Previous PageTable Of ContentsNext Page


J-François Dejoux1, Frédéric Ferré1 and J-Marc Meynard1

1 Unité d'Agronomie INRA-INAPG – BP 01 – F-78850 Thiverval Grignon – France
E-mail :


'Pre-emergence' herbicides are generally used on rapeseed crops because 'post-emergence' herbicides are of very limited efficacy. As it is necessary to reduce herbicide use for environmental and economic reasons, it is important to increase and predict situations where rape is competitive enough to control weed emergence, growth and reproduction. In experimental plots, we studied the effects of sowing dates and nitrogen level in autumn and their interaction on the number, growth, height and yield of rape and weeds. Six weed species, 2 summer annuals and 4 biennials, were sown separatly, in mixture with the rape crop.

Summer annual weeds (such as Amaranthus reflexus L.) emerged almost only in very early sowing dates (beginning of august) and not at normal sowing dates (beginning of september) whereas biennials weeds (especially winter annuals) had a longer emergence period, whith more plants emerging in autumn than in the end of summer.

For both sowing dates, nitrogen effect was predominant. With high N availibility, rape growth limited weeds emergence, survival and individual growth of biennials, leading to clean plots after winter. On the contrary, with low N availibility, both sowing dates were infested by biennials, which could produce grains before rape harvest. Summer weeds were higher than rape in autumn, even with high N availibility and summer weeds produced grains before they were killed by winter frost. However, the risk of seed bank increase by summer weeds needs to be verified with viability tests.

We found the same results on a farmer field-trials network. Therefore, we conclude that it is possible not to use pre-emergence herbicides if there is enough nitrogen available in autumn and sufficient rape growth.

KEYWORDS. competition, weed emergence, weed growth, Amaranthus reflexus L., Geranium dissectum L.


'Pre-emergence' herbicides are generally used on rapeseed crops because 'post-emergence' herbicides are of very limited efficacy. As it is necessary to reduce herbicide use for environmental and economic reasons, it is important to increase and predict situations where rape is competitive enough to control weeds without herbicides. This may allow a reduction in 'Pre-emergence' herbicide use and make a 'post-emergence' strategy more realistic, which is environmentally more acceptable. The hypothesis is based on a better growth and LAI development of rape in autumn and early spring, which may reduce emergence, growth and grain production of weeds.

In this paper, we present results from a first year of field experiments on rapeseed sowing dates and nitrogen level in autumn and their interactive effects on emergence, growth and grain production of 4 weeds, as measured in monospecific mixture.

Matériels et méthodes

Twenty experimental treatments, with 5 replicates in a randomized block lay-out, were setup by complete crossing of 3 factors: rape and weed sowing dates, sowing applied N and weed species. The 2 sowings dates were on 1 August (Sowing 1) and 1 September (Sowing 2) and the 2 sowing N rates were: 0 (N0) et 200 (N200) kg N ha-1. The 5 "weed modalities'' were: control without weeds, Amaranthus reflexus L., winter barley (Hordeum vulgaris L.), Stellaria media L. and Geranium dissectum L.

Field experiments were conducted in Grignon in the Paris Basin (France, 48.5° latitude North) in 1997-1998 on a deep loamy soil. Except for Amaranthus which was spontaneous, weeds were sown with winter oilseed rape (Brassica napus cv Goëland) between 1 and 2 cm deep, and nitrogen was applied on soil surface. Number of sown grains per m² was 70 for rape and 130 for weeds. In spite of germination tests, the level of sown weeds dormancy was not completely controlled. Manual weeding was performed to keep plots with only one weed species. Crop was irrigated in autumn to ensure good germination of weeds and rape, but not in spring. The crop was fully protected against pests and diseases and spring N was applied for a reasonnable yield target. Samplings were made 3 times in autumn for sowing 1 and 2 for sowing 2, and 2 times in spring: once at mid-stem elongation and once 6 weeks before harvest. Soil mineral N, plant number, stages and height were measured for rape and weeds as well as root and aerial accumulated dry matter and N, green leaf area, grain yield and grain number at maturity.

Résults and discussion

Effect of rapeseed sowing date on weed population density

Advancing the sowing date has induced an almost specific emergence of Amaranthus, a summer weed, whereas it has conversly reduced biennials weeds emergence (Figure 1). Because of biological and thermal requirements, Amaranthus germinate preferentially in spring and summer, therefore they rapidly lost their ability to germinate between early August (Sowing 1) and early September (Sowing 2).

Figure 1. Weed number in autumn in N0. For each weed and treatment, the maximal value measured in autumn has been retained. Vertical bars represent the standart deviation.

On the contrary, as Stellaria and Geranium require rather low temperatures to germinate (Jauzein, 1995), few plants had emerged until mid-September, but more until end of November. Leaf cover, which was already well established in Sowing 1 when these weeds could germinate (Table 1), has also probably limited these weeds emergence because of shade (Béres et Sardi, 1998).

The treatment ''winter barley'' was intended the effect of the preceding crop's voluteer growth on rapeseed. Therefore we sowed grains which had just been harvested on the same field before rape was sown. For the first sowing, some grains were probably still in dormancy, which induced a much lower emergence rate than for sowing 2. The stage of dormancy of different weeds according to sowing date would have to be investigated under different conditions because numerous factors interact.

Table 1. Rape leaf area index and weed grain number for the 4 treatments. Grain number of Amaranthus was measured on 22 October and 6 weeks before harvest for Geranium, according to flower numbers, assuming that each flower contains 5 grains.


Sowing 1

Sowing 2






Rape leaf area index in control

- on 1 September

- on 6-14 October









Grain number per m² of


Amaranthus reflexus L.





Geranium dissectum L.





Short cycle weeds control

Amaranthus had systematically disappeared at the end of winter. The risks of spring growth and yield reduction of rapeseed seems to be low because of its large capacity to recover in spring. It is therefore possible to spare herbicides with these weeds if they have not produced significant amounts of grains before dying, as it has been for winter barley.

Sowing date was of major importance on Amaranthus dry matter (Figure 2) and grain production (Table 1). Thus, the risk of grain seeding was almost nil with Sowing 2, due to a low emergence rate (cf. Figure 1) and low individual growth. Indeed, Amaranthus had a very short cycle because they flower very quickly, after only 2 produced leaves, which impeds elongation of the main stem. Therefore, they remained very low and were systematically shaded by rape, independently of rape growth. Last, grains maturation was probably too late to be complete before the first frost.

Amaranthus had conversely a very quick growth with Sowing 1, strongly influenced by available N. Indeed, the duration of the emergence to flowering period was much longer, which allowed the main stem to elongate and even overtop rape for certains weeds. Moreover, Amaranthus is a nitrophile species (Jauzein, 1995) as is rapeseed and these two species had a strongly increased growth in N200 compared to N0 (Table 1), but rapeseed was not able to overtop Amaranthus. Amaranthus grain production was strongly correlated with its dry matter. Seeding occured just before first frosts, which killed all Amaranthus. We should check if seeding would be possible with intermediate sowing date.

Long cycle weeds control

Unlike summer weeds, emerged biennials (Hordeum vulgaris L., Stellaria media L. and Geranium dissectum L.) had not all disappeared after winter. They can thereby been damped by rape at spring regrowth. Otherwise, at crop scale, they can reduce rape spring growth and yield by competing for light, water and nitrogen, but we do not present these results. At rotation scale, they can increase seed-bank by seeding. Geranium has been chosen for illustration of biennials comportment because of its weeding difficulties.

Unlike Amaranthus, Geranium dry matter and grains production (Table 1) were strongly influenced by sowing applied N and few by sowing date, especially in spring (Figure 3). Indeed, Geranium dry matter accumulation was reduced over time in N200, but earlier and more completely with early sowing. Geranium dissectum is not nitrophile and is on rosette form during autumn and winter where its growth was not much increased by more available N.

Therefore, the strong development of rape leaves in N200 (Table 1) might have reduced by shading the growth of Geranium , its building-up of reserves in the tap-root and thus its ability to survive winter and growth in spring. The low competition for light of N0 rapeseed allows on the contrary good spring regrowth of Geranium.

Figure 2. Amaranthus reflexus dry matter

Figure 3. Geranium dissectum dry matter

Consequently, Sowing 1 N200 was clean of Geranium even before winter and it remained so until harvest with an almost nil grain production by Geranium (Table 1), which led to a seeding bank reduction (80 emerged Geranium per m²). As it death was later and less complete in Sowing N200, Geranium seeded a little more grains than emerged plants. Last, Geranium spring growth was very high in N0 for both sowing dates, which led to a potentially enormous grain seeding (Table 1).


The interests of very early sowing dates and high N availibility vary according to weed type. In autumn poor N situations which are frequent with French rape cropping systems, it is not possible to spare herbicides for biennials. In autumn rich N situations, after summer slurry spreading for example, it seems necessary to sow rape very early in order to strongly increase N uptake before winter drainage, which enables a major reduction of the N leaching risk (Dejoux et al., 1999). These are also the most interesting situations for competition with non-nitrophile or biennials weeds, and this is particularly interesting for Geranium which presents major difficulties for post-emergence control by herbicides. In France, this kind of weeds are more frequent in rotations based on winter wide-crops, like the spread winter wheat-barley-rape rotation. On the other hand, seeding by nitrophile summer weeds may be largely increased in early sowing of rape with high available N, therefore herbicides must be used and targeted on these species if necessary. These weeds may be abundant if summer crops like maize are included in rotations.


1. Béres L. et Sardi K., 1998. Effect of light and nitrate on the germination of weeds (Matrcaria inodora L., Ambrosia artemisiifolia L., Abutilon theophrasti Medic.). Z. PflKrankh. PflSchutz, Sonderh, 16, 129-132.

2. Dejoux J. F., Meynard J. M. et Reau R., 1999. Rapeseed new crop management with very early sowing in order to reduce N-leaching, N-fertilization. In "'New horizons for an old crop', Proc. of the 10th Inter. Rapeseed Congress", Canberra-Australia, 26-29/09/99. GCIRC (this proceeding).,

3. Jauzein P., 1995. “Flore des champs cultivés”. INRA, Paris, 898 p.

Previous PageTop Of PageNext Page