Characterisation of Australian isolates of Actinobacillus capsulatus, Actinobacillus equuli, Pasteurella caballi and Bisgaard Taxa 9 and 11

Objective The objective of this work was to perform a comprehensive phenotypic characterisation of 16 isolates of bacteria previously identified as Actinobacillus equuli.
Design The 16 isolates that had been obtained from Australian animals – 15 from horses and one from a rabbit – were compared with reference strains of A equuli, A capsulatus, Pasteurella caballi and Bisgaard Taxa 9 and 11.
Results The characterisation study demonstrated that only nine of the isolates were A equuli . The other isolates were identified as A capsulatus (the isolate from rabbit), P caballi (one isolate), Bisgaard Taxon 11 (two isolates) and Bisgaard Taxon 9 (one isolate). The final two isolates could not be assigned to any recognised species or taxa.
Conclusion This study has highlighted the importance of a complete characterisation of Actinobacillus -like organisms isolated from horses and rabbits. The study represents the first time that A capsulatus, P caballi and Bisgaard Taxa 9 and 11 have been recognised as being present in Australia.     

Classification of Actinobacillus spp isolates from horses involved in mare reproductive loss syndrome

OBJECTIVE: To identify Actinobacillus spp isolates recovered from fetuses and pericardial fluid from horses affected with mare reproductive loss syndrome (MRLS) and determine whether these bacterial species are the same as those isolated from clinically normal horses. SAMPLE POPULATION: Isolates of actinobacilli recovered from 18 horses with pericarditis and 109 fetuses aborted by mares affected by MRLS. Procedures-Actinobacillus spp isolates were identified to the level of species or subspecies by use of conventional phenotypic tests and biochemical and enzyme test kits. The 16S rRNA gene from selected isolates was amplified, purified, and sequenced. Sequence data were compared with sequence data for actinobacilli in GenBank. RESULTS: Of the 109 isolates obtained from fetuses, 14 were Actinobacillus equuli subsp equuli, 65 were A equuli subsp haemolyticus, 28 were Bisgaard taxon 10-like bacterium, and 2 were Actinobacillus genomospecies 1. Of the 18 isolates from horses with pericarditis, 4 were A equuli subsp equuli, 13 were A equuli subsp haemolyticus, and 1 was Bisgaard taxon 10-like bacterium. Comparisons with published data and GenBank data revealed that the isolates recovered from horses with MRLS were the same as those isolated from the oral cavity or alimentary tract of healthy horses. CONCLUSIONS AND CLINICAL RELEVANCE: Actinobacillus spp isolates recovered from fetuses and pericardial fluid samples of horses affected by MRLS in 2001 to 2003 were identical to Actinobacillus spp found in the oral cavity and alimentary tracts of healthy horses.     

Characterisation of a novel Mannheimia sp from Australian feedlot cattle

OBJECTIVE: To characterise eight isolates of a Gram-negative organism obtained from the upper respiratory tract of cattle showing evidence of mild upper respiratory tract disease. DESIGN: The isolates were compared with the five recognised species within the genus Mannheimia - M haemolytica, M glucosida, M granulomatis, M ruminalis and M varigena--using a range of phenotypic and genotypic methods. RESULTS: Phenotypic characterisation indicated that the isolates belonged to the trehalose-negative [Pasteurella] haemolytica complex. This complex has recently been reorganised into five species within the new genus Mannheimia. Ribotyping performed using HindIII and a computerised analysis system indicated that the eight Australian isolates formed a distinct cluster that was related to, but different from, the five recognised species of Mannheimia. The 16S rRNA sequence of one isolate (BNO311) was determined and a phylogenetic analysis performed. Isolate BNO311 was distinct from the five named Mannheimia spp but did join a larger cluster consisting of rRNA cluster IV (M varigena) and the unnamed rRNA cluster V of Mannheimia. DNA:DNA hybridisation between isolate BNO311 and M haemolytica NCTC 9380T, M granulomatis P411 and Actinobacillus ligniersii NCTC 4189T all suggested similarities of approximately 30%. CONCLUSIONS: These phenotypic and genotypic characterisation studies suggest that the eight Australian isolates represent a new species of Mannheimia. Until further characterisation studies are performed, we are unwilling to propose a name for this taxon, preferring to refer to this possible new species as Bisgaard taxon 39 of cluster V of Mannheimia.     

Actinobacillus suis-like organisms and evidence of hemolytic strains of Actinobacillus lignieresii in horses

Thirty-seven local isolates of Actinobacillus suis-like organisms from diseased and clinically normal horses and 1 llama were compared with reference strains of A suis, A lignieresii, A equuli, A capsulatus, A hominis, A (Pasteurella) ureae, and equine A suis-like organisms (ASLO) previously described in literature. Comparison was by cultural characteristics, carbohydrate fermentation, enzyme profiles, and whole-cell protein polyacrylamide gel electrophoresis. Carbohydrate fermentation, determined by API-CH gallery, divided 36 equine ASLO isolates into 6 API-CH biotypes. The llama isolate was an additional distinct biotype. The biochemical comparisons between A suis and ASLO did not reveal remarkable and consistent differences. Enzyme analysis revealed 5 API-ZYM biotypes, one of which included the same strains as one of the API-CH biotypes and consisted in both instances of 4 esculin-negative ASLO cultures and the reference strain of A lignieresii. We conclude that the 4 strains were hemolytic variants of A lignieresii. Protein electrophoresis disclosed 15 banding patterns, 10 of which represented equine ASLO strains. The reference strains of A suis shared the pattern predominant among equine ASLO. Four of the remaining reference strains of Actinobacillus species each had a unique profile, whereas the type strain of A capsulatus and the llama isolate had similar profiles. The groupings of cultures resulting from the different testing methods had little relation to each other and to the anatomic source of the strains except the strains comprising API-CH biotype II, which originated in the equine respiratory tract, and the A lignieressi cluster.     

Isolation of pure Babesia equi and Babesia caballi organisms in splenectomized horses from endemic areas in South Africa

Both Babesia equi and Babesia caballi are endemic in large parts of South Africa. Attempts were made to obtain pure local isolates of both B. equi and B. caballi for the purpose of developing serological tests to study the epidemiology of equine babesiosis in this country. The indirect fluorescent antibody test was used to screen horses for B. equi and B. caballi in an endemic area. Seven horses and 3 donkeys between 3 and 36 months of age that tested negative were subsequently splenectomized. The splenectomy operation was performed through the abdominal approach. A 100% survival rate was achieved through this method, probably because it reduced the risk involved in the operation. Blood collected from naturally infected horses and passaged in fully susceptible splenectomized horses and a donkey, under laboratory conditions, produced 2 isolates of Babesia caballi and 1 of B. equi. Microscopical and serological examinations confirmed that these were pure isolates.     

Mesenteric lymphangitis and sepsis due to RTX toxin-producing Actinobacillus spp in 2 foals with hypothyroidism-dysmaturity syndrome

Actinobacillus suis-like organisms (ASLOs) have been isolated from the genital, respiratory, and digestive tracts of healthy adult horses, horses with respiratory disease, and septic foals. Two foals with congenital hypothyroidism-dysmaturity syndrome from separate farms developed ASLO infection. At necropsy, both had contracted carpal flexor tendons, thyroid hyperplasia, and thrombotic and necrotizing mesenteric lymphangitis and lymphadenitis; one foal also had mandibular prognathism. Numerous ASLOs were isolated from tissues from both foals, including intestine. Biochemical testing and mass spectrometric analysis of the two Actinobacillus isolates did not allow unequivocal identification. Comparative genetic analysis was done on these and similar isolates, including phylogeny based on 16S rRNA, rpoB and recN genes, as well as RTX (repeat in toxin) toxin typing of apxIA-apxIVA and aqxA genes. One isolate was identified as Actinobacillus suis sensu stricto, based on the presence of apxIA and apxIIA but not aqxA, whereas the other isolate had aqxA but neither apxIA nor apxIIA, consistent with A equuli ssp haemolyticus. Based on genotypic analysis of the isolates included for comparison, 3 of 3 equine ASLOs and 2 of 5 A equuli isolates were reclassified as A equuli subsp haemolyticus, emphasizing the importance of toxin genotyping in accurate classification of actinobacilli.     

Serological characterisation of Actinobacillus pleuropneumoniae isolated from pigs in 1993 to 1996

OBJECTIVES: To clarify the serological identity of the prototype strain of a group of Actinobacillus pleuropneumoniae isolates that could not be serotyped in previous studies and to establish the serovar of 378 isolates of A pleuropneumoniae obtained from pigs in Australia over the period 1993 to 1996. DESIGN: After initial validation, QGD and IHA tests were used to characterise the prototype isolate (HS143) selected to represent the cross-reacting isolates that were found in a previous study. Next, 378 recent field isolates of A pleuropneumoniae were characterised using the existing gel diffusion serotyping technique and/or the IHA or QGD tests. RESULTS: The indirect haemagglutination test was shown to be capable of correctly recognising the reference strain for all serovars except serovar 11. While the quantitative gel diffusion test was not as effective as indirect haemagglutination, it could recognise serovar 11. When the two tests were used to examine the prototype strain (HS143) of the cross-reactive isolates, the results indicated that HS143 is serologically distinct from all 12 of the recognised serovars of A pleuropneumoniae. However, as HS143 was subsequently identified as serovar 12 by one of the leading international reference laboratories, the antiserum to isolate HS143 was used as the serovar 12 antiserum. A total of 346 of the 378 A pleuropneumoniae field isolates examined could be confidently serotyped with almost 90% of the isolates being either serovar 1 (104 isolates); serovar 7 (83 isolates) or serovar 12 (142 isolates). A range of other serovars and some cross-reactive isolates made up the remainder of the isolates. CONCLUSION: The serovar 12 antiserum produced against the international reference strain (1096) does not recognise Australian serovar 12 isolates. The antiserum raised against isolate HS143 does recognise the Australian serovar 12 isolates. The dominant serovars of A pleuropneumoniae infecting Australian pigs are (in decreasing order) serovars 12, 1 and 7.     

Revised definition of Actinobacillus sensu stricto isolated from animals. A review with special emphasis on diagnosis

The taxonomy of the members of the genus Actinobacillus associated with animals has been reviewed with focus on classification and identification including molecular based characterization, typing and identification. Out of the 22 species or species like taxa reported as Actinobacillus, 19 are associated with animals. When classified on the basis of 16S rRNA sequence based phylogenetic analysis, DNA-DNA hybridizations and phenotypic analysis, Actinobacillus sensu stricto is restricted to include A. lignieresii, A. pleuropneumoniae, A. equuli subsp. equuli, A. equuli subsp. haemolyticus (taxon 11 of Bisgaard), A. hominis, A. suis, A. ureae, A. arthritidis (taxon 9 of Bisgaard), Actinobacillus genomospecies 1 and 2 and the taxa 8 and 26 of Bisgaard. The remaining 11 species of Actinobacillus are unrelated to A. sensu stricto and should consequently be grouped with other genera or be renamed as new genera depending on new data. Identification of members of Actinobacillus at species level is possible through phenotypic characterization combined with information on host of isolation. PCR tests are available for specific detection of A. pleuropneumoniae. Only A. pleuropneumoniae is presently considered as a primary pathogen. Based on different types of RTX genes it is possible to PCR type A. pleuropneumoniae to serotype level. PCR might also be used for the specific detection of A. equuli subsp. haemolyticus. Epidemiological investigations and surveillance have so far included serotyping, multilocus enzyme electrophoresis (MLEE), ribotyping and restriction fragment length profiling.     

Comparative molecular characterization of Corynebacterium pseudotuberculosis of different origin

Ribotyping and susceptibility to 17 antimicrobial agents were used to compare 37 isolates of Corynebacterium pseudotuberculosis (28 from horses, 1 from cattle, 3 from sheep and 5 from goats) derived from various types of lesions, and different geographic locations. According to the presence of nitrate reductase, all but one isolate from horses reduced nitrate (nitrate-positive), whereas all isolates from sheep and goats were unable to reduce nitrate (nitrate-negative). The ribotype of the nitrate-negative isolate from a horse with ulcerative lymphangitis was identical to all the other isolates from horses, and different than the ribotype of nitrate-negative isolates from sheep and goats. Ribotyping with one of the restriction endonucleases, Apa I, revealed differences between, but not within, the two biotypes. However, ribotyping with Pst I endonuclease revealed one variant within the equine biotype and one variant within the ovine biotype. The minimum inhibitory concentration (MIC; μg/ml) of antimicrobial agents against isolates from nitrate-negative and nitrate-positive groups was very similar, with the exception of isolates from sheep and goats which had a higher MIC for amikacin than isolates from horses and cattle.     

Effects of Actinobacillus equuli culture supernatants on equine neutrophil functions and survival

After exposure of equine granulocytes from both foals and adult horses to culture supernatants from clinical isolates of Actinobacillus equuli, phagocytic capacity and respiratory burst was examined by flow-cytometry and a chemiluminescence assay, respectively. One haemolytic isolate of an equine Actinobacillus was also included in the study. An average decrease of 22% in total number of granulocytes, in the flow cytometric assay (P < 0.01), and an average decrease of 26% in light emission, in the chemiluminescence assay (P < 0.001), was seen after exposure to bacterial culture supernatants of A. equuli, indicating that the supernatants contained leukotoxic bacterial products. Supernatants from the haemolytic isolate appeared to contain a higher amount or more potent leukotoxic metabolites when haemolysis was expressed, causing a decrease in total number of granulocytes of 44% (P < 0.01) and a decrease in light emission of 52% (P < 0.01). Evaluation of the stability of the methods used revealed that within-method variation was far less than the observed results. The leukotoxic effects of A. equuli culture supernatants were mainly reflected in the decreased survival of neutrophils and not in neutrophil functions.     

Differentiation of Australian isolates of Mycobacterium paratuberculosis using pulsed-field gel electrophoresis

Objectives To examine strain variation amongst Australian isolates of Mycobacterium paratuberculosis .
Design Pulsed field gel electrophoresis was optimised for differentiation of M paratuberculosis strains, and this typing technique was then applied to a collection of Australian isolates.
Procedure DNAs from 35 Australian isolates of M para-tuberculosis and a UK reference strain were digested with one or other of three restriction endonucleases. The banding patterns obtained after pulsed field gel electrophoresis of the DNA fragments were compared.
Results The Australian isolates were divided into two groups on the basis of their DNA banding pattern. Both were different from the UK reference strain. Seven isolates from cattle in Victoria and the Northern Territory had the same pattern as five isolates from alpacas in Victoria and Western Australia. Another 20 isolates from cattle in Victoria, Western Australia and the Northern Territory had the same pattern as isolates from two sheep and a goat in New South Wales.
Conclusion Pulsed field gel electrophoresis was a useful tool for strain typing of M paratuberculosis , and could be used to study the transmission of strains in Australia.     

Repeated high dose imidocarb dipropionate treatment did not eliminate Babesia caballi from naturally infected horses as determined by PCR-reverse line blot hybridization

Imidocarb treatment of horses infected with Babesia caballi is supposed to eliminate the infection, but data on the efficacy of this treatment is scarce. The study presented here concerns four Paso Fino horses, which were imported into the island of Curacao on the basis of a piroplasmosis negative complement fixation test (CFT). Upon re-testing with an indirect fluorescent antibody test immediately after arrival in Curacao, two horses appeared to have antibodies to B. caballi and all horses had antibodies to Theileria equi. Subsequent testing with polymerase chain reaction combined with a reverse line blot yielded positive results for both agents in all four horses. Treatment with five consecutive doses of imidocarb dipropionate (4.7 mg/kg BW im q 72 h), temporarily resulted in negative results, but B. caballi and T. equi were detected again in the samples taken at 6 and 18 weeks after completion of the treatment. These results confirm that the CFT is not a suitable test for pre-import testing and that even high dose treatment with imidocarb may not be capable of eliminating B. caballi and T. equi infections from healthy carriers.     

Genomic analyses of bovine viral diarrhea viruses isolated from cattle imported into Japan between 1991 and 2005

Thirty-one isolates of bovine viral diarrhea virus (BVDV) isolated within the past 15 years from imported cattle by the Japanese Animal Quarantine Service (AQS) were used in this study in which a 5'-untranslated region of each isolate was genetically analyzed. Twenty-six of the 31 isolates were classified as BVDV1 and the remainder as BVDV2. Phylogenetic analysis of the RT-PCR fragments amplified from the isolates showed the presence of viruses belonging to the BVDV1a, BVDV1b, BVDV1c, unclassified BVDV1 genotypes, and BVDV2. From the cattle of Australian origin, 16 of 17 isolates were classified as BVDV1c. This result was in agreement with a report showing that BVDV1c was a predominant subgenotype in Australia. From the cattle of North American origin, BVDV1 and BVDV2 species were both found. BVDV2 from the North American cattle was identified as the same cluster as the BVDV 890 strain, which is the prototype of BVDV2. These results suggest that the BVDVs isolated from exported cattle at the AQS reflect the predominant genotypes of BVDVs found in the exporting countries. The unclassified BVDV1 genotype of Chinese origin was in the same cluster as the ZM-95 strain, which was isolated from pigs in China. In this study, the genomic properties of 31 isolates of BVDV collected in the AQS were investigated. We concluded that isolates are genetically heterogeneous but geographically restricted. The information obtained from this report will be useful when carrying out epidemiological surveys of BVDV isolated in Japan.     

Actinobacillus suis infection of horses

Nineteen isolates of Actinobacillus suis were recovered from horses during the period October 1978-December 1980. Animals varied in age from a full term foetus to 12 years. One isolate was obtained from the nose of an apparently healthy horse, the remainder were obtained from still-born foetuses (2), foals dying within a week of birth (5), older animals with respiratory (6) or genital infections (3) or abscesses in the jaw (1). One isolate was obtained from the lung of a 2-week-old foal which had shown diarrhoea. The bacteriological characteristics of the isolates and the pathological lesions present in eight cases are described. The organism has a wide geographical distribution in New Zealand, and in the northern part of the North Island appears to be more common than A. equuli.     

Characterisation of avian adenoviruses associated with inclusion body hepatitis

Eleven avian adenoviruses were isolated in monolayer cultures of specific pathogen free chicken kidney cells which were inoculated with suspensions of liver, intestine or bursa obtained from 15 broiler flocks experiencing outbreaks of inclusion body hepatitis (10 isolates) and from five unaffected flocks (one isolate). Of the 11 isolates obtained, nine were identified by virus neutralisation tests as serotype 8, one as serotype 1 and one as serotype 12. Adeno-associated viruses were only observed in combination with adenoviral particles of the serotype 12 isolate which was derived from a relatively mild outbreak of inclusion body hepatitis. Only the serotype 1 isolate, obtained from the unaffected broiler flock, consistently caused the death of embryos with marked pathological changes. All of the isolates produced basophilic intranuclear inclusion bodies surrounded by clear halos in chicken kidney cell cultures. DNA preparations, obtained from six strains of serotype 8 avian adenovirus (two New Zealand isolates, three Australian isolates and the reference strain HVI) after digestion with the restriction enzymes EcoRI and BamHI, gave electrophoretic patterns showing the New Zealand isolates to be similar to one another and to strain HVI, but quite distinct from the Australian isolates.     

A survey for piroplasmids in horses and Bactrian camels in North-Eastern Mongolia

Equine piroplasmosis caused by Babesia caballi and Theileria equi is widespread in Asia. The presence of these haemozoans in Mongolia was previously confirmed in domestic as well as in reintroduced Przewalski horses in which they cause significant pathology. The data on occurrence of piroplasms from Bactrian camels in Asia is lacking. A total of 192 horses, 70 Bactrian camels, and additional 16 shepherd dogs from the Hentiy province were included in our study. No clinical signs typical for piroplasmid infection were observed during the field survey. Microscopic examination revealed the presence of T. equi in blood smears from 67% of examined horses, with camels and dogs being negative. A two step PCR approach was used to detect piroplasms in peripheral blood. In the first "catch all" PCR reaction, amplification of the 496 bp-long fragment of the SSU rRNA gene enabled the detection of Babesia and Theileria spp. Second round multiplex PCR reaction used for species discrimination allowed the amplification of T. equi- and B. caballi-specific 340 bp and 650 bp-long regions of the SSU rRNA, respectively. This assay detected T. equi in 92.7% of horses, while the infections with B. caballi and dual infections were rare. In both PCR setups, camels and dogs were negative indicating that in the studied region, these hosts do not share piroplasms with horses.     

Peritonitis associated with Actinobacillus equuli in horses: 51 cases

OBJECTIVE: To review the clinical findings, diagnosis and treatment of 51 horses with peritonitis attributed to Actinobacillus equuli. DESIGN: Retrospective study of clinical cases. METHODS: Breed, age and gender of horse, history, physical examination findings, treatment and outcome were determined from the hospital records of 51 horses in which a diagnosis of peritonitis attributed to A. equuli was made between January 1993 and June 1999. Results of abdominal fluid cytology and bacteriology, antimicrobial sensitivity patterns, haematology and faecal egg counts, when performed, were also retrieved. RESULTS: There was a variety of breeds of horses affected. There were 35 male and 17 female horses, aged from 9 months to 22 years, presented. Lethargy, signs of depression with mild to moderate signs of abdominal pain and inappetence were the most common reasons for presentation. Most horses had elevated heart and respiratory rates, an elevated rectal temperature and reduced intestinal borborygmi heard on auscultation of the abdomen. Abnormal colour with an elevated protein were features of an abdominal fluid sample in 98% of horses and a marked elevation in nucleated cell count was present in all samples. Pleomorphic gram-negative rods were seen on cytology in 53% of samples and a positive culture of A. equuli was returned in 72% of samples. Other laboratory findings in some horses included mild haemoconcentration, hypoproteinaemia, an elevated circulating nucleated cell count with a left shift, an elevation in fibrinogen concentration and an elevated faecal egg count. All horses demonstrated a rapid response to treatment with procaine penicillin alone, or a combination of procaine penicillin and gentamicin sulphate. Where antimicrobial sensitivity tests were performed, all but two isolates were sensitive to procaine penicillin. All horses responded to antimicrobial and supportive therapy and were discharged from hospital. CONCLUSION: Horses with A. equuli peritonitis present with similar clinical signs as horses with other causes of abdominal pain. However, these signs, when evaluated in conjunction with the results of abdominal fluid analysis and response to treatment, are characteristic of A. equuli peritonitis. Pleomorphic gram-negative bacteria may be seen on a cytological preparation of the abdominal fluid sample, and a positive bacterial culture may be obtained in some, but not all, cases. Most isolates are sensitive to procaine penicillin, so treatment with procaine penicillin and gentamicin sulphate is recommended until antimicrobial sensitivity is known.     

A comparative study on the prevalence of Theileria equi and Babesia caballi infections in horse sub-populations in Turkey

Blood and serum samples were taken from 481 horses, from a stud farm or a racecourse, and tested by microscopic examination of blood smears and cELISA for Theileria equi (T. equi) and Babesia caballi (B. caballi) infections. At the time of sampling, animals were also examined for tick infestations and clinical disease, which were not observed in any of the sampled horses. During the microscopic examination of thin blood smears, parasites were detected in the three horses from the racecourse. Overall seroprevalence of infection was detected as 18.50% (89 of 481 horses) by cELISA, with T. equi being significantly more prevalent than B. caballi. Of the 481 blood samples, 78 (16.21%) were serologically positive for T. equi and 4 (0.83%) were serologically positive for B. caballi. In addition, 7 (1.46%) samples were positive for both T. equi and B. caballi antibodies. Seropositivity rates in the racecourse horses were higher than those determined in the stud farm horses. The rates for T. equi, B. caballi and both species were 13.39, 0.52 and 0% in the horses from the stud farm and 27, 2 and 7% in the racecourse horses, respectively. These results indicate that equine piroplasmosis is more common in racehorses than studhorses and therefore it might be a serious concern in horses that participate to international races.     

Diagnosis of swine dysentery and spirochaetal diarrhea. 2. Effort of macrorestriction analysis for differentiation of intestinal Serpulina relative to determination by culture-biochemical markers following species classification]

Genotypic differentiation by means of macrorestriction fragment profile analysis using Mlul restriction enzyme was carried out differentiating 41 Serpulina field strains from swine (38), dog (2) and a rat as well as ten type and reference strains into 40 electrophoretic types. A dendrogram was created using the average linkage between groups method. At a level of 50% similarity the patterns could be divided into six groups that roughly corresponded to the results yielded by cultural and biochemical methods formerly (FELTRUP et al. 1999). Five of these clusters corresponded to the five known porcine Serpulina species, one cluster contained the S. pilosicoli isolates from dog and rat included in this study. Interestingly all nine investigated indole negative, strongly haemolytic isolates were clustered together in one group with the S. hyodysenteriae strains, so that incidence of indole negative variants of S. hyodysenteriae was confirmed. Because of being grouped together with two S. intermedia isolates, the suitability of B 256 as S. innocens type strain is--in accord to investigations carried out by PETTERSSON et al (1996)--called in question.