2.1.1. Processing Tomato

In ORG and ORG+, processing tomato was preceded by an autumn-sown cover crop of field pea (*Pisum arvense* L., cv Arkta) and barley (*Hordeum vulgare* L., cv Amyllis) in mixture (barley at 25% + pea at 75% of their ordinary full sowing rates, i.e., 100 and 75 seeds m<sup>−</sup>2, respectively) while, in INT, the soil was left bare and weed-free (by mechanical control). Cover crop termination was carried out traditionally in ORG: aboveground biomass of the mixture was mowed, finely chopped (0.02–0.1 m) and immediately incorporated into the soil (0.2 m depth) by a rotary cultivator equipped with tines and a back-roller. In ORG+, cover crop biomass was roll-crimped and left on the soil surface as dead mulch. Cover crop termination in ORG+ was generally postponed compared to ORG, because of the slower plant development in the conservative system (Table S1). The processing tomato was transplanted at 3.3 plants m<sup>−</sup><sup>2</sup> into single rows 1 m apart by a standard machinery in the INT and ORG and a no-till direct transplanter in ORG+.

All systems received a fertilization of 150 kg N ha−1, which was distributed by means of fertigation (details on rate, scheduling and methods in Farneselli et al. [17]) using a synthetic fertilizer (Radicon N30, Green Has Italia spa, Italy) in the INT system and an organic fertilizer (Ilsadrip Forte, Ilsa spa, Italy) in the ORG and ORG+ systems. In the case of ORG, N content in the legume component of the CC (pea) was measured prior to termination and the corresponding amount was subtracted from the aimed rate of 150 kg N ha−1. This difference was distributed to the cash crop. In the case of ORG+, only 50% of N accumulation in pea was subtracted from the aimed rate, in order to account for the lack of CC biomass incorporation into the soil [18].

Concerning the other macronutrients, 150 kg ha−<sup>1</sup> of P2O5 and K2O were broadcast at cover crop sowing (in ORG and ORG+) and at final seedbed preparation in INT.

## 2.1.2. Durum Wheat

Durum wheat was grown as the sole crop in the INT and ORG (single rows 0.15 m apart); in ORG+, durum wheat was temporary intercropped (TIC, Guiducci et al. [19]) with faba bean (*Vicia faba* L. var. *minor* Beck. cv Scuro di Torrelama) in alternate rows, 0.45 m apart with faba bean sown in the middle of the wheat inter-row space. Sowing density was 400 kernels m<sup>−</sup><sup>2</sup> for wheat (in all systems) and 90 seeds m<sup>−</sup><sup>2</sup> for faba bean (in ORG+).

Concerning wheat N fertilization, in the INT system, 160 kg N ha−<sup>1</sup> was applied as urea in two applications (half dose at tillering and half at shooting, following the regional recommendation for durum wheat N fertilization management). In ORG, 40 kg N ha−<sup>1</sup> was broadcast just before seedbed preparation as poultry manure (N = 4%). In ORG+, at the beginning of wheat shooting (Table S1), faba bean plants were incorporated into the top soil (0.10 m depth) by split rotary hoeing. Thus, in ORG+, durum wheat N fertilization came entirely from the incorporated faba bean plants (relying on an expected amount of approximately 50 kg N ha−<sup>1</sup> [19])

## *2.2. Plant Sampling*

Each year, the aboveground biomass accumulation of cash crops was determined before harvest by sampling plants from two subplots with 1.2 m<sup>2</sup> area per plot. The harvested aboveground biomass was separated in residues (straw, non-marketable fruits and vegetative parts) and yield (grains and marketable fruits). The cover crop and faba bean biomass was determined just before the termination (pea and barley in the mixture were kept separated). Weed biomass was also determined at each sampling operation. Plant samples were oven dried at 80 ◦C to determine dry matter content, then ground to a fine powder and stored. A reduced-N concentration of Kjeldhal digests, prepared following the method proposed by Isaac and Johnson [20], was measured by using an automatic analyzer (FlowSys, Systea, Italy).

#### *2.3. NO3-N Leaching, N Balance and N Use E*ffi*ciency*

Every year, two lysimeters consisting of porous, ceramic cups (32 mm external diameter by 95 mm length) were installed [21] in the core part of each plot at a depth of 0.9 m. The cups were installed just after sowing by drilling the soil vertically at a depth of approximately 1.0 m. The excavated topsoil and lower subsoil were kept separate. Before placing the porous cup, thick slurry prepared from the lower subsoil was poured into the hole. The repacked soil was then added and consolidated with care in order to avoid preferential water flow. The ceramic cups (SDEC, Tauxigny, France) were joined to a capillary tube, long enough to emerge from the soil surface and sealed at the end by an iron clamp. Samples of the soil solution at 0.9 m were taken using a portable vacuum pump, and then transferred to a storage pot. The NO3-N concentration in the soil solution was determined by an ion-specific electrode meter (Cardy, Spectrum Technologies, Inc., Plainfield, IL, USA), calibrated at the beginning of each measurement and set by using the standard solutions provided with the testing kits [22]. According to the method proposed by Gabriel et al. [23], NO3-N concentration data were recorded only when all soil lysimeters could provide drainage water, which occurred after rainfall events of adequate intensity.

A simplified model was adopted to estimate the drainage volumes (for further details see Tosti et al. [22]). As proposed by Gabriel et al. [23], the NO3-N leached over the time intervals between soil solution samplings was calculated as the product of mean NO3-N concentration in the soil solution and the daily drainage obtained for the sampling interval.

As reported by De Notaris et al. [16], N balance was calculated as the di fference between N input and output. The input included: N in manure or mineral fertilizer (i.e., extra-farm N), atmospheric N deposition [24], N derived from atmosphere via symbiotic fixation (Ndfa) and N in seeds. The output consisted of N removed from the field (i.e., leaching losses and N in yield). The N surplus was generated by the combination of the input and output values. For each year, inputs and outputs were determined, for each system, as averages across crops.

Ndfa was considered equal to 90% of the total N accumulation in the pea and faba bean above-ground biomass, according to the findings reported by Antichi [25] and Saia et al. [26] for similar climatic conditions. N use e fficiency at system scale was assessed by two indices: yield to N leaching loss ratio (Y/Nloss, kg kg−<sup>1</sup> N) and yield to extra-farm-N input (Y/Nextra, kg kg−<sup>1</sup> N).
