© S. Cadoux et al., published by EDP Sciences 2015

1 Introduction

Despite significant gains in productivity, intensive agriculture is criticized for its harmful effects on the environment (Tilman et al., 2002). In today’s environment of ongoing rapid growth in world population, agriculture must succeed in reconciling the production of food, feed and biomass, and the conservation of the environment (Robertson and Swinton, 2005). New agricultural systems are required, and one way is to base them on ecological intensification i.e. favoring services provided by biotic interactions in terms of soil fertility and pest control (Doré et al., 2011; Gaba et al., 2014). Plant species diversity in agroecosystems is often presented as a pillar of agroecological approach (Vandermeed et al., 1998; Ratnadas et al., 2012; Gaba et al., 2014). Mixing plant species can result in improving productivity, pest regulation, carbon and nutrient sequestration (Malézieux et al., 2009).

The use of living mulches, defined as “a cover crop sown either before or with a main crop and maintained as a living ground cover throughout the growing season” (Hartwing and Ammon, 2002), is a promising solution. Many authors demonstrated that the introduction of living mulch provides control of weeds or, at least, a low chemical weed control (Abdin et al., 1997; Agegnehu et al., 2008; Hiltbrunner et al., 2007; Ilnicki and Enache, 1992), to decrease insect damage (Costello, 1994; Hooks and Johnson, 2004; Theunissen, 1994), and to increase N availability (Thériault et al., 2009; Thorsted et al., 2006). However, the introduction of living mulch could induce competition between the cash crop and the living mulch (Teasdale, 1996). Most studies reported a yield reduction of the cash crop (Bergkvist, 2003; Carof et al., 2007; Picard et al., 2010; White and Schott, 1991), even if a few ones reported no yield loss (Ilnicki and Enache, 1992). Nevertheless, the yield reduction mentioned before occurred for systems where the living mulch was still growing during the main phases of yield components establishment particularly in the spring (Carof et al., 2007; Hiltbrunner et al., 2007).

In France, winter oilseed rape (WOR) is the second arable crop in terms of cultivated areas. It is also the second arable crop in terms of treatment frequency index (5.5 on average) and nitrogen applied per hectare (162 kg/ha on average) (AGRESTE, 2014). This leads to environmental impacts, notably pesticide losses in water (Sausse et al., 2012). Indeed, in the French context (soil, climate, cropping systems, etc.), rapeseed is particularly subject to weed infestation (Primot et al., 2006) and insect pressure (Valantin-Morison, 2012). The specificity of winter oilseed rape (e.g. taproots, breaking cereal diseases cycles, nitrogen catchment, etc.) makes it an important crop in French rotations dominated by straw cereals (AGRESTE, 2014) and justifies the design of innovative cultural practices (Valantin-Morison, 2012). Studies of brassica species based on mixed cropping are rare (Banik et al., 2000; Schröder and Köpke, 2012), especially with WOR (Bergkvist, 2003). Jamont et al. (2013) demonstrated the niche complementarity of winter oilseed rape intercropped with faba bean in the early growth but this study only focused on nitrogen use and was carried out under controlled conditions. The particular interest in mixing winter oilseed rape with frost-sensitive legume crops is due to different potential positive interactions: (i) the early sowing of WOR seems favorable to a significant growth of legume crops; (ii) the competitive effect of WOR to N acquisition in the soil should favor, the N fixation of legume crops, and thus its potential competition; (iii) the likely destruction of the legume crop during the winter should favor its N release, with no competition against WOR and no additional herbicide application. Therefore, the benefit for WOR could be better than “permanent” living mulch.

The aim of this study was to assess the effect of three mixtures of frost-sensitive legume crops – faba bean + lentil (BL); grass pea + fenugreek + lentil (GFL); purple vetche + common vetch + berseem clover (VVT) – intercropped with winter oilseed rape on the growth of the intercrop, the weed cover, the damage caused by rape winter stem weevil in autumn, and on the aerial nitrogen content in spring and the yield of winter oilseed rape.

2 Materials and methods

2.1 Site characteristics and experimental design

The experiments were performed at two sites (Murs and Villedieu-sur-Indre, 36, central France) in 2011 and at four sites (Murs and Villedieu-sur-Indre, 36, central France; Chambon, 17, western France; Morville-sur-Seille, 54, northeastern France) in 2012, 2013 and 2014. The main characteristics of each site are presented in Table 1.

Winter oilseed rape (Brassica napus) – subsequently referred to as “WOR” – was grown as sole crop and intercrop with three different legume mixtures: (i) faba bean (Vicia faba) + lentil (Lens culinaris) – subsequently referred to as “BL” –; (ii) grass pea (Lathyrus sativus) + fenugreek (Trigonella foenum-graecum) + lentil – subsequently referred to as “GFL” –; and (iii) purple vetche (Vicia benghalensis) + common vetch (Vicia sativa) + berseem clover (Trifolium alexandrinum) – subsequently referred to as ‘VVT’. In terms of sowing densities, the intercrop was based on the additive principle. The sowing density of the WOR varied according to the trial (Tab. 2). The sowing density of BL, GFL and VVT was 55 kg ha^-1, 35 kg ha^-1 and 20 kg ha^-1, respectively.

The experimental plots (120 to 200 m²) were set up in a randomized block design with WOR, WOR +BL, WOR +GFL, WOR +VVT as treatments with three replicates. Other intercrops were studied in each trial but varied between sites and years and were thus not included in that analysis.

2.2 Management practices

In each trial, sole WOR was managed according to local agriculture guidelines and conditions. Soil tillage was based either on plowing (P), reduced tillage (RT) or direct seeding (DS). The nitrogen (N) fertilization rate was calculated according to the balance sheet method, which determines the a priori optimal fertilization rate (subsequently referred to as “X” rate), with the CETIOM method “réglette azote” (Lagarde and Champolivier, 2006). The management of intercrops was the same as for sole WOR except in terms of broad leaf herbicide applications – reduced when the doses to be applied were considered potentially phytotoxic for legumes- and nitrogen fertilization rates – systematic reduction of 30 kg ha^-1 (subsequently referred to as “X-30” rate) –, assuming a contribution of legumes to WOR N nutrition and seeking to offset the extra cost of legume seeds. The main characteristics of the management practices are presented in Table 2.

2.3 Measurements and calculations

Four types of measurements were carried out: aerial dry weights and nitrogen contents at two dates (November and G4 stage); weed competition, estimated by the percentage of ground cover by weeds in Autumn; insect damage; and yield. All these measurements are detailed thereafter. They were not carried out on all the sites described before; the distribution of the measurements is detailed in Table 3.

The aerial biomass was collected manually in three areas of 0.4 m² per plot, and then weighed, at two dates: November (before the first significant frost) and early G4 stage (BBCH 73), considered as the stage of maximum N accumulation in the aerial parts of the WOR. For the first measurement, in November, the collected biomass was separated into two fractions, i.e. winter oilseed rape and companion plants. The dry matter content was obtained after drying a representative subsample at 70 °C during 48 h. The aerial dry weight (ADW) was calculated as the product of fresh weight and dry matter content. The total nitrogen concentration in aerial biomass (r_N) was determined using the Dumas method. The aerial N content (ANC) was calculated as the product of aerial dry weight and N concentration.

These measurements were also carried out at F2 stage (BBCH 61) in order to calculate the nitrogen nutrition index (NNI), used as an indicator of underestimation or overestimation of the a priori optimal fertilization rate ‘X’ (NNI< 1 or NNI> 1, respectively), with the equation (1) defined by Colnenne et al. (1998): (1)where NNI is the nitrogen nutrition index, r_N is the nitrogen concentration in aerial biomass and ADW is the aerial dry weight.

The percentage of ground cover by weeds was estimated in subplots, representing approximatively half of the plots (60 to 100 m²), receiving no broad leaf herbicide. These time-consuming measurements were only carried out in four trials: Chambon 2012/2013 (in October); Villedieu-Sur-Indre 2012/2013, 2013/2014; and Murs 2013/2014 (in November). The relationship between the percentage of ground cover by weeds and the aerial dry matter in November has only been evaluated in the three trials where both measurements were carried out in November.

Damage caused by rape winter stem weevil (Ceutorhynchus picitarsis) was used to estimate the effect of intercropping WOR with companion plants on insects. Indeed, rape winter stem weevil is considered as one of the most harmful pests of WOR in its main cropping area in France (central and eastern France). After laying in autumn, larvae can migrate in the spring, toward the terminal bud and destroy it, leading to bushy plants with reduced fertility. We thus used the percentage of bushy plants at F2 stage (BBCH 61) as indicator of insect damage. It was estimated at both sites where the pressure was the highest (Villedieu-Sur-Indre and Murs), in 2011/2012, 2012/2013 and 2013/2014, on 100 plants per plot. In the six trials with measurements, no autumn insecticide had been applied, except in Villedieu-Sur-Indre in 2012 where a “Karate zéon” (λ-cyhalothrine) was applied on all plots at a rate of 0.075 l ha^-1.

At maturity of the WOR, 20 m² were harvested. The standard yield was calculated after determination of the actual dry matter content (h) and the actual impurity content (i) according to the equation (2): (2)where h is the actual dry matter content and i is the actual impurity content.

The grain nitrogen use efficiency (NUE) was calculated according to Good et al. (2004) as: (3)where NUE is the nitrogen use efficiency, and ‘Fertilizer N’ is the amount of fertilizer N applied per hectare.

In addition to the absolute value, three variables of the intercrops: ADW in November, ANC at G4 stage and grain yield were also calculated, in each trial and for each block, as the deviation from the value of the sole WOR. These calculations in relative value were used to assess the competition/facilitation effect of the companion plants.

All measurements were not carried out in all trials. In particular, no measurements were carried out in Chambon in 2013 due to unsuccessful destruction of the companion plants and a consecutive high competition of the WOR in the spring. All details are given in Table 3.

2.4 Statistical analysis

Statistical analyses by ANOVA were performed, after verifying that distributions were normal and variances homogeneous, using R software (“aov” function) to assess the effects of the four treatments, year, site and site*year interaction, on different variables (Tab. 4), according to the equation (4): (4)where Y_ijkl is the criterion variable (see Tab. 4), μ is the mean, T_i stands for the treatment condition i, S_j for the site j, Y_k for the year k, B_l for the replicate l and ε_ijkl is the residual. All terms are considered as fixed effect.

The significance of the differences between treatments were estimated using the Student, Newman & Keuls (“SNK.test” function) range test with α = 0.05.

The link between variables was assessed with simple and multiple regressions of R software (“lm” function), according to the general equation (5): (5)where Y_i is the response variable value for measure i, x_1i, x_2i, ..., x_ki are the regressors, β₀,β₁,β₂, ..., β_k are intercepts and ε_i is the residual.

3 Results

3.1 Growth in autumn

In November, the aerial dry weight of the sole WOR averaged 152 g m^-2 (Fig. 1). The effects of site and year were significant (Tab. 4) with Villedieu-Sur-Indre = Murs > Chambon = Morville-Sur-Seille and 2011 2012 = 2014 > 2013, respectively. The aerial dry weight of the intercropped WOR was slightly lower (135 g m^-2 on average), whatever the companion crop (Fig. 1; Tab. 4). The aerial dry weight of the companion plants was approximately half of that of the WOR, and was significantly higher for BL compared to GFL and VVT (Fig. 1; Tab. 4). The effects of site and year were significant (Tab. 4) with Murs > Chambon > Villedieu-Sur-Indre > Morville-Sur-Seille and 2012 > 2014 > 2011 > 2013, respectively. Whatever the companion plant, the total aerial dry weight was significantly higher for the intercrops (WOR + companion plants) compared to sole WOR and ranked as follows: WOR + BL = WOR + VVT > WOR + GFL > Sole WOR (Fig. 1; Tab. 4).

Fig. 1 Aerial dry weight of the winter oilseed rape (WOR), the companion plants (CP) and the total WOR + CP, in November and for the four experimental treatments.

The competition/facilitation due to the companion plants in autumn was estimated through the deviation of aerial dry weight from sole WOR. A competition effect (decrease in WOR ADW with the increase in CP ADW) should be due to the share of water, nutrient or radiation. A facilitation effect (increase in WOR ADW with the increase in CP ADW) should be due to an improvement of WOR root exploration. The deviation of aerial dry weight from sole WOR decreased very slightly but significantly (p = 0.01103) with the increase in the aerial dry weight of the companion plants (Fig. 2). However, in absolute terms, the aerial dry weight of the sole WOR and the companion plants were positively and significantly (p = 1.438e-05) correlated (Fig. 3).

Fig. 2 Relationship between the aerial dry weight deviation from sole winter oilseed rape (WOR) and the aerial dry weight of the companion plants (CP) in November.

Fig. 3 Relationship between the aerial dry weight from sole winter oilseed rape (WOR) and the aerial dry weight of the companion plants (CP) in November.

Fig. 4 Aerial N content of the winter oilseed rape (WOR), the companion plants (CP) and the total WOR + CP, in November and for the four experimental treatments.

In November, the aerial N content of the sole WOR averaged 42.6 kg ha^-1 (Fig. 4). The aerial N content of the intercropped WOR did not differ significantly from the sole WOR, whatever the companion crop (Fig. 4; Tab. 4). The N concentration of the WOR was significantly higher for the WOR + BL and WOR +VVT treatments compared to sole WOR and WOR +GFL treatments (Tab. 4). The aerial N content of the companion plants was significantly higher for BL compared to GFL and VVT (Fig. 4; Tab. 4). The total aerial N content was significantly higher for the intercrops (WOR + companion plants) compared to sole WOR and ranked as follows: WOR + BL = WOR + VVT > WOR + GFL > Sole WOR (Fig. 4; Tab. 4).

3.2 Autumn insect damage

The damage caused by the rape winter stem weevil in autumn was estimated through the percentage of bushy plants at flowering. It was lower for intercrop treatments than for sole WOR treatments (Fig. 5). The difference was not significant, but the p-value was low (0.095; Tab. 4).

Fig. 5 Percentage of bushy plants at flowering for the four experimental treatments.

Fig. 6 Relationship between the percentage of bushy plants at flowering and the total aerial dry weight of sole winter oilseed rape (Sole WOR) or winter oilseed rape + companion plants (WOR + CP) in November.

The percentage of bushy plants decreased with the increase in total aerial dry weight of plants, whatever its source (Sole WOR or WOR + companion plants). When considering intercrops, it was below 20% from 200 g m^-2 of total aerial dry weight (Fig. 6).

Fig. 7 Percentage of weed cover in Autumn for the four experimental treatments.

3.3 Autumn weed cover

The percentage of weed cover in Autumn was significantly lower for intercrop treatments than for sole WOR treatments (Fig. 7). It decreased with the increase in total aerial dry weight of plants whatever its source (Sole WOR or WOR +companion plants). When considering intercrops, it was below 20% from 200 g m^-2 of total aerial dry weight (Fig. 8).

Fig. 8 Relationship between the percentage of weed cover in November and the total aerial dry weight of sole winter oilseed rape (Sole WOR) or winter oilseed rape + companion plants (WOR + CP) in November.

3.4 WOR Aerial dry weight and N content at G4 stage

At the early G4 stage, the treatment effect on the aerial dry weight was significant (Tab. 4). The values were higher for the intercrop WOR + BL than for the other treatments (not shown). The differences of aerial N contents between the other treatments were not significant, despite the reduction of the N fertilization rate (–30 kg per hectare) for the intercrops (Fig. 9; Tab. 4). In order to analyze the effect of a possible overestimation of the a priori optimal N fertilization rate ‘X’ which could explain this absence of difference, the trials were divided into two categories: (i) trials in which the NNI at flowering of the sole WOR was below 1 (i.e. ‘X’ was not considered as overestimated) and (ii) trials in which the NNI at flowering of the sole WOR was above 1 (i.e. ‘X’ was considered potentially overestimated). In both cases, no significant differences were found between treatments on the aerial N content (Fig. 9; Tab. 4). More surprisingly, the trend was an increase in the aerial N content of the intercrops compared to sole WOR in the trials with NNI< 1, and a decrease in the aerial N content of the intercrops compared to sole WOR in the trials with NNI> 1.

Fig. 9 Aerial N content of rapeseed at early G4 stage in all trials, in the trials in which the NNI of the sole WOR was < 1 and in the trials in which the NNI of sole WOR was > 1, for the four experimental treatments.

Fig. 10 Relationship between the aerial N content deviation from sole winter oilseed rape (WOR) at early G4 stage and the aerial dry weight of the companion plants (CP) in November.

The main hypothesis to explain the contribution of companion plant to WOR N nutrition is its mineralization which should increase with the increase in the CP N content and the CP N concentration. In out experiment, the aerial N content deviation from sole WOR was not explained by the aerial dry weight of the companion plant in November (Fig. 10) neither by its N concentration nor the interaction between both variables (Tab. 5).

3.5 WOR yield

Despite the reduction of the N fertilization rate (–30 kg per hectare) for the intercrops, the grain yield was significantly higher for WOR +BL than for the sole WOR (+0.14 t ha^-1). No significant differences were found between the two other intercrops and the sole WOR (Fig. 11, Tab. 4). The effects of site and year were significant (Tab. 4), with Chambon > Murs > Villedieu-Sur-Indre = Morville-Sur-Seille and 2014 = 2012 > 2011 > 2013, respectively. Due to similar or even higher yield with reduced N fertilization rate, the grain NUE was significantly improved with intercropped WOR compared to sole WOR (Tab. 4), and ranked as follows: WOR +BL > WOR + VVT = WOR + GFL > Sole WOR.

Fig. 11 Standard grain yield for oilseed rape in all trials, in the trials in which the NNI of the sole WOR was < 1, and in the trials in which the NNI of sole WOR was > 1, for the four experimental treatments.

In order to analyze the effect of a possible overestimation of the a priori optimal N fertilization rate ‘X’ which could explain the absence of difference in yields, the trials were divided into two categories: (i) trials in which the NNI of the sole WOR at flowering was below 1 (i.e. ‘X’ was not considered as overestimated) and (ii) trials in which the NNI of the sole WOR at flowering was above 1 (i.e. ‘X’ was considered potentially overestimated). Surprisingly, the grain yield was systematically higher for the intercrops than for the sole WOR in the trials in which the NNI of the sole WOR was below 1 (insignificant difference). The grain yield deviation of the three intercrops from sole WOR decreased in the trials in which the NNI of the sole WOR was above 1 (Fig. 11, Tab. 4).

Considering all trials, the grain yield deviation from the sole WOR was significantly affected by site and year and ranked as follows: Chambon (+0.428 t ha^-1) > Villedieu-Sur-Indre (+0.040 t ha^-1) = Murs (–0.032 t ha^-1)   = Morville-Sur-Seille (–0.089 t ha^-1) and 2013 > 2011 = 2012 > 2014, respectively (Tab. 4). It decreased significantly (p = 1.814e-06) with the increase in the grain yield of the sole WOR (Fig. 12). The multiple regression showed that it was also affected by both the aerial dry weight (negatively) and the aerial N concentration (positively) of the companion plants in November (Fig. 13, Tab. 5).

Fig. 12 Relationship between the grain yield deviation from sole winter oilseed rape (WOR) and the grain yield of the sole WOR.

Fig. 13 Relationship between the grain yield deviation from sole winter oilseed rape (WOR) and the aerial dry weight of the companion plants (CP) in November.

4 Discussion

4.1 Competition/facilitation effect of the companion plants

In autumn, the slight competitive effect of the companion plants on the aerial dry weight of winter oilseed rape had no consequence on its aerial N content. This was due to an increase in the aerial N concentration of the WOR in the intercrop treatments (except for WOR +GFL). These opposite effects could be explained by a likely competition of the companion plants for light on the one hand and facilitation for root exploration on the other hand. Indeed, net N transfer from the legume mixtures to the WOR in autumn is unlikely to be significant (Jamont et al., 2013). The aerial dry weight of the companion plants largely offset the slight reduction in the aerial dry weight for winter oilseed rape; intercropping was thus a way to maximize the total aerial dry weight and the total aerial N content (winter oilseed rape + companion plants) in autumn.

The companion plants were destroyed during winter in all trials (except Chambon in 2013), either by freezing or by herbicides. The aerial dry weight deviation from the sole WOR treatments was not observed any longer at early G4 stage, considered as the stage of maximum aerial N content in the WOR. The reduction of the N fertilization rate of 30 kg per hectare for the intercrops was partially or even completely compensated in the intercropped WOR. The first hypothesis explaining this effect is the mineralization of the companion plant residues, and the absorption of this N by the WOR. Indeed, Redin et al. (2014) showed a fast mineralization of crop residues decomposing at the soil surface, in particular with Fabaceae crops. The absence of relationship between the deviation of the aerial N content from the sole WOR treatments and the companion plants characteristics (aerial dry weight, aerial N concentration and interaction between both variables) could be due to the importance of the multiple factors controlling the mineralization of mulches: biochemical residue quality, residue placement, soil water content (Coppens et al. 2007), or climatic conditions during mineralization. However the mineralization of companion plant residues is not likely to be the only factor explaining the compensation in N acquisition in the WOR. Indeed, in the experiment of Redin et al. (2014) the C mineralization was not complete after 120 days under subtropical conditions. Moreover, the aerial N content of the companion plants in our experiment was lower than 30 kg per hectare on average. A possible facilitation for P extraction is unlikely to explain any difference since our experiments were all carried out in well supplied soils. An indirect effect through an improvement of root exploration is thus a possible hypothesis. Indeed, Shroder and Kopke (2012) showed that intercropping faba bean and oil crops resulted in more regular horizontal root distribution and enhanced the root length density in the subsoil. This phenomenon could also explain the equivalent or even higher grain yields in intercropped treatments compared with sole WOR treatments, despite a slightly lower aerial N content at the stage of maximum aerial N content. Intercropping winter oilseed rape with frost-sensitive legume crops leads to improve its grain nitrogen use efficiency.

However it is interesting to highlight that the aerial N content deviation from sole WOR decreased with the increase in sole WOR NNI, and that the yield deviation from sole WOR decreased with the increase in sole WOR NNI and yield. This suggests that the benefit of intercropping WOR with legume crops increases with the decrease in soil fertility or nitrogen availability. This could be explained again by an improvement of WOR root exploration due to companion plants, which could be beneficial or have almost no effect when soil N and water content availability are respectively limiting or not limiting. This is consistent with Hauggaard-Nielsen (2005) who showed, in intercrops, that critical interspecific competition for plant growth factors in nutrient poor soils and in low input systems explained the greater importance of facilitative root interactions in such conditions.

4.2 Effect of companion plants on the weed control in autumn

Our results showed the contribution of the companion plants in the control of weed cover. This had yet been demonstrated with living mulches in no-till cropping systems (Ilnicki and Enarche, 1992; Hiltbrunner et al. 2007), and with intercropping (Agegnehu et al., 2008; Bulson et al., 1997; Hauggaard-Nielsen et al., 2001). Lorin et al. (2014) showed that, in the case of winter oilseed rape intercropped with frost-sensitive legume crops, the main factor explaining such effects was the increased competition for light, itself linked with the biomass production of the companion plants and the complementarity of its shape with the one of WOR. Our results suggest a possible substitution of the biomass of both intercropped plants. This is particularly interesting since the threshold of 200 g of aerial dry weight per square meter for an efficient competition vs. weeds was reached on average by intercropped treatments but not by sole WOR treatments. This contribution of the companion plants to weed control could compensate a reduction in herbicide applications, necessary to avoid phytotoxicity against legumes (Sauzet, pers. comm.).

4.3 Effect of companion plants on damage caused by rape winter stem weevil

Our results show a likely effect of the companion plants on damage caused by the rape winter stem weevil. This positive effect of the companion plants had been reported for different brassicas and different insects (Theunissen, 1994; Costello, 1994; Hooks and Johnson, 2004), and explained by a variety of phenomena: visual and olfactory confusion, mechanical barrier, increased presence of natural enemies, etc. (Theunissen, 1994). In our experiment, there was a clear effect of the total aerial dry weight (Sole WOR or WOR +CP) on the reduction of the percentage of bushy plants. The superposition of the observations related to both sole WOR and WOR + companion plants suggests a substitutive effect of both the companion plants and the WOR biomass, when the total weight exceeds 200 g m^-2. The biomass effect of the WOR is likely due to the increase in the difficulty for the larvae to reach the terminal bud with the increase in stem size (C. Robert, pers. comm.). The biomass effect of the companion plants should be due to an increase in the phenomena described by Theunissen (1994) with the increase in aerial weight. Through an increase in the total aerial dry weight in November, intercropping legume crops with winter oilseed rape appears thus as a way to mitigate damage caused by rape winter stem weevil. However, this effect, alone, seems too limited and variable to advise reduce insecticide applications. It could rather be a contribution to an integrated pest management strategy.

5 Conclusion

In this study, we demonstrated the various benefits of mixing frost-sensitive legume crops with winter oilseed rape. We showed the higher total aerial dry weight and total aerial nitrogen content in the intercrops compared to sole winter oilseed rape in November. The companion plants contributed to the control of weeds and the mitigation of the rape winter stem weevil damage, notably through the increase in the total aerial weight. In spring, after destruction of the companion plants, the intercrops had partially compensated a reduction of the N fertilization rate (–30 kg per ha^-1) in terms of aerial nitrogen content in rapeseed, with no consequences on the yield which was maintained or even increased. The grain nitrogen use efficiency was consequently improved in intercrops.

Further research is needed to better understand the mechanisms involved in these effects and to better analyze the variability observed and the suitability in different contexts of soils and climate. The effect of legume species has to be analyzed in order to clarify which plant traits have to be chosen in several combinations of companion plants. Finally, since the mineralization of the living mulch during the rapeseed growth seems incomplete, an evaluation of the fate of N after harvest would be necessary.

Acknowledgments

This work was carried out under the project “Phyto-sol”, which is part of the French “Ecophyto plan” led by the Ministry of Agriculture, with the financial support of the National Agency for Water and Aquatic Environments by credits from the fee for diffuse pollution. The authors are grateful to N. Urvoit, D. Poisson, R. Lange, D. Forget, D. Collin for their technical assistance, and to L. Lot for English revision.

References

All Tables

All Figures

	Fig. 1 Aerial dry weight of the winter oilseed rape (WOR), the companion plants (CP) and the total WOR + CP, in November and for the four experimental treatments.
In the text

	Fig. 2 Relationship between the aerial dry weight deviation from sole winter oilseed rape (WOR) and the aerial dry weight of the companion plants (CP) in November.
In the text

	Fig. 3 Relationship between the aerial dry weight from sole winter oilseed rape (WOR) and the aerial dry weight of the companion plants (CP) in November.
In the text

	Fig. 4 Aerial N content of the winter oilseed rape (WOR), the companion plants (CP) and the total WOR + CP, in November and for the four experimental treatments.
In the text

	Fig. 5 Percentage of bushy plants at flowering for the four experimental treatments.
In the text

	Fig. 6 Relationship between the percentage of bushy plants at flowering and the total aerial dry weight of sole winter oilseed rape (Sole WOR) or winter oilseed rape + companion plants (WOR + CP) in November.
In the text

	Fig. 7 Percentage of weed cover in Autumn for the four experimental treatments.
In the text

	Fig. 8 Relationship between the percentage of weed cover in November and the total aerial dry weight of sole winter oilseed rape (Sole WOR) or winter oilseed rape + companion plants (WOR + CP) in November.
In the text

	Fig. 9 Aerial N content of rapeseed at early G4 stage in all trials, in the trials in which the NNI of the sole WOR was < 1 and in the trials in which the NNI of sole WOR was > 1, for the four experimental treatments.
In the text

	Fig. 10 Relationship between the aerial N content deviation from sole winter oilseed rape (WOR) at early G4 stage and the aerial dry weight of the companion plants (CP) in November.
In the text

	Fig. 11 Standard grain yield for oilseed rape in all trials, in the trials in which the NNI of the sole WOR was < 1, and in the trials in which the NNI of sole WOR was > 1, for the four experimental treatments.
In the text

	Fig. 12 Relationship between the grain yield deviation from sole winter oilseed rape (WOR) and the grain yield of the sole WOR.
In the text

	Fig. 13 Relationship between the grain yield deviation from sole winter oilseed rape (WOR) and the aerial dry weight of the companion plants (CP) in November.
In the text