Linkage mapping of maize and barley myo-inositol 1-phosphate synthase DNA sequences: correspondence with a low phytic acid mutation

 We sequenced and genetically mapped the myo-inositol 1-phosphate synthase (MIPS) genes of maize (Zea mays L.) and barley (Hordeum vulgare L). Our objective was to determine whether the genetic map positions of these MIPS loci correspond with the location of the low phtyic acid 1 (lpa1) mutations that were previously identified in maize and barley. Seven MIPS-homologous sequences were mapped to positions on maize chromosomes 1S, 4L, 5S, 6S, 8L, 9S and 9L, and a similar number of divergent MIPS sequences were amplified from maize. To the extent that we can compare across different genetic mapping populations, the position of the MIPS gene on maize chromosome 1S is identical to the location of the maize lpa1 mutation. However, only one MIPS sequence was identified in barley and this gene was mapped to a locus on chromosome 4H that is separate from the barley lpa1 mutation on chromosome 2H. Although several RFLP markers linked to the barley MIPS gene on chromosome 4H also detect loci near barley lpa1 on chromosome 2H, our experiments failed to reveal a second MIPS gene that could be associated with the barley lpa1 mutation. Therefore, genetic mapping results from this study support the MIPS candidate-gene hypothesis for maize lpa1, but do not support the MIPS candidate-gene-hypothesis for barley lpa1. These opposing results contradict the hypothesis that maize lpa1 and barley lpa1 are mutations of orthologous genes, which is suggested by the similar biochemical phenotypes of these mutants. Yet, comparisons of RFLP mapping studies show loci that are homologous between maize chromosome 1S, barley chromosome 4H and barley chromosome 2H, including regions flanking the respective MIPS and/or lpa1 loci. This putative relationship, between the regions flanking the lpa1 mutations on maize 1S and barley 2H, also supports the assertion that these mutations are orthologous despite contradictory results between our maize and barley candidate-gene experiments. Received: 24 August 1998 / Accepted: 19 December 1998     

Deletion-based physical mapping of barley chromosome 7H

Chromosomal mutations in barley (Hordeum vulgare, 2n=2x=14, HH) chromosome 7H added to the common wheat (Triticum aestivum, 2n=6x=42, AABBDD) cultivar Chinese Spring were induced genetically by the gametocidal activity of certain alien chromosomes derived from wild species of the genus Aegilops. The rearranged barley chromosomes were characterized by C-banding, FISH and GISH. Twenty two deletion or translocation chromosomes in a hemizygous condition were selected for deletion mapping of 17 AFLP and 28 STS markers that are specific to 7H. Of the 22 breakpoints in chromosome 7H, seven involved the short arm (7HS), 12 the long arm (7HL) and three were in the centromeric region. The seven 7HS breakpoints separated all four 7HS-specific AFLP markers and split the 21 STS markers into six groups. One breakpoint occurred between two STS markers formerly occupying the same position in the genetic map. All seven 7HS breakpoints were separated from each other by either the AFLP or STS markers. The 12 breakpoints in 7HL divided the 13 7HL-specific AFLP markers into seven groups, and the seven STS markers into three groups. On the other hand, the 12 breakpoints in 7HL were divided into six groups by the AFLP markers and into two groups by the STS markers. This deletion-based map was in accordance with previously published genetic and physical maps using the same STS markers. The breakpoints, AFLP markers and STS markers were arrayed in a consistent order. Received: 5 February 2001 / Accepted: 19 February 2001     

RFLP mapping of manganese efficiency in barley

In many cropping regions of the world, yield is limited by the availability of micronutrients, and micronutrient-efficient cultivars provide a yield advantage. Traditional methods of testing cultivars for micronutrient efficiency are time-consuming and laborious. Molecular markers linked to loci controlling micronutrient efficiency will allow more rapid and efficient selection and introgression of these traits than is currently possible. Using a pot-based bioassay and bulked segregant analysis of an F₂ population, we have identified several RFLPs (grouped distally on chromosome 4HS) linked to a locus for manganese efficiency in barley. This manganese efficiency locus has been designated Mel1. Pot bioassay analysis of intercrosses suggests that three useful sources of manganese efficiency are likely to be allelic at the Mel1 locus. Field evaluation of marker selected F₄ progeny supports the major role of Mel1 in the genetic control of manganese efficiency. Adoption of marker assisted selection for this trait in the Southern Australian barley breeding program has occurred. This has been facilitated by the demonstration that the Mel1 allele of Amagi Nijo can be distinguished from 95 other locally useful varieties and breeder’s lines on the basis of RFLPs identified by just two molecular markers. Received: 20 October 1999 / Accepted: 18 February 2000     

EST analysis in barley defines a unigene set comprising 4,000 genes

We report the generation of 13,109 EST (Expressed Sequence Tag) sequences from barley as a first step towards the generation of a unigene set for this organism. Sequences were generated from three libraries encompassing 7,568 cDNA clones. Comparisons to nucleic acid and protein sequence databases enabled the assignment of putative functions to the mRNAs. The results of the searches against protein databases were parsed and built into a regularly updated database, available over the World Wide Web. The Stack_Pack clustering system has been applied to survey the level of redundancy, which was calculated to amount to 69%, thus we identified 4,000 different barley genes. To prove the usability of the results of the clustering process for further experiments, we subjected alignments with sequences similar to elongation factor 1 alpha to additional analysis. These sequences represented the largest group with identical putative functions (228 members) and clustering based on the analysis of 3′ sequences subdivided the group into five different assemblies. Alignments of the consensus sequences facilitated the development of PCR assays suitable for genetic mapping of four of the different gene-family members, which reside on chromosomes 2H, 4H and 5H, thus demonstrating the suitability of the cluster-results as a basis for in-depth analyses of barley gene families. Received: 15 March 2001 / Accepted: 18 April 2001     

Molecular mapping of the photoperiod response gene ea7 in barley

 The gene ea ₇ determining photoperiod insensitivity under short day length was mapped on the short arm of chromosome 6H near the centromere. The gene was linked to the two flanking markers Xmwg2264 and Xmwg916 by 6.7 and 13.0 cM, respectively. Compared to Ppd-H1 (chromosome 2H) and Ppd-H2 (chromosome 1H), ea ₇ determines the strongest effect on flowering time with 55 and 18 days difference compared to photoperiod sensitive genotypes grown under short and long photoperiods, respectively. Allelic and homoeologous relationships to major genes and quantitative trait loci controlling flowering time in barley and wheat are discussed. Received: 10 March 1998 / Accepted: 7 April 1998     

Molecular breeding for grain yield in barley: an evaluation of QTL effects in a spring barley cross

 We report results from a breeding strategy designed to accumulate favorable QTL alleles for grain yield identified in the SteptoeבMorex’ (SM) barley germplasm. Two map lines (SM73 and SM145) from the original mapping population were selected based on their marker genotype and QTL structure. When crossed, these lines would be expected to produce progeny with most favorable QTL alleles. One hundred doubled haploid (DH) lines from the F₁ hybrid of this cross were genotyped with ten RFLP markers and one morphological marker defining grain yield to monitor QTL segregation. A subset of 24 lines representing various combinations of putatively favorable and unfavorable QTL alleles, together with Steptoe, ‘Morex’, SM73, and SM145, were phenotyped for grain yield in five environments. Multiple regression procedures were used to explore phenotype and genotype relationships. Most target QTLs showed significant effects. However, significance and magnitude of QTL effects and favorable QTL allele phase varied across environments. All target QTLs showed significant QTL-by-environment interaction (QTL×E), and the QTL on chromosome 2 expressed alternative favorable QTL alleles in different environments. Digenic epistatic effects were also detected between some QTL loci. For traits such as grain yield, marker-assisted selection efforts may be better targeted at determining optimum combinations of QTL alleles rather than pyramiding alleles detected in a reference mapping population. Received: 2 June 1998 / Accepted: 17 September 1998     

Identification and mapping of a new leaf stripe resistance gene in barley (Hordeum vulgare L.)

Pyrenophora graminea is the seed-borne pathogen causal agent of barley leaf stripe disease. Near-isogenic lines (NILs) carrying resistance of the cv ”Thibaut” against the highly virulent isolate Dg2 were obtained by introgressing the resistance into the genetic background of the susceptible cv ”Mirco”. The segregation of the resistance gene was followed in a F₂ population of 128 plants as well as on the F₃ lines derived from the F₂ plants; the segregation fitted the 1:2:1 ratio for a single gene. By using NILs, a RAPD marker associated with the resistance gene was identified; sequence-specific (STS) primers were designed on the basis of the amplicon sequence and a RILs mapping population with an AFLP-based map were used to position this molecular marker to barley chromosome 1 S (7HS). STS and CAPS markers were developed from RFLPs mapped to the telomeric region of barley chromosome 7HS and three polymorphic PCR-based markers were developed. The segregation of these markers was followed in the F₂ population and their map position with respect to the resistance gene was determined. Our results indicate that the Thibaut resistance gene, which we designated as Rdg2a, maps to the telomeric region of barley chromosome 7HS and is flanked by the markers OPQ-9₇₀₀ and MWG 2018 at distances of 3.1 and 2.5 cM respectively. The suitability of the PCR-based marker MWG2018 in selection- assisted barley breeding programs is discussed. Received: 22 June 2000 / Accepted: 16 October 2000     

Stability of Hor1-specific YAC-clones and physical mapping of Hor1-loci in barley

 The hordeins are the major class of storage proteins in barley and are encoded by multigene families. Two YAC-clones specific for the C-hordein-coding Hor1-locus of barley (Hordeum vulgare L.) were selected. The clones were constructed with DNA from the cultivars ‘Franka’ and ‘Hockey’ and have insert sizes of 330 kb and 350 kb, respectively. Performing partial digestions and hybridizations with vector-specific probes, a restriction analysis was conducted using restriction enzymes with a 8-bp recognition sequence. Both clones cover the complete region of the Hor1-locus, but exhibit a different pattern of restriction sites reflecting the polymorphic nature of the locus on the scale of long-range restriction mapping. The maximal extent of the regions homologous to the Hor1-specific probe, pBSC5, was 105 kb in the ‘Hockey’-derived YAC and 190 kb in the yeast artificial chromosome constructed with ‘Franka’-DNA. Furthermore the high degree of instability observed with the Hor1-specific YAC-clones is discussed in conjunction with the structure of the Hor1-locus. Received: 19 December 1996 / Accepted: 31 January 1997     

Further evidence supporting Morocco as a centre of origin of barley

 Thirty-five populations of H. spontaneum from nine countries, encompassing almost all the known range of distribution of the species, Afghanistan, Crete (Greece), Cyprus, Iran, Iraq, Israel, Libya, Morocco and Turkey, were studied utilizing RFLP markers (21 probes with three restriction enzymes) distributed across all seven barley chromosomes in an attempt to unveil the genetic dissimilarities existing among them. UPGMA clustering, based on the Nei and Li (1979) similarity coefficient, produced a dendrogram where three clusters could be defined: two with a clear geographical distinction (Morocco and Cyprus) and another one grouping all the Asian/Middle Eastern populations, except for an accession from Iran that clustered separately. These results confirm our previous work and suggest that barley domestication could also have taken place outside the Fertile Crescent, particularly in Morocco. Received: 10 October 1998 / Accepted: 27 October 1998     

Detection of loci controlling seed dormancy on group 4 chromosomes of wheat and comparative mapping with rice and barley genomes

Three quantitative trait loci (QTLs) controlling seed dormancy were detected on group 4 chromosomes of wheat (Triticum aestivum L.) using 119 doubled haploid lines (DHLs) derived from a cross between AC Domain and Haruyutaka. A major QTL, designated QPhs.ocs-4A.1, was identified within the marker interval between Xcdo795 and Xpsr115 in the proximal region of the long arm of chromosome 4A. Two minor QTLs, QPhs.ocs-4B.2 on 4B and QPhs.ocs-4D.2 on 4D, were flanked by common markers, Xbcd1431.1 and Xbcd1431.2 in the terminal region of the long arms, suggesting a homoeologous relationship. These three QTLs explained more than 80% of the total phenotypic variance in seed dormancy of DHLs grown in the field and under glasshouse conditions. The AC Domain alleles at the three QTLs contributed to increasing seed dormancy. Comparative maps across wheat, barley and rice demonstrated the possibility of a homoeologous relationship between QPhs.ocs-4A.1 and the barley gene SD4, while no significant effects of the chromosome regions of wheat and barley orthologous to rice chromosome 3 region carrying a major seed dormancy QTL were detected. Received: 5 June 2000 / Accepted: 31 August 2000     

Characterization of two antifungal endochitinases from barley grain

A basic chitinase (chitinase T, EC 3.2.1.14, molecular mass 33 kDa, pI 9.8) was isolated and compared with a previously described chitinase (chitinase C, molecular mass 28 kDa, pI 9.7). The two chitinases were isolated in homogeneous form from barley ( Hordeum vulgare L.) Bomi mutant 1508 grains either by two cation exchange steps or by one affinity step followed by cation exchange. Both chitinases are endochitinases with specific activities of 168 and 54 nkat (mg protein)⁻¹ for chitinase T and chitinase C, respectively. Both inhibit the growth of Trichoderma viride efficiently. The lysozyme activity of both chitinases is 10⁴ times lower than that of hen egg-white lysozyme as measured by lysis of cell walls of Micrococcus lysodeikticus . The amino acid composition and two partial amino acid sequences of chitinase T were determined. A 23 residue sequence of the N-terminal domain of chitinase T, which was not present in chitinase C, showed 73% identity with domain B of wheat germ lectin and 65% identity with the N-terminal domain of an endochitinase from bean leaves (deduced from cDNA). A 9 amino acid sequence of a cyanogen bromide fragment of chitinase T was identical with a cDNA deduced sequence of a barley aleurone endochitinase but differed in one residue from chitinase C. Generally, the two grain chitinases have physico-chemical and enzymatic properties similar to the plant leaf chitinases characterized. Both chitinases are localized in the aleurone layer and starchy endosperm of developing and germinating grain, but not in the embryo. The appearance of chitinases T and C at a late state of grain development suggests a role for these enzymes as a defense against fungi in the quiescent and germinating grain.     

Optimization of gene transfer into barley (Hordeum vulgare L.) mature embryos by tissue electroporation

 This study was conducted to detect the optimum conditions for DNA transfer into mature embryos of barley via electroporation. Cultured mature embryos of barley were directly electroporated in the presence of the pBI 121 vector carrying both the β-glucuronidase (GUS) and neomycin phosphotransferase II (npt II) genes. It was found that 500 v/cm and 500 μFd capacitance was the optimum combination for healthy germination of the transformed plants from mature electroporated embryos. Effects of culture duration before electroporation and selection antibiotic concentrations on germination were also examined. Gene transfer performed on 3-day-old cultures resulted in the highest germination frequencies. GUS expression was observed on transversal sections of embryos and mature leaves from 3 month-old regenerants. PCR and Southern blot analyses show the presence of the npt II transgene in the genome of a plant. Received: 15 June 1999 / Revision received: 27 September 1999 / Accepted: 26 October 1999     

Molecular mapping of the Rph7.g leaf rust resistance gene in barley (Hordeum vulgare L.)

In many temperate areas of the world, leaf rust is becoming an important disease of barley. In the last decade, new races of Puccinia hordei G. Otth have emerged which are virulent against the so-far most-effective race-specific resistance genes, such as Rph7. Marker-assisted selection greatly facilitates the pyramidization of two or more resistance genes in a single variety in order to achieve a more comprehensive resistance. Such a strategy requires the development of efficient and reliable markers. Here, we have developed a linkage map and found RFLP markers closely linked to the Rph7.g resistance gene on chromosome 3HS of barley. The receptor-like kinase gene Hv3Lrk that maps at 3.2 cM from Rph7.g was used to develop a PCR-based marker by exploiting a single nucleotide polymorphism. This marker was detected in 11 out of 12 (92%) barley lines having Rph7 and represents a valuable tool for marker-assisted selection. In addition, the identification of markers flanking Rph7.g provides the basis for positional cloning of this gene. Received: 1 December 1999 / Accepted: 28 February 2000     

Mapping and sequence analysis of barley hordoindolines

Barley (Hordeum vulgare L.) cultivars vary in traits such as grain hardness and malt quality. However, little is known about the genetic basis of these grain quality traits in barley, while more is known about the basis of grain hardness in wheat (Triticum aestivum L.). Puroindolines are endosperm-specific proteins found in wheat and barley, as well as other members of the Triticeae. In wheat, variation of puroindoline sequence is associated with most of the variability in wheat grain texture. However, no information exists on sequence variation of the barley homologs of puroindolines, the hordoindolines. We have therefore chosen to isolate and characterize the hordoindoline (hin) sequences of eight North American barley cultivars. The barley sequences contain numerous non-conservative amino-acid substitutions relative to their wheat counterparts. However, no significant rearrangements were found in either hinA or hinB of barley. Three hinA and two hinB sequence types were found among the eight barley cultivars examined, indicating substantial allelic variation at this locus. The hinB sequence variability was used to map hinB to the short arm of chromosome 5H in a Steptoe/Morex mapping population, which is coincident with the previously mapped location of hinA and Gsp (grain-softness protein). This chromosomal location also coincides with a small barley malt-extract QTL, suggesting that hordoindoline sequence variation may play a small role in barley grain quality. Efforts to correlate barley seed textural differences and malting quality with hordoindoline sequence type are ongoing. Received: 25 May 2000 / Accepted: 21 September 2000     

The two genes homologous to Arabidopsis FAE1 co-segregate with the two loci governing erucic acid content in Brassica napus

 KCS (β-keto-acyl-CoA synthase) has been proposed as a candidate gene for explaining the erucic acid level in rapeseed. Degenerate PCR primers corresponding to the FAE1 gene have been designed. Two B. napus genes BN-FAE1.1 and BN-FAE1.2, corresponding to the parental species B. rapa and B. oleracea FAE1 genes, were amplified. Polymorphism was revealed for these two genes by acrylamide electrophoresis of the amplification products. These two genes could then be mapped and a co-segregation of these genes with the E1 and E2 loci controlling erucic acid content was found. Furthermore, mutations observed for one of these genes could explain part of the low erucic trait of the three LEAR types used in this study. Received: 3 November 1997 / Accepted: 25 November 1997     

Molecular characterization and genetic mapping of random amplified microsatellite polymorphism in barley

 This study has analyzed the molecular basis and genetic behaviour of the polymorphism generated by the amplification of barley genomic DNA with primers complementary to microsatellites. Primers anchored at the 5′ end, used alone or in combination with arbitrary sequence primers, generated random amplified microsatellite polymorphisms (RAMPs). Unanchored primers were also used as single primers in a microsatellite primed-PCR (MP-PCR). Twenty six randomly selected RAMP DNA fragments which showed polymorphism between the cultivars Steptoe and Morex were cloned and sequenced. All sequences showed the expected repeated motif at the end of the insert, with the number of repeats ranging from five to ten. Genomic sequences containing low numbers of microsatellite motifs were preferentially amplified; therefore, only a fraction of the polymorphism could be attributed to variation in the number of microsatellite motifs at the priming site. Some sequences contained either cryptic simple sequences or members of families of repeated DNA. Polymorphism at the internal cryptic simple sequences was detected by RAMP bands inherited as co-dominant markers. Four MP-PCR bands were cloned and sequenced. A number of repeats identical to the primer itself were found at each end of the insert. Two allelic bands were polymorphic for an internal microsatellite. The potential use of cloned bands as fingerprinting tools was investigated by employing them as hybridization probes in Southern blots containing digested barley DNA from a sample of cultivars. RAMP probes produced complex hybridization band patterns. MP-PCR probes produced either a highly variable single locus or low-copy number loci. Segregations for 31 RAMPs and three MP-PCR bands were studied in a population of 70 doubled-haploids from the Steptoe/Morex cross. One third of all markers were co-dominantly inherited. Markers were positioned on an RFLP map and found to be distributed in all barley chromosomes. The new markers enlarged the overall length of the map to 1408 cM. Received: 6 May 1998 / Accepted: 20 July 1998     

Strong extracellular nuclease activity displayed by barley (Hordeum vulgare L.) uninucleate microspores

 Electroporation is becoming an increasingly important technique for plant transformation. Nevertheless, no positive results were achieved in barley when uninucleate microspores were used as target cells. Since it was previously demonstrated that electric shocks create pores in the microspore cell wall, experiments were designed to verify the presence of nucleases in the electroporation mix. Aliquots of all the solutions used for microspore extraction, purification and transformation were collected and analysed using supercoiled pBI 221 as a substrate; a nuclease activity was detected in all samples. Though microspore rinsing removed most nucleolytic activity in the supernatants, DNA preservation in the electroporation buffer was difficult to achieve, because microspores appeared capable of synthesising and releasing endonucleases at any time. Microspore chilling at 0°C was fairly effective in reducing nuclease secretion in the mix, whereas 1%PEG or 10 mM EDTA maintained most of the DNA in a supercoiled or circular relaxed form. EDTA effects were counterbalanced by Mg²⁺, but not Ca²⁺ or Zn²⁺, and enhanced by Mn²⁺. Barley microspore nucleases actively degraded different DNAs as well as TMV RNA, and apparently had a molecular weight above 30 kDa. Nuclease inactivation with EDTA did not alter microspore viability and allowed a transient expression of the uidA gene in electroporated barley microspores. Received: 13 January 1997 / Accepted: 28 February 1997     

The maize rp1 rust resistance gene identifies homologues in barley that have been subjected to diversifying selection

A number of agronomically important grasses (sorghum, wheat, panicum, sugar cane, oats, rice and barley) are shown to contain sequences homologous to rp1, a maize gene that confers race-specific resistance to the rust fungus Puccinia sorghi. Mapping of rp1-related sequences in barley identified three unlinked loci on chromosomes 1HL, 3HL and 7HS. The locus located on chromosome 7HS comprises a small gene family of at least four members, two of which were isolated and are predicted to encode nucleotide binding site-leucine-rich repeat (NBS-LRR) proteins that are respectively 58% and 60% identical to the maize rp1 protein. Evidence of positive selection for sequence diversification acting upon these two barley genes was observed; however, diversifying selection was restricted to the carboxy terminal half of the LRR domain. One of these rp1 homologous genes cosegregated with the barley Rpg1 stem rust resistance gene amongst 148 members of the Steptoe × Morex double haploid mapping family. Three other unrelated resistance gene-like sequences, potentially encoding NBS-LRR proteins, are also shown to be linked to the Rpg1 locus but not cosegregating with the gene. Received: 2 August 1999 / Accepted: 28 September 1999     

Inhibition of xanthine oxidase by phytic acid and its antioxidative action

We examined if phytic acid inhibits the enzymatic superoxide source xanthine oxidase (XO). Half inhibition of XO by phytic acid (IC50) was about 30 mM in the formation of uric acid from xanthine, but generation of the superoxide was greatly affected by phytic acid; the IC50 was about 6 mM, indicating that the superoxide generating domain of XO is more sensitive to phytic acid. The XO activity in intestinal homogenate was also inhibited by phytic acid. However, it was not observed with intestinal homogenate that superoxide generation was more sensitive to phytic acid compared with the formation of uric acid as observed with XO from butter milk. XO-induced superoxide-dependent lipid peroxidation was inhibited by phytic acid, but not by myo-inositol. Reduction of ADP-Fe3+ caused by XO was inhibited by superoxide dismutase, but not phytic acid. The results suggest that phytic acid interferes with the formation of ADP-iron-oxygen complexes that initiate lipid peroxidation. Both phytic acid and myo-inositol inhibited XO-induced superoxide-dependent DNA damage. Mannitol inhibited the DNA strand break. Myo-inositol may act as a hydroxyl radical scavenger. The antioxidative action of phytic acid may be due to not only inhibiting XO, but also preventing formation of ADP-iron-oxygen complexes.     

Novel mutations affecting leaf stearate content and plant size in Arabidopsis

 The fab2-1 mutant of Arabidopsis is an extreme dwarf as a direct result of an increase in the levels of stearate (18 : 0) in membrane lipids. We isolated a series of lines in which second-site suppressor mutations partly alleviate the dwarf phenotype. In all four of the suppressor lines examined, restoration of more normal morphology is accompanied by decreases in leaf 18 : 0 content. Three of the isolated suppressors suppress the high stearate phenotype in both leaves and seeds. The effects of one of the suppressors, TW2-1, is limited to the leaves. A second allele at the fab2 locus, fab2-2, was also identified and plants homozygous for this allele where intermediate in both plant size and 18 : 0 content between wild-type Arabidopsis and fab2-1 mutants. The alleles at fab2 and the suppressor mutations provided a total of nine genotypes which were analyzed to demonstrate a clear-cut relationship between leaf 18 : 0 content (0.7–19.6% of total leaf fatty acids) and reductions in plant size (24–4 mm). These results illustrate the utility of suppressor analysis for addressing problems in biochemistry and plant biology. They also indicate that the genetic control of plant lipid composition is more complex than previously appreciated. Received: 24 January 1997 / Accepted: 14 February 1997