Background

Cyclamen persicum Mill. (cyclamen) is a popular and economically significant pot plant crop in Japan, Germany, Italy, the Netherlands and North America. Flower colour in commercial lines ranges from white, through red, pink, reddish-purple to purple. The pigments present are predominantly anthocyanins and there have been several studies on anthocyanin and flavonoid pigmentation in cyclamen [1–5]. The main anthocyanins are 3,5-di-O-glucosides of peonidin, cyanidin and malvidin (Figure. 1). There are two missing colour groups in cyclamen, the orange-red of pelargonidin-derived anthocyanins [6] and blue, even though some delphinidin-derived anthocyanins often associated with blue flower colours are present in maroon to purple cultivars [1–3, 6].

Figure 1
figure 1

A simplified version of a section of the flavonoid biosynthetic pathway. The flavonols kaempferol, quercetin and myricetin are formed from dihydrokaempferol, dihydroquercetin and dihydromyricetin, respectively, by flavonol synthase (FLS). The double arrows show the points of possible action of multiple enzymes for formation and modification of the anthocyanins. Abbreviations are as follows: CHS, chalcone synthase; CHI, chalcone isomerase; F3H, flavanone 3-hydroxylase; DFR, dihydroflavonol 4-reductase; ANS, anthocyanidin synthase; F3'H, flavonoid 3'-hydroxylase; F3'5'H, flavonoid 3',5'-hydroxylase; F3GT, flavonoid 3-O-glucosyltransferase; A5GT, anthocyanin 5-O-glucosyltransferase; A3'OMT, anthocyanin 3'-O-methyltransferase; A3'5'OMT, anthocyanin 3',5'-O-methyltransferase. The numbering of the 3', 4' and 5' carbon positions is shown on the anthocyanin structure.

Full size image

To date there has only been one reported molecular breeding experiment involving flavonoid pigments for cyclamen. It was focused on the generation of yellow flower colours through the production of yellow flavonoid pigments [7]. Our interest is in altering the anthocyanin-based colours [8]. In flower colour modification studies in general, particular attention has been paid to the enzymes responsible for the hydroxylation of the B-ring of the flavonoid molecule, namely F3'H and F3'5'H (Figure. 1) because of their key influence on the colour of anthocyanin pigments [9]. Specific experiments to accumulate delphinidin-derived anthocyanins by over expression of a F3'5'H transgene have been reported for carnation [10] and rose [11], while inhibition of both the F3'H and the F3'5'H genes has been used to modify colour and promote cyanidin- and pelargonidin-based pigment accumulation in flowers in the genera Torenia [12], Nierembergia [13] and Osteospermum [14].

Our strategy for modification of flower colour in cyclamen focused on the F3'5'H. Substrate feeding experiments with DHK and the F3'H/F3'5'H inhibitor tetcyclacis indicate that the cyclamen DFR can use DHK and that cyclamen has the ability to make pelargonidin-derived anthocyanins (K. Schwinn, unpublished data). The cloning of a F3'5'H cDNA and our cyclamen genetic transformation system [15] have allowed us to investigate flower colour formation in cyclamen. In this study we report on the effects of antisense suppression of F3'5'H on flavonoid end-product accumulation and flower colour.

Results

Isolation and sequence analysis of a cyclamen flavonoid 3', 5'-hydroxylase cDNA

A putative full-length cDNA for F3'5'H (CpF3'5'H) was isolated from a cDNA library made from mixed flower bud stages of C. persicum 'Sierra Rose'. The complete nucleotide sequence has 1719 nucleotides with a single major ORF encoding 508 amino acid residues (GenBank accession GQ891056).

When the deduced amino acid sequence for CpF3'5'H was used in a BLAST search of GenBank http://www.ncbi.nih.gov/blast/, the closest sequence was the putative F3'5'H from Camellia sinensis (GenBank accession AAY23287), with 83% amino acid identity. The Lasergene program MegAlign (DNASTAR Inc., Madison, USA) was used to compare the CpF3'5'H deduced amino acid sequence with ten F3'5'H sequences (the CYP75A group), ten F3'H sequences (CYP75B) and two 'outlier' cytochrome P450 sequences (data not shown). Amino acid identity of CpF3'5'H to other F3'5'H sequences was in the range from 75-82%, except for the Campanula medium F3'5'H sequence (BAA03440), 68% identity, which is suggested to have a distinct F3'5'H structure [16] and the monocot Phalaenopsis hybrida F3'5'H sequence (AAZ79451, 50% identity) [17]. A phylogenetic tree was formed using the CLUSTAL W algorithm http://www-bimas.cit.nih.gov/clustalw/clustalw.html with the MegAlign data (Figure. 2). The F3'5'H sequences form a distinct cluster, which includes the cyclamen sequence. Based on the amino acid and phylogeny analysis the evidence supports CpF3'5'H as encoding a F3'5'H enzyme.

Figure 2
figure 2

A phylogenetic tree inferred using CLUSTAL W from the deduced amino acid sequences for F3'5'H , F3'H and C4H. F3'5'H (mauve shading), F3'H (red shading), C4H (cinnamate 4-hydroxylase, no shading), a less closely related sequence for a cytochrome P450 enzyme involved in flavonoid biosynthesis. The phylogenetic tree shows bootstrap values. The F3'5'H sequences form a distinct cluster, which includes the cyclamen sequence. Based on the amino acid and phylogeny analysis the evidence supports CpF3'5'H as encoding F3'5'H.

Full size image

Generation of transformed lines and transgene expression analyses

Antisense CpF3'5'H transformants were produced from the 'Purple' cultivar using constructs pPN48/51, and from the 'Wine-Red' cultivar using pLN96/pPN50 (Figure. 3A). Flowers from several of the transgenic lines showed significant changes in colour, both in hue and intensity (chroma) (Figure. 4). No other phenotypic alterations were observed when compared with wildtype plants.

Figure 3
figure 3

Northern analysis of Minicyclamen transgenic lines. A) Schematic diagrams of the T-DNA regions of binary vectors, pLN96, pPN48, pPN50 or pPN51. These binary vectors harboured in their T-DNAs the cyclamen antisense F3'5'H gene under a CaMV35S promoter and either nptII or hpt selectable marker genes under a NOS promoter. B) Northern RNA blot analysis of hpt selectable marker expression in control and transgenic lines of cv 'Purple' (left) and RT-PCR analysis of nptII selectable marker expression of cv 'Wine-Red' (right). The expected size of the hpt signal was 1.4kb and the expected size of the nptII signal was 600bp. C) Northern RNA blot analysis of sense and antisense CpF3'5'H transcript in transgenic and control lines of cv 'Purple' (left) and cv 'Wine-Red' (right).

Full size image
Figure 4
figure 4

Flower colour phenotypes of selected transgenic lines. Cv 'Purple' (A-D); A-#31704 regeneration control line, B-#31674, C-#31682, D-#31683, antisense CpF3'5'H transgenic lines. Cv 'Wine-Red' (E-H): E-#29009 regeneration control line; F-#31691, G-#31695, H-#31698, antisense CpF3'5'H transgenic lines.

Full size image

Northern blot analysis of cultivar (cv) 'Purple' transformants showed that eight lines were transgenic for the hygromycin selectable marker (Figure. 3B). RT-PCR analysis of the nptII selectable marker showed the three cv 'Wine-Red' lines were also transgenic as expected (Figure. 3B).

Northern blot analysis with a mixed sense and antisense CpF3'5'H probe, (1.7 kb XbaI-EcoRI fragment, Figure. 3A), showed that two F3'5'H specific transcripts were detected (Figure. 3C). There was a marked reduction in endogenous CpF3'5'H transcript in all antisense lines of both cultivars. Antisense CpF3'5'H transcript was detected only in the transgenic lines and the levels varied between lines.

Flavonoid analyses

Anthocyanin content in the petals of the transgenic lines changed in both concentration and profile. The anthocyanins detected in the flower tissue of the regeneration control plants and transgenic lines are shown in Figure. 5 and 6 and are listed in Table 1. Anthocyanin identities were assigned by retention times and mass spectrometer data and were consistent with the anthocyanins identified previously for cyclamen, predominantly the 3-mono and 3,5 di-glucosides of peonidin, cyanidin and malvidin [2, 4]. Malvidin 3-O-glucoside was the predominant anthocyanin in cv 'Wine-Red' while malvidin 3,5 di-O-glucoside was the predominant anthocyanin in cv 'Purple'.

Table 1 HPLC-MS2 based identifications of the main anthocyanins detected in petal tissue.
Full size table
Figure 5
figure 5

HPLC chromatograms for petal extracts from selected transgenic lines. Untransformed control lines were #29009 and #31704. Absorbance was monitored at 530 nm. The major anthocyanins were identified as: (1) Petunidin 3-O-glucoside; (2) Malvidin 3-O-glucoside; (3) Malvidin rhamnosyl-glucoside; (4) Peonidin 3- O -glucoside; (5) peonidin rhamnosyl-glucoside; (6) Malvidin 3-5-di-O-diglucoside; (7) Cyanidin 3-5-di-O-diglucoside; (8) Peonidin 3-5-di-O-diglucoside.

Full size image

A change in anthocyanin profile was found in the petal tissue of the transgenic lines as might be expected with a reduction in F3'5'H activity (Figure. 6A). Delphinidin-derived (malvidin- or petunidin-based anthocyanins) pigment levels decreased as a proportion of the total anthocyanins in petal tissue of most of the transgenic lines while the proportion of cyanidin-derived pigments (peonidin- and cyanidin-based anthocyanins) increased. This shift in anthocyanin profile correlates with a loss of expression of the endogenous CpF3'5'H transcript (Figure. 3C). The greater the loss of expression, e.g. cv 'Purple' line #31685 and cv 'Wine-Red' line #31691, the greater the change in anthocyanin profile (Figure 6B). Pelargonidin, an anthocyanin pigment with a mono-hydroxylated B-ring, was not produced in the transgenic lines of either cultivar.

Figure 6
figure 6

Anthocyanin profiles of minicyclamen transgenic flowers. A) Relative proportions of delphinidin-derived and cyanidin-derived anthocyanins in the flower petals. Control lines are denoted C and transgenic lines by T. B) Total anthocyanin concentrations in the flower petals of transgenic lines. Colour bars are representative of actual petal colour of each line. Control lines are denoted C and transgenic lines by T. Mean ± SEM, n = 2. Values significantly different from the control at the 5% level have been indicated by a superscript a.

Full size image

There was also a marked reduction in total anthocyanin concentration in petal tissue of the transgenic lines. Lines with modified flower colour showed a decrease in total anthocyanin concentration of up to 80% of that in untransformed controls (Figure. 6B). The difference in anthocyanin concentrations between the transgenic lines and their respective controls were statistically significant at the 5% level.

Flavonol profiles were also examined. Flavonols in the untransformed and transgenic lines were putatively identified as kaempferol and quercetin 3-glucosides, rutinosides and acylated rutinosides (data not shown). This is consistent with previous studies [2]. Total flavonol concentration in the transgenic lines showed a statistically significant increase in most lines (Table 2). The quercetin/kaempferol ratio also increased significantly in most transgenics lines of cv 'Purple' but decreased significantly in all the transgenic lines of cv 'Wine-Red' (Table 2).

Table 2 Flavonol concentration and ratios in petals of transgenic lines (mg.g.DW-1) (Mean ± SEM, n = 2).
Full size table

Flower colour analysis

Expression of the introduced antisense CpF3'5'H transgene and resulting flavonoid concentration and profile changes in the transgenic lines were translated into visible flower colour changes (Figure. 4). Cultivar 'Purple' lines showed a loss of purple colour and became pink, while the cv 'Wine-Red' lines remained a similar pinkish hue but with reduced intensity (chroma).

The change in colour observed by eye was quantified by colour measurements using a colorimeter. The colour parameters, lightness [L], chroma [C] and hue angle [H°] were statistically significantly different from controls in lines with modified flower colour in most lines (Table 3). The exceptions were lines 31675 (cv 'Purple') and 31698 (cv 'Wine-Red') for L and C values. Both lines exhibited the least change in their anthocyanin profiles (Figures 6A, B). The majority of transgenic lines of both cultivars showed an increase in lightness (L) and a reduced intensity of colour (C). This is consistent with the decreased anthocyanin concentration in the petal tissue from the transgenic lines. There was also a clear shift in H° away from purple in the control line towards red in the transgenic lines of the 'Purple' cultivar. This change in hue angle correlates with a decrease in the proportion of delphinidin-derived anthocyanins. However line #31685, which had the largest proportion of cyanidin-derived pigments, did not have the largest shift in H°. Similarly the only line of the cv 'Wine-Red' transgenics showing a shift from delphinidin- to cyanidin-derived pigments (line #31691) did not show a significant change in H° while the other two transgenics did. The shift in hue angle for the cv 'Wine-Red' transgenics was in fact back to the purple region of the colour wheel. The overall shift, however, was very small and hue angle remained in the red/pink region.

Table 3 Flower colour characteristics for petal tissue of the control and transgenic lines.
Full size table

Discussion

Antisense suppression of CpF3'5'H was successful in changing anthocyanin profiles and flower colour in cyclamen. A shift from predominantly delphinidin-derived pigments to a greater relative proportion of cyanidin-derived pigments was achieved and in general this showed up as a concomitant shift in H°, the parameter indicating colour group. It is interesting that the degree of change in H° did not correlate with the degree of shift in pigments. The fact that the transformants also showed variable drops in total anthocyanin levels and changes in flavonol level and type illustrates both the links between the different pools of flavonoid substrates and the importance of the roles that anthocyanin concentration and flavonol copigmentation play in flower colour.

Similar changes in anthocyanin concentration and the accumulation of cyanidin-derived anthocyanins were seen for the two different minicyclamen cultivars and yet the greatest change in H° was seen in the lines of the purple cultivar. This is most likely due to a reduction in the predominant anthocyanin, malvidin 3-5-di-O-glucoside in these lines. This anthocyanin has been reported as being bluer in colour than malvidin mono-glucosides [3]. The predominant anthocyanin in the 'Wine-Red' cultivar is malvidin 3-O-glucoside and this has been reported to give pink/purple colours, closer to the colour associated with cyanidin and peonidin pigments [3].

Pelargonidin-based pigments were not detected in the flowers of the transgenics. One explanation for their absence is that suppression of F3'5'H activity was not complete, as evidenced by the presence of delphinidin-derived anthocyanins. This may be either due to inefficiency of the antisense approach (as opposed to hairpin RNA-induced RNAi [18]), effects due to transgene insertion or copy number [19–21], or the presence of other unaffected F3'5'H family members. The presence of a F3'H enzyme in the petals could have also removed substrate for pelargonidin production. We have searched for a cyclamen F3'H cDNA and found one (GenBank GU808358) with high deduced amino acid similarity to known F3'H sequences of other species (81% identity with the F3'H of gentian). However, transcript levels for this particular F3'H gene were not detectable by northern analysis during cyclamen petal development (unpublished data).

Substrate specificity is an important consideration regarding pelargonidin production. In some species, such as petunia [22], cymbidium [23, 24] and Osteospermum [14], synthesis of pelargonidin-based anthocyanins is limited by the substrate specificity of the endogenous DFR. Our substrate feeding experiments (mentioned previously) showed that cyclamen has the ability to make pelargonidin-derived anthocyanins. It is still possible, however, that cyclamen DFR has low substrate specificity for DHK and the action of flavonol synthase (FLS), F3'H and F3'5'H means that the DHK substrate is not used for the synthesis of pelargonidin. Retransformation of an antisense F3'5'H line from this study, with a transgene encoding a DFR known to efficiently catalyse the reduction of DHK to leucopelargonidin [25–27] could result in transgenic plants accumulating pelargonidin derivatives in flowers, as successfully demonstrated for Osteospermum [14]. It remains to be resolved whether there is a F3'H functioning in the flower. The presence of cyanidin-based pigments in the flowers of the antisense CpF3'5'H lines suggests F3'H activity. Thus, inhibition of either F3'H or FLS gene activity to reduce enzymatic competition for DHK substrate may also be necessary to promote pelargonidin production in DFR/antisense F3'5'H transgenics.

In the cyclamen transgenic lines, total anthocyanin levels decreased markedly while flavonol levels increased and the quercetin/kaempferol ratio changed. Similar results were reported for Nierembergia flowers modified with an antisense F3'5'H construct and were suggested to be due to a modified flow through the flavonoid pathway [13]. A block in F3'5'H activity resulted in an increase in pelargonidin precursors. Low F3'H activity coupled with a DFR that putatively does not recognise DHK, was suggested to have led to limited substrate flow toward pigment production and an increase in the sustrate pool for FLS [13]. The flavonoid enzyme kinetics are not known for cyclamen. However, if the cyclamen DFR has a low specificity for pelargonidin or cyanidin precursors (as the reduction in total anthocyanins (Figure. 6B) suggests) this would provide extra substrate for the FLS enzyme and explain the increased flavonol levels. Competition for substrate between FLS and DFR has also been shown to occur in petunia [28, 29].

It is interesting that while flavonol levels generally increased in the transgenics, there were differences in the quercetin/kaempferol ratios between the lines of the different cultivars. Quercetin flavonols increased in cv 'Purple' lines while kaempferol types increased in cv 'Wine-Red' lines. This inverse result and the consistency of the ratio change within lines of each cultivar argues against the suppression of F3'5'H activity directly altering the balance of DHK and DHQ, and thus what is available for the FLS. Furthermore, differing substrate specificities of their respective FLS cannot account for the observed results. Differing specificities of other enzymes are likely to be the cause. The probable candidate is F3'H, which in other species can not only alter the balance between DHK and DHQ, but also convert kaempferol to quercetin [30]. Further studies of cyclamen flower colour would warrant a continued search for a F3'H.

Conclusions

We report here the first successful alteration of cyclamen anthocyanin pigmentation using genetic modification techniques. Our results highlight the intricate interplay between type and concentration of both anthocyanin pigments and flavonol co-pigments in flower colour and illustrate the complexity involved in modifying a biosynthetic pathway with multiple branch points to different end products.

Methods

Cloning of F3'5'H cDNA and sequence analysis

A cDNA library from mixed flower stages of C. persicum 'Sierra Rose' petals was made using a lambda ZAPII bacteriophage vector kit (Stratagene, USA). This library was first screened with a heterologous clone of F3'H from petunia (Florigene Flowers, Australia) and a partial F3'5'H cDNA was found. The partial F3'5'H cDNA was used to rescreen the cDNA library to obtain a full length CpF3'5'H cDNA.

The MegAlign programme of Lasergene (DNASTAR Inc., Madison, USA) was used to compare the CpF3'5'H deduced amino acid sequence to ten known F3'5'H sequences (Camellia sinensis AAY23287; Campanula medium BAA03440; Catharanthus roseus CAA09850; Eustoma grandiflorum BAA03439; Glycine max ABQ96218; Gossypium hirsutum AAP31058; Petunia hybrida CAA80266; Phalaenopsis hybrida AAZ79451; Solanum tuberosum AAV85473; Vitis vinifera BAE47007), ten F3'H sequences (Antirrhinum majus ABB53383; Arabidopsis thaliana NP_196416; Glycine max ABW69386; Ipomoea tricolor BAD00192; Matthiola incana AAG49301; Perilla frutescens BAB59005; Petunia hybrida AAD56282; Populus trichocarpa XP_002319761; Sorghum bicolor ABG54321; Vitis vinifera ABH06586), cinnamate 4-hydroxylase from Arabidopsis thaliana (AAC99993) and flavone synthase II from Medicago truncatula (ABC86159).

Construction of binary vectors

The CpF3'5'H cDNA was cloned into the EcoRI multiple cloning site of pART7 [31] in the antisense orientation to form pLN95. The NotI fragment from pLN95, which contains the 35S:antisenseF3'5'H:Ocs expression cassette, was ligated into the binary vectors; pART27 [31] to make pLN96, pMOA33 [32] to make pPN50, pMOA 34 [32] to make pPN51, and BJ49 [31] to make pPN48 (Figure. 3A). These binary vectors carried either the nptII or hpt selectable marker genes under a NOS promoter (Figure. 3A).

Transformation with Agrobacterium tumefaciens

Etiolated hypocotyls of two parental lines of F1 hybrid minicyclamen cv 'Purple' and cv 'Wine-Red' were used as explants for transformation experiments. A. tumefaciens strain EHA105 containing either pLN96, pPN48, pPN50 or pPN51 were used to inoculate explants. The transformation protocol used was that reported by Boase et al. [15] except that hygromycin was used as the selection agent for cv 'Purple' lines using a range of concentrations: 5mg/l to day 12 after Agrobacterium inoculation, 20mg/l to day 77 after inoculation, then 15mg/l until shoots were recovered.

Northern blot analyses

RNA was extracted from petal tissue for northern blot analysis using a modified hot borate method [33, 34]. RNA was separated by electrophoresis on a 1% agarose RNA gel and subsequently transferred to Hybond XL nylon membranes using a SSC overnight blotting method. The membranes were hybridized with appropriate radioactively-labelled probes. The probe for hpt was a 1.1 kb XhoI fragment digested from pCAMBIA1301, which contained the hpt gene. The probe for F3'5'H was a 1.7 kb XbaI-EcoRI fragment digested from pLN95. Both membranes were also rehybrised to a cDNA probe corresponding to a 25/26S rRNA (pTip6) from Asparagus officinalis, to show RNA loadings. Autoradiography was conducted at -80°C using Kodak Biomax X-ray film.

RT-PCR analysis of nptIImRNA transcripts

To investigate the expression of the introduced nptII selectable marker recombinant gene, RT-PCR analysis was performed on RNA extracted from petals using a modified hot borate method [33, 34]. Three independent transgenic lines of cv 'Wine-Red' (#31691, #31695 and #31698) and one untransformed control (#29009) were tested. First strand cDNA was reverse transcribed from 100ngRNA per sample using Superscript II (Invitrogen USA) and oligo dT primer, and then 1 μl of the resulting cDNA per line was used for the PCR. For PCR, initial denaturation was at 94°C for 2 min followed by 40 cycles of melting (94°C/30 s), annealing (50°C/30 s) and extension (72°C/2 min). The nptII primers used were: forward 5'-ATGACTGGGCACAACAGACCATCGGCTGCT-3' and reverse, 5'-CGGGTAGCCAACGCTATGTCCTGATAGCGG-3'.

PCR products were separated electrophoretically on a 1% (w/v) NaB agarose gel stained with Sybr®safe (Invitrogen USA).

Flavonoid analyses

Flavonoids were analysed by high performance liquid chromatography (HPLC) and liquid chromatography mass spectrometry (LC-MS). Freeze dried tissue was used for the analysis. Samples of ground freeze-dried petal tissue (50mg DW) were extracted initially in 2ml of methanol:acetic acid:water (70:3:27) and then re-extracted in 2 ml methanol:acetic acid:water (90:1:9). The combined supernatants were concentrated in vacuo and made up to a final volume of 1ml. HPLC analysis was carried out using a Waters 600 solvent delivery system with a Phenomenex Prodigy (5 μm, 250 × 4.6 mm) RP-18 endcapped column (column temperature 30°C) and a Waters 996 PDA detector. Solvent systems, flow rates and gradients are as described by Bloor et al. [35]. Flavonoids were detected at 350nm and anthocyanins at 530nm. Flavonoid levels were determined as quercetin-3-O-rhamnoglucoside (Apin Chemicals, Abingdon, Oxon, UK) equivalents, and the anthocyanins as cyanidin 3-O-glucoside (Extrasynthese, Genay, France) equivalents. Results are reported as the mean of the two replicates.

Separate extracts were analysed by electrospray mass spectrometry with a Thermo Finnigan LTQ ion-trap mass spectrometer. A Synergi Fusion RP80, 4 μm, 150 × 2.1 mm column with 4 × 2 mm guard cartridge from Phenomenex Ltd was used for separation. The mobile phase consisted of water (A) and acetonitrile (B) both containing 1% formic acid (FA). Extracts were injected at 5 μL volumes with a gradient program from 95% A to 50% A over 50 min. The column was washed by ramping to 90% B for 5 min and then re-equilibrated to the starting conditions for a further 5 min. Compound elution was monitored by PDA detector scanning the range 250-600 nm and by mass scanning from m/z 150-1500 to collect parent, MS2 and MS3 data in positive and negative ion (additional run) selection modes.

Flower colorimeter analysis

Colours in all lines were quantified by measurement of three petals of each flower, three flowers per line with a Minolta CR-200 tristimulus colorimeter, set on CIELab D65 light source and 0° observer angle. Lightness (L) represents the proportion of total incident light that is reflected. Chroma (C) describes the degree to which selective absorption occurs i.e. colour saturation in relative intensity units. Hue angle (H) is derived from a CIELAB colour space wheel with values stepped counterclockwise from red at 0°/360°, yellow at 90°, bluish-green at 180° and blue at 270° [36].

Statistics

A one-way ANOVA was performed on each data set shown in Tables 2 & 3 and Figure 6B followed by a comparison of means using either a 5% Fisher's Least Significant Difference (5% LSD) to compare each line with a single control, or contrasts to compare each line with the combined mean of two controls. Lines with values significantly different from their control (or pair of controls) at the 5% level have been indicated by adding a superscript a to the means in Tables 2 and 3, and in Figure 6B. All analyses were performed using GenStat statistical software [37].

Authors' information

Ms Marshall and Ms Patel are former team members of the New Zealand Institute for Plant and Food Research Ltd.