**3. Results**

In our study, all quantitative traits had a normal distribution. The ANOVA indicated a statistically significant influence of years and the years' × cultivar interaction for all eleven traits (Table S1).

#### *3.1. Leaf Area and Leaf Greenness Indices*

The leaf area index (LAI1) of the oat-vetch mixture measured in the tillering phase of oats was significantly differentiated (Table 1). The LAI1 of the mixtures in the conventional farming system was significantly higher (by 60%) than in the organic one. Additionally, the LAI1 was affected by the weather conditions, being the highest in the optimal year 2014 (1.60 m<sup>2</sup> m<sup>−</sup>2), and the lowest in the year 2013 (0.90 m<sup>2</sup> m<sup>−</sup>2), most probably due to a very dry April (Figure 4).

Interactions also differentiated the LAI1. Particularly, the interaction of oat cultivars and years was important, i.e., a significantly larger LAI1 was found in the mixture with cv. Celer in 2012, cv. Grajcar in 2013, and in 2014 the LAI1 was similar for both mixtures.

The LAI2 of the oat and vetch mixtures, measured at oats' grain watery ripe (BBCH 71), was also significantly differentiated by the examined factors (Table 1). A higher LAI2 was again found in the conventional farming; however, the system's difference diminished to 5%. Additionally, on average, the LAI2 of the mixture with oats cv. Celer was 6% higher, compared to the one with cv. Grajcar. It is worth mentioning that the LAI2 of mixtures with cv. Celer was similar, regardless of the farming system, whereas the LAI1 and LAI2 of mixtures with cv. Grajcar were higher in the conventional system by 41 and 11% compared to the organic one. The highest LAI2 value was again in a regular year, 2014, and the lowest in a dry 2012 year.


**Table 1.** Leaf area index (m<sup>2</sup> m<sup>−</sup>2) of the oat-vetch mixture, measured at the oats tillering: LAI1 and the grain watery ripe phase; LAI2 for the farming system in 2012–2014.

1 SD: standard deviation; 2 ns: non-significant. Homogeneous groups were created for the main factors. Mean values marked with the same letters are not significantly different according to Tukey's test at a significance level *p* ≤ 0.05; Three-factors of experiment: (1) farming system variant—organic or conventional (letters A, B); (2) oat cultivars—Celer or Grajcar (letters a, b); (3) years—2012, 2013, and 2014 (letters x–z).

> The oats' leaf relative chlorophyll content (SPAD) was differentiated by the examined factors and their interactions (Table 2). In the oats tillering phase (o1), the oats leaf greenness index in the organic farming system was 6% higher than in the conventional farming. However, in the second term (o2), the difference between the farming systems diminished. Additionally, a significant difference was noted between the oats' cultivars. Each time, higher SPAD values were found for the oats cv. Celer as compared to cv. Grajcar.

**Table 2.** The leaf chlorophyll content (relative content of chlorophyll) of oats in the mixtures with vetch, SPADo1—measured at oats tillering and SPADo2—measured at oats grain watery ripe phase, depending on the farming system and the oat cultivar in 2012–2014.


For explanation, see Table 1.

An interesting pattern was found for the oats' SPAD concerning the years. In the oats' tillering phase, higher chlorophyll content was noted in a regular 2014 year; however, in the watery ripe phase, the oats' SPAD values were highest in the dry and warm 2012, i.e., by 18% compared to the 2014 year.

The chlorophyll content of the vetch was also significantly differentiated (Table 3). Contrary to oats, higher SPAD values for vetch were found in the conventional system, compared to the organic one, by 4% in v1 and v2 terms. A selection of oat cultivars to the mixture with vetch also differentiated the vetch's chlorophyll content; in the v1 term, it was higher in the mixture with cv. Celer in comparison to the v2 term in the mixture with cv. Grajcar.

**Table 3.** The leaf chlorophyll content (relative content of chlorophyll) of vetch in the mixtures with oats measured at oats tillering (SPADv1) and oats grain watery ripe phase (SPADv2), depending on the farming system and oats cultivar in 2012–2014.


For explanation, see Table 1.

An interesting pattern of vetch's chlorophyll content was noted concerning the years. In the v1 term, the SPAD of the vetch was similar for all the years. Contrarily, in the v2 term, the highest vetch SPAD values were noted in a regular year, 2014, and the lowest were noted in the dry 2012. This is the reverse of the oat's SPAD values in the same term (SPADo2) (Table 4).

**Table 4.** Seed yield (t ha−1) of oat-vetch mixtures depending on the farming system and oat cultivar in 2012–2014.


For explanation, see Table 1.

#### *3.2. Yield of Mixtures and Their Components*

The mixtures' yield was 24% higher in the conventional system than the organic one (Table 4). An interaction was found for oat cultivars and years, e.g., the yield of the mixture with oats cv. Celer was significantly higher in a dry 2012 and a regular 2014, compared to 2013.

A significantly higher, by 38%, share of vetch seeds in the seed yield of mixtures was found in the organic system compared to the conventional one (Table 5). Additionally, on average, a higher share of vetch seeds was found in the mixture with oats cv. Grajcar, compared to oats cv. Celer. The share of vetch seeds in the yield was lowest in the dry 2012 and highest in the year 2013.

**Table 5.** The share of vetch seeds (%) in the oat-vetch mixture yields depending on the farming system and oat cultivar in 2012–2014.


For explanation, see Table 1.

Oats produced more tillers per plant and more panicles per unit area in the conventional system (Table 6). Interestingly, even though oats cv. Grajcar produced more tillers in the mixture, as compared to the oats cv. Celer, Grajcar still had a lower number of panicles per area in comparison with Celer. The highest number of oats' tillers and panicles was noted for both cultivars and farming systems in the dry year 2012. Despite a similar number of oats' tillers in 2013 and 2014, there was a significant drop in the number of oat panicles per unit area in 2013, regardless of the farming system and oat cultivar.

Like the seed yield and the number of panicles per area, a significantly greater number of grains per oat panicle (by 31%) were present in the conventional system compared to the organic one (Table 7)—oats cv. Celer developed by 38% more grains per panicle in the mixtures, compared to the cv. Grajcar. It was found that the number of grains of cv. Celer was significantly higher in conventional farming, by 43%, compared to the organic one, whereas the number of grains of the cv. Grajcar was similar in both farming systems. The number of grains in the panicles was highest in the regular year 2014. In the other two years, the number of grains per panicle was similar.


**Table 6.** The average number of tillers per oat plant and number of oats panicles per m<sup>−</sup><sup>2</sup> in the oat-vetch mixtures, depending on the farming system and oat cultivar in 2012–2014.

> For explanation, see Table 1.

**Table 7.** The number of grains per oat panicle in the oat-vetch mixtures, depending on the farming system and oat cultivar in 2012–2014.


For explanation, see Table 1.

The number of vetch pods per m<sup>−</sup><sup>2</sup> and the number of vetch seeds per pod (Table 8) followed, to some extent, the pattern of the share of vetch seeds in the mixture's yield (Table 5). Compared to the conventional system, the number of vetch pods was 53% higher in the organic one. The highest number of vetch pods was found in 2013 in the mixture with cv. Grajcar. However, a significantly higher number of seeds per pod was noted in conventional farming over organic. The highest number of vetch seeds per pod was found in the mixture with cv. Grajcar in the regular year 2014. The weather also influenced the vetch pod and seed per pod production in a significant way. Interestingly, the highest number of pods per m<sup>−</sup><sup>2</sup> was found in the 2013 year, but the highest number of seeds per pod was found in the regular 2014 year (Table 8).


**Table 8.** The pod number per m<sup>2</sup> and seed number per pod of vetch grown in the oat-vetch mixtures, depending on the farming system and oat cultivar in 2012–2014.

> For explanation, see Table 1.

The thousand-grain mass (TGM) of oats was higher in the conventional system, whereas for vetch this was in the organic one (Table 9). Simultaneously, higher TGMs of both oats and vetch were noted in the mixtures with cv. Celer. The TGM of oat cv. Celer was similar, regardless of the farming system, but in the case of cv. Grajcar was by 7% higher in the conventional system than in the organic one. The TGM of vetch fitted well to this pattern, as it was similar in the mixture with Celer, but 13% lower in the mixture with cv. Grajcar in the conventional system, compared to the organic one. On average, the TGM ofbothoatsandvetchwaslowestinthedry2012andhighestintheregular2014year.

**Table 9.** The thousand-grain mass (TGM) of oats and vetch (g) in the oat-vetch mixture, depending on the farming system and oat cultivar in 2012–2014.


For explanation, see Table 1.

The Pearson correlation coefficient analyses revealed several statistically significant interdependencies between the observed traits (Table S2, Figure 5). LAI1 (leaf area index in the oats' tillering phase BBCH 29) was significantly positively correlated with: LAI2, leaf area index in the oats BBCH 71 phase; SPADo2, relative chlorophyll content in oat leaves in the oats BBCH 71 phase; yd, mixtures yield; no-p, number of oats panicles per m2; no-gr, number of oats grains per panicle, and no-sd, number of vetch seeds per pod. LAI2

was positively correlated with: SPADv2, relative chlorophyll content in vetch leaves in the oat BBCH 71 phase; sh-v, share of vetch in the mixture's yield; no-gr; TGWo, thousandgrain mass of oats; TGWv, thousand-grain mass of vetch; no-pod, number of vetch pods per m2; and no-sd. SPADo2 was positively correlated with: yd and no-p; and negatively correlated with: SPADv2, sh-v, and no-pod. SPADv2 was positively correlated with: sh-v, TGWo, TGWv, no-pod, and no-sd; and negatively with yd and no-p. The yd was positively correlated with no-p and negatively correlated with sh-v, TGWv, no-pod, and no-sd. The shv was negatively correlated with no-p (−0.691) and positively with TGWo, TGWv, no-pod, and no-sd. The no-p negatively correlated with no-pod and no-sd. TGWo was positively correlated with no-gr, TGWv, and no-sd. TGWv positively correlated with no-pod and no-sd, and additionally, no-sd correlated with no-pod. SPADo1 was positively correlated with: LAI2, sh-v, no-gr, TGWo, TGWv, and no-sd; and negatively with yd and no-p. SPADv1 correlated positively with SPADo2 and SPADv1; and negatively with no-pod (Figure 5, Table S2).

**Figure 5.** Heatmap for linear Pearson's correlation coefficients between observed traits; *rcr* = 0.2875.

The greatest diversity in all eleven traits, measured with Mahalanobis distances, was observed for the combination co-ce-12 (conventional variant-Celer-2012) and or-ce-13 (organic variant-Celer-2013) (Table S3). The Mahalanobis distance between them amounted to 74.44. The greatest similarity (distance: 11.73) was observed between co-ce-14 (conventional variant-Celer-2014) and co-gr-14 (conventional variant-Grajcar-2014).

The canonical analysis was performed to present the tested mixtures' overall performances, based on all of the tested traits, for all of the three factors of this experiment

(Figure 5). The first two canonical variates explained jointly 85.6% of the total variation between the treatments. The greatest, significant linear relationship with the first canonical variate was found for SPADv2, the share of vetch in the mixture's yield, TGWv, number of vetch pods, number of vetch seeds per pod (positive dependencies), and SPADo2, the yield of mixture and number of panicles per m<sup>2</sup> (negative dependencies). The second canonical variate was significantly positively correlated with LAI1, LAI2, and the number of vetch seeds per pod. The results point to the best performance of the mixtures in the conventional variant of the farming system and during the regular year 2014 (Figure 6). However, both mixtures performed well also in the organic system in 2014. The mixtures performed worst in both organic and conventional systems in 2012.

**Figure 6.** The distribution of all 12 combinations of farming systems, cultivars, and years of study in the two first canonical variates, based on all tested traits. In the diagram, the coordinates of a given combination of treatments are values of the first and second canonical variates. Co: conventional farming; or: organic farming; ce: Celer, gr: Grajcar; 2–4: years 2012–2014.
