Results demonstrated that there were differences between hybrids for many phenomorphological traits and competitive factors, that some phenomorphological traits accounted for differences in competitive ability in the hybrids, and that, for some of these traits, hybrid performance can be predicted based on parentage.

Consistent differences between hybrids existed for most phenomorphological traits and competitive factors. Hybrid effects existed for all traits except ear length, ear length stability, and ear width stability (Table 1, Table 2). Hybrid-by-environment (H × E) effects existed for 100 kernel mass, ears per plant, early leaf area, late leaf area, average leaf height, early height, mid-season height, and yield stability. However, Spearman rank-change H × E effects existed only for ears per plant and average leaf height. In all traits that had hybrid effects, GCA effects also existed, except for 100 kernel mass stability. In all traits that had hybrid effects, SCA effects also existed, except for 100 kernel mass stability, average leaf height, average number of tillers, mid-season height, ear number stability, and early sorghum biomass.

Correlations were seen between the sorghum-free plot means of many of the phenomorphological traits and the three factors used to assess competitive ability: early sorghum biomass, late sorghum biomass, and yield stability (Table 3). The three traits most correlated with reduced early-season sorghum mass were mid-season plant height (r = −0.62) (Figure 1), early-season leaf area (r = −0.61) (Figure 2), and yield (r = −0.60) (Table 3). Stepwise regression found that early-season plant height and early-season leaf area combined accounted for 67% of the variation present in early-season sorghum mass (p < 0.01), and additional traits did not significantly improve the model (data not shown). Early-season plant height (r = −0.78) (Figure 3), 100 kernel mass (r = −0.62), and yield (r = −0.61) were most correlated with reduced late-season sorghum biomass. Early-season plant height accounted for 61% of variation in late-season sorghum biomass, with additional traits not significantly improving the regression model. Mid-season plant height (r = 0.58) (Figure 4), late-season plant height (r = 0.56), and upper leaf angle (r = −0.53) were the three factors most correlated with yield stability. Mid-season plant height accounted for 33% of variation in yield stability, with additional traits not significantly improving the regression model.

Large differences were seen in the potential combining abilities for the inbred parents both for the factors measuring competitive ability, and for the phenomorphological traits that correlate with competitive ability. For example, averaged across all environments, the range of GCA for yield, i.e., the difference between the inbred with the highest GCA for yield and the inbred with the most negative GCA for yield, was 1341 kg ha⁻¹ (Table 4), the range for early leaf area GCA was 256 cm² (Table 4), the range for late plant height GCA was 26 cm (Table 5), the range for early height was 8.4 cm (Table 5), the range for early sorghum biomass GCA was 3.0 grams (Table 6), the range for late sorghum biomass was 26 grams (Table 6), and the range for yield stability GCA was 21% (Table 7). By comparing the mean squares of GCA to total hybrid mean squares (2MSGCA: 2MSGCA + MSSCA), it can be shown that most of the variation in competitive ability and related traits is predictable based on inbred GCA alone (Table 8).

Some of the genotype by environment interaction effects seen in many of the traits may be due to the variable growing conditions experienced between environments. Rainfall and temperature may have contributed to creating these different growing environments.

Some previous conclusions about traits relating to competitive ability were supported by the results of this study. Although, unlike Roggenkamp et al. [20], upright leaves were not found to be correlated with WSA, upright leaves were found to be correlated with increased yield stability. While Makus [21] and Williams et al. [12,13] found greater WSA in late hybrids and Begna [22] found greater WSA in early hybrids, no differences in WSA based on maturity were observed in this study. Although late-season leaf area was not found to be correlated to WSA, early-season leaf area was, which supports Woolley and Smith’s [23] finding that leafy corn had greater WSA. In contrast to Makus’s [21] finding that yield was reduced more in a taller cultivar than a shorter one, this study found that, in general, plant height correlated with both increased CT and increased WSA. This study had more entries present than some previous work [8,12,13,19,20,21,22,24,25], allowing greater legitimacy to inferences made relating traits such as plant height to competitive ability.

If breeders are choose a few traits to use as a basis of selection for improved competitive ability, which should they be? Plant height, especially early- and mid-season plant height, showed the greatest correlation with both reduced sorghum biomass and yield stability (Table 3, Figure 1, Figure 3 and Figure 4), and that hybrid plant height was highly predictable based on the heights of the inbred parents (Table 8). Correlation is not causation and plant height may simply be the measured trait that is linked to other unmeasured factors that cause increases in competitive ability. For example, a recent modeling study [26] suggested that root characteristics may have been more important than canopy characteristics in the improvement of stress tolerance in commercial field corn hybrids. However, the results of this study suggest that regardless of the underlying factors responsible for competitive ability, selection for early- and mid-season plant height may lead to more competitive sweet corn.

Can a diallel analysis help breeders identify superior inbreds for competitive ability? Taking this study as an example, it can be shown that C68 stands out as a competitive inbred. It produced hybrids which on average had the highest yield, the highest yield stability, and the lowest early- and late-season sorghum biomass (Table 4, Table 6 and Table 7). C68 also had the largest GCA in the positive direction for early- and late-season plant height and for vertical leaves (Table 5, data not shown). C68 produced both the tallest and the most tolerant hybrids, which is consistent with So et al.’s [27] finding that plant height, along with other ‘late canopy and maturity’ factors, was most effective in explaining the differences in CT in sweet corn.

While C68 stands out as most competitive, P39 stands out as least competitive. P39 produced hybrids which had the lowest yield, yield stability and the most early- and late-season sorghum biomass (Table 4, Table 6 and Table 7). P39 also contrasts with C68 in that P39 produced hybrids which were, on average, shortest in the early season and had the most horizontal leaves (Table 5, data not shown).

Two key parameters influence the context of this study and should be considered when making inferences from this work about competitive ability in other systems. First, the interaction of two organisms depends partially on the relative fitness of each organism in a particular environment. The combination of the interactions that the two organisms have with each other, and the interactions that each have with the environment can make results obtained from evaluating the relationship between two organisms more sensitive to environmental variability than research focused on a single organism. In this study, competitive ability was most influenced by plant height. However, care must be taken when making any generalizations to other environments, especially those environments in which other resources, such as water or nutrients, may be more limited. In order to broaden the scope of the conclusions, similar experiments could be conducted in varied environments.

The second parameter that influences the context of this study is the choice of model weed. Forage sorghum is a good model due to its high competitiveness, accelerated growth and substantial biomass production and shading ability. It also offers the advantage of being able to be managed similarly to corn. However, can information gained from research with sorghum be applied to other weeds? Sorghum is a C4 grass similarly to corn. It has narrow, long leaves and has upright growth ability. But sorghum has a determinate flowering pattern, which limits plant height. Many broadleaf weeds have an indeterminate growth habit resulting in greater plasticity in plant height. Despite the differences between sorghum and broadleaf weeds, some inferences may be justifiable. Previous studies have found that, despite large differences in morphology, weed species can present similar competitive pressures. Moechnig et al. [28,29] found that common lambsquarters (Chenopodium album) and giant foxtail (Setaria faberi) had similar competitive abilities in corn, depending on the environment.

An additional difference between sorghum and wild weeds is that sorghum will germinate consistently. It is useful for a model weed, as it allows uniform pressure to be created. However, weeds and other wild plants generally have extended germination periods, emerging throughout the season. Corn competitive traits that act on young weed seedlings would only be important against sorghum at the beginning of the season, but would continue to be important through-out the season when in competition with wild weeds. Conducting a similar experiment with weeds which had different emergence patterns would allow broader conclusions to be made.

Nineteen hybrids were developed from an incomplete seven-line half-diallel without parents (Griffing’s Method 4) [30]. The following seven sugary1 (su1) inbreds served as parents: Ia5125, C40, C68, We10, Ill101t, P39, and P51. These inbreds were chosen to represent a wide range of morphologies, and a diversity of sweet corn ancestors. Jubilee, a commercial su1 hybrid, was also included as a check to provide a point of reference. The sorghum variety used was Silo 700D, a short-statured forage sorghum, chosen for its abundant biomass production and similarity in plant height to the sweet corn hybrids.

The experiment was a split block randomized complete block design with three replications per environment. The main-block factor was presence or absence of sorghum and the sub-block factor was hybrid. Plots were four rows, with rows 3.5 m long, 0.76 m between rows, and 0.91 m alleys. Each plot consisted of one hybrid and each split block contained the set of nineteen hybrids and was interplanted either with or without sorghum. The experiment was conducted in 2007 at the Arlington Agricultural Research Stations in 2008 at the University of Wisconsin-Madison West Madison. The soil type at all locations was Plano silt loam (fine-silty, mixic mesic Typic Argiudoll).

The sweet corn hybrids were planted at Arlington, WI (ARL07) on 3 May in 2007 and at West Madison, WI (WM08) on 6 May in 2008. Because of limited seed, only 19 of the 21 possible diallel entries were planted in ARL07, with We10 x IL101t and C40 x IL101t not planted. Jubilee was planted in place of the missing entries.

The target density of sweet corn after emergence was approximately 93,600 plants ha⁻¹ (25 seeds row⁻¹) Around V6 [31], the center two rows of each plot were thinned to 14 plants row⁻¹ (52,400 plants ha⁻¹), while the outer two rows were thinned to 16 plants row⁻¹ (59,900 plants ha⁻¹), leaving extra plants for destructive leaf area sampling. After leaf area sampling (V8-VT, see below) all rows had a final density of 14 plants row⁻¹, or 52,400 plants ha^-1.

In the sorghum treatment sub-blocks, after the corn had emerged (V1–V2), sorghum was hand planted in between the corn rows in two rows spaced 10 cm between sorghum rows and 20.5 cm away from the nearest corn row. The sorghum was thinned to a final density of 36 plants row⁻¹ or 135,300 plants ha⁻¹. This density was chosen based on Myers et al. [32], which found that field corn yields were reduced 20% to 40% when sorghum was interplanted at a density of approximately 140,000 plants ha⁻¹.

Throughout the growing seasons of 2007 and 2008, phenomorphological, crop tolerance and weed suppressive traits were measured on the sweet corn hybrids and sorghum. Phenomorphological traits measured were: early-season, mid-season, and late-season plant height; height of tallest and second-tallest tillers; average tiller height; date of anthesis; leaf angle; leaf area; yield; ear length; ear width; mass of 100 kernels; and ears per plant. Traits measuring CT were: yield stability, 100 kernel mass stability, ear length stability, ear width stability, and ears per plant stability. The trait measuring WSA was sorghum biomass. Early-season [ARL07: V8, 519 growing degree days (GDD, 10°C/30° C); WM08: V6, 482 GDD] and mid-season (ARL07: V12, 762 GDD; WM08: V12, 814 GDD) sweet corn plant heights were visually estimated as an average of the center two rows of each plot from the soil surface to the height of the tallest leaf at its apex by placing a measuring stick in front of each plot. Late-season plant height was taken post anthesis (ARL07: 1630 GDD, WM08: 1509 GDD) and measured from the soil surface to the collar of the leaf subtending the tassel on five random plants from the third row of each plot. Number of tillers was recorded on the same plants as the late-season plant height and at the same time. Date of anthesis was recorded when 50 percent of tassels in the center two rows had exserted 50% of their anthers. Sweet corn leaf angles were measured post-anthesis (ARL07: 1630 GDD, WM08: 1141 GDD) on five random plants in the center two rows of each plot. The angle was measured with a protractor, with the zero center placed at the intersection of the leaf above the ear leaf and the stem, and the angle recorded as degrees from parallel to the stem [33]. Total leaf area was measured early in the season (ARL07: V6, 397 GDD; WM08: V4, 316 GDD) and later in the season (ARL07: VT, 990 GDD; WM08: V8, 557 GDD). Two sweet corn plants from each plot were harvested from each of the outer two rows during the early harvest and from each of the center two rows for the late harvest.

Sweet corn yield was measured on ears harvested from the center two rows after physiological maturity and dried to constant moisture. The following yield components were also measured for each plot: uppermost ear length, uppermost ear width, the mass of 100 kernels, number of ears per plant. The stability of yield and yield components (100 kernel mass, ear length, ear width, and ears per plant) were calculated as the value measured for a hybrid in the sorghum treatment divided by the value of the same hybrid in the non-sorghum treatment of the same block, and expressed as a percentage. Sorghum biomass was measured twice in each experiment, once near sweet corn anthesis (ARL07: 991 GDD, WM08: 1106 GDD) and once at physiological maturity (ARL07: 1876 GDD, WM08: 2045 GDD). Sorghum was taken from between sweet corn rows one and two during sweet corn anthesis and between rows three and four at sweet corn maturity. Sorghum was harvested by cutting all plants at 0.30 m from the soil surface and weighing in the field. A subsample was weighed, dried completely and weighed again to estimate dry mass for each plot.

Linear mixed model analysis was conducted for all traits measured using the PROC MIXED in the SAS statistics package (SAS Institute, Cary, NC, USA). Block effects were considered random, with all other effects considered fixed. For all traits with H × E effects, Spearman rank correlations were calculated to determine if hybrid-by-environment effects were due to a change in magnitude or a change in rank. For each trait, environments were pooled if there were no environment-by-treatment or environment-by-hybrid effects, or if a Spearman rank correlation between environments for each trait or treatment condition demonstrated that the interaction was due to a change in magnitude and not a change in rank. Prior to calculating mean squares, residuals were analyzed to ensure they meet the normality and equal variance assumptions.

General combining ability (GCA) is the deviation of the mean performance of all crosses derived from an inbred line from the mean of all crosses in a diallel. GCA can be used to predict how a hybrid will perform based on the performance of the inbred parents. Specific combining ability (SCA) is the deviation in performance of a particular cross from its expected performance based on the mean of all crosses and the GCAs of its parents. If hybrid effects were significant, they were partitioned into GCA and SCA effects based on Griffing Model 2, Method IV [30]. Because a complete diallel of all seven lines was not represented in all environments, a design matrix for the SCA and GCA components was constructed based on the procedures in Wu and Matheson [34] and calculated based on Zhang and Kang [35] using CONTRAST and ESTIMATE statements in PROC MIXED in SAS (SAS Institute, Cary, NC, USA).

Means were compared using protected least significant differences (LSD) at p < 0.05 significance level. Correlations were calculated for traits with significant (p < 0.05) hybrid effects, as determined by the ANOVA. Correlations were done on phenotypic means using PROC CORR in SAS (SAS Institute, Cary, NC, USA). In order to remove the compounding effects of sorghum on phenomorphological traits that may affect weed competitiveness, sorghum plots were excluded from the means of the phenomorphological traits when calculating correlations between phenomorphological traits and competitive traits. Regressions between traits were calculated using stepwise regression with PROC REG in SAS (SAS Institute, Cary, NC, USA). The predictability of GCA for determining progeny performance was calculated based on the ratio of 2MSGCA to 2MSGCA + MSSCA [36].