Introduction

Contamination of maize grain with mycotoxins, such as aflatoxins and fumonisins is a global food and feed safety risk. Aflatoxins produced by Aspergillus flavus (Link: Fr.) and fumonisins produced by Fusarium verticillioides (Sacc.) Nirenberg, are carcinogenic, teratogenic, mutagenic and immunosuppressive to both humans and livestock (Mesterházy et al. 2012; Wild and Gong 2010). Therefore health authorities worldwide regulate aflatoxin and fumonisin levels in human food and animal feed (FAO 2003). The United Nations food standards body Codex Alimentarius Commission has set maximum levels of total aflatoxin (Aflatoxin B1+B2+G1+G2) in food as 15 parts per billion and 5 ppb for aflatoxin B1. Aflatoxin B1 is the most toxic and carcinogenic (Wu 2014). Maximum levels for total fumonisin (Fumonisin B1+B2+B3) are set at 4 ppm in raw maize grain and 2 ppm in maize flour and maize meal (Codex 2014). The European Union has stricter regulation with maximum levels for total aflatoxin and fumonisin set at 10 ppb and 2 ppm, respectively (EC 2007). In the United States of America the Food and Drug Administration (U.S.FDA) has set the limit for total aflatoxin at 20, 5 ppb for aflatoxin B1 and 4 ppm for total fumonisin (U.S.FDA 2001).

The standards applied in many African countries are the same as recommended by Codex Alimentarius (Codex 2014). However, regulations are not usually effectively enforced in Africa. In South Africa the limit for total aflatoxin is 10, 5 ppb for aflatoxin B1 and 4 ppm for total fumonisin (Marasas et al. 2012). Of significant concern is the lack of resources to monitor and enforce these regulations in sub-Saharan Africa (SSA) (Shephard 2003; Wagacha and Muthomi 2008). The disturbing outcome of these strict regulations is that contaminated grain rejected on the international market would be retained and subsequently be available to communities, which are already exposed to higher levels of mycotoxins. Therefore millions of people, especially in southern Africa where maize has the highest per capita consumption in the world (Warburton and Williams 2014), could be consuming contaminated grain without being aware of the health risks associated with mycotoxins. There is little information on the occurrence of mycotoxins in maize hybrids currently used in most parts of Southern Africa including South Africa.

When conditions are favourable for fungal growth, no cultural practice or chemical control can prevent the contamination of maize grain by aflatoxins and fumonisins. The development of resistant germplasm will be the most economically efficient control measure for these diseases of maize for all markets. Decades of research have resulted in the identification of maize inbred lines that are resistant to either aflatoxin or fumonisin accumulation (Afolabi et al. 2007; Mayfield et al. 2012; Menkir et al. 2006) but not to both. All resistant breeding lines identified to date contain tropical germplasm in their backgrounds. Thus, they tend to be tall, late and prone to lodging, in addition to lower yielding than locally adapted commercial hybrid controls in temperate regions. In the current study diverse inbred lines were crossed to diverse donors. Genetic diversity serves as a way for a breeding population to adapt to changing environments (Clements et al. 2004). With more variation, chances are high that some progenies in a population will possess variations of alleles that are suited for the environment (Robertson et al. 2006). Therefore unique and genetically broad-based germplasm used in this study would increase the chances of finding alternative sources of mycotoxin contamination resistance. The aim of this study is to transfer the aflatoxin and fumonisin resistance genes from tropical maize inbred lines into adapted subtropical and temperate maize germplasm grown in southern Africa.

The objectives of this study were to (i) determine the incidence of ear rot fungi associated with mycotoxin contamination of maize grain on germplasm used across southern Africa, (ii) stack the resistance genes in a single product, through introgression of aflatoxin and fumonisin resistance genes from tropical inbred lines into adapted inbred lines used in the subtropical and temperate conditions of southern Africa, (iii) identify novel families that combine resistance to both aflatoxin and fumonisin accumulation (iv) calculate the heritability and genetic gain as a measure of the breeding progress attained in this study. Overall the study aimed to demonstrate the potential of enhancing mycotoxin resistance in maize hybrids through stacking of aflatoxin and fumonisin resistant genes in the end product.

Materials and methods

Nursery and field experiments

Germplasm

In order to determine the natural incidences of ear rot in maize hybrids, the hybrids with different levels of physiological maturities (early, medium and late) from different breeding programmes in four countries in southern Africa (Zimbabwe, Zambia, Malawi and South Africa) were observed in the regional trials, organised by the International Maize and Wheat Improvement Center (CIMMYT), Zimbabwe Station. A set of 50 early, 63 medium and 54 late-maturing hybrids were evaluated in 2012/13. In 2013/14, 60 early, 60 medium- and 40 late-maturing hybrids were evaluated. The same set of hybrids were used in both seasons. The hybrids 11XH1579 and 11XH1774 were used as local controls. These are medium maturing hybrids which are adapted to South Africa warm temperate and sub-tropical environments. They have high yield potential but have not been tested for resistance to ear rot.All the experimental and control hybrids that were used in the study have white grain, which is preferred by consumers in southern Africa.

For the second experiment, the maize inbred lines, AFD01 and AFD02 were used as aflatoxin contamination resistance sources. These maize inbred lines combine temperate and tropical genomes in their background, which makes them suitable sources for breeding aflatoxin resistance in South African germplasm lines. In addition, they contain sufficient genes of resistance to major foliar diseases, which compromise grain yield in the southern Africa region. A different set of the maize inbred lines, FD01 and FD02 were used as fumonisin contamination resistance sources. These lines are resistant to FER infection, which is associated with contamination of grain with fumonisins and other mycotoxins. They have white grain and high yield potential in hybrids. They are adapted to the medium altitude environments (800–1600 masl.), in eastern and southern Africa. Forty-one adapted inbred lines from the maize programme at UKZN were used as recipients of genes for resistance to contamination by aflatoxin and fumonisin. These 41 lines had diverse genetic backgrounds, and are adapted to South Africa warm temperate and sub-tropical environments. They have high yield potential but have also not been tested for resistance to ear rot. The donor lines were also from different genetic backgrounds. As result different hybrids could be created between these two sets of germplasm lines.

The genes for resistance to aflatoxins and fumonisins were stacked into South African adapted maize inbred lines by crossing them with the resistant inbred lines in a factorial mating design. A total of 41 recipient lines were crossed to two fumonisin resistant inbred lines (donors): FD01 and FD01, to generate 82 F1 single crosses, during the 2012 winter season, at Makhathini Research Station (27°39`S; 32°10`E; altitude 72 m). These 82 single crosses were crossed with two aflatoxin resistant inbred lines: AFD01 and AFD02, in the greenhouse during the 2012/13 summer season. Not all crosses were successful, this resulted in only 44 three-way cross hybrids stacked with aflatoxin and fumonisin resistant genes were successful. The three-way cross is equivalent to the S1 generation seed. These three-way crosses were self-pollinated to produce the S1:2 generation seed at Makhathini Research Station, during the 2013 winter season. The resultant 146 S1:2 families were then advanced to S2:3 at the Cedara Research Station (29°54`S, 30°26`E, altitude 1066 m), during the 2013/14 summer season.

Experimental design and management

The seeds of the hybrids were planted at Cedara Research Station (29°54`S, 30°26`E, altitude 1066 m), during the 2012/13 and 2013/2014 maize growing season. Cedara Research Station was chosen as a suitable site to conduct the study. It is situated in the Natal midlands’ mist-belt which provides favourable conditions for the development of ear rot diseases. The three trial sets (comprising early, medium and late hybrids) were arranged in an alpha lattice design and replicated three times. Plots consisted of two rows of 5 m length, 0.75 m between the rows and 0.3 m within the rows. The trials were planted by hand with two seeds per station and thinned down to one plant, 3 weeks after seedling emergence. Each row had a maximum potential of 17 plants, resulting in 34 plants per plot. The experiment was conducted under dryland conditions and was inoculated naturally. Basal fertilizer (2,3,4 compound) was applied and provided 75 kg N, 50 kg P, 25 kg K per hectare before planting. The top dressing of 120 kg per hectare in the form of Limestone Ammonium Nitrate, LAN (28 % N) was applied four weeks after crop emergence. Hand weeding and other cultural practices were conducted as and when they were required. The cultural practices which are recommended for maize in South Africa were followed. This included the application of insecticide granules to control stalk-borer and the application of herbicides, such as atrazine and gramoxone, to control weeds. The trials were manually harvested during May 2013 and May 2014, respectively.

The 82 single crosses were planted in a randomised complete block design (RCBD) with four replications in the greenhouse during the 2012/13 summer season. The trial was set up in 30 cm plastic pots filled with composted pine bark growth media. Two replications were crossed with the aflatoxin resistant inbred lines to generate three-way crosses. The other two replications were artificially inoculated with F. verticillioides. The 82 single crosses were also planted in the field at the Makhathini Research Station, during the 2012/13 summer season, under rain-fed conditions. The trial was planted in single row plots, with 17 plants per row. Plots were spaced 0.9 m between rows and 0.3 m between stations, arranged in a RCBD with two replications. Both replications were artificially inoculated with F. verticillioides. The field trial on three-way crosses was conducted under irrigation at the Makhathini Research Station, during 2013 winter season.

The 44 three-way crosses from the greenhouse were planted at the Makhathini Research Station during the 2013 winter season (Jun–Oct). The trial was planted in single row plots, with 17 plants per row. Plots were spaced 0.75 m between rows and 0.3 m between stations, arranged in a RCBD with two replications. Both replications were artificially inoculated with F. verticillioides and A. flavus. In each replicate both pathogens were inoculated onto ears by partitioning a plot with 17 plants, such that eight plants were inoculated with F. verticillioides and another eight plants with A. flavus, leaving one plant in-between. The 146 S2:3 families were planted at the Cedara Research Station, during the 2013/14 summer season (Dec–May), under rain-fed conditions. The inbred lines FD01 and FD02 were used as positive controls for fumonisin contamination resistance. AFD01 and AFD02 were used as positive controls for aflatoxin contamination resistance. Three S1:2 family bulk of the AFD01 × AFD02 cross was also included as a positive control, which was at the same level of inbreeding with the test entries. Sixteen drought-tolerant S1:2 families (MTX-144-153, MTX-155-7), which were not introgressed with aflatoxin and fumonisin resistant genes, were used as negative controls or susceptible controls for both aflatoxin and fumonisin contamination. The experiment was planted in single row plots by hand, with two seeds per station and thinned down to one plant at 3 weeks after seedling emergence. Each row plot had a maximum potential of 17 plants. Plots were 5 m long, with spacing of 0.75 m between rows and 0.3 m between stations, arranged in a RCBD. In each plot, five plants were self-pollinated to advance to the S3:4 generation and the remaining 12 plants were partitioned such that six plants were artificially inoculated with F. verticillioides and another six plants with A. flavus. The nursery was manually harvested on 28 May 2014.

Field data collection

The monthly rainfall, temperature and relative humidity data for 2012/13 and 2013/14 seasons were recorded (Figs. 1, 2, 3, 4).The data on experimental hybrids was collected on a whole plot basis for each hybrid. At harvest, diseased ears per plot were counted and categorised based on the visual symptoms of the ear rots. The symptoms were classified as Aspergillus, Fusarium, Diplodia or Gibberella ear rot. The characteristic symptoms of Aspergillus ear rot (AER) are yellow-green mycelia growth on the kennels. Fusarium ear rot (FER) is characterised by cottony, whitish-pink growth, typically occurring individually or in groups scattered randomly on the ear. Diplodia ear rot is dense, whitish fungal growth matted between the kernels and between the ear and the husk, beginning at the base of ear and progressing towards the tip. Pink to reddish mould, usually starting at the tip of the ear, characterise Gibberella ear rot. Grain yield was measured as plot weight and transformed to t ha−1. Data on insect damage was taken in the 2013/14 season because of the observed high incidence. Insect damage ratings were recorded from 1 = no damage or tunnelling, to 9 = heavy damage or tunnelling (Badu-Apraku et al. 2012). Grain texture was recorded using a rating scale of from 1 = hard, completely rounded flint kernel, 2 = semi-flint, 3 = semi-dent, 4 = soft, 5 = very soft, distinct dent (CIMMYT 1985).

Fig. 1
figure 1

Monthly rainfall (in bars) and mean temperatures (line graph) during the experimental period (2012/13 and 2013/14) at Cedara Research Station, South Africa. (Source Agricultural Research Council, 2014)

Full size image
Fig. 2
figure 2

Monthly mean relative humidity during the experimental period (2012/13 and 2013/14) at Cedara Research Station, South Africa. (Source Agricultural Research Council, 2014)

Full size image
Fig. 3
figure 3

Monthly rainfall (in bars) and mean temperatures (line graph) for Cedara Research Station, during the experimental period, November 2013 to May 2014 (Source Agriculture Research Council, 2014)

Full size image
Fig. 4
figure 4

Monthly mean relative humidity for Cedara Research Station during the experimental period, November 2013 to May 2014 (Source Agriculture Research Council, 2014)

Full size image

For the second experiment ears were also manually harvested at maturity. At each harvest date, ears in each plot were hand-picked, dehusked and evaluated for severity of ear rot symptoms. Disease severity of FER and AER was assessed by determining the percentage of each ear covered by symptoms using a 7–class rating scale, in which 1 = no infection, 2 = 1–3 %, 3 = 4–10 %, 4 = 11–25 %, 5 = 26–50 %, 6 = 51–7 %, and 7 = 76–100 %, as described by Afolabi et al. (2007). Grain texture was also recorded. Ears were oven-dried, after disease assessment, to approximately 14 % grain moisture content. Grain was hand-shelled and bulked by plots.

Laboratory experiments

Isolation and morphological identification of cultures

All kernels with suspected fungal infection were surface sterilised with 2 % sodium hypochlorite (NaClO) for one minute, before being washed three times with distilled water. The kernels with suspected Fusarium spp. infection were cultured on a selective media, Synthetic Nutrient Deficient Agar (SNA– glucose 0.2 g, sucrose 0.2 g, KH2PO4 1 g, KNO3 1 g, MgSO4 0.25 g, KCl 0.5 g, agar 14 g/L). Plates were incubated at 25 °C under UV light, for 14 days. Fusarium verticillioides was identified morphologically by the formation of micoconidia chains using a light microscope (magnification ×40). Sickle-shaped macroconidia observed under a light microscope (magnification ×40) were used as characteristic tools for confirmation of F. graminearum. As well as the deep purple colour formed on the agar plate was used for identification (Leslie and Summerell 2006).

The A. flavus Selective Media (AFPA—dichloran 0.002 g, ferric ammonium citrate 0.5 g, peptone 10 g, K2HPO4, MgSO4.7H20 0.5 g, chloramphenicol 0.2 g, agar 15 g/L) was used to culture the suspected A. flavus infected kernels. Plates were incubated at 28 °C for 7 days. Yellow-green moulds on media and the observation of a spore-bearing structure (the aspergillum) were used as morphological characteristic tools to confirm A. flavus in the laboratory (Klich 2002).

Suspected Diplodia ear rot kernels were plated on potato dextrose agar (PDA). The plates were incubated at 28 °C, for seven days. Black spherical pycnidia on kernels surface and straight conidia curved or irregular, septate, smooth-walled and pale-brown, with rounded or truncated ends observed under a light microscope (magnification ×40) were used as the morphological characteristics to identify Stenocarpella maydis (Flett et al. 2001).

Molecular characterisation of cultures

Molecular techniques, which include deoxyribonucleic acid (DNA) extraction, amplification and sequencing, were employed. For Fusarium sp. part of the transcript elongation factor (TEF) was amplified using the primer set elongation factor 1(EF1) [5′-CGAATCTTTGAACGCACATTG-3′] and EF2 (5′-CGTGTTTCAAGACGGG -3′) (O’Donnell and Cigelnik 1998). For A. flavus and S. maydis internal transcribed spacer (ITS) region primers, ITS1(5′- TCCGTAGGTGAACCTGCGG- 3′) and ITS4 (5′- TCCTCCGCTTATTGATATGC-3′) were used for the amplification of the target genomic regions of the fungal isolates (White et al. 1990). Consensus sequences were compiled in BioEdit and BLAST comparisons done on Fusarium spp., Aspergillus spp. and Stenocarpella spp. parts of MycoBank database (http://www.mycobank.org). A 99 % match confirmed the species as F. verticillioides (Accession no. KF562131.1), F graminearum (JX118859.1), A. flavus (KF309063.1) and S. maydis (KC311732.1).

Isolation, inoculum preparation and inoculation of pathogens

Isolates of F. verticillioides and A. flavus for inoculation in the greenhouse and in the field were obtained from naturally infected maize ears at Cedara Research Station near Pietermaritzburg. The fungal pathogens were isolated and identified, as described in Sect. 2.1.5 and 2.1.6.

Fungal isolates were routinely maintained in 15 % glycerol at −80 °C. Prior to inoculation F. verticillioides and A. flavus were cultured on SNA and AFPA, respectively. Conidia were washed from the surface of the agar media with sterile distilled water and filtered through two layers of sterile cheesecloth. The concentration was adjusted to 1 × 106 conidia/ml, using a haemocytometer; 0.1 % Tween 20 (Fisher Biotech, Fairlawn, NJ) per litre was added as surfactant. Inoculum suspension was used within 24 h of preparation.

At flowering, the maize ears were inoculated using a 10 ml syringe 7–10 days after mid-silking. Conidia suspension (5 ml) was injected down the silk channel of primary ears of all plants at the blister (R2) growth stage (Chungu et al. 1997). Ears were covered after inoculation with plastic shoot bags for 2 days, to maintain high humidity and to protect the inoculum from being drained off by rain or dried by excessive heat.

Aflatoxin and fumonisin analysis

Bulked grain from each plot was ground using a Cyclotech sample mill to pass through a 1-mm mesh and stored in a cold room at 4 °C. Two replicates of 5 g samples were used for aflatoxin and fumonisin extraction. Ground samples were extracted with 25 ml of 70 % methanol using a multi-tube shaker for 20 min. The samples were then centrifuged for 10 min at 4000 rpm and 1 ml of the obtained supernatant was diluted with 1 ml of distilled water. Contamination of grain samples by aflatoxin B1 and total fumonisin was quantified using a direct competitive enzyme-linked immunosorbent assay (ELISA) technique following the methods described in the manufacturer’s instructions (BIOO Scientific Corporation 2014).

Statistical analysis

All statistical analysis of all data of hybrids and S2:3 families were performed in GenStat (version 14, VSN International). Differences between hybrids were determined with Fisher’s unprotected least significant differences (LSD) test. Data on ear rot incidence of hybrids was transformed to arcsine of the percentage of ear rots incidence. Although the experiment was designed as an alpha-lattice, the trial was analysed as RCBD, without violating any of the assumptions for the statistical model.

For introgressed hybrids, linear correlation coefficients were determined for the relationship among ear rots severity, aflatoxin and fumonisin concentration. For S2:3 families, linear correlation coefficients were determined for the relationship between aflatoxin and fumonisin concentration with AER, FER and grain texture using Pearson correlation coefficients based on untransformed means.

The REML tool was used to estimate the variance components. The replications were considered as fixed effects and hybrids effects were treated as random. Heritability was calculated on an entry mean basis using the following formula (Nyquist 1991):

$$H^{2} = \frac{{\sigma_{G}^{2} }}{{\sigma_{G}^{2} + \frac{{\sigma_{e}^{2} }}{r}}}$$

where H2 is the broad sense heritability, \(\sigma_{G}^{2}\) is the genotypic variance, \(\sigma_{e}^{2}\) is the overall error variance and r is the number of replications for the experiment.

The realised genetic gain was calculated as the difference between the population mean and the mean of the selected hybrids or families (Nyquist 1991):

$$\Delta {\mathbf{G}} \, = \, {\varvec{\upmu}}_{{{\mathbf{2}} }}- {\varvec{\upmu}}_{{\mathbf{1}}}$$

where µ2 = mean of selected hybrids or families, µ1 = population or trial mean.

The predicted genetic gain was estimated (Nyquist 1991):

$$\Delta G = R = i\sigma_{p} H^{2}$$

where ∆G is the genetic gain, R is the response to selection, i is the selection intensity, H2 is the broad sense heritability and σp is the phenotypic variance. Ten percent of the hybrids or families were selected for advancement in the breeding programme. The selection intensity (i) of 1.76 was used to predict the genetic gain.

Results

Natural incidence of ear rots in regional maize hybrid

Temperature, rainfall and relative humidity data are presented in Figs. 1 and 2. In the 2012/13 summer season, S. maydis was the most prevalent (42.61 %) ear rot causing fungus, followed by F. verticillioides (28.19 %), A. flavus (15.06 %) and F. graminearum (14.32 %). In the 2013/14 summer season, the most prevalent fungus was F. verticillioides (48.7 %), followed by S. maydis (22.13 %), F. graminearum (16.62 %) and A. flavus (12.52 %). Over the two seasons, F. verticillioides recorded the highest mean (38.46 %), S. maydis was second (32.37 %), F. graminearum was third (15.45 %) and A. flavus had the least incidence (13.79 %).

Ear rots incidence in maize hybrids was more prevalent during the 2013/14 summer season than the 2012/13 summer season. There was significant variation (P < 0.001) among hybrids for ear rots resistance over the two seasons. The lowest ear rots incidence was observed in early-maturing hybrids for both seasons (12.99 and 28.71 %), with a combined mean of 20.85 %. Medium-maturing hybrids had an incidence of 13.27 % and 32.13 for the 2012/13 and 2013/14 summer seasons, respectively, with a combined mean of 22.7 %. The highest incidence were observed in late-maturing hybrids for both seasons (14.17 and 37.3 %), with a combined mean of 25.73 %.

Hybrids evaluation for ear rot severity under artificial conditions

Highly significant (P < 0.001) differences were observed in FER severity and fumonisin concentration among single cross hybrids in both the green house and field trial at Makhathini Research Station. Forty-nine percent of the hybrids showed low disease severity (≤3) in both the greenhouse and in the field. The mean visual rating of FER on maize ears in the green house and the field trials was 2.4 and 3.4, respectively (Table 1). At least five single cross hybrids consistently showed low fumonisin concentration levels <4 ppm in both the greenhouse and field, which is the Codex Alimentarius Commission legal limit of total fumonisin in grain (Codex 2014). The hybrids are: FUMH03, FUMH10, FUMH30, FUMH47 and FUMH73 (Table 1). Only FUMH47 exhibited lower accumulation levels than the resistant control in both the greenhouse and the field (Table 1). Four single cross hybrids (FUMH1, FUMH17, FUMH71 and FUMH74) showed moderate resistance (>4–10 ppm) in both environments. Fourteen percent of the hybrids were susceptible in the two environments, accumulating between 10 and 20 ppm. Highly susceptible hybrids, which consistently accumulated >30 ppm in both environments accounted for 17 % of the single cross hybrids. Hybrid FUMH53 had the highest average in both environments (Table 1).

Table 1 Evaluation of 72 single cross maize hybrids for resistance to Fusarium ear rot and fumonisin contamination in the greenhouse and field
Full size table

Correlation between greenhouse and field results

The reaction of the hybrids following artificial inoculation with F. verticillioides in the greenhouse correlated well (r = 0.7, P < 0.01) with the reaction of the same hybrids when artificially inoculated in the field. A significant correlation was observed between FER and fumonisin concentration in the greenhouse (r = 0.49. P < 0.01) and field trial (r = 0.58, P < 0.01). The trend was not consistent because some hybrids showed relatively high disease severity score and low fumonisin contamination. In sharp contrast, another set of hybrids exhibited a trend of low disease severity rating and high fumonisin contamination.

Estimates of genetic gain and heritability

When the top 10 % resistant single crosses were selected, high genetic gains were realised for fumonisin contamination resistance in both the greenhouse and field trials. The realised genetic gains for FER were moderate and low in the greenhouse and in the field, respectively. Low genetic gains were predicted in both environments for fumonisin contamination resistance and FER. Heritability estimates of fumonisin contamination resistance and FER were very high in both trials (Tables 2, 3).

Table 2 Breeding gains and heritability in single cross maize hybrids in the greenhouse
Full size table
Table 3 Breeding gains and heritability in single cross maize hybrids in the field at Makhathini Research Station
Full size table

Evaluation of three-way cross hybrids for ear rots and mycotoxin contamination

Differences among three-way cross hybrids were highly significant (P < 0.001) in AER severity, aflatoxin concentration and fumonisin concentration. Differences in FER were significantly high (P < 0.01). AER and FER severity on the three–way cross hybrids ranged from 1.0 to 6.5, with means of 3.4 and 3.8, respectively (TableS 4, 5). Eighteen percent of the hybrids showed low disease severity (≤3) for both AER and FER. Mycotoxin analysis showed that at least three hybrids, such as FUMH/AFTX11, FUMH/AFTX12 and FUMH/AFTX18, have low concentrations of aflatoxin (≤5 ppb) and fumonisin (≤4 ppm) contamination, respectively. However, there were some hybrids that had low aflatoxin concentration but high fumonisin concentration, for example FUMH/AFTX26, which had an aflatoxin content of 4.1 ppb and fumonisin content of 37.0 ppm. No hybrid showed high aflatoxin content and low fumonisin content (Tables 4, 5).

Table 4 Aspergillus ear rot and aflatoxin contamination resistance of selected three-way cross maize hybrids in the field at Makhathini Research Station
Full size table
Table 5 Fusarium ear rot and fumonisin contamination of selected three-way cross hybrids in the field at Makhathini Research Station
Full size table

Four hybrids: FUMH/AFTX39, FUMH/AFTX22, FUMH/AFTX23 and FUMH/AFTX34, showed moderate resistance (>5–10 ppb) to aflatoxin contamination, 32 % were susceptible (>10–20 ppb). Only FUMH/AFTX31 was highly susceptible to aflatoxin contamination, accumulating ≥30 ppb. Four hybrids: FUMH/AFTX03, FUMH/AFTX2, FUMH/AFTX38 and FUMH/AFTX16, showed moderate resistance (>4–10 ppm) to fumonisin contamination. Five hybrids: FUMH/AFTX13,FUMH/AFTX19, FUMH/AFTX24, FUMH/AFTX17 and FUMH/AFTX15) were susceptible (>10–20 ppm). However, 66 % of the hybrids showed high susceptibility to fumonisin contamination, accumulating ≥30 ppm.

Correlation between ear rot and mycotoxin contamination

A significant and positive correlation was observed between FER and fumonisin contamination of maize hybrids (r = 0.65, P < 0.01). Not all hybrids followed a similar trend, for example, FUMH/AFTX01 had a low mean FER score of 2.0, but accumulated a high fumonisin content of 29.42 ppm. The linear correlation between AER and aflatoxin contamination (r = 0.7, P < 0.01) was higher compared to the relationship between FER and fumonisin contamination. Few hybrids did not follow this trend, for example, FUM/AFTX 14 and FUM/AFTX32 had low ear rot severity scores, but accumulated high aflatoxin content. A weaker, but significant correlation (r = 0.39, P < 0.05) was observed between fumonisin contamination and aflatoxin contamination among the hybrids. Some hybrids, however, displayed a strong correlation in the accumulation of aflatoxins and fumonisins. An example is FUMH/AFTX12, which accumulated low concentration for both mycotoxins.

Estimates of genetic gain and heritability

When the top 10 % three-way crosses were selected, high genetic gains were realised for aflatoxin and fumonisin accumulation. Moderately low genetic gains were predicted for both aflatoxin and fumonisin contamination resistance. Heritability estimates for aflatoxin and fumonisin contamination resistance were both high, whereas those for AER and FER were moderate. Moderately high genetic gains were realised for AER and FER. Low genetic gains were predicted for both AER and FER (Table 6).

Table 6 Breeding gains and heritability in three-way cross maize hybrids
Full size table

Evaluation of S2:3 families for ear rots and mycotoxin contamination

The weather data for the 2013/14 summer season is shown in Figs. 3 and 4. Highly significant (P < 0.001) differences were observed in FER, AER, aflatoxin and fumonisin content and graintexture. Sixty-nine percent of the families had a low ear rot severity score (≤3) for AER. Fourteen percent of the families performed better than the resistant controls, which had a mean score of 1.5. Eighty-four percent of the families were more susceptible than the resistant control(Table 7).

Table 7 Means of aflatoxin B1 content and agronomic traits of the top 15 and bottom 5 S2:3 families relative to the controls
Full size table

Fifty-six percent of the families had a low ear rot severity score (≤ 3) for FER. Sixteen percent of the families performed better than the resistant controls, which had a mean score of 2.0. Seventy-seven percent of the families were more susceptible than the resistant control (Table 8). Aflatoxin and fumonisin contamination was observed in all families and resistant controls, following artificial field inoculation with A. flavus and F. verticillioides. There was no family that had complete resistance to both aflatoxin and fumonisin contamination. However, 23 families with concentration levels below the Codex Alimentarius legal limits of 5 ppb for aflatoxin B1 were identified, while nine families with contamination levels below the set limit of 4 ppm for total fumonisin were identified.

Table 8 Means of total fumonisin content and agronomic traits of the top 15 and bottom 5 S2:3 families relative to the controls
Full size table

Five families MTX-90, MTX-79, MTX-12, MTX-67 and MTX-36 (Table 7), exhibited lower levels of aflatoxin B1 contamination compared to the resistant controls. The five families, MTX-45, MTX-90, MTX- 67, MTX- 79, MTX-36, accumulated lower levels of fumonisin contamination than the resistant controls (Table 8). The families, MTX-90, MTX-79, MTX-67 and MTX-36, showed potential resistance to both aflatoxin and fumonisin contamination. Unexpectedly, the inbred line, FD02, which was used as a resistant control for fumonisins, also accumulated low levels of aflatoxins. A high level of contamination was observed for both aflatoxin and fumonisin in the families that were not introgressed with resistant genes. The families MTX-144 toMTX-149 and MTX-158 to MTX-164 (negative controls, Tables 7, 8) were not introgressed. Some families that were introgressed, for example MTX-53 and MTX-54, accumulated high levels of aflatoxin and fumonisin more than the non-introgressed families (Tables 7, 8). Potential resistance to one trait and susceptibility to another trait was also observed among families. For example, MTX-4 accumulated low levels of fumonisin (4.8 ppm), but high levels of aflatoxin, 13.1 ppb. In contrast, families MTX-25 and MTX-113 accumulated low levels of aflatoxin, 3.4 and 3.5 ppb, respectively, but high levels of fumonisin contamination, 15.9 and 21.8 ppm, respectively. Several families had a lower FER score, but a high fumonisin concentration. For example, MTX–126 had a low ear rot severity score of 1.3, but accumulated 48.7 ppm of fumonisin.

Correlations between ear rot and mycotoxin contamination

Aflatoxin B1 contamination showed a significant and positive correlation (r = 0.20, P < 0.001) with AER, as shown, for example, by MTX-79, which accumulated 1.2 ppb and an ear rot rating of 1.5 (Table 7). A weak, but significant, positive correlation (r = 0.15, P < 0.001) was observed between aflatoxin B1 contamination and grain texture.

A significant and positive correlation(r = 0.33, P < 0.001) was observed between fumonisin contamination and FER severity. This result is shown, for example, by MTX–90, which had a mean FER rating of 1.0 and fumonisin contamination of 1.9 ppm (Table 8). Fumonisin contamination also showed a significant and positive correlation (r = 0.51, P < 0.001) with grain texture and aflatoxin B1 contamination (r = 0.49, P < 0.001).

Heritability estimates and genetic gain

When the top 10 % resistant families were selected, very high genetic gain was realised for aflatoxin and fumonisin contamination resistance. Low genetic gains were realised for AER and FER. Low genetic gains were predicted for aflatoxin and fumonisin contamination, AER and FER. Heritability estimates were also very high for aflatoxin and fumonisin contamination resistance, AER and FER (Table 9).

Table 9 Breeding gains and heritability in S2:3 families
Full size table

Discussion

Natural incidence of ear rots in southern African maize hybrids

Stenocarpella maydis accounted for the highest incidences of ear rot diseases in 2012/13, compared to the 2013/14 season. This can be explained by the variability of weather patterns between the two seasons (Figs. 1, 2, 3, 4). Ears are most susceptible to Diplodia infection during the first 3 weeks of silking and when the green silks start to turn brown (White 1999). Dry weather prior to silking, followed by wet weather during silking, seems to increase the incidence and severity of Diplodia ear rot (Steckel 2003). The co-ordinated timing of spore release and plant silking could be responsible for this relationship (Agrios 1997). In the present study, a period of dry spell with high temperatures was experienced in February to early March, during the 2012/13 summer season. High rainfall and high relative humidity was recorded during the rest of March, providing a conducive environment for S. maydis development. The period coincided with the silking dates of the hybrids. In the 2013/14 summer season, such a pattern was not observed.

In 2013/14, the most isolated fungus was F. verticillioides. Dry conditions at silk emergence are known to favour the spread of F. verticillioides microconidia (Munkvold 2003). Little rainfall, coupled with higher temperatures and relative humidity in March 2014 compared to March 2013, was recorded. Conditions in 2013/14 were, therefore more favourable for F. verticillioides than S. maydis, compared to the 2012/13 season. Overall, F. verticillioides accounted for the highest incidence of ear rot diseases in hybrids. Results in the present study are in agreement with previous studies that reported the high occurrence of F. verticillioides and fumonisins in maize and maize based food in Zambia(Mukangaet al. 2010) and in Limpopo and Eastern Cape provinces of South Africa (Ncube et al. 2011; van der Westhuizen et al. 2010).

Results from the present study showed that late-maturing hybrids are more susceptible to ear rots. As the water available for fungal growth plays a key role, late-maturing maize genotypes, in which grain moisture content decreases slowly are most likely to be susceptible (Eller and Holland 2008). The occurrence of late-season decline in rainfall in April and May 2013 might explain the high incidence of A. flavus in the 2012/13 season, compared to the 2013/14 season. Although low ear rot incidence was observed in early maturing hybrids, the hybrids displayed a high negative correlation between yield and ear rot incidence compared to the medium and late hybrids. Gasura et al. (2010) reported that early-maturing maize genotypes yields 20–30 % less than late-maturing genotypes. Thus despite their ability to escape ear rots infection and provide food early, the challenge is to increase yield while maintaining earliness.

The fact that a high number of hybrids were susceptible to fungal ear rots has implications for both breeders and consumers of maize grain. First, it is a matter of great concern that millions of people might be consuming contaminated maize grain and maize-based foods daily, without being aware of the danger. In terms of breeding, the way forward is to continue screening more experimental hybrids and breeding populations to find safe and high-yielding varieties.

Development of mycotoxin resistant hybrids

Single-cross maize hybrids, with consistently low levels of fumonisins in grain, were identified in the field and greenhouse experiment, giving hope that resistant commercial hybrids can be developed and that shuttling the breeding between the greenhouse and field testing would be effective. Use of greenhouses would be important to bridge the summer seasons, because the logistics of conducting off-station winter nurseries are very expensive. Three-way cross hybrids, with low levels of fumonisin and aflatoxin contamination were also identified in the field experiment, adding further credence that resistance can be stacked in the end-product. The three-way cross hybrids are the predominant form of hybrids which are grown in SSA. The cost of producing single-cross hybrids is very high. Therefore the development of three-way crosses would be ideal for the resource-poor farmers in SSA, who are the target for the deployment of mycotoxin resistant maize hybrids. Families with potential resistance to both aflatoxin and fumonisin contamination in grain were also identified.

A survey of the literature suggests that this is the first study to report the identification of maize hybrids and S2:3 families with potential resistance to both fumonisin and aflatoxin contamination in a single product, which are adapted to subtropical and temperate environments of southern Africa. Previous studies have reported on the development of maize inbred lines for either fumonisin or aflatoxin, but not both, in a single genotype (Afolabi et al. 2007; Small et al. 2012). The hybrids and families with potential mycotoxin resistance, which were identified in this study, will be subjected to further evaluation across seasons and over an extended range of conditions, representative of the localities where they will eventually be grown, to further confirm their resistant status and value for cultivation.

Seemingly, mycotoxin resistance is quantitative, because none of the hybrids and families evaluated in this study were completely resistant to fumonisins or to both fumonisin and aflatoxin contamination. The involvement of polygenic inheritance can be inferred from the continuous distributions of fumonisin and aflatoxin data in the genotypes. The observation of a lack of complete resistance is in agreement with previous studies (Abbas et al. 2002; Henry et al. 2009). It is suggested that both parents of single cross hybrids or all three parents of the three-way cross should be improved for resistance. As a result, resistance should be improved in all heterotic groups which will be used to make hybrids among them.

The current study indicates that genotype x environment (G×E) interaction would present fewer complications in selection of hybrids, because there was a significant and positive correlation between results from the greenhouse and the field. Previous studies have reported high G×E interactions for grain contamination by aflatoxins and fumonisins contributing to low heritability estimates for ear rots and mycotoxin contamination (Abbas et al. 2002). Genotype x environment interactions especially drought stress, have therefore been attributed as the main reason for the lack of consistency of the performance of maize genotypes for resistance to mycotoxin contamination (Abbas et al. 2002; Cotty and Jaime-Garcia 2007). In sharp contrast, results from the present study displayed high heritability estimates for ear rots and mycotoxin contamination. Thus results from the present study indicate that, although ear rots and mycotoxin resistance are complex traits, selection of best-performing genotypes could speed up breeding for resistance to mycotoxin contamination in hybrids or populations.

Rapid and cheaper screening methods are required to scale up breeding for mycotoxin resistance in maize hybrids. In this regard, the findings from the current study reveal a promising trend. Total fumonisins and disease severity on single cross hybrids were positively correlated in both the field and greenhouse experiments. Similar results were observed in three-way crosses. The strong correlations between FER and AER with fumonisin and aflatoxin contamination observed in this study suggest that selection against FER and AER should also result in reduced susceptibility to fumonisin and aflatoxin contamination. This is a favourable situation because ear rots are easy to score in the field, relatively across environments, and much less expensive to phenotype than mycotoxin contamination.

The findings from the current study are consistent with previous investigations, which have reported a similar correlation between fumonisin and ear rot severity (Robertson et al. 2006). Although this relationship describes a general trend, a number of hybrids in the present study (e.g. FUMH52 and FUMH81 in the greenhouse experiment) had low ear rot severity, with very high levels of fumonisin content in grain. A similar observation that identified high levels of fumonisins in apparently symptomless kernels has previously been reported (Mukanga et al. 2010; Small et al. 2012). This might be difficult to explain, but Bacon et al. (2008) reported on the endophytic colonisation of maize plants by F. verticillioides and presumed that this can contribute to mycotoxin contamination of grain. As global warming seems to be durable, this problem may become more serious in the tropics and subtropical areas mainly in SSA. This underscores the need for fumonisin analyses when evaluating maize resistance. Unfortunately laboratory assays are still very expensive and this can impact negatively on research, especially in developing countries.

Results from the present study suggest the existence of some common resistance mechanisms to these two fungi. Genes that control plant stress reactions may contribute to the correlation between resistances to these two fungi. For example, stress due to weather conditions influences fumonisin and aflatoxin contamination (Lobell et al. 2008; Murillo-Williams and Munkvold 2008).

The relationship between aflatoxin B1 and total fumonisin concentration was significant and positively correlated. Consistent with this observation, the maize inbred line FD02, which was used as a resistant control for fumonisin accumulation in this study also showed resistance to aflatoxin contamination. However, some families with high aflatoxin B1 concentration accumulated low levels of fumonisin concentration and vice versa. Selecting for aflatoxin resistance does not necessarily result in fumonisin resistant genotypes, and therefore the grain must be evaluated separately, for both aflatoxin and fumonisin contamination.

Earlier studies evaluated maize germplasm for drought tolerance and concluded that this character appeared tightly linked with mycotoxin reduction (Henry 2013; Williams and Windham 2001). In the current study, drought tolerant lines which were not introgressed with aflatoxin and fumonisin genes were included. None of these inbred lines were resistant. They accumulated high levels of both aflatoxin and fumonisin concentration. It is, however, imperative to evaluate sources of aflatoxin and fumonisin contamination resistance identified in this study under drought conditions.

Previous studies have shown that harder (flint) kernels minimise insect damage and impede fungal penetration compared to the soft (dent) kernels (Wit et al. 2011). In the present study, sources of aflatoxin and fumonisin resistance were all flint. However, there are several susceptible inbred lines among the flints and dent pools indicating that resistance should be evaluated in both the flint and dent maize germplasm. This might explain why resistant families identified in this study were also flint. The non-introgressed lines were predominantly dent and showed high susceptibility. However, a high correlation was observed between FER and fumonisin accumulation, compared to AER and aflatoxin accumulation, suggesting that selecting for the flint trait might prove valuable for resistance to fumonisin accumulation.

Heritability estimates of resistance to aflatoxin and fumonisin contamination and low severity of AER and FER were high in the germplasm. These results indicate that selection of potentially resistant families identified in this study provide a greater chance of attaining genotypes with high levels of resistance to both mycotoxin accumulation and ear rots severity. Robertson et al. (2006) found high heritability estimates for FER and fumonisin contamination of maize inbred lines from two maize populations. A high genetic gain realised in aflatoxin and fumonisin contamination suggest significant breeding progress can be made through introgression of resistant genes into adapted lines. Although results predict low genetic gains, further research efforts to improve aflatoxin and fumonisin accumulation through selection may result in genotypes with higher levels of mycotoxin resistance.

Conclusion

The present study succeeded in determining the potential incidence levels of ear rots in southern African maize hybrids and demonstrates feasibility of introgressing mycotoxin resistance genes in maize hybrids. In this regard the concept of stacking aflatoxin and fumonisin resistance genes in maize hybrids has been proven. The potential resistant families which were identified in this study will be advanced and fixed in the breeding program. Ultimately the materials would be valuable breeding stocks for use as parents of new hybrids and synthetic populations. However, further resources should be mobilised to upscale the effort to breed new varieties which can reach the farmer. The F. verticillioides was the most prevalent ear rot causing fungus, followed by S. maydis, F. graminearum and A. flavus suggesting that breeding programmes should target mycotoxins which are caused by these fungi in South Africa.