Previous PageTable Of ContentsNext Page

The effect of faba bean plant population on yield, seed quality and plant architecture under irrigation in southern NSW.

PW Matthews1, EL Armstrong2, CJ Lisle2, ID Menz2 and PL Shephard2 and BC Armstrong2

1NSW Department of Primary Industries, Temora Agricultural Research and Advisory Station, Temora, NSW
NSW Department of Primary Industries, Wagga Wagga Research Institute, Wagga Wagga, NSW


The effect of plant population on the grain yield, seed quality and plant architecture of faba beans (Vicia faba) was evaluated in the Murrumbidgee Irrigation Area of southern NSW in 2006. The faba bean cultivars Farah, Fiesta VF and Nura were sown under irrigated conditions at target plant populations of 10, 15, 20, 25, 30 and 35 plants/m2. To understand how plant density influenced yield, plant samples were collected and mapped for branch number, height, location of flowering nodes, pod position, pod set, seed number per pod and seed size. Statistical analysis of grain yield showed all varieties responded similarly to changes in plant density, although yields were significantly different between varieties. Pod mapping data showed yield architecture differs with plant population. From the fitted yield model, optimal grain yield was achieved from targeting 22-24 plants/m2 for the varieties Farah, Fiesta VF and Nura.

Key Words

Grain legumes, sowing rate, Vicia faba, yield components


The area sown to faba bean in southern NSW has been increasing over the last 10 years (excluding the severely droughted years of 2006 and 2007), largely due to expansion into southern irrigation districts. Here, stable high yields are possible even under drought conditions due to the availability of supplementary water and because the crop is ideally suited to the heavier textured soils that predominate in the region. The crop has also gained popularity with the release of superior varieties, particularly those with improved disease resistance to the main faba bean diseases of southern NSW - chocolate spot (Botrytis fabae) and ascochyta (Ascochyta fabae).

Despite the availability of irrigation water, grain yields can fluctuate widely, between 2.5 t/ha - 7 t/ha. While there are undoubtedly several significant management issues behind these fluctuations, many growers and agronomists recognize most agronomic studies in the past have focused on dryland production. For example, while most irrigated growers target plant populations between 15-20 plants/m2, dryland research shows 30-35 plants/m2 are needed to maximise yield under these conditions (Matthews et al 2001). Also, irrigated faba beans can suffer from increased plant lodging and we do not fully understand the impact of plant population on this. Plant population can also affect canopy development, plant architecture and distribution of pods

The influence of plant population on faba bean yield has been well documented across a range of Australian environments including southern and central NSW (Matthews et al 2001), Western Australia (Loss et al 1998), South Australia (Adisarwanto and Knight, 1997), northern NSW (Marcellos and Constable, 1986) and irrigation areas in northern Victoria (Drew, 1994). Most of these studies focused on dryland systems using older varieties such as Fiord with very few looking at the impact of plant population on yield, yield components and seed quality of irrigated crops, especially in southern NSW. Previous studies found optimum plant density for dryland faba bean varied between 20-45plants/m2 depending on factors such as sowing time and seasonal moisture.

We undertook this study to fill these gaps and to provide agronomists and growers with information on optimum plant densities for modern faba bean cultivars under irrigated conditions in southern NSW. Mapping pods, nodes and heights of stems helped explain these yield responses to changing density and to help understand the resulting differences in plant architecture.


The faba bean cultivars Farah, Fiesta VF and Nura were sown on irrigated beds at targeted plant populations of 10, 15, 20, 25, 30 and 35 plants/m2. Plots were 10 metres long and two rows wide to suit 1.83m raised beds. Fiesta VF and Farah are 7-10 days earlier flowering than Nura. Two one-metre rows were cut at physiological maturity from each plot to measurer dry matter production and for mapping nodes and pods along the stem. Ten stems from each cut were taken for these studies.

Site details

The site was located south east of Griffith on a grey cracking clay soil (Table 1) that had been formed into 1.83m raised beds. The surrounding paddock was sown to a commercial crop of Fiesta VF faba bean.

Table 1. Soil analysis results for plant population trial site in 2006.

Soil property

0-10 cm

10-60 cm

EC dS/m



pH (CaCl2)



Colwell Phosphorus (mg/kg)



Cation Exchange Capacity cmol(+)/kg



Calcium/Magnesium Ratio



Sodium Percentage (ESP) %



Fallow rainfall was 43 mm (January - April) with in crop rainfall 119.2 mm (May - November). The trial site was pre-irrigated on 1st April with 1.5 ML/ha of water, and subsequently irrigated on 27th August, 20th September and 6th October with 0.73 ML/ha, 0.68 ML/ha and 0.61 Ml/ha water respectively.

Experimental Design, Analysis and Data handling

The trial was designed in 6 ranges (columns) by 9 rows with 3 replicates with treatments blocked in 2 directions. Varieties were blocked within the intersections, with the plant densities randomised within these blocks.

All analyses were performed using ASREML, a mixed model analysis program (Gilmour et al. 2002). Traits were analysed using mixed model cubic smoothing spline analyses (Verbyla et al. 1999). The use of non-parametric regression, via cubic smoothing splines, enables the fitting of nonlinear relations without necessitating assumptions on the shape of the relation. This method incorporates the splines into a mixed model framework, enabling their use in combination with other design and treatment factors in complex experiments. After preliminary inspection of the data, knot points for the density spline were chosen to be 4, 10, 20, 30, and 43 plants/m2. Random components are shown in bold (see formulae below). The linear and non-linear (spline) effects of crop density are denoted ‘lin(density) ’ and ‘ spl(density)’ respectively. ‘Cultivar’ is a factor representing the effects of cultivar. The design stratum component comprises factors that are necessary to create the strata inferred by the design.

Trait ~ cultivar + lin(density) + cultivar:lin(density) + spl(density) +cultivar:spl(density) + {design stratum} + error


Predicted responses of yield to increasing plant density are shown in Figure 1. The response curvatures of all varieties were similar, reaching maximum yield at densities between 22-24 plants/m2. Yield of Farah and Fiesta VF were similar and significantly higher than Nura.

Dry matter responded to increasing plant density in a more linear fashion compared to yield but the response curvatures across varieties were significantly different (Figure 1). Nura showed a much flatter response compared to both Farah and Fiesta VF and consequently produced significantly less dry matter at higher densities. Plant heights of all varieties were unaffected by plant density (Table 2), however Nura was

significantly shorter than Farah and Fiesta VF. In contrast, height to lowest fertile pod did increase significantly (and with a similar curvature) as plant density increased across all varieties (Figure 3).

Seed weight per plant decreased significantly (but with similar curvature) as plant density increased, (Figure 1). For all three cultivars, 100 seed weights increased significantly as plant density increased (Figure 2). While response curvatures of all varieties were similar, 100 seed weights of each cultivar were very different (mean 69.9g, 66.3g and 62.5g per 100 seeds for Farah, Fiesta VF and Nura respectively).

Figure 1. Fitted grain yield, total plant drymatter and total seed weight response curves (solid line) of Farah, Fiesta VF and Nura across the range of plant populations tested, Griffith 2006. (Confidence intervals shown with broken lines).

Figure 2. Fitted seed weight response curves (solid line) of Farah, Fiesta VF and Nura to increasing plant population. (Confidence intervals shown with broken lines).

Figure 3. Relationship of height to the lowest fertile pod and changes in plant population of three cultivars of faba bean. (Confidence intervals shown with broken lines).

Table 2. Measured plant densities and stem height for each of the cultivars and target plant populations, Griffith 2006


Fiesta VF



Target Plant
Population (plants/m2)

Plant Density


Average Stem

Plant Density


Average Stem

Plant Density


Average Stem











































Table 3. Statistical analysis and significance levels# of measured traits


Grain Yield

Total Plant Drymatter

Total Plant Seed Weight

100 Seed Weight

Height to 1st Flower

Height to 1st Fertile Pod

Stem Height

















Var X Den












Var X spl(Den)


# Significance levels: *** - 0.001; ** - 0.01 and * - 0.05


Under irrigated conditions in southern NSW the optimal plant population for this study was 22-24 plants/m2 across all three cultivars tested. These finding are in general agreement with earlier studies by Adisarwanto and Knight (1997) and Marcellos and Constable (1986) who concluded 20-35 plants/ m2 was the optimal density range across the eastern states with the lower densities more appropriate under more favourable conditions. However, these densities are below those suggested by Matthews et al (2001) for dryland farming systems in southern NSW. Here, 30-35 plants/m2 were optimal across a range of varieties and seasonal conditions from drier to wetter regions. However both the current study and that of Matthews et al (2001) show similar response curves to increasing plant density for all tested varieties with only overall genetic yield potentials varying. The significant difference in yield potential between Farah/Fiesta VF and Nura in this study is supported by the 7 year across-sites analysis for irrigated trials in southern western NSW showing Nura to yield 5 and 6% below that of Fiesta VF and Farah respectively (McRae et al 2008).

The significant increase in dry matter production at physiological maturity as plant density increases agrees with the previous studies by Loss et al (1998) and Adisarwanto and Knight (1997) using the cultivar Fiord. The divergence of dry matter production between the cultivars Farah and Fiesta VF and the cultivar Nura as plant density increases may be explained by the genetically different backgrounds of the cultivars and the significantly lower plant height expressed by Nura. The reduction of total seed weight per a plant as plant density increases is supported by the previous studies mentioned, however Loss et al (1998) found that increases in plant numbers at higher densities more than compensated for a lower seed weight per plant, effectively producing higher yields per hectare. However in this study the predicted yield response curve peaked and turned as plant density increased above the optimum range. In the experiments of Loss et al (1998), yield response curves continued to rise as densities reached the upper values used (60 plants/m2).

Seed weights increased as plant densities increased and this conflicts with the studies of Loss et al (2001) and Adisarwanto and Knight (1997) who reported mean seed weight to be largely unaffected by sowing rate in both Western Australia and South Australia respectively. Height to the lowest pod increased by around 13 cm as plant density increased from 10 to 35 plants/m2. This finding is supported by Loss et al (1998) who showed increases in height to the lowest pod of about 8 cm as seeding rate increased from 70 - 270 kg/ha.


The yield model used here suggests that the optimum plant population for irrigated faba bean grown under similar conditions in southern NSW to be 22-24 plants/m2, which is below the previous population targets suggested by earlier work targeting dryland farming systems in southern NSW. However, the current grower practice of targeting 15-20 plants/m2 under irrigated conditions appears below the optimum and growers could well be loosing yield if conditions were similar to this study.

The trend shown here of lower seed weight at lower plant density is a concern should growers achieve only low plant populations and minium receival standards for seed size were set by marketers. The lowing of first pod height with reductions in plant densities is of practical concern at harvest. However heights of first pods recorded in this irrigated study are well above minimum cutter bar heights.


Adisarwanto T. and Knight R. (1997). Effects of sowing date and plant density on yield and yield components in the faba bean. Aust. J. Agric. Res. 48, 1161-8.

Drew N. (1994). Irrigated Faba Bean Growers Handbook (Victorian Department of Agriculture).

Gilmour AR, Cullis BR, Welham, SJ. and Thompson R. (2002). ASReml Reference Manual 2nd ed. Release 1.0. NSW Agriculture Biometrical Bulletin 3. Orange, NSW, Australia: NSW Agriculture.

Loss SP, Siddique, KHM, Jettner, R and Martin, LD. (1998). Response of faba bean (Vicia faba L.) to sowing rate in south-western Australia I. Seed yield and economic optimum plant density. Aust. J. Agric. Res. 49, 989-97.

Loss SP, Siddique KHM, Martin LD and Crombie A. (1998). Response of faba bean (Vicia faba L.) to sowing rate in south-western Australia II. Canopy development, radiation absorption and dry matter partitioning. Aust. J. Agric. Res. 49, 999-1008.

Marcellos H and Constable GA.. (1986). Effects of plant density and sowing date on grain yield of faba beans (Vicia faba L.) in northern New South Wales. Aust. J. Exp. Agric. 26, 493-6.

Matthews PW, Carpenter DJ, Smith A and Fettell N. (2001). Faba bean seeding rates for central and southern NSW. Proceedings of the 10th Australian Agronomy Conference, Hobart. Australian Society of Agronomy.

McRae FJ, McCaffery DW and Matthews PW (2008) Winter Crop Variety Sowing Guide 2008, NSW Department of Primary Industries, 74-80

Verbyla AP, Cullis BR, Kenward MG and Welham SJ. (1999). The analysis of designed experiments and longitudinal data using smoothing splines (with discussion). Applied Statistics 48, 269–312.

Previous PageTop Of PageNext Page