**4. Discussion**

#### *4.1. In-Season Variability in the Growth Factor—Nitrate Nitrogen*

The yield variability of winter oilseed rape (WOSR) was evaluated based on the in-season variability in the N-NO3 content, i.e., the nitrogenous growth factor (NGF), directly impacting the rate of WOSR plant growth during the growing season. Yield responded to the applied N, irrespective of its chemical form, but significantly only to its lowest rate of 60 kg ha−1. This N rate resulted in a yield increase of 60% in 2016, a year with drought in June, and a 115% increase in 2017. The recorded yield increase fully corroborates the opinion on the high sensitivity of WOSR to drought during the end of inflorescence emergence and the onset of flowering [18,39]. The strong response of WOSR to the lowest N rate is typical for soil naturally poor in inorganic N, or soils strongly depleted of this N form [43]). The second problem, which became apparent during the study, refers to the lack of WOSR response to increasing N rates, applied as ammonium nitrate (AN) or digestate, which in the basic form is ammonium [44]. The application of 240 kg ha−<sup>1</sup> of N as AN resulted in a yield reduction. The observed tendencies are broadly explained by assuming an imbalance of applied nutrients or referring to the low nitrogen use e fficiency (NUE) of WOSR plants [10,45–49]. In the studied case, this observation was fully corroborated by the trend of partial factor productivity of N indices, which were calculated based on the N-NO3 content in the soil at BBBCH 30. The observed PFPN30 stabilization on treatments with the highest N rates was due to the lower rate of the NN content decrease during STME, subsequently leading to the better N utilization by WOSR during the seed filling period (SFP). The net N increase during the SFP, as has been documented recently by Łukowiak and Grzebisz [43], is the prerequisite of the high yield of WOSR.

One of the most important targets of soil or plant N monitoring during the growing season is to make a reliable yield prognosis [3,50]. In the study, the N-NO3 content in the soil profile during two of the cardinal stages of WOSR growth, i.e., rosette (BBCH 30) and the onset of flowering (BBCH 60), significantly limited seed yield, and therefore can be used as a yield predictor (Table A1). The relationship obtained was best described using the quadratic regression model:

$$\text{BRCH:}\\30; \text{ Y-N}\_{30} = -0.00012 \text{N}\_{30} \\ \text{ ${}^2$ } + 0.042 \text{N}\_{30} + 0.76 \text{ for } n = 15, \text{ R}^2 = 0.72, \text{ P} \le 0.05 \tag{9}$$

$$\text{BBCH 60: Y-N}\_{60} = -0.022 \text{N}\_{60} \\ \text{ ${}^2$ } + 0.21 \text{N}\_{60} - 1.56 \text{ for } n = 15, \text{ ${}^2$ } = 0.66, \text{ ${}^p$ } \le 0.05 \tag{10}$$

As results from equations developed, the N-NO3 content of 150 kg ha−1, as recorded at BBCH 30, resulted in a theoretical Ymax of 3.91 t ha−1. The calculated yield is very close to that achieved in the OM-NFS on the plot with the N rate of 180 kg ha−1. The yield prognosis at the onset of flowering was significantly lower when compared to BBCH 30, as indicated by the Ymax of 3.451 t ha−1. This yield could be achieved provided the N-NO3 amounted to 47.7 kg ha−1. Based on these two data sets, it can be concluded that a typical feature of N managemen<sup>t</sup> by WOSR during the stem elongation phase is a significant decline in the N-NO3 content. The average N-NO3 content decrease, based on the respective N optima, was 102.3 kg ha−<sup>1</sup> (= 150 kg ha−<sup>1</sup> at BBCH 30 minus 47.7 kg ha−<sup>1</sup> at BBCH 60). The study showed that the change in the N-NO3 content (ΔN-NO3) within the STME can be determined based on the N-NO3 content precisely at BBCH 30:

$$\text{AN-NO}\_3 = -0.85\text{N}\_{30} + 23.4 \text{ for } n = 15, R^2 = 0.98, P \le 0.01 \tag{11}$$

The direction coefficient of the linear regression model obtained clearly indicates a high efficiency of N-NO3 during STME, which reached 85% of the N-NO3 content in the whole analyzed soil profile. This value can be used as an index of nitrate N uptake efficiency by WOSR.

The analysis of the N-NO3 content at BBCH 30 and its rate of change during STME can be used as a basis for the revision of the current hypothesis about the critical stages of WOSR yield development. In fact, it is well-documented that two phases, i.e., inflorescence emergence and flowering, are crucial for the establishment of basic yield components. Any disturbance in plant growth during these two phases, either abiotic or nutritional, leads to yield reduction [18,39,51]. The observed dependence of the N-NO3 decrease during STME on its content in the soil profile at BBCH 30, as documented in this study, explicitly indicates the rosette stage as the cardinal for exploiting WOSR yielding potential [52]. In the well-established WOSR canopy, as in this study, the NN content underwent significant depletion, irrespective on its amount and N form (Figure 7).

**Figure 7.** Yield prediction based on the N-NO3 managemen<sup>t</sup> during stem elongation phase of WOSR growth.

#### *4.2. In-Season Variability in Soil Fertility Factor—Available Nutrients*

The yield of WOSR as discussed in the previous section responded significantly to the applied fertilizer N, irrespective of its chemical form. The key question remaining is to what the extent the soil fertility factor (FF), as defined by the in-season status of four basic nutrients (P, K, Mg, Ca), was related

to the content of N-NO3, defining the NGF status, and in consequence, seed yield. The applied PCA was revealed as a useful tool to evaluate the nitrogen fertilization system based on the above described factors. It was also a useful tool to define the production role of particular subsoil layers with respect to the availability of a given nutrient, or set of nutrients, as yield limiting factors. So far, knowledge about nutrient availability from deeper soil layers, especially during the growing season, is low [53].

The analysis of the NGF, i.e., variability in the N-NO3 content during the growing season, clearly indicates the rosette and the onset of flowering stages as cardinal for determining WOSR yield. Based on the PCA, it can be concluded that at the onset of STME, the primary set of soil nutrients, including the content of available Ca and Mg, representing the FF significantly limited yield, irrespective of the studied NFS. It is necessary to stress that the importance of both nutrients for NFS productivity resulted more strongly from their contents in the subsoil than in the topsoil. This conclusion stresses the importance of Ca as the yield nutritional factor. In fact, the yield forming function of Ca, in spite of the high requirements of some crops, such as oilseed rape, is weakly recognized [23,53]. In contrast to Ca, the yield forming functions of Mg are well recognized, but are mostly related to the onset of flowering and during the SFP [18,49].

The detailed analysis of the impact of FFs on yield for each of the studied NFSs showed a significantly di fferent response to the applied N form. The general pattern of the FF relation with Y, as presented above, was to a grea<sup>t</sup> extent typical for the M-NFS. The shortage of Ca and Mg in connection with a shortage of NN resulted in the insu fficient exploitation of both P and especially K from the second subsoil layer. The content of available K in this layer in the M-NFS can be used as a single yield (Y) predictor at BBCH 30:

$$\text{Y} = -0.68\\\text{K}\_{\text{c}} + 6.32 \text{ for } n = 10, R^2 = 0.85, \text{P} \le 0.01 \tag{12}$$

The negative sign of the direction coe fficient implicitly indicates that the low exploitation of K from the subsoil, just at the rosette stage of WOSR, is the reason for the yield decline. The OM-NFS yielded higher, because N was better balanced with basic nutrients, such as P and K, which were more strongly exploited by WOSR plants from the subsoil as compared to the M-NFS (Table S1b and Table 3d). It was found that a higher content of available K in the subsoil resulted in favorable conditions for the uptake of Ca and Mg. This hypothesis was fully corroborated at the onset of flowering, when available Ca positively a ffected N availability, finally resulting in a yield increase. The same e ffect in the OM-NFS on N availability was exerted by P and Mg. It is well-documented that all these three nutrients are strongly exploited by WOSR from subsoil [54–57].

A significantly di fferent pattern of FF impact on the content of NN and yield was observed in the O-NFS. In this particular NFS, yield significantly depended on the content of N-NO3, i.e., the NGF was found to be the direct yield driver (Table S1c; Figure S1c). The productivity of this system was also dependent on the content of Mg and Ca in the deep subsoil layer (c). The higher availability of both nutrients in this layer significantly a ffected the unit productivity of N (PFPN30). This was the key reason for the much higher productivity of the O-NFS system as compared to the M-NFS.

This study explicitly shows that a reasonable, but significant, decrease in the NN content during STME is the prerequisite of high WOSR yield (Figure 7). A similar decrease in the content of other nutrients can therefore be assumed. K requires special attention, because during this period it reaches both the maximum rate of uptake and the maximum value of its accumulation just at the end of the inflorescence emergence (INFE) [23,56]. In 2016, a net K content increase was recorded in all three soil layers. The same trend was observed in 2017 for treatments fertilized fully or partly with organic N. The observed phenomenon was not, however, related to the amount of K applied in digestate. Therefore, it cannot be explained by the direct impact of this fertilizer on K availability through K input, or indirectly by ammonium oxidation and its subsequent impact on the cation exchange [58]. The only reasonable explanation for the K content increase with soil depth is the WOSR activity in the rhizosphere, including its acidification [59].

At the onset of flowering, being at the same time the end phase of intensive uptake of N by WOSR, the key limiting nutrient for the whole system became the content of available Mg, which controlled the amount of NN, the direct yield driver. The result obtained corroborates the importance of Mg for stabilizing the nutritional status of WOSR plants at this particular phase [18]. The insu fficient availability of Mg on the one hand and excess of available P in the subsoil on the other points to FFs as critical factors for WOSR yield performance, irrespective of the NFS. Magnesium content and to a lesser extent Ca in the subsoil exerted a positive impact on the content of NN, a direct growth factor. Therefore, it can be concluded that the higher the content of available Mg in the subsoil, the better the supply of N to WOSR plants during the INFE, the phase responsible for seed set [49]. In the O-NFS, the impact of the key FF variables on the NFS functioning revealed a positive impact of Kc on the content of NN at the end of the INFE phase, concomitant with a significant improvement in N productivity, which was exerted by the Mg and Ca present in the second subsoil layer (Table S2c). The positive impact of Mg content on the NN content was fully corroborated in the OM-NFS. The higher content of both nutrients at the onset of flowering resulted in a positive yield response. The yield increase also resulted from the concomitant positive impact of available Ca on N productivity. In both years, the highest net increase in the content of available Ca during the STME was recorded on plots fertilized with the highest N rate, which resulted in the highest yield. These three simultaneously occurring processes led to a higher seed yield and the advantage of the OM-NFS over other tested NFSs.

The dominant impact of FFs as limiting production factors was fully corroborated at WOSR physiological maturity. The entire NFS production e fficiency was governed by the content of available Ca and P in the subsoil. It is well documented that in soil of a neutral pH range, the content of available P depends on the content of Ca and vice versa (Table S3a) [17]. The negative relationship between the content of available K in the second subsoil layer with yield clearly indicates the stability of the soil/plant system during the spring growing season:

$$\text{Y} = -0.0015 \text{K}\_c{}^2 + 0.15 \text{K} \text{-} 0.334 \text{ for } n = 10, R^2 = 0.55, \text{P} \le 0.05 \tag{13}$$

The presented equation shows that the depletion of the K content in the second subsoil layer at WOSR physiological maturity to 50.3 mg kg−<sup>1</sup> is the prerequisite of the maximum yield of 3.466 t ha−1. A quite di fferent set of variables were responsible for O-NFS. In the final phase of WOSR growth, the two decisive variables for yield were the amount of NN, K, and also P. The importance of this set of nutrients for exploiting WOSR yielding potential corroborates the latest study by Grzebisz et al. [49]. The key importance of K during the SFP results from its physiological function, as related to the transport of assimilates to the growing pods and seeds [60]. The better the supply of assimilates to the growing pods and seeds, the higher the number of seeds, subsequently resulting in a higher yield. The shortage of NN at this stage led to the weak exploitation of P from the whole soil profile. This observation is in agreemen<sup>t</sup> with Grzebisz et al. [61], who showed that the exploitation of P from its soil resources depends on the WOSR sink strength, which is related to the number of seeds per unit area.
