The hypothesis that predation by feral cats and introduced foxes reduces population sizes of small, native vertebrates was supported by results of a predator-removal experiment at Heirisson Prong, a semi-arid site in Western Australia. The methods of control used against cats and foxes to protect native mammals reintroduced to Heirisson Prong produced three broad ‘predator zones’: a low-cat and low-fox zone, where foxes were eradicated and spotlight counts of cats declined after intensive cat control; a high-cat and low-fox zone where spotlight counts of cats increased three-fold after foxes were controlled; and a zone where numbers of cats and foxes were not manipulated. Small mammals and reptiles were monitored for one year before and three years after predator control began. Captures of small mammals increased in the low-cat and low-fox zone, but where only foxes were controlled captures of small mammals declined by 80%. In the absence of cat and fox control, captures of small mammals were variable over the sampling period, lower than where both cats and foxes were controlled, yet higher than where only foxes were controlled. The capture success of reptiles did not appear to be related to changes in predator counts. This study presents the first experimental evidence from mainland Australia that feral cats can have a negative impact on populations of small mammals.

International differences in practices and attitudes regarding pet cats' interactions with wildlife were assessed by surveying citizens from at least two cities in Australia, New Zealand, the UK, the USA, China and Japan. Predictions tested were: (i) cat owners would agree less than non-cat owners that cats might threaten wildlife, (ii) cat owners value wildlife less than non-cat owners, (iii) cat owners are less accepting of cat legislation/restrictions than non-owners, and (iv) respondents from regions with high endemic biodiversity (Australia, New Zealand, China and the USA state of Hawaii) would be most concerned about pet cats threatening wildlife. Everywhere non-owners were more likely than owners to agree that pet cats killing wildlife were a problem in cities, towns and rural areas. Agreement amongst non-owners was highest in Australia (95%) and New Zealand (78%) and lowest in the UK (38%). Irrespective of ownership, over 85% of respondents from all countries except China (65%) valued wildlife in cities, towns and rural areas. Non-owners advocated cat legislation more strongly than owners except in Japan. Australian non-owners were the most supportive (88%), followed by Chinese non-owners (80%) and Japanese owners (79.5%). The UK was least supportive (non-owners 43%, owners 25%). Many Australian (62%), New Zealand (51%) and Chinese owners (42%) agreed that pet cats killing wildlife in cities, towns and rural areas was a problem, while Hawaiian owners were similar to the mainland USA (20%). Thus high endemic biodiversity might contribute to attitudes in some, but not all, countries. Husbandry practices varied internationally, with predation highest where fewer cats were confined. Although the risk of wildlife population declines caused by pet cats justifies precautionary action, campaigns based on wildlife protection are unlikely to succeed outside Australia or New Zealand. Restrictions on roaming protect wildlife and benefit cat welfare, so welfare is a better rationale.

The Index of Relative Importance (IRI) is a composite measure that reduces bias in descriptions of animal dietary data. The two papers introducing the IRI in 1971 had been cited a total of 214 times by the end of 2001 and proposed as a standard methodology. However, 180 of these citations concerned the description of the diets of fish, indicating that the IRI is not well known outside fisheries biology. This illustrates how the interests of researchers in a narrow range of taxa may restrict the application of a useful technique to particular groups of animals. Here we apply the IRI to dietary data from one mammal species, two bird species and two species of geckoes to illustrate its applicability to a wide range of taxa. We believe the approach should be considered seriously by terrestrial ecologists concerned about the biases inherent in single-index approaches to describing animal diets.