**1. Introduction**

The response to a growing demand for food under climate change and environmental risks connected to intensive agriculture requires more e fficient and sustainable agricultural practices. Ecological intensification has been proposed as a solution to these challenges, and organic farming principally relies on this, by promoting biodiversity and soil health [1–4]. Organic farming is increasingly growing in Europe, with almost 14.6 million hectares of agricultural land as of 2017 and a corresponding growth of more than 75% in a decade [5].

However, the current organic managemen<sup>t</sup> practices have been an object of debate recently. Organic farmers still rely heavily on conventional intensive tillage to incorporate crop residues, organic fertilizers, and cover crops and most importantly to control weeds [6]. The intensive tillage leaves the

soil exposed to wind and water erosion; destroys soil structure; and accelerates organic carbon loss due to oxidation, leaching, and translocation [7]. Reduced or no-tillage has been proposed in conventional agriculture to solve these problems. These practices are the key of "conservation agriculture" (CA), which not only contributes to sustain soil health and labor savings but also was shown to decrease greenhouse gas emissions through carbon sequestration and the reduction of energy use [8–10]. The implementation of reduced or no-tillage in organic agroecosystems may provide additional benefits to soil quality and may enhance resource use e fficiency compared to conventional tillage systems [11,12]. However, these systems are challenged with nutrient availability at key crop growth stages and weed pressure, which are di fficult to cope with in the absence of synthetic fertilizers, herbicides, and an adapted direct weed control [6]. These problems, often compromising crop yields, have limited so far the adoption of no-till by organic farmers, who in a recent survey showed interest to conservation practices for soil building purposes [13].

According to the Food and Agriculture Organization of the United Nations (FAO) definition [14], conservation agriculture systems imply also the diversification of cropping systems (e.g., by intercropping, cover cropping, agroforestry, and mixed farming) and permanent soil cover with crop residues or mulching material. The intensive use of cover crops has the potential to comply with these two principles. In organic farming, cover crops are normally used as green manures to provide nutrients to the cash crops and to increase soil fertility. Legume cover crops provide also by their N2-fixing ability additional soil N for cash crops upon incorporation and decomposition [15–17]. Nevertheless, cover crops are well-known to provide a wide range of ecological services, such as protection against soil erosion, reduction of leaching and increased availability of nutrients, improvement of soil and water quality, and weed and pest control [18]. Moreover, it has been suggested that cover cropping e ffect would be higher by decreasing tillage intensity through an expected higher ecological intensification, which could alleviate weed and crop nutrition problems related to reduced tillage or no-till [19].

Cover crop-based no-till is one of the forms in which conservation practices can be integrated in organic farming. In these systems, cover crops could be grown to remain at the surface either as dead or living mulch when cash crop is to be planted [20]. Cover crops in no-till can reduce weed infestation during their growth and/or by their residues making a physical barrier, preventing sunlight reaching the soil surface or through allelopathy [21]. However, residues left on the soil surface generally slows down the decomposition rate and nitrogen release compared to their incorporation [22]. It has been shown the importance of cover crop managemen<sup>t</sup> such as the selection, planting, and termination time and a complex rotation for the success of organic no-till, especially in vegetable systems which could su ffer more from competition, nutrient shortage, and weeds [20].

On the other hand, the e ffects of conservation agriculture practices on soil greenhouse gas (GHG) emissions are still uncertain, especially when adopted within organic agriculture. Cover crops may impact soil processes in ways that could potentially increase or decrease GHG emissions [23]. Tillage also may indirectly a ffect GHG fluxes by altering soil biological and physical parameters in a variable way, demonstrating di fferent responses across cropping systems [24]. Increased emissions in no-till farming were reported previously compared to conventionally tilled systems linked to soil types, climatic conditions, and the duration of conversion, although some studies showed lower emissions or no consistent e ffect [24–26].

Despite the environmental and economic promises that may hold, limited knowledge is available on organic cover crop-based no-till in Europe with far too little information concerning their performance in Mediterranean climate zone [18,26–28]. Thus, further investigation is needed as the success of organic reduced tillage systems may depend also on local pedoclimatic conditions. In this research, we aimed to study the transitional agronomic and environmental e ffects of the implementation of an organic conservation system (ORG+) within an organic vegetable rotation under Mediterranean conditions, with respect to an integrated managemen<sup>t</sup> system (INT) and a standard organic one (ORG). Our objectives were to evaluate their performance in terms of (a) crop production, (b) nutrient availability, (c) N budget and use e fficiency, and (d) weed infestation.

#### **2. Materials and Methods**

#### *2.1. Site Characteristics*

A three-year field experiment (2014–2017) was carried out on two adjacent fields (F1 and F2) at the Centre for Agri-environmental Research "Enrico Avanzi" of the University of Pisa, San Piero a Grado, Pisa, Italy (43◦40 N Lat; 10◦19 E Long; 1 m above mean sea level and 0% slope). The climate is typical Mediterranean with seasonal peaks of rainfall in spring and fall. Total average annual rainfall is 907 mm, and mean annual temperature is 15 ◦C. The soil was classified as Typic Xerofluvent, according to the USDA taxonomy [29]. The soil texture was loam sandy. Averaged over the two fields and two soil depths (i.e., 0–10 and 10–30 cm), soil organic matter content (Walkley–Black method) was 2.3 g 100 g<sup>−</sup><sup>1</sup> soil, total N (Kjeldahl method) was 1.25 mg g<sup>−</sup><sup>1</sup> soil, and available P (Olsen method) was 4.25 mg kg−<sup>1</sup> soil. More details on the main parameters of soil fertility measured at the beginning of the experiment are reported in Reference [26].

#### *2.2. Experimental Design and Crop Management*

The field experiment was based on the following crop sequence: Savoy cabbage (*Brassica oleracea* var. sabauda L. cv. Famosa F1, Bejo), spring lettuce (*Lactuca sativa* L. cv. Justine, Clause), fennel (*Foeniculum vulgare* Mill. Cv. Montebianco F1, Olter), and summer lettuce (*L. sativa* L. cv. Ballerina RZ, Rijk Zwaan). The experimental field was split in two fields in order to rotate the crops both in space and time. For this reason, in 2014, the rotation started with fennel in the first field (F1) and with cabbage in the second field (F2). The experiment layout was explained in a recent paper focusing on GHG emissions in two years (2014–2016) [26]. Conversely, this paper covers the entire duration of the crop rotation that was replicated for three years (2014–2017). In F1, the crop sequence was fennel (2014), summer lettuce (2015), savoy cabbage (2015), spring lettuce (2016), fennel (2016), and summer lettuce (2017). In F2, the crop sequence was savoy cabbage (2014), spring lettuce (2015), fennel (2015), summer lettuce (2016), savoy cabbage (2016), and spring lettuce (2017).

Three different cropping systems were assigned to the experimental plots according to a randomized complete block (RCB) design [30] with one factor (i.e., the cropping system) and with three levels (i.e., the treatments) and three blocks per field following the main soil gradient of each field. The elementary plots were of size 63 m<sup>2</sup> (21 m long × 3 m wide) and were separated by alleyways of 3 m width within the blocks and 5 m width between the blocks.

The three cropping systems compared were a control, represented by an integrated farming system (INT) based on conventional tillage practices (i.e., spading, rotary cultivation, and hoeing), mechanical and chemical weed control, chemical pesticide, and mineral fertilizer (ammonium nitrate, superphosphate, and potassium sulphate) use; a standard organic cropping system (ORG), built upon the same tillage practices as INT, mechanical weed control, fertilization based on commercial solid organic fertilizers (pelleted dried manure, blood meal, rock phosphate, and potassium sulphate) and on the use of cover crops incorporated as green manures, and crop protection by substances and biocontrol agents admitted according to the Reg. CE 2007/834 and Reg. CE 2008/889; and an organic conservation system (ORG+) including continuous no-tillage, use of cover crops managed as living or dead mulches, reduced organic fertilizer application (same products used in ORG), cultural and thermal (i.e., flaming) weed control, and crop protection strategy as described for ORG.

The three tested cropping systems differed not only in terms of tillage intensity and use of agrochemicals but also more generally on external resource use. The ORG+ system was mainly designed to exploit internal natural resources. That is why fertilization levels were kept at a minimum, aiming to support naturally occurring soil-fertility-building processes rather than directly supplying nutrients to the crops. The total amounts of N supplied as fertilizers for the entire crop rotation cycle were 303, 155.6, and 55.5 kg N ha−<sup>1</sup> respectively for INT, ORG, and ORG+. For P fertilizers, totals of 292, 192, and 87 kg P2O5 ha−<sup>1</sup> were supplied respectively to INT, ORG, and ORG+. K fertilizers were supplied at 603, 385, and 120 kg K2O ha−<sup>1</sup> respectively to INT, ORG, and ORG+. The intensive use of

cover crops in ORG+ was designed to replace several passes of mechanical weeding and herbicides. In the ORG system, a spring green manure mixture of field peas (*Pisum sativum* L.) and faba beans (*Vicia faba* var. *minor* Beck.) and a summer green manure mixture of red cowpea (*Vigna unguiculata* L. Walp), buckwheat (*Fagopyrum esculentum* L.), grain millet (*Panicum miliaceum* L.), and foxtail millet (*Setaria italica* L.) were chopped and incorporated into the soil before summer lettuce and fennel, respectively. In the ORG+ system, red clover (*Trifolium pratense* L.) was directly seeded and established as a living mulch for both summer lettuce and cabbage whilst the same summer cover crop mixture included in the ORG system was directly seeded on spring lettuce residues and terminated as dead mulch before the transplanting of fennel. In the ORG+ system, the termination of the dead mulch as well as the managemen<sup>t</sup> of crop residues and living mulch was implemented by rolling with roller crimper alternated with flaming. The roller crimper used was the Clemens Eco-Roll (Clemens Technologies, Wittlich, Germany), whilst the flaming machine was a prototype developed by Maito (MAITO Srl., Arezzo, Italy). The two machines and operating conditions are described in detail in References [28,31] and in the Table S7. Direct transplanting of the field vegetables into the untilled soil in the ORG+ system was successfully implemented by using the modified version of the FAST transplanting machine produced by Fedele Mario Costruzioni (Fedele Costruzioni Meccaniche, Lanciano, Chieti, Italy) and developed by the University of Pisa [32].

Sprinkler irrigation was applied to all the treatments at the same volume in the ten days after transplant and, afterwards, every 3 days until harvest only during summer (May to September).

Detailed information on the managemen<sup>t</sup> of each crop grown in each year/treatment in the two experimental fields is reported in Table S7.

#### *2.3. Sampling Protocol and Measurements*

Crop biomass production was assessed at harvest time (field vegetables) or before managemen<sup>t</sup> (cover crops) by sampling the aboveground biomass of each crop on 3 areas per elementary plot. Plant samples collected in each sampling area were processed separately and the data were then averaged to obtain one value per each parameter per elementary plot (i.e., the block). The size of the sampling areas varied according to the spatial arrangemen<sup>t</sup> of the crops. For the cover crops, the biomass produced by each cover crop species and the total biomass were assessed on 0.5 m<sup>2</sup> (1 m wide × 0.5 m long) areas. For large row vegetables (i.e., fennel and savoy cabbage) transplanted at 0.75 m between the rows, the sampling areas covered two crop rows and were of the size 1.5 m<sup>2</sup> (1.5 m wide × 1 m long), including 4 and 8 plants per sampling area, respectively, for savoy cabbage and fennel. Likewise, for the lettuce crops, the sampling areas covered 2 rows but the size of the areas was 1 m<sup>2</sup> as the crops were transplanted with an inter-row space of 0.5 m. In this case, the total number of plants sampled was 10 per sampling area. In the same areas, the total aboveground biomass of weeds was also collected.

In the lab, crop and weed biomass were manually separated and the fresh weight of their total biomass was measured. Crop biomass was subdivided in marketable product (i.e., corymbs for savoy cabbage, swollen bases for fennel, and heads for lettuces) and residues (i.e., discarded products, outer leaves, and rotten/diseased/damaged plant biomass) that were fresh weighted separately.

The mean fresh weight of each marketable product unit (i.e., corymbs, swollen bases, and heads) was determined by dividing the total fresh weight of the marketable product by the number of product units. The mean dimension of the marketable products was assessed by measuring the lengths of the two main orthogonal diameters of each corymb, swollen basis, and head that were finally averaged to obtain one value for the mean diameter (cm) of the product unit of each crop.

From each of the three total plant samples collected in each plot, a representative subsample of each component (marketable product and residues, for the crops, and total aboveground biomass, for the weeds) was fresh weighted and oven-dried at 60 ◦C until constant weight. The dry material was then weighted to obtain the dry weight and the percentage of moisture. The Harvest Index was calculated as the ratio between the dry matter of marketable yield and the dry matter of total aboveground crop biomass. On the dry samples of the two components of the biomass of each vegetable crop (i.e., marketable product and residues), total nitrogen (Kjeldahl method) and total phosphorus (colorimetric method) concentration were determined.
