Plant Breeding, 136, 372–378 (2017) doi:10.1111/pbr.12469 © 2017 The Authors Plant Breeding Published by Blackwell Verlag GmbH This is an open access article under the terms of the Creative Commons Attribution-NonCommercial License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited and is not used for commercial purposes.
Weed tolerance in soybean: a direct selection system B E R N D H O R N E B U R G 1,4
, S A B R I N A S E I F F E R T 1, J E N N I F E R S C H M I D T 2
, M O N I K A M . M E S S M E R 3 and
KLAUS-PETER WILBOIS2 1
Division of Plant Breeding, Section of Genetic Resources and Organic Plant Breeding, Georg-August-Universit€at, Von-Siebold-Str. 8, G€ottingen D - 37075, Germany; 2FiBL Deutschland e.V., Kasseler Str. 1a, Frankfurt D-60486, Germany; 3Research Institute of Organic Agriculture (FiBL), Ackerstrasse 113, Frick CH-5070, Switzerland; 4Corresponding author, E-mail:
[email protected] With 2 figures and 4 tables Received November 4, 2016 / Accepted January 18, 2017 Communicated by R. Singh
Abstract Weed competition can severely reduce soybean (Glycine max (L.) Merr.) yields, particularly in organic systems. An efficient screening and breeding approach is needed to increase breeding progress for weed tolerance. This study sought to (i) establish a system for direct selection of competitive genotypes, (ii) evaluate genotypic differences in weed tolerance among six early-maturing genotypes and (iii) assess the contribution of selected morphological traits to weed tolerance. A direct selection system providing two different levels of weed competition through all development stages of a soybean crop was developed, using mixtures of selected crop species as sown competitors. Two resulting mixtures induced intermediate (50%) yield reduction, respectively. This selection system can be applied in screening and breeding programmes to facilitate breeding for weed tolerance. No significant difference in weed tolerance was detected between six soybean genotypes of maturity groups 000 to 00. Morphological traits that might influence competitive ability, for example light absorption, leaf area and lateral shoots, were assessed, and their potential for indirect selection for weed tolerance is discussed.
Key words: breeding — direct selection — soybean — soybean morphology — weed competition — weed tolerance Competition with weeds is a major challenge in organic soybean cultivation. This is especially true in central Europe, where the cool, moist climate creates conditions of high weed pressure (Vollmann et al. 2010). Weed pressure reduces soybean yields, lowers harvesting efficiency, changes seed composition and raises the likelihood of diseased or damaged seed (Vollmann et al. 2010). Weed tolerance is often defined as the ability of a crop to maintain high yields despite the presence of competition (Goldberg and Landa 1991). That strict definition is extended here to include suppressive ability, the ability of crop plants to restrict weed growth by out-competing weed plants for light or nutrients. Suppression may provide longerterm weed control by reducing seed banks, but weed-suppressive traits such as rapid early growth and large leaf area could represent a trade-off with yield or other desirable characteristics like lodging resistance. Tolerance may be more important where weather conditions restrict early tillage (Vollmann and Menken 2012). Thus, both suppression and tolerance are of interest for breeding soybean genotypes with high yields under weed pressure. Breeding soybean genotypes with improved competitive ability are a promising alternative or addition to chemical or
mechanical methods of weed control, especially if used in combination with management techniques such as early sowing and narrow row spacing (Pester et al. 1999). However, the success of breeding programmes depends on the availability of weed-tolerant genotypes, and knowledge of such genotypes is lacking in central Europe. This study sought to (i) establish a direct selection system for weed tolerance, (ii) use this system to evaluate genotypic differences in weed tolerance among six soybean genotypes and (iii) assess whether certain morphological or developmental traits can be used as indirect selection criteria for weed suppression. Direct selection studies under on-station organic farming conditions with natural weed pressure, commonly used to screen for weed tolerance, often fail to provide information about weed tolerance under real-world field conditions. Studies relying on natural weed pressure may provide heterogeneous selection environments, due to spatial and temporal variation in natural weed pressure (Burnside 1972, Gibson et al. 2008). Using a single weed species allows for greater uniformity and control over the level of competitive pressure (Knake and Slife 1969, McWhorter and Barrentine 1975, James et al. 1988), but does not necessarily reflect crop performance against diverse natural weed communities. Soybean crops respond differently to competing species with different morphological and developmental traits (Burnside and Moomaw 1984, Monks and Oliver 1988, Bussan et al. 1997), such that no single species can be used as a universal representative of all weeds. Mixtures of weed species can be used (Gibson et al. 2008), but arbitrary or unclear criteria for species selection may reduce the relevance of the results. The establishment of a direct selection system that uses a mixture of sown competitors to establish consistent competition would more accurately reflect natural weed pressure and could be used to compare genotypic differences across environments. Indirect selection for traits related to weed tolerance could be quicker and more efficient than direct selection, but there is a lack of consensus over the most suitable traits to use. Plant height (Jannink et al. 2000), rapid and dense canopy development (Pester et al. 1999), leaf surface area (Jordan 1993) and light interception (Yelverton and Coble 1991) have all been proposed as predictors of weed competitiveness. However, some studies have found no significant correlation between competitive ability and canopy area, early groundcover, height or volume (Bussan et al. 1997, Norsworthy and Shipe 2006). Leaf
Breeding for weed tolerance in soybean
373
shape could be predicted to influence weed development due to its effects on light absorption and juvenile development. Lanceolate leaf morphology allows a greater proportion of light to pass to lower leaves and leads to better yield stability, whereas ovate leaves absorb a greater proportion of light and show faster juvenile development (Rotzler et al. 2009). In a three-year field trial, this study established a direct selection method for the selection of genotypes with high weed tolerance that can be upscaled in future soybean breeding programmes. A novel direct selection method was developed by creating two mixtures of sown competitors providing two levels of competitive pressure (strong or intermediate) consistently over the course of the growing season.
Materials and Methods Study site: The study was conducted over three years at Reinshof experimental farm near G€ottingen, Germany (N51° 300 5.796″ E9° 550 17.047″) at 150 m above sea level. Soils are a fertile silty loam derived from an alluvial loess layer >2 m deep. Fields used in this experiment were under certified organic management. Precrops in 2011 were winter rye (2010) and field pea (2009); in 2012 winter rye (2011) and spring wheat (2010); and in 2013 winter wheat (2012 and 2011). The usual crop rotation is a clover–grass mix followed by winter wheat, pea, winter rye and spring wheat. Climate data for G€ottingen 2001–2013 show an average annual temperature of 9.5°C and a mean annual precipitation of 690 mm. Genetic material: Soybean genetic material was obtained from multiple sources and included three breeding lines as well as three commercially available cultivars (Table 1). Genotypes were selected for earliness and morphological variation. Genotypes No. 78 and No. 82 represented lanceolate leaf morphology, whereas ‘Merlin’, ‘Proteix’ and ‘Klaxon’ had ovate leaves, and No. 73 was intermediate. In total, ten plant species (winter rye [Secale cereale L.], winter canola [Brassica napus L.], camelina [Camelina sativa L.], mustard [Sinapis alba L.], buckwheat [Fagopyrum esculentum L.], einkorn wheat [Triticum monococcum], spring wheat [Triticum aestivum L.], foxtail millet [Setaria italica L.], Phacelia tanacetifolia Benth. and forage chicory [Cichorium intybus L.]) were used as competitors to simulate weed pressure. Seed was obtained from multiple providers in all years (Table 2). Experimental design in 2011 to develop a direct selection system: In 2011, five plant species (winter rye, winter canola, camelina, mustard, buckwheat) were sown together with soybean cultivars ‘Merlin’ and ‘Proteix’ to simulate weed pressure. A split–split-plot design with two replicates was used with weed treatment (five competitor species and a weed-free control) as whole plot, weed sowing time as split-plot factor and soybean cultivar as split–split factor. Five additional species (einkorn wheat, spring wheat, foxtail millet, Phacelia and forage chicory) and the natural weed community were tested without replication. Plots were established using Øyjord technology at a size of 5 m2 (1.5 9 3.33 m) with four rows spaced 30 cm apart. Soybean seeds were sown on 11 May 2011 with a sowing density of 69 seeds/m2, and competitors in the Table 1: Origin of six soybean genotypes of early maturity groups 00 and 000 tested for weed tolerance Genotype Klaxon Merlin Proteix No. 73 No. 78 No. 82
Breeder Cerom Canada/RAGT France Saatbau Linz, Leonding, Austria Agroscope, Changins/Delley Samen and Pflanzen DSP, Switzerland Agroscope, Changins, Switzerland Agroscope, Changins, Switzerland Agroscope, Changins, Switzerland
Maturity group 000 000 00 000/00 000/00 000/00
Year of release 2005 1997 2009 – – –
Table 2: Origin of 10 species used to simulate weed competition in soybean Species Buckwheat (Fagopyrum esculentum L.) Forage chicory [Cichorium intybus L.] cv. Puna Einkorn wheat [Triticum monococcum L.] Foxtail millet (Setaria italica L. cv. Herbstfeuer) Mustard [Sinapis alba L.] Phacelia tanacetifolia Benth. Spring wheat (Triticum aestivum L. cv. Passos) Camelina [Camelina sativa L.] Winter canola (Brassica napus L. cv. Express) Winter rye (Secale cereale L. cv. Vitallo)
Origin Camena, Lauenau, Germany Olssons Fr€o, Helsingborg, Sweden Division of Plant Breeding Dreschflegel GbR, Witzenhausen, Germany Dreschflegel GbR, Witzenhausen, Germany Dreschflegel GbR, Witzenhausen, Germany Seed saved from Reinshof Division of Plant Breeding Inbred line, Division of Plant Breeding KWS Saat AG, Einbeck, Germany
simultaneous sowing treatment were sown at the same time also at a sowing density of 69 seeds/m2 and at a depth of 4 cm. Soybean seed was inoculated immediately before sowing with NPPL Hi-Stick inoculant according to manufacturer’s instructions (Becker Underwood, Germany). Competitor species in the delayed treatment were sown on 1 June 2011 at the unifoliate leaf stage of the soybean plants at a density of 270 seeds/m2. Competitor and soybean height were measured on two dates (5 July 2011 and 2 August 2011) as well as lodging, maturity, soybean yield and seed weight. Experimental design in 2012 and 2013 to validate a direct selection system: In 2012 and 2013, the experiment was laid out in a split-plot design with genotype as the main plot factor and competitor treatment as subplot. The six genotypes ‘Klaxon’, ‘Merlin’, ‘Proteix’, No. 73, No. 78 and No. 82 (Table 1) were sown in two replicates in 2012 and in four replicates in 2013. The same plot design was used as in 2011. In both years, spring wheat was planted around the soybean plots to divert field mice. The three competitor treatments were Phacelia mixture (PM; Phacelia, buckwheat and winter canola), grain mixture (GM; spring wheat, winter rye and foxtail millet) and weed-free control. In 2012, 69 soybean seeds and 69 competitor seeds (23 seeds per competitor species) were sown per m2; in 2013, 70 soybean seeds and 23 seeds per competitor species and m2 were sown. In 2012, the field was cultivated with a rotary harrow before planting. In 2013, it was worked with a combination of a spring-tine harrow and roller harrow on April 9 and before sowing. Seeds were inoculated immediately before sowing with NPPL Hi-Stick inoculant according to manufacturer’s instructions (Becker Underwood, Germany). Seeds were sown on May 2 in both years. After sowing, rolling was applied on 3 May 2013. The entire study site was covered with netting from 2 May to 21 May 2012 and 6 May to 21 May 2013. In 2012, a wheel hoe was used on May 22 and all plots were weeded once by hand over the course of June. In 2013, hand weeding occurred twice, at the end of May and the end of June, and a wheel hoe was used on June 6 on all plots. Harvesting was performed in both years using a Hege 160 plot combine (AgrarTest Gmbh, Germany), at 160 and 169 days after sowing (DAS) in 2012 and at 145 DAS in 2013. In 2012, ‘Proteix’ and No. 78 were harvested 9 days later than the other genotypes due to delayed ripening. In 2012, the following characteristics were measured: soybean and total aboveground (soybean + competitor) dry matter at 56 DAS by harvesting 1 m2 within each plot, height at 51 DAS and yield in dry mass at maturity (160–169 DAS). Dry matter was determined after drying at 105°C. In
374
B. HORNEBURG, S. SEIFFERT, J. SCHMIDT et al.
2013, more characteristics were recorded: soybean and total aboveground (soybean and competitor) dry matter (45 DAS), soybean height (72, 97, 115, and 139 DAS), light absorbance (42, 61 and 85 DAS), leaf surface area (average of 10 leaves) (113–119 DAS), yield (145 DAS) and number of lateral shoots with pods (145 DAS). Light absorption measurements were conducted using AccuPar LP-80 (Decagon Devices, Inc., USA) in 2013. Leaf surface area was calculated as the average of 10 leaves from the middle of plants in the centre of control plots. Relative yield was calculated by dividing yield for a given treatment by the yield of the weed-free control of the same genotype and replicate. Statistical methods: Analysis of variance was conducted in JMP 5.0.1 (SAS Institute Inc) using the EMS (expected mean square) method. For the 2011 data, a split–split-plot model was used with weed treatment as a whole plot factor, weed sowing time as split-plot factor, soybean cultivar as split–split factor and replication as a random. For the 2012 and 2013 data, a split-plot model was used with genotype as a main plot factor, competitor treatment as a subplot factor and replication as a random factor. The soybean biomass data from 2012 were transformed with an inverse transformation to achieve normally distributed residuals. Pairwise comparisons were conducted using the Tukey HSD (honestly significant difference) function in JMP, with a familywise error rate of 5%. Correlations between yield relative to the weedfree control and proposed indirect selection criteria were determined using Spearman’s rank correlation in JMP.
Results Establishment of a direct selection system Pairwise sowing in 2011 was used to determine the level of competition provided by different competitor species. Yields of soybean varied significantly by competitor species and sowing date in 2011 (Table 3). Delayed sowing of the competitor species did not result in statistically significant soybean grain yield reduction, whereas clear differences were detected by simultaneous sowing. There was no significant genotype x competitor species interaction; this was expected because only two soybean cultivars were used at this stage. Averaged across cultivars buckwheat (54% yield reduction in relation to weed-free control) and mustard (49%) were strongest competitors using simultaneous sowing. In contrast, winter rye (5%) and camelina (11%) had no significant effect on soybean grain yield. In the unreplicated trial, Phacelia showed highest yield reduction (70%); chicory and einkorn wheat had little effect on soybean yield (Table 3).
Using these results, two competitor mixtures of different species were designed to provide different levels of competition: Phacelia mixture (PM; Phacelia, buckwheat and winter canola) and grain mixture (GM; spring wheat, winter rye and foxtail millet). The mixtures were chosen to provide consistent competition over the course of the growing season, with early competition from winter rye and winter canola; mid-season competition from spring wheat, foxtail millet, buckwheat and Phacelia; and late season competition from buckwheat and Phacelia. PM was designed to provide strong competition and GM medium competition. Validation of a direct selection system The effectiveness of this system was shown in 2012 and 2013: Early biomass and grain yield of soybean were strongly affected by competition (Fig. 1). In 2012, GM produced more aboveground competitor biomass than PM (75.62 g/m2 GM vs. 63.83 g/m2 PM), but the order was reversed in 2013 (14.05 g/ m2 GM vs. 18.73 g/m2 PM). PM reduced soybean grain yields to 44% of control in 2012 and 42% in 2013 as compared to 71% and 75%, respectively, for GM. The percentage relative yield for each competitor mixture was similar across years, despite large variation in absolute biomass for each year. Competition was observed during the entire growing season: early competition was mainly induced by the big rosettes of winter canola and heavily tillering winter rye, respectively. Spring wheat did compete with soybean in mid-season but was outcompeted towards the end of the season. Buckwheat and Phacelia did grow taller than soybean and induced heavy competition until full maturity of soybean. In 2012, buckwheat was heavily damaged by frost in May. In both years, only a few plants of foxtail millet developed due to its slow early plant development. Genotypic differences in relative yield The six genotypes tested in 2012 and 2013 did not differ significantly in seed yield relative to the weed-free control (=100%) (Fig. 2), either with individual competitor mixtures (PM, GM) or on average. High relative yields across years and competitor mixtures were achieved by No. 73, No. 78 and ‘Proteix’, with ‘Klaxon’ performing consistently poorly. Averaged across years, competitor biomass was highest in ‘Klaxon’ and lowest in No.
Table 3: Soybean seed yield (dry matter g/m2) for two cultivars and 10 competitor species with simultaneous and with 21 days delayed competitor sowing 2011
Cultivar Merlin Merlin Merlin Merlin Merlin Merlin Cultivar mean (g/m2) Merlin Merlin Merlin Merlin
Competitor species Control Mustard Buckwheat Camelina Winter canola Winter rye Einkorn wheat Foxtail millet Natural weeds Phacelia
Yield (g/m2) Simultaneous Sowing
Yield (g/m2) Delayed Competitor Sowing)
426a 174d 156d 374ab 322abc 365abc 303 382 275 281 127
434a 416a 364a 423a 426a 409a 412 418 402 267 358
Cultivar
Competitor species
Proteix Proteix Proteix Proteix Proteix Proteix
Control Mustard Buckwheat Camelina Winter canola Winter rye
Proteix Proteix
Chicory Spring wheat
Yield (g/m2) Simultaneous Sowing
Yield (g/m2) Delayed Competitor Sowing)
Mean yield across cultivars (g/m2) simultaneous sowing
429a 260bcd 239cd 385ab 428a 447a 365 475 301
442a 414a 392a 406a 411a 446a 419 466 420
427.5a 217.0bc 197.5c 379.0a 375.0ab 406.0a
Values followed by the same superscript letter do not differ significantly at P = 0.05 according to the Tukey HSD for all 12 values per sowing date. Additional competitor species tested against one cultivar in one replication were not included in the statistical analysis.
Breeding for weed tolerance in soybean
375
400
Dry aboveground biomass/seed yield (g/m2)
GM
PM
a
Control
350 300
b
250 200
a c
150
b
100 50 0
a a
b b
a
Soybean 2012
b b a
c
b
b a
c
Soybean Competitor Competitor Soybean Soybean 2013 2013 2012 grain 2012 grain 2013
Fig. 1: Early soybean and competitor dry aboveground biomass and soybean seed yield in 2012 and 2013 in a weed-free control and with two mixtures of competitor species (GM and PM). Soybean and competitor biomass were measured at 56 (2012) and 45 (2013) days after sowing and grain yield at maturity. Error bars represent standard error. Bars designated with the same letter are not significantly different according to the Tukey HSD with p = 0.05
100
80 70 60 50 40
Discussion
10
No. 73 No. 78 No. 82 Klaxon Merlin Proteix
20
No. 73 No. 78 No. 82 Klaxon Merlin Proteix
30 No. 73 No. 78 No. 82 Klaxon Merlin Proteix
Relative soybean seed yield (%)
90
biomass when grown with competitors (only 2013), light absorption, leaf surface area, number of lateral shoots and soybean yield (Table 4). Height differed significantly among genotypes in 2012 (early plant growth) and in 2013 (mid-season and late development) (Table 4). Light absorption differed significantly on the third of three measurements in 2013 (Table 4). The first measurement showed a similar trend in the order of genotypes but was not statistically significant (P = 0.10). Leaf surface area, also measured only in 2013, was significantly different between genotypes (Table 4). ‘Proteix’ and No. 73 had the greatest leaf surface area (0.0137, 0.0136 m2), followed by No. 78, No. 82, ‘Merlin’ and ‘Klaxon’ (0.0127, 0.0114, 0.0100, and 0.0082 m2). Number of lateral shoots differed significantly in 2013; lateral shoot production was not measured in 2012 (Table 4). ‘Klaxon’ produced with 3.5 the most lateral shoots by far, with No. 78 and ‘Proteix’ producing the fewest (1.8, 1.5). Early soybean biomass differed significantly between ‘Klaxon’ (18.54 g/m2) and No. 73 (13.23 g/m2) in 2013. In 2013, early aboveground soybean biomass when grown with competitors was highest for ‘Klaxon’ (16.82 g/m2) and lowest for No. 73 (11.68 g/m2) and ‘Merlin’ (Table 4). Although these characteristics have been proposed as criteria for indirect selection for genotypes with high weed tolerance, none of the above parameters were significantly correlated with relative yield.
2012
2013
Mean
0
The goals of this study were to (i) establish a system for the direct selection of competitive genotypes, (ii) evaluate genotypic differences in weed tolerance among six early-maturing genotypes and (iii) assess the contribution of selected morphological traits to weed tolerance.
600.00
Establishment of a direct selection system
400.00 300.00
0.00
No. 73 No. 78 No. 82 Klaxon Merlin Proteix
100.00
No. 73 No. 78 No. 82 Klaxon Merlin Proteix
200.00 No. 73 No. 78 No. 82 Klaxon Merlin Proteix
Weed dry biomass (g)
500.00
2012 (56 DAS)
2013 (45 DAS)
Mean
Fig. 2: Soybean seed yield relative to the weed-free control (=100%; top) and early aboveground competitor dry biomass 56 (2012) and 45 (2013) days after sowing (bottom) for six soybean genotypes. Error bars represent standard error
73 and ‘Proteix’. However, genotypic differences were not significant in either year or on average (Fig. 2). Relative yield was not significantly correlated with absolute yield. Indirect selection criteria Significant differences were observed among genotypes in plant height, early soybean biomass (only 2013), early soybean
The first goal was achieved with the direct selection system established based on results obtained in 2011 and validated in 2012 and 2013. The proposed PM and GM competitor mixtures provided two stress levels of consistent competition in early, mid and late season. This is an advantage to previous studies, which simulated weed pressure with a single sowing winter oilseed rape (Vollmann et al. 2010). An environment of continuous weed pressure is appropriate for evaluating weed tolerance, as it closely resembles natural conditions. In central Europe, natural weeds in soybean fields include morphologically and phenologically diverse species like Chenopodium album L., Atriplex spec., Amaranthus spec., Galinsoga spec., Convolvulus arvensis L., Polygonum spec., Galium aparine L., Galeopsis spec., Agropyron repens (L.) P.B., Echinochloa crusgalli (L.) P.B. and Setaria spec. (own observations, personal communication with soybean growers and extension services). The fact that yield depression for each competitor mixture was similar across years, despite large variation in absolute biomass for each year, is an indication of the applicability of this selection system in diverse environments. Still, the direct selection system discussed here has not been validated in other regions. Weed communities vary largely depending on local pedoclimatic conditions, farming methods and crop rotations. We do think, however, that a careful selection of the components of a
Lateral shoots, leaf surface area and light absorption were measured in the control treatment only. Soybean biomass with competitor is measured as the average of the two treatments with competitor. Soybean yield is given as the mean of all treatments including control. Values followed by the same superscript letter do not differ significantly at P = 0.05 according to the Tukey HSD. DAS = days after sowing. *P = 0.05; **P = 0.01; ***P = 0.001; NS = not significant.
237.36b 261.26b 261.94b 246.86b 276.18ab 302.83a 264.41 0.0004***
