**1. Introduction**

Worldwide, land under certified organic production reached 698 million hectares in 2017 [1]. Across the global organic land base, the production of organic soybean [*Glycine max* (L.) Merr.] is increasing, with 429,621 ha under production in 2017 [1]. With more than 39,996 ha of organic soybean grown in 2014, the United States is the third largest producer of organic soybean [1–3]. In recent years, the European market is also rapidly expanding, with 72,710 ha of organic soybean production in 2016 [1,4]. Within Europe, France leads organic soybean production with 24,615 ha.

Improved weed managemen<sup>t</sup> and increased crop productivity have emerged as two main levers to facilitate the expansion of organic soybean acreage and meet the production demand [5,6]. As the prohibition of most synthetic substances is included in global organic regulatory frameworks, alternative techniques have been developed to manage weeds, including mechanical cultivation, strategic crop rotation, and the use of cover crops [7–10]. For most organic farmers, soil tillage is necessary to manage weeds, prepare the seedbed, and incorporate organic inputs [11]. However, intensive soil disturbance may decrease soil quality (e.g., reducing organic matter, increasing soil erosion, etc.), thereby raising concerns on the sustainability of organic farming practices [12].

To maintain soil fertility, organic farmers are encouraged by the Food and Agriculture Organization of the United Nations (FAO) to reduce soil tillage, improve soil coverage and diversify crop rotation [13–15]. Among all the techniques developed to reduce tillage, organic cover crop-based rotational tillage systems (organic CCBRT) has emerged as a practice of grea<sup>t</sup> interest. These systems reduce tillage through the establishment of cash crops into high residue cover crops terminated with a roller-crimper [16–19]. The cover crop mulch remains on the soil surface until cash crop harvest, preventing weed emergence, and thus eliminating the need for mechanical weed management, maintaining soil quality while reducing labor and fuel consumption. In addition to creating a physical barrier which reduces weed emergence, an additional mechanism of weed suppression includes the competition of the cover crop with weeds for water, nutrients, and light [20,21]. Further, weed control may also be enhanced through allelopathic compounds released by the cover crop, which can inhibit weed germination [20,22–26].

Currently, reduced tillage practices implemented within conventional row crop systems are highly dependent on the use of chemical herbicides [17,27–29]. Growing concerns about the detrimental impacts of herbicides and the increasing occurrence of herbicide-resistant weeds have stimulated research interest for CCBRT in both organic and conventional production systems, especially in the United States where this technique has seen significant growth over the past decade [19,30,31]. The technique is less developed in Europe, but farmers' interest in preserving soil quality is increasing, as shown by a European survey conducted in 2012 on organic conservation practices [32].

Previous research has shown that e ffective weed control can be achieved through CCBRT until crop harvest if the cover crop biomass reaches from 8000 to 10,000 kg ha−<sup>1</sup> according to conditions (e.g., climate, weed infestation, weed species) before termination [16,33]. Cover crop species selection also serves as a fundamental tool to (1) optimize cover crop biomass, (2) inhibit weed germination through the release of allelopathic compounds and (3) ensure adequate termination of the cover crop with a roller-crimper [34–36]

Some cereal grain cover crops perform well in CCBRT systems with soybean cash crops, including cereal rye (*Secale cereale* L.), triticale (*x Triticosecale Wittmack*), barley (*Hordeum vulgare* L.), oat (*Avena sativa* L.)], and winter wheat (*Trticum vulgare* L.) [37,38]. Their main advantages over legume species are the high biomass production and consistent termination with a roller-crimper.

Among the cereal grain cover crops, cereal rye has consistently superior performance in the organic CCBRT system, producing high amounts of dry matter and reaching anthesis (Zadoks stage 69) [36,37], the stage of maturity necessary for mechanical termination, earlier than other cereals [19,21,39,40]. Cereal rye has also exhibited a high degree of allelopathy, inhibiting weed seed germination [41,42]. Incomplete mechanical termination of cereal rye in organic CCBRT may result in volunteer cereal rye plants in subsequent phases of the crop rotation, which results in contamination of following crops with rye grain, a ffecting both quality and yields of subsequent crops [18,40,43,44]. Thus, in recent years, triticale and barley, species with lesser propensity to produce volunteer plants, have been explored as alternative cover crops to rye. Additionally, the more prostrate growth habit and wider leaves characteristic of these species may provide greater light interception, improving early season weed control [39,45]. However, a dearth of references exists on the comparative performances of di fferent cereal species in organic CCBRT systems.

While previous studies have demonstrated the ability of cereal rye cover crops to suppress weeds, the success of the CCBRT technique remains highly variable across years and location [18,46]. Investigation of the performance of organic CCBRT systems over a broad range of pedoclimatic conditions with the comparison of some cereal grain cover crops is needed to understand the reasons for failures and achieve more consistent success. Alternative cereal grain species such as triticale could

provide similar results than cereal rye and provide benefits for facing soil and climate condition to reach consistent cover crop performance. The objective of this study was to examine the performance of two cover crop species used in combination with soybean in an organic CCBRT system under di fferent pedoclimatic conditions through a multi-site experiment over two years: (1) In the Upper-Midwestern USA and (2) Southeastern France. This study aimed (i) to determine which cover crop species leads to the highest soybean success rate and (ii) determine the drivers of variability in cover crop performance, weed control and soybean yields observed in di fferent pedoclimatic conditions.

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

#### *2.1. Site Description*

The trials were conducted on certified organic land at two locations in 2017 and four in 2018, located in the upper Midwestern U.S. and in Southern France. The US sites are characterized by short growing season with high seasonal rainfall, cold winter conditions, and warm summer temperatures, as compared to the European sites which were defined by a more temperate climate, with consistent cool conditions and lower precipitation.

Site A is the University of Wisconsin Arlington Agricultural Research Station (UW-AARS) in Arlington, WI, USA. The four other locations are in Southern France Rhône-Alpes region, with site B in Drôme, site C in Northwestern Isère, site D in Ain and site E in Northeastern Isère. Soil types and climates are presented in Table 1. At Arlington (site A), fields have been certified organic since 2009 and were under alfalfa cover crop from 2014 through 2016. The organic CCBRT system trial was initiated in 2017 and relies on the common four-year rotation practiced in the upper Midwestern U.S., including, corn, soybean, fallow, and small grain [19]. In Southern France (sites B, C, D, and E), annual trials were implemented in the typical crop rotation practiced by farmers under organic grain system which is based on similar crops rotation as encountered in the upper Midwestern U.S. (i.e., winter wheat, corn, soybean, alfalfa). At sites B and E, reduced tillage was practiced throughout the prior 10 year period, while the historical managemen<sup>t</sup> practices at C and D sites relied on traditional tillage. Sites B, C, and D have been certified organic for 13–27 years, while sites E has been managed organically for three years.


**Table 1.** Description of the six experimental sites (soil and climate conditions).

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

At each location, two cover crop species were compared (cereal rye and triticale) using a randomized complete block design with four replications. The detailed field operations are presented in Table 2. Site A (Arlington, WI, USA) was a 0.48 ha field with 67 m × 9 m sub-plots. The French sites (B, C, D and E) were 0.23 ha fields with 24 × 12 m sub-plots. Winter rye, ('*Aroostook*' (site A), '*Dukato*' (site C, D, E), *'Ovid'* (site B)) and winter triticale ('*NE426GT'* (site A), *'Vuka'* (site B, C, D, E)) were planted at the end of summer or early fall of 2016 and 2017 (Table 2). Di fferent 3 m wide drills were used depending on the location (site A-Model 750, John Deere, Moline, IL, site B and E-Sulky Master,

site C-Saphir 7/400-DS 125, Lemken). On site D, the drill wasa4m wide Vitasem 402 A, Pottinger. Planting depth was standardized at 2.5 cm.

Roller-crimpers of di fferent widths, weight and manufacturers were used to terminate the cover crops (site A 4.6 m, 1360 kg, I and J Manufacturing, Gap, PA, sites B, C and E 3 m, 1400 kg, University of Lyon 1, Rhône-Alpes region, France, site D-6 m, 3300 kg, FACA, Sky Agriculture). Soybeans were planted and cover crops were terminated when the latter reached 50% to 100% anthesis (Zadoks growth stage 65–69) both years, thus resulting in di fferent soybean planting dates depending on year.

Soybeans were planted with a 4.6 m wide conservation tillage planter in Wisconsin (site A) (Model 1750 Max Emerge Plus, Conservation Tillage, John Deere, Moline, IL),a6m wide no-till drill on site C and D (Easydrill W 6000, Sky Agriculture), a 3-m wide no-till drill on site E (Easydrill 3000 Fertisem, Sky Agriculture), anda4m wide planter on site B (Maxima 2 TI M, Kuhn) (see Table 2 for row spacing). Crimping and planting were performed the same day in two separate passes across the field, except for 18-Frce E site where both crimping and planting were performed as a one-pass operation.
