Population dynamics ofAzospirillum brasilense and its bacteriophage in soil

Summary This study examined the ecology and interaction ofAzospirillum brasilense and its bacteriophage in soil. Four Chernozemic soils from Canada, a Latosol and three Podzolic soils from Brazil were assayed for phage. Only the Latosol containedA. brasilense phage. None of the soils contained phage for otherA. brasilense orA. lipoferum strains tested. Recovery of phage from soil depended on the growth of indigenous or added host cells. A phage isolated from the Latosol had a hexagonal head of 100 nm and a tail of 200 nm. This phage was morphologically distinct from previously described Azospirillum phage and its host range was limited toA. brasilense strains 29145 and 29711.Survival and recovery of phage added to phage-free soil was dependent on the phage, the initial phage population, the presence of host cells and nutrients, and the soil. Phage persisted in soils at undetectable levels for at least seven weeks, but were still able to interact with multiplying host cells and exhibit a 1000-fold increase in number. Phage required a host cell population of at least 100–1000 per g of soil in order to multiply. The phage burst detected under these conditions increased as the cell to phage ratio increased. Long term incubation studies showed that the activity of phage in soil closely followed the activity of host cells and thus both were manipulated by appropriate amendments to soil.     

Ensifer adhaerens Predatory Activity Against Other Bacteria in Soil, as Monitored by Indirect Phage Analysis

An indirect phage analysis procedure was used to detect and follow the activity of the bacterial predator Ensifer adhaerens in situ in natural soil. The soil was percolated with an aqueous suspension of washed bacterial host cells so that the E. adhaerens cells naturally present in the soil would multiply in response to the host cells. The natural phage development which ensued against these multiplying E. adhaerens cells in the soil was then monitored by noting plaques which developed when the percolation fluid was plated with laboratory strains of E. adhaerens on laboratory media. The activities of the other members of the predation system that includes E. adhaerens (Streptomyces sp. strain 34 and a myxobacter) could not be monitored directly by phage analysis because phage were not found for them. Indirect monitoring was possible, however, because they were susceptible to attack by E. adhaerens. In general, the results were in agreement with previous observations by other methods of the predation sequence. E. adhaerens attacked Micrococcus luteus, Streptomyces sp. strain 34, and the myxobacter but did not attack several other possible species of hosts. It also did not respond to percolation of the soil with various nutrient solutions. E. adhaerens phage activity was not present in half of the soils percolated with M. luteus cells. This seemed to reflect too great a phage-host specificity for the technique as regards these soils, because E. adhaerens-like bacteria other than the strains used for plaquing were present in at least some of these soils. Although E. adhaerens did not attack Escherichia coli or Pseudomonas aeruginosa in soil, there was an overproduction of E. adhaerens phage if these bacteria were percolated simultaneously with M. luteus cells. The possibility is discussed that this represents an activation by M. luteus (or by a heat-extractable factor from it) of other bacterial predators that attack E. coli or P. aeruginosa and that these predators subsequently are themselves attacked by E. adhaerens.     

Ecology of Indigenous Soil Rhizobia: Response of Bradyrhizobium japonicum to Readily Available Substrates

Populations of indigenous Bradyrhizobium japonicum serocluster 123 and serogroups 110 and 138 were studied after various sugars were added to their soil habitat. Loam soil with approximately 10⁴ cells of each group per g of soil were amended every 3 days with 0.1% glucose, sucrose, arabinose, xylose, or galactose. Enumerations of the populations were made every 12 days by immunofluorescence assay. Each B. japonicum population in the sugar-treated soils increased by about 1 log during the first 12 days, to a maximum of about 10⁶ cells by day 36 or 48, irrespective of the sugar added. Maximum growth rates were similar for each group and occurred during the 12-day incubation period. The most rapid growth was in response to arabinose, with a mean generation time of about 3.0 days. Other mean doubling times were 4.0 days with glucose and galactose treatments, 4.5 days with xylose treatment, and 5.4 days with sucrose amendment. These data provide the first direct evidence that indigenous soil rhizobia can compete successfully with other soil bacteria for readily available substrates in soil in the absence of host legume roots or other rhizospheres. The growth rates in soil of the specific B. japonicum populations studied were nearly the same with a given sugar treatment but varied considerably with different sugars. The mean generation times of 3 to 5 days are among the first reported growth rates for heterotrophic bacteria in natural soil.     

Rapid Detection of Escherichia coli O157:H7 by Using Green Fluorescent Protein-Labeled PP01 Bacteriophage

A previously isolated T-even-type PP01 bacteriophage was used to detect its host cell, Escherichia coli O157:H7. The phage small outer capsid (SOC) protein was used as a platform to present a marker protein, green fluorescent protein (GFP), on the phage capsid. The DNA fragment around soc was amplified by PCR and sequenced. The gene alignment of soc and its upstream region was g56-soc.2-soc.1-soc, which is the same as that for T2 phage. GFP was introduced into the C- and N-terminal regions of SOC to produce recombinant phages PP01-GFP/SOC and PP01-SOC/GFP, respectively. Fusion of GFP to SOC did not change the host range of PP01. On the contrary, the binding affinity of the recombinant phages to the host cell increased. However, the stability of the recombinant phages in alkaline solution decreased. Adsorption of the GFP-labeled PP01 phages to the E. coli cell surface enabled visualization of cells under a fluorescence microscope. GFP-labeled PP01 phage was not only adsorbed on culturable E. coli cells but also on viable but nonculturable or pasteurized cells. The coexistence of insensitive E. coli K-12 (W3110) cells did not influence the specificity and affinity of GFP-labeled PP01 adsorption on E. coli O157:H7. After a 10-min incubation with GFP-labeled PP01 phage at a multiplicity of infection of 1,000 at 4°C, E. coli O157:H7 cells could be visualized by fluorescence microscopy. The GFP-labeled PP01 phage could be a rapid and sensitive tool for E. coli O157:H7 detection.     

Protozoan Response to the Addition of Bacterial Predators and Other Bacteria to Soil

Representatives of several categories of bacteria were added to soil to determine which of them might elicit responses from the soil protozoa. The various categories were nonobligate bacterial predators of bacteria, prey bacteria for these predators, indigenous bacteria that are normally present in high numbers in soil, and non-native bacteria that often find their way in large numbers into soil. The soil was incubated and the responses of the indigenous protozoa were determined by most-probable-number estimations of total numbers of protozoa. Although each soil was incubated with only one species of added bacteria, the protozoan response for the soil was evaluated by using most-probable-number estimations of several species of bacteria. The protozoa did not respond to incubation of the soil with either Cupriavidus necator, a potent bacterial predator, or one of its prey species, Micrococcus luteus. C. necator also had no effect on the protozoa. Therefore, in this case, bacterial and protozoan predators did not interact, except for possible competition for bacterial prey cells. The soil protozoa did not respond to the addition of Arthrobacter globiformis or Bacillus thuringiensis. Therefore, the autochthonous state of Arthrobacter species in soil and the survival of B. thuringiensis were possibly enhanced by the resistance of these species to protozoa. The addition of Bacillus mycoides and Escherichia coli cells caused specific responses by soil protozoa. The protozoa that responded to E. coli did not respond to B. mycoides or any other bacteria, and vice versa. Therefore, addition to soil of a nonsoil bacterium, such as E. coli, did not cause a general increase in numbers of protozoa or in protozoan control of the activities of other bacteria in the soil.     

Effect of the fuel system icing inhibitor diethylene glycol monomethyl ether on the biodegradation of jet fuel in soil

Microbial degradation of jet fuel leads to the accumulation of sludge in fuel distribution systems and storage tanks. To prevent this phenomenon, the biocidal anti-icing inhibitor diethylene glycol monomethyl ether (DiEGME) is routinely added to the fuel. The fate of DiEGME in soil and its consequent effect on the biodegradation of jet fuel by indigenous soil microflora have not been investigated. The aim of this work was to study the kinetics of biodegradation of jet fuel in dark rendzina soil, as affected by the presence of DiEGME. Our data show that the degradability in soil of jet fuel amended with DiEGME was tenfold higher than that of non-amended fuel. Consequently, there was an increase in the jet-fuel-utilizing soil microbial populations during the 100 days of incubation of soil samples amended with jet fuel containing DiEGME. Gas chromatograms of distilled fractions of jet fuel extracted from the soil demonstrated that most of the light fractions' extracts could not be detected at the end of the 100-day incubation. The relative concentration of aromatic compounds in the soil contaminated with DiEGME-amended jet fuel increased during incubation, demonstrating the lower biodegradation rate of these components compared with other fuel components. DiEGME was partially degraded by the general microbial population of the soil. Maximal DiEGME degradation was obtained with specific jet-fuel-utilizing microbial strains – Pseudomonas aeruginosa and Cladosporium resinae – that were added to a carbon-free mineral medium. The degradation rate of DiEGME by specific strains or by soil mixed populations bore an inverse relationship to the DiEGME concentration. The finding that DiEGME can be degraded by indigenous soil microorganisms may have facilitated its utilization also by jet-fuel-degrading microorganisms.     

Fate of Escherichia coli O157:H7 in Manure-Amended Soil

Escherichia coli O157:H7 cells survived for up to 77, >226, and 231 days in manure-amended autoclaved soil held at 5, 15, and 21°C, respectively. Pathogen populations declined more rapidly in manure-amended unautoclaved soil under the same conditions, likely due to antagonistic interactions with indigenous soil microorganisms. E. coli O157:H7 cells were inactivated more rapidly in both autoclaved and unautoclaved soils amended with manure at a ratio of 1 part manure to 10 parts soil at 15 and 21°C than in soil samples containing dilute amounts of manure. The manure-to-soil ratio, soil temperature, and indigenous microorganisms of the soil appear to be contributory factors to the pathogen's survival in manure-amended soil.     

Development and Use of Field Application Vectors To Express Nonadaptive Foreign Genes in Competitive Environments

Many potential applications of genetically engineered microorganisms in environmental and agricultural biotechnology involve introducing genetic capabilities into nonsterile competitive environments in which they provide no advantage to the host. Field application vectors have been designed for the purpose of creating a temporary niche for the host in such environments. This technique involves the addition to the target environment of a selective substrate readily utilizable by the host microorganism but unavailable to most indigenous species. Thirteen nonionic and anionic detergents, representing a wide range of structural complexities and molecular weights, were screened as potential selective substrates. Competition experiments in soil, using Warburg respirometry, indicated that isolates from six different detergent enrichment cultures were more active on their corresponding detergents than the indigenous microorganisms. Detergents of intermediate structural complexities and molecular weights were most effective for use as selective substrates. A field application vector that utilizes 1.0% Igepal CO-720 (detergent) as the selective substrate and Pseudomonas paucimobilis 1IGP4 as the host was tested for its ability to increase the presence of nonadaptive tetracycline resistance marker genes in soil. In soil amended with the selective substrate, strain 1IGP4 plate counts increased by three orders of magnitude and tetracycline-resistant transformant (pRK293) counts increased from 1.8 × 10⁶/g of soil to 4.3 × 10⁸/g in 2 days. Inoculation in the absence of substrate amendment or amendment with a nonselective substrate did not result in growth of strain 1IGP4. These results demonstrate the effectiveness of field application vectors for increasing the concentration of nonadaptive genes in competitive environments.     

Interactions between a genetically markedPseudomonas fluorescens strain and bacteriophage ΦR2f in soil: Effects of nutrients, alginate encapsulation, and the wheat rhizosphere

The introduction of bacteriophages could potentially be used as a control method to limit the population size of engineered bacteria that have been introduced into soil. Hence, the ability of a species-specific phage, R2f, to infect and lyse its host, a Pseudomonas fluorescens R2f transposon Tn5 derivative, in soil, was studied. Control experiments in liquid media revealed that productive lysis of host cells by phage R2f occurred when cells were freely suspended, whereas cells present in alginate beads resisted lysis. The presence of nutrients enhanced the degree of lysis as well as the production of phage progeny, both with the suspended cells and with cells escaped from the alginate beads. Experiments in which host cells and phage R2f were introduced into two soils of different texture revealed that host cells were primarily lysed in the presence of added nutrients, and phage reached highest titres in these nutrient-amended soils. Encapsulation of the host cells in alginate beads inhibited lysis by the phage in soil. Populations of free host cells introduced into soil that colonized the rhizosphere of wheat were not substantially lysed by phage R2f. However, P. fluorescens R2f populations colonizing the rhizosphere after introduction in alginate beads were reduced in size by a factor of 1,000. Cells migrating from the alginate beads towards the roots may have been in a state of enhanced metabolic activity, allowing for phage R2f infection and cell lysis. Correspondence to: J.D. van Elsas     

Isolation of Arthrobacter Bacteriophage from Soil

Soil was percolated with water and various nutrient solutions, and then the percolates were analyzed for bacteriophages which produced plaques on various Arthrobacter strains. The water percolates did not contain detectable phage. In contrast, phages for A. globiformis strains ATCC 8010 and 4336, and for several recent Arthrobacter species soil isolates, were easily detected in nutrient broth, soil extract, and cation-complete medium percolates. These percolates did not contain phage that produced plaques on A. oxydans and a recent Arthrobacter species soil isolate. Percolation with a selective nicotine-salts solution was required for demonstrating phage for these bacteria. None of the percolates contained phage for five additional named Arthrobacter species. In addition, phages were not detected for A. crystallopoietes in a 2-hydroxypyridine percolate of soil. Based on their lytic spectra, the phage isolates from this soil were relatively host specific.     

Effect of Inoculation and Leaf Litter Amendment on Establishment of Nodule-Forming Frankia Populations in Soil

High-N₂-fixing activities of Frankia populations in root nodules on Alnus glutinosa improve growth performance of the host plant. Therefore, the establishment of active, nodule-forming populations of Frankia in soil is desirable. In this study, we inoculated Frankia strains of Alnus host infection groups I, IIIa, and IV into soil already harboring indigenous populations of infection groups (IIIa, IIIb, and IV). Then we amended parts of the inoculated soil with leaf litter of A. glutinosa and kept these parts of soil without host plants for several weeks until they were spiked with [¹⁵N]NO₃ and planted with seedlings of A. glutinosa. After 4 months of growth, we analyzed plants for growth performance, nodule formation, specific Frankia populations in root nodules, and N₂ fixation rates. The results revealed that introduced Frankia strains incubated in soil for several weeks in the absence of plants remained infective and competitive for nodulation with the indigenous Frankia populations of the soil. Inoculation into and incubation in soil without host plants generally supported subsequent plant growth performance and increased the percentage of nitrogen acquired by the host plants through N₂ fixation from 33% on noninoculated, nonamended soils to 78% on inoculated, amended soils. Introduced Frankia strains representing Alnus host infection groups IIIa and IV competed with indigenous Frankia populations, whereas frankiae of group I were not found in any nodules. When grown in noninoculated, nonamended soil, A. glutinosa plants harbored Frankia populations of only group IIIa in root nodules. This group was reduced to 32% ± 23% (standard deviation) of the Frankia nodule populations when plants were grown in inoculated, nonamended soil. Under these conditions, the introduced Frankia strain of group IV was established in 51% ± 20% of the nodules. Leaf litter amendment during the initial incubation in soil without plants promoted nodulation by frankiae of group IV in both inoculated and noninoculated treatments. Grown in inoculated, amended soils, plants had significantly lower numbers of nodules infected by group IIIa (8% ± 6%) than by group IV (81% ± 11%). On plants grown in noninoculated, amended soil, the original Frankia root nodule population represented by group IIIa of the noninoculated, nonamended soil was entirely exchanged by a Frankia population belonging to group IV. The quantification of N₂ fixation rates by ¹⁵N dilution revealed that both the indigenous and the inoculated Frankia populations of group IV had a higher specific N₂-fixing capacity than populations belonging to group IIIa under the conditions applied. These results show that through inoculation or leaf litter amendment, Frankia populations with high specific N₂-fixing capacities can be established in soils. These populations remain infective on their host plants, successfully compete for nodule formation with other indigenous or inoculated Frankia populations, and thereby increase plant growth performance.     

Effect of Prophage W on the Propagation of Bacteriophages T2 and T4

Studies have been undertaken to determine whether the temperate phage ω present in Escherichia coli strain W is responsible for the inability of this strain to act as a host for T2 and T4. E. coli WS, cured of phage ω, was sensitive to T2 and T4. Lysogenation of E. coli C and WS with phage ω resulted in loss of ability to plate T2 and T4. However, E. coli K-12 lysogens still served as hosts for the T -even phage. Two of three WS lysogens studied resembled strain W at the biochemical level. They converted about 30% of infecting T2 deoxyribonucleic acid (DNA) to acid-soluble fragments and limited macromolecular synthesis to a few minutes after infection. The third lysogen did not degrade phage DNA, and nucleic acid and protein synthesis continued for some time, although no phage production occurred. It is concluded that phage ω plays a role in the restriction of virulent phage but that it is not the only factor involved. Since acid solubilization was not observed in all cases of phage ω-mediated restriction of T -even phage, a hypothesis for the restriction has been proposed which is based on an alteration in the cell envelope after lysogenation with phage ω.     

Bacterial Predators of Micrococcus luteus in Soil

Micrococcus luteus cells died relatively rapidly when they were added to natural soil. Microscopic observation showed that the cells were being physically destroyed by bacterial predators in the soil. Two of these predators were responsible for the initial, main attack, and they were isolated. The isolates on laboratory media lysed M. luteus cells in a manner similar to the attacks that occurred in soil. Neither predator was obligate, however, nor were they nutritionally fastidious. One of these bacteria produced mycelium and conidia. Under nutritionally poor conditions it used slender filaments of mycelium to seek out host cells. It had at least some of the characteristics of a Streptoverticillium species. The other bacterium was a short, gram-negative rod that did not easily fit into any of the known groups of gram-negative bacteria. It attached to host cells, but its mechanism of lysing these cells is not known.     

Effect of spermidine on the development of bacteriophage SP6

Bacteriophage SP6 is a virulent phage of Salmonella typhimurium which behaves differently than other phages of the same host. The effect of spermidine on SP6 infection of S. typhimurium has been found to depend on the time of addition of spermidine with respect to the time of addition of the phage and also on the composition of the growth medium. If spermidine was added prior to or within a short time after infection, the cells survived. Under this condition the invading DNA appeared to remain trapped in the cell membrane, and there was no expression of the phage genome. If spermidine was added after the initiation of the infection process, the replication of the phage was inhibited but the cells did not survive. Furthermore, if spermidine was added after DNA synthesis was over, there was no effect of spermidine on phage multiplication. Spermidine was found to affect phage DNA synthesis but not host DNA synthesis.     

Survival of Rhizobium in Acid Soils

A Rhizobium strain nodulating cowpeas did not decline in abundance after it was added to sterile soils at pH 6.9 and 4.4, and the numbers fell slowly in nonsterile soils at pH 5.5 and 4.1. A strain of R. phaseoli grew when added to sterile soils at pH 6.7 and 6.9; it maintained large, stable populations in soils of pH 4.4, 5.5, and 6.0, but the numbers fell markedly and then reached a stable population size in sterile soils at pH 4.3 and 4.4. The abundance of R. phaseoli added to nonsterile soils with pH values of 4.3 to 6.7 decreased similarly with time regardless of soil acidity, and the final numbers were less than in the comparable sterile soils. The minimum pH values for the growth of strains of R. meliloti in liquid media ranged from 5.3 to 5.9. Two R. meliloti strains, which differed in acid tolerance for growth in culture, did not differ in numbers or decline when added to sterile soils at pH 4.8, 5.2, and 6.3. The population size of these two strains was reduced after they were introduced into nonsterile soils at pH 4.8, 5.4, and 6.4, and the number of survivors was related to the soil pH. The R. meliloti strain that was more acid sensitive in culture declined more readily in sterile soil at pH 4.6 than did the less sensitive strain, and only the former strain was eliminated from nonsterile soil at pH 4.8; however, the less sensitive strain also survived better in limed soil. The cell density of the two R. meliloti strains was increased in pH 6.4 soil in the presence of growing alfalfa. The decline and elimination of the tolerant, but not the sensitive, strain was delayed in soil at pH 4.6 by roots of growing alfalfa.     

Rapid detection of Escherichia coli O157:H7 by using green fluorescent protein-labeled PP01 bacteriophage

A previously isolated T-even-type PP01 bacteriophage was used to detect its host cell, Escherichia coli O157:H7. The phage small outer capsid (SOC) protein was used as a platform to present a marker protein, green fluorescent protein (GFP), on the phage capsid. The DNA fragment around soc was amplified by PCR and sequenced. The gene alignment of soc and its upstream region was g56-soc.2-soc.1-soc, which is the same as that for T2 phage. GFP was introduced into the C- and N-terminal regions of SOC to produce recombinant phages PP01-GFP/SOC and PP01-SOC/GFP, respectively. Fusion of GFP to SOC did not change the host range of PP01. On the contrary, the binding affinity of the recombinant phages to the host cell increased. However, the stability of the recombinant phages in alkaline solution decreased. Adsorption of the GFP-labeled PP01 phages to the E. coli cell surface enabled visualization of cells under a fluorescence microscope. GFP-labeled PP01 phage was not only adsorbed on culturable E. coli cells but also on viable but nonculturable or pasteurized cells. The coexistence of insensitive E. coli K-12 (W3110) cells did not influence the specificity and affinity of GFP-labeled PP01 adsorption on E. coli O157:H7. After a 10-min incubation with GFP-labeled PP01 phage at a multiplicity of infection of 1,000 at 4 degrees C, E. coli O157:H7 cells could be visualized by fluorescence microscopy. The GFP-labeled PP01 phage could be a rapid and sensitive tool for E. coli O157:H7 detection.     

Behavior of a temperate bacteriophage in differentiating cells of Bacillus subtilis.

During the first 6 hr of sporulation, infection of Bacillus subtilis by by phi105 wild type or the clear-plaque mutant phi105 c30 was nonproductive, but phage DNA was trapped inside developing spores. After infection with either wild-type or mutant phage at early times of sporulation (T1-T3), phage DNA entered the developing spores in a heat-stable form, which may represent integration of the phage DNA into the host chromosome. Phage DNA in carrier spores produced by infection at later times (T4-T6) was much more heat sensitive. Spore preparations containing either phi105 wild type or phi105 c30 carrier spores gave rise to a spontaneous burst of phage during outgrowth, although the fraction of carried wild-type phage that chose lysis over lysogeny at germination has not been determined. Heat induction of the thermoinducible lysogen 3610 (phi105 cts23) was also abortive during sporulation. Furthermore, induction neither prevented eventual spore formation nor resulted in the conversion of prophage DNA to the carrier state; during outgrowth, the previously induced lysogenic spores remained stable lysogens. However, if the sporulating lysogenic cells were plated immediately after induction, they did not form colonies at high efficiency, as though transfer to fresh medium allowed sufficient phage expression to kill the host.     

Effect of spermidine on bacteriophage P22 infection.

The effect of spermidine on phage P22 infection of Salmonella typhimurium has been found to depend on the time of addition of spermidine with respect to the time of addition of the phage and also on the composition of the growth medium. If spermidine was added prior to or within a short time after infection, the cells survived. Under this condition the invading DNA appeared to remain trapped in the cell membrane, and there was no expression of the phage genome. If spermidine was added after the initiation of the infective process, the replication of the phage was inhibited but the cells did not survive. If spermidine was added after DNA synthesis was over, there was no effect of spermidine on phage multiplication. Spermidine was found to affect phage DNA synthesis but not host DNA synthesis.     

Effects of Triton X-100 and Quillaya Saponin on the ex situ bioremediation of a chronically polychlorobiphenyl-contaminated soil

The possibility of enhancing the ex situ bioremediation of a chronically polychlorinated biphenyl (PCB)-contaminated soil by using Triton X-100 or Quillaya Saponin, a synthetic and a biogenic surfactant, respectively, was studied. The soil, which contained about 350 mg/kg of PCBs and indigenous aerobic bacteria capable of growing on biphenyl or on monochlorobenzoic acids, was amended with inorganic nutrients and biphenyl, saturated with water and treated in aerobic batch slurry- and fixed-phase reactors. Triton X-100 and Quillaya Saponin were added to the reactors at a final concentration of 10 g/l at the 42nd day of treatment, and at the 43rd and 100th day, respectively. Triton X-100 was not metabolised by the soil microflora and it exerted inhibitory effects on the indigenous bacteria. Quillaya Saponin, on the contrary, was readily metabolised by the soil microflora. Under slurry-phase conditions, Triton X-100 negatively influenced the soil bioremediation process by affecting the availability of the chlorobenzoic acid degrading indigenous bacteria, whereas Quillaya Saponin slightly enhanced the biological degradation and dechlorination of the soil PCBs. In the fixed-phase reactors, where both the surfactant availability and the mixing of the soil were lower, Triton X-100 did not exert inhibitory effects on the soil biomass and enhanced significantly the soil PCB depletion, whereas Quillaya Saponin did not influence the bioremediation process. Received: 28 April 1998 / Received last revision: 15 July 1998 / Accepted: 29 July 1998