Pre-attachment Striga hermonthica resistance of New Rice for Africa (NERICA) cultivars based on low strigolactone production

Authors


Author for correspondence:
Harro J. Bouwmeester
Tel: +31 317 489859
Email: harro.bouwmeester@wur.nl

Summary

  • Striga hermonthica (Striga) is an obligate hemiparasitic weed, causing severe yield losses in cereals, including rice, throughout sub-Saharan Africa. Striga germination depends on strigolactones (germination stimulants) exuded by the host roots. The interspecific New Rice for Africa (NERICA) cultivars offer a potentially interesting gene pool for a screen for low germination-inducing rice cultivars.
  • Exudates were collected from all NERICA cultivars and their parents (Oryza sativa and Oryza glaberrima) for the analysis of strigolactones. In vitro and in situ Striga germination, attachment and emergence rates were recorded for each cultivar.
  • NERICA 1 and CG14 produced significantly less strigolactones and showed less Striga infection than the other cultivars. NERICAs 7, 8, 11 and 14 produced the largest amounts of strigolactones and showed the most severe Striga infection. Across all the cultivars and parents, there was a positive relationship between the amount of strigolactones in the exudate and Striga germination, attachment and emergence rates.
  • This study shows that there is genetic variation in Striga pre-attachment resistance in NERICA rice. Cultivars combining this pre-attachment resistance with post-attachment resistance (already identified) can provide a key component for durable integrated management of this noxious weed in cereal production systems in sub-Saharan Africa.

Introduction

Rice is an increasingly important staple crop in sub-Saharan Africa (Balasubramanian et al., 2007). Over the last three decades, the harvested area has increased by c. 105%, whereas production has increased by 170% (FAO, 2008). However, the current domestic production still only covers c. 40% of the regional rice consumption (Anonymous, 2007). The Africa Rice Center and partners have recently developed a collection of interspecific upland rice cultivars, named NEw RICe for Africa (NERICA) cultivars. These cultivars were developed with the aim of combining the high yields from Asian rice species Oryza sativa (WAB56-104, WAB56-50 and WAB181-18) with the ability of the African species Oryza glaberrima (CG14) to resist local stresses (Jones et al., 1997a,b). To date, 18 interspecific upland cultivars are available to rice farmers. They are popular among rice farmers in the region and are partly responsible for the recent increase in rice area under rain-fed upland conditions (Rodenburg et al., 2006c; Balasubramanian et al., 2007; Wopereis et al., 2008).

However, average rain-fed rice productivity in the region is low as a result of a myriad of production constraints (Balasubramanian et al., 2007), of which weed competition is the most severe (Rodenburg & Johnson, 2009). Parasitic plants of the Orobanchaceae, such as Striga spp., are becoming troublesome weeds in rain-fed rice in sub-Saharan Africa (Rodenburg et al., 2010). Of the Striga genus, Striga hermonthica (Del.) Benth. is the most damaging species. It is an obligate hemiparasite, attacking the roots of several cereal crop species, leading to severe yield losses (Parker, 1991). Striga spp. only germinate on exposure to host root-derived chemicals, such as strigolactones (Bouwmeester et al., 2003), which are apocarotenoid signalling molecules (Matusova et al., 2005) released by the host plant into the rhizosphere. After seed germination has been triggered, the radicle of the germinating seed penetrates the host root and forms a haustorium to establish a xylem–xylem connection with the host to withdraw water and nutrients. Subsequently, the parasite emerges above the soil surface, flowers and produces thousands of seeds. Most of the damage to the host occurs between attachment and emergence. Hence, a promising opportunity to minimize losses would be to avoid the triggering of Striga seed germination, for instance by reducing strigolactone production.

Effective Striga control should be based on multiple, simultaneously applied, strategies. In such an integrated approach, the use of resistant cultivars could be one cost-effective element (Scholes & Press, 2008). Although genetic variation in Striga resistance has been reported in rice in several studies (Harahap et al., 1993; Johnson et al., 1997), only few adapted resistant rice cultivars have been identified to date (Rodenburg et al., 2010). Moreover, in contrast with other host crops, such as maize and sorghum, little is known about Striga resistance mechanisms in rice. Only two mechanisms have been identified so far: a post-vascular connection resistance (Yoshida & Shirasu, 2009) and an incompatibility reaction (Gurney et al., 2006). Both post-attachment resistance mechanisms were found in the rice cv Nipponbare which is not adapted to upland rice-growing environments.

The existence of pre-attachment resistance, as found in other cereals, is yet to be confirmed in rice. As S. hermonthica germination is dependent on the quantity and quality of strigolactone production (Sun et al., 2007; Jamil et al., 2010, 2011), genetic variation in this trait could potentially confer pre-attachment resistance. One of the reasons for the relatively slow progress in the identification of resistant materials and resistance mechanisms is probably the lack of simple and effective screening methods (Ejeta, 2007).

The current study aims to determine pre-attachment Striga resistance in rice cultivars and to discover whether this resistance is based on strigolactone production. To this end, the complete set of 18 upland cultivars of NERICA and their parents were screened for strigolactone production and Striga infection characteristics, such as germination, attachment, emergence and Striga dry biomass. The development of a strigolactone analysis-based screening method and the identification of resistant germplasm would benefit breeding efforts targeted at the development of Striga-resistant cultivars. Furthermore, on confirmation of effectiveness in the field, the identification of germplasm with efficient pre-attachment resistance could directly benefit rice farmers in Striga-prone areas. This study was carried out in parallel with a study on post-attachment resistance in the NERICAs (Cissoko et al., 2011). The combination of pre- and post-attachment resistance mechanisms against Striga spp. into new varieties could lead to more durable resistance against the scourge of sub-Saharan Africa.

Materials and Methods

Germplasm and chemicals

Seeds of 18 upland NERICA cultivars and their O. sativa L. parents WAB56-104, WAB56-50, WAB181-18 and O. glaberrima Steud. parent CG14 were provided by the Africa Rice Center. Seeds of S. hermonthica (Delile) Benth. used in the attachment study were collected from a sorghum field near Cinzana, Mali (courtesy of Cheickna Diarra), and S. hermonthica seeds used in germination bioassays with rice root exudates were collected from a sorghum field near Wad Medani, Sudan (courtesy of Professor Abdel Gabar Babiker). Striga seed viability was c. 60–70%. A standard of orobanchol was provided by Koichi Yoneyama (Weed Science Center, Utsunomiya University, Utsunomiya, Japan), 2′-epi-5-deoxystrigol by Kohki Akiyama (Osaka Prefecture University, Osaka, Japan) and D6-2′-epi-5-deoxystrigol (which was synthesized as described by Ueno et al. (2010)) by T. Asami (Department of Applied Biological Chemistry, The University of Tokyo, Tokyo, Japan).

Analysis of root exudates

For the collection of exudates, 25 germinated seeds of each rice cultivar were planted in a 3-l plastic pot filled with 1.5 l of sand. After 1 wk, plants were thinned to 20 plants per pot. Half-strength modified Hoagland’s nutrient solution with corresponding phosphorus concentration was applied to each pot (500 ml at 48-h intervals). The plants were allowed to grow in a climate room (supplemented with artificial lighting, 450 μmol m−2 s−1) under controlled conditions (28°C (light) 10 h : 25°C (dark) 14 h, at 70% relative humidity) for 4 wk. In the fifth week, phosphorus deficiency was created in each pot to increase strigolactone production (Lopez-Raez et al., 2008). Three litres of phosphorus-deficient nutrient solution (half-strength modified Hoagland’s nutrient solution minus phosphate) were added to the top of each pot and allowed to drain freely through the holes in the bottom of the pot to remove phosphorus from the sand. The plants were kept under phosphorus deficiency for 1 wk. In the sixth week, the same draining with 3 l of phosphorus-deficient nutrient solution was again added to remove any accumulated strigolactones. Finally, 48 h later, root exudates were collected in a 1-l plastic bottle by passing nutrient solution without phosphate through each pot. The collected root exudates were then run through an SPE C18 column (500 mg per 3 ml) and strigolactones were eluted using 6 ml of 100% acetone.

The strigolactones orobanchol, 2′-epi-5-deoxystrigol and three methoxy-5-deoxystrigol isomers (Fig. 1) (C. Cardoso et al., unpublished) were identified and quantified using ultra-performance liquid chromatography tandem mass spectrometry (UPLC-MS/MS) as described previously by Jamil et al. (2011). The samples were analysed by a Waters Xevo triple quadrupole tandem mass spectrometer (Waters, Milford, MA, USA) equipped with an electrospray ionization (ESI) source and coupled to an Acquity UPLC system (Waters). Multiple reaction monitoring (MRM) was used for the quantification of strigolactones in rice root exudates. Data acquisition and analysis were performed using Mass Lynx 4.1 (TargetLynx) software (Waters). The biological activity of the exudates of each cultivar was studied using a Striga germination bioassay with pre-conditioned S. hermonthica seeds, as described previously (Matusova et al., 2005).

Figure 1.

Chemical structure of orobanchol, 2′-epi-5-deoxystrigol and methoxy-5-deoxystrigol in the root exudates of NEw RICe for Africa (NERICA) cultivars and their parents.

Striga hermonthica infection

Plastic pots (1.5 l), with a perforated plastic sheet in the bottom, were filled with 100 ml of clean, Striga-free river sand. On top of this, 500 ml of sand mixed with 25 mg (50 000 seeds) of S. hermonthica seeds were added. One pre-germinated seed of each rice cultivar was planted in the middle of each pot and covered by another 100 ml of Striga-free sand. Plants were grown in a temperature-controlled glasshouse, where natural sunlight was supplemented with artificial light (28°C (light) 10 h : 25°C (dark) 14 h, with 70% relative humidity). Half-strength modified Hoagland’s nutrient solution was applied in the first week (250 ml at 48-h intervals). For the remainder of the experiment, a nutrient solution with 20% phosphorus was applied to stimulate strigolactone exudation (250 ml to each pot at 48-h intervals). At 8 wk after planting (WAP), rice plants were carefully removed from the sand, the roots were carefully washed and S. hermonthica attachments were counted under a stereomicroscope.

Striga hermonthica performance and rice tillering

The third experiment was conducted to determine the final expression of resistance of all cultivars to S. hermonthica in terms of parasite emergence and biomass production. To this end, 25 mg of S. hermonthica seeds were thoroughly mixed in 500 ml of a (1 : 1) potting compost (Lentse Potgrond)/sand mixture which was placed in a plastic pot (1.5 l) with a perforated plastic sheet in the bottom. A pre-germinated seed of each rice cultivar (three replicates) was planted in the middle of each pot at 2 cm depth and water was applied regularly (250 ml at every 48-h interval), allowing excess water to drain. The plants were grown in a glasshouse, supplemented with artificial lighting under controlled conditions (10 h at 28°C (light) : 14 h at 25°C (dark) at 70% relative humidity). Starting at 4 WAP, S. hermonthica emergence was recorded weekly for each cultivar for a period of 8 wk. At 12 WAP, all emerged Striga plants were uprooted, oven dried at 70°C for 72 h and weighed to determine the total dry biomass of parasite per rice plant. To assess the tillering phenotype, a pre-germinated seed of each cultivar (in six replicates) was planted in the middle of a pot containing 500 ml of Striga-free sand. Nutrient solution with 20% phosphorus was applied (250 ml at every 48-h interval) and the number of tillers per plant was counted at 12 WAP.

Statistical analysis

The Genstat 9.2 (VSN International Ltd., Hemel Hempstead, Hertfordshire, UK) statistical package was used for ANOVA on Striga germination, attachment, emergence and dry biomass data, followed by a post hoc least-significant difference (LSD) test for comparison of means. To select the strigolactones that contribute significantly to the explanation of the variation in Striga germination, attachment, emergence and dry biomass in the rice cultivars, a stepwise algorithm, based on the Akaike information criterion (AIC), was used to fit a linear regression model (Akaike, 1974). The statistical package R was used for correlation analysis to assess how the amounts of different strigolactones correlated with each other. The peak areas of the strigolactones of the NERICAs and their parents were used in redundancy analysis (RDA), employing CANOCO (ter Braak, 1988) to visualize the distance between samples and correlations between explanatory (strigolactones) and response (germination, attachment, emergence, dry biomass and tiller numbers) variables.

Results

Strigolactone production

In the UPLC-MS/MS (MRM) chromatograms of the root exudates of the NERICAs and their parents, five single intense peaks were detected in the different MRM channels. 2′-Epi-5-deoxystrigol was detected at 12.19 min at MRM channels m/z 331 > 234, 331 > 217 and 331 > 97, orobanchol was detected at 7.29 min at MRM channels m/z 347 > 233, 347 > 205, 347 > 97, and three methoxy-5-deoxystrigol isomers 1–3 were detected at Rt 9.18, 9.78 and 10.33 min at MRM channels m/z 361 > 247 and 361 > 97 (Figs 1, 2). The same strigolactones were also found in the root exudates of the rice cv Nipponbare and IAC 165 (Jamil et al., 2011).

Figure 2.

Liquid chromatography-mass spectrometry (LC-MS) analysis using multiple reaction monitoring (MRM) of rice root exudates. The MRM transitions for 2′-epi-5-deoxystrigol (a, b), orobanchol (c, d), methoxy-5-deoxystrigol isomers 1–3 (e) and total ion current (TIC) (f) obtained for rice cv NERICA 7 root exudates are shown as examples.

A strong quantitative variation in strigolactone production was observed between the NERICA cultivars (Fig. 3; Supporting Information Table S1). WAB56-50 and NERICA cultivars 7, 8, 11 and 14 represented the top five highest strigolactone producers (Fig. 3). NERICA cultivars 6, 10, 15, 2, 9 and 5 showed intermediate production levels, whereas CG14, WAB56-104 and NERICA 1 produced the smallest amount of strigolactones (Fig. 3; Table S1). In addition to differences in the amount of strigolactones, the cultivars also showed differences in the composition of the strigolactone blend. For example, the proportion of strigolactones in the total strigolactone blend varied between cultivars from 6.4% to 50.4% for orobanchol, 25.5% to 60.4% for 2′-epi-5-deoxystrigol, 3.5% to 20.9% for methoxy-5-deoxystrigol isomer 2 and 8.8% to 29.8% for methoxy-5-deoxystrigol isomer 3, indicating substantial genetic variation in strigolactone composition (Fig. 3; Table S1).

Figure 3.

Production of 2′-epi-5-deoxystrigol, orobanchol and methoxy-5-deoxystrigol (DS) isomers 1–3 by NEw RICe for Africa (NERICA) cultivars and their parents (Oryza sativa parents (WAB56-50, WAB56-104, WAB181-18) and Oryza glaberrima parent (CG14)). The purified root exudates were analysed using multiple reaction monitoring-liquid chromatography-mass spectrometry (MRM-LC-MS) (see the Materials and Methods section). Bars represent means of peak areas of the individual strigolactones as determined by MRM-LC-MS in triplicate.

Striga hermonthica germination, attachment, emergence and dry biomass

Considerable variation was observed in Striga germination rates induced by the root exudates of the different cultivars (Table 1). WAB56-50 and NERICAs 7, 11, 14 and 8 induced the highest Striga germination (≥ 40%), whereas others, such as CG14 and NERICA 1, stimulated significantly less Striga seed germination than the majority (18) of the other cultivars. The last two, as well as NERICAs 4 and 3, induced < 25%Striga germination.

Table 1.   Tillers per plant and Striga hermonthica germination, attachment, emergence and dry biomass (means) for rice cultivars
RiceStriga hermonthica
Rice cultivarsTillers per plant (12 WAP)aGerminationAttachment (8 WAP)Emergence (12 WAP)Dry biomass (12 WAP)
NameTillersRankb%RankNo. per plantRankNo. per plantRankmgRank
  1. aWAP, weeks after planting. bRanking was performed from ‘resistant’ to ‘susceptible’, with rank 1 indicating the highest resistance genotype against S. hermonthica infection and 22 the most susceptible; values are the means of three replicates (for Striga) or six replicates (for Tillers). cLSD, Least significant difference of means (5% level); dSED, Standard error of difference of means.

NERICA016.1811.721.722.01982
NERICA025.61039.5175.7911.715128013
NERICA035.11722.946.31410.0117278
NERICA045.41316.736.01110.011132214
NERICA055.31526.865.038.374195
NERICA064.92036.9157.31713.317140215
NERICA074.42250.02111.72219.022225522
NERICA085.01941.5189.32014.319175020
NERICA096.4634.2116.3158.06150616
NERICA105.61037.8166.0128.78155417
NERICA116.9244.6208.71914.319169418
NERICA126.4630.895.378.784416
NERICA135.61035.8145.7109.0102394
NERICA145.31542.0197.71816.021173119
NERICA155.11735.4136.01312.714106911
NERICA164.92027.175.047.04124412
NERICA176.6428.785.057.042353
NERICA185.7934.5126.31610.0116617
WAB56-1046.7326.355.065.7385210
WAB56-505.41354.0229.32113.318183921
WAB181-186.6432.7105.3812.7167939
CG149.416.610.712.01111
P< 0.01 < 0.001 < 0.05  < 0.01 < 0.01 
LSD 5%c0.9 11.4 6.0 8.3 603 
SEDd0.4 5.7 3.0 4.1 299 

Striga attachment was shown to be a less discriminative parameter (< 0.05) than the Striga germination rate (< 0.001), but, nevertheless, the top five cultivars with the highest germination rates (WAB56-50 and NERICAs 7, 8, 11 and 14) also showed the highest Striga attachment (Table 1). However, of the cultivars showing the lowest germination rates, only CG14 and NERICA 1 also had a similar ranking based on attachment, although the number of attachments on CG14 and NERICA 1 was only significantly different from 20% of the other cultivars. NERICAs 3 and 4, which showed low germination rates, had an intermediate number of attachments. For most of the intermediate cultivars, the ranking based on attachment was similar to the position based on the germination rates.

NERICA 7 had the highest Striga emergence, followed by NERICA cultivars 14, 8, 11 and WAB56-50 (> 13 plants per pot) (Table 1; Fig. 4). NERICA 1 and O. glaberrima parent CG14 showed significantly (< 0.01) lower Striga emergence than 55% of the other cultivars. The other parent of NERICA 1, WAB56-104, also showed a relatively low emergence (ranked third with less than six plants per pot) (Table 1; Fig. 4).

Figure 4.

Striga hermonthica emergence in NEw RICe for Africa (NERICA) cultivars (N1–N18), Oryza sativa parents (WAB56-50, WAB56-104, WAB181-18) and Oryza glaberrima parent (CG14).

In line with the emergence data, WAB56-50 and NERICAs 7, 8, 14 and 11 supported the highest S. hermonthica biomass, whereas the lowest Striga biomass was found on CG14 and NERICA 1 (Table 1). The last two supported significantly (< 0.01) lower Striga biomass than 75% of the other cultivars. NERICAs 17, 13, 5 and 12 supported significantly lower Striga biomass than 12 of the 14 least-performing cultivars (Table 1).

Tiller numbers

As there is a relationship between strigolactones and tiller numbers (Umehara et al., 2008), the numbers of tillers for the rice cultivars were assessed. Some of the NERICA cultivars, such as 6, 7 and 16, had less than five tillers per plant, whereas CG14 of the African rice species O. glaberrima had as many as nine tillers per plant (Table 1). The rest of the NERICA cultivars and parents produced between five and seven tillers per plant (Table 1).

Relationship between strigolactones and S. hermonthica infection

Striga hermonthica germination, attachment and emergence correlated positively with the peak areas of 2′-epi-5-deoxystrigol, orobanchol and methoxy-5-deoxystrigol isomers 2 and 3 in the exudates of the NERICAs and their parents (Fig. 5). Striga dry biomass correlated positively with the peak area of orobanchol and 2′-epi-5-deoxystrigol in the root exudate. Methoxy-5-deoxystrigol isomer 1 did not show a significant correlation with any of these parameters. Linear regression showed that the peak areas of 2′-epi-5-deoxystrigol, orobanchol and methoxy-5-deoxystrigol isomer 2 contributed significantly to the explanation of the variation in S. hermonthica germination induced by the NERICAs and their parents (Table 2). Only orobanchol contributed significantly to the explanation of the variation in S. hermonthica attachment and emergence, and orobanchol and 2′-epi-5-deoxystrigol showed a significant contribution to the explanation of variation in S. hermonthica dry biomass (Table 2).

Figure 5.

Relationship between the amounts of 2′-epi-5-deoxystrigol, orobanchol, methoxy-5-deoxystrigol (DS) isomers 1–3 and Striga hermonthica germination, attachment, emergence and dry biomass for all the NEw RICe for Africa (NERICA) cultivars and their parents. The amounts of various strigolactones and S. hermonthica germination, attachment, emergence and dry biomass were log transformed (log10(x)) and related by correlation analysis.

Table 2.   Contribution of strigolactones to the explanation of variation in Striga hermonthica germination, attachment and emergence
 GerminationAttachmentEmergenceDry biomass
  1. *, < 0.05; **, < 0.01; NS, nonsignificant. The strigolactone peak areas were log transformed ((log(x + 0.01)). Linear regression models were fitted to relate S. hermonthica germination (logit transformation), attachment and emergence with the best combination of strigolactones.

2′-Epi-5-deoxystrigol**NSNS*
Orobanchol*******
Methoxy-5-deoxystrigol-isomer 1NSNSNSNS
Methoxy-5-deoxystrigol-isomer 2**NSNSNS
Methoxy-5-deoxystrigol-isomer 3NSNSNSNS

The correlation between the peak areas of the strigolactones was also determined (Table S2). Methoxy-5-deoxystrigol isomers 2 and 3 and 2′-epi-5-deoxystrigol correlated highly significantly (< 0.001) with each other. The correlations between orobanchol, the methoxy-5-deoxystrigol isomers 2 and 3 and 2′-epi-5-deoxystrigol were weaker, but also significant (Table S2). Methoxy-5-deoxystrigol isomer 1 showed a weaker correlation with most of the other strigolactones, possibly because its concentration was close to the detection level (Tables S1, S2).

Classification of cultivars

The peak areas of the strigolactones of the NERICAs and their parents were used in RDA to visualize the distance between samples and the correlations between strigolactones (as explanatory variables) and germination, attachment, emergence and tiller numbers (as response variables) (Fig. 6). The scores and loadings biplot shows clustering of NERICA cultivars and their parents based on strigolactone production. The NERICA parent CG14 and NERICAs 1, 3, 4, 5 and 17 – all low producers of strigolactones – clustered separately to the right of principal component 1 (PC1), whereas high producers, such as WAB56-50 and NERICAs 7, 11 and 14, clustered to the left of PC1. The rest of the NERICA cultivars were found in between these two groups. The loadings plot shows that 2′-epi-5-deoxystrigol, methoxy-5-deoxystrigol isomers 2 and 3 and orobanchol contributed more than methoxy-5-deoxystrigol isomer 1 to the explained variation and separation of samples along PC1 because of the larger loading scores on PC1. Orobanchol contributed more to PC2, explaining more variation and separation of the samples along PC2 than the other strigolactones. The angle between the arrows in the biplot represents the correlation between the corresponding variables, with 0° and 180° indicating the maximum positive and negative correlation and 90° indicating no correlation. Thus, Striga germination correlated best with methoxy-5-deoxystrigol isomers 2 and 3, followed by 2′-epi-5-deoxystrigol and methoxy-5-deoxystrigol isomer 1, but less with orobanchol. Methoxy-5-deoxystrigol isomers 1–3 also showed positive correlations with attachment, emergence and dry biomass of Striga (> 2′-epi-5-deoxystrigol), whereas orobanchol showed the strongest correlation with these parameters. Similarly, strigolactones correlated negatively with tiller numbers. Of the different strigolactones, orobanchol showed the strongest negative correlation with tiller number. Striga attachment, emergence and dry biomass correlated strongly with each other.

Figure 6.

CANOCO redundancy analysis scores and loadings biplot showing clustering of NEw RICe for Africa (NERICA) cultivars (N1–N18) and their parents Oryza sativa (W56-104, W56-50, W181-18) and Oryza glaberrima (CG14) based on the amounts of 2′-epi-5-deoxystrigol (2′-epi-5-DS), orobanchol (Orob) and methoxy-5-deoxystrigol isomers 1–3 (M-1, M-2, M-3) as explanatory variables to explain the variation in the response variables Striga germination (Germ), attachment (Atta), emergence (Emer), dry biomass (Dryb) and host tillers per plant (Tillers). The explanatory variables (strigolactones) are indicated with dashed arrows and the response variables with solid arrows.

Discussion

The present study shows that there is a large variation in pre-attachment Striga resistance in the NERICA collection of rice cultivars derived from crosses between O. sativa and O. glaberrima species because of differences in strigolactone secretion into the rhizosphere. The method used to screen for pre-attachment resistance is based on UPLC-MS/MS analysis, which enables the quantification and identification of Striga seed germination stimulants, the strigolactones, present in the root exudates of Striga hosts. The method tested in this study was shown to be very accurate for the identification of potentially resistant genotypes.

The first critical step in the life cycle of Striga, the germination of its seed, is regulated by the strigolactones (Bouwmeester et al., 2003; Akiyama & Hayashi, 2006). It has been hypothesized previously that such dependence on host-derived signals could be used to the advantage of the crop, for instance through breeding for low germination stimulant-producing cultivars (Lynn & Chang, 1990; Hess et al., 1992; Bouwmeester et al., 2003; Ejeta, 2007). The results of the current study, showing significant variation among NERICA cultivars and their parents for strigolactone production and Striga germination, confirm that such an approach is feasible in rice. The results show that the strigolactone production of a rice cultivar to a large extent explains the level of resistance against Striga. A lower production of strigolactones results in a lower percentage of Striga germination, contributing to a more resistant phenotype.

Based on the present findings, breeding for low strigolactone production is likely to result in a certain degree of Striga resistance. Previously, significant variation was found for the amount of germination stimulant in sorghum genotypes (Netzly et al., 1988; Weerasuriya et al., 1993; Rich et al., 2004). Sorghum genotypes with low production of the germination stimulant have been shown to be resistant to Striga in the field (Ramaiah, 1987; Hess et al., 1992; Ejeta, 2007). Low production of the Striga seed germination stimulant in sorghum is inherited as a single recessive gene (Vogler et al., 1996). The high-yielding sorghum cultivars containing the low germination stimulant (lgs) gene have been shown to be very effective against Striga, and these cultivars have been introduced in many African countries (Ejeta et al., 2000; Ejeta, 2005). For rice, only a few resistant materials and resistance mechanisms have been identified so far (Rodenburg et al., 2010). The present study confirms the existence of lgs as a resistance mechanism in rice cultivars. The methodology tested in the present study showed significant differences in pre-attachment resistance among a group of rice cultivars. The screening of a wider pool of germplasm for the levels of strigolactone production could be helpful for the selection of even more resistant genotypes in the future. The development of cultivars combining resistance with high levels of tolerance (Rodenburg et al., 2005, 2006a; Rodenburg & Bastiaans, 2011), or combining pre- and post-attachment resistance (Cissoko et al., 2011), seems necessary in the near future if varietal control of Striga is to become an important component in integrated management. Such an approach would require screening procedures for individual mechanisms (Rodenburg et al., 2005, 2006a). Post-attachment resistance (Gurney et al., 2006) can, for instance, complement partial resistance because of low strigolactone production, and enhance the durability of the resistance as a result of the multigenic nature of combined resistance mechanisms.

Arbuscular mycorrhizal fungi (AMF), which are important for providing mineral nutrients to plants through symbiosis, also recognize their host plants through strigolactones (Akiyama et al., 2005; Harrison, 2005; Bouwmeester et al., 2007). Hence, symbiosis in low-strigolactone-producing hosts may be affected negatively. Careful attention should therefore be paid to the link between strigolactones, Striga germination and the colonization by AMF. It might be interesting to test whether AMF and Striga are triggered by exactly the same types of strigolactone and, consequently, whether it would be an option to identify cultivars that produce the types of strigolactone that stimulate AMF without triggering Striga germination. In a study by C. Cardoso et al. (unpublished), activity profiling of rice root exudates with HPLC showed that some fractions induce high S. hermonthica germination, but relatively low AMF branching, whereas other fractions show much lower germination stimulatory activity and induce high AMF branching. Akiyama et al. (2010) confirmed that strigolactones, such as 5-deoxystrigol and orobanchol, are more active in inducing AMF hyphal branching than others, such as strigol and sorgomol. By using this knowledge, plant breeders could select cultivars that produce only the desired types of strigolactone that cause maximum AMF hyphal branching, but do not induce Striga germination.

In addition to quantity, the quality and composition of strigolactones in the root exudates of a host might also be important to explain the differences in Striga incidence (Netzly et al., 1988; Siame et al., 1993). The differences in the significance of correlations between various types of strigolactone and Striga infection, as found in the current study, strengthen this assumption (Table 2). Regression analysis shows that 2′-epi-5-deoxystrigol, orobanchol and methoxy-5-deoxystrigol isomer 2 contribute significantly to the explanation of the variation in Striga germination (Table 2), whereas RDA suggests that 2′-epi-5-deoxystrigol and methoxy-5-deoxystrigol isomers 2 and 3 are most important for the induction of Striga germination (Fig. 6). The latter fits with earlier observations in our laboratory that the methoxy-5-deoxystrigol isomers are more active stimulants of Striga germination than is orobanchol (C. Cardoso et al., unpublished). Screening for pre-attachment resistance based on strigolactone production should hence focus particularly on finding germplasm with a low production of these strigolactones. The fact that there is substantial genetic variation in the composition of the strigolactone blend (Table S1; Fig. 3) suggests that this is certainly feasible. Clearly, the biological activities of individual as well as combinations of different strigolactones need to be studied further to clarify the specific function of individual strigolactones in the rhizosphere.

The negative correlation between tillering and strigolactones and Striga infection suggests that higher tillering cultivars have better Striga resistance because of lower strigolactone production. It is remarkable that the arrows for tillering and orobanchol in the biplot almost completely correlate negatively (Fig. 6). Interestingly, orobanchol has been shown recently to be present in the xylem of tomato and Arabidopsis, suggesting that it is at least one of the strigolactones involved in the control of shoot branching (Kohlen et al., 2011). This could suggest that, in rice also, orobanchol is the strigolactone that regulates tillering, assuming that there is a correlation between the amount of orobanchol in the exudate – which we measured – and in the xylem. Intriguingly, the arrows for attachment, emergence and dry biomass show a slightly different direction from the germination arrow in Fig. 6, and these parameters seem to correlate more strongly with orobanchol than with the other strigolactones. This could be explained by the presence of other resistance mechanisms in NERICAs 2, 9, 10, 12, 13 and 17 and WAB56-50 and CG14, which cluster on the low end of PC2, that are absent in NERICAs 3, 4 and 7, clustering at the high end of PC2, rather than a causal relation with orobanchol. For some of the former cultivars, Cissoko et al. (2011) indeed showed that they have increased post-attachment resistance. Therefore, it is clear that screening for Striga resistance in rice should not only be based on strigolactone production, but should also examine post-attachment resistance and the expression of these mechanisms in the field. This is predictable as not all of the germinated seeds will establish attachment and, also, not all of them will emerge from the soil. Clearly, field testing of germplasm selected through screening methods as described here, and by Cissoko et al. (2011), would still be necessary before the formulation of recommendations to farmers or breeders. For this field testing, other factors should be closely monitored. For example, phosphorus deficiency causes increased production of strigolactones (Yoneyama et al., 2007a,b; Lopez-Raez et al., 2008) and a pre-attachment resistance trait might not be, or not as easily be, identified in field screening trials with high phosphorus application. Similarly, the Striga seed bank density in the soil might affect the apparent pre-attachment resistance. Cultivars with pre-attachment resistance may become ineffective under high Striga seed bank density, and hence fail to show their resistance phenotype because of heavy soil infestation (Ejeta, 2007).

In addition to pre-attachment resistance against Striga, associated with lgs production, other resistance mechanisms, such as low production of the haustorial initiation factor (lhf), or an incompatibility response (ir) to Striga parasitism by the host, could act as additional factors in the selection of resistant genotypes (Ejeta et al., 2000). For haustorial initiation, the parasite requires an additional host signal and, on host roots of genotypes possessing lhf resistance, germinated parasitic seeds fail to form haustoria and to attach to their potential host and, consequently, die (Riopel & Timko, 1995). Similarly, because of the incompatibility reaction in some genotypes, penetrated Striga seedlings show stunted growth and stop further development after the first emergence of leaves as a result of a lack of nutrients from the host. The incompatibility reaction mainly occurs as a result of inadequate xylem–xylem connections between host and parasite. In addition, in rice, post-attachment Striga resistance has been identified (Gurney et al., 2006; Yoshida & Shirasu, 2009). The symptoms for this are necrosis around the site of parasite attachment and inability of the parasite to penetrate the endodermis. In addition, in other plant species, other mechanical barriers, such as lignification (Maiti et al., 1984), cellulose accumulation (Olivier et al., 1991) and encapsulation of the parasite (Labrousse et al., 2001), have been found to be responsible for failure of the connection of the parasite’s vascular tissue to that of the host. Strigolactone production correlates well with the level of resistance against Striga, suggesting that germination is an important component of resistance. However, other parameters affecting the attachment and emergence success further determine whether or not successful Striga–host relations are established. It is interesting to note that some of the cultivars that produce small amounts of strigolactones also exhibit good post-attachment resistance to several different Striga ecotypes (Cissoko et al., 2011).

In conclusion, pre-attachment Striga resistance, caused by low induction of Striga germination as a result of low strigolactone production, exists in the NERICA collection. This pre-attachment resistance can be determined through the identification and quantification of strigolactones present in the root exudates by LC/MS analysis. Based on this information, a rice cultivar with pre-attachment Striga resistance can be identified and recommended to breeders or farmer communities. The proposed screening method might prove to be a rapid and efficient approach that could be highly relevant for future screening and breeding programmes, and help breeders to screen for pre-attachment resistance for the development of Striga-resistant cultivars. As Striga populations are genetically diverse, for durable resistance, it would be advisable to combine the pre-attachment resistance reported in this article with the post-germination resistance mechanisms reported by Cissoko et al. (2011).

Acknowledgements

Funding through the Higher Education Commission (HEC to M.J.) Pakistan and the Netherlands Organization for Scientific Research (to H.J.B., VICI grant 865.06.002 and Equipment grant 834.08.001) is kindly acknowledged. This project is (co) financed by the Centre for BioSystems Genomics (CBSG), which is part of the Netherlands Genomics Initiative/Netherlands Organization for Scientific Research. We thank the Africa Rice Center for providing seeds of NERICA and their parents. We also thank Cheickna Diarra and Abdel Gabar Babiker for providing S. hermonthica seeds. For useful suggestions and comments, we acknowledge Aad van Ast (CCSA, Wageningen University). We are also thankful to Binne Zwanenburg (Department of Organic Chemistry, Radboud University, Nijmegen, the Netherlands), Koichi Yoneyama (Weed Science Center, Utsunomiya University, Japan) and Kohki Akiyama (Osaka Prefecture University, Japan) for kind provision of GR24, standards of orobanchol, 2′-epi-orobanchol, 5-deoxystrigol and 2′-epi-5-deoxystrigol. For valuable help and comments on RDA, we are grateful to Benyamin H. Houshyani (Laboratory of Plant Physiology, Wageningen University, the Netherlands).

Ancillary