PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of annbotAboutAuthor GuidelinesEditorial BoardAnnals of Botany
 
Ann Bot. 2010 June; 105(7): 1199–1210.
Published online 2009 October 8. doi:  10.1093/aob/mcp253
PMCID: PMC2887060

Genetic analysis of potassium use efficiency in Brassica oleracea

Abstract

Background and Aims

Potassium (K) fertilizers are used in intensive and extensive agricultural systems to maximize production. However, there are both financial and environmental costs to K-fertilization. It is therefore important to optimize the efficiency with which K-fertilizers are used. Cultivating crops that acquire and/or utilize K more effectively can reduce the use of K-fertilizers. The aim of the present study was to determine the genetic factors affecting K utilization efficiency (KUtE), defined as the reciprocal of shoot K concentration (1/[K]shoot), and K acquisition efficiency (KUpE), defined as shoot K content, in Brassica oleracea.

Methods

Genetic variation in [K]shoot was estimated using a structured diversity foundation set (DFS) of 376 accessions and in 74 commercial genotypes grown in glasshouse and field experiments that included phosphorus (P) supply as a treatment factor. Chromosomal quantitative trait loci (QTL) associated with [K]shoot and KUpE were identified using a genetic mapping population grown in the glasshouse and field. Putative QTL were tested using recurrent backcross substitution lines in the glasshouse.

Key Results

More than two-fold variation in [K]shoot was observed among DFS accessions grown in the glasshouse, a significant proportion of which could be attributed to genetic factors. Several QTL associated with [K]shoot were identified, which, despite a significant correlation in [K]shoot among genotypes grown in the glasshouse and field, differed between these two environments. A QTL associated with [K]shoot in glasshouse-grown plants (chromosome C7 at 62·2 cM) was confirmed using substitution lines. This QTL corresponds to a segment of arabidopsis chromosome 4 containing genes encoding the K+ transporters AtKUP9, AtAKT2, AtKAT2 and AtTPK3.

Conclusions

There is sufficient genetic variation in B. oleracea to breed for both KUtE and KUpE. However, as QTL associated with these traits differ between glasshouse and field environments, marker-assisted breeding programmes must consider carefully the conditions under which the crop will be grown.

Key words: Arabidopsis, Brassica oleracea, genetics, potassium (K), potassium use efficiency (KUE), quantitative trait loci (QTL), shoot

INTRODUCTION

Potassium (K) is an essential mineral element for plant growth, development and fecundity (White and Karley, 2009). It is required in large amounts by crop plants and, because many agricultural soils lack sufficient phytoavailable K for maximal crop production, it is generally supplied as K-fertilizers in both intensive and extensive agricultural systems (Lægreid et al., 1999; Pettigrew, 2008; Rengel and Damon, 2008; Fageria, 2009). However, there are both financial and environmental costs, inherent in the energy required for their production, distribution and application, to the consumption of K-fertilizers (Lægreid et al., 1999). In the immediate future, a scarcity of K-fertilizers is unlikely, but unstable energy prices, which affect the mining, distribution and application of K-fertilizers, and the introduction of financial instruments associated with meeting climate change and other environmental targets, will determine their cost, availability and usage. For these reasons, K-fertilizers must be deployed efficiently.

Breeding crops that acquire and/or utilize K more effectively is one strategy that could reduce the use of K-fertilizers (Baligar et al., 2001; Trehan, 2005; Pettigrew, 2008; Rengel and Damon, 2008; Fageria, 2009; Szczerba et al., 2009). Agronomic K use efficiency is defined as crop dry matter yield per unit K supplied (g DM g−1 Ks). This is numerically equal to the product of plant K content per unit K supplied (g K g−1 Ks), which is referred to as plant K uptake efficiency (KUpE), and crop yield per unit plant K content (g DM g−1 K), which is referred to as plant K utilization efficiency (KUtE). In general, plant K content can be estimated from shoot K content, and KUtE can then be expressed as the reciprocal of shoot K concentration ([K]shoot) if the entire shoot is harvested (White et al., 2005).

Potassium is the most abundant inorganic cation in plants, comprising up to 10 % of a plant's dry weight (Watanabe et al., 2007), and [K]shoot is often higher in plants from the Brassicaceae than in those from many other angiosperm families when grown under comparable conditions (Broadley et al., 2004). Variation in [K]shoot has been observed among genotypes of several plant species grown in the same environment (Baligar et al., 2001; Pelletier et al., 2008; Pettigrew, 2008; Rengel and Damon, 2008; Fageria, 2009), including Brassica oleracea kales and collards (Kopsell et al., 2004; Vilar et al., 2008), Brassica rapa (Wu et al., 2008), Brassica napus (Brennan and Bolland, 2007; Damon et al., 2007; Rose et al., 2007; Bhardwaj and Mamama, 2009) and Brassica juncea (Shi et al., 2004).

This paper describes the genetic variation in, and chromosomal quantitative trait loci (QTL) associated with, [K]shoot and therefore KUtE in Brassica oleracea, a major edible crop worldwide that includes several morphologically distinct subtaxa, such as acephala (kale/collards), alboglabra (oriental kale), botrytis (cauliflower), capitata (cabbage), gemmifera (Brussels sprout), gongylodes (kohlrabi), italica (broccoli/calabrese), sabauda (Savoy cabbage) and sabellica (borecole/curly kale), in which the entire shoot is generally harvested (Broadley et al., 2008). Brassica oleracea is a diploid plant species for which significant genetic resources are available, including (1) a diversity foundation set (DFS) thought to contain most of the common allelic variation within this species (Broadley et al., 2008), (2) genetic mapping populations suitable for the identification of chromosomal loci affecting quantitative traits (Sebastian et al., 2000; Pink et al., 2008), (3) collections suitable for isolating mutants in specific genes (Himelblau et al., 2009), (4) extensive genome sequence (Schranz et al., 2007) and (5) routine techniques for genetic transformation (Cogan et al., 2004). In addition, the genus Brassica contains the closest crop relatives of the model plant Arabidopsis thaliana and genomic relationships between A. thaliana, B. oleracea (Brassica C genome) and other Brassica crops, such as B. rapa (A genome), B. napus (AC genome) and the B genome-containing mustards (B. nigra, B. juncea, B. carinata), are becoming increasingly well characterized (Parkin et al., 2005; Schranz et al., 2006, 2007). Ultimately, this knowledge of comparative genomics will facilitate the identification of genes affecting KUpE and KUtE among the Brassicaceae and the breeding of Brassica crops that utilize K-fertilizers more efficiently.

MATERIALS AND METHODS

Plant material

The plant material used in this study consisted of: (1) a DFS of 376 Brassica oleracea L. accessions, selected from the >4300 accessions held in the Warwick-HRI Genetic Resources Unit and thought to contain most of the common allelic variation within this species (Broadley et al., 2008); (2) a set of 74 commercial genotypes sampled to represent the distinct major B. oleracea morphotypes in current or recent cultivation in northern Europe (Broadley et al., 2008); (3) the 90 most informative lines from the ‘AGDH’ genetic mapping population, to enable the identification of QTL associated with shoot K concentration in B. oleracea. The AGDH population was generated through anther culture of the F1 of a cross between a DH rapid-cycling accession of B. oleracea var. alboglabra (A12DHd) and a DH accession derived from an F1 hybrid calabrese cultivar, ‘Green Duke’, B. oleracea var. italica (GDDH33). A linkage map of 906 cM for the AGDH mapping population has been developed, with a mean distance between marker loci of 1·92 ± 3·49 cM, such that approx. 90 % of the genome is within 5 cM of a marker (Sebastian et al., 2000; Broadley et al., 2008; Pink et al., 2008). There were physical limitations on the number of AGDH lines that could be grown in these experiments. Hence, a subset of AGDH lines was selected that maximized the range of recombination break points, as well as the marker scoring density. This subset has been shown to be adequate for the successful detection of QTL for shoot mineral concentrations (Broadley et al., 2008; Hammond et al., 2009). (4) A set of 20 recurrent backcross substitution lines (the ‘AGSL’ population), each containing chromosomal segments of GDDH33 introgressed into the A12DHd background, was used to validate the location of QTL in the AGDH population (Rae et al., 1999; Broadley et al., 2008). (5) A common reference set of genotypes consisting of A12DHd, GD33DH and eight commercial B. oleracea cultivars allowed comparisons between experiments (Greenwood et al., 2005, 2006; Broadley et al., 2008; Hammond et al., 2009).

Field and glasshouse experiments

Plants were grown in a series of field and glasshouse experiments at Wellesbourne, UK (52°12′30″N, 1°36′39″W, 45 m above sea level), each temporally arranged over several occasions, as described by Broadley et al. (2008). Each occasion represented an independent experimental run, containing a subset of the accessions being screened. Sets of plants were grown with different applications of P-fertilizers to investigate the effect of plant growth rate on [K]shoot. Although most plants were grown successfully, not all of the accessions sown in each experiment survived to harvest, especially in the field. The experiments were as follows. (1) A glasshouse experiment (GE1), in which three replicates of the 376 DFS accessions and nine replicates of the 74 commercial cultivars were sown over six occasions between June, 2003 and July, 2004 in a 40-m2 ‘Cambridge’-type glasshouse compartment that was set to maintain temperatures of 24 °C by day and 15 °C at night using automatic vents and supplementary heating. Daylight was supplemented by artificial lighting (Son-T 400-W Philips phi 0·85i, Groote, Noort, The Netherlands) to maintain 16 h light per day above a photosynthetically active radiation (PAR) of 300 W m−2. Plants were grown in compressed polystyrene pots (dimensions 11 × 11 × 12 cm; Desch Plantpak Ltd, Mundon, Maldon, UK), filled to a depth of approx. 0·5 cm below the rim with a peat-based compost. The compost contained either 5·25 mg L−1 (low [P]ext) or 15·75 mg L−1 (high [P]ext) of added P following the incorporation of 0·075 and 0·225 g of sieved (500 µm) single superphosphate [SSP, CaSO4 + Ca(H2PO4)2, containing 7 % P] per litre of compost (Greenwood et al., 2005). Other nutrients were incorporated in the potting-mix in sufficient amounts to prevent mineral deficiencies. Analysis of compost samples gave Olsen's extractable P values of 9·2 and 20·2 mg L−1 for low and high [P]ext composts, respectively. The average water-extractable K concentration in these composts was 183 ± 3 mg L−1 (mean ± s.e.m., n = 34), resembling soils of high K-fertility. Plant shoots were sampled at similar developmental stages, 39, 47, 49, 49, 42 and 37 d after sowing on the six occasions. (2) A field experiment (FE1) conducted in Wharf Ground, Wellesbourne, between May, 2004 and May, 2005 in which three replicates of the 74 commercial cultivars were sown over three occasions at four [P]ext using an alpha design (Patterson and Williams, 1976). The Wharf Ground soil is a sandy loam Inceptisol in the Wick series of English classification (Whitfield, 1974). Supplementary irrigation was supplied via oscillating lines when required, and pesticide applications were made according to horticultural best-practice. The [P]ext treatments were imposed by incorporating triple superphosphate [TSP, Ca(H2PO4)2, containing 21 % P] equivalent to 0, 298, 1125 or 2713 kg TSP ha−1 to a depth of 0·10 m using a power harrow (Greenwood et al., 2005). Analysis of soil samples (to a depth of 30 cm) from these plots gave average Olsen's extractable P values of 40·7, 39·6, 81·7 and 152·1 mg P L−1 for the four [P]ext treatments. Unfertilized soils had ammonium nitrate-extractable K concentrations of 59 ± 1·8 mg L−1 (mean ± s.e.m., n = 40) and there was an annual overall dressing of 289 kg N ha−1 and 250 kg K2O ha−1 to the Wharf Ground field. Plant shoots were sampled after 101, 97 and 93 d growth on the three occasions. These timings were chosen to represent pre-commercial maturity. (3) A second glasshouse experiment (GE2), in which nine replicates of the 90 AGDH lines plus the A12DHd and GDDH33 parents of the AGDH population and eight reference commercial cultivars were sown over three occasions between February and July, 2005 in the same glasshouse compartment and at the same two [P]ext as GE1 using an alpha design. Plant shoots were sampled at a comparable growth stage, after 50, 50 and 34 d growth on the three occasions. (4) A second field experiment (FE2), conducted on Wharf Ground between March and May, 2006, in which three replicates of 72 genotypes (62 AGDH lines, A12DHd, GDDH33 and eight reference commercial cultivars) were sown at the same four [P]ext levels, and with the same amounts of N and K fertilizer, as FE1 using an alpha design. Plant shoots were sampled after 105 d growth. (5) A third glasshouse experiment (GE3), undertaken between March and May, 2006, in which three replicates of 30 genotypes (20 AGSLs, A12DHd, GDDH33 and eight reference commercial cultivars) were sown in the same glasshouse compartment and at the same two [P]ext as GE1 and GE2. Plant shoots were sampled 39 d after sowing.

Potassium analysis

In all experiments, shoot fresh weight (FW), comprising all above-ground biomass, was recorded immediately upon harvesting and shoot dry matter (DM) was determined after oven-drying at 60 °C for 72 h. For GE1, [K]shoot was determined by a commercial foliar analysis laboratory (Yara Phosyn Ltd, Pocklington, York, UK). For all other experiments, [K]shoot was determined at Warwick-HRI using inductively coupled plasma emission spectrometry (JY Ultima 2, Jobin Yvon Ltd, Stanmore, Middlesex, UK) following the digestion of dried plant material using the micro Kjeldahl method (Bradstreet, 1965).

Data analysis

Data were analysed using REML procedures in GenStat (Release 9·1·0·147, VSN International, Oxford, UK) to allocate sources of variation and estimate accession means for individual experiments (Patterson and Thompson, 1971; Robinson, 1987). QTL mapping was performed with QTL Cartographer 2·0 (Wang et al., 2004), using the composite interval mapping (CIM) option as described previously (Broadley et al., 2008; Hammond et al., 2009). Summary statistics of [K]shoot for the DFS, commercial cultivars and AGDH lines are expressed as mean ± s.e. for n genotypes.

RESULTS

Species-wide genetic variation in shoot K concentration in B. oleracea

Species-wide genetic variation in [K]shoot of B. oleracea was quantified in glasshouse (GE1) and field (FE1) experiments that included substrate P concentration ([P]ext) as a treatment factor (Fig. 1). There was substantial variation in [K]shoot among the 343 B. oleracea genotypes of the DFS grown in GE1, among the 75 commercial cultivars grown in GE1 and among the 72 commercial cultivars grown in FE1.

Fig. 1
Shoot K concentrations of Brassica oleracea genotypes represented in (A) the structured diversity foundation set (DFS) in glasshouse experiment one (GE1; n = 343), in B. oleracea subtaxa surveyed in GE1 (sabellica, n = 6; acephala, n = 40; italica, n ...

In GE1, 18·3 % of the total variation in [K]shoot was attributed to the accession, [P]ext and [P]ext × accession terms (Table 1). The genetic variance component was highly significant (P < 0·001), accounting for 16·8 % of the total variation in [K]shoot (Table 1). [K]shoot was greater at high [P]ext for most accessions (Fig. 2A), but [P]ext × accession interactions were not significant (P > 0·05). In general, therefore, [K]shoot of B. oleracea genotypes did not respond differently to altered [P]ext. The [K]shoot of B. oleracea genotypes grown in GE1 ranged from 2·72 to 6·56 %DM at low [P]ext and from 2·06 to 6·94 %DM at high [P]ext (Figs 1 and and2A).2A). [K]shoot differed between subtaxa, with botrytis (cauliflower) and gongylodes (kohlrabi) subtaxa having highest values and sabellica (borecole/curly kale), acephala (kale/collards) and italica (broccoli/calabrese) the lowest values (Fig. 1). Averaged across both [P]ext, [K]shoot among DFS accessions ranged from 2·92 to 6·64 %DM (n = 343) and the [K]shoot of commercial cultivars ranged from 3·27 to 5·94 %DM (n = 75). Thus, the extent of variation in [K]shoot observed among commercial cultivars was 72 % of the species-wide variation in [K]shoot. The effect of shoot DM accumulation on [K]shoot was tested within subtaxa, to avoid confounding effects of shoot morphology (data not shown). These relationships were rarely significant. At low [P]ext, [K]shoot was significantly (Fprob < 0·05) inversely correlated with shoot DM only within the alboglabra subtaxon. At high [P]ext, [K]shoot was significantly (Fprob < 0·05) inversely correlated with shoot DM within the acephala and capitata subtaxa.

Table 1.
Variance components analyses of shoot K concentrations (%DM) of Brassica genotypes grown in the three glasshouse experiments (GE1, GE2 and GE3) and the two field experiments (FE1, FE2)
Fig. 2
Shoot K concentrations of Brassica oleracea genotypes grown at either low or high external P concentrations ([P]ext) in (A) glasshouse experiment one (GE1) and (B) GE2, and at the lowest and highest P-fertilizer application rate ([P]ext) in (C) field ...

There was substantial variation in [K]shoot among the 72 commercial cultivars grown in FE1 (Figs 1 and and2C).2C). Variance components attributed to accession, [P]ext and [P]ext × accession terms accounted for 16·4 % of the total variation in [K]shoot (Table 1). Genetic variance components for [K]shoot were highly significant (P < 0·001) in FE1 and accounted for 10·5 % of the total variation. The [P]ext × accession interactions for [K]shoot were marginally significant (P < 0·05). There were significant positive correlations between [K]shoot at different [P]ext among the 72 commercial cultivars grown in FE1 (e.g. Fig. 2C; P < 0·001). However, in contrast to values in GE1, the [K]shoot of most genotypes grown in FE1 was greater at the lowest [P]ext than at the highest [P]ext (Fig. 2C). In general, the [K]shoot of plants grown in the field were lower than the [K]shoot of plants grown in the glasshouse, which might, in part, be attributed to differences in available K and/or the presence of competing cations between the two environments. Averaged across all [P]ext, [K]shoot of B. oleracea genotypes grown in FE1 ranged from 1·72 to 2·70 %DM, and [K]shoot were significantly (P < 0·001) positively correlated among the 70 cultivars grown successfully in experiments GE1 and FE1 (Fig. 3A), indicating that genotypic differences in [K]shoot were consistent between field and glasshouse environments. However, the [K]shoot of all commercial cultivars was higher in GE1 than in FE1 (Fig. 3A). Although there was a significant (P = 0·0011) negative relationship between [K]shoot and shoot DM among commercial cultivars when grown at high [P]ext in the glasshouse, no significant relationships were obtained between [K]shoot and shoot DM among commercial cultivars when grown at low [P]ext in the glasshouse or at any [P]ext in the field (data not shown).

Fig. 3
(A) Shoot K concentrations averaged across all P-fertilizer application rates ([P]ext) of 70 Brassica oleracea genotypes grown in both field experiment one (FE1) and glasshouse experiment one (GE1). The fitted line represents a significant linear regression ...

QTL affecting shoot K concentration in B. oleracea

Chromosomal QTL associated with [K]shoot were mapped using an informative subset of 90 DH lines from the AGDH population in both glasshouse (GE2) and field (FE2) experiments that included [P]ext as a treatment factor (Table 2). These loci were confirmed and resolved using substitution lines in a further glasshouse experiment (GE3).

Table 2.
Chromosomal quantitative trait loci (QTL) associated with shoot K concentration (%DM) of Brassica oleracea grown in glasshouse experiment two (GE2) and field experiment two (FE2)

A significant (P = 0·014) positive correlation was observed for [K]shoot among the nine reference B. oleracea accessions grown in both GE1 and GE2 (Fig. 3B), suggesting that these experiments were comparable. In GE2, the variance components attributed to accession, [P]ext and [P]ext × accession terms amounted to 25·8 % of the total variation in [K]shoot (Table 1). The genetic variance component was highly significant (P < 0·001) and accounted for 22·2 % of the total variance in [K]shoot. However, in contrast to DFS accessions and commercial cultivars grown in GE1, [P]ext × accession interactions were marginally significant (P < 0·05) for [K]shoot in GE2, despite the highly significant (P < 0·001) positive relationship between [K]shoot at low and high [P]ext among the genotypes of the AGDH population (Fig. 2B). [K]shoot varied significantly among the 92 DH lines and parents of the AGDH population studied in GE2 and, as was observed for the DFS and commercial cultivars studied in GE1, [K]shoot was greater at high [P]ext for most AGDH lines (Figs 1 and and2C).2C). The [K]shoot of the DH lines grown in GE2 ranged from 3·93 to 6·50 %DM at low [P]ext and from 3·90 to 6·23 %DM at high [P]ext (Figs 1 and and2C).2C). Averaged across both [P]ext, the range of [K]shoot among the 92 B. oleracea genotypes sampled in GE2 (4·12–6·36 %DM) suggested that this population approximated 60 % of the species-wide variation in [K]shoot observed in the glasshouse.

A significant (P = 0·039, n = 28) positive correlation was observed for [K]shoot among the seven reference B. oleracea accessions successfully grown in both FE1 and FE2 (Fig. 3C), suggesting that these experiments were comparable. The variance components attributed to accession, [P]ext and [P]ext × accession terms amounted to 27·6 % of the total variation in [K]shoot in FE2 (Table 1). The genetic variance component was highly significant (P < 0·001) and accounted for 23·7 % of the total variation in [K]shoot (Table 1). Averaged over all [P]ext, there was a significant (P < 0·001) positive correlation in the [K]shoot of the 62 genotypes from the AGDH population grown in FE2 and GE2 (Fig. 3D), which is consistent with the significant positive correlation in [K]shoot of the 70 commercial cultivars grown in FE1 and GE1 (Fig. 3A). In addition, [K]shoot of all the AGDH population was lower in FE2 than in GE2 (Fig. 3D), as was observed for the commercial cultivars grown in FE1 and GE1 (Fig. 3A). The [P]ext × accession interactions were not significant (P > 0·05) in FE2. Thus, [K]shoot did not respond differently to altered [P]ext among the AGDH lines grown in FE2. Shoot K concentration varied significantly among the 62 lines of the AGDH population studied in FE2 and, as was observed for the commercial cultivars studied in FE1, [K]shoot was greater at low [P]ext for most of the AGDH population (Figs 1 and and2).2). There were also significant positive correlations between [K]shoot at different [P]ext among the 62 lines of the AGDH population grown in FE2 (e.g. Fig. 2D; P < 0·001). The [K]shoot of the 62 AGDH lines grown in FE2 ranged from 2·04 to 4·11 %DM with the lowest rate of P-fertilizer application and from 2·05 to 3·57 %DM with the highest rate of P-fertilizer application (Figs 1 and and2D).2D). Highly significant (P < 0·001) negative relationships were observed between [K]shoot and shoot Ca and Mg concentrations in the AGDH population grown in the glasshouse (Fig. 4), and negative relationships were also observed between [K]shoot and shoot Ca (P = 0·0018) and Mg (P = 0·176) in the field (data not shown).

Fig. 4.
Relationships between shoot K concentration ([K]shoot) and (A) shoot Ca concentration ([Ca]shoot), (B) shoot Mg concentration ([Mg]shoot) and (C) shoot Na concentration ([Na]shoot) averaged across both P-fertilizer application rates ([P]ext) for the 90 ...

In both glasshouse and field experiments, A12DHd had a consistently higher [K]shoot than GDDH33 (e.g. Fig. 5). Significant (log-likelihood of there being one vs. no QTL, LOD > 3) or indicative (LOD > 2) QTL associated with [K]shoot were detected on six of the nine linkage groups of B. oleracea (Table 2). No QTL associated with [K]shoot were detected on chromosomes C5, C6 or C8. The observed QTL accounted for 83 % of the additive genetic variance component (VA) for [K]shoot for plants grown in the glasshouse (GE2) and 44 % of the VA in [K]shoot for plants grown in the field (FE2). However, no QTL associated with [K]shoot were identified in both environments, despite the significant positive correlation in [K]shoot among genotypes grown in the glasshouse and field (Fig. 3). In GE2, there was a positive additive effect of the A12DHd allele on [K]shoot at QTL on chromosomes C3, C4 and C7, and a negative effect of the A12DHd allele on [K]shoot at QTL on chromosomes C1 and C9 (Table 2). In FE2, there was a positive effect of the A12DHd allele on [K]shoot at QTL on chromosome C2 and a negative effect of the A12DHd allele on [K]shoot at the QTL on chromosome C9 (Table 2).

Fig. 5.
Shoot K concentrations ([K]shoot) in A12DHd (A alleles), GDDH33 (G alleles), and the AGSL substitution lines AGSL118, ASGL119, AGSL121, AGSL122, ASGL129, AGSL165 and AGSL168 in which chromosomal segments of GDDH33 are introgressed into the A12DHd background ...

The presence of QTL associated with [K]shoot in glasshouse-grown plants was tested using the AGSL substitution lines, in which segments of GDDH33 are introgressed into the A12DHd background (Rae et al., 1999; Broadley et al., 2008; Hammond et al., 2009). As was observed in previous experiments in the glasshouse, A12DHd had a higher [K]shoot than GDDH33 (Fig. 5). Of the AGSL lines screened, AGSL118 and ASGL119 were potentially informative for a putative QTL associated with [K]shoot on chromosome C1 (91·2 cM) and AGSL121, AGSL122 and ASGL129 were potentially informative for a putative QTL associated with [K]shoot on chromosome C9 (33·5 cM), although the QTL lie within chromosomal regions whose parental identity has not yet been attributed in the AGSL lines, and AGSL165 and AGSL168 were informative for a putative QTL associated with [K]shoot on chromosome C7 (62·2 cM). The phenotypes of AGSL118, ASGL119, AGSL121, AGSL122 and ASGL129 were similar to A12DHd and therefore did not confirm the putative QTL associated with [K]shoot on chromosome C1 (91·2 cM) or C9 (33·5 cM). However, the [K]shoot of AGSL165 and AGSL168 (line 27) were lower than A12DHd, apparently confirming the putative QTL associated with [K]shoot on chromosome C7 (62·2 cM).

Potassium use efficiency in B. oleracea

Agronomic K use efficiency is the product of KUpE and KUtE of a crop. The KUtE of a plant can be estimated as the reciprocal of [K]shoot. There is therefore significant genetic variation in KUtE, and QTL associated with KUtE correspond to QTL associated with [K]shoot (Table 2). The KUpE of a plant can be estimated as the shoot K content and calculated as the product of shoot DM and [K]shoot. When grown in the glasshouse or field, there is considerable variation in KUpE among genotypes of B. oleracea. For example, in GE1 KUpE varied between 1·7 and 62·5 mg K per plant when grown at low P and between 6·1 and 108·9 mg K per plant when grown at high P (Fig. 6A). Subtaxa differed in KUpE, with botrytis and italica subtaxa having the lowest KUpE, and capitata, sabauda and tronchuda subtaxa having the highest KUpE. KUpE was higher in commercial cultivars than in the DFS when grown in the glasshouse at low [P]ext (30·8 ± 0·73 mg K per plant, n = 79, vs. 27·8 ± 0·47 mg K per plant, n = 340) or high [P]ext (71·4 ± 1·76 mg K per plant, n = 79, vs. 59·9 ± 1·00 mg K per plant, n = 341). In FE1, KUpE varied between 138·3 and 368·0 mg K per plant among commercial cultivars when grown at the lowest P-fertilizer application and between 110·3 and 300·3 mg K per plant when grown at the highest P-fertilizer application (data not shown). The KUpE of commercial cultivars was significantly (P = 0·0012, n = 70) correlated between glasshouse and field experiments performed with an adequate P supply (data not shown). Several QTL associated with KUpE were identified using the AGDH population, which differed between glasshouse and field environments (Table 3). None of these coincided with QTL for KUtE in this population.

Fig. 6.
Relationships between shoot biomass and (A) potassium uptake efficiency (KUpE), expressed as plant K content, and (B) potassium utilisation efficiency (KUtE), expressed as the reciprocal of shoot K concentration, among B. oleracea genotypes assayed in ...
Table 3.
Chromosomal quantitative trait loci (QTL) associated with K+ uptake efficiency (KUpE) of Brassica oleracea grown in glasshouse experiment two (GE2) and field experiment two (FE2)

DISCUSSION

Genetic factors affecting shoot K concentration in B. oleracea

This study has demonstrated up to 3·4-fold variation in [K]shoot among B. oleracea genotypes when grown under identical conditions (Fig. 1), a significant proportion of which could be attributed to genetic factors (Table 1). This is consistent with previous studies showing 1·4- to 2·3-fold variation in [K]shoot among B. oleracea kales and collards grown together in the glasshouse or field (Kopsell et al., 2004; Vilar et al., 2008). The [K]shoot of B. oleracea genotypes correlated between glasshouse and field environments (Fig. 3A, D), and was relatively insensitive to P supply (Table 1, Fig. 2). When grown in the glasshouse, the variation in [K]shoot among commercial cultivars was about 72 % of the species-wide variation in [K]shoot, and the variation in [K]shoot among AGDH lines was about 60 % of the species-wide variation in [K]shoot (Fig. 1). This suggests an opportunity to alter [K]shoot through the introduction of alleles from the wider gene pool into elite germplasm.

Several QTL associated with [K]shoot were identified using the AGDH population, although these differed between glasshouse and field environments (Table 2). It is most likely that environmental factors affecting, for example, plant growth rates, rooting volume or differences in K availability determined the most influential physiological traits and therefore the QTL associated with [K]shoot in the glasshouse and field environments. It is unlikely that the different QTL arose as an artefact from the different numbers of AGDH lines employed for mapping QTL in glasshouse and field environments. One of the QTL associated with [K]shoot in glasshouse-grown plants (on chromosome C7 at 62·2 cM) was confirmed from the phenotypes of AGSL165 and AGSL168 (Fig. 5). This locus falls largely in a region of co-linearity with a section of arabidopsis chromosome 4 (Parkin et al., 2005). The closest genetic marker on the B. oleracea map is an orthologue of arabidopsis AtIRT1 (At4g19690) at 61·3 cM, and a gene encoding the putative plasma membrane K+ transporter AtKUP9 (At4g19960) is in the vicinity of AtIRT1 (White and Karley, 2009). Genes encoding the K+ channels AtAKT2 (At4g22200), AtKAT2 (At4g18290) and the putative K+ channel AtTPK3 (At4g18160) are also in close proximity to AtIRT1 (White and Karley, 2009). Similarly, this marker is present within a sequenced B. rapa BAC (KBrB085J21), which contains an orthologue of AtKUP9. The arabidopsis genome contains approximately 32 000 genes and about 450 of these are located between At4g18160 and At4g222000. Based on a two-way contingency table, the probability that four of the 111 genes encoding K+-transporters are found in this region (P = 0·048) suggests that it contains more K+-transporters than would be expected by chance (Karley and White, 2009). To confirm and resolve QTL associated with [K]shoot to smaller candidate regions a further backcrossing programme would now be appropriate using a subset of the AGSL lines.

No QTL associated with [K]shoot in the AGDH population observed in either the glasshouse or the field co-localized with any QTL associated with shoot DM accumulation or shoot P, Mg or Ca concentrations (Broadley et al., 2008; Hammond et al., 2009), despite highly significant (P < 0·001) negative relationships between [K]shoot and shoot Ca and Mg concentrations in these plants in the glasshouse (Fig. 4). Similarly, no QTL associated with [K]shoot in the AGDH population observed in either the glasshouse or the field co-localized with any QTL associated with shoot Na concentration (Table 4), despite the significant (P = 0·0042) positive correlation between [K]shoot and shoot Na concentration among AGDH lines assayed in the glasshouse (Fig. 4C) and the significant (P = 0·008) negative correlation between [K]shoot and shoot Na concentration among AGDH lines assayed in the field (data not shown). Nor did any QTL associated with [K]shoot co-localize with any QTL associated with aspects of seedling vigour previously identified in this population (Bettey et al., 2000). No QTL associated with [K]shoot in the AGDH population co-localized with any QTL associated with leaf N concentration in another B. oleracea genetic mapping population (NGDH) grown in the glasshouse (Hall et al., 2005), nor did any QTL associated with [K]shoot identified in the AGDH population co-localize with any of the QTL for the 17 morphological and developmental traits scored in the NGDH population (Sebastian et al., 2002). Thus, it would appear that the QTL associated with [K]shoot reported here were not associated with any QTL associated with plant growth rate or morphology, nor were they the consequence of pleiotropic effects of a non-specific QTL.

Table 4.
Chromosomal quantitative trait loci (QTL) associated with shoot Na concentration (%DM) of Brassica oleracea grown in glasshouse experiment two (GE2) and field experiment two (FE2)

Significant genetic variation in [K]shoot has also been observed among arabidopsis accessions, and genetic variation in [K]shoot among a recombinant inbred population of arabidopsis developed from a cross between the Landsberg erecta (Ler) and Cape Verde Island (CVI) accessions has allowed several QTL associated with this trait to be mapped (Harada and Leigh, 2006). Four QTL associated with [K]shoot expressed on a fresh weight basis were mapped to chromosomes 2 (at 40 cM), 4 (at 43 cM) and 5 (at 82 and 106 cM), and three QTL associated with [K]shoot in dry matter were mapped to chromosomes 3 (at 0 cM), 4 (at 43 cM) and 5 (at 106 cM). None of these QTL co-localized with QTL associated with either FW or DM accumulation, suggesting that [K]shoot and biomass accumulation were genetically independent under the environmental conditions in which these plants were grown. However, the QTL on chromosome 2 co-localized with a QTL for seed K concentration occasionally observed in this population (Waters and Grusak, 2008) and the QTL on the bottom of chromosome 3 co-localized with one of several QTL associated with seed K concentration in both this and three other genetic mapping populations of arabidopsis (Vreugdenhil et al., 2004; Ghandilyan et al., 2009). On chromosome 5 the QTL at 82 cM co-localized with a QTL for seed K concentration occasionally observed in two other genetic mapping populations of arabidopsis (Waters and Grusak, 2008; Ghandilyan et al., 2009), and the QTL on chromosome 5 at 106 cM co-localized with a QTL associated with shoot Cs concentration expressed on a fresh weight basis in the Ler × CVI population (Payne et al., 2004). Candidate genes within these chromosomal QTL include many putative cation transporters, including AtAKT1, AtAKT6, AtKAT1, AtSKOR, AtCNGC1, AtCNGC5, AtCNGC6, AtCNGC14, AtTPK1, AtTPK2, AtKCO3, AtHAK5, AtKEA5 and AtNHX3 (Harada and Leigh, 2006). Recently, large-scale ionomic profiling programmes have collected data on [K]shoot for many arabidopsis accessions and for several arabidopsis genetic mapping populations (Baxter et al., 2007). Rapid progress in linkage mapping and association genetics in arabidopsis (Nordborg and Weigel, 2008), together with the use of comparative genomic information (Parkin et al., 2005; Schranz et al., 2006, 2007), will facilitate the transfer of knowledge of QTL and genes affecting [K]shoot between arabidopsis and other species in the Brassicaceae.

Potassium use efficiency in B. oleracea

The efficiency with which K-fertilizers are used in crop production is of both commercial and environmental importance. Considerable variation in both KUpE and KUtE was observed among genotypes of B. oleracea when grown in either the glasshouse or the field (Fig. 6). Several QTL associated with KUpE (Table 3) and KUtE (1/[K]shoot, Table 2) were identified using the AGDH population, but these QTL differed between glasshouse and field environments, suggesting that they are influenced profoundly by environmental vagaries. Furthermore, no QTL for KUpE coincided with QTL for KUtE in either environment, and there was a strong negative correlation between these traits (Fig. 6C). These observations imply that several QTL will have to be combined to achieve improved KUpE and/or KUtE across diverse environments, and that the environmental factors influencing the relative importance of specific QTL will have to be defined.

Greater KUpE is generally attributed to better K+ acquisition from the soil solution (Jungk and Claassen, 1997; Baligar et al., 2001; Trehan, 2005; Rengel and Damon, 2008; White and Karley, 2009). Theoretical models suggest that diffusion through, and mass flow of, the soil solution contribute most to the delivery of K+ to the root surface (Jungk and Claassen, 1997). Differences in K+ acquisition between genotypes might therefore be attributed to: (1) the rate of K+ uptake across the plasma membrane of root cells, which reduces the K+ concentration in the rhizosphere solution and increases diffusional K+ fluxes; (2) the release of non-exchangeable K+ by root exudates, which increases K+ concentration and availability in the soil solution; (3) the proliferation of roots into the soil volume, which increases the area for K+ uptake and also reduces the distance required for K+ diffusion and water flow; and (4) the transpiration rate of the plant, which drives mass flow of the soil solution to the root (Jungk and Claassen, 1997; Baligar et al., 2001; Høgh-Jensen and Pedersen, 2003; Trehan, 2005; Rengel and Damon, 2008; White and Karley, 2009). It has been observed that Brassica genotypes differ greatly in their ability to reduce K+ concentrations in the rhizosphere (Shi et al., 2004), that members of the Brassicaceae access considerable quantities of soil K from the non-exchangeable fraction (Jungk and Claassen, 1997; Shi et al., 2004), that the identity and quantities of organic acids exuded by roots differ markedly between Brassica genotypes (Akhtar et al., 2006, 2008), and that the growth rate and architecture of the root system differ markedly between genotypes of B. oleracea (Hammond et al., 2009). Future studies should investigate these properties in B. oleracea genotypes with contrasting KUpE.

Greater KUtE, especially at low K supply, can be achieved by better K redistribution within a plant to tissues with immediate K requirements and/or by improving a plant's ability to maintain appropriate cytoplasmic K concentrations, either by anatomical adaptations or by the substitution of K+ for other solutes, such as Ca2+ or Na+, in the vacuole (Rengel and Damon, 2008; White and Karley, 2009). Future studies should investigate these properties in B. oleracea genotypes with contrasting KUtE.

There was a highly significant (P < 0·001) negative correlation between KUpE and KUtE among B. oleracea genotypes in both the glasshouse and the field (e.g. Fig. 6C). In addition, there was a highly significant (P < 0·001) correlation between shoot biomass and KUpE (Fig. 6A), but no relationship between shoot biomass and KUtE (Fig. 6B). This suggests that shoot biomass and KUtE are genetically independent. This contrasts with a previous study of 84 canola (B. napus) genotypes, which suggested that growth responses to low soil K phytoavailability in glasshouse trials could be determined by either KUpE or KUtE depending upon the genotype (Damon et al., 2007). In their study, Damon et al. (2007) classified canola genotypes as K-efficient based on high values for the shoot DM ratio at deficient versus adequate K supply. They observed that growth rates of K-efficient genotypes differed considerably at low K supply and concluded that K-efficient genotypes with high growth rates at low K supply have the ability to improve yields irrespective of K supply. Several B. oleracea accessions studied here can similarly be identified as having high yields and high KUtE or KUpE when grown with an adequate P supply (Fig. 6).

CONCLUSIONS

Considerable variation in [K]shoot was observed among B. oleracea genotypes grown in the glasshouse or the field, a significant proportion of which (between 10 and 25 %) could be attributed to genetic factors. This should be sufficient for breeding for KUtE in B. oleracea. However, although several QTL associated with [K]shoot were identified using the AGDH genetic mapping population, and despite a significant correlation in [K]shoot among these genotypes grown in the glasshouse and field, QTL differed between glasshouse and field environments. One of the QTL associated with [K]shoot in glasshouse-grown plants (chromosome C7 at 62·2 cM) was confirmed from the phenotypes of AGSL165 and AGSL168. This QTL corresponds to a segment of arabidopsis chromosome 4 that contains genes encoding the plasma membrane K+-transporter AtKUP9 (At4g19960) and the K+ channels AtAKT2 (At4g22200), AtKAT2 (At4g18290) and AtTPK3 (At4g18160). Agronomic K use efficiency is the product of KUpE and KUtE. In B. oleracea, KUpE correlated strongly with shoot biomass, but KUtE (1/[K]shoot) did not. This implies that KUtE and biomass can be genetically manipulated independently. In the context of conventional agriculture, breeding for increased KUtE and KUpE will decrease crop K requirements and K-fertilizer applications. However, as QTL impacting these traits differ between glasshouse and field environments, marker-assisted breeding programmes must consider carefully the conditions under which the crop will be grown.

ACKNOWLEDGEMENTS

This work was supported by the UK Biotechnology and Biological Sciences Research Council, the UK Department for Environment, Food and Rural Affairs, and the Scottish Government Rural and Environment Research and Analysis Directorate.

LITERATURE CITED

  • Akhtar MS, Oki Y, Adachi T, Murata Y, Khan MHR. Phosphorus starvation induced root-mediated pH changes in solubilization and acquisition of sparingly soluble P sources and organic acids exudation by Brassica cultivars. Soil Science and Plant Nutrition. 2006;52:623–633.
  • Akhtar MS, Oki Y, Adachi T. Genetic variability in phosphorus acquisition and utilization efficiency from sparingly soluble P-sources by Brassica cultivars under P-stress environment. Journal of Agronomy and Crop Science. 2008;194:380–392.
  • Baligar VC, Fageria NK, He ZL. Nutrient use efficiency in plants. Communications in Soil Science and Plant Analysis. 2001;32:921–950.
  • Baxter I, Ouzzani M, Orcun S, Kennedy B, Jandhyala SS, Salt DE. Purdue Ionomics Information Management System. An integrated functional genomics platform. Plant Physiology. 2007;143:600–611. [PubMed]
  • Bettey M, Finch-Savage WE, King GJ, Lynn JR. Quantitative genetic analysis of seed vigour and pre-emergence seedling growth traits in Brassica oleracea. New Phytologist. 2000;148:277–286.
  • Bhardwaj HL, Mamama AA. Effect of cultivar and growing location on the mineral composition of canola sprouts. HortScience. 2009;44:508–511.
  • Bradstreet RB. The Kjeldahl method for organic nitrogen. London: Academic Press; 1965.
  • Brennan RF, Bolland MDA. Comparing the potassium requirements of canola and wheat. Australian Journal of Agricultural Research. 2007;58:359–366.
  • Broadley MR, Bowen HC, Cotterill HL, et al. Phylogenetic variation in the shoot mineral concentration of angiosperms. Journal of Experimental Botany. 2004;55:321–336. [PubMed]
  • Broadley MR, Hammond JP, King GJ, et al. Shoot calcium and magnesium concentrations differ between subtaxa, are highly heritable, and associate with potentially pleiotropic loci in Brassica oleracea. Plant Physiology. 2008;146:1707–1720. [PubMed]
  • Cogan NOI, Newbury HJ, Oldacres AM, et al. Identification and characterization of QTL controlling Agrobacterium-mediated transient and stable transformation of Brassica oleracea. Plant Biotechnology Journal. 2004;2:59–69. [PubMed]
  • Damon PM, Osborne LD, Rengel Z. Canola genotypes differ in potassium efficiency during vegetative growth. Euphytica. 2007;156:387–397.
  • Fageria NK. The use of nutrients in crop plants. Boca Raton, FL: CRC Press; 2009.
  • Ghandilyan A, Ilk N, Hanhart C, et al. A strong effect of growth medium and organ type on the identification of QTLs for phytate and mineral concentrations in three Arabidopsis thaliana RIL populations. Journal of Experimental Botany. 2009;60:1409–1425. [PubMed]
  • Greenwood DJ, Stellacci AM, Meacham MC, Broadley MR, White PJ. Phosphorus response components of different Brassica oleracea genotypes are reproducible in different environments. Crop Science. 2005;45:1728–1735.
  • Greenwood DJ, Stellacci AM, Meacham MC, Mead A, Broadley MR, White PJ. Relative values of physiological parameters of P response of different genotypes can be measured in experiments with only two P treatments. Plant and Soil. 2006;281:159–179.
  • Hall NM, Griffiths H, Corlett JA, Jones HG, Lynn J, King GJ. Relationships between water-use traits and photosynthesis in Brassica oleracea resolved by quantitative genetic analysis. Plant Breeding. 2005;124:557–564.
  • Hammond JP, Broadley MR, White PJ, et al. Shoot yield drives phosphorus use efficiency in Brassica oleracea and correlates with root architecture traits. Journal of Experimental Botany. 2009;60:1953–1968. [PubMed]
  • Harada H, Leigh RA. Genetic mapping of natural variation in potassium concentrations in shoots of Arabidopsis thaliana. Journal of Experimental Botany. 2006;57:953–960. [PubMed]
  • Himelblau E, Gilchrist EJ, Buono K, et al. Forward and reverse genetics of rapid-cycling Brassica oleracea. Theoretical and Applied Genetics. 2009;118:953–961. [PubMed]
  • Høgh-Jensen H, Pedersen MB. Morphological plasticity by crop plants and their potassium use efficiency. Journal of Plant Nutrition. 2003;26:969–984.
  • Jungk A, Claassen N. Ion diffusion in the soil-root system. Advances in Agronomy. 1997;61:53–110.
  • Karley AJ, White PJ. Moving cationic minerals to edible tissues: potassium, magnesium, calcium. Current Opinion in Plant Science. 2009;12:291–298. [PubMed]
  • Kopsell DE, Kopsell DA, Lefsrud MG, Curran-Celentano J. Variability in elemental accumulations among leafy Brassica oleracea cultivars and selections. Journal of Plant Nutrition. 2004;27:1813–1826.
  • Lægreid M, Bøckman OC, Kaarstad O. Agriculture, fertilizers and the environment. Wallingford, UK: CABI Publishing; 1999.
  • Nordborg M, Weigel D. Next-generation genetics in plants. Nature. 2008;456:720–723. [PubMed]
  • Parkin IAP, Gulden SM, Sharpe AG, et al. Segmental structure of the Brassica napus genome based on comparative analysis with Arabidopsis thaliana. Genetics. 2005;171:765–781. [PubMed]
  • Patterson HD, Thompson R. Recovery of inter-block information when block sizes are unequal. Biometrika. 1971;58:545–554.
  • Patterson HD, Williams ER. A new class of resolvable incomplete block designs. Biometrika. 1976;63:83–92.
  • Payne KA, Bowen HC, Hammond JP, et al. Natural genetic variation in caesium (Cs) accumulation by Arabidopsis thaliana. New Phytologist. 2004;162:535–548.
  • Pelletier S, Bélanger G, Tremblay GF, Virkajärvi P, Allard G. Timothy mineral concentration and derived indices related to cattle metabolic disorders: a review. Canadian Journal of Plant Science. 2008;88:1043–1055.
  • Pettigrew WT. Potassium influences on yield and quality production for maize, wheat, soybean and cotton. Physiologia Plantarum. 2008;133:670–681. [PubMed]
  • Pink D, Bailey L, McClement S, et al. Double haploids, markers and QTL analysis in vegetable brassicas. Euphytica. 2008;164:509–514.
  • Rae AM, Howell EC, Kearsey MJ. More QTL for flowering time revealed by substitution lines in Brassica oleracea. Heredity. 1999;83:586–596. [PubMed]
  • Rengel Z, Damon PM. Crops and genotypes differ in efficiency of potassium uptake and use. Physiologia Plantarum. 2008;133:624–636. [PubMed]
  • Robinson DL. Estimation and use of variance components. Statistician. 1987;36:3–14.
  • Rose TJ, Rengel Z, Ma Q, Bowden JW. Differential accumulation patterns of phosphorus and potassium by canola cultivars compared to wheat. Journal of Plant Nutrition and Soil Science. 2007;170:404–411.
  • Schranz ME, Lysak MA, Mitchell-Olds T. The ABC's of comparative genomics in the Brassicaceae: building blocks of crucifer genomes. Trends in Plant Science. 2006;11:535–542. [PubMed]
  • Schranz ME, Song B-H, Windsor AJ, Mitchell-Olds T. Comparative genomics in the Brassicaceae: a family-wide perspective. Current Opinion in Plant Biology. 2007;10:168–175. [PubMed]
  • Sebastian RL, Howell EC, King GJ, Marshall DF, Kearsey MJ. An integrated AFLP and RFLP Brassica oleracea linkage map from two morphologically distinct doubled-haploid mapping populations. Theoretical and Applied Genetics. 2000;100:75–81.
  • Sebastian RL, Kearsey MJ, King GJ. Identification of quantitative trait loci controlling developmental characteristics of Brassica oleracea L. Theoretical and Applied Genetics. 2002;104:601–609. [PubMed]
  • Shi W, Wang X, Yan W. Distribution patterns of available P and K in rape rhizosphere in relation to genotypic difference. Plant and Soil. 2004;261:11–16.
  • Szczerba MW, Britto DT, Kronzucker HJ. K+ transport in plants: physiology and molecular biology. Journal of Plant Physiology. 2009;166:447–466. [PubMed]
  • Trehan SP. Nutrient management by exploiting genetic diversity of potato – a review. Potato Journal. 2005;32:1–15.
  • Vilar M, Cartea ME, Padilla G, Soengas P, Velasco P. The potential of kales as a promising vegetable crop. Euphytica. 2008;159:153–165.
  • Vreugdenhil D, Aarts MGM, Koornneef M, Nelissen H, Ernst WHO. Natural variation and QTL analysis for cationic mineral content in seeds of Arabidopsis thaliana. Plant, Cell and Environment. 2004;27:828–839.
  • Wang S, Basten CJ, Zeng Z-B. Windows QTL Cartographer 2·0. Raleigh, NC: Department of Statistics, North Carolina State University; 2004.
  • Watanabe T, Broadley MR, Jansen S, et al. Evolutionary control of leaf element composition in plants. New Phytologist. 2007;174:516–523. [PubMed]
  • Waters BM, Grusak MA. Quantitative trait locus mapping for seed mineral concentrations in two Arabidopsis thaliana recombinant inbred populations. New Phytologist. 2008;179:1033–1047. [PubMed]
  • White PJ, Karley AJ. Potassium. In: Hell R, Mendel R, editors. Cell biology of metals and nutrients in plants. Dordrecht: Springer; 2009. in press.
  • White PJ, Broadley MR, Greenwood DJ, Hammond JP. Proceedings of the International Fertiliser Society 568. Genetic modifications to improve phosphorus acquisition by roots. York, UK: International Fertiliser Society; 2005.
  • Whitfield WAD. Report of the National Vegetable Research Station for 1973. Stratford-upon-Avon, UK: Herald Press; 1974. The soils of the National Vegetable Research Station, Wellesbourne; pp. 21–30.
  • Wu J, Yuan Y-X, Zhang X-W, et al. Mapping QTLs for mineral accumulation and shoot dry biomass under different Zn nutritional conditions in Chinese cabbage (Brassica rapa L. ssp. pekinensis) Plant and Soil. 2008;310:25–40.

Articles from Annals of Botany are provided here courtesy of Oxford University Press