Kangaroo

The overall survey data were non-normally distributed, as indicated by Kolmogorov–Smirnov [48 ] and Shapiro–Wilk [49 ] testing, so we decided to analyse the impact of Human Position, Featured Animals, and other demographics on each agreement statement with a linear regression approach.
Following the approach of Van der Meer, Botman, and Eckhardt [38 (link)], we collapsed the 7-point Likert scale categories into 5 for each of the agreement statements, to allow for an adequate number of events for each variable. Thus, “strongly disagree” and “disagree” were combined, as were “strongly agree” and “agree”.
Pearson’s chi-square tests were then conducted to compare the two samples across each of the demographic questions. The results showed significant differences between the two samples for most demographic questions, so the samples were separated for further analyses. A principal components analysis was conducted of the three agreement statements for both samples, to determine if the agreement statements were measuring different perceptions. Results indicated that none of the three agreement statements were closely correlated with each other for either sample, so distinct statistical analysis was appropriate.
As the responses to each of the agreement statements represented a scale of agreement, we fitted ordinal regression models to each of the three statements for each sample, as done by Van der Meer, Botman, and Eckhardt [38 (link)], for a total of 6 models. As demographic information such as gender, age, and education level has been shown to impact our perceptions and attitudes towards animals [50 ,51 ], all collected demographic information, including zoo visitation and zoo membership, was added to the original models, as well as the Human Position and Featured Animal that each respondent saw in their allocated image. All possible interactions between each of these variables were also added. Ordinal regression modelling was conducted with SPSS software version 25.0 [52 ] using forward selection and then a GENLIN procedure, which is centred on likelihood testing. An ordinal regression was not suitable for the second model that examined the relationship between the variables on agreement that the zoo cares for the featured animal for the General Public (Pearson’s goodness of fit test p < 0.05, test of parallel lines p < 0.05). Instead, a multinomial regression was fitted. For all models, any variables and interaction terms that did not have a significance level (p-value) of > 0.05 were removed via forward selection (for a full list of variables tested, see Supplementary File 1, and any non-significant interaction effects). For comparison analyses, the Kangaroo was set as the reference animal as it was the most recognised species and least likely to be a pet. The animal-alone condition was also set as a reference category for comparison of analyses of Human Position effects.
For both samples the model fit to the statement “the featured animal is cared for by the zoo” and for the Zoo Community samples’ model fit to the statement “the featured animal would make a good pet” included a significant interaction term between the Featured Animal and the Human Position, so post-hoc Pearson’s chi-square testing with Bonferroni corrections were conducted to determine the nature of these interactions.

Pellets were collected from Cleland captive koalas (n = 4) and Kangaroo Island koalas (rescued wild koalas, n = 3) housed at the Cleland Wildlife Park (34°58′01.5″S, 138°41′49.0″E) in the Adelaide Hills, SA, Australia (Table 1). The koalas were housed individually for the duration of the study, as they were being kept isolated during testing for koala retrovirus and Chlamydia. Between April and May 2020, freshly defecated faecal pellets were collected every 3 to 4 days for a period of one month into 5 mL tubes and immediately stored in a −20 °C freezer (Table S2). All koalas were supplied with branches of a range of eucalypt species daily, primarily including river red gum (E. camaldulensis), manna gum (E. viminalis), and South Australian blue gum (E. leucoxylon) (Table S2). Cleland sampling was conducted with University of Adelaide Animal Ethics Committee Approval S-2016-169 and Department for Environment and Water scientific permit Y26054.