3.3.3. Spring Lettuce

For the spring lettuce crop grown before fennel, the effect of the cropping system was not significant only for the P concentration in marketable yield. Besides P concentration in heads, the year did not significantly affect also the harvest index (HI) and the N accumulation in heads (Naccy) (Table S3).

The results of biomass production of spring lettuce at harvest time are shown in Table 6. For this crop, the performances of the ORG+ system were particularly negative. All the biomass components were significantly depleted by the ORG+ system, which was always lower than ORG and INT.

The concentration of N in heads and residues was highest in INT in all three years. INT did not differ from ORG only in 2017 whilst differed from ORG+ in 2015 and in 2017 (for crop residues) (Table 7). The amount of N accumulated in marketable product and residues was significantly higher in INT and ORG than ORG+ in all three years. Naccr and Nacct were not statistically lower in ORG than INT only in 2017. For P, the concentration in heads of lettuce was higher in ORG and ORG+ than INT in 2015, lower in ORG and ORG+ than INT in 2016, and not different among the systems in 2017. For crop residues, the P concentration was significantly higher in ORG+ than ORG and INT. The P accumulation in marketable yield was higher in INT than ORG and in ORG than ORG+ in 2015 and 2016. In 2017, there were no differences between INT and ORG, which both outperformed ORG+. For crop residues, INT was still the treatment with the highest P accumulation levels, being higher than ORG and ORG+ in 2015 and 2016 and higher than ORG+ alone in 2017. ORG and ORG+ did not differ from each other only in 2015. As a result, the total accumulation of P in the aboveground biomass of lettuce was higher in INT and ORG than ORG+ in 2014 and 2016 and was not different between INT and ORG only in 2016. P accumulation in total biomass in ORG+ was always lower than ORG.


**Table 6.** Least squares means and standard errors of marketable fresh yield (Y), dry matter of marketable yield (dwy), dry matter of residues (dwr), total aboveground dry matter (dwt), mean fresh weight of heads (MFW), Harvest Index (HI), and mean diameter of heads (MD) in spring lettuce. Confidence level: 95%.

Means followed by different letters are statistically different (95% confidence interval).

**Table 7.** Least squares means and standard errors of N concentration in marketable yield (Nconcy) and residues (Nconcr); N accumulation in marketable yield (Naccy), residues (Naccr), and total aboveground dry matter (Nacct); P concentration in marketable yield (Pconcy) and residues (Pconcr); P2O5 accumulation in marketable yield (P2O5accy), residues (P2O5accr), and total aboveground dry matter (P2O5acct) in spring lettuce. Confidence level: 95%.



**Table 7.** *Cont*.

Means followed by different letters are statistically different (95% confidence interval).

#### 3.3.4. Summer Lettuce

For the lettuce crop grown in the summer before savoy cabbage, the statistical analysis gave significant results for all the parameters, except HI (as affected by the cropping system), Pconcy, Pconcr, and P2O5acct (Table S4).

As for the marketable yield (expressed as fresh matter or dry matter), in 2014, INT and ORG were superior to ORG+ whereas, in 2016, there were no significant differences between ORG and ORG+ but only with INT and, in 2017, we did not find any difference among the treatments (Table 8). A similar trend was also identified for dry matter production of crop residues with the exception of 2016, when INT was higher than ORG+ only. For the total biomass of the crop, we found the same trend as for the dry matter of heads. The mean fresh weight of each lettuce head was found to be higher in INT and ORG than ORG+ in 2015, and higher in INT than ORG and ORG+ in 2016. No differences were found among treatments in 2017. The mean diameter of lettuce heads was higher in INT than ORG and higher in ORG than ORG+ in 2015 and 2016, whereas in 2017, ORG was equivalent to ORG+.

The concentration of N in the marketable yield was higher in INT than ORG and ORG+ in 2016 and 2017 (Table 9). In 2015, ORG was also higher than ORG+. The residues were richer in N in the INT plots, as well. INT was not different from ORG+ in 2015 and from ORG in 2017. The N accumulation in marketable yield was higher in INT than ORG+ in all three years. In 2015 and 2017, INT did not differ from ORG, only from ORG+. For N accumulated in crop residues, in 2015, INT and ORG were significantly higher than ORG+ whereas, in 2016, ORG+ was equivalent to ORG and, in 2017, there were no differences among the cropping systems. The total N accumulation in aboveground biomass of the lettuce was higher in INT and was significantly lower in ORG+. Nevertheless, in 2016 and 2017, the lettuce crop achieved N accumulation levels equivalent to ORG. For P concentration, summer lettuce showed normally higher levels for ORG and ORG+ with respect to INT. In 2016 and 2017, we did not find any significant differences among treatments. The residues showed a similar trend. Due to the lower biomass production, the N accumulation was very low in the ORG+ system anyway. Only in 2017 (marketable product and total biomass) and 2016 (residues), no differences were found among the three cropping systems tested.


**Table 8.** Least squares means and standard errors of marketable fresh yield (Y), dry matter of marketable yield (dwy), dry matter of residues (dwr), total aboveground dry matter (dwt), mean fresh weight of heads (MFW), Harvest Index (HI), and mean diameter of heads (MD) in summer lettuce. Confidence level: 95%.

Means followed by different letters are statistically different (95% confidence interval). \* Value statistically not different from zero.

**Table 9.** Least squares means and standard errors of N concentration in marketable yield (Nconcy) and residues (Nconcr); N accumulation in marketable yield (Naccy), residues (Naccr), and total aboveground dry matter (Nacct); P concentration in marketable yield (Pconcy) and residues (Pconcr); P2O5 accumulation in marketable yield (P2O5accy), residues (P2O5accr), and total aboveground dry matter (P2O5acct) in summer lettuce. Confidence level: 95%.



**Table 9.** *Cont*.

Means followed by different letters are statistically different (95% confidence interval). \* Value statistically not different from zero.

#### *3.4. Weed Biomass at Harvest Time of the Field Vegetables*

The effect of the cropping system on the dry matter produced by weeds at harvest time of savoy cabbage, fennel, and spring and summer lettuce is reported in Tables S1–S4, respectively. Only in the case of savoy cabbage, there were no significant differences among treatments. Neither were there differences due to the year. The interaction between the cropping system and year was not significant only in the case of summer lettuce.

In savoy cabbage, the organic conservative system (ORG+) did not perform worse than the other two systems in terms of weed suppression (Table 10). Only in 2016, we highlighted significantly higher weed biomass at harvest than in ORG and INT, although far under the 1 Mg ha−<sup>1</sup> of dry matter. The level of weed biomass was higher in fennel in INT and ORG+ plots, whilst on average, the ORG plots showed lower values than in savoy cabbage. In one year (2014), weed biomass in ORG reached a level statistically not different from 0, resulting in a weed biomass significantly lower than INT and far lower than ORG+. In 2015, weeds were significantly more abundant in INT plots whilst ORG and ORG+ were statistically not different from each other. In 2016, we did not find any difference among the treatments, but the level of weed biomass was ca. 50% less in ORG than INT and ORG+. In the lettuce crops (i.e., spring and summer lettuce), ORG+ showed everytime higher levels of weed biomass, with only two years (i.e., 2015 for spring lettuce and 2016 for summer lettuce) with values below 1 Mg ha−1. As expected, the INT system reached very low levels of weed infestation, accounting for 4 out of 6 cases for a level statistically not different from 0. The performance of ORG dramatically varied upon the lettuce crops, with significantly higher values than INT and equal to ORG+ registered in summer lettuce in 2015 and 2017. In spring lettuce, only in 2017, the ORG plots showed a mean value higher than 2 Mg ha−<sup>1</sup> that was significantly higher than INT.



Means followed by different letters are statistically different (95% confidence interval). \* Value statistically not different from zero

#### *3.5. Total Biomass Production and Nutrient Uptake at Crop Sequence Level*

The results of the statistical analysis of the performances of the cropping systems at the level of the entire crop sequence are reported in Tables S5 and S6. As shown in Table S5, the cropping system significantly affected all the tested variables other than the yield-related ones, except the dry biomass of the weeds. The inclusion of cover crops in the analysis of the performances of the cropping system at the crop sequence level significantly affected all the tested variables, whereas the position in the crop sequence (i.e., the field) was shown to be significant only for the dry matter produced by the weeds. For yield-related variables (Table S6), the cropping systems significantly affected all the parameters whilst the position in the sequence (field) affected only the N accumulation in marketable yield.

For the total fresh marketable yield of all the crops grown in the entire crop rotation in the three years (Figure S1), overall, the INT system outperformed ORG by 12.5% and ORG+ by 161% whereas ORG was superior to ORG+ by 132%.

In Figure 2, the interaction effects between cover crops and cropping system on total production of dry matter in marketable yield (dwy), residues (dwr), total aboveground biomass (dwt), and weeds (dww) is shown.

The total dry matter marketable yield production did not differ between INT and ORG, whereas it was lower in ORG+, whatever the level of cover crops. For total dry matter residue production, the highest value was shown by ORG CC+. ORG+ CC+ was not different from INT CC+, INT CC<sup>−</sup>, and ORG CC− but was higher than ORG+ CC<sup>−</sup>. As a result, total aboveground dry matter production of the crop sequence followed the same trend as dwr. The net gains in total crop dry matter production due to inclusion of cover crops in calculations were 8.05 and 5.17 Mg ha−1, respectively, for ORG and ORG+. Total weed dry matter was significantly lower in INT CC<sup>−</sup>, where the dry matter of weeds collected in the inter-crop period was not considered, than all the other treatments. The highest weed abundances were observed in INT CC+ and ORG+ CC+. Intermediate results were achieved by the remaining treatments.

In Figure 3, we reported the interaction effects between cover crops and cropping system on total N accumulation (kg N ha−1) in marketable yield (Naccy), residues (Naccr), and total aboveground biomass (Nacct).

**Figure 2.** Interaction between cover crops (without (CC−) vs. with (CC+)) and cropping system (INT vs. ORG vs. ORG+) on dry matter production (Mg ha−1) of marketable yield (dwy), residues (dwr), total aboveground biomass (dwt), and weeds (dww) at the level of entire crop sequence: Within the same dependent variable, bars with different letters are significantly different (confidence level 0.95).

**Figure 3.** Interaction between cover crops (without (CC−) vs. with (CC+)) and cropping system (INT vs. ORG vs. ORG+) on total N accumulation (kg N ha−1) in marketable yield (Naccy), residues (Naccr), and total aboveground biomass (Nacct) at the level of entire crop sequence: Within the same dependent variable, bars with different letters are significantly different (confidence level 0.95).

Total N accumulation in marketable product was significantly higher in INT than ORG, irrespective of cover crops level. Averaged over cover crops, ORG+ accumulated less than 100 kg N ha−1, resultingly significantly lower than INT and ORG. For crop residues, we observed a different trend, with the highest N accumulation observed for ORG CC+, followed by INT CC+ and INT CC− and different from ORG CC− and ORG+ CC+. ORG CC− showed the lowest value. Total N accumulation in aboveground crop biomass followed the same trend as Naccr. ORG CC+ was the only treatment that accumulated more than 500 kg N ha−1. Averaging cover crops levels, INT accounted for around 450 kg N ha−<sup>1</sup> whereas the best performing ORG+ treatment (i.e., ORG+ CC+) accounted only for 236 kg N ha−1. The net gain in N accumulation due to inclusion of cover crops in calculations accounted for 166 kg N ha−<sup>1</sup> for ORG and 78 kg N ha−<sup>1</sup> for ORG+.

#### *3.6. Nitrogen Use E*ffi*ciency*

In Table 11, the results of the analysis of N use efficiency of the single crops averaged over the three experimental years and of the entire crop sequence, considering or not considering the contribution of cover crops, are shown.


**Table 11.** N use efficiency indicators averaged over the three years for savoy cabbage, fennel, spring lettuce, summer lettuce, and the entire crop sequence with and without the contribution of cover crops.

\* N surplus calculated on N accumulation in total crop biomass (Nsurplusti) and N accumulation in marketable yield (Nsurplusyi); N utilization efficiency (NUtEi); N Recovery Efficiency of total N inputs (NREaci) and of fertilizers only (NREacfi); and Partial Factor Productivity of total N inputs (PFPi) and of fertilizers only (PFPfi).

For savoy cabbage, the N budget (i.e., the difference between all the N inputs and N accumulation in total biomass) was positive only for ORG and ORG+. In particular, ORG+ resulted in the lowest value, with about 22 kg N ha−<sup>1</sup> of surplus. Overall, for cabbage, the three systems did not overconsume or exploit N. Nevertheless, the important contribution of N from sources other than fertilizers was clearly shown by the negative values of N surplus calculated in terms of total N accumulation for all three systems (Nsurplusti). The fertilizers covered actually the N accumulation of corymbs in ORG and ORG+ whilst gave a surplus of around 50 kg N ha−<sup>1</sup> in INT (Nsurplusyi). Apparently, the efficiency in converting into marketable yield the unit of N accumulated in the biomass was not different among the systems (NUtE) and accounted for around 0.2 Mg f.m. kg−<sup>1</sup> total N accumulation. The recovery of total N inputs was close to 1 (i.e., the level at which N accumulated in total biomass was equal to the N inputs) for INT and ORG, whilst ORG+ accumulated only 67% of total N inputs. If considering only N from fertilizers, the three systems clearly all showed they accumulated also N from other sources, as they all showed values far higher than 1. The efficiency in converting the unit of N supplied in marketable yield (PFP) was higher in the ORG system, either considering the totality of N inputs or only the fertilizers. Interestingly, ORG+ outperformed INT when considering only N from fertilizers as an input.

For fennel, the N budget was sensibly more positive than for savoy cabbage. The ORG+ revealed an N surplus close to zero when considering the total N accumulation (only 5.23 kg N ha−1). The NUtE was slightly higher in the ORG+ and converted more e fficiently the N accumulated into swollen bases (+0.1 Mg f.m. kg−<sup>1</sup> N). The PFP was lower than in savoy cabbage and reached the maximum in ORG+.

For spring lettuce, ORG+ was the only treatment showing a slightly positive N budget (5.61 kg N ha−1), but when considering as N inputs, only the N from fertilization of all the treatments gave negative values, meaning N outputs were higher than inputs due to low values of N from fertilizers. The NUtE results did not show any di fference among the systems and averaged around 0.4 Mg f.m. kg−<sup>1</sup> N. The lettuce in ORG+ plots did not uptake 22% of the N supplied as total inputs. NReacfi and PFPfi were not calculated for ORG+ as N fertilizers were not applied. More than double the N accumulated in crop biomass in ORG and INT came from sources other than fertilizers (NReacfi). ORG+ was the less e fficient system in terms of conversion of N supplied into marketable yield.

In summer lettuce, the INT system resulted in a positive N budget (+14.53 kg N ha−1) and surplus (4.22 and 14.33 kg N ha−1, respectively, for Nsurplusti and Nsurplusfi) whereas ORG and ORG+ always gave negative values due to nonuse of fertilizers. The NUTe was not di fferent among the systems and reached the highest values in the crop rotation (around 0.5 Mg f.m. kg−<sup>1</sup> N). ORG (2.79) and ORG+ (1.19) showed the highest e fficiency in recovery of N supplied as total inputs, whilst INT did not reach the tie value of 1 even when considering only N from fertilization. The productivity of N units (PFP) was higher in ORG than ORG+ and then INT.

Considering the entire crop sequence, it is clear how all the systems produced high N surplus expressed as N budget that peaked 540 kg N ha−<sup>1</sup> in INT, 356 kg N ha−<sup>1</sup> in ORG, and 313 kg N ha−<sup>1</sup> in ORG+. If considering also N fixation of legume cover crops, the N budget of ORG and ORG+ reached, respectively, 457 and 365 kg N ha−1. Interestingly, the two organic systems di ffered from INT in terms of N surplus that was close to 0 but still positive for INT and very negative for ORG and ORG+, especially when considering also N accumulated by cover crops, as we did not distinguish between N accumulation derived from N fixation. This means the two organic systems strongly relied on N sources other than fertilizers. When considering only N accumulation in marketable yield, the N surplus was close to 0 for ORG+, positive for ORG (around 43 kg N ha−1), and still high for INT (227 kg N ha−1). Averaged over crops, NUtE was around 0.30 Mg f.m. kg−<sup>1</sup> N for all the systems when not considering cover crops, whereas it became 0.1 Mg f.m. kg−<sup>1</sup> N lower in ORG and ORG+ when including N from cover crops in calculations. The N recovery was far lower from 1 in all the systems when considering total N inputs, with ORG showing the highest value (0.65 Mg f.m. kg−<sup>1</sup> N). Including N accumulated by cover crops increased the e fficiency of ORG and ORG+, with ORG reaching 0.81 Mg f.m. kg−<sup>1</sup> N. If accounting only N from fertilization, the results clearly showed how INT was able to accumulate 99% of fertilizer N whilst ORG and ORG+ were underfertilized and relied upon additional N from other natural sources. Finally, the PFP of total N inputs was comparable among the systems and a bit lower in ORG+ than ORG and INT. Nevertheless, the PFP of fertilizers only clearly segregated among INT and the two organic systems. Due to the low N fertilization rates, ORG+ resulted in being the most productive system per unit of N supplied as fertilizers (+0.1 Mg f.m. kg−<sup>1</sup> N with respect to ORG and +0.31 Mg f.m. kg−<sup>1</sup> N with respect to INT).
