichloran rose bengal (MDRB) agar (Horn & Dorner, 1998). The medium was prepared by mixing 10 g glucose, 2·5 g peptone, 0·5 g yeast extract, 1 g KH2PO4, 0·5 g MgSO4.7H2O, 20 g agar and 25 mg rose bengal in 1 L of distilled water. The pH of this medium was then adjusted to 5·6 using 0·01 m HCl. The medium was autoclaved for 20 min at 121°C, and cooled in a water bath at 50°C. To inhibit bacterial growth and ensure the medium was semiselective for Aspergillus section Flavi fungi, 5 mL of 4 mg L−1 dichloran (in acetone), 40 mg L−1 streptomycin and 1 mg L−1 chlortetracycline were added to the medium through a sterile 0·25 μm syringe filter. The medium was then poured on to 90 mm-diameter Petri plates and allowed to settle for 2–3 days before use.

Preparation of samples for plating was performed by thoroughly mixing the 1 kg sample from each household. Two subsamples (100 g each) were blended in a kitchen grinder (Kanchan Multipurpose Kitchen Machine, Kanchan International Limited). Grinders were wiped with 70% ethanol between samples to minimize cross-contamination. From each of the 100 g ground samples, 10 replicates of 0·25 g each were placed in calibrated centrifuge tubes, into which 10 mL of 2% water agar solution was then added and mixed thoroughly. An aliquot of the suspension (0·2 mL) was then pipetted, spread onto MDRB plates, and incubated for 3 days at 30°C, after which the colonies were identified and classified as described below. Total colony counts, and colony counts for A. flavus L-strain, A. flavus S-strain, A. parasiticus, A. alliaceus, A. tamarii, A. niger and Penicillium spp., were recorded. The presence or absence of Rhizopus spp. was also noted.

Czapek yeast extract agar (CYA) was prepared by mixing 1 g K2HPO4, 10 mL Czapek concentrate, 5 g powdered yeast extract, 30 g sucrose and 15 g agar in 1 L of distilled water. The pH of the medium was adjusted to 7·2 before autoclaving for 20 min at 121°C. The medium was allowed to cool in a water bath to 60°C and approximately 20 mL were poured into 90 mm Petri dishes and left to cool overnight under a laminar flow hood. Pure colonies on MDRB agar medium were then streaked onto the CYA plates, and placed into an incubator at 30°C for 7 days. Species of Aspergillus section Flavi were distinguished based on colony colour, texture and conidial morphology characteristics (Klich, 2002), and by comparison with reference strains obtained from Dr B. Horn (USDA National Peanut Research Lab, Dawson, GA, USA).

Screening isolates of A. flavus and A. parasiticus for aflatoxin production

Screening of isolates for aflatoxin production was done in high sucrose yeast extract (YES) liquid medium (Horn & Dorner, 1998). The YES medium was prepared by dissolving 150 g sucrose, 20 g yeast extract (Difco), 10 g soystone (BD Bacto) and 40 g glucose in 1 L distilled water, and the pH adjusted to 5·9 with 0·01 m HCl. Aliquots of 2 mL of the broth were dispensed into 6 mL vials and autoclaved for 30 min at 121°C. Conidia from uncontaminated colonies of A. flavus and A. parasiticus were inoculated into the vials containing 2 mL YES medium. Inoculated vials were incubated in the dark at 30°C for 7 days, with intermittent shaking using a vortex shaker. Subsequently, the vials were removed from the incubator and 2 mL of chloroform was pipetted into each vial. The mixture was vortexed for about 60 s per sample and left to stand overnight in a fume hood. Using a micropipette, 5 μL of the chloroform extract was spotted on silica gel 60 TLC plates (EMD Chemicals Inc.), along with analytical grade standards of aflatoxin B1, B2, G1 and G2. Toxigenic strains were used as positive controls. The plates were developed in chloroform, acetone and distilled water, in a ratio of 88:12:1·5 respectively, until the solvent covered about 90% of the plate length. The plates were transferred to a darkroom and scored for the presence or absence of the four aflatoxins under UV light. This method has a detection limit of 0·5 μg kg−1 (Horn & Dorner, 1998).

Statistical analyses

Incidence of fungal species was determined by means of contingency tables and the number of samples from which a particular species was isolated, recorded as a percentage of the total number of samples assayed in each district. Two binary variables were also constructed by grouping samples in one of two categories (0 or 1) depending on whether they contained: (i) ≥10 colony forming units (CFU) g−1 of A. flavus S-strain for the first variable, or (ii) ≥25 CFU g−1 for A. flavus L-strain for the second variable. These thresholds were based on median counts of colonies of each strain in preliminary analyses. Associations between the incidence of a particular fungal species, CFU g−1 of sample for the A. flavus strains, and various categorical variables were investigated using analysis of contingency tables with appropriate chi-squared tests (Stokes et al., 2000). Associations between the incidence of aflatoxin B1, B2, G1 and G2 and the categorical variables were also assessed with analysis of contingency tables. The categorical variables evaluated in these tests included: (i) district of sample origin, (ii) agro-ecological zone (AEZ) from which sample was collected (three lower midlands AE zones [LM1, LM2 and LM3] in increasing order of dryness as described in Ngugi et al. (2002) were compared), (iii) specific cultivar, (iv) cultivar type (i.e. whether a local landrace or improved), (v) whether or not crop rotation was practised during the growing period of the sample, (vi) whether or not the farmer used commercial fertilizer in peanut production, (vii) farmer’s membership of a producer marketing group (PMG), (viii) number of times a peanut crop was weeded (0, 1–2 or 3 times), (ix) whether or not grading of the nuts was carried out, (x) whether or not the nuts were sorted and (xi) a binary variable for total aflatoxin in samples (0 for <10 μg kg−1 and 1 for ≥10 μg kg−1). The 10 μg kg−1 threshold is the Kenyan 2007 regulatory limit for nuts intended for human consumption (Anonymous, 2007).

The relationship between the incidence of samples with ≥10 μg kg1 and the incidences of aflatoxin-producing Aspergillus spp., the district of sample origin and all factors with significant correlations based on the chi-squared tests was investigated with logistic regression analysis methods. Logistic regression analysis was also used to model the relationship between the incidence of production of specific aflatoxin types (B1, B2, G1 and G2), and the likelihood of a sample’s total aflatoxin exceeding the regulatory limit. In these analyses, the response was the binary variable describing whether or not a sample exceeded the 10 μg kg1 limit. All statistical analyses were carried out using sas v.9·2 (SAS Institute).

Results

Incidence of various fungi recovered from peanuts in the Busia and Homa bay districts

Based on the 436 samples assayed, the most commonly isolated species in decreasing order of incidence were A. flavus L-strain, A. flavus S-strain, A. niger, Rhizopus spp., Penicllium spp., A. parasiticus, A. tamarii, A. alliaceus and A. caelatus (Table 1). With the exception of A. alliaceus, Penicillium spp. and Rhizopus spp., all the fungal species were recovered at relatively similar incidences in peanut samples from the two districts (Fig. 1a). There was a high prevalence of both strains of A. flavus, with an incidence of 77·8% and 69·1% for L- and S-strains, respectively, in the Busia district, while the incidence of these strains in Homa bay was 78·8% and 65·8% respectively (Fig. 1a). Other fungal species observed at high levels of incidence were: A. niger (67·5% and 62% in Busia and Homa bay, respectively), Penicillium spp. (50% and 34·2%), and Rhizopus spp. (50·8% and 33·7%; Fig. 1a). Aspergillus caelatus, A. alliaceus and Atamarii were the least recovered members of section Flavi.

Table 1.   Factors associated with the incidence of Aspergillus and other fungal species isolated from peanut samples obtained from the Busia and Homa bay districts of western Kenya in 2006
Fungal speciesOverall incidence (%)aImproved cultivarb,cMembership of a PMGb,dGradingb,e
YesNoYesNoYesNo
  1. aBased on a total of 436 samples (252 and 184 from the Busia and Homa bay districts, respectively) assayed with 10 replicates analysed for each sample.

  2. bValues are the incidence (%) of a fungal species in samples obtained from growers in each response category. Asterisks indicate a significant association based on chi-squared analysis (*< 0·05; **< 0·01; ***< 0·001).

  3. cCultivars are grouped either as improved in a breeding programme or as local landraces.

  4. dProducer marketing group (PMG): a group of local peanut farmers brought together for the purposes of sourcing markets and to facilitate technology transfer (Mutegi et al., 2007).

  5. eFarmers were asked if they had graded the peanuts crop from which the sample came from.

A. flavus L-strain78·273·281·6*69·683·6***74·385·5**
A. flavus S-strain67·765·968·660·772·0*64·174·3*
A. parasiticus 21·823·520·021·422·021·522·4
A. tamarii 10·8 9·511·810·111·211·39·9
A. caelatus  1·6 1·7 1·6 2·41·1 0·72·1
A. alliaceus  6·0 5·6 5·9 6·65·6 5·37·2
A. niger 65·153·672·9***53·672·4***63·468·4
Penicillium spp.43·444·142·838·146·644·740·8
Rhizopus spp.43·637·448·2*36·947·8*42·944·7
Figure 1.

 Incidence of Aspergillus and other fungal species recovered from peanut samples obtained in a farmer survey in the Busia and Homa bay districts (a), and in three different agroecological zones (b) in western Kenya in 2006. Values are based on 252 and 184 samples from the Busia and Homa bay districts, respectively, and ten replicates were assayed for each sample.

Among the Aspergillus species, only A. parasiticus was isolated at significantly different incidences in samples from the three agro-ecological zones represented in the survey (χ2 = 6·63; = 0·01; Fig. 1b). The incidence of A. parasiticus was lowest in samples obtained from the more humid LM-1 (14·3%) zone, and highest in the dryer LM-3 zone (27·9%), with an incidence of 23·6% from the intermediate LM-2 zone (Fig. 1b). Statistically significant differences were also observed in the incidences of Penicillium spp. and Rhizopus spp. but no obvious pattern associated with levels of AEZ moisture gradients was discernable for these fungi (Fig. 1b).

Factors associated with the incidence of fungi recovered

There was a significant association between cultivar improvement status and A. flavus L-strain (χ2 = 4·31; = 0·038), A. niger2 = 17·20; < 0·001) and Rhizopus spp. (χ2 = 5·07; = 0·026), with improved cultivars having lower incidences of contamination compared to local landraces (Table 1). Membership of a producer marketing group (PMG) was significantly correlated with the presence of four fungi A. flavus L-strain, A. flavus S-strain, A. niger and Rhizopus spp. (Table 1), with higher incidences of the four being recorded among farmers who did not belong to a PMG. For example, while A. flavus L-strain was isolated from 83·6% of the samples from non-PMG farmers, the incidence was reduced to 69·6% among farmers who belonged to PMGs (χ2 = 11·75; < 0·001). Furthermore, while the incidence of A. flavus S-strain in samples from non-PMG farmers was 72·0%, this strain was isolated from only 60·7% of samples from farmers belonging to a PMG (χ2 = 6·01; = 0·014). Grading of peanuts as a post-harvest practice significantly reduced contamination by A. flavus L-strain (χ2 = 7·31; = 0·007) and A. flavus S-strain (χ2 = 4·75; = 0·029; Table 1).

When all the factors associated with the incidence of A. flavus S-strain were evaluated together in a logistic regression model, only the membership of a PMG had a significant effect (Wald χ2 = 6·07; = 0·014). Based on the resulting model, samples from farmers who were not members of a PMG were >1·5 times more likely to be contaminated with A. flavus S-strain compared to peanuts from farmers in a PMG (odds ratio = 1·67; 95% confidence interval (CI) = 1·11–2·52). No significant effect on the incidence of fungi isolated from samples was noted as a result of either sorting or practising crop rotation in the fields where the peanuts were planted (data not shown).

Association between aflatoxin-producing species and total aflatoxin content

The total aflatoxin content was highly variable among samples, ranging from 0 to 2687·6 μg kg−1 and 0–1838·3 μg kg−1 in samples from the Busia and Homa bay districts, respectively. Only 32% of the samples contained detectable levels of aflatoxin. Both the incidence and the number of colonies of A. flavus S-strain were significantly and positively correlated with levels of total aflatoxin in the samples (Fig. 2). The strongest association was between the incidence of A. flavus S-strain and total aflatoxin content; 99·3% of the samples containing <10 μg kg1 of total aflatoxin were not contaminated with A. flavus S-strain (χ2 = 19·59; < 0·001; Fig. 2a). Furthermore, 17·2% of the samples that exceeded the 10 μg kg1 aflatoxin threshold were contaminated with ≥10 CFU g1 of A. flavus S-strain (χ2 = 32·61; < 0·001; Fig. 2c) whereas only one sample (i.e. >1%) exceeding this threshold contained fewer than 10 CFU g1. In spite of the relatively high overall incidence among samples, and although A. flavus L-strain was isolated at a higher frequency from contaminated (>11%) than from safe samples (5%), there was no significant statistical association between the incidence of isolation of this strain and that of the sample contamination (χ2 = 2·88; = 0·089; Fig. 2b). However, a significant association was noted between the number of A. flavus L-strain colonies and the incidence of sample contamination with total aflatoxin. While only 5·9% of samples that had fewer than 25 CFU g1 of A. flavus L-strain were found to have more than 10 μg kg1 of total aflatoxin, 13·4% of samples with ≥25 CFU g1 of A. flavus L-strain were found in the same aflatoxin category (χ2 = 6·98; = 0·008; Fig. 2d).

Figure 2.

 Association between the incidences of Aspergillus flavus S-strain (a & c), or A. flavus L-strain (b & d) with the incidence of samples exceeding Kenya’s total aflatoxin regulation threshold of 10 μg kg−1 of total aflatoxin. Values are based on a total of 436 peanut samples obtained from the Busia (= 252) and Homa bay (= 184) districts of western Kenya in 2006.

Based on logistic regression analysis, the district of sample origin and the incidence of A. flavus S-strain were significantly related to the likelihood of a sample being contaminated (Table 2). Furthermore, samples from the Busia district were three times (odds ratio = 3·01; 95% CI = 1·39–6·53) as likely to contain more than 10 μg kg−1 of total aflatoxin (i.e. deemed unsuitable for human consumption) as samples from the Homa bay district (Table 2). The model also predicted that samples contaminated with A. flavus S-strain were 96% more likely to exceed the aflatoxin regulatory limit compared with samples from which this fungus was not isolated (odds ratio = 0·04; 95% CI = 0·01–0·32; Table 2). Interestingly, the incidences of the other aflatoxin-producing Aspergillus spp. (A. flavus L-strain and A. parasiticus) were not related to the likelihood of a sample containing ≥10 μg kg−1 of total aflatoxin.

Table 2.   Parameter estimates and odds ratio statistics for a logistic regression model relating the district of sample origin and the incidence of sample contamination with Aspergillus flavus S-strain to the likelihood of the total aflatoxin content of a sample exceeding 10 μg kg−1
Factord.f.EstimateSEWald χ2 > χ2Odds ratioa95% CIb
  1. aComparison of the Busia versus Homa bay district, or of samples with no A. flavus S-strain versus samples contaminated with the fungus.

  2. bCI: confidence interval for the odds ratio estimate.

Intercept1−3·5660·52146·92<0·0001
District of sample origin10·5510·1977·810·00523·011·39–6·53
Incidence of A. flavus S-strain1−1·5700·5099·510·00200·040·001–0·32

Incidence of specific aflatoxin types in A. flavus and A. parasiticus isolates

Of the 1458 isolates assayed, 43·1% produced at least one of aflatoxin B1, B2, G1 or G2, with all A. flavus S-strain cultures producing at least one of the aflatoxins. The most common aflatoxin type was B1, followed by B2, G1 and G2, with an incidence of 37·6%, 20·0%, 15·2% and 8·7%, respectively, among the 1458 isolates screened. There was no significant association between the incidences of specific aflatoxin types and AEZ. Numerically more isolates from samples obtained in the Busia district (39·5%) were toxigenic for aflatoxin B1 compared with isolates from samples sourced in the Homa bay district (34·6%) but this correlation was not statistically significant (χ2 = 3·49; = 0·062). However, there was a significant association between the district of sample origin and aflatoxin B2 contamination (χ2 = 12·72; < 0·001), with 23·0% of samples from the Busia district producing B2 compared to 15·3% of samples from the Homa bay district (Fig. 3). There was also a significant association between the district of sample origin and incidence of sample contamination with aflatoxin G12 = 5·48; = 0·019), with 18·4% of samples from the Busia district producing G1 compared to 11·2% of samples from the Homa bay district (Fig. 3). This was in spite of the fact that there was no significant difference in prevalence of A. parasiticus between the districts.

Figure 3.

 Incidence of different aflatoxin types (a), toxigenic isolates (i.e. producing at least one of the four aflatoxin types) (b), isolates toxigenic for B2 toxin (c), or G1 toxin (d) among cultures of A. flavus and A. parasiticus isolated from peanut samples obtained from farmers in the Busia and Homa bay districts of western Kenya in 2006. A total of 1458 fungal isolates from 436 peanut samples (252 and 184 from the Busia and Homa bay districts, respectively) were screened for production of the four toxin types with a thin-layer chromatography protocol. Asterisks in (a) indicate significantly (at < 0·001 and < 0·05, for B2 and G1 respectively) different incidences between the two districts.

Among the factors assessed, only crop rotation and cultivar improvement status were significantly associated with the incidence of production of specific aflatoxin types. Crop rotation was strongly correlated with reduced incidences of the B, but not with the G, aflatoxins. The percentage of A. flavus and A. parasiticus isolates testing positive for B1 and B2 toxins was significantly higher in samples obtained from farmers who did not practise crop rotation compared to those that rotated their crops (Fig. 4). The odds of a sample obtained from a farmer who did not practise crop rotation containing at least one aflatoxin-producing isolate were about one-and-a-half times those of a sample from those practising crop rotation (odds ratio = 1·42; 95% CI = 1·04–1·95; Wald χ2 = 4·85; = 0·028; Table 3). Interestingly, while cultivar improvement status had no effect on the incidence of B aflatoxins, it was strongly correlated with increased production of the G aflatoxins. For example, while 86·3% of cultures that did not produce aflatoxin G1 were isolated from local landrace cultivars, only 80·8% of the isolates from improved cultivars tested negative for G12 = 10·68; < 0·001).

Figure 4.

 Effect of crop rotation on the incidence of aflatoxin B1 (a), and aflatoxin B2 (b) production in isolates of Aspergillus flavus and A. parasiticus from peanut samples obtained from the Busia and Homa bay districts of western Kenyan in 2006. The values are based on a total of 1458 isolates screened for toxin production and the scale of the vertical axis is set from 50% for clarity.

Table 3.   Parameter estimates for a logistic regression model describing the relationship between the incidence of peanut sample contamination with at least one Aspergillus flavus toxigenic isolate, the district of sample origin and crop rotation
Parameterd.f.EstimateSEWald χ2 P > χ2Odds ratioa95% CIb
  1. aComparison of the Busia versus Homa bay district, or of samples from farmers who did not practise crop rotation versus those who rotated peanuts with other crops.

  2. bCI: confidence interval for the odds ratio estimate.

Intercept1−2·1890·0825·260·0218
District10·1790·05610·320·00231·431·15–1·78
Crop rotation10·1770·0804·850·02771·421·04–1·95

The incidence of samples exceeding 10 μg kg−1 of total aflatoxin was significantly (7·39 ≤ χ2 ≤ 9·86; 0·002 ≤  0·007) correlated with the incidence of all aflatoxin types except aflatoxin G2. Samples with isolates producing at least one aflatoxin type also had significantly (χ2 = 18·96; < 0·001) higher levels of total aflatoxin (Fig. 5). For example, 91·7% of the samples containing <10 μg kg1 did not yield aflatoxin-producing isolates compared with 80·2% for samples with at least one toxigenic isolate (Fig. 5). When the incidences of the toxin types were analysed as explanatory variables in a logistic regression model, only the district of sample origin and the presence of at least one toxigenic isolate in the sample were related to the risk of a sample exceeding 10 μg kg−1 of total aflatoxin (Table 4; Wald χ2 = 15·08; < 0·001). This model also showed that the odds of a sample with no toxigenic isolate exceeding 10 μg kg1 of total aflatoxin were about a half (odds ratio = 0·54; 95% CI = 0·39–0·74) those of a sample with at least one toxigenic isolate (Table 4).

Figure 5.

 Association between the incidence of toxin producing isolates (Aspergillus flavus or A. parasiticus) and the incidence of peanut contamination with aflatoxin. Samples were considered contaminated if they exceed 10 μg kg−1 of total aflatoxin, an isolate was rated as toxigenic if it was positive for at least one of the aflatoxin types.

Table 4.   Parameter estimates from a logistic regression relating the likelihood of a peanut sample being contaminated with aflatoxin (>10 μg kg−1) with the district of sample origin and presence of at least one Aspergillus flavus toxigenic isolate
Parameterd.f.EstimateSEWald χ2 P > χ2Odds ratioa95% CIb
  1. aComparison of the Busia versus Homa bay district, or of samples with no toxigenic isolates versus samples with at least one toxin-producing isolate.

  2. bCI: confidence interval for the odds ratio estimate.

Intercept1−2·2570·113397·6<0·0001
District10·8060·113 50·77<0·00015·013·22–7·81
Presence of a toxigenic isolate1−0·3110·080 15·080·00010·540·39–0·74

Discussion

This study investigated the factors associated with prevalence of fungal species in peanuts from western Kenya, focusing on section Flavi of the genus Aspergillus and fungi from other genera that may produce mycotoxins. The predominant species across the two districts (>60% incidence) were A. flavus L-strain, A. flavus S-strain and A. niger. A strong relationship between the incidence of A. flavus S-strain and peanut contamination with aflatoxin was also noted. Although these fungi have previously been isolated at similar levels of incidence in peanuts (Awuah & Kpodo, 1996; Gachomo et al., 2004), this is the first study to quantify the association between the incidences of specific species with levels of aflatoxin in peanut samples in Kenya.

The high incidence of A. flavus-S strain, for which all characterized isolates produce aflatoxin, implies an ever-present risk of aflatoxin contamination for peanuts from western Kenya. The data indicated that contamination with A. flavus S-strain increased the likelihood of a sample having ≥10 μg kg−1 of total aflatoxin by as much as 96% relative to a sample from which this strain was not isolated. In this regard, the data concur with those of Probst et al. (2007) who attributed the risk of maize samples from Kenya being contaminated with aflatoxin almost entirely to colonization by A. flavus S-strain. Although the majority of the peanut samples in this study were safe according to both the EU and KEBS regulatory limits (Mutegi et al., 2009), the high incidence of A. flavus L- and S-strains implies a likelihood of increased aflatoxin levels if safe post-harvest management practices are not followed. Because the factors that trigger aflatoxin production are not well understood, vigilance in pre- and post-harvest handling is needed to minimize the risk of contamination of peanuts with aflatoxin.

The observation that a high incidence of A. flavus L-strain was not positively correlated with levels of total aflatoxin in the samples could be attributed to the fact that most of the L-strain isolates were atoxigenic. Several studies have reported that L-strains tend to be atoxigenic (Horn & Dorner, 1998; Probst et al., 2010). Furthermore, the significant positive correlation between A. flavus L-strain CFU and incidence of sample contamination documented in this study was not unexpected. As the number of CFU per sample increased, so did the likelihood of encountering a toxigenic isolate, although this relationship was not fully quantified.

This is the first study to document the incidence of A. caelatus, A. alliaceus, and only the second to report occurrence of A. tamarii in Kenya. Despite their low prevalence, A. caelatus, A. alliaceus and A. tamarii were isolated from samples from both the Busia and Homa bay districts, and their occurrence at low incidence is in line with the observations of Horn (2005), who documented these species in the USA. Probst et al. (2010) also reported that A. tamarii occurred at a relatively low incidence in maize samples from several provinces in Kenya.

The finding of A. tamarii (which produces cyclopiazonic acid, Horn et al., 1996) and A. alliaceus (which produces ochratoxin A, Varga et al., 1996; Bayman et al., 2002), underscores the need to screen peanuts not just for aflatoxin but also for other mycotoxins. On their own, A. tamari and A. alliaceus pose a minimal risk given their relatively low incidence. Nevertheless, the need to test peanuts from western Kenya for other mycotoxins is supported by the high incidence of A. niger (65%) and Penicillium spp. (43%) in the samples. Indeed, because the semiselective medium used for isolating fungal species in this study is favourable to recovery of Aspergillus section Flavi (Horn & Dorner, 1998), it is possible that the data underestimate the incidence of species such as A. niger. Aspergillus niger and certain Penicillium species produce ochratoxin A (Abarca et al., 1994; Bayman et al., 2002). Certain Penicillium species also produce patulin (Spadaro et al., 2009) and/or citrinin (Singh et al., 2008). The high incidence of these species in peanut samples suggests that such toxins could be present in peanuts from western Kenya, even though their occurrence was not investigated in this study.

The most common aflatoxin type was aflatoxin B1 and the data is in agreement with results of other studies (Lisker et al., 1993; Awuah & Kpodo, 1996). The predominance of B1 can be explained in part by the high incidence of A. flavus S-strain in the samples. As the total aflatoxin levels increased, the incidence of aflatoxin B1 production generally increased, which suggests B1 accounted for most of the toxins in the samples. Furthermore, the presence of G1 and G2 aflatoxins in the samples can be explained by the presence of A. parasiticus. Hill et al. (1985) reported the presence of G1 and G2 aflatoxins in peanut kernels infected with 10–30%A. parasiticus.

Although there were no significant differences in the incidences of A. flavus S- and L-strains or A. parasiticus between districts or agro-ecological zones, there was a significantly higher incidence of strains toxigenic for aflatoxin types B2 and G1 in the Busia district compared to the Homa bay district. When isolates were grouped as either toxigenic or atoxigenic, the incidence of toxigenic isolates was significantly higher in peanuts from Busia than from Homa bay. In addition, the incidence of isolates producing at least one of the B aflatoxins was also significantly higher in peanuts from the Busia district. These results, along with the numerically higher, albeit not statistically significant, incidence of isolates toxigenic for aflatoxin B1 in samples from the Busia district, most likely account for the higher levels of total aflatoxin contamination in peanuts from Busia (Mutegi et al., 2009). Growth conditions for the Aspergillus Flavi fungi in the Busia district (wetter and more humid than the Homa bay district) could have predisposed the peanuts to colonization by the toxigenic strains of these fungi, although other unknown factors cannot be ruled out.

As expected, membership of a PMG was associated with reduced incidence of all but two fungal species assessed. PMGs were established to assist farmers to strengthen their crop marketing abilities and to increase their profit margins, by improving both pre- and post-harvest handling practices (Mutegi et al., 2007). Thus, farmers who belong to PMGs are expected to embrace practices that improve peanut quality and safety through proper drying, grading and storage. Such practices have consistently been shown to reduce the level of contaminated peanuts (Gowda et al., 2002; Turner et al., 2005; Waliyar et al., 2008). It is surprising, however, that the reduction in incidence of fungal contamination did not result in a significant association between PMG membership and levels of total aflatoxin in samples as documented previously (Mutegi et al., 2009), suggesting the role of other factors in production of aflatoxin.

Almost all farmers interviewed sorted their peanuts, so there was no association between sorting and the incidence of fungal species, probably due to the skewed sample sizes. However, there was a significantly higher incidence of both A. flavus L- and S-strains among ungraded peanuts. Grading is mainly conducted for marketed peanuts (Mutegi et al., 2007) and includes assessing parameters such as size of nuts, discoloration and broken nuts: the process is likely to eliminate shrivelled or visibly mouldy nuts. Such grading criteria are, therefore, likely to result in the discarding of nuts contaminated with aflatoxin-producing fungi and have been demonstrated to be negatively correlated with levels of aflatoxins (Waliyar et al., 2008; Mutegi et al., 2009). The fact that grading was not a significant term in a logistic regression model relating membership to a PMG with the incidence of A. flavus S-strain does not contradict this conclusion. Rather, it indicates that, in the model, the effects of grading were confounded with those of PMG membership: a likely outcome given that PMG members are trained to reduce aflatoxin contamination via methods such as grading and sorting (Mutegi et al., 2007).

Cultivar improvement status and crop rotation were significantly associated with the incidences of A. flavus, A. niger and Rhizopus spp. recovered from the samples, and also with the incidence of toxigenic isolates among the A. flavus and A. parasiticus cultures screened. The higher incidence of A. flavus, A. niger and Rhizopus spp. in local landraces compared to the improved varieties is consistent with observations in previous studies that documented higher incidences of fungal contamination among landrace peanut varieties than improved cultivars elsewhere (Middleton et al., 1994). Variety improvement programmes are generally tailored to reducing susceptibility to diseases, and this could explain why improved varieties were likely to have a lower incidence of fungal contamination compared to the local varieties. Crop rotation may lower the rate of between-season survival of different species and/or strains, especially if it involves crops that are non-host to Aspergillus species, but this hypothesis was not investigated in the present study.

In summary, the results demonstrate a high risk of aflatoxin contamination of peanuts in western Kenya, especially in the more humid areas of the Busia district. Furthermore, the incidence of peanut contamination with aflatoxin in the region can be accounted for mainly by a high incidence of the toxigenic A. flavus S-strain. However, the occurrence of other toxigenic Aspergillus species underscores the need to screen peanuts from the region for other mycotoxins in addition to aflatoxin. In the meantime, the risk of peanut contamination and subsequent human exposure to aflatoxin can be mitigated by practising crop rotation, planting improved cultivars and grading the harvested nuts.

Acknowledgements

We thank the data collection team that included staff from the Ministry of Agriculture, Kenya and Catholic Relief Services (CRS). We also thank CABI for allowing use of their laboratory facilities. Special thanks go to Lucy Karanja for technical assistance in the laboratory. This study was jointly funded by ICRISAT and Kenya Agricultural Research Institute (KARI).

Ancillary