Phenotypic characterization of southeastern United States open-pollinated maize landraces

Maize is the most important crop in the United States, but food production and niche uses such as distilling often rely on varieties that are not bred explicitly for these purposes. Farmers targeting niche food grain markets have expressed interest in historical open-pollinated varieties of maize, but few such populations are widely available, and even fewer are well-characterized. We planted field trials of a subset of 76 open-pollinated heirloom varieties available from catalogs and the USDA germplasm collection along with four F1 hybrid cultivars. We measured 24 traits across three years and three locations to characterize the selected varieties and measure their phenotypic relationships. We identified a subset of 19 traits useful for phenotypic analysis. Our results demonstrate that many historical accessions contain substantial genetic variation that should permit improvement from within-population selection. Variety name and origin are often not useful indicators of phenotypic relationships or potential crop value. Cluster analysis identifies nine morphologically distinct groups within the varieties tested, which are not fully in agreement with currently accepted landrace groupings, suggesting a need for genetic analysis of larger samples of USA open-pollinated populations to better define the natural classification of maize indigenous to the United States.

Methods
In 2017, a single plot of each variety was planted at Clayton, NC.  Each plot consisted of four 7.6-m rows of the same variety, seeded at a rate of 25 seeds per plot (100 seeds total per population). Seed was produced from each plot by bulking pollen within the plot each day and pollinating as many ears as possible by hand. Seed was bulked from each population and used as source seed for subsequent experiments. Check plots of T175 × T177 and T177 × T179 F1 hybrids were planted in three replicates distributed throughout the experiment field. 

In 2018 and 2019 an 8×10 alpha-lattice design with three full replications was planted in each of three locations: Clayton, NC; Pendleton, SC; and Columbia, MO. Experimental units were single 7.6-m rows sown with 25 seeds of each variety.  Experimental fields were irrigated as needed with overhead irrigation at Columbia and Clayton and drip irrigation at Pendleton.  At Clayton, fields received nitrogen at 53.2 kg ha-1 pre-planting, and then 190.5 kg ha-1 in a split application during the growing season.  At Columbia, fields received 134.5 kg ha-1 N pre-planting, then 11.2 kg ha-1 in a single application during the growing season. At Pendleton, fields received 44.8 kg ha-1 N pre-planting.  

We collected data on a subset of the morphological traits previously recommended for maize racial classification by Sánchez G. et al. (1993) from each of the four rows of the seed increase plots in 2017 and from each experimental plot in 2018 and 2019 (Table S1). Days to anthesis (DTA), days to silking (DTS), anthesis-silk interval, mean numbers of ears and tillers per plant, and lodging percentage were measured on a plot-basis. DTA and DTS were the first dates on which half or more of the plants in a plot were shedding pollen or silking, respectively. Anthesis-silking interval (ASI) was calculated as the difference between DTS and DTA. Mean numbers of ears and tillers per plant were measured as the total count of ears or tillers per plot, divided by number of plants per plot. The end plants in each plot were not included in counts of ears or tillers or in the total count of plants per plot. Lodging percentage was the proportion of plants within a plot that exhibited root lodging (stalks leaning from base of the plant by more than 30° from upright) or stalk lodging (stalks broken below the primary ear or dropped primary ear).

Other traits were measured on one randomly selected individual plant per row in 2017 and on two plants per row in 2018 and 2019. Plant and ear heights were measured approximately two weeks after pollination. Plant height was measured from base to node of the leaf subtending the tassel; ear height was measured from base to the primary ear node. Ear position, the ear height relative to plant height, was calculated as (Ear height/Plant height)*100. Leaf length was measured along the midrib from ligule to tip on the leaf above the primary ear. Leaf width was measured at the widest point on the same leaf. Tassels were detached from the plant and measured in the field.  The number of primary branches per tassel were counted.  Tassel length was measured from lowest branch node to tip of main spike.

            In year 2017, one sib-pollinated ear was selected from each row to measure ear and kernel traits. In years 2018 and 2019, two plants per plot were randomly chosen for self- or sib-pollination, and at maturity, two self-pollinated ears from each plot were collected, along with two open-pollinated ears from other plants in the plot (not including the end plants) for ear and trait measurements.  Sib- or self-pollinated ears were used to measure husk and shank traits, and the range of kernel colors among all sibbed/selfed ears per population were noted (Table 1).  Open-pollinated ears generally had more complete pollination and remained healthier during maturation, and thus were used to measure other ear and kernel traits in 2018 and 2019.

            Husk color was recorded as a binary trait for presence or absence of anthocyanin in the outer husk leaves.  Husk number was the number of husk leaves covering each ear.  Shank length was measured with a Fowler Xtra-Valu caliper (54-101-300-1; Newton, MA). Ear diameter was measured by caliper at the middle of the ear with seeds attached to the cob.  Kernel row number was counted at the middle of the ear.  Each ear was assigned a percentage ear rot score based on the proportion of the ear exhibiting signs of rot diseases incited by Fusarium, Aspergillus, or Diplodia fungi.  Ears were dried in a forced-air drier for two weeks and kernels were removed from ears to permit cob and kernel measurements. Cob length was measured with a caliper from the base of the ear to the tip after removal of staminate tips (where present). Cob color was noted on a 5-point scale from 1 (white cob, no pigment) through 5 (dark red; Figure S1).  Cob diameter was measured at the middle of the cob with a caliper. Kernel width was estimated as the mean of four individual kernel values measured with calipers per ear. A sample of 100 kernels was randomly selected from each ear to measure 100-kernel weight.