Phenotypic characterisation and evaluation of resistance to Fusarium ear rot, fumonisin contamination and agronomic traits in a collection of maize landraces

Germplasm

A set of 33 traditional maize landraces from the Emilia-Romagna (Italy) region was evaluated under field condition (Table 1). The landraces presented different origins: the ‘Va’ series was retrieved from the germplasm bank of CREA-Cereal and Industrial Crops (CREA-CI) in Bergamo, and was collected during the 1954 census; the ‘EMR’ series was provided by the Plant Germplasm Bank of the Department of Earth and Environmental Sciences at Università degli Studi di Pavia, Pavia, Italy, and is of recent sampling; five additional landraces (Quarantina Genovese-Q_G, Mais da pipe-M_P, Ustnenina gialla-U_G, Ustneina arancione-U_A, and Ustneina rossa-U_R) derived from the germplasm collection maintained at Department of Sustainable Crop Production at Università Cattolica del Sacro Cuore, Piacenza, Italy. The landrace ‘Mais Nostrano di Storo’ (N_S), grown in Valli Giudicarie, Trento, Italy, and a commercial hybrid of the FAO maturity class 300 were also included in this study and used as comparisons representing a well-known and valorised landrace and a modern flint hybrid, respectively (Table 1).

Field management

Field trials were carried out in 2 years, 2019 and 2020, at Centro Ricerche Zootecniche (45.005066° N, 9.704206° E, San Bonico, Piacenza, Italy). In 2019, fields were sown on 17 April, and in 2020, on 15 April. Each experimental field represented a completely randomised design with four replications originated by the blockdesign package (Edmondson 2019) of the R software (R Core Team 2017). Each plot consisted of 4 rows of 5 m long, spaced 0.8 m apart, and separated by a 1 m aisle, and 25 seeds were planted for each row. The central rows of each plot were used in order to assess agronomic traits relying on the indications of the UPOV-CPVO TP/2/3 protocol as follows: tasselling (50% of plants shedding pollen), silking (50% of plants with visible silks) and physiological maturity (presence of the black layer on mid-ear kernels). Plant height (tassel included), primary ear height and ear/plant insertion ratio were measured on five plants per plot at 3 weeks after pollination, and the number of plants per plot and the percentage of lodged plants were recorded over the total plot as well. At maturity all ears of the central rows were hand harvested and shelled. Grain moisture (GM; %) and hectolitre weight (kg/hL) were measured with a GAC® 2100-AGRI grain moisture analyser (DICKEY-john, Auburn, IL, USA). Yield potential was calculated at 14% GM and adjusted to a final density of 70,000 plants per hectare.

Artificial inoculation, Fusarium ear rot severity and fumonisin content analysis

Artificial inoculation was performed according to the pin-bar inoculation method at 15 days after pollination (DAP) (Stagnati et al. 2020b; Guche et al. 2022). Briefly, the primary ear of five plants per plot was inoculated with a spore suspension of F. verticillioides MPVP 294 (ITEM 10027). This strain is maintained in the culture collection of the Department of Sustainable Crop Production, Università Cattolica del Sacro Cuore of Piacenza, Italy. The inoculum was produced as described in Lanubile et al. (2013, 2021) to a final concentration of 1 × 10⁶ conidia/mL. Inoculated ears were hand harvested at maturity and air dried in a barn.

Fusarium ear rot (FER) severity was visually evaluated, assessing the percentage of the rotted surface of the ear, using a seven-point severity grid and assigning 1 for the absence of symptoms, 2 for 1–3% infection, 3 for 4–10% infection, 4 for 11–25% infection, 5 for 26–50% infection, 6 for 51–75% infection, and 7 for 76–100% infection (Fig. 1; Stagnati et al. 2020b; Guche et al. 2022).

Fig. 1.

Rating scale of Fusarium ear rot severity and spread of the infection (%) on the F. verticillioides inoculated maize ears. Class 1 = no symptoms; Class 2 = 1–3% infection; Class 3 = 4–10% infection; Class 4 = 11–25% infection; Class 5 = 26–50% infection; Class 6 = 51–75% infection; Class 7 = 76–100% infection.

For the mycotoxin analysis, grains derived from the inoculated ears belonging to the same plot were bulked and further dried for 2 days at 65°C. A random subsample of 150–200 g was milled with a laboratory mill (Cyclotec™ 1093 Sample Mill, FOSS, Hilleroed, Denmark) using a 1 mm mesh. To avoid cross-contamination of kernels showing different disease levels special attention was taken in the milling procedure. Total fumonisins (B₁ + B₂ + B₃) were determined in kernels using the VICAM Fumo-V AQUA strips (VICAM, Watertown, MA, USA), a fluorometric-immunocapture assay having a limit of detection of 0.2 mg/kg and a quantitation range from 0 to 100 mg/kg (Guche et al. 2022). Values of fumonisins given in the text are expressed in mg/kg.

Statistical analysis

Data manipulation and visualisation was carried out using the software R (R Core Team 2017). Before proceeding with ANOVA analysis all data were transformed using the Box–Cox function (Morales et al. 2018; Stagnati et al. 2020b) using the MASS (Venables and Ripley 2002) and rcompanion packages (Mangiafico 2018). λ coefficients for transformations were: −5.2 for days to tasselling, −3.2 for days to silking, 3.3 for days to physiological maturity, 1.2 for plant height, 1 for ear height, 1.3 for ear/plant insertion ratio, 0.1 for percentage of lodged plants, 0.5 for yield potential, −0.6 for grain moisture, 0.1 for both FER and fumonisin content. A +1 constant was added for calculating the percentage of lodged plants to correct for a 0 score, in the case of plots with no lodging.

Two-way ANOVA and Fisher’s least significant difference (l.s.d.) test with the Bonferroni correction were performed using the car (Fox and Weisberg 2011) and agricolae (de Mendiburu 2017) R packages.

Correlation between different traits and years were investigated by the chart.correlation function available in the PerformanceAnalytics (Peterson and Carl 2018) package.

Evaluation of Fusarium ear rot severity and fumonisin accumulation

Significant differences in the expression of FER symptoms were revealed among landraces and years (P = 8.5 × 10⁻⁶ and 1.0 × 10⁻⁷, respectively), but not for the landrace × year interaction (P = 4.1 × 10⁻¹) (Table 2). Mean values of FER severity were 27.2 and 17.2 in 2019 and 2020, respectively (Supplementary Table S1). In 2019, U_G and Va229 showed the lowest and highest FER percentage (6.9% and 69.6%, respectively), whereas in 2020, EMR07 was the most resistant landrace (6.2% of FER), followed immediately after by U_G (6.4%), and EMR06 was the most susceptible (56.0% of FER).

Artificially inoculated landraces were also analysed for the content of total fumonisins (B₁ + B₂ + B₃) in kernels and significant differences were observed among landraces, years and landrace × year interaction, with the strongest trait effect due to the year (P < 2.0 × 10⁻¹⁶) (Table 2). Fumonisin levels ranged from 0.7 (Va221) to 12.0 (Va223) mg/kg in 2019 (mean value = 3.7 mg/kg), and from 4.2 (Va219) to 65.5 (Va213) mg/kg in 2020 (mean value = 20.3 mg/kg) (Table S1). Va211 and Va223 were found to be contrasting landraces considering an α = 0.07 (Table 2).

A good correlation (r = 0.66, P ≤ 0.001) was observed between FER severity and fumonisin accumulation in 2019, and between FER severity of both years (r = 0.40, P ≤ 0.05) (Fig. 2), whereas no correlation was observed between the two traits when both years were analysed together (Fig. 3).

Fig. 2.

Correlation between Fusarium ear rot (FER, %) and fumonisin (B₁ + B₂ + B₃) content (FUM, mg/kg) traits measured in 33 traditional maize landraces, Mais Nostrano di Storo and the commercial hybrid FAO 300 during 2019 and 2020 field trials. The distribution of each trait is shown on the diagonal. On the bottom of the diagonal the bivariate scatter plots with a fitted line are displayed. On the top of the diagonal the value of the correlation plus the significance level are reported as stars (*** P ≤ 0.001; * P ≤ 0.05).

Fig. 3.

Correlation among Fusarium ear rot (FER, %), fumonisin (B₁ + B₂ + B₃) content (FUM, mg/kg), days to tasselling (T, days), days to silking (S, days), days to physiological maturity (PM, days), plant height (PH, cm), ear height (EH, cm), ear/plant insertion ratio (E/P_IR, %), percentage of lodged plants (L, %), grain yield (Y, t/ha), and grain moisture (GM, %) traits measured in 33 traditional maize landraces, Mais Nostrano di Storo and the commercial hybrid FAO 300 during 2019 and 2020 field trials. The distribution of each trait is shown on the diagonal. On the bottom of the diagonal the bivariate scatter plots with a fitted line are displayed. On the top of the diagonal the value of the correlation plus the significance level are reported as stars (*** P ≤ 0.001; * P ≤ 0.05).

Evaluation of agronomic performance

The agronomic parameters of the maize landrace collection are described in Tables 2 and S1. In general, great phenotypic variability was observed for all agronomic traits measured in 2019 and 2020.

Significant differences were detected among the 33 maize landraces, N_S and the commercial hybrid for the traits associated with flowering, days to tasselling and days to silking, at both landrace (P = 6.7 × 10⁻¹⁶ and 1.9 × 10⁻¹⁴) and year (P = 5.5 × 10⁻⁴ and 1.4 × 10⁻⁵) level, whereas the landrace × year interaction was not significant (P = 1.3 × 10⁻¹ and 2.9 × 10⁻¹) (Table 2). During 2019, the earliest and latest landraces were U_G and Va217, and the mean values were 70.9 and 75.0 days to tasselling and silking, respectively (Table S1). During 2020, Va217 was confirmed as the latest landrace, whereas the earliest anthesis and silking were recorded for M_P and EMR10, respectively. On average, anthesis was reached at 69.8 days and silking at 73.3 days (Table S1).

Cycle length is determined at the appearance of the so called ‘black layer’ at the kernel pedicel. This trait significantly differed among landraces (P = 2.0 × 10⁻¹⁶), years (P = 2.0 × 10⁻¹⁶), and landrace × year interaction (P = 3.9 × 10⁻¹³) (Table 2). For both years, the earliest landrace was M_P, which reached maturity at 108.5 and 113.7 days, whereas the latest was confirmed as Va217, maturing at 135.5 and 125.5 days for 2019 and 2020, respectively (Table S1). Considering the entire germplasm collection, physiological maturity occurred at approximately 123 days (Table 2). The commercial hybrid reached maturity at 127.5 and 123.5 days in the 2 years 2019 and 2020, respectively; this is classified in the early maturity class FAO300 and suggests that the maize collection from Emilia-Romagna is characterised by a widespread presence of early landraces.

As expected, cycle traits (days to tasselling and silking, and physiologic maturity) were positively and significantly correlated with each other (P ≤ 0.001), and the highest correlation was observed for anthesis and silking traits (r = 0.94) facilitating the proper pollination of the collection (Fig. 3).

The mean values of plant and ear height in 2019 and 2020 varied significantly among landraces at both genotype and year level (P = 2.0 × 10⁻¹⁶), by contrast a significant landrace × year interaction was found only for plant height (P = 4.8 × 10⁻⁷), as reported in Table 2. The average of both traits was higher in 2020 than 2019, with mean values of 199.2 cm and 77.3 cm in 2019, and 227.5 cm and 92.8 cm in 2020 for plant and ear height, respectively (Table S1). The shortest variety was M_P for both traits in both years, whereas the hybrid was found to be the tallest genotype (230.9 cm and 259.7 cm in 2019 and 2020, respectively). The most elevated ear height was recorded for Va217 (108.6 cm and 120.7 cm in 2019 and 2020, respectively) (Tables 2 and S1). Plant and ear height were found to be significantly and positively correlated each other (r = 0.68; P ≤ 0.001) and with cycle duration traits (Fig. 3), illustrating how early landraces were generally characterised by reduced height.

From plant and ear insertion height it was possible to compute an index of plant architecture that is the ear/plant insertion ratio. This trait was influenced by landrace (P = 2.0 × 10⁻¹⁶), year (P = 9.1 × 10⁻³) and landrace × year interaction (P = 2.9 × 10⁻⁶), as shown in Table 2. Mean values of the collection were similar during the 2 years (Table S1), with EMR10 and Va217 being the two most distant landraces for this trait (Table 2). Moreover, ear insertion height and ear/plant insertion ratio showed significant associations with the different kernel types (dent, intermediate and flint) of landraces (P = 0.038 and 0.0068, respectively) (Fig. 4a, b).

Fig. 4.

Relationship between: (a) ear height (EH, cm) and kernel type (dent, intermediate, and flint); (b) ear/plant insertion ratio (E/P_IR, %) and kernel type; (c) percentage of lodged plants (L, %) and ear morphology; (d) grain yield (Y, t/ha) and kernel type; (e) altitude (ALT, m) and kernel type; (f) altitude (ALT, m) and ear morphology. Differences among groups were detected by a pairwise-t-test. The used bowblot convention was 1.5 IQR.

A further agronomic trait related to plant architecture is the susceptibility to lodging. In both years, a moderate percentage of lodging was recorded (Table 2). In 2019, lodging percentage varied from a minimum of 0.3% for EMR01 to 40.0% for Va218 with a mean percentage of 14.6% (Table S1). During 2020, the percentage of plant lodging was lower, varying from 0% for Va220, EMR01, EMR04, EMR07, N_S and the commercial hybrid, to 21.7% for Va211, with a mean percentage of 7.8% (Table S1). Significant differences were observed at both landrace (P = 2.0 × 10⁻¹⁶) and year (P = 1.6 × 10⁻⁸) level, whereas landrace × year interaction was not significant (Table 2). The higher incidence of lodging during 2019 was probably due to an adverse meteorological event like strong wind and heavy rains that occurred after flowering. Significant moderately negative correlations were found between lodging and time to tasselling (r = −0.29; P ≤ 0.05), and lodging and physiological maturity (r = −0.34; P ≤ 0.05) (Fig. 3). It was previously reported that early root lodging occurred during wind and rainstorm events around flowering time when corn was in an awkward (top-heavy) stage, and this phenomenon differed depending on the years and localities (Troyer 2000).

Lodging percentage was related to ear shape of landraces and significant differences among the three different ear morphologies (slightly conical, conical and cylindrical; P = 0.018) were observed (Fig. 4c). Conversely, kernel types and sampling location did not affect this trait (data not shown). Landraces with conical ears were more prone to lodging (Fig. 4c), and this could be due to a modification of the centre of gravity of the plants, mainly in those with a tall architecture.

Grain yield represents the driving factor for plant breeding and one of the key elements to take into account for landrace exploitation. Average production was about 4.7 t/ha in 2019, and 6.0 t/ha in 2020 (Table S1). The best performing landrace was EMR04 (6.3 t/ha) and Va217 (8.9 t/ha) in 2019 and 2020, respectively. Landraces and years were significantly different (P = 2.0 × 10⁻¹⁶), whereas landrace × year interaction was not significant (Table 2). Significant positive correlations were observed between yield and days to tasselling (r = 0.35; P ≤ 0.05), silking (r = 0.36; P ≤ 0.05), and physiological maturity (r = 0.66; P ≤ 0.001), as well as between yield and plant and ear heights (r = 0.60 and 0.62, P ≤ 0.001, respectively) (Fig. 3). Similar findings have been already reported (Cömertpay et al. 2012; Stagnati et al. 2020b), highlighting how a longer vegetative phase allows bigger plants and higher grain yields. A significant moderately negative correlation (r = −0.35; P ≤ 0.05) was detected between yield and FER disease, in line with Stagnati et al. (2020b) results. To understand the potential of the landrace collection as source of genetic variability without any detrimental effect on grain yield, the commercial hybrid characterised by a short cycle was used as comparison. Nearly all landraces showed average productions comparable to that of the hybrid, and some landraces displayed yield values higher than the hybrid, such as Va211, Va217, Va218, Va222, Va225, Va227, EMR01 and EMR04 (Table 2). Even though a few of them still present some morphological defects such as open leaf angle, lodging susceptibility and big tassels, these landraces represent good sources of genetic diversity and may be used to extract new inbred lines or improved ‘per se’ through mass selection.

Differences in yield values were found among landraces characterised by different kernel types (P = 0.006), as reported in Fig. 4d. Dent and intermediate landraces were characterised by a similar yield, whereas the flint ones had reduced yields. Further considerations are necessary for M_P and EMR06 landraces, which were removed from the correlation analysis. M_P showed very bad agronomic performance, explained by the presence of silk-cut and popped kernel symptoms and a low seed set under hot climate conditions. Even including these landraces in the analysis, yield distribution results according to kernel types were not affected (P = 0.0024), and the comparison between dent and intermediate kernel type landraces still remained not significant (data not shown). Interestingly, altitude of landrace sampling locations was significantly related to kernel types (P = 0.039) (Fig. 4e).

Average grain moisture was 14.8% in 2019 and 17.6% in 2020, with an effect of the year (P = 2.0 × 10⁻¹⁶) higher than that of landrace (P = 3.0 × 10⁻²) (Tables 2 and S1). Interestingly, in both years the landrace with the highest moisture content was M_P, an accession characterised by a strongly conical ear morphology. Therefore, a significant relationship between ear morphology and altitude of sampling location was displayed (P = 0.037) (Fig. 4f). Landraces characterised by the presence of conical ears were the most abundant and sampled from plain to high mountains (over 1000 m above the sea), landraces with slightly conical ears were few and generally sampled in areas between 500 and 700 m above the sea, whereas maize with cylindrical ears were localised mainly in the plain or hills (altitude less than 500 m).