Influence of socio‐economic and agronomic factors on aflatoxin and fumonisin contamination of maize in western Kenya

Abstract Consumption of maize contaminated with mycotoxins has been associated with detrimental health effects. A farm survey covering 116 push‐pull and 139 non‐push‐pull cropping systems was conducted to determine the socio‐economic and agronomic factors that influence farmers’ knowledge on incidence and contamination of maize by ear rots and associated mycotoxins in western Kenya. All the respondents were smallholder farmers between the ages of 23 and 80 years, with 50% of them being female. Maize samples were collected from the standing crop in the field of each interviewed farmer and analyzed for aflatoxin and fumonisin. Only a small proportion of farmers had knowledge of aflatoxin and ear rots in maize. Overall, less than 20% of maize samples were contaminated with both aflatoxin and fumonisin, and more maize samples were contaminated with fumonisin as compared to aflatoxin. Proportions of maize samples containing higher than the acceptable Kenyan regulatory threshold (10 µg/kg) for aflatoxin and European Commission regulatory threshold (1,000) µg/kg for fumonisin were lower in maize samples from push‐pull cropping system. Age of farmer and county of residence were significantly and positively associated with knowledge of aflatoxin, while cropping system, county of residence, and level of education were positively associated with knowledge of maize ear rots. There was strong correlation between knowledge of maize ear rots and knowledge of aflatoxin. Levels of both aflatoxin and fumonisin were significantly and positively associated with the use of diammonium phosphate (DAP) fertilizer at planting. Aflatoxin levels were also positively associated with stemborer damage. Agronomic practices were not significantly different between push‐pull and non‐push‐pull farmers. However, use of DAP fertilizer was the most important agronomic factor since it was associated with both aflatoxin and fumonisin contamination of maize. These results imply that creating awareness is key to mitigation of ear rots and mycotoxin contamination of maize. The results also suggest that the levels of aflatoxin and fumonisin in maize in western Kenya were influenced both by pre‐harvest agronomic practices and by the cropping system adopted, push‐pull or not.


| INTRODUC TI ON
Maize (Zea mays L.) is the staple food for majority of households in Kenya, with nearly all agricultural households growing the crop.
Compared to other food crops, maize production occupies more land, with over 70% of its production being done by smallholder mixed farmers (Keya & Rubaihayo, 2013;Kibet, 2011). Maize is used as food for humans, as feed for livestock, and as industrial raw material for many manufactured maize-based products. However, the quality of maize is compromised by numerous constraints, which include fungal diseases, such as ear rots and kernel infections. Ear rot is a major disease of maize worldwide that is visually characterized by moldiness of kernels at different points of the cob depending on the causative fungi during the growth of the crop, at harvest and during storage (Bigirwa, Kaaya, Sserurwa, Adipala, & O. S., 2007). Moldy kernels have low integrity compared to healthy kernels. The predominant maize ear rot causative fungi are determined by the climatic conditions of a region. Common types of maize ear rots are Fusarium, Penicillium, Aspergillus, Diplodia, and Giberella (www.krugerseed.com). Some ear rot fungi such as Fusarium and Aspergillus spp. pose health risks by contaminating maize with associated mycotoxins under appropriate conditions of moisture and temperature (Dragich & Nelson, 2014). Maize ear rots also reduce grain yield, since the rotten grains are removed during shelling.
Mycotoxins are toxic compounds naturally produced by certain fungi as secondary products of their biosynthetic pathway (Hendrich, 2017;WHO, 2018). When produced in food like maize and maize-based food products, mycotoxins become a threat to food and feed safety because they are poisonous to humans and animals.
Consumption of mycotoxin-contaminated foods has been linked with adverse health effects such as exacerbation of the symptoms of diseases like HIV/AIDS and malaria, suppression of the immune system, cancers of vital organs, and even death in humans (Lewis et al., 2005;WHO, 2018;Williams, Aggarwal, Jolly, Phillips, & Wang, 2005). Humans can also be exposed to mycotoxins through inhalation or contact with contaminated foods (Paterson & Lima, 2010).
Mycotoxin contamination of maize is usually influenced by factors such as climatic conditions during the cropping season, production practices, and physical factors, mainly temperature and moisture. However, drivers for contamination are different for different mycotoxins. Warm temperature and erratic weather patterns encourage infection of maize grains by toxigenic fungi such as A. flavus (Cotty & Jaime-Garcia, 2007). A previous report indicated that proper crop management practices effectively controls A. flavus and F. verticillioides and associated mycotoxins (Bruns, 2003). Tillage practices, handling of maize stovers after harvest, host susceptibility, use of fertilizers, rate of maturity of cultivar, kernel breakage, shape of kernel, presence of rotten kernels at harvest, and continuous cultivation also influence proliferation of some ear rot fungi (Mutiga et al., 2017(Mutiga et al., , 2014. For instance, farmer practice of handling maize stovers after harvest could influence population of ear rot fungi and associated mycotoxins, since most mycotoxigenic fungi survive as saprophytes on crop residues (Chulze et al., 2015;Mutiga et al., 2017). Drying duration after harvest for the maize also significantly influences production of aflatoxin because of A. flavus infection at specific moisture and humidity level (Mbuge et al., 2016). Edwards (2004) reported that exposure to mycotoxins can also be due to insect damage to maize cobs. Feeding insects can either act as vector of fungal pathogens by transferring the fungi or expose the cob to infection from the atmosphere. Insect damage is of most important concern for Fusarium spp. mycotoxins such as fumonisin (Sobek & Munkvold, 1999).
A recent study reported reduced incidence and severity of maize ear rots and associated mycotoxins with the push-pull technology (Owuor, Midega, Obonyo, & Khan, 2018). However, cropping systems and management practices as drivers of mycotoxin contaminations, including underlying mechanisms, have remained unstudied in the region. Push-pull is a farming system that involves intercropping cereals with a "push" crop and a border "pull" crop around the plot (Khan, Pickett, Berg, Wadhams, & Woodcock, 2000). In maize farming, the system has three components: the maize, Desmodium spp. (commonly known as desmodium) as the intercrop, and Brachiaria or Napier grass (Pennisetum purpureum) as the border crop. Desmodium roots produce allelopathic chemicals which induce suicidal germination of striga seeds, thus suppressing the development of the weed (Khan, Hassanali, Pickett, Wadhams, & Muyekho, 2003). Desmodium also improves soil nutrition through biological nitrogen fixation and phosphorus availability (Khan et al., 2000). Desmodium foliage emits semiochemicals that repel gravid stemborer moths (Midega, Jonsson, Khan, & Ekbom, 2014;Midega, Pittchar, Pickett, Hailu, & Khan, 2018), which are simultaneously attracted to the "pull" plants where they lay their eggs. The border crop, however, has characteristics that cause high mortality of the larvae (Khan et al., 2000). The objective of this study was to establish socio-economic and agronomic factors aggravating aflatoxin and fumonisin contamination of maize in western Kenya, as well as compare the practices of push-pull and non-push-pull farmers.
Farmers' knowledge of the two mycotoxins and maize ear rots and the association of their knowledge with farmers' socio-economic status were also determined.

K E Y W O R D S
agronomic practices, management, mycotoxins, push-pull, Zea mays 2 | MATERIAL S AND ME THODS

| Description of the Study sites
The study was conducted in selected sub-counties in five counties of western Kenya: Kisumu, Vihiga, Siaya, Kakamega, and Migori. These five counties represent counties where smallholder maize farmers have widely adopted the push-pull cropping system for stemborer and striga management. In the five counties, maize is the main food crop and source of income. The agro-ecology parameters of the counties are shown in Table 1. A total of 255 farmers were randomly selected from a list provided by the International Centre of Insect Physiology and Ecology (icipe) field staff located in the counties; 116 of these were farmers who had adopted the push-pull technology, while 139 had not (had plots planted to maize monocrop or intercropped with a food legume).

| Field survey and collection of maize samples
A household survey was conducted using a pre-tested semi-structured questionnaire between January and February 2017. Face-toface interviews were conducted by trained enumerators to obtain information on socio-economic factors and agronomic practices of maize farmers that influence understanding of and be associated with aflatoxin and fumonisin contamination of maize, respectively.
Socio-economic data collected included age of the farmer, sex of the farmer, level of education, acreage under maize production, membership of welfare group(s), experience in maize farming, knowledge of aflatoxin and fumonisin, source of funding for farm inputs, and average annual family income. Agronomic data collected included cropping system (push-pull or otherwise), varieties of maize grown, sources of seeds, any crop rotation program, intercropping practice, use of fertilizers, tillage method, use of maize stovers after harvest, knowledge of maize ear rots, and causes and management of maize ear rots. The questionnaire was set in English, and the enumerators interpreted the questions to farmers in local languages of the study area. Ten to 20 maize cobs, depending on the size of cob, were collected from standing crop from farms of each respondent. The cobs were sun-dried, manually shelled and finely ground (Bunn-O-Matic Corporation Coffee Mill, G3-000) before storage at 4°C until analyses.

| Detection and quantification of aflatoxin levels in maize
Twenty grams sub-samples were weighed in duplicate for each sample and extracted with 100 ml of 70% methanol. The samples were mixed by shaking in sealed containers for 2 min. The particulate matter was allowed to settle; the extracts filtered through Whatman no.1 filter paper and the filtrate collected for testing.

| Detection and quantification of fumonisin levels in maize
Twenty grams sub-samples were weighed in duplicate for each sample and extracted with 40 ml of 90% methanol. The samples were mixed by shaking in sealed containers for 1 min, after which the particulate matter was allowed to settle. The extracts were filtered through Whatman no. 1 filter paper and the filtrate collected for testing. The sample extracts were further diluted with distilled water in the ratio of 1:20. Fumonisin levels were quantified by direct competitive enzyme-linked immunosorbent assay (ELISA) (Helica Biosystems Inc) following manufacturer's instructions. The lower and the upper limits of detection for fumonisin test kits were 100 and 6,000 µg/kg, respectively. A calibration curve for the fumonisin standards was plotted and used to compare the optical densities of the samples with those of the standards. Samples with fumonisin levels above the upper limit of detection were diluted, and the toxin levels quantified again.

| Statistical data analyses
The survey data were analyzed using SPSS version 22 (IBM Corp, 2013). Percentages were used for basic description of socio-economic characteristics and agronomic practices of the farmers.

| Socio-economic characteristics of maize farmers in five counties in western Kenya
All the respondents were smallholder farmers with low average annual income and dependent on farm produce for own consumption and sale for financing farm operations as well as supporting family financial needs (Table 2). Respondents varied in age, between 23 and 80 years. 50% of the respondents were female, between the age of 31 and 60 years. The proportion of female respondents was, however, higher (70%) for non-push-pull as compared to push-pull (57%) respondents. The number of push-pull respondents below the age of 30 years was significantly lower (p < 0.05) than non-push-push respondents. Over 80% of the respondents belonged to one or more welfare groups. As shown in

| Agronomic practices in maize production
Ninety-nine percent of the respondents spent the period between cropping seasons clearing and digging the fields in preparation for the subsequent season. In addition, some farmers applied compost and farmyard manure (<10%), while others grew short duration crops such as vegetables and sweet potatoes (≈15%). Tillage was mainly by use of hand hoe with significantly higher proportion of push-pull farmers using the tool (  (Table 3).
Approximately 40% of the farmers applied diammonium phosphate (DAP) fertilizer and a significant proportion of both push-pull and non-push-pull farmers applied farmyard and compost manure, although a higher proportion of push-pull farmers used compost manure as compared to non-push-pull farmers.
The two most common practices of harvesting maize by the respondents were (a) dehusking maize cobs in the field, drying, and then manual shelling and (b) cutting stovers with cobs, stoking for drying, dehusking, and then manual shelling (Table 3). Upon harvesting, most respondents either harvested the maize stovers for hay or left the stovers in the farm and plowed in during cultivation.
The proportion of push-pull respondents that plowed in maize stovers was significantly lower (p < 0.05) compared to non-push-pull respondents. Other ways of handling maize stovers after harvesting included direct grazing of cattle, burning in the field and use as firewood. The harvested maize was mainly stored as shelled maize grains in polythene sacks on raised floors in the house by over 80% Over 50% of respondents mentioned that they encountered rotten cobs, however, in low incidence and severity. About 95% of respondents hand-sorted rotten and unwanted cobs before shelling the maize. The rotten and unwanted grains were mainly fed to livestock-cattle, poultry, and pigs (Table 3). Some respondents also used rotten maize grains for local brewing, while others mixed with clean ones for cooking or milling. The respondents estimated that grain yields were significantly higher (p < 0.05) in push-pull than in non-push-pull cropping system, across the counties.

| Prevalence of aflatoxin and fumonisin in maize
Overall, <20% of maize samples were contaminated with both aflatoxin and fumonisin, with proportion of maize samples without aflatoxin contamination being significantly higher (p < 0.05) from push-pull farms (Table 4). Across the counties, there were less than 10% of push-pull maize samples contaminated with aflatoxin, except in Siaya where the proportion was 12%. The proportion of non-pushpull maize samples contaminated with aflatoxin across the counties ranged from 6% in samples from Vihiga to 25% in samples from Siaya. All the push-pull samples from the five counties had aflatoxin levels below the Kenya Bureau of Standards (KEBS) recommended level (10 µg/kg), while a proportion of non-push-pull samples from Kakamega, Siaya, and Vihiga had aflatoxin levels above the limit.
Conversely, there were relatively higher proportions of maize samples from both push-pull (17.2%) and non-push-pull (21.6%) cropping systems that were contaminated with fumonisin (Table 5).
Across the counties, the proportions of push-pull samples contaminated with fumonisin varied between 5.6% and 23.8% in samples from Kakamega and Kisumu, respectively. The proportion of nonpush-pull maize samples contaminated with fumonisin ranged from 11.3% and 37.5% in samples from Kakamega and Vihiga, respectively.
A higher proportion of non-push-pull samples had fumonisin levels above the European Commission (EC) recommended level (1,000 µg/ kg) compared to push-pull samples across the counties.

| Relationship between socio-economic characteristics of farmers and knowledge on aflatoxin and maize ear rots
Farmers' knowledge on aflatoxin increased with increase in age, irrespective of whether the farmer practiced push-pull or non-push-pull cropping system (Table 6). Specifically, the respondents between the ages of 46 and 60 years constituted 48% of the proportion of farmers with knowledge on aflatoxin and were significantly more (p < 0.05) knowledgeable about aflatoxin as opposed to younger age groups.
The farmers from Kakamega, Migori, and Siaya were the least knowledgeable on aflatoxin compared to farmers from Vihiga. Push-pull farmers were 0.34 times significantly less (p < 0.05) knowledgeable of maize ear rots compared to non-push-pull farmers. The average proportion of push-pull respondents knowledgeable on ear rots was 4% lower than proportion of non-push-pull respondents. The farmers' knowledge of maize ear rot was approximately 5 times higher in farmers who had primary education compared to farmers who had tertiary education. In fact, respondents who completed primary school education constituted 65% of the total number of respondents knowledgeable on ear rots. Like knowledge on aflatoxin, farmers from Kakamega, Migori, and Siaya were the least knowledgeable on ear rots. Farmers' knowledge on aflatoxin and maize ear rots had a significant positive correlation (r = 0.338, n = 253, p = 0.01).

| Association between agronomic practices of farmers and levels of aflatoxin and fumonisin
The levels of aflatoxin in maize samples were 3.9 times higher than 10 µg/kg in farms where DAP fertilizer was applied at planting (p < 0.05; TA B L E 5 Fumonisin levels (µg/kg) in maize samples under push-pull and non-push-pull cropping systems in five counties in western Kenya Table 7). Likewise, aflatoxin levels were higher in maize sampled from farms infested by stemborer (2 times) and in maize intercropped with food crops such as sorghum and cassava (0.3 times; p < 0.05). Use of hand hoe tillage, compost manure, intercropping maize with beans, harvesting maize stovers as hay for cattle, plowing in maize stovers in subsequent season, and directly grazing livestock on maize stovers in the field after harvest increased the odds of maize contaminated with high levels of aflatoxin. Fumonisin levels were 0.3 times higher than 1,000 µg/kg in maize planted with DAP fertilizer (p < 0.05). High levels of fumonisin in maize samples were also to some extent positively influenced by most of the agricultural practices shown in Table 7.

| D ISCUSS I ON
Prevalence of mycotoxins, especially aflatoxin and fumonisin in western Kenya, has been reported in a number of studies (Mutiga, Hoffmann, Harvey, Milgroom, & Nelson, 2015;Mutiga et al., 2014;Owuor et al., 2018). Most of these studies targeted stored maize and, therefore, provided no premise for developing mitigation strategies, especially preharvest. A recent study, however, reported low incidence of ear rots and levels of mycotoxins in maize grown under the push-pull system (Owuor et al., 2018). The current study adds The results of the survey suggest that women were the main managers of farming activities in western Kenya, which concurs with the findings of previous studies (Midega, Murage, Pittchar, & Khan, 2016;Sofa & Doss, 2011). Adults between the ages of 31 and 60 years constituted the largest age group involved in small holder   TA B L E 7 Association between levels of aflatoxin and fumonisin and agronomic practices of farmers in five counties in western Kenya seeds, even asymptomatically, could act as a source of secondary inocula (Parsons & Munkvold, 2012). Agricultural practices that remove, burry, or destroy infected maize stovers are likely to minimize the amount of inocula of saprophytic fungi (Edwards, 2004). Ndemera, Nyanga, Saeger, Boevre, and Landschoot (2018) reported that choice of seeds for planting significantly affects the levels of fumonisin B1 in maize.
Majority of respondents applied different types of soil amendments, both organic and mineral. This is a good practice to enhance productivity. However, the amounts applied, time of application, and status of soil fertility are not properly taken into consideration.
Previous studies have reported that insufficient or excessive application of fertilizers in soil may end up enhancing contamination of maize with different mycotoxins such as fumonisin, aflatoxin, and ochratoxin (Arino et al., 2009;Blandino, Reyneri, & Vanara, 2008;Hassegawa et al., 2008). Even in this study, application of DAP fer- Maize farming technologies that reduce incidence of maize ear rots would, therefore, reduce the levels of mycotoxins in maize as rotten maize has higher levels of mycotoxin than clean maize (Alakonya, Monda, & Ajanga, 2009).
Our findings suggest that fumonisin was a more economically important mycotoxin of maize in western Kenya during the study period as only a small proportion of maize samples was contaminated with aflatoxin, while a higher proportion was contaminated with fumonisin. Even so, a higher proportion of non-push-pull samples had aflatoxin and fumonisin levels above 10 µg/kg and 1,000 µg/ kg, the limits set Kenya Bureau of standards (KEBS) and European Commission (EC), respectively. The proportion of maize samples with aflatoxin and fumonisin levels above the regulatory threshold was, however, lower in this study compared to some previous studies (Mutiga et al., 2015;Sirma, 2016). This was possible because in this study, samples were collected from standing crop while in the previous studies the samples were collected from storage. Maize grains contaminated with mycotoxin-producing fungi if not properly stored would have higher levels of associated mycotoxins than they would at harvest. The amount of mycotoxins would also have been influenced by the climatic conditions of temperature and rainfall patterns during the cropping season during which the samples were collected (Tirado, Clarke, Jaykus, McQuatters-Gollop, & Frank, 2010;Viegas, Meneses, & Viegas, 2016).
Older farmers were significantly more likely to know about aflatoxin than younger farmers. This could possibly be because the older farmers may have learnt about aflatoxin in the national news through local media during the 2004-2005 and previous aflatoxin outbreaks in lower eastern Kenya (Lewis et al., 2005). This could also be supported by the fact that knowledge of aflatoxin was not associated with the level of education. Unlike many of agronomic practices by the respondents, knowledge of existence of maize ear rots was significantly influenced by the farmers` cropping system, push-pull or non-push-pull, and push-pull farmers were less knowledgeable about maize ear rots at harvest compared to non-push-pull farmers.
Additionally, farmers with education levels lower than the secondary level were less knowledgeable on maize ear rots as compared to those with tertiary education. The association between highest levels of education and knowledge of maize ear rot implies that literacy is key to management of maize ear rots.
Although socio-economic and agronomic practices influenced the levels of aflatoxin and fumonisin, the maize samples from pushpull fields had lower levels of the two mycotoxins. Push-pull cropping system integrates insect pest management, striga control, and improvement of soil nutrition (Khan et al., 2000;. As a result, the companion cropping system results in increased maize grain yield. Insect larvae infesting maize cobs are known to act as vectors for mycotoxigenic fungi such as F. verticillioides (Sobek & Munkvold, 1999), and therefore, it is possible that through insect control functionality, the push-pull cropping system could contribute to reduction in levels of mycotoxins in maize.

| CON CLUS IONS
From the study, we conclude that socio-economic and agricultural factors influence mycotoxin contamination of maize in western Kenya. Indeed, some of the agronomic practices by the farmers were significantly associated with aflatoxin and fumonisin levels in the maize samples. Most of the other practices which were not significantly associated with the levels of the two toxins had reasonable (95%) confidence interval, suggesting that the practices were also important. Farmers, therefore, can play an important role in management of mycotoxins through manipulations of such practices. There is, therefore, a need to invest in mycotoxin awareness training to sensitize the farmers on good agricultural practices and management of mycotoxin.
Despite the predisposing farming practices to mycotoxin contamination of maize by both push-pull and non-push-pull farmers, there were still lower levels of aflatoxin and fumonisin in maize from push-pull cropping system. This suggests presence of several mechanisms that suppress toxigenic fungi and associated mycotoxins under push-pull cropping system. Possibly, the less tendency of push-pull farmers to plow-in maize stovers after harvest could have contributed to the lower levels of the two mycotoxins. Further studies are recommended to evaluate push-pull technology for mechanisms of management of mycotoxigenic fungi and associated mycotoxins in maize. This will inform on necessity for implementation of the cropping system as part of integrated management strategy for mycotoxin control in the region.
Additionally, mycotoxin surveillance will be necessary in order to avoid future acute mycotoxicosis and help in development of robust integrated mycotoxin management tools.

ACK N OWLED G EM ENTS
We thank German Academic Exchange Service (

CO N FLI C T O F I NTE R E S T
The authors declare that they do not have any conflict of interest.

E TH I C A L A PPROVA L
This study does not involve any human or animal testing.