In the present study the minimum NDVI values for achieving the highest yields were 0.7–0.75 at GS32, GS37, and GS65. However, the NDVI threshold values at GS30 were not clear (0.4–0.65). In a previous paper [
14], good correlations between the NNI and NDVI were found; therefore, those correlations were used to compare the results of this study with the ones published by Ravier et al. [
10]. For achieving the NNI values proposed by Ravier et al. [
10], the NDVI values should be 0.5 at GS30, 0.75 at GS32, 0.75 at GS37, and 0.8 at GS65. Similarities and differences were found when comparing the results of the present study with the ones proposed by Ravier et al. [
10]. At GS30, the results obtained in the present study and the ones proposed by Ravier et al. [
10] diverged. However, at GS32, GS37, and GS65, the values obtained from this study were similar to the ones obtained by Ravier et al. [
10].
Focusing on the GS30 growing stage, in all treatments in 2016 and only for the conventional treatments in 2017 and 2015 were the NDVI values higher than the ones proposed by Ravier et al. [
10], 0.4 for the NNI and 0.5 for the NDVI according to the relationship established by Aranguren et al. [
14]. The same happened for all overfertilized treatments, where the NDVI values were 0.75, 0.70, and 0.6 in 2016, 2017, and 2015, respectively. Tillering (GS21 to GS29) has been reported to be important for reaching an adequate wheat yield [
6,
20], as a good tiller establishment allows a greater potential for biomass accumulation [
6]. At this early growing stage (before GS30), when the soil water content is high, the most limiting contributor to slow tillering could be the soil temperature and anaerobic conditions [
6]. The tillering period is related with wheat yield as it is involved in the determination of the spike number [
21]. Prolonging the tillering period results in a higher number of tillers formed [
6], which is probably what happened in 2016. The time elapsed from GS21 to GS30 was longer in 2016 (56 days) than in 2017 (35 days) and in 2015 (30 days) (
Table 1), promoting more efficient N incorporation (higher N uptake), leading to a higher photosynthetically active biomass. The longer the time elapsed from GS21 to GS30, the higher the achieved NDVI values (0.65 in 2016, 0.45–0.6 in 2017, and 0.38–0.55 in 2015). Otherwise, in 2015 and 2017, it rained much more than in 2016, with a difference of 353 mm and 103 mm, respectively, from the same period in 2016 (
Table 2). These heavy rains caused soil waterlogging, visible puddles, and anaerobic conditions, explaining the lower NDVI values. Soil waterlogging in winter in this area is not uncommon in very wet years. In fact, the experimental area is located in the Quaternary aquifer of Vitoria-Gasteiz, where the existence of several historically built trenches that 0.5 m deep with the purpose of avoiding flooding in agricultural areas have been described [
22]. Besides, those environmental conditions have provoked less N availability from the organics applied as initial fertilizers, explaining the lower NDVI values in the treatments with organics in 2015 and 2017 (0.4–0.45) than the conventional treatments (
Table 3 and
Table 5). Otherwise, those values were lower than the threshold values proposed by Ravier et al. [
10], 0.4 for NNI and 0.5 for NDVI according to the relationship established by Aranguren et al. [
14]. In 2015, the yields were comparable between organic treatments and conventional treatments for each N rate applied at GS30. That fact suggests that a N deficiency at GS30 can be partially corrected if the plant is able to absorb enough N in the following growing stages, as in the year 2015 (0.4, 0.7, 0.7 (organics) and 0.55, 0.7, 0.7 (conventional) for GS30, GS32, and GS37, respectively). Similarly, Morris et al. [
23] reported that a N deficiency in early growing stages results in maximum or near maximum yields in winter wheat. Conversely, in 2017, the yields in organic treatments were significantly lower than the yields in conventional treatments for the same N rate applied at GS30, as the NDVI values in conventional treatments remained higher than in organics, at least until GS37 (
Table A4; discussed in the following paragraphs). Yields in 2016 were generally higher than in 2015 and 2017, suggesting that the period before GS30 always has an important effect on crop yield if it maintained in the following key growing stages (from GS32 to GS65). No specific value was found for the GS30 growing stage, as the data volume was small for all possible options presented in the following key wheat growing stages (GS32, GS37, and GS65). That high variability made the establishment of the GS30 value difficult, thus, new data obtained from more experiments might help in its adjustment. Ravier et al. [
10] also remarked that the N threshold at GS30 was not so clear in terms of the comparison with the other stages.
Focusing on stem extension (GS32 to GS37), the values detected in the present study for achieving maximum yields and the ones proposed by Ravier et al. [
10] were similar (a NDVI value around 0.75). Adequate N availability at stem extension promotes crop growth through the development of viable tillers, increasing the future sink capacity of the plant, being related to the spike numbers and the grain number per unit area [
9]. A N deficiency in this period leads to the greatest losses in grain number [
9]. It is relevant that in 2017, during this period, the NDVI values were very low (
Table 5) in most of the treatments (0.4–0.55) because of the lack of rain after N application at GS30 (
Table 1), justifying the lower yields achieved in 2017 in comparison with 2015 and 2016. In fact, the treatments with organics presented lower NDVI values (NDVI = 0.40–0.45) than the conventional treatments (NDVI = 0.52–0.55), explaining the significantly higher yields in conventional treatments, as mentioned in the previous section. The NDVI values were significantly higher in the 280N treatment (
Table A2, NDVI = 0.65–0.70) because of the high N rate applied at GS21 (80 kg N ha
−1), explaining the higher yields achieved (8020 kg ha
−1). Anyway, these values were lower than the threshold values proposed by Ravier et al. [
10] and the ones detected in the present study (NDVI around 0.75). Villegas et al. [
24] showed that drought stress at stem extension resulted in lower rates of biomass accumulation. Adequate water availability for crops after the application of N fertilizer can be beneficial for a good N status and for achieving high yields [
25], as in 2015 and 2016 (
Table 1), where after the mineral fertilizer application at GS30 the soil was sufficiently moist, making the N uptake possible. The rainfall quantity necessary for the absorption of a N application by the crops is at least 15 mm in the fortnight after the application [
26].
At the GS65 growing stage, the NDVI values achieved at the treatments with the highest yields were similar to the ones proposed by Ravier et al. [
10], with NDVI values around 0.75. In the 2017 growing season, this was the first growing stage where differences in the NDVI values among different N rates were appreciable (
Table 5). The previous dry period and the rainfall that occurred after GS37 made the uptake of the N applied at GS30 viable. The N absorption by the crop from GS37 to GS65 was evident when comparing the yields obtained at the lower and higher doses applied at GS30, although it has been reported by other authors that this is late for increasing yields [
27]. Similar to the present study, Asseng et al. [
28] reported that it is a viable method to improve yields via nitrogen application at GS59 (before flowering) if the soil is wet. Anyway, in 2017, there was a smaller increase in yield per kg of N applied at GS30 in comparison with the 2015 and 2016 growing seasons, which is appreciable in the lower yields achieved and in the narrow difference of up to 2000 kg ha
−1 in yield between the treatments with the maximum yields and the lowest yields.