Location of experimental work

The experiment was carried out in the 2019-2020 winter-spring cycle at the Faculty of Agricultural Sciences, dependent on the Autonomous University of the State of Mexico, located 18 km north of the city of Toluca, with geographical coordinates of 19° 15’ 33” north latitude, 99° 39’ 38” west longitude and at an altitude of 2 640 masl. The predominant climate is type C (w2) (w) b (i), which corresponds to a subhumid temperate climate, the most humid of the subhumid climates, with rainfall in summer and little rainfall in winter (5%), little thermal oscillation, average annual temperature of 12.8 ºC and average annual rainfall of 900 mm.

The predominant soil of the region is of the pelic vertisol type of volcanic origin, it has a mineral horizon with a low content of organic matter that ranges from 1.01 to 2.36%. The color of the dry surface horizon is dark grayish-brown or dark gray with clay contents of 20 to 36.4%. In the soil profile, a horizon with tillage disturbances showing compaction due to plow floor can be distinguished, the amount of organic matter is very low, it ranges from 0.07 to 1.01 (^{Gil et al., 2014}).

Treatments and experimental design

Two sowing systems were evaluated, DRBS: double-row bed system (24 plots) and FBS: flat bed system (24 plots). The treatments derived from the factorial arrangement of two canola genotypes, Hyola 61 (hybrid) and Bioaureo 2486 (open pollination), and 3 densities were established under a randomized complete block design with four repetitions in each of the aforementioned systems (two serial experiments) considering the two sowing systems.

General conditions of the experiment

The sowing was carried out on December 7, 2019, in two sowing systems, the first made up of plots of 3 m in length with three beds 80 cm apart with two rows, each 20 cm apart (7.2 m²) double-row bed system (DRBS), while the second had plots of 5 rows of 3 m in length and separated from each other at 0.3 m (4.8 m²) flatbed system (FBS). Three sowing densities were used, 50, 75 and 90 seeds m^-2. To ensure the three sowing densities, the seed was placed on biodegradable paper tapes at a distance that would guarantee the required density. The crop was kept with moisture at field capacity during its cycle and without nutritional restrictions, so it was fertilized with a formula of 150-60-30 (NPK), using urea as source of nitrogen, triple calcium superphosphate as a source of phosphorus, potassium chloride as a source of this element.

Biomass production, yield and its components

At physiological maturity, the plants in a linear meter of the central furrow of each plot were cut and separated into branches and main stem, separating the siliques of each stratum. The dry matter of each stratum (branches, main stem and siliques) was determined after drying in a forced air oven for 72 h at 60 °C. The siliques of each stratum were counted and threshed separately to obtain the dry weight of seed (yield). The number of siliques, the number of seeds per silique and the number of seeds per m² were determined. The weight of 1 000 seeds was determined by counting and weighing 1 000 seeds of each experimental unit.

Statistical analysis

The response variables of the experiment were subjected to a combined analysis of variance and when the F test of the analysis of variance was significant, the mean test was performed using the honest minimum significant difference (HMSD) or Tukey test at a significance level of 5% (^{Palaniswamy and Palaniswamy, 2006}), using the statistical package R for Windows version 4.0.5 based on the following model:

Where:

is the general mean,

is the effect of the i-th system,

is the effect of the j-th block within systems and was the error term used to test systems,

is the effect of the k-th density,

is the effect of the l-th genotype,

is the effect of the interaction of the k-th density on the i-th system,

is the effect of the i-th system on the l-th genotype,

is the effect of the k-th density on the l-th genotype,

is the interaction of the i-th system in the k-th density in the l-th genotype and

is the general error term. The relationships between variables were determined by linear regression and the association between them by correlation analysis (^{Jandel, 1991}).

Climatic conditions

The climatic conditions during the development of the experiments are shown in Figure 1, in which it is observed that the emergence occurred 14 days after sowing. On the other hand, it can be seen that the beginning of the reproductive period in both genotypes coincided with favorable temperatures for the crop. It should be noted that precipitation during the reproductive period was very scarce, but this was not a limitation since the crop was established under irrigation conditions. Towards physiological maturity, an increase in the recorded precipitation levels was observed, which did not affect the development of the experiment. In general, factors such as precipitation and temperature were favorable for the development of the experiment.

The results of the analysis of variance presented in Table 1 show that for the factor sowing system, highly significant differences (p< 0.01) were only observed for the variables biomass, weight of 1 000 seeds and number of branches. On the other hand, for genotypes, highly significant differences (p< 0.01) were observed in biomass, weight of 1 000 seeds and number of branches and significant differences (p< 0.05) were only observed for harvest index. For the interactions, no significant effects were observed for most of the variables, except for the interaction system x density x genotype, which showed highly significant effects (p< 0.01) for number of siliques. Finally, the coefficients of variation ranged between 1.2 and 20.8, which correspond to the variables weight of 1 000 seeds and biomass at physiological maturity, these coefficients are considered acceptable given the nature of the experiment.

Figure 1 Agroclimatic conditions occurred during the conduct of the experimental work. (S= sowing; Eme= emergence; BFV= visible flower bud; IF= beginning of flowering; IFR= beginning of fruiting; MF= physiological maturity). 

Table 2 shows the comparison of means for the variables evaluated, where the flat bed system (FBS) was significantly higher than the double-row bed system (DRBS) in biomass production, weight of 1 000 seeds and number of branches per plant, for the rest of the characters both sowing systems were statistically the same. The superiority of the flat bed system over the double-row bed system lies in the fact that, in the first, the separation of the rows was 30 cm, while in the second, the separation was 20 cm, but in beds 80 cm apart. Evidence in the literature shows that a smaller spacing between rows produces a higher seed yield as a result of a higher number of siliques and seeds per m² (^{Shahin and Valiollah, 2009}; ^{Uzun et al., 2012}).

Increases in plant density are reflected in higher biomass production and seed yield, as a result of a higher number of seeds per m² (^{Shahin and Valiollah, 2009}; ^{Uzun et al., 2012}). The data reveal that D90 (90 plants m^-2) was statistically higher than D75 (75 plants m^-2) and D50 (50 plants m^-2) in most of the variables evaluated, except for the harvest index, weight of 1 000 seeds and number of seeds per silique.

The genotypic effect on population density in canola has been analyzed in many studies, where differences in seed yield have been minimal (^{Różyło and Pałys, 2014}), while others mention that differences in yield may be due to the different structure of the plant (^{Waseem et al., 2014}). In this sense, our data show that Hyola 61 was statistically higher (14%) than Bioaureo 2486 in harvest index, while Bioaureo 2486 was statistically higher than Hyola 61 in biomass production (18%), weight of 1 000 seeds (3%) and number of branches (11%) (Table 2). The differences in some characteristics between both cultivars may be due to the different plant structure that exists between them since Hyola 61 produces fewer branches than Bioaureo 2486.

The production of biomass at maturity and the harvest index are very important characteristics to reach high yields in grain crops and particularly in canola (^{Zhang and Flottmann, 2016}). The contribution of both characters to seed yield strongly depends on the type of cultivar (hybrid or open pollination), but usually biomass production largely explains changes in seed yield (^{Li et al., 2016}). The data indicate that, when considering sowing systems, densities and genotypes, seed yield was mainly explained by the changes observed in biomass production at maturity (r²= 0.81, p< 0.001) and not by the harvest index (Figure 2), apparently the latter was more stable, as has been reported in other grain crops such as triticale (^{Estrada et al., 2012}) and wheat (^{Parry and Hawkesford, 2010}).

It has been established that the distance between rows is a very important management factor since it affects the seed yield and the components of the yield in individual plant in canola (^{Uzun et al., 2012}; ^{Różyło and Pałys, 2014}). Our data revealed that when densities and genotypes are considered, the flat bed system (5 rows to 30 cm) obtained a higher biomass production compared to the double-row bed system (Table 3; Figure 2a).

Figure 2 Relationships between seed yield and biomass produced at maturity (a) and with harvest index (b) for two canola cultivars grown under three plant densities and in two sowing systems in Toluca, Mexico. Solid and dashed lines in the graph are the regression models fitted for DRBS (SCDH) and FBS (SCP), respectively. 

On the other hand, increases in sowing density significantly increased seed yield in both sowing systems, with increases of 22 and 20% when going from 50 to 90 plants m^-2 in Bioaureo 2486 and Hyola 61, respectively, for the double-row bed system. While in the flat bed system, the increases in yield, when going from 50 to 90 plants m^-2, were 23 and 28% for Bioaureo 2486 and Hyola 61, respectively (Table 3).

Positive effects in reducing the distance between rows have been reported in numerous studies, for instance, ^{Shahin and Valiollah (2009}), when evaluating three distances between rows (12, 18 and 24 cm), found significant increases in seed yield in rows 12 cm apart. On the other hand, ^{Uzun et al. (2012}), when evaluating different spacings between plants (5, 10, 15 and 20 cm) and between furrows (10, 20, 30 and 40 cm), found a higher yield with distances between plants of 5 to 10 cm and separation between furrows of 10 cm. In the same way, ^{Różyło and Pałys (2014}), when evaluating different spacings between furrows, found that greater separation decreased both biomass production and yield per m². Recently, ^{French et al. (2016}), when evaluating densities between 7 and 180 plants m^-2, found that the optimal average density was 32 plants m^-2 and as the density increased there was a positive effect on yield.

The analysis of the numerical components of yield in grain crops has shown that the most important component in determining yield is the number of grains per m² (^{Cheng-dong et al., 2019}). As expected, changes in seed yield, when considering sowing systems, densities and cultivars, were explained by the number of seeds per m² (r²= 0.99, p< 0.001; Figure 3a), since no degree of association was observed for individual seed weight (Figure 3b). Between both cultivars, it was observed that the increase in sowing density positively affected the number of seeds per m², mainly in the double-row bed system, however, this did not statistically affect the individual seed weight (Figure 3b).

Figure 3 Relationships between seed yield and number of seeds per m² (a) and with individual seed weight (b) for two canola cultivars grown under three population densities and in two sowing systems in Toluca, Mexico. The solid line in the graph represents the regression model fitted for both systems. 

It should be noted that the number of seeds per m² had a moderate and positive association with the number of siliques per m² (r²= 0.41 p< 0.05) (Figure 4a) when considering the joint effect of sowing systems, densities and cultivars, while for the same number of seeds per silique, changes in the number of seeds per m² were observed (Figure 4b). These results highlight the importance of the number of siliques as a characteristic strongly associated with seed yield through the number of seeds (^{Shahin and Valiollah, 2009}). The number of siliques per plant depends on the number of branches, number of flower clusters and how many of these siliques set (^{Shahin and Valiollah, 2009}; ^{Assefa et al., 2018}) and is a character that compensates for losses in yield due to low density in potentially good environments, because a greater number of siliques per productive branch is achieved when the sowing density is reduced (^{Assefa et al., 2018}).

Figure 4 Relationships between the number of seeds per m² with the number of siliques per m² (a) and with the number of seeds per silique (b) for two canola cultivars grown under three population densities and in two sowing systems in Toluca, Mexico. The solid line in the graph represents the regression model fitted for both systems. 

The number of productive branches in canola is a characteristic that is largely associated with seed yield, and it also responds significantly to changes in the environment and management practices (^{Shahin and Valiollah, 2009}; ^{Hua et al., 2014}). In this sense, reductions in the number of branches are observed when the sowing density is increased (^{Uzun et al., 2012}; ^{Vincze, 2017}) or when there are delays in the sowing date (Hua et al., 2014; Vincze, 2017). Our data reveal that the number of siliques per m² was associated with the changes observed in the number of productive branches, mainly in the flat bed system (r²= 0.83, p< 0.01) (Figure 5).

Figure 5 Relationship between the number of siliques per m² and the number of branches per m² for two canola cultivars grown under three population densities and in two sowing systems in Toluca, Mexico. Solid and dashed lines in the graph are the regression models fitted for DRBS (SCDH) and FBS (SCP), respectively. 

Seed yield was positively associated with biomass production at physiological maturity, with the number of seeds per m² and with the number of siliques per m² (Table 4), which is consistent with numerous studies in the literature that indicate that these attributes are defining in canola seed yield (^{Shahin and Valiollah, 2009}; ^{Uzun et al., 2012}; ^{Mamun et al., 2014}; ^{Yang et al., 2014}). On the other hand, the biomass at physiological maturity was positively correlated with the number of seeds, with the weight of 1 000 seeds and with the number of siliques per m², while a negative association was observed with the harvest index.

As expected, the number of seeds per m² was positively correlated with the number of siliques per m², which indicates the great influence of the number of siliques in the determination of the number of seeds and the yield in canola, coinciding with previous evidence in the literature (^{Shahin and Valiollah, 2009}; ^{Kazemeini et al., 2010}). On the other hand, there was a negative association between the number of siliques per m² and the number of branches. This is a very important aspect to consider when varying density, since the number of productive branches tends to be affected both by density and by delays in the sowing date (^{Kazemeini et al., 2010}; ^{Vincze, 2017}; ^{Assefa et al., 2018}). The weight of 1 000 seeds showed a positive association with the number of branches.