Journal of the Royal Society of Western Australia, 90(3), September 2007 
plan prepared for the then Rothiest Island Board (Iveson 
(& Bradshaw 1978). 
Over the subsequent period 1972-2000, a total of 
31 laboratory-confirmed cases of human salmonellosis 
were diagnosed in residents and visitors to the island 
involving 9 Salmonella serotypes. These comprised 5. 
javiana 16, S. typhimurlum 6, S. muenchen 5, S. 
saintpaul 3, S. Chester!, S. oranienburg, S. wandsbeck 
1, S. waycross 1, and S. Illb diarizonae. Salmonella 
javiana has caused severe gastro-enteritis in infants 
exposed to quokka droppings and hospitalisation has 
been needed to treat the infection. Two patients were 
infected with multiple serotypes. Salmonella javiana 
and S. typhimurlum Phage Type 202 have been the 
major strains implicated in human infections and 
non-human isolations on Rottnest Island have been 
recorded from quokkas, horses, seagulls, swallows, 
reptiles, water supplies and sewage (Iveson & Hart 
1983). The finding of S. javiana infections in quokkas, 
horses, humans and well waters at the Riding School, 
resulted in its closure and the removal of horses from 
the Island. The undoubted virulence of S. javiana and 
its absence from the food chains of humans and 
livestock on the mainland also resulted in a test 
programme and quarantine measures to exclude 
infected quokkas being translocated to mainland 
fauna reserves (Short Bradshaw Giles Prince & 
Wilson 1992). 
This unusual pattern of exposure in humans to 
reservoirs of infection in wildlife on the island differs 
from the usual pattern of food-borne Infections that is 
commonly seen in urban areas on the mainland. 
Salmonella javiana and S. typhimurlum PT 202 are rarely 
isolated from humans resident in mainland Australia and 
have not previously been detected in domesticated 
animals or in indigenous fauna. A total of 10 serovars 
classed as exotic isolates introduced by humans and 
domestic animals has been discovered in the period 1972 
to 2000 on Rottnest Island and comprise S. agona, S. 
bovismorbificans, S. derby, S. havana, S. infantis, S. 
javiana, S. Uvingstone, S. meleagridis, S. newport and S. 
typhimurlum. The majority of these serovars has been 
recorded from quokkas in settlement areas and all of the 
serotypes were also isolated from sewage effluent, 
suggesting latent infections in humans either resident or 
visiting the island. 
Although monitoring of sewage effluent provides 
evidence of silent carrier infections in humans on the 
island, symptoms of gastroenteritis in residents, 
visitors and, occasionally, research workers on the 
island have rarely been matched by collection and 
examination of patient specimens. Surprisingly, the 31 
cases of salmonellosis and 35 isolations recorded from 
humans on the island were derived from only 52 faecal 
samples and the majority of cases was diagnosed prior 
to the upgrading of public facilities on the island, as 
recommended in the Rottnest Island Management Plan 
and reviewed by (Portlock 1991). Collective studies of 
Salmonella infections in settlement areas on Rottnest 
Island commenced in 1972 and revealed infection rates 
in quokkas ranging from 20-30%. Specific information 
on the gain and loss of serotype infections was lacking, 
however, and following the discovery of 20 (74%) S. 
Adelaide infections in quokkas in the main settlement 
area on the Island this suggested a possible epizootic 
in quokkas. 
In view of the risks to public health, data were 
obviously needed on the seasonal incidence of 
Salmonella infections in quokkas inhabiting settlement 
areas and other contact areas on the Island. The aims of 
the study were thus: 
to monitor infection rates in individually-tagged 
quokkas frequenting the settlement areas on Rottnest 
Island over a 12-month period, and to compare these 
with infection rates of quokkas resident in other parts of 
the Island. 
To assess the health threat to tourists visiting the 
Island of the practice of hand-feeding quokkas at the bus 
stop adjacent to Lighthouse swamp. 
Materials and methods 
The Thompson Bay settlement and associated 
camping area on Rottnest Island (31°60'S, 115°30'E) 
occupies an area of approximately 2 km^and is bordered 
by the water-catchment area, sewage facility, golf course 
and Garden Lake (see Figure 1). Remnants of native 
vegetation provide daytime habitat for quokkas that are 
attracted to feeding sites in the early evening (Kitchener 
1972). Distances between sampling locations ranged from 
100-500 m and were subject to a degree of overlap 
during capture and recapture over the year. 
Quokkas were caught with hand-held nets between 
1800-0200 hours in the settlement and Kingston 
Barracks/Bickley Swamp area, and between 1100-1600 
hours at the tourist feeding site, close to Lighthouse 
Swamp. Animals were ear-tagged individually, scored as 
adults or juveniles, sexed and duplicate rectal swabs 
were placed in 5 mL of strontium chloride B enrichment 
broth (Iveson 1971) and isolation procedures for 
serotypes are detailed in Hart Iveson Bradshaw & Speed 
(1982). Average times for the tourist and vacation feeding 
activities averaged from 15 min to 45 min on 1-4 daily 
visits during the peak summer period and involved 
hand-feeding in roadside areas littered with many 
droppings. A number of animals was also sampled at the 
alternate Longreach and Geordie Bay settlements, located 
on the north coast, approximately 2.5 km from the main 
settlement area in Thompson Bay. Samples were 
collected on 10 occasions over a 12-month period from 
March 2, 1984 to March 8, 1985. The sampling effort was 
deliberately concentrated over the summer months when 
public numbers were greatest and the infection rate in 
quokkas was expected to be at its highest. Capture and 
recapture of tagged quokkas allowed approximate 
estimates of local populations to be calculated, using a 
simple Lincoln Index, as modified by Bailey (Bailey 1952; 
Lincoln 1930). Most individuals were recaptured less 
than 100 m from the site of their original capture, but a 
small number of individuals was highly mobile and any 
quokkas that were recaptured more than 500 m from 
their initial capture site were classified as 'mobiles' are 
analysed separately. 
Data on Salmonella infections were analysed 
statistically using ANOVA and Chi-squared tests with 
Yate's correction. 
128 
