**4. Discussion**

In this work, we studied the agronomic performances of an organic conservative managemen<sup>t</sup> of a two-year field vegetable crop rotation compared to a standard organic and an integrated managemen<sup>t</sup> system in Mediterranean conditions.

Our study confirms previous evidences (e.g., References [18,27,28,39]) that organic no-till systems are promising strategies to improve the sustainability of organic field vegetable systems but still need strong development and further investigations. In our experiment, we designed the organic conservative system with the main aims to reduce GHG emissions, to produce nonrenewable energy saving, and to emphasize use of internal natural resources in compliance with the principles of organic

no-till [6,20] and agroecology [27,39]. This resulted in very basic application of organic fertilizers in the ORG+ plots, where most of the regulating services (i.e., nutrient availability and weed suppression) were supposed to be provided by the cover crops grown as living or dead mulch. Given the not so high and stable biomass production of the dead mulch and the variable growth of the living mulch grown in the ORG+ plots (Table 1), the provision of agroecological services was not expressed enough to enhance system performances. Further research e fforts are still needed to identify the cover crop species and managemen<sup>t</sup> options most adapted to no-till conditions in order to enhance the level and the stability of ecosystem service delivery by cover crops. Selection of cover crop species should be done taking into account traits related to rooting capacity, high nutrient uptake and mobilization, N2 symbiotic fixation, low water consumption, quick soil cover, and creeping habitus (especially for living mulch use) [20]. The level of biomass production of the cover crops is indeed the crucial factor behind the functioning of no-till, cover crop-based cropping systems, as also pointed out by Reference [28].

For savoy cabbage, our results clearly showed how lack of nitrogen was likely the most important limiting factor for crop yield and N accumulation in ORG+. In the best years (2014 and 2015), the cabbage managed under ORG+ conditions yielded 50% lower than INT, whereas it was almost unable to complete the reproductive phase in 2016 (Table 2). Nevertheless, the results achieved by the ORG and INT systems were overall far below the standard for Central Italy [40] but in line with other similar experiments [41]. It is noteworthy that the harvest index was normally higher in ORG+ plots than in INT (Table 2). This result, combined with the low dry matter production of corymbs, highlights that also the vegetative growth was not well completed by the crop in the conservative system, likely due to a lack of readily available nitrogen from the initial stages after transplanting. To prove this hypothesis, further investigations are needed, looking at early indicators of crop nutrient status at the vegetative stage, such as the NDVI (Normalized Di fference Vegetation Index) or the LAI (Leaf Area Index), as suggested by Reference [42].

Besides the concentration of total N in the root zone, the mineralization rate of soil organic matter and of organic fertilizers applied uniquely at the transplanting stage also might have di ffered between ORG+ and tilled systems. In the ORG+ system, the N fertilizers were broadcast spread over the soil just before transplanting the cabbage and this might have likely caused poor contact with the soil and consequently a slowdown in the mineralization rate of the fertilizers. The N use e fficiency indicators studied (Table 11), in particular, the NReacc (showing a 33% of reduced recovery of N applied as fertilizers in ORG+), clearly support this hypothesis. In a recent paper published on GHG emission in the same experiment [26], a lower N2O emission from ORG+ plots than in INT and ORG was demonstrated. This might have been due to the low supply of fertilizers in the organic conservative system. Nevertheless, in the same paper, peaks of N2O emission after application of organic fertilizers have been reported, possibly due to no incorporation of the fertilizers. Our findings thus support the option to increase N fertilization rates at least in the transition phase to no-till in order to better support plant growth, given the uncertainty of the mineralization rate of the organic fertilizers in untilled soils. Another important option to be tested is to place fertilizers directly into the crop furrow [43] on the transplanting date in order to enhance contact between nutrients and roots and, more importantly, to stimulate the mineralization of the organic material and fertilizer N uptake while preventing high N2O emissions. Given the organic nature of the fertilizers and their low N concentration, a significant caustic e ffect of N on crop roots should not occur with in-furrow applications.

Among other potential stressors, excluding any e ffect of noxious organisms (e.g., pests and diseases), we can argue that also soil compaction caused by no-till in first 0–10 cm soil layer might have played a role. This is well known in literature on no-till, especially in the transitional stage from inversion tillage to conservative managemen<sup>t</sup> as in our case [6,44]. The use of the modified transplanting machine mounting shank openers [32] should have been reduced at least at the beginning problems of compaction for rootlets. At later stages, the persistence of shallow compaction might have caused a limited vertical root development that we were not able to assess.

Actually, neither weeds nor P seemed to have been the real limiting factors for cabbage. For weed biomass at harvest (Table 10), we did not find clear and strong di fferences between ORG+ and tilled systems. Likewise, for P concentration in plant tissues (Table 3), we did not observe significant depletion in ORG+ plots, but rather often an increase. This trend was also evident for fennel (Table 5), for which we did not detect any significant di fferences between ORG+ and INT. This is an interesting issue that is worth further investigating in the future. We hypothesize that, besides a concentration effect due to poorer crop biomass produced, there might have been other reasons for this increase in P content under no-till plus living mulch, first of all symbiosis with arbuscular mycorrhizal fungi (AMF) and soil acidification mediated by the living mulch root exudates [45].

Nevertheless, besides P availability, the contribution of the studied living mulch of red clover was not appreciable in terms of crop advantages. Other authors [46] highlighted how the managemen<sup>t</sup> of living mulch plays a key role in determining its ability to grow without competing for resources with the crop and to deliver weed suppression and nutrient mobilization. In particular, it was shown how a sowing date of the living mulch earlier than the transplanting date of the vegetable crop can reduce the performance of the living mulch compared to sowing contemporary to transplanting [46]. In our case, the red clover did not always establish well after its direct seeding due to soil compaction and poor seed-soil contact. Then, the clover covered the soil pretty well at crop transplanting (summer lettuce and cabbage). Anyway, in many cases, we observed the clover outgrowing the cabbage and, more often, summer lettuce at early development stages of the crops, whereas competition with weeds at later stages was not satisfactory. This was also because the in-crop managemen<sup>t</sup> of the living mulch, i.e., inter-row flaming, was feasible and e ffective only until the crop did not cover the rows, but in many cases, the most aggressive weeds started to grow only later in these stages. This was especially the case of summer lettuce, for which we observed repeatedly problems of summer grass weed species (e.g., *Digitaria sanguinalis* (L.) Scop., *Setaria italica* subsp., and *viridis* (L.) Beauv.) escaping from control with flaming. Another weakness of this managemen<sup>t</sup> system was the control of the weeds within the row that was simply not feasible with the operating machines available. Developing machinery for e ffective weed and living mulch managemen<sup>t</sup> also at later stages and within the row is thus required to improve no-till systems based on living mulch. To enhance the living mulch establishment, as its direct seeding can be problematic, and to reduce soil compaction due to passes of heavy machinery like direct drillers, alternative strategies, e.g., testing permanent living mulch lasting for years before regeneration as suggested by References [6,47], should be tested. This could also help solve the problem of controlling weeds within the crop row as creeping, permanent living mulch can have enough time to cover the entire soil surface during their growth.

For fennel and, more importantly, on the two lettuce crops, we identified in weed managemen<sup>t</sup> the most important factor together with nutrient managemen<sup>t</sup> a ffecting crop yield in the organic conservative system. As pointed out by Reference [48], weed control represents an important yield determinant in organic fennel due to scarce competitiveness of the crop. For lettuce, weed competition is a major issue given its short cycle, small plant height, low soil cover capacity, and shallow root [49].

Weed biomass at harvest in spring lettuce was always higher than the other systems except for in 2016 (not di fferent from ORG) (Table 10). In summer lettuce, ORG+ had more weeds than INT in all years, but not significantly in 2015. ORG+ did not di ffer from ORG in all three years (Table 10). The high weed presence and the low yield observed in summer lettuce might be partially explained by the performances of the living mulch of red clover. The lowest yield observed in 2015 (4.41 Mg f.m. ha−1, a value not statistically di fferent from 0) can be linked, in our opinion, to the overgrowth of the living mulch, which peaked 2.27 Mg d.m. ha−<sup>1</sup> (Table 1) and was too competitive with the crop from early stages. In 2016, when the summer lettuce reached the highest yield (Table 8), the biomass production of the living mulch was almost null. Nevertheless, the weeds were also not so aggressive (Table 10), maybe because weather conditions more favorable to the crop. In 2017, when the weed biomass reached a peak over 2 Mg d.m. ha−<sup>1</sup> (Table 10), the crop yield was not much depleted, maybe

because of favorable weather conditions and also a slightly higher N availability from the living mulch, the biomass of which was a bit higher than in 2016 (Table 1).

For spring lettuce, the relationships between crop yield and weed biomass were more evident than in summer lettuce, as also shown by the huge yield depletion in the ORG system in 2017 in the presence of the highest weed abundance (Tables 6 and 10). The absence of a new cover crop grown immediately before the spring lettuce and the poor regeneration of the red clover after cabbage harvest might have led to insu fficient weed control.

In our experiment, the weed biomass at harvest of fennel was higher in ORG+ than ORG and INT only in 2014 (Table 10), leading to a marketable yield much lower than achievable [40,41] (Table 4). The yield was still low in 2016, although weed biomass was half that of 2014 (Table 10). Nevertheless, in 2015, the marketable yield of fennel in the conservative system reached a peak and did not di ffer from the standard organic system (ORG) (Table 4). If we look at the composition and biomass produced by the summer cover crop mixture grown before fennel (Table 1), we can easily argue that the performances of the fennel were very related to the growth of the cover crops grown before its transplant and terminated as dead mulch. In 2014, the low yield of fennel and the high weed biomass at harvest can be explained by the low biomass produced by the dead mulch (only 1.31 Mg d.m. ha−1). For dead mulch, the importance of achieving high amounts of biomass production of cover crops to produce good soil cover, weed suppression, and nutrient release has been reported by many authors [6,50] and, in a recent paper, produced in similar conditions [28]. The summer cover crop mixture adopted in this study performed quite well but with the contribution of 2–3 species, whilst one of them (i.e., buckwheat) was very scarce in the canopy (Table 1). This finding emphasizes the need to investigate further in species/varieties of cover crops adapted to use as dead mulches in no-till systems. Besides quantity, also quality of cover crop biomass can be a key factor in terms of service delivery. In 2015 and 2016, when the cover crop biomass production was satisfactory, we can identify two di fferent compositions of the mixture, with grass species (i.e., foxtail millet and grain millet) dominating in 2015 and a more balanced composition in 2016 (Table 1). This di fference might have led to di fferent killing rates and di fferent kinetics of decomposition [31]. The termination technique (i.e., roller crimping plus flaming [28,31]) was very e ffective in both years in terms of killing rate, which was proximate to 95%. In 2015, the more abundant dead mulch obtained allowed the fennel to complete regularly the first stages after transplanting and to establish well, given the good soil moisture level conserved below the mulch and the thickness of the mulch, which prevented weeds from emerging at early stages. The not-so-quick mineralization rate of the biomass, characterized by a dominance of grass plants (i.e., high C:N ratio), avoided quick disappearance of the mulch that was as e ffective as at early stages in reducing weed emergence and growth, especially at later stages. Probably, this did not happen in 2016, when a higher presence of red cowpea biomass in the mixture might have led to a quicker decomposition of the dead mulch, freeing space for weeds to develop earlier than in 2015. In fennel, inter-row flaming was not as e ffective as in cabbage due to the presence of the dead mulch layer, which did not allow to enhance exposure time because of the burning risk. Thus, our findings encourage further research e fforts aimed at identification and testing in di fferent pedoclimatic conditions of high biomass producing cover crops with high long-lasting capacity. Di fferent termination techniques and machinery should be also tested in order to allow for distribution of the dead mulch along the crop furrow. Monitoring of mineralization rates of dead mulch provided by di fferent cover crop species managed di fferently is also recommended.

As shown in Table 5, in 2016, the concentration of N in swollen bases and residues of fennel in ORG+ reached the lowest values, revealing insu fficient crop N uptake and a nonrelevant contribution from the cover crops in terms of N supply. Red cowpea, the only legume in the mixture, which was supposed to deliver N to fennel through N2 biological fixation, did not produce nodules in the first year, maybe due to low presence of the required *Rhizobium* strain, whereas it showed regularly root nodules in the second and third year. This behavior might have produced di fferent interactions with the grass companion cover crops (foxtail millet and grain millet) in terms of N availability, leading to di fferent levels and quality of service provision to the fennel.

In terms of nutrient management, for fennel and cabbage, our findings highlighted how a unique application of N fertilizers at crop transplant could not be enough to sustain the crop during its growth. Fertigation with organic soluble fertilizers can be an option to achieve an increased nutrient availability for the vegetables, to simultaneously reduce water volumes for irrigation, and to reduce water availability for weeds by concentrating irrigation on the crop row [51]. Nevertheless, this option may interfere with mechanical/thermal weed control due to the presence of the irrigation hoses on topsoil. Subirrigation combined with no-till can be an alternative valuable option in that sense [51].

Overall, from an agroecological point of view, our results demonstrate that the total biomass production of the low-input organic conservative systems can be as high as in the standard organic and integrated systems if cover crop biomass is also considered. Cover crops thus were confirmed to be indispensable tools in conservative low-input systems. What clearly made a huge di fference was the proportion of marketable yield on total biomass, which was normally higher in the tilled systems because of faster mineralization of crop residues and organic fertilizers and lower weed abundance. Insisting on the fine-tuning of organic conservation systems is thus worth to be pursued in order to enhance the marketable productivity of these systems.
