Department for Environment, Food and Rural Affairs

Project AR0138 is covered in the report as well as other projects on related work covering in total a 10 year period, from 1993 to 2003 (see Annexe 1). The report is in the form of a review document, considering the main agronomic aspects in sequence.

Report on DEFRA funded research on autumn sown determinate lupins.

1. Crop Establishment

1.1. Sowing Date.
From the very earliest field experiments (1992 to 1994) conducted it was obvious that the sowing date had a very strong effect upon the probability that a crop would survive the winter and upon the structure of the individual plants in the following spring and summer (Shield et al. 1996).

The first autumn sown determinate cultivar, Lucyanne, was extensively tested at Rothamsted and at the INRA station at Lusignan, western France. Basic models were constructed that described the relationship between sowing date, autumn weather and plant development. The models used accumulated thermal time (above a base temperature of 3°C) to predict the development and the extent of lignification of the root parenchyma (crucial to cold tolerance, Huyghe and Papineau 1990) and leaf production (Huyghe 1991 and 1993). A vernalisation model used accumulated thermal time (between temperature limits of 1 and 14°C) to determine when the main stem apex turned floral (Huyghe 1991 and 1993). It was demonstrated that the number of main stem leaves initiated prior to floral initiation was closely related to the structure of the whole plant (Julier and Huyghe 1993, Shield et al. 1996). These underlying principles were behind the modified models of the effect of the vernalisation response on plant architecture proposed by Shield et al. (in press), and shown in Annexe 2.

The models were consequently re-worked using long term temperature records to predict optimal sowing windows for the different regions of the UK (Milford et al. 1996). The recommended calendar dates varied with regional climate, late August and early September in cooler northern England and Scotland and late September and early October in warmer southern and south-western England.

Those sowing windows formed part of the management guidelines for a series of experiments testing the geographic range of cvs Lucyanne and Ludet (a sister of Lucyanne) conducted between 1994 and 1996. Despite the results showing that these cvs were suited to most arable sites in England and Wales they were rapidly replaced by the dwarf determinate cvs Lucille and Lunivers. The dwarf determinates showed greater cold tolerance and resistance to lodging.

It was demonstrated that the dwarf determinate cvs behaved similarly to Lucyanne and that the same sowing windows could be recommended (Shield et al. in press). Latterly (from 2001 onwards) it was observed (not measured from a designed experiment) that sowing dwarf determinate cvs on clay soils produced smaller plants than sowing on sandy soils. These plants had similar structure in terms of leaf and branch number but with shorter inter nodes and slightly smaller leaflets. This lead to the recommendation that crops of dwarf determinate cvs could be safely sown earlier in the autumn than originally recommended via the sowing windows without fear of large over winter losses or lodging the following summer.

1.2. Row Spacing and Seed Rate.

Early field experiments (1991 to 1994) were sown on wider rows than usual for combinable crops in the UK (35 cf. 8-12 cm), as this was the practice in France. Experiments conducted by ADAS in 1995 and 1996 (at Rosemaund and Arthur Rickwood, unpublished data) and by Rothamsted and INRA (1996-98, Shield et al. 2002) demonstrated that the use of wide rows was unnecessary and possibly undesirable. Herbicide efficacy was greater and light distribution potentially better (see canopy architecture below) in crops grown on 12 cm rows.

The original seed rate recommendation for the UK (of 40 seeds m^-2) was derived empirically from early experience, but stood the test of much subsequent detailed investigation. Over winter plant losses (see below) in crops sown within the sowing window were approximately 40%, leaving 24 plants m^-2 in spring. If sown early in the sowing window and resulting in large individual plants this could be too many plants for optimal canopy architecture (see below). From 1999 onwards it was recommended that in good sowing conditions, such as early sowings where over winter plant losses were predicted to be low, as few as 30 seeds m^-2 may be sown.

2. Over-winter losses.

Over-winter plant losses have typically been 40% of the sown seeds in latter years, the causes have varied with season. The three principal causes were frost, bean seed fly (Delia platura) larvae and slugs. Fungal pathogens were only occasionally the primary cause of plant loss, but were able to infect sites of damage caused by frost or invertebrates. Rooks and crows pulled emerging seedlings from the ground but were unable to remove seedlings with two or more true leaves emerged. Damage to lupin crops by other birds was rare. There was occasional grazing by deer and rabbits. Hares could be a particular problem in spring when they grazed the extending main stem.

2.1. Cold tolerance.

Early experiments (1991 to 1993) with Lucyanne demonstrated that young seedlings were sensitive to relatively modest frosts (Shield et al. 1996). Prior to lignification of the root parenchyma the effect of frost penetrating to 4 or 5 cm soil depth could be to kill the plant (Huyghe and Papineau, 1990). Three or more consecutive nights with air minima of <–3°C and intervening cold days could result in severe crop damage (Shield et al. 2000). Once the root parenchyma was fully lignified the plants could withstand air temperatures of –12°C (Shield et al. 2000, but see below).
The autumns of 1994 and 1995 were unusually warm. All autumn sown lupin crops were at an advanced stage of development when the first frosts were experienced. Stem extension was observed in cvs Lucyanne and Ludet when sown at the beginning of the sowing window for a region. These extending plants suffered damage from relatively mild frosts of –3°C.

F
igure 1. The relationship between thermal time from sowing to the first severe frost (<-3°C) and the percentage of plants killed by that frost for two cultivars of autumn sown lupin at multiple sites in contrasting seasons (from Shield et al. 2000).The polynomial model fitted was y = 0.0004008 x² - 0.425 x + 137.1 which explained 55% of the variance in y.
The autumn of 1993 was a complete contrast to 1994 and 1995 and autumn sown crops experienced cold weather between sowing and the first frost. The network of experimental sites sown to define the geographic range of the crop provided excellent data to demonstrate frost tolerance in the first autumn sown determinate cultivars (Shield et al. 2000 and Figure 1.). The lupin plant became progressively more frost hardy with time after sowing (as the root parenchyma became lignified) until a turning point was reached after which the plant became progressively more sensitive to frost (as stem extension began).

The percentage of emerged plants lost to frost was variable. However, the data in Figure 1 were in agreement with controlled environment studies reported by Leach et al. (1997). These studies showed that lignification of the root parenchyma and adequate frost tolerance was achieved approximately 400 DDA3°C after sowing

2.1.1. Sowing late in the autumn.

In the autumn of 1996 a demonstration area was sown at Rothamsted (Figure 2). It included three cultivars (Lumineaux, an old indeterminate, Lucyanne and Lunivers) sown at 6 dates. The objective was to demonstrate our work on Genotype x Environment interactions and the effect canopy architecture. The ‘treatments’ were organised for visual effect and it was in no way a designed experiment. However, it was quite clear that the cv Lumineaux survived the winter from a later sowing date than either of the other two cvs. This suggested that lignification of the root parenchyma occurred earlier in the life of the plant.

Figure 2. a).


Figure 2.b).
Figure 2. a) A field demonstration of the effects of sowing date and genotype. From left to right are 6 sowing dates, 14 days apart from 12^th August (left) to 22^nd October (right) and from front to back are 3 genotypes Lunivers (foreground blue flowering), Lucyanne (middle ground) and Lunineaux (background, both white flowering). b) The same demonstration photographed from the bottom left in a) above.

Lumineaux was able to survive when sown on 7^th October, whereas Lucyanne and Lunivers were not able to survive.
Two years of field experiments (Table 1) failed to confirm this observation. The field experiments were coupled with controlled environment experiments following the protocols described by Leach et al. (1997). Lunivers was replaced with a new breeding line (CH 1391) thought to be more frost tolerant than many other genotypes (Huyghe and Harzic, INRA Lusignan, personal communication). Plants were grown in soil filled boxes through which a glycol cooling pipe passed. This allowed controlled freezing of the soil in the root and hypocotyl zone when plants were different ages. Plants were removed from the soil tanks immediately before a freezing treatment was applied and sections of root stained and examined under a low power light microscope to determine the extent of lignification of the root parenchyma. It was not possible to detect any differences between genotypes in their ability to survive soil freezing or in the timing or extent of lignification of the root parenchyma.
Table 1. The percentage of established plants in January dying during the winter. Four sowing dates of three genotypes in two years at Rothamsted. SED for 1998-99 = 13.02 (22 df) and for 1999-00 = 12.64 (24 df).

	Sowing date
1998-99	11 Sept	25 Sept	9 Oct	23 Oct
Lucyanne	3.1	15.0	21.4	52.3
Lumineaux	8.2	14.8	23.0	50.7
CH 1391	6.0	14.6	24.0	53.3
1999-00	1 Sept	16 Sept	28 Sept	19 Oct
Lucyanne	16.4	1.2	12.2	44.5
Lumineaux	11.0	24.4	40.9	73.8
CH 1391	6.9	1.8	24.1	22.6

2.1.2. Sowing early in the autumn.

A
utumn 1995 was the first time that the dwarf determinate cultivars Lucille and Lunivers were sown on a large scale at Rothamsted. In warm conditions they were shown to remain in the rosette stage longer than Lucyanne or Ludet, and therefore to tolerate frost that caused severe damage to those non-dwarf cvs (Figure 3).
Figure 3. The effect of mean plant height (from soil surface to top of plant) 655 DDA3°C (November) from sowing on the number of frost damaged plants m^-2 in spring. Cultivars Lucyanne, Ludet (non-dwarf) Lucille and Lunivers (dwarf) sown at 40 seeds m^-2 in autumn 1995.
In several cases of plants that began stem extension in autumn the frost damage occurred above the basal rosette of leaves, killing the main stem but not the whole plant. Branches grew from the basal leaf axils and contributed to yield. A comparison of seed yield per plant showed that such plants produced on average 70% of the yield of undamaged plants (unpublished data).
Generally the stage of development of the lupin plants when the first frosts of <-3°C (first severe frost in Figure 1) of each winter were experienced determined the fate of the crop. In most years the plants then remained at that stage of development until spring (Shield et al. 2000).
It was possible to detect an effect of the duration of the cold temperatures on plant survival. An extended cold period could damage plants more than short duration extreme temperatures. In the winter 1996-97 at INRA Lusignan in France the minimum temperature was –12.5°C, the temperature remained <0°C for 12 days and almost all autumn sown lupins were killed, except for a small number of potentially very valuable breeding lines. One of those lines, CH 1391, was tested alongside Lumineaux and the control cv Lucyanne in the experiments described above, where all three failed to show any notable differences in cold tolerance.
In Herefordshire in the winter of 1999-2000 the minimum temperature was –14.5°C, but as part of a cold spell that only lasted 3 days. Crops of Lucille in the area survived undamaged. The following year temperatures fell to similar values but as part of a longer cold spell and crops of Lucille were damaged.

It is recognised that snow cover forms a useful insulator for autumn sown crops. At Ancenis in France, during the same cold period described at Lusignan above, 10 cm depth of snow was sufficient to prevent the severe damage to lupins recorded at Lusignan (Shield et al. 2002).

2.2. Pests.
Bean seed fly larvae were the most serious insect pest of autumn sown lupins (Bateman et al. 1997). They were present at all sites where the crop was sown, but in variable numbers from season to season. In the warm autumns of 1994, 1995 and 1997 they were active in October, but more normally only in September around the time of sowing and emergence. Normally the larvae tunnelled into the root / hypocotyl from the soil resulting in plant death, more unusually surface activity resulted in damage to the plant apex. Experiments were conducted at Rothamsted between 1997 and 1999, and at ADAS Rosemaund and Starcross (Devon) in 1998 and 1999, to evaluate potential control methods. Unfortunately there was very little insect activity in the latter two years.





Figure 4. The effect of potential control methods for bean seed fly (Delia platura) larvae in autumn sown lupins, 26^th September 1997.

Figure 4 shows the results from Rothamsted in autumn 1997, when bean seed fly larvae were abundant. The data clearly shows that it is not possible to react to visual signs of damage. The treatments applied at the 2-4 true leaf stage showed no advantage over the ‘no control’ treatment. It was vital to incorporate chlorpyrifos into the soil to prevent rapid breakdown, when incorporated it was very effective. Bendiocarb was the only insecticide seed treatment available in the UK with a label recommendation for bean seed fly control, but not for use in lupins. Ultimately, due to pesticide registration costs, seed was imported from France with an alternative carbamate insecticide seed treatment applied (furathiocarb).

Several cultural control measures were tested including sowing old seed that had been stored for a number of years. Of these the only successful method was the stale seedbed technique. If a seedbed were created 3 weeks prior to seed sowing, and the only soil movement on the day of sowing was caused by the drill, damage could be lessened. It is thought that this was due to the release of volatiles from soil organic matter during and immediately following cultivations. These volatiles attract the female fly to lay her eggs. If seed sowing is then delayed for 3 weeks that batch of eggs have completed their development and have pupated and migrated as adults. Minimal soil disturbance by the drill prevents new egg laying from taking place.

The limitations of such a technique were demonstrated the following year on the silt soils at ADAS Rosemaund where the soil structure was not able to sustain good seedbed conditions for three weeks after cultivations. However, at Lusignan in France, in 1999 the technique was again demonstrated successfully as part of a larger experiment on bean seed fly control. A stale seedbed was also successfully created by ADAS in Devon, but unfortunately there was very little bean seed fly activity. The soils at Rothamsted and Lusignan are clay loams and in Devon the soil had a relatively coarse texture. All three soils had greater natural structure and were able to sustain seedbed conditions for three weeks.

Slugs have also been serious pests of autumn sown lupins. They often fed below the soil surface and therefore could not be controlled with surface applied pellets. Early in the life of the plant small scale grazing of the hypocotyl could result in death, later the plants appeared more able to withstand grazing. It is thought possible that this could coincide with the lignification of the parenchyma, as described for the root above. Above ground, plants damaged by bean seed fly larvae or by slugs, were observed to wilt. It was necessary to dig them up in order to determine the cause of wilting. When bean seed fly was the cause the hypocotyl and root were usually wet and rotten. Often it was possible to recover a larva from the hypocotyl. When slugs were the cause the plant material remained intact except for the immediate point of damage.
Occasionally plants were damaged above ground by Thrips angusticeps. Affected plants produced progressively smaller and thickened leaves until no further leaves emerged and the meristem appeared black and dead. Very occasionally such plants could produce a basal branch from the axil of an early emerging leaf and survive the damage. More typically they died. Generally thrip damage was considered insufficient to justify a pesticide application. However, two sowings were completely destroyed by thrips, one of the first commercial crops sown in conjunction with Dalgety in Herefordshire in 1993-94, and an experiment at Rothamsted in 1995-96. On the second occasion only the middle sowing date of three (each 14 days apart) was affected.

2.3. Diseases.

Plant losses during the winter months caused by fungal diseases were relatively uncommon. Fungi infected the sites of damage by animals (principally slugs and bean seed flies) and by frost. Fusarium spp. were commonly, but not exclusively, found on plant parts below the soil surface (Bateman 1997). Botrytis was more common above the soil surface. Etheridge and Bateman (1999) working in controlled conditions demonstrated that an artificial damage site on the hypocotyl increased the number of plants infected with these fungi and the numbers ultimately dying.

Occasionally Botrytis was pathogenic in its own right, especially in late winter and spring, when Botrytis cinerea infected the newly extending stem tissue. In a series of experiments studying canopy architecture and light interception it was found that Botrytis infection was least in the more open canopies. Coincidentally these were the canopies that produced the greatest seed yield and were therefore recommended in our agronomic advice to farmers and growers.

In Australia intensive lupin growing has increased the incidence of Pleiocheata root rot, a seedling disease. To date this disease has not been recorded in the UK. When causing Pleiocheata root rot the fungus infects the hypocotyl beneath the soil surface. It can be transmitted on lupin seed or survive for a number of years (up to 5) on crop residues in the soil. Despite not having been recorded in the UK, prevention of Pleiocheata root rot is the principal reason for the recommendation that lupin crops are grown no more frequently than once every 5 years on a piece of land.

The causal fungus, Pleiocheata setosa, has caused ‘brown spot’ symptoms on pods and branches in late summer in the UK. It was the subject of a number of field experiments at Rothamsted in the early 90s, but in subsequent years was not considered sufficiently detrimental to the seed yield to require fungicide application. In the production of seed for re-sowing control of Pleiocheata setosa infection is very much more important to prevent seed transmission of disease. Fortunately it is very sensitive to applications of tebuconazole used primarily to control rust (see summer diseases below).

3.0. Weed control.
The combination of the difficulties of registering pesticides for additional uses and the sensitivity of lupins to herbicides generated a large amount of work to identify suitable herbicides and sequences to achieve weed control. Fortunately products with label approval for use on combining peas and beans were given automatic off label approval for use on lupins. However lupins are sensitive to a number of herbicides commonly used in pea and bean crops. An added complication was that certain herbicides are no longer approved or were not submitted for re-registration.

ADAS, PGRO and one or two agrochemical distributors assisted Rothamsted Research by also carrying out herbicide screens (Knott 1996 and Shield et al. 2000). Most herbicides available in the mid 1990s were screened by one or another of these partners. Many of the new introductions since that time have also been screened.

As with many broad-leafed crops the selective control of grass weeds did not prove to be difficult. However, from the point of view of herbicide resistance amongst the grass weeds the narrow range of chemical groups (mainly ACC inhibitors, the Fops and Dims, and one amide) available for grass weed control in lupins is a concern.

At the present time only the following broad leafed herbicides can be recommended to growers of winter lupins, pre crop emergence; clomazone, simazine, simazine + trietazine, terbuthylazine + terbutryn, and trifluarlin, post crop emergence; simazine. This list contains exclusively residual acting herbicides, which means that they must be applied pre weed emergence and rely on good seedbed conditions and adequate soil moisture. They are largely unsuitable for use on high organic matter soils.