Introduction
The primary objective of weed management is to reduce the negative effects of weeds on crop production. Effectively, the potential crop loss due to weeds is estimated up to 34% each year (Oerke, Reference Oerke2006) and weeds are often recognized as the most detrimental threat to crop production in systems with reduced or no use of agricultural chemicals, including organic farming (Uchino et al., Reference Uchino, Iwama, Jitsuyama, Ichiyama, Sugiura and Yudata2011). This detrimental effect is mainly related to the ability of weeds to compete with crops for limiting resources (i.e., light, water, and nutrients). The relative competitive ability of weeds is mainly determined by both the species characteristics (i.e., botanical and physiological aspects) and the cropping system management (Lundkvist and Verwijst, Reference Lundkvist, Verwijst and Nokkoul2011). These aspects are particularly relevant for vegetable crops, which are commonly weak competitors against weeds (Masiunas, Reference Masiunas1998), especially in low input systems (den Hollander et al., Reference den Hollander, Bastiaans and Kropff2007a). Therefore, by cropping system optimization, it is possible to aim at a reduction of weed establishment through crop competition (Blackshaw et al., Reference Blackshaw, Anderson, Lemerle, Upadhyaya and Blackshaw2007). Furthermore, according to agro-ecology principles, the implementation of systematic preventive and cultural practices including crop rotations, cultivar choice, cover crops, tillage systems (e.g., minimum tillage), as far as fertilization and irrigation programs, is needed for an effective long-term weed management in crop production systems (Gaba et al., Reference Gaba, Fried, Kazakou, Chauvel and Navas2014). In particular, the introduction of cover crops may aid in creating an unfavourable environment for weeds while ensuring a greater level of biodiversity and soil protection. For their ability to provide a wide range of services (e.g., pest control and soil quality) (den Hollander et al., Reference den Hollander, Bastiaans and Kropff2007b), including agro-ecological weed management, the most recent scientific literature defined cover crops as agro-ecological service crops (ASC) (Canali et al., Reference Canali, Diacono, Campanelli and Montemurro2015). Accordingly, in the present paper, we decided to use this terminology.
Intercropping a cash crop with ASC as a living mulch (LM) may potentially reduce weed infestation by limiting seed germination by ASC light interception (Phatak, Reference Phatak1992) or secretion of allelochemicals (White et al., Reference White, Worsham and Blum1989). Several studies have reported effective weed suppression with LM (Brainard and Bellinder, Reference Brainard and Bellinder2004; Araki and Tamura, Reference Araki and Tamura2008; Pouryousef et al., Reference Pouryousef, Yousefi, Oveisi and Asadi2015) as well as yield loss due to direct competition between the ASC and the cash crop (den Hollander et al., Reference den Hollander, Bastiaans and Kropff2007b; Kolota and Adamczewska-Sowińska, Reference Kolota and Adamczewska-Sowińska2013). As a consequence, a successful LM system should provide a balance between competition against weeds and acceptability for the cash crop with respect to limiting resources (Chase and Mbuya, Reference Chase and Mbuya2008). This depends on finding mechanisms for uncoupling weed and cash crop suppression by ASCs (Teasdale, Reference Teasdale, Hatfield, Buhler and Stewart1998). Delayed ASC seeding was demonstrated to be effective to minimize the yield losses in many crops, allowing the weed control in the first, critical period of weed infestation during the cash crop growth (Brainard and Bellinder, Reference Brainard and Bellinder2004; Kolota and Adamczewska-Sowińska, Reference Kolota and Adamczewska-Sowińska2013). Since the LM can compete with cash crop for belowground resources (Båth et al., Reference Båth, Kristensen and Thorup-Kristensen2008), the choice of ASC with low demand for nitrogen (e.g., leguminous) or water may partially overcome this competition effect (Mohammadi, Reference Mohammadi and Price2012). Otherwise, the introduction of LM strips alternated with cash crop rows can contribute to reduce a negative interference of ASC on crop production (Kolota and Adamczewska-Sowińska, Reference Kolota and Adamczewska-Sowińska2013). Cultivar or crop genotypes selections that possess traits conferring a greater ability to grow with ASCs, also should be considered, in addition to environmental conditions and management (Bárberi, Reference Bárberi2002). In such a context, the Competitive Balance (Cb), which quantifies the ability of a crop to compete with plant competitors (weed as well as ASC), can be a useful tool in comparing different treatments and cultivars for their ability to compete with LM in weedy fields (Paolini et al., Reference Paolini, Faustini, Saccardo and Crinò2006; Ciaccia et al., Reference Ciaccia, Montemurro, Campanelli, Diacono, Fiore and Canali2015).
In order to investigate the relative effectiveness of the LM to improve crop performance and competitive ability, the following specific objectives were pursued: (1) identifying the LM management with the highest weed suppression and cash crop yield potentials; (2) assessing the LM management with the lowest detrimental effect on cash crop growth; (3) evaluating the genotype effect on the cash crop-neighbour plants (i.e., weed and ASC) competitive relationships. In order to reach these objectives, in organically managed vegetable cropping systems, two 2-year field experiments were carried out on transplanted leek (Allium porrum L.) and cauliflower (Brassica oleracea L. var. botrytis) in Denmark and Central Italy, respectively. Since biomass of a plant gives an integrated estimation of both ability to grow and to compete with neighbours, the investigation of its production were used to provide insight into the effect of the analyzed treatments on the weed-cash crop-ASC relationships. The calculation of indices of competition was also performed with the aim to deeply analyze the effectiveness of the LM management on weed-crop competition reduction. In addition, the identification of the weed species was performed to provide additional information on the effect of LM introduction on the field plant community.
Materials and Methods
Sites and experimental designs
The research was carried out during 2011–2012 and 2012–2013 seasons (indicated as 2012 and 2013, respectively) in two sites: (1) the Research Centre of Aarslev on the island of Funen in Denmark (55°18′N, 10°27′E); (2) the MOnsampolo VEgetable organic long-term field experiment (MOVE LTE) located in the CREA-Research unit for vegetable production (CREA-ORA) of Monsampolo del Tronto, in central Italy (42°53′N, 13°48′E).
The Aarslev experimental site is characterized by a temperate climate in the nemoral zone (Ahti et al., Reference Ahti, Hämet-Ahti and Jalas1968), with an average total annual precipitation of 634 mm and temperatures averaging 17 and 5°C in the May–November growing period (Fig. 1a). Soil type at the field trial site was a sandy loam (Typic Agrudalf) (Soil Survey Staff, 1996). The experimental layout was a completely randomized design with three replicates (blocks). Each experimental plot was the combination of the LM and genotype (CV) treatments. The following LM treatments were compared: (1) no LM or control (no LM), (2) LM early sowing (sown 5 weeks delayed after cash crop transplanting––es LM) and (3) LM late sowing (sown 8 weeks delayed after cash crop transplanting––ls LM). In the es LM and ls LM treatments, the design was a substitution design where the rows of LM, dyers woad (Isatis tinctoria L.), replaced every third row of leek which reduced the crop density by one-third compared with the no LM treatment. In order to meet the specific needs of this study, an additional control was introduced with no LM in which every third crop row was substituted with an empty row, reaching the same crop density for es LM and ls LM treatments. The obtained no LM treatment was considered in substitution of the original control (hereafter Standard no LM), and used in the assessment reported in this paper. Two different leek cultivars were compared (Hannibal––Ha––an open-pollinated cultivar, and Runner––Ru––a hybrid one). Each experimental plot was divided in two sub-plots, in one of which (3.2 × 6.5 m2) weeding was performed during the leek cropping cycle (i.e., the standard agronomic practice(s) for leek production in Denmark)––Weeding––and not performed (3.2 × 3.5 m2)––No Weeding. A combination of mechanical and manual weeding was performed four times over the season, middle of June, early July, end of July and end of August (2012) or early September (2013) by use of a weed brush machine and hand hoeing. The es LM and ls LM were weeded until the sowing of the LM strips in the No Weeding sub-plots. Leek transplants were 10 weeks old and were hand-transplanted at an inter-row × row distance spacing of 0.50 × 0.08 m2 on May 23 and 25 in 2012 and May 31, 2013, respectively. Sowing of dyers woad in LM strips occurred on July 4 and 23, 2012 in es LM and ls LM, respectively, and July 5 and 26, 2013 in es LM and ls LM, respectively. In 2012, the harvest was carried out on October 24–25, with a cropping cycle of 153 days. In 2013, the harvest occurred on September 25, with a cropping cycle of 117 days. The field management was since 1996 performed according to the Danish organic management regulations without the use of pesticides or inorganic fertilizers. All treatments were irrigated when precipitation deficits exceeded 30 mm. Dry chicken manure was applied as N fertilizer before transplanting leek to a total of 200 kg N ha−1 (spring soil inorganic N in the soil layer of 0-0.3 m and total manure N).
Mean monthly temperature and rainfall at the Aarslev experiment during May 2012–November 2013 compared with long-term (28 year) mean values (a) and mean monthly temperature and rainfall at the MOVE long term experiment during July 2011–February 2013 compared with long-term (30 year) mean values (b). The leek-growing period was May–November in both experimental years at the Aarslev site. The cauliflower-growing period was August–February in both experimental years at the MOVE long term experiment.
The MOVE Experimental site is characterized by a ‘thermomediterranean’ climate (UNESCO–FAO, 1963), with an average total annual precipitation of 564 mm and temperatures averaging 9 and 20°C in the October–March and April–September growing periods, respectively (Fig. 1b). Soil type at the field trial site was fine-loamy, mixed thermic Typic Calcixerepts (U.S. Department of Agriculture, 1996). The experimental design was a split-plot with two factors and three replications. The main factor was the type of LM management: (1) no LM, (2) burr medic (Medicago polymorpha L., var. anglona) LM sowed contemporary to the cash crop transplanting (es LM), and (3) burr medic LM sowed at least 3 weeks after the cash crop transplanting (ls LM). The subplot, randomized within each LM treatment, was the cauliflower (Brassica oleracea L., var. botrytis) genotype, and three different cultivars were compared––a hybrid cultivar Emeraude (EHF1), and two open-pollinated, locally adapted cultivars (ORA1 and ORA2). All es LM and ls LM plots (5 × 20 m2) were weeded by hoeing until the LM sowing. In the no LM treatment weeding was performed according to standard agronomic practices used by organic farmer in the area (i.e., last hoeing 5 and 6 weeks after cauliflower transplanting in September 2011 and 2012, respectively), thus ensuring low weed presence in the critical period of the cash crop cycle. Cauliflower transplants were 4 weeks-old and were hand-transplanted at an inter-row × row distance spacing of 0.7 × 0.6 m2 on August 17 and 13 in 2011 and 2012, respectively. Burr medic was broad sowed in the LM plots contemporary to cauliflower transplanting in es LM and on September 8, 2011 and September 12, 2012 in the ls LM treatment. In the 2012 season, the harvest began on December 16, 2011 in both the ls LM and no LM treatments and was completed on January 24, 2012, with a maximum cropping cycle of 164 days, whereas it began on January 17, and completed on February 23, in the es LM, with a cropping cycle of 195 days. In the following season, the harvest occurred from November 16 to December 4, with a cropping cycle of 111 days in both the ls LM and no LM treatments, and from November 23 to December 20, with a maximum cropping cycle of 128 days in the es LM.
In order to examine in detail crop-weed competition, additional extra plots were included in both the experiments next to main plots in the experimental layouts as: (1) weed stands without cash crops (pure weed stands); (2) unweeded LM stands without cash crop (LM-weeds mixed stands) and (3) without cash crop and weeds (LM pure stands). In the Aarslev experiment, pure weed stands (3.2 × 1.5 m2) were obtained as a sub-area in the No weeding subplot of the no LM treatment. An additional no crop plot (i.e., crop rows substituted by empty rows) followed the same randomization of experimental plots. The two sub-plots were the LM pure and LM-weeds mixed stands.
In the MOVE experiment, where the weeding was not included as subplot, cash crop stands without weeds (manual weeding) for each LM × cultivar combination were also included (pure cash crop stands and LM-cash crop mixed stands) as strips (2.1 × 18 m2) on the sides of the experimental plots. Cash crop spacing in these stands was identical to the main experiment. Overall plant density in LM/weed/cash crop mixtures was estimated in the weed seedling stage and was similar to the sum of the density of LM, cash crop and weeds in their pure stands, following as close as possible an additive design for determining competition indices (Snaydon, Reference Snaydon1991; Paolini et al., Reference Paolini, Faustini, Saccardo and Crinò2006). The pure weeds and LM-weeds mixed stands were included as strips (21 × 2 m2) in the experimental layout. The LM pure stands (manual weeding; 1 × 1 m2) were obtained in the experimental plots, following their randomization.
Plant sampling
At the end of the harvest period, in the two experiments, aboveground LM, weed and cash crop residue and yield biomasses were separately collected in the experimental and extra plots. In the Aarslev experiment, two rows × 3 m were harvested by hand for leek while a 0.5 × 1 m2 area was cut at soil surface for LM and weeds. In the MOVE experiment, 0.7 × 0.6 m2 areas, for plot with cash crop, were randomly selected and plant materials were collected by cutting all plants at soil surface. Similarly, half-meter-square quadrats were randomly sampled for the pure-weed-stand, LM-pure-stand and LM-weed mix-stand areas. In each cash crop-pure-stand area a single cauliflower plant was also collected. All the sampled areas of the two experiments were randomly individuated in the middle of both the experimental plots and stands, in order to avoid the plot edges and reduce the border effect on obtained data. The obtained biomasses were dried at 80 and 70°C for 48 h to obtain dry weights. The crop yield and plant biomass were summed to obtain total crop aboveground biomass.
Indices of competition measurement
In order to measure cash crop yield losses as affected by weed competition and quantify the relative ratio of weed, LM and crop biomass in each plot (Paolini et al., Reference Paolini, Principi, Froud-Williams, Del Puglia and Biancardi1999), indices of competition comparing data from experimental plots and additional extra plots were calculated for each treatment. Using yield data from the weed-free areas and the main plots, an Agronomic Tolerance to Competition (ATC) value was calculated where a high value symbolized a low competitive effect of weeds and LM on crop yield. In the MOVE experiment, using yield data from the pure-crop stands the ‘ATC lm’ index was used to evaluate the tolerance of cash crop to the LM presence without weeds.
Other indices, comparing the cash crop, LM and weed biomass, were also used. For ease of comparison, the indices of competition assumed the weed community as a single species, like in previous studies (Paolini et al., Reference Paolini, Faustini, Saccardo and Crinò2006). The following indices were calculated (Table 1): Relative Biomass of Crop (RBc; Aarslev experiment), Relative Biomass of Weeds and LM (RBn; Aarslev experiment), Relative Biomass Total (RBT; Aarslev experiment) as sum of the relative RBc and RBn and Cb index (Aarslev and MOVE experiments). The RBc provided a comparison of the effect of neighbour plants (i.e., weeds and/or LM) competition on cash crop biomass production, while the RBn quantified the effect of cash crop competition on neighbour biomasses. A high Cb value signified low neighbour biomass produced in the presence of the cash crop, thereby reducing weed reproduction rates and risk of infestation in subsequent years. Competitive indices were also used to: (1) determine the competitive relationships between cash crop and LM without weeds (i.e., Cbc-lm to assess the competitive ability of cash crop against the LM––both the experiments); (2) determine the competitive relationships between the LM and the weeds (i.e., Cb lm-w to assess the competitive ability of LM against weeds––MOVE experiment).
Crop–weed/living mulch indices of competition utilized in the two experiments, Aarslev (DK) and MOVE (IT), 2012–2013.
Weed community and coverage assessment
In the Aarslev experiment, the main species for each LM treatment were recorded in order to put in evidence variation due to the treatments.
In the MOVE experiment, in order to evaluate cash crop/LM-weed competition during cauliflower plant cycle, a coverage assessment was estimated in each LM management treatment in two phases: cauliflower-head emergence (i.e., head emergence) and final harvest (i.e., harvest). At both the phases weed species were recognized and total weed coverage for each plot was estimated according to the Braun-Blanquet scale for visual evaluation (Braun-Blanquet, Reference Braun-Blanquet1932), as modified by Pignatti (Reference Pignatti and Cappelletti1976): 5 for coverage ranging from 75 to 100% of the plot; 4 for 50–74%; 3 for 25–49%; 2 for 5–24%; 1 for 1–5% and + for coverage <1%. At each sampling time, a cover abundance/dominance index for each species was also evaluated. Each Braun-Blanquet class was converted to its midpoint coverage value and graphed as coverage over time according to Wikum and Shanholtzer (Reference Wikum and Shanholtzer1978).
Statistical analysis
Biomass parameters and indices of competition were analyzed by analysis of variance (ANOVA) using LM treatment, cultivar and year as factors. In the Aarslev experiment, also the weeding factor was used for leek yield data, whereas the other parameters were evaluated considering the No Weeding sub-plot results. The Duncan Multiple Range Test (DMRT) was performed for treatment mean comparisons (P ≤ 0.05 probability level). Before analysis, data for ATC (range 30–80%) required angular transformation (Gomez and Gomez, Reference Gomez and Gomez1984). The Kruskal–Wallis H-test, based on rank transformation, was applied for the analysis of cover abundance/dominance (Hahn and Scheuring, Reference Hahn and Scheuring2003) in the MOVE experiment. The pairwise comparisons per cover management factor were processed by the Mann–Whitney post hoc test (Lehmann and D'Abrera, Reference Lehmann and D'Abrera1975). All analyses were performed with the Statistical Package for Social Science (SPSS) 16.0 package.
Results and Discussion
Climatic variations were different over the 2 experimental years in the two sites (Fig. 1). In the Aarslev experiment air temperature was near the long-term (28 years) mean in both years, but greater and more consistent rainfall during the cropping cycle in 2012 than in 2013 (475 and 326 mm from May to October in 2012 and 2013, respectively) was recorded (Fig. 1a). In the MOVE experiment slight differences for air temperature compared with the long-term (30 years) mean was recorded, whereas the August 2011–January 2012 period showed lower rainfall than the same period of the following season (120 and 507 mm, respectively) and the long-term mean (293 mm), and a very consistent rainfall was recorded in September 2012 (Fig. 1b).
The ANOVA results for the tested parameters of Aarslev experiment are reported in Table 2. Significant differences were observed in 7 out of 8 Y × LM management interactions, with the exception of Cbc-lm. The LM factor showed a significant effect in Cbc-lm, while Cultivar (CV) differences were significant only for the total crop aboveground dry biomass and the Cb.
Results of ANOVA for tested parameters in the Aarslev vegetable experiment, Aarslev, Denmark, 2012–2013.
ATC, agronomic tolerance to competition; RBc/RBn, relative biomass for crop/living mulch with weeds; RBT, relative biomass total; Cb, competitive balance; Cbc-lm, competitive balance of crop against the living mulch; Crop, cash crop; lm, living mulch; w, weeds; CV, cash crop genotype; Ha, Hannibal; Ru, Runner; Y, year.
1 LM, living mulch. No weeding subplot data.
2 no LM, control with no living mulch (every third crop row substituted with an empty row); es LM, dyers woad living mulch strips sown 5 weeks delayed after cash crop transplanting; ls LM, dyers woad living mulch strips sown 8 weeks delayed after cash crop transplanting.
3 Standard error of the mean.
4 Mean values in each column followed by a different letter are significantly different according to DMRT (more than two values) or LSD (two values); n.s. = not significant; ***P ≤ 0.001; **P ≤ 0.01; *P ≤ 0.05.
5 Aboveground dry weed biomass.
In the MOVE experiment, LM factor affected all the tested parameters (Table 3). There was a significant effect of CV on the aboveground biomass of cash crop, LM and weeds, while the year affected all parameters with the exception of weeds aboveground biomass, ATC and the Cblm-w. The LM management × CV interaction was significant for Cb and Cbc-lm, while LM × Y interaction was significant for Cbc-lm and Cblm-w. Moreover the 3-factors interaction was significant for Cbc-lm.
Results of ANOVA for tested parameters in the MOVE long-term organic vegetable experiment, Monsampolo, Italy, 2012–2013.
ATC, agronomic tolerance to competition; ATClm, agronomic tolerance to competition with living mulch; Cb, competitive balance; Cbc-lm, competitive balance of crop against the living mulch; Cblm-w, competitive balance of living mulch against the weeds; Crop, cash crop; lm, living mulch; w, weeds; CV, cash crop genotype; ORA1, ORA2, open pollinated locally adapted cultivar; EHF1, Emeraude; Y, year.
1 LM, living mulch; no LM, control with no living mulch; es LM, burr medic living mulch sown at cash crop transplanting; ls LM, bur medic living mulch sown at least 3 weeks delayed after cash crop transplanting.
2 Standard error of the mean.
3 Mean values in each column followed by a different letter are significantly different according to DMRT (more than two values) or LSD (two values); n.s. = not significant; ***P ≤ 0.001; *P ≤ 0.05.
Differences observed for rainfall between the two experimental seasons may have led to biomass differences between years either when considered as a single factor or as interaction with LM or CV.
Effects of LM on plant growth and crop yield
Aarslev experiment
In the Weeding subplots, total leek crop aboveground dry biomass showed significant differences in both the experimental years (Table 4). The ls LM treatment had the greatest leek biomass both in 2012 and 2013 (4.93 and 4.73 Mg ha−1, respectively) and values 112 and 93% higher than the no LM and es LM ones in 2012. On the other hand, in 2013 the es LM significantly increased leek biomass close to the ls LM one (4.53 and 4.73 Mg ha−1, respectively) with higher value than the no LM treatment (1.75 Mg ha−1). Significant difference in crop biomass was observed between the two tested cultivars, with higher values in Ha (3.66 Mg ha−1) than in Ru (3.29 Mg ha−1; Table 2). As far as the Standard no LM is concerned, no crop biomass difference was observed between cultivars in No Weeding (3.01 and 2.83 Mg ha−1 for Ha and Ru, respectively) and in Weeding (6.52 and 5.77 Mg ha−1 for Ha and Ru, respectively) subplots, while significantly higher yield was observed in 2012 (3.22 Mg ha−1) than in 2013 (2.62 Mg ha−1) in No Weeding subplots––data not shown.
Results for LM treatment × year of tested parameters in the Aarslev vegetable experiment, Aarslev, Denmark, 2012–2013.
ATC, agronomic tolerance to competition; RBc/RBn, relative biomass for crop/living mulch with weeds; RBT, relative biomass total; Cb, competitive balance; Crop, cash crop; lm, living mulch; w, weeds; Y, year factor.
1 LM, living mulch treatment. No weeding subplot data.
2 no LM, control with no living mulch (every third crop row substituted with an empty row); es LM, dyers woad living mulch strips sown 5 weeks delayed after cash crop transplanting; ls LM, dyers woad living mulch strips sown 8 weeks delayed after cash crop transplanting.
3 Standard error of the mean.
4 Mean values in each column followed by a different letter are significantly different according to DMRT (more than two values) or LSD (two values); n.s. = not significant; ***P ≤ 0.001; **P ≤ 0.01; *P ≤ 0.05.
5 Aboveground dry weed biomass.
Leek yield showed significant interaction between the LM and the weeding factor (P ≤ 0.01; Table 5). In particular, the es LM and the ls LM yields were not affected whether the weeding was interrupted or not after the LM sowing. On the contrary, the no LM showed a significant reduction (64%) when weeding was not performed. The same reduction (64%) was also observed for the Standard no LM, highlighting no increased detrimental effect of weeds due to the empty row space in the no LM plot. On the contrary, es LM and ls LM treatments did not show difference with the no LM weeded subplots, putting in evidence the absence of a detrimental effect on crop yield of the LM strips introduction. This result is in accordance with previous studies (Ilnicki and Enache, Reference Ilnicki, Enache, Paoletti and Pimentel1992; Båth et al., Reference Båth, Kristensen and Thorup-Kristensen2008), reporting no reduction in crop yield when LM strips were included in the layout. As a consequence, it is possible to hypothesize a low root competition between the ASC and the cash crop as an effect of the delayed LM sowing respect to leek transplanting, which influenced the distribution of the roots of the two species in the intercropping system. Significant difference was also observed with respect to the CV factor (3.09 and 2.68 Mg ha−1 for Ha and Ru, respectively; P ≤ 0.01) and the Y (3.09 and 2.88 Mg ha−1 for 2012 and 2013, respectively; P ≤ 0.05)––data not shown.
Interaction LM × W on leek dry yield and Standard no LM yield in the Aarslev organic vegetable experiment, Aarslev, Denmark, 2012–2013.
1 Dyers woad living mulch strips sown 5 weeks delayed after cash crop transplanting.
2 Dyers woad living mulch strips sown 8 weeks delayed after cash crop transplanting.
3 No living mulch with every third crop row substituted with an empty row (2/3 of Standard no LM density).
4 No living mulch with no crop row substitution (full density).
5 Mean values in each column followed by a different letter are significantly different according to DMRT; n.s. = not significant; ***P ≤ 0.001; **P ≤ 0.01; *P ≤ 0.05.
A significant reduction in neighbour plants (LM and weeds) biomass was observed for es LM and ls LM treatments in 2013 (−72 and 85% compared with 2012 values, respectively), while, in the same year, the no LM (weeds) showed the highest biomass (5.55 Mg ha−1; Table 4). The significant difference in neighbour plant development patterns between the two experimental years was probably due to the opposite rainfall trends recorded in June–July, higher in 2012 than in 2013 and respect to the long-term mean (Fig. 1a). This circumstance could have reduced the LM and weed growth difference among systems in the first cycle. Otherwise, the low rainfall recorded in July–August 2013 could have maximized the competition for water between LM and weeds, which shared the same ecological niches in the field, determining the total neighbour plant biomass reduction of the second cycle. These evidences are in accordance with the goal proposed by Samarajeewa et al. (Reference Samarajeewa, Horiuchi and Oba2006) of replacing an unmanageable weed population with a manageable cover crop (i.e., ASC). Indeed, the introduction of the dyers woad as LM was able to ensure higher Total crop above ground biomass in ls LM in 2012 and in both the LM treatments in 2013 with respect to no LM. As a consequence, our results highlighted how LM introduction could reduce to 8 (ls LM) or 5 (es LM) weeks the critical period for weed control in leek that Tursun et al. (Reference Tursun, Bükün, Karacan, Ngouajio and Mennan2007) considered between 7 and 85 days after transplanting.
MOVE experiment
Total cauliflower aboveground dry biomass differed among the LM treatments: the ls LM produced 6.40 Mg ha−1, which was equivalent to the no LM, with 6.88 Mg ha−1 (Table 3). The es LM, which produced 2.13 Mg ha−1, was significantly lower than the other two treatments. The EHF1 cultivar showed the highest cauliflower biomass, higher than ORA1 and ORA2 cultivars (+59 and 48%, respectively), which were equivalent. The pure-crop stands across all cultivar treatments did not show significant differences for total cauliflower biomass and averaged 7.6 ± 0.5 Mg ha−1, while significant difference (P ≤ 0.05) was found for the pure-crop stand yields: the EHF1 cultivar produced 2.22 Mg ha−1, higher than the ORA1 (1.47 Mg ha−1) and ORA2 (1.49 Mg ha−1) ones, which were similar––data for crop-pure stands not shown. LM also affected neighbour plants aboveground biomass (Table 3). Plots where burr medic ASC was sowed as LM contemporary to cauliflower transplanting (es LM) developed the highest weed biomass (2.27 Mg ha−1), more than 3 times higher than plots where LM sowing were delayed 3 weeks (ls LM). Furthermore, the ls LM showed equivalent weed biomass of the no LM control (where last weeding was 6 weeks after cauliflower transplanting). On the other hand, the es LM showed also higher LM biomass (0.83 Mg ha−1) than ls LM treatment (0.57 Mg ha−1). The cauliflower biomass reduction in es LM is in accordance with many data from the literature that demonstrated the decrease of vegetable crop production due to LM presence (Galloway and Weston, Reference Galloway and Weston1996; Adamczewska-Sowińska and Kolota, Reference Adamczewska-Sowińska and Kolota2010; Kolota and Adamczewska-Sowińska, Reference Kolota and Adamczewska-Sowińska2013). This result is related to both the LM and weed development in the early critical period of cauliflower growth and the reduced weed suppressiveness of the treatment (Zimdahl, Reference Zimdahl2007). Similarly, the equivalent results of cash crop and weed biomass for ls LM and no LM confirmed the effective role of delayed sowing of cover crop at least 21 days after transplanting to minimize cash crop losses (Brainard et al., Reference Brainard, Bellinder and Miller2004; Zimdahl, Reference Zimdahl2007).
Effects of LM on crop-weed competition
Aarslev experiment
All the studied indices of competition, except for RBT, differed across the three LM treatments in the 2 experimental years (Table 4). The low ATC values in the no LM treatment in 2012 (21%), followed by the no LM in 2013 (53%), along with the es LM in 2012 (56%), demonstrated a competitive effect of neighbour plants (weeds in no LM and dyers woad and weeds in es LM) on crop yield, compared with the 2012–2013 ls LM and the es LM one in 2013, which did not show yield reduction compared with pure cash crop. Similarly, the low Cb values for the no LM in 2013 (–1.04), followed by the same treatment in 2012 (−0.68) along with the es LM treatment (−0.73) also demonstrated a less competitive leek crop in these treatments compared with the neighbour plants. The positive Cb values of the ls LM treatment in both years and the es LM one in 2013 put in evidence a positive competitive ability in those conditions against neighbour plants. As far as RBT parameter was concerned, all values ranged between 1 and 2, representing a condition of partial competition and partial complementarity in resource use between crops and neighbours. Complementarity in resource use occurs when competitors (e.g., leek, dyers woad plants and weeds) utilize limited resources (e.g., soil nutrients and moisture) to produce similar biomass in the presence of each other or when grown in pure stands (Snaydon, Reference Snaydon1991; Paolini et al., Reference Paolini, Principi, Froud-Williams, Del Puglia and Biancardi1999; Weigelt and Jolliffe, Reference Weigelt and Jolliffe2003). Despite the RBT results showing the highest value of the ls LM treatment in 2012 (1.87) and the lowest one in 2013 (1.25), more information is provided by differences observed for RBc and RBn. In particular, the ls LM treatment, which had the highest values for ATC and Cb in 2012 as far in 2013 along with the es LM one, also had a RBc value close to 1 (absence of biomass reduction compared with sole crop condition). Moreover in 2013, the es LM and ls LM showed the lowest RBn (0.25 and 0.30, respectively), putting in evidence the strong LM and weeds biomass decrease in leek crop presence in the second experimental season. This result may have been maximized by the lower rainfall recorded after the LM sowing (July–August 2013; Fig. 1a) even though irrigation was performed whenever water deficit was recorded. By comparing the competitive ability of crop against the dyers woad in the crop-LM weeded subplots, the ls LM treatment showed higher Cbc-lm than es LM (0.78 and 0.48, respectively), even if the crop positive competitive ability against LM was observed in both the treatments (Table 2). Competitive ability against neighbour plants also differed between leek cultivars, with Ru exhibiting greater Cb value than Ha (0.10 compared with 0.00).
MOVE experiment
Low ATC and ATC lm values in the es LM treatment, demonstrated a competitive effect of burr medic alone (ATC lm) or together with weeds (ATC) on crop yield, compared with the ls LM. Moreover, no difference was observed for ATC between ls LM and the no LM treatment, confirming a reduced burr medic effect on crop yield in this LM management strategy (Brainard et al., Reference Brainard, Bellinder and Miller2004). Despite no difference for ATC and ATC lm being observed among cultivars, they exhibited different competitive ability between the es LM and ls LM as far as between years (Cb values in Figs. 2a and b, respectively). For all the tested cultivars, low negative Cb values were observed in es LM treatment, while higher competitive ability was shown in the ls LM one (Fig. 2a). The EHF1 in the ls LM showed the highest Cb value, while the Cb of EHF1 in the es LM was not statistically different from those of the other two cultivars in the ls LM treatment and close to 0 value. Moreover, the EHF1 showed the highest competitive ability in the first experimental season, while no difference among the ORA 1 and ORA 2 in 2012 and all the three cultivars in 2013 was observed (Fig. 2b). Significant LM × CV × Y interaction was showed for competitive ability of cauliflower in ‘LM-cash crop mix stand’ (Cb c-lm; Fig. 3), where, in both the years, higher competitive ability for all the cultivar was observed in ls LM than es LM treatment. No difference was observed among cultivars in each LM treatment × Y, while the EHF1 cultivar exhibited the highest Cb c-lm value in 2012 and a significant decrease in 2013. By overlapping the Cb and Cb c-lm results, it is possible to infer that cauliflower was a weak competitor against LM, in particular when LM was sowed at planting. On the other hand, the cultivar choice could modify the competitive relationship with neighbours, maximizing the positive effect by delaying the LM sowing.
Significant LM × Y interaction found for Cb lm-w showing, in 2012, a more effective and strong competitive ability of burr medic against weeds in ls LM than es LM, while no difference was observed for 2013, when both the es LM and the ls LM treatments showed positive values (Fig. 4). In 2012, the burr medic was observed being a weak competitor against weeds in the es LM, probably due to the low rainfall of the first experimental season (Fig. 1b). In such a scenario, the burr medic introduction demonstrated to exhibit weed suppressive potential comparable with standard practice treatment (no LM) when the sowing was delayed 3/4 weeks after cauliflower transplanting. On the other hand, the burr medic significantly reduced cauliflower performances when sowed at cash crop planting, in accordance with the result of De Haan et al. (Reference De Haan, Sheaffer and Barnes1997), who used burr medic as LM in corn.
Effects of living mulch treatment (a) and year (b) on cultivar Cb in the MOVE organic vegetable experiment, Monsampolo del Tronto, Italy. EHF1 = Emeraude cultivar; ORA1 = CRA-ORA 1; ORA2 = CRA-ORA 2; es LM = early sowing living mulch; ls LM = late sowing living mulch; Cb crop = Competitive balance of crop. Bars represent mean values, with different letters significantly different according to the DMRT test per P ≤ 0.05.
Interaction among living mulch treatment, cultivar and year on the competitive ability of crop against the burr medic living mulch in the MOVE organic vegetable experiment, Monsampolo del Tronto, Italy. EHF1 = Emeraude cultivar; ORA1 = CRA-ORA 1; ORA2 = CRA-ORA 2; es LM = early sowing living mulch; ls LM = late sowing living mulch; Cb crop versus lm = Competitive balance of crop against the burr medic living mulch. Bars represent mean values, with different letters as significantly different according to the DMRT test per P ≤ 0.05.
Effect of living mulch treatment on the competitive ability of the burr medic living mulch against the weeds in the different living mulch treatments in the MOVE organic vegetable experiment, Monsampolo del Tronto, Italy. es LM = early sowing living mulch; ls LM = late sowing living mulch; Cb crop versus lm = Competitive balance of crop against the burr medic living mulch. Bars represent mean values, with different letters as significantly different according to the DMRT test per P ≤ 0.05.
Effects of LM on weed community and weed coverage
Aarslev experiment
In both the years of the experiment, no difference was observed for main species presence across the LM treatment at harvest. The main weed species were: Chenopodium album L. (common lambsquarters), Stellaria media L. (Common chickweed), Capsella bursa-pastoris L. (shepherd's-purse), Polygonum spp. and Veronica spp. The Poa annua L. (meadow bluegrass annual) was observed as main species only in the 2012 season.
MOVE experiment
Weed coverage was higher in the es LM than in the other two treatments in both the head emergence and harvest phases in 2012 as well as in 2013 (Table 6). Ten weed species were present at different levels in this experiment. Identified species were representative of common weeds in vegetable crop production area in central Italy (Pignatti, Reference Pignatti1982). Main species included Amaranthus retroflexus L. (redroot pigweed), Chrysanthemum segetum L. (marigold corn), Raphanus raphanistrum L. (radish wild) and Stellaria media L. Weed species abundance differed among LM treatments, especially during the second year, when the es LM treatment contained high coverage of all the species observed in this treatment except for the Portulaca oleracea L. (common purslane). This result was attributable to the greater rainfall that occurred in the first phases of cauliflower crop during the second year (Fig. 1b). In comparison a larger number of species of weeds developed in the ls LM treatment, but without an apparently dominant one. A similar result was also obtained in no LM treatment, where weed control was performed until 6 weeks after the cauliflower transplanting. Based on these results, we speculate that the burr medic sowing delay can contribute to significantly reduce the weed development, although without modifing the weed species presence.
Weed species and coverage estimates by species in the MOVE long-term organic vegetable experiment, Monsampolo, Italy, 2012–2013.
1 Living mulch treatment.
2 Coverage index: 5 = 75–100%; 4 = 50–74%; 3 = 25–49%; 2 = 5–24%; 1 = 1–5%; + = <1%; r = sporadic presence.
3 Mean values in each column followed by a different letter are significantly different according to Mann–Whitney test n.s. = not significant; **P ≤ 0.01; *P ≤ 0.05.
Conclusions
Managing weeds in organic farming is challenging and requires the use of many techniques and strategies to achieve ecological sustainability as well as economically acceptable weed control and crop yields. Our study demonstrated the potential for introducing LM to reduce weed emergence and development as well as their detrimental effect on cash crop performance, thus limiting the need for mechanical cultivation to control weeds in organic vegetable systems, under different climates.
In particular, the substitutive approach (LM strips replacing every third row of cash crop) was observed to give a consistent reduction in critical period for weed control in a vegetable cropping system in Northern Europe, offering a trade-off to the crop yield reduction due to the lower crop density than in standard systems. Similarly, in Mediterranean conditions, the delayed sowing of LM, when broad-sowed, was found to be effective in reaching the double objective of reducing the critical period for weed control and avoiding competitive effect of LM versus the cash crop. In addition, both the F1 hybrids and open-pollinated local adapted cultivars highlighted tolerance to the LM introduction, eventhough the F1 hybrids showed the highest suitability. Future experiments aiming to compare different LM species and mixture of species should be promoted, seeking to provide alternative LM for cash crops in different environments, while also analyzing the potential allelopathic effect of ASC species on both cash crops and weeds.
Acknowledgements
The authors wish to thank Astrid Bergmann, Fabio Fusari, and Aldo Bertone for their contributions in field operations, soil and plant sampling, and analysis. This study has been carried out in the frame of the InterVeg research project: Enhancing multifunctional benefits of cover crops––vegetables intercropping (Core Organic II ERA-NET).
