Aspergillus section Flavi community structure in Zambia influences aflatoxin contamination of maize and groundnut

Aflatoxins are cancer-causing, immuno-suppressive mycotoxins that frequently contaminate important staples in Zambia including maize and groundnut. Several species within Aspergillus section Flavi have been implicated as causal agents of aflatoxin contamination in Africa. However, Aspergillus populations associated with aflatoxin contamination in Zambia have not been adequately detailed. Most of Zambia's arable land is non-cultivated and Aspergillus communities in crops may originate in non-cultivated soil. However, relationships between Aspergillus populations on crops and those resident in non-cultivated soils have not been explored. Because characterization of similar fungal populations outside of Zambia have resulted in strategies to prevent aflatoxins, the current study sought to improve understanding of fungal communities in cultivated and non-cultivated soils and in crops. Crops (n = 412) and soils from cultivated (n = 160) and non-cultivated land (n = 60) were assayed for Aspergillus section Flavi from 2012 to 2016. The L-strain morphotype of Aspergillus flavus and A. parasiticus were dominant on maize and groundnut (60% and 42% of Aspergillus section Flavi, respectively). Incidences of A. flavus L-morphotype were negatively correlated with aflatoxin in groundnut (log y = 2.4990935 − 0.09966x, R2 = 0.79, P = 0.001) but not in maize. Incidences of A. parasiticus partially explained groundnut aflatoxin concentrations in all agroecologies and maize aflatoxin in agroecology III (log y = 0.1956034 + 0.510379x, R2 = 0.57, P < 0.001) supporting A. parasiticus as the dominant etiologic agent of aflatoxin contamination in Zambia. Communities in both non-cultivated and cultivated soils were dominated by A. parasiticus (69% and 58%, respectively). Aspergillus parasiticus from cultivated and non-cultivated land produced statistically similar concentrations of aflatoxins. Aflatoxin-producers causing contamination of crops in Zambia may be native and, originate from non-cultivated areas, and not be introduced with non-native crops such as maize and groundnut. Non-cultivated land may be an important reservoir from which aflatoxin-producers are repeatedly introduced to cultivated areas. The potential of atoxigenic members of the A. flavus-L morphotype for management of aflatoxin in Zambia is also suggested. Characterization of the causal agents of aflatoxin contamination in agroecologies across Zambia gives support for modifying fungal community structure to reduce the aflatoxin-producing potential.


Introduction
Maize and groundnut are important crops for both commercial and smallholder farmers in Zambia. Maize is cultivated by > 80% of the farmers in all agroecologies for self-consumption, sale or both (Tembo and Sitko, 2013) and contributes up to 50% of daily calorie intake (FAO, 2014). Groundnut is the second most widely cultivated crop and is grown in all the agroecologies of Zambia (Tembo and Sitko, 2013). International demand for groundnut provides an important potential source of income. Groundnut and maize are susceptible to aflatoxin contamination and heavy dependence on these two crops in Zambia may result in significant aflatoxin associated hazards.
Consumption of aflatoxin-contaminated food may cause cirrhosis, liver cancer, stunting, reduced immunity, reduced weight-gain and/or rapid death (Gong et al., 2004;Lewis et al., 2005;Liu et al., 2012; products and reduced income (van Egmond et al., 2007;Wu, 2014). Europe and South Africa, with regulatory limits of 4 and 10 ppb total aflatoxin, respectively, have been important markets for agricultural commodities from Zambia. The country exported over 8000 metric tons of groundnut to Europe in the 1960s. However, this market collapsed due in part to enforcement of aflatoxin regulations in Europe (Sitko et al., 2011). Improved knowledge of the etiology of aflatoxin contamination in Zambia may reveal management options .
Aflatoxin contamination is caused by crop infection by one or more species in Aspergillus section Flavi. The fungi disperse from soil, organic matter, and alternative hosts to developing crops. Crop infection and subsequent aflatoxin production are high when conditions are hot and dry during crop development and warm and humid after crop maturation and/or harvest (Cotty and Jaime-Garcia, 2007). The species most notorious for crop contamination are Aspergillus flavus (produces only B aflatoxins), A. parasiticus (produces both B and G aflatoxins) and two unnamed taxa S B (only B aflatoxins) and S BG (both B and G aflatoxins; Cotty et al., 2008;Probst et al., 2010). Aflatoxin-producers are often sorted on the basis of sclerotial morphology (Cotty, 1989). L morphotype fungi produce few large sclerotia (average diameter > 400 μm) and S morphotype fungi produce numerous small sclerotia (average diameter < 400 μm; (Cotty, 1989). Fungi with S morphology frequently produce large quantities of aflatoxins. Molecular phylogenetic studies suggest S morphotype aflatoxin-producers are actually several species: a) A. flavus S strain; b) Lethal Aflatoxicosis Fungus (LAF) S B that severely contaminated maize and led to many deaths in Kenya (Probst et al., 2007); c) the un-named taxon S BG from West Africa (Cotty and Cardwell, 1999); and d) A. minisclerotigenes (Pildain et al., 2008). Aspergillus parasiticus is also frequently described as an etiologic agent of groundnut aflatoxin contamination (Horn and Dorner, 1998). Although all of these aflatoxin-producers may cause dangerous aflatoxin levels in crops when present in a conducive environment, genotypes vary in average aflatoxin-producing potential and the relative importance of specific etiologic agents may vary from one region to another . Frequencies of aflatoxin-producers on crops and relationships of fungal communities in non-cultivated soils to those resident in cultivated soils have not been characterized in Zambia. Non-cultivated areas, such as forests, may be reservoirs for aflatoxin-producers that may either move into cropping systems or cause contamination of non-cultivated fruits and grains (Boyd and Cotty, 2001). Potential causal agents of aflatoxin contamination in cultivated and non-cultivated plants in Zambia need characterization, and the relationship of fungal community structure to aflatoxins in groundnut and maize needs investigation .
In order to explore possibilities for limiting aflatoxin contamination in Zambia, compositions of Aspergillus section Flavi communities associated with aflatoxin contamination infecting maize and groundnut were explored and aflatoxin production by these communities was characterized and related to Aspergillus section Flavi resident in noncultivated areas. Aspergillus parasiticus was found to be an important etiologic agent for both maize and groundnut and communities of Aspergillus section Flavi resident in native, non-cultivated areas appear to have influenced compositions of fungi infecting crops.

Study area
Zambia lies between 8°and 18°South, and 22°and 34°East of the Greenwich meridian and has three agroecologies designated I, II, and III (Bunyolo et al., 1995). Agroecology III is the northern most with elevation 1100 to 1700 masl, annual rainfall > 1000 mm, and average annual temperature, 30-33°C (Bunyolo et al., 1995). Agroecology II covers most of the land in agricultural production and all of central Zambia. Elevation extends from 900 to 1300 masl with 800-1000 mm annual rain, and 30-32°C average annual temperature. Agroecology I extends across southern Zambia with elevations below 900 masl, < 800 mm average annual rainfall, and 30-36°C average annual temperature (Bunyolo et al., 1995).

Sampling
Maize (n = 250) and groundnut (n = 162) samples from a previous study  representing 27 districts and all three agroecologies of Zambia (Tables 1 and 2) were included in the current study. In addition, 220 soils were sampled from cultivated fields (n = 160) and from non-cultivated areas (n = 60), in 16 districts covering all three agroecologies (Fig. 1). Briefly, at least 4 locations or fields were sampled in each district. Three composite soil samples (100-175 g each) were obtained from each field by scooping soil subsamples at three random locations in each field to a depth of 2 cm (Cotty, 1997). In each agroecology, crop and soil samples were collected during the same trip with sampling occurring during January and 2.3. Isolation and identification of fungi from maize, groundnut and soils Maize and groundnut samples were ground in a knife mill (Grindomix GM200, Retsch GmbH, Haan, Germany) to pass a #12 sieve, and homogenized. Fungi were recovered from ground crop material and dry soil using dilution plate technique on modified rose Bengal agar (Cotty, 1994). Briefly, ground crop material and soil (0.1 to 10 g) were shaken in 50 ml sterile distilled water (20 min, 100 rpm) on a reciprocal shaker (KS-501, IKA Works Inc., Wilmington, NC, USA). Dilution plating was performed on modified rose Bengal agar in triplicate. Plates were incubated (3 days, 31°C, dark) and up to eight colonies of Aspergillus section Flavi were transferred to 5-2 agar (5% V8 Vegetable Juice (Campbell's Soup Company, Camden, N.J., USA); 2% agar, pH 5.2, Cotty, 1989). Fungi were stored in sterile water (2 ml) as plugs of sporulating culture after incubation for 7 days at 31°C (Cotty, 1988). Isolations were performed at least twice from each sample. Aspergillus species and strains were identified using both macroscopic and microscopic characteristics (Cotty, 1989(Cotty, , 1994Klich and Pitt, 1988;Probst et al., 2007).

Community composition of Aspergillus section Flavi from soils of cultivated and non-cultivated areas
Quantities and community composition of Aspergillus section Flavi from cultivated and non-cultivated areas were compared. The total quantity of section Flavi fungi from each crop and soil sample was calculated as Colony Forming Units (CFU) per gram. Community composition of section Flavi was described as percent of A. flavus Lmorphotype (Cotty, 1989), undelineated S-morphotype species (Probst et al., 2007), A. parasiticus, and A. tamarii recovered from each sample. Quantities of section Flavi members were calculated as the percent detected during isolation multiplied by total section Flavi CFU/g.

Aflatoxin producing potential of A. parasiticus from cultivated fields and non-cultivated areas
Aspergillus parasiticus isolates from maize (6), groundnut (6), and either cultivated (16) or non-cultivated soil (35) were assayed for aflatoxin-producing potential on sterile maize and groundnut. Fungi were inoculated onto undamaged, sterile maize and groundnut kernels (10 g/250 ml Erlenmeyer flask) previously autoclaved for 60 min, cooled to room temperature, and adjusted to 30% water content. The following steps were performed to adjust moisture content of the maize and groundnut kernels post-autoclaving and to inoculate: (1) The initial moisture after autoclaving was measured using an HB43 Halogen Moisture Analyzer (Mettler Toledo, Columbus, OH) and the amount of water needed to raise the moisture of kernels to 30% determined; (2) Conidia of each isolate from 7-day-old cultures (grown on 5% V8-juice; 2% agar, pH 5.2, Cotty, 1989) were harvested into sterile, deionized water (10 ml); (3) concentrations of conidia were estimated with turbidity (Orbeco-Hellige turbidimeter TB300IR; Orbeco Analytical Systems, Farmingdale, NY) and using a nephelometric turbidity unit (NTU)-versus-CFU standard curve, where y is equal to 49,937x (x is NTU, and y is conidia/ml) (Probst et al., 2010); (4) A spore suspension containing 1 × 10 6 conidia (usually about 500 μl) was mixed with water to bring the final volume to that determined in step 1 above, added to 10 g of kernels in a flask and swirled to coat the kernels. Inoculated grains were incubated (7 days, 100% RH, 31°C). After incubation, sample cultures were blended in 50 ml of 70% methanol. The slurry was allowed to separate for 30 min and the supernatant was spotted directly onto thin-layer chromatography (TLC) plates (Silica gel 60, EMD, Darmstadt, Germany) adjacent to aflatoxin standards (Aflatoxin Mix Kit-M, Supelco) containing known quantities of aflatoxins B 1 , B 2 , G 1 and G 2 . Plates were developed in ethyl ether-methanol-water, 96:3:1, air-dried and aflatoxins visualized under 365-nm UV light. Aflatoxins were quantified directly on TLC plates using a scanning  P.W. Kachapulula et al. International Journal of Food Microbiology 261 (2017) 49-56 densitometer (TLC Scanner 3, Camag Scientific Inc., Wilmington, N.C.). Five market samples found in a previous study to contain > 500 ppb total aflatoxins were subjected to aflatoxin analyses by TLC shortly after grinding to evaluate presence of B and G aflatoxins. Fifty grams of ground crop were extracted with 70% methanol (250 ml) and the methanol extract was directly spotted onto TLC plates and separated and quantified as above. The specific aflatoxins were identified by comparison with standards spotted on the same plate.

Data analysis
Statistical analyses were performed with JMP 11.1.1 (SAS Institute, Cary, NC). Means for fungal frequencies in cultivated and adjacent noncultivated soils in each district were compared using paired t-test were compared using the paired t-test and multiple comparisons were performed with Analysis of Variance (general linear models) followed by mean separation with Tukey's HSD test. Relationships between crop aflatoxin concentration and quantities of each member of Aspergillus section Flavi were investigated with regression analyses. Data were tested for normality and, if required, log (for aflatoxin concentrations and propagules per gram) or arcsine (for percentages) transformed to normalize distributions. Actual means are presented for clarity. All tests were performed at α = 0.05.

Fungi in maize and groundnuts
Aspergillus section Flavi was recovered from all maize and groundnut samples. A total of 4099 isolates were characterized from 412 samples (Tables 1 and 2). The frequencies of occurrence of Aspergillus section Flavi members in maize differed significantly (ANOVA, F 3,48 = 18.6842, P < 0.001) across agroecologies, with the A. flavus Lmorphotype dominating communities (60%), followed by A. parasiticus (24%) and S-morphotype fungi (16%, Table 1). In all agroecologies, the A. flavus L-morphotype was the most common member of Aspergillus section Flavi on maize making up 91%, 44%, and 45% of section Flavi in agroecologies III, II, and I, respectively. L morphotype frequencies in region III were higher than either region II or I (Tukey's HSD, P < 0.05). Aspergillus parasiticus and fungi with S morphology were more common in agroecologies II and I (agroecology averages = 22 to 34% of section Flavi) than in agroecology III (average 4 to 6%).

Aspergillus section Flavi from cultivated and non-cultivated soils
A total of 2128 Aspergillus section Flavi isolates were obtained from 220 soil samples (Table 5). One or more Aspergillus section Flavi species was recovered from all soils. Species differed in overall frequency in both un-cultivated (F 3,168 = 101.2705, P < 0.001) and cultivated (F 3,324 = 113.0661, P < 0.001) soils (Table 5). Aspergillus parasiticus was the most frequent Aspergillus section Flavi species in both soil types (58% in cultivated and 69% in non-cultivated soils), and in all agroecologies, with the exception of cultivated soils in agroecology I where A. flavus L-strain morphotype was most frequent (Table 5). Aspergillus parasiticus was most frequent in the coolest agroecology, agroecology III, and, overall, least frequent in the agroecology with the highest average temperature and the least rainfall, agroecology I ( Table 5). The frequency of Aspergillus section Flavi with S morphology did not differ among agroecologies in both cultivated (ANOVA, F 2,79 = 2.7790 P = 0.0682) and non-cultivated soils (ANOVA, F 2,40 = 0.3660, P = 0.6958, Table 5).

Discussion
Dangers aflatoxins pose to human health, livestock productivity and trade are widely recognized (Gong et al., 2004;Lewis et al., 2005;Liu et al., 2012;Probst et al., 2007;Reddy and Raghavender, 2007;Turner et al., 2003;van Egmond et al., 2007;Williams et al., 2004;Wu, 2014). Recent deaths from consumption of highly contaminated food in Kenya (Lewis et al., 2005;Probst et al., 2007) and Tanzania have increased recognition of the need to understand the etiologic agents of aflatoxin contamination in many parts of sub-Saharan Africa and the world. Although A. flavus, A. parasiticus, and the two unnamed taxa (S B and S BG ) are associated with aflatoxin contamination of crops in warm parts of the world, the most important etiologic agents vary from one region to another (Cotty and Cardwell, 1999;Horn and Dorner, 1998; , 2007). In addition, some aflatoxin-producers have also been recovered from non-cultivated areas (Boyd and Cotty, 2001) from which they may move into cropped areas, and provide region-specific influences on composition of communities of aflatoxin-producing fungi infecting and contaminating crops. The current study provides insights into compositions of communities of Aspergillus section Flavi in Zambia resident on maize, groundnut and soils. Compositions were found to influence observed aflatoxin concentrations and insights were developed into how communities of Aspergillus section Flavi from non-cultivated areas might shape those observed in cultivated areas and on crops.

Composition of Aspergillus section Flavi fungi in maize and groundnuts
Communities of Aspergillus section Flavi on maize consisted mostly of A. flavus (Table 1), while those on groundnut were dominated by A. parasiticus (Table 2). Aspergillus flavus L strain morphotype is reported to be the most prevalent member of Aspergillus section Flavi in soils and crops from many areas including maize, groundnut, cottonseed, rice, sorghum and almonds (Donner et al., 2015;Purcell et al., 1980;Schroeder and Boller, 1973). Aspergillus flavus is also a much more aggressive colonizer than A. parasiticus on groundnut (Horn, 2005). The current study stands in contrast to the aforementioned reports in that A. parasiticus occurred in frequencies higher than A. flavus in groundnut and at levels higher than has been observed on maize in East (Probst et al., 2007;Probst et al., 2010) or West Africa (Atehnkeng et al., 2008). The agroecologies in Zambia apparently differ from previously examined systems. This may reflect in part the percent of land in Zambia not in crop production. Indeed in agroecology I, the A. flavus L strain morphotype has displaced A. parasiticus in cultivated but not in noncultivated soils ( Table 5). As agriculture becomes more intensive, this trend may become more widespread.

Association between quantity of Aspergillus section Flavi and aflatoxin concentration
To assess the risk that a given fungal group poses to aflatoxin contamination of a crop, the aflatoxin-producing potential and frequency of occurrence of the fungus in contaminated crops need to be considered (Mehl et al., 2012;Probst et al., 2007). Regression analyses were conducted to examine agents associated with actual contamination events in market places previously reported in maize and  Mansa  24  338  2  0  0  0  96  100  2  0  33  82  Mpongwe  23  312  12  7  13  12  75  81  0  0  11  179  Chibombo  23  275  0  11  0  46  100  43  0  0  30  511  Chongwe  21  208  23  28  0  1  59  67  18  4  5  101  Kaoma  30  277  76  2  2  8  7  65  15  25  19  20  Senanga  24  273  28  8  10  19  62  73  0  0  10  182  Sesheke  21  294  33  68  3  12  64  19  0  1  65  150  Average  24  281  25  18  4  14  66  64  5  4  25 b  175 b a Percent data were arcsine transformed and CFU/g data were log transformed prior to statistical comparisons. b Non-cultivated and cultivated values differ by paired t-test (α = 0.05). c NC = non-cultivated soil, CV = cultivated soil.  . Percentages of the Aspergillus section Flavi community composed of the L strain A. flavus were inversely related to aflatoxin concentrations in groundnut in agroecology I (Table 4). This suggests that other components of these communities are more important causal agents. Increases in S-morphology fungi positively explained aflatoxin concentrations only in maize from agroecology II, indicating the potential of these fungi to exacerbate contamination. However, this was observed in only one area. Prevalence of A. parasiticus positively explained aflatoxin concentrations in groundnut in all agroecologies and in maize from agroecology III (Table 4). Atoxigenic A. parasiticus are rare (Horn et al., 1996). All A. parasiticus isolates in the current study were highly toxigenic (Table 7). High toxigenicity combined with high prevalence in infected crops suggests that A. parasiticus is the most important etiologic agent of aflatoxin contamination in Zambia. This was further supported by presence of both B & G aflatoxins in 5 highly contaminated maize and groundnut samples from markets.

Aspergillus section Flavi from cultivated and non-cultivated soils
Incidence of A. parasiticus in soils of Zambia were much higher than incidences previously observed in soils from East or West Africa (Donner et al., 2009). Natural dominance of A. parasiticus in the soil could contribute to crop from Zambia having higher levels of this species than has been observed elsewhere (Donner et al., 2009;Horn and Dorner, 1998). Horn and Dorner (1998) observed that A. parasiticus incidences were higher in fields cultivated to peanuts than in those where maize was grown. Most of the small scale farms sampled in the current study had mixed cropping, where maize or other cereals were grown in combination with groundnut (Tembo and Sitko, 2013). Wide cultivation of groundnut in agroecologies of Zambia (Sitko et al., 2011) may contribute to high incidences of A. parasiticus in the soil and eventually on the crop. Within the agroecologies of Zambia, A. parasiticus incidences were highest in agroecology III and lowest in I (Table 5). Soils in the three regions differ in pH and temperature, with region III being the coolest and most acidic (Bunyolo et al., 1995). Low temperature promotes crop colonization by A. parasiticus (Donner et al., 2015;Horn and Dorner, 1998) and could contribute to the higher frequencies of this fungus observed in agroecology III of Zambia (Table 5). However, perhaps the most important factor influencing incidences of A. parasiticus in crops in Zambia is the natural distribution of this species as reflected in composition of the fungal community in noncultivated soils.
Fungi are capable of dispersal under natural conditions and mixtures of different aflatoxigenic and atoxigenic fungi are found in both non-cultivated and cultivated fields (Boyd and Cotty, 2001). In Zambia, many small-scale farmers grow their crops adjacent to national forests. The current study reports similar patterns of community composition in non-cultivated and cultivated areas (Table 6) suggesting that Aspergillus species endemic to non-cultivated areas in Zambia influence compositions of fungal communities in cultivated areas to a greater extent than Aspergillus introduced with crops. Application of atoxigenic A. flavus based biocontrol products may prevent this movement and associated crop contamination.

Aflatoxin production by A. parasiticus from crops and soils
Assessing the risk a given fungal group poses to aflatoxin contamination requires knowledge of the aflatoxin-producing potential and frequency of occurrence of the fungus Probst et al., 2007). High frequencies of A. parasiticus were observed in crops and soils from Zambia (Tables 1-3, 5-6) and were associated with contaminated crops (Table 4). Aspergillus parasiticus from Zambia was highly toxigenic (Table 7) irrespective of the fungus source (i.e. maize, groundnut, or soil), substrate used for assaying aflatoxigenicity (maize or groundnut) or temperature at which aflatoxigenicity assessments were conducted. Aspergillus parasiticus from non-cultivated areas was just as aflatoxigenic as those from crops or cultivated soil, suggesting that fungi endemic to non-cultivated areas in Zambia are potential reservoirs from which aflatoxigenic fungi disperse to crops. Similar to what has been reported before where aflatoxigenicity of A. parasiticus was the same on maize and groundnut at 31°C , no differences were observed on the two crops at 25°C, 30°C or 35°C (Table 7). However, higher concentrations of aflatoxins were produced on groundnut than maize at 20°C. The mechanisms behind higher aflatoxigenicity on groundnut than maize at lower temperatures are currently unknown.
Aflatoxin-producing fungi are common in maize and groundnut, and in cultivated and non-cultivated soil in all agroecologies of Zambia. Aspergillus parasiticus and fungi with S strain morphology are important etiologic agents of crop contamination. Aspergillus flavus L strain morphotype fungi were associated with lower aflatoxins in crops, possibly because this group of aflatoxin-producers are of lower average aflatoxin-production potential in Zambia . L strain isolates with low aflatoxin-producing potential are also known to interfere with crop contamination by highly toxigenic fungi (Atehnkeng et al., 2014). Methods to increase incidences of atoxigenic L strain isolates, such as biocontrol (Atehnkeng et al., 2016), may lower aflatoxin contamination in Zambia. Cultivated and non-cultivated areas had comparable community structures of Aspergillus section Flavi. Moreover, fungi from non-cultivated areas were just as toxigenic as those from cultivated areas. Fungi endemic to non-cultivated areas in Zambia may influence the compositions of fungal communities in cultivated areas and are potential reservoirs from which toxigenic fungi disperse to crops. Treatments with atoxigenic genotypes of the A. flavus L strain morphotype may reduce effects of fungal communities from non-cultivated areas on fungi infecting crops.