Previous PageTable Of ContentsNext Page

Varietal variation in allelopathic activity in wheat and barley and possibilities to use this in breeding

Nils-Ove Bertholdsson

Svalöf Weibull AB, SW Laboratory, 268 81 Svalöv, Sweden. E-mail


Most breeding traits are the same in conventional and organic cereal breeding. However, a trait that is very important in organic and low-input farming, but until now been neglected in breeding is weed competitive ability. Morphological and physiological characters, mainly influencing either the biomass growth or the uptake of nutrients and water, could be used as potential breeding traits. However, the significance of allelopathy, i.e. root exudates inhibiting weed growth, is more and more envisaged. In both barley (Hordeum vulgare L.) and wheat (Triticum aestivum L.) there is a large variation in potential allelopathic activity, measured as the root growth inhibition of ryegrass. Nordic barley landraces and old cultivars are in general more allelopathic than new bred cultivars, but there are exceptions that could be used in breeding. In wheat there is an opposite trend as Swedish landraces and old cultivars are less allelopathic than new ones. Swedish cultivars are also in general less allelopathic than cultivars from western and central Europe and North America. In a screen of 813 spring wheat cultivars and lines two cultivars were found to be as allelopathic as the most allelopathic barley and are now used in a conventional breeding program to improve the allelopathic proprieties of the Swedish spring wheat.

Media summary

New gene sources for high allelopathic activity are found in spring wheat and this may reduce the herbicide use in the future wheat production.

Key Words

Allelopathy, weed competition, genetic variation, barley, wheat, breeding


Allelopathy is more important in the crop-weed inference complex than generally thought of. Olofsdotter et al. (2002) also suggested that this could be one reason for the limited success in previous breeding programmes for weed competition. The significance of the root development and indirect probably also root exudation of allelochemicals was shown in a study of breeding traits for cereal cultivars for organic production (Jönsson et al. 1994). The study, including barley (Hordeum vulgare L.) and oat (Avena sativa L) cultivars, showed that the differences in root growth among barely cultivars explained about 50% of the variance in weed biomass, while the part explained by shoot growth was insignificant. In oat, on the other hand, differences in early shoot and root development explained equally parts. Competition for nutrients and water could explain some of the results, but also differences in allelopathic activity may have played a significant role. The significance of allelopathy is also stressed in a recent study of breeding traits in barley and wheat (Triticum aestivum L.) (Bertholdsson 2005). Using stepwise multiple regression analysis it was shown that only early vigour, measured as the biomass about two weeks prior anthesis and allelopathy, measured as the growth inhibition of ryegrass roots, explained a significant part of the differences in weed biomass of the studied cultivars. The two traits were of more or less equal significance. In this paper some results are presented from studies of genetic variations in allelopathic activity in barley and wheat and possibilities to use these in breeding of more allelopathic cultivars. Results are also presented to justify a breeding programme for more allelopathic cultivars.



In spring barley two different materials were used. The 1st material consisted of 127 old and new barley cultivars from Finland, Sweden, Denmark and the Baltic states obtained from the Nordic Gene Bank in Alnarp, Sweden and field propagated in the south of Sweden (Kolodinska Brantestam 2005). The collection from each country consisted of two-row barley landraces and cultivars, except for the Finnish collection, which also contained six-row barley cultivars. In the Swedish collection one newly released cultivar and two breeding lines were included. One newly released cultivar was also included in the Danish collection. The year of introduction for the local landraces was set to 1890 since there are cultivars selected from these landraces dating back to 1900 (see Bertholdsson 2004). The 2nd material consisted of 83 old and new cultivars from Europe and tested in organic trials in Denmark. All field data of the 2nd material was kindly provided by Preben K.Hansen and Jakob Willas in the Danish DARCOF II project BAROF (


In wheat 4 different materials were used. The 1st material was 813 spring wheat cultivars from the gene pool collection at Svalöf Weibull AB. One part of the collection was old and new Swedish spring wheat, another part originated from Ethiopia (about 400 cultivars/lines) and the rest from Central and Eastern Europe, Asia, Africa, North and South America. The 2nd material was new cultivars and breeding lines of spring wheat from Svalöf Weibull AB. Some of these were also tested in organic yield trials during 1-3 yrs. The 3rd and 4th materials were new winter wheat cultivars tested in official trials during 2003 in Sweden and Germany, respectively.

Evaluation of allopathic activity

An agar based method adapted from Wu et al. (2000) was used to evaluate the potential allelopathic activity (PAA). The method was evaluated against the weed data from an organic yield trial in 1999 and different parameters were changed to obtain highest possible correlation with observed variation in the ability to suppress weeds. Parameters evaluated were number of plants, number of receiver plants, gel concentration and gel quantity, length of test period and type of receiver plant. In the final bioassay, plastic tissue culture vials (Phytotech, 400 mL) were filled with 30 mL 0.3 % water agar, and six pre-germinated barley or wheat seedlings were planted circular along the vial wall with 10 pre-germinated perennial ryegrass (Lolium perenne cv. Helmer) seedlings, planted in the centre of the vial. The cereal and ryegrass seed were pre-germinated on filter paper in darkness for two days at 25°C. Seeds were in general free from seed-borne diseases and thus, no seed sterilisation was necessary. The vials were loosely sealed with a lid and placed in a growth chamber with a light/dark cycle of 16/8 hr, at a temperature of 20°C and inflorescent light of 52 µmol x m-2 s-1. After seven days, the root area in mm² of ryegrass was measured using an image analyser (DIAS, Delta-T Devices, Cambridge, England). Vials with only ryegrass were used as controls. The barley and wheat roots were dried at 80°C for 48hr and dry weight was measured after removal of the root crown. In some of the materials fresh weights was used instead of dry weights. PAA was calculated according to PAA = (1-A1/A2)*100 with A1= ryegrass root area in presence of barley or wheat and A2= ryegrass root area without barley or wheat. Based on PAA, specific potential allelopathic activity (SPAA) was calculated as SPAA=PAA/root dry weight or fresh weight. All tests were done with three to eight replicates depending on material. It may be agued that some competition could occur in the modified method, but a test with donor and receiver plants separated in time gave similar rankings as if donor and receiver plants were grown together.

Results and Discussions

Genetic variation in spring barley

In the study of material 1 with landraces, old and new cultivars bred and released in the Nordic and the Baltic, allelopathy was showing a decreasing trend, with some deviations from this trend especially among more recent released cultivars (Bertholdsson 2004). One obvious explanation of the negative trend could be a reduced root growth and hence less exudation of allelochemicals of newer cultivars. However, there was no such trend in root growth with one exception of the Danish cultivars. Because of this SPAA, i.e. PAA divided with the root weight also showed a negative trend with exception of the Danish material. In both Swedish and Danish barley cultivars, however, there is a negative trend in thousands kernel weight that may have influenced the results in some way. Again the different trends in root growth, but similar trends in PAA do suggest that PAA is not related either to differences in root growth or to seed size. This is also suggested by Jensen et al. (2001) in rice and Wu et al. (2000) in wheat.

The historic Nordic material showed a normal distribution pattern centered around 40-45 percentage ryegrass root inhibition with the landraces and old cultivars at the upper end and most of the newer cultivars at the lower end (Figure 1). The pattern is normal distributed even if some cultivars with low activity are missing. Normal distribution often indicates that the trait is quantitatively inherited (Olofsdotter 2003; Wu et al. 2000).

Figure 1. The distribution pattern of the potential allelopathic activity (PAA) of 127 Nordic barley cultivars. Standard error of the mean (SEM) =4.3

PAA of material 2 showed also a normal distribution pattern with an optimum at 45 to 55 percent reduction of the ryegrass root growth. There were several new cultivars among those with a high PAA value, as e.g. Simba and Wikingett. However, a pre-dominant part of new cultivars and number cultivars were found at the lower end, e.g. SJ5519 and SW2511 (Figure 2). The higher PAA values in material 2 than in material 1 are probably more an effect of differences in seed vigour because of year and/or time of the bioassay. Similar situations are also observed in rice and it is therefore difficult to compare tests (Maria Olofsdotter, personal). Some cultivars were common in both materials, but with seeds from different years and locations. eg. cv. Pallas showed a PAA value of 54 in material 1 and 65 in material 2. A similar difference was also observed among other cultivar that occurred in both materials.

Figure 2. Frequency distribution of 83 European barley cultivars. It should be noted that the higher PAA values than reported in Figure 1 is more related to seed vigour or others factors than to genotypic differences (see the text). SEM=2.7

Most of the cultivars in material 2 have also been studied in hydroponics for root development. The root biomass without any interference with ryegrass showed a weak but significant positive correlation with PAA (r=0.24, P≤0.05)(Figure 3). The root biomass from the bioassay was, however, not correlated with PAA (r=0.19, P≤0.05). Therefore, some differences in PAA could be related to differences in root growth, but most are not related.

Figure 3. The relationship between the root biomass per plant after 14 days cultivation in a hydroponic system and the potential allelopathic activity (PAA) of the barley cultivars in material 2. SEM=1.7 for the root biomass and SEM=2.7 for PAA. *P≤0.05

Genetic variation in spring wheat

Opposite to barley there is a positive trend in PAA as new cultivars are released. The Swedish spring wheat landraces here dated as released 1900 (Figure 4, left) all showed a low PAA value and during 100 years of breeding PAA has increased from 10 to 20 %. In other parts of Europe PAA also show a similar trend but at a slightly higher level (Figure 4, right).

Figure 4. Variation in potential allelopathic activity in cultivars bred in Sweden (left) and other parts of Europe (right). Bars=2 x SE. *P≤0.05; ***P≤0.001.

A similar trend is also observed among cultivars and breeding lines coming from USA, Canada and Mexico. Some of this material was Triticale lines, which in general showed high PAA values (Figure 5). The frequency distribution of PAA follows like in spring barley a normal distribution pattern with peaks between 15 and 25 (Table 5b). There are, however, a few cultivars with a PAA in the levels of the best barley. These cultivar are now used in a breeding program to improve the allelopathic activity in Swedish spring wheat

Figure 5. A (Left);Variation in potential allelopathic activity in spring wheat cultivars bred in North America. Black symbols denote triticale lines. Year of introductions is in this case year of introduction of the material in the Svalöf Weibull AB gene pool. Bars=2 x SE. **P≤0.01.

B (right); Frequency distribution of 813 spring wheat cultivars and lines. SEM= 3.6 – 4.1

Most of the Swedish advanced breeding lines showed a PAA value of 15 to 20. About 10 % show a doubled activity (Figure 6, left). Even if the PAA values a rather low in spring wheat as compared with barley, wheat cultivars with slightly higher allelopathic activity in general showed less weed growth (Figure 6, right). The relationship is weak but significant. If also differences in early growth are accounted for the correlation coefficient is improved to 0.46 (P≤0.01)

Figure 6. Frequency distribution of the genetic variations of potential allelopathic activity (PAA) of new breeding lines of Swedish spring wheat. SEM=4 (left) and the relationship between PAA and weed biomass in plots with new breeding lines of Swedish spring wheat (SEM = 5.4-6.8 for weeds and SEM=3.8 – 6.8 for PAA, depending of no. of years the line has been studied (right). *P≤0.05.

Genetic variation in winter wheat

In winter wheat no historical germplasm has been studied like in barley or spring wheat. Both Swedish and German cultivars, either already on the market or near introduction, showed similar patterns of distribution of PAA (Figure 7). PAA was low like in Swedish spring wheat and the lack of cultivars with high PAA indicates that it is probably necessary to introduce new genes to be able to improve PAA of future cultivars.

Figure 7. Frequency distribution of 24 winter wheat cultivars either on or will be introduced on the Swedish market (left) and 55 cultivars already or will be introduced on the German market (right). SEM=2.65

PAA and its relation to weed competitive ability

One question that is frequently asked is if a bioassay with a certain receiver plant can be used for screening of cultivars with an un-specific allelopathic activity or even if it is possible to transfer bioassay results to field conditions. The first answer is probably a no and we are therefore talking about potential allelopathic activity. All results from the lab need further evaluation out in field. However, the bioassay was not optimized for ryegrass, but for a natural mixture of weeds in organic trials and therefore hopefully the bioassay may be less specific in this way. PAA or SPAA and weed ground coverage of the barley cultivars in material 2 showed low correlation coefficients that were significant only in one of the years (Table 1). Early shoot weight, measured during the bioassay, and straw length were both strongly correlated with weed ground coverage. One problem is that some cultivars may be weak competitors because of low vegetative growth at an early stage or show some other features negative to weed competitive ability. In this case a high allelopathic activity may not be enough to suppress the weeds and the relation between PAA and weed competitive ability will be low. Therefore, the weed competitive ability influenced by morphological characters should be combined with the allelopathic activity in a multivariate or multiple regression models. Doing this Bertholdsson (2005) showed that allelopathy when added to early vigour and early growth explained 44-69% of the observed genotypic variance in weed competitive ability in barley. The two traits separately explained 24-57 % and 7-58%, respectively, depending on year. Hence, with PAA in the model another 12-26 % was explained. In spring wheat the figures were lower: 14-21 % for early vigour, 0-21 % for PAA and 27-37 % when combined. See also Figure 6, right. In this graph the spring wheat results presented in Bertholdsson (2005) are extended with one more year of field trials and the results are also combined over years.

Table 1. Genotypic correlations between weed ground cover and four traits of interest for weed competitive ability of barley cultivars tested in organic yield trials at 2-3 locations each year. Field data of weed ground coverage are kindly provided by Preben K.Hansen and Jakob Willas from the Danish DARCOF II project BAROF.





PAA (%)




SPAA (% x root weight-1) Early shoot weight (mg)




Straw length




Number of cultivars




n.d. no data
* P≤0.05; **P≤0.01; ***P≤0.001

In 2004 the ground coverage of various weeds were also analyzed. If sub-divided on weeds PAA or SPAA was negatively correlated with veronicas. Other weeds did not show any significant correlations with PAA or SPAA , except for PAA that was positively correlated with rape seeds and “other” weeds. The positive correlation with rape seed is unexpected since rape seeds were tested instead of ryegrass as receiver plants during the development of the bioassay and in this case gave similar results as with ryegrass. If PAA and SPAA are combined with early vigour the relationships will probably be further strengthened as shown by Bertholdsson (2005).

Table 2. Genotypic correlations between weed ground cover of different weeds and SPAA of barley cultivars tested in organic yield trials in 2004. No. of trials was 1-2 depending of weeds. Field data of weed ground coverage are kindly provided by Preben K.Hansen and Jakob Willas in the Danish DARCOF II project BAROF. N=29







Trifolium sp.



Rape seeds



Veronica spicata



Matricaria L
Atriplex laciniata



Stellaria media L.






Persicaria L.




*P≤0.05; ** P≤0.01

PAA and effects on grain yield

Another question that is often asked is if it will cost yield if allelopathy is increased. There are no studies concerning this, but in theory exudation of allelochemicals could affect yield in several ways; metabolites and energy are lost, there is a risk of over-sizing the root system, negative effects on the micro flora of the soil and risk of auto-toxicity (Olofsdotter personal; Ben-Hammouda et al. 2002). The negative breeding trends in barley may imply that high allelopathic activity is something negative in the modern agricultural practice. However, since there are new modern cultivars with high PAA values it seems possible to combine the two traits. The breeding trends in wheat also indicate that there is no negative relationship to yield. On the other hand, PAA is much lower in wheat than in barley and the question is what will happen if PAA is increased to the level of the most allelopathic barley.

There are yield data from material 2 in barely. If the whole material is considered there are no relation to yield (Figure 8). However, if only breeding lines and cultivars to be released or newly released in Sweden or Danmark are considered (Figure 8, left) there is a clear negative relation. If the negative trend is a causal relationship is not known, but the fact that there are cultivars that do not follow this relationship implies that other factors may be involved. Correlation breakers like cv. Simba are therefore interesting even if the weed competitive ability of this cultivar is not particularly good. The reason for this is probably that the early growth rate and straw length is below average of this cultivar.

Figure 8. Relationship between PAA and grain yield of some old cultivars, cultivars on the market and cultivars and breeding lines to be introduced on the market (left) and breeding lines and a few new EU cultivars that have been tested during the last years in Swedish or Danish official trials (right). The three cultivars marked with names are not included in the regression analysis. The yield data is from organic BAROF yield trials 2002 and 2003. Means of 2 x 3 trials. SEM=2.9 for PAA

In winter wheat there is a weak but significant positive relation between yield and PAA of new cultivars tested in Sweden and no relation at all among the cultivars tested in Germany (Figure 9). This may indicate, as in spring wheat, that an allelopathic activity at this low level is either not negative for yield or that there are no direct relationship with yield.

Figure 9. Relationships between potential allelopathic activity (PAA) and grain yield of new cultivars in official trials in South Sweden (left) and Germany (right). LSD0.05 = 4 for yield in Sweden, SEM=2.6 for PAA in Sweden and German materials.

Possibilities to breed for improved competitive ability

There are genetic sources for improved allelopathy available in both barley and wheat. Genetic studies in rice (Jensen et al. 2001) and wheat (Wu et al. 2003) implies that the trait is quantitative inherited but some major genes may be involved (Wu et al. 2000). In barley allelopathy is already rather high also in some adapted cultivars and it may be difficult to improve this further without finding new gene sources. However, if the allelochemicals differ in various landraces it should be possible to combine these to improve allelopathy further. One such example is Pallas that combines genes from both Swedish and German landraces (Bertholdsson 2004). A relatively small change could improve the weed competitive ability as shown by Bertholdsson (2005). In this study multiple regression models were used to predict the improvement of weed competitive ability if e.g. both early vigour and PAA was increased with 20 %. According to these calculations the weed biomass could be reduced by 31-54 %. A prerequisite is that there will be no auto-toxicity or other negative effects of the growth of the plants.

In wheat a breeding program would be even more promising. The allelopathic activity is much lower than in barley and there are non-adapted high allelopathic spring wheat cultivars identified and also gene sources found in winter wheat (Wu et al. 2000). In similar calculations as in barley an improvement of PAA to the level of the most allelopathic barley would decrease the weed biomass with 56-66 % and if early vigour also is improved the weed competitive ability could be improved even further (Bertholdsson 2005). In case of spring wheat non-adapted materials has to be introduced and therefore genetic markers could be useful both for the background selection to recover the recipient parent genotype but also the foreground selection if the efficiency and cost is lower than for the bioassay.


The varietal variation in allelopathic activity is high both in barley and wheat. In Nordic barley the high allelopathic activity originates from several local landraces and the trait has been eroded during a century of breeding. Still it is rather high and there are modern cultivars that show high allelopathic activity. The potential allelopathic activity was low or medium low in most tested new winter wheat for the Swedish and German market. The trait was also low in Swedish spring wheat, but contrary to barley the activity has been slowly increased during a century of breeding but is still much lower than in barley. A few high allelopathic cultivars mainly from North Africa and Asia are identified and now used in a breeding program for improving the allelopathic activity of Swedish spring wheats. It is predicted that an improvement of the allelopathic activity to the levels of the most allelopathic barley could reduce the weed problem in wheat with 58-66 % under the assumption that there are no negative effects on growth. There are no negative effects on yield at the present levels of allelopathy. The question is if it will be the same if allelopathic activity is doubled. In barley there is a negative relation especially among new Danish breeding lines.


The Swedish Research Council for Environment, Agricultural Sciences and Spatial Planning (Formas) is acknowledged for funding this research. The breeders at Svalöf Weibull AB, Agnese Koldinska Brantestam, SLU, Alnarp, Sweden, Hanne Oestergaard and co-workers in the project BAR-OF and SUSVAR, Denmark are acknowledged for providing seeds and field data and Eva Nordqvist for all technical assistance.


Ben-Hammouda M, Ghorbal H, Kremer RJ and Queslatt O (2002). Autotoxicity of barley. Journal of Plant Nutrition 25, 1155-1161.

Bertholdsson NO (2004). Variation in allelopathic activity over one hundred years of barley selection and breeding. Weed Research 44, 78-86.

Bertholdsson NO (2005). Early vigour and allelopathy – two useful traits for enhanced barley and wheat competitiveness with weeds. Weed Research 45, 94-102.

Jensen LB, Courtois B, Shen LS, Li Z, Olofsdotter M and Mauleon RP (2001). Locating genes controlling allelopathic effects against barnyardgrass in upland rice. Agronomy Journal 93,21-26.

Jönsson R, Bertholdsson, NO, Enqvist G and Åhman I (1994). Plant characters of importance in ecological farming. Journal of the Swedish Seed Association 104, 137-148.

Kolodinska Brantestam A (2005). A Century of Breeding – is Genetic Erosion a Reality? Temporal diversity changes in Nordic and Baltic barley. Doctorial dissertation ISSN 1652-6880, ISBN 91-576-7029-3, Alnarp, Sweden.

Olofsdotter M, Jensen LB and Courtois B (2002). Improving crop competitive ability using allelopathy – an example from rice. Plant Breeding 121, 1-9.

Wu H, Pratley J, Lemerle D and Haig T (2000). Evaluation of seedling allelopathy in 453 wheat (Triticum aestivum) accessions against annual ryegrass (Lolium rigidum) by the equal-compartment-agar method. Australian Journal of Agricultural Research 51, 937-944.

Wu H, Pratley J, Ma W and Haig T (2003). Quantitative trait loci and molecular markers associated with wheat allelopathy. Theoretical and Applied Genetics 107,1477-1481.

Previous PageTop Of PageNext Page