Genetic diversity among North African faba bean landraces for competitive ability against weeds

Genetische Diversität Nordafrikanischer Ackerbohnen-Landrassen für Konkurrenzkraft gegen Unkraut

Faba bean production is adversely affected by weeds, especially during the early growth stages (Frenda et al., 2013). Weeds compete with the crop for limited resources, such as nutrients, water and light. They can cause heavy yield losses and decrease yield quality (Naylor, 2008). Weeds were estimated to cause seed yield reduction in faba bean of up to 60% (Frenda et al., 2013; Ghaouti et al., 2016).

Farmers in traditional agroecosystems in Morocco very rarely use chemical and/or physical control options for weed management. Weeding is very rarely practiced because of high labor costs (Alaoui, 2007). Hence, the utilization of varieties that can tolerate weeds or reduce weed growth and seed production could, therefore, be a relevant, integrated weed-management strategy (Fisher et al., 2001; Lemerle et al., 2001a). The genetic improvement of competitive ability is an attractive environment-friendly weed-management option for both low and high-input cropping systems (Cousens and Mokhtari, 1998; Dingkuhn et al., 1999).

One can select for competitive ability directly in the presence of weeds or indirectly under non-competitive conditions, based on secondary traits related to WSA. A large number of studies have been conducted to determine the competitive ability of crop species and the association of crop competitive ability with specific morphological and crop-growth traits (Place et al., 2011; Zystro et al., 2012; Korres et al., 2016). In crops such as wheat (Triticum aestivum L.) and rice (Oryza sativa L.), relationships between crop competitive ability and plant traits have been reported, e.g., plant height, tiller number, speed of development, seed size and leaflet (Lemerle et al., 2001b; Zhao et al., 2006b; Saito et al., 2010).

Variability among cultivars in their ability to compete with weeds has been reported in many crops: wheat (Lemerle et al., 1996; 2001a; 2001b), rice (Johnson et al., 1998; Caton et al., 2003; Rodenburg et al., 2009; Saito et al., 2010), corn (Zea mays L.) (Zystro et al., 2012), and soybean (Glycine max L.) (Place et al., 2011). However, very few studies have evaluated the same in faba bean (Ghaouti et al., 2016) and none has addressed the competitive ability of faba bean landraces against weeds. Most of the faba bean breeding programs have concentrated on improving yield, disease resistance, drought tolerance and resistance against broomrapes, with little consideration given to tolerance to weed stress and to crop’s ability to suppress weeds.

The Moroccan local faba bean landraces contain a great deal of genetic diversity for morphological, agronomic and physiological traits (Sadiki et al., 2002; Sadiki and El Bouhmadi, 2002; Belqadi, 2003). Therefore, the objectives of this study were (1) to evaluate the weed compe­titive ability of Moroccan faba bean landraces and (2) to identify superior local landraces that are suitable for direct use on farmers’ fields and/or for use in breeding programs.

Materials and methods

The size of each subplot was 6.0 m² (four 3-m long rows, with an inter-row spacing of 50 cm). Mixed plots (weedy, i.e. with S. alba) and pure plots (weed-free) had a sowing density of 27 faba bean seeds/m², with the addition of 550 seeds/m² of white mustard in the mixed plots. First, faba bean seed was sown manually at a depth of 10 cm; white mustard was sown as second step at a depth of 2 cm in the same row. The model weed and faba bean were hence grown as additive mixture following the ideas of Snaydon (1991) and Gibson et al. (1999). The trials were sown between November and December and harvested between May and June.

The analysis of variance (ANOVA) was performed among the 60 local populations. The two non-landrace checks were excluded from ANOVA because they may not fully represent the envisaged basic pool of local faba bean landraces. PLABSTAT v.2H (L) (Utz, 2005) was employed for statistical analysis using the following linear model:

Y_ijk = µ + e_i + w_j + (we)_ij + g_k + (gw)_jk + (ge)_ik + e_ijk;

where Y_ijk is the observation of the genotype k in environment i and the weed treatment j; μ is the general mean, e_i is the environmental effect, w_j is the weed treatment effect assigned to the main plot, (we)_ij is the main plot error, g_k is the genotype effect assigned to the subplot, (gw)_kj is the interaction between the genotype k and the weed treatment j, (ge)_ik is the interaction between the genotype k and the environment i, and e_ijk is the subplot error. The four location-year combinations were modelled as four environments. Genotypes were tested for significance against the G × E interaction. As the experiment was conducted with one replicate per environment, the single plot error variance could not be explicitly estimated. Instead, it became part of the G × W × E inter­action variance. This term, given as e_ijk in the linear model, is the error term for testing the G × W interaction for significance.

Correlation coefficients were calculated to explore relationship between the grain yield of the 62 bean genotypes, their different indices (see below) and agromorphological traits measured in weed-free conditions, based on genotype’s means across environments. The hypothesis is that highly competitive faba beans had specific agronomic traits, which distinguished them from those that were less competitive. Taking this line of reasoning, such agronomic traits were observed under weed-free conditions.

The yield loss attributable to the competition between faba bean genotypes and the white mustard was calculated at each environment using the following equation:

YL_i (%) = [(Y⁰_k – Y¹_k)/Y⁰_k] × 100 = [1 – (Y¹_k/Y⁰_k)] × 100

where Y⁰_k is the grain yield of the faba beans (k) in the weed-free condition and Y¹_k is the grain yield of the faba beans (k) in weedy condition.

Mean productivity (MP_k) was measured in each of the four environments using the following equation:

MP_k = [Y⁰_k+ Y¹_k]/2

To directly compare the relative performance of the faba beans with the relative performance of the white mustard in the weedy plots, the so-called weed competitive index (CI) as used by Langeroudi and Kamkar (2009), was employed, using the following equation:

CI_k = [Y¹_k/Ymean]/[W_k/Wmean]

where Y¹_k is yield of the faba bean genotype (k) under the weedy condition, Ymean is the mean yield of all the faba bean genotypes in the presence of weeds in each environment, W_k is the grain yield of white mustard, when grown mixed with the faba bean genotypes (k) and Wmean is the mean yield of white mustard in all mixed plots in each environment.

Results

In weedy plots, the 60 faba bean genotypes and the two checks matured, on average, 3 days earlier. They showed an increase in plant height by about 14% and a reduction in tiller number by about 37% under weedy conditions (Table 4). The pod size was reduced under weedy conditions by about 13%. Pod number per main stem, at secondary stem, and per plant decreased under weedy conditions by about 39%, 80% and 57%, respectively. Seed number per plant decreased by about 66%, seed number per pod by about 23% and TSW decreased by 13% due to presence of S. alba (Table 4).

Average grain yield decreased by 69% because of weeds. Although to different extents, the grain yield of all faba bean local landraces and the two checks decreased under weedy conditions, except for Pop.2 (Fig. 1). In weedy plots, the average grain yield varied from about 0.5 t/ha to 1.5 t/ha with an average of nearly 0.9 t/ha (Table 4). In these conditions, the lowest faba bean grain yields were recorded for Pop.10, Pop.16, Pop.13 and the check Fouila Defes. The local landraces Pop.48 and Pop.54 exhibited the highest grain yield. On the other hand, there was a greater range in grain yield in weed-free plots from about 1.2 t/ha for Pop.2 to 5.0 t/ha for Pop.38. The local landraces Pop.38, Pop.47, Pop.49, Pop.17, Pop.48 and Pop.57, had a relatively high yield both under weedy and weed-free conditions (Fig. 1).

Fig. 1. Relationship of yield of the 60 faba bean local landraces and the two checks with vs. without whi­te mustard as model weed. Symbols in grey present best genotypes in weedy and weed-free conditions (cf. text); Pearson (r) and Spearman (r_S) coeffi­cient.

Using the average of the four environments, the yield losses of the 60 faba bean genotypes and the two checks varied from 7% to 81% with an average of 69% (Table 5). Pop.2 and Pop.15 obtained the lowest and the highest yield loss, respectively (Fig. 2 and 3). The average mean productivity of the 60 genotypes and the two checks in the four environments varied from about 1.2 t/ha to 3.1 t/ha with an average of about 1.9 t/ha (Table 5; Fig. 2). The check Fouila Defes presented the lowest mean productivity whereas Pop.38 had the highest mean productivity (Fig. 2).

Fig. 2. Relationship between yield loss and mean produc­tivity of the 60 faba bean local landraces and the two checks; Pearson (r) and Spearman (r_S) coeffi­cient.

Fig. 3. Relationship between yield loss and the weed com­petitive index of the 60 faba bean local landraces and the two checks; Pearson (r) and Spearman (r_S) coefficient.

The weed competitive index ranged from 0.24 to 3.10 with an average of 1.28 (Table 5). Pop.4 obtained the lowest weed competitive index, whereas Pop.53 presented the highest competitive index (Fig. 3), resulting from both (1) its high grain yield under weedy conditions (1.1 t/ha) compared to the average (0.9 t/ha) (Fig. 1) and (2) its pronounced ability to reduce the grain yield of white mustard from an average of 0.6 t/ha down to 0.3 t/ha; as comparison, Pop.49 achieved a similar CI, combining even higher yield under weed stress with a weaker weed suppressive ability (details not shown).

Fig. 4. Average white mustard grain yield when it was grown in mixture with one of the 60 faba bean local landraces or with one of two checks.

A positive relationship was found between weed-free grain yield and weedy grain yield of the 62 bean genotypes (r = 0.453**). When selecting the 12 best genotypes, according to either weed-free or weedy grain yield, these two groups of selection share six common local landraces (cf. Fig. 1).

Yield loss was correlated to weed-free grain yield of the faba bean landraces and the two checks (r = 0.435**). Genotypes with good grain yield performance in weed-free conditions tended to lose more yield due to weed. Yield loss was also positively correlated with plant height (r = 0.264*), tiller number (r = 0.268*) and seeds per plant (r = 0.353**) under weed-free conditions. Tall and highly tillering faba beans tended to lose more yield in competition with S. alba.

Genotypes with high mean productivity were characterized by high yield component performances: large pods, many tillers and many seeds per plant, seeds per pod and high TSW.

The weed competitive index of faba bean landraces and the two checks was negatively correlated with faba bean yield loss (r = –0.371**) and with yield of the model weed S. alba (r = –0.550**), reflecting the algebra of this index. No relationship was found between the yield of the model weed S. alba and the other traits recorded.

In the present study, maturity was not found to be an indicator of the level of tolerance to weed stress. In contrast, Huel and Hucl (1996) found that increased weed tolerance was associated with early maturity in wheat.

Tiller number of faba bean genotypes was reduced in the presence of S. alba. Such tiller number reductions have been demonstrated in wheat, barley and oat cultivars (Andrew et al., 2015). Lemerle et al. (1996) reported that high tiller number was associated with low yield reduction for wheat varieties. In the present study, no significant difference between faba bean local landraces was recorded for the number of tillers. Moreover, the genotypes did not interact significantly with the presence or the absence of weed stress for their number of tillers.

The variance of grain yield among faba bean local landraces was higher under weed-free conditions compared to weedy conditions. Indeed, the standard deviation among faba bean genotypes was 0.66 t/ha under weed-free conditions versus only 0.23 t/ha in weedy conditions; yet, the corresponding coefficients of variation were quite similar (Table 4). A smaller variation under stress is in agreement with Zhao et al. (2006b) who found such pattern for aerobic rice cultivars. Other authors observed that the genetic variation was higher under weedy than under weed-free conditions for faba bean (Ghaouti et al., 2016) and for lowland rice varieties (Rodenburg et al., 2009).

A positive relationship was observed between grain yield of faba bean local landraces and the two checks under weedy and weed-free conditions (r = 0.453**, Fig. 1). Strong positive correlations between weed-free grain yield
and weedy grain yield were observed for wheat (Cousens and Mokhtari, 1998; Lemerle et al., 2001b) and for faba bean (Ghaouti et al., 2016). Conversely, Langeroudi and Kamkar (2009) found no relationship between grain yield under weed-free and weedy conditions, when canola (Brassica napus) was grown in the presence of wild mustard (Sinapis arvensis). Although the correlation between grain yield under weedy and weed-free conditions was significant, it was not very high (r = 0.453**; Fig. 1). Highest yielding genotypes in weed-free conditions were not necessarily highest yielding in weedy conditions (e.g. Pop.52; Fig. 1).

Average yield loss ranged from 7% to 81% (Table 5). Frenda et al. (2013) reported a mean yield reduction of 60% in faba bean when it was left to compete with wild mustard and Italian ryegrass as artificial weed species. Yield loss of faba bean local landraces and the two checks was positively correlated with their yield performance in the absence of white mustard. This is in agreement with the findings of De Lucas and Froud-Williams (1994) who found that wheat varieties which produced the greatest grain yield in weed-free plots were most severely affected by weed competition. Ghaouti et al. (2016) reported that yield loss of faba bean genotypes because of the competition with weeds was negatively correlated with their yield performance in the absence of weeds. This result was clearly because of different groups of genotypes with different inbreeding levels that were collected into a joint analysis (F1 hybrids, pure lines and partly heterotic genotypes). Other authors reported that no relationship was observed between the level of grain yield in optimal conditions and yield loss for wheat (Cousens and Mokhtari, 1998; Lemerle et al., 2001b; Coleman et al., 2001) and barley (Didon and Boström, 2003).

Genotypes with high mean productivity tended to have higher yield losses in presence of the model weed S. alba (Fig. 2). Pop.2 was an outlier and the coefficient of correlation was recalculated excluding this landrace. As a result, a non-significant correlation was found (r = 0.179, details not shown). This indicates that both indices are independent and are providing complementary information on the level of mean productivity versus the level of weed tolerance. Hence, it should be possible to find genotypes that combine both high yield and high weed tolerance. Pop.48 and the variety Aguadulce correspond to this objective.

No significant relationship was found between the weed-free grain yield of faba bean and the yield of white mustard, when grown in mixture (r = –0.006, Table 7). Highest yielding faba bean genotypes under weed-free conditions were not necessarily the most suppressive genotypes of S. alba. No significant relationship was found between the grain yield of the model weed S. alba in mixtures and the yield loss of faba beans (r = 0.027; Table 7). This is in agreement with the findings of Langeroudi and Kamkar (2009).

A highly significant, negative relationship was found between the yield loss and the CI (r = –0.371**; Table 7). Yet, this association is far from complete. Some faba beans (e.g. Pop.45 and Pop.23) showed in spite of low yield loss rather low CI values (Fig. 3), in these cases because these two populations allowed very high seed yield to S. alba (Fig. 4). The coefficient of correlation was recalculated between the yield loss and weed competitive index excluding the Pop.2. The coefficient of correlation remained nearly the same, yet significant only at P = 0.05 (r = –0.323*, details not shown). The negative relationship between the competitive index and yield loss seems rather to be a general trend of the studied genotypes set.

The present study did not show any significant relationship between the weed-free grain yield Y⁰_k and the weed suppressive ability. This is true, whether the focus is on faba bean-influenced weed yield directly (W_k) or on the more complex parameter CI: r (Y⁰_k, W_k,) = –0.006 and r (Y⁰_k, CI) = 0.226 (Table 7). Weed-free yield was not an indicator of the suppressive ability of a genotype, corroborating Langeroudi and Kamkar (2009; canola cultivars competing with wild mustard).