Samples of DNA from a panel of Giardia isolated from humans and animals in Europe and shown previously to consist of 2 major genotypes–‘Polish’ and ‘Belgian’–have been compared with human-derived Australian isolates chosen to represent distinct genotypes (genetic groups I–IV) defined previously by allozymic analysis. Homologous 0·52 kilobase (kb) segments of 2 trophozoite surface protein genes (tsa417 and tsp11, both present in isolates belonging to genetic groups I and II) and a 1·2 kb segment of the glutamate dehydrogenase (gdh) gene were amplified by the polymerase chain reaction (PCR) and examined for restriction fragment length polymorphisms (RFLPs). Of 21 ‘Polish’ isolates that were tested, all yielded tsa417-like and tsp11-like PCR products that are characteristic of genetic groups I or II (15 and 6 isolates respectively) in a distinct assemblage of G. intestinalis from Australia (Assemblage A). Conversely, most of the 19 ‘Belgian’ isolates resembled a second assemblage of genotypes defined in Australia (Assemblage B) which contains genetic groups III and IV. RFLP analysis of gdh amplification products showed also that ‘Polish’ isolates-were equivalent to Australian Assemblage A isolates (this analysis does not distinguish between genetic groups I and II) and that ‘Belgian’ isolates were equivalent to Australian Assemblage B isolates. Comparison of nucleotide sequences determined for a 690 base-pair portion of the gdh PCR products revealed ≥ 99·0% identity between group I and group II (Assemblage A/‘Polish’) genotypes, 88·3–89·7% identity between Assemblage A and Assemblage B genotypes, and ≥ 98·4% identity between various Assemblage B/‘Belgian’ genotypes. The results confirm that the G. duodenalis isolates examined in this study (inclusive of G. intestinalis from humans) can be divided into 2 major genetic clusters: Assemblage A (= ‘Polish’ genotype) containing allozymically defined groups I and II, and Assemblage B (= ‘Belgian’ genotype) containing allozymically defined groups III and IV and other related genotypes.

Infection of suckling mice with Giardia trophozoites recovered from the intestines of 11 dogs autopsied in Central and Southern Australia in each case produced an established isolate. In contrast, only 1 isolate was obtained by inoculation of faecal cysts. The organisms grew poorly in comparison with isolates from humans or non-canine animal hosts. Light microscopy revealed that the trophozoites had median bodies with the ‘claw hammer’ appearance typical of G. intestinalis (syn. G. duodenalis, G. lamblia) but that they differed in shape and nuclear morphology from axenic isolates of human or canine origin. Allozymic analysis of electrophoretic data representing 26 loci and phylogenetic analysis of nucleotide sequences obtained from DNA amplified from the glutamate dehydrogenase locus showed that the 11 isolates examined from Australian dogs were genetically distinct from all isolates of G. intestinalis that have been established previously from humans and animals, and also from G. muris. Both analytical methods placed 10 of the Australian canine isolates into a unique genetic lineage (designated Assemblage C) and the eleventh into a deep-rooted second branch (designated Assemblage D), each well separated from the 2 lineages (Assemblages A and B) of G. intestinalis that encompass all the genotypes known to infect humans. In contrast, 4 axenic isolates derived from dogs in Canada and Europe (the only other isolates to have been established from dogs) have genotypes characteristic of genetic Assemblages A or B. The findings indicate that the novel Giardia identified in these rural Australian dogs have a restricted host range, possibly confined to canine species. The poor success rate in establishing Giardia from dogs in vitro suggests, further, that similar genotypes may predominate as canine parasites world-wide. The absence of such organisms among isolates of Giardia that have been established from humans by propagation in suckling mice indicates that they are unlikely to infect humans. However, infection of humans by those dog-derived genotypes that grow in vitro cannot be excluded.