We selected publications that described the effect of reduction or addition of avian or mammalian predators on avian or mammalian prey, excluding livestock and other non-native prey species. Studies that had removed both native and introduced predators were excluded, if the effects of these predator groups could not be separated. Acceptable prey responses to predator manipulations were classified as either population size or reproductive responses. Population size responses included those measured directly, as density, minimum numbers known to be alive, numbers of breeding pairs (as an index of population size), rate of increase or survival; and catch-per-unit effort indices such as the number of animals per area, trapline or transect. Reproductive responses included numbers of juveniles or broods produced, numbers of females with young, nesting success, survival of young and mean recruitment. Per capita measures, such as brood size per hen, number of juveniles per hen, number of broods per pair, number of fledglings/ducklings per pair, number of chicks fledged per pair, number of fawns/100 does, etc., were not included. The studies also had to have been run for long enough (one prey generation or more) for a prey demographic response to be possible. The studies measuring other parameters or using other units than those described were omitted. No authors were contacted to obtain missing data.

Necessary data (sample sizes, means of controls and treatments and their standard deviations/standard errors/confidence limits) were extracted from the text, tables or figures of the articles. In cases where error bars were not symmetrical about means, variances were calculated conservatively using the longest bars provided and, if no variances were given, these were calculated from raw results. Where possible, we used data taken at the end of experiments, but otherwise used an arithmetic mean of responses over the course of the studies. In cyclically fluctuating prey species, such as some small mammals, data were taken for consistency from the peak phase of the cycle. Reproductive responses were taken separately for each year, and an arithmetic mean calculated across the duration of the study to obtain one effect size per study.

Publications were scored also for the type of prey (e.g. rodent, ungulate, marsupial), prey class (bird/mammal), origin of predator (native/introduced), predator class (bird/mammal/both), the method of manipulation (addition/removal), location (mainland/island), habitat, spatial and temporal scales of the experiment (manipulation area and manipulation time), the continent in which the study was conducted and whether it was conducted in an exclosure or open terrain (manipulation type). We also recorded the mean weight of prey and predator species to calculate a predator/prey weight ratio for each study. Predators were considered to be introduced or native based on definitions provided in each study and confirmed using Long (2003). ‘Predator addition’ means either release of predators into experimental areas or attraction of predators (e.g. attraction of raptors with perch sites). ‘Predator removal’ refers to either exclosure experiments or manipulations, where predators were killed or relocated. Two persons (M.N. and P.S.) were responsible for data collection.

The 45 replicated experiments (table S1 in electronic supplementary material) were examined for publication bias using the normal quantile plot method (Wang & Bushman 1998), and no evidence of publication bias was found (figure S1 in electronic supplementary material). This analysis was not possible for unreplicated studies and, therefore, the publication of such studies may have been biased towards large positive effects of predator removal on prey. However, it is rather unlikely that the results, whether significant or not, of very expensive, long-lasting predator manipulation experiments would remain unpublished, strongly reducing the likelihood for the file-drawer problem (Rosenthal 1979) particularly in this meta-analysis.

For each replicated study, we calculated the standardized effect size as Hedges' d using MetaWin v. 2.1 (Rosenberg et al. 2000). There are also other metrics available for this type of primary data (means, variances and sample sizes), such as the log response ratio lnR (Rosenberg et al. 2000), but we chose d because our data were not suitable for use of the response ratio (e.g. in some studies, the control group value was zero; Hedges et al. 1999). Positive values of d indicate that the predator treatment had a positive effect on prey species, zero means that there was no difference between treatment and control, and negative values signify a greater response in controls. For studies that reported the responses of multiple prey species to predator manipulation, we used the mean effect size across all species to retain independence. In one study, predators had been both added and removed; also here a mean effect for the whole study was calculated from the effect sizes of both treatments.

Our first prediction was that introduced predators should have more pronounced effects than native predators on the population sizes and reproductive outputs of their prey. To test this prediction with the 45 replicated experiments, we carried out a categorical summary analysis using the homogeneity statistic, Q, in MetaWin v. 2.1. As with variance in ANOVA, the total heterogeneity Q_T can be partitioned into Q_M, the variation explained by the model, and Q_E, the residual error variance (Rosenberg et al. 2000). Continuous summary analysis (weighted linear regression) was used to determine whether d was affected by the spatial or temporal scale of the studies. We used random effects models and conducted resampling tests with 4999 iterations. Bias-corrected confidence intervals were used to evaluate the probability at 0.05. All tests were two-tailed.

To expand the coverage of research that has evaluated the impacts of predation, we conducted a similar analysis on the unreplicated predator removal experiments. Altogether, 34 unreplicated studies fulfilled the criteria, but two were excluded as they reported earlier stages of experiments that were represented in the analysis by later, more inclusive papers. In two cases, different aspects of the same experiment were reported in separate papers, which were then combined to gain one effect size. One study consisted of six experiments at different locations, which were therefore treated as independent studies in the dataset. Hence, the final dataset has 35 rows (table S2 in electronic supplementary material).

We classified the traits of the unreplicated experimental systems as described previously and defined the effect size as X_e/X_c, where X_e and X_c are the treatment and control prey responses, respectively. A ratio over 1 means that predator manipulation had a positive effect on the prey species, while a ratio up to 1 means that manipulation did not affect the prey species or the effects were negative. These unreplicated data cannot be analysed using typical meta-analysis approaches; therefore, we tested for differences in effect size in the study traits using Student's t-test with the Satterthwaite option for heteroscedastic variances (procedure TTEST, SAS Statistical Package, v. 9.1; SAS Institute, Cary, NC, USA). Effect size was ln transformed to meet the assumptions of normality.

Finally, a generalized linear model was built in order to further test our first and second predictions (i.e. that alien predators would have more impact than native predators on prey populations and that predation impacts would be greater on prey in island ecosystems compared with mainland ecosystems), and to explore possible interactions of the different explanatory variables. Neither MetaWin nor t-test allows the simultaneous analysis of multiple factors, and the sample sizes of replicated and unreplicated experiments alone were too small for such an analysis. Therefore, we pooled population size responses of the replicated and unreplicated experiments using X_e/X_c as the effect size measure. The model was fitted with a negative binomial distribution of the response and a log link function with the negative binomial GLM (glm.nb) procedure in the MASS library of S-Plus (v. 6, Insightful Corporation, Seattle, USA).

In the replicated experiments, the effects of introduced predators on prey were more than double those of native predators (figure 1a; Q_M=4.96, d.f.=1, p=0.020). Further partitioning revealed very striking effects of introduced predators in Australia compared with other parts of the world. The mean effect size of prey responding to the removal of introduced predators in Australia (all mammalian responses) was twofold higher than that of prey responding to introduced predators elsewhere and threefold higher when compared with native predators generally (figure 1b; Q_M=7.07, d.f.=2, p=0.022).

There was a significant difference between population size effects and reproductive responses in the experiments, the latter being larger (table 1). Therefore, we reanalysed the data from only those experiments where population size responses were measured. The analysis of population size responses revealed the significant overall difference between alien and native predators again and also a significant difference between native and alien predators outside Australia (table 1). This is probably because most of the studies on reproductive responses had manipulated native predators (table S1 in electronic supplementary material).

Experiments in exclosures may be confounded by the ‘fence effect’, whereby the enclosed populations reach high densities when emigration is prevented (Krebs et al. 1969). To explore the influence of this phenomenon on our results, we tested whether there were differences in predation impacts between exclosure and open-area experiments. Interestingly, the effect sizes of prey populations in exclosure experiments were similar to those in open terrain, but removal of exclosure experiments from the analysis increased the difference in effect sizes produced by introduced and native predators (table 1). We also checked for possible biases arising from differences among studies in their duration and spatial extent by regressing effect size on time and area of study (continuous meta-analysis). There was no bias arising from either study duration (slope=0.017, d.f.=1, p=0.99, n=45) or spatial coverage (slope=0.001, d.f.=1, p=0.51, n=45).

Of the 35 unreplicated experiments included in the analyses, 13 removed alien predators (table S2 in electronic supplementary material). Six studies were from islands, 12 were from Australia and New Zealand and 17 were from mainland areas. Twenty-two studies examined the responses of mammalian prey, while the rest investigated birds. Eighteen studies recorded population size responses, and six reported reproductive responses. In 11 studies, both responses were measured but, as in the replicated studies, only population size responses were used. There was no difference between the two responses in the pooled data (t=1.18, d.f.=13.7, p=0.258; reproduction: back-transformed ln(X_e/X_c)=1.681, lower 95% CL=0.881, upper 95% CL=3.210, n=6; population size: back-transformed ln(X_e/X_c)=2.486, lower 95% CL=1.449, upper 95% CL=3.863, n=29).

Analyses of the unreplicated studies yielded remarkably similar results to those of the replicated experiments (figure 2). The effects of introduced predators were on average 2.5 times higher than for native predators (figure 2a; t=2.22, d.f.=16.3, p=0.041), but again the difference appeared to be influenced mainly by the large effects in Australian studies (figure 2b). There were no obvious biases in effect size resulting from study duration or spatial coverage (Spearman's rank correlation, r_s=0.246, p=0.16, n=35 and r_s=0.184, p=0.31, n=33, respectively).

Despite the skew in availability of research on alien predators, there was evidence that the greater impact of alien predators is not simply restricted to Australia. Outside of Australia, there was also a higher impact of alien predators compared with native predators on prey population responses in the replicated studies (table 1). Furthermore, analyses of the different explanatory variables and their interactions in the broader pooled data from replicated and unreplicated studies revealed that only predator origin (native versus alien) had a highly significant effect on population size responses of prey animals. Although the interaction between predator origin and location (mainland versus island) was also slightly significant, these results together suggest that, in general, alien predators may have more severe impacts on prey than do native predators. One potential cause of the consistently large impact of alien predators on prey in Australia is small sample size bias. Three of the eight replicated studies were from Australia and were ranked second, sixth and seventh in magnitude of effect sizes. The seven unreplicated Australian studies also revealed an identical phenomenon, making a total of 10 studies that met our strict criteria for inclusion in the analysis. These studies were diverse; they spanned the continent, were conducted in environments ranging from temperate forest through to deserts and agricultural areas and involved prey body sizes from small rodents to large marsupials. We surmise that the geographical and faunal characteristics of Australia strongly contribute to especially large alien predator impacts there (discussed in detail below) and, therefore, feel confident that the 10 studies are a good representation of the situation there.

In communities where predators and prey have coexisted for long periods, prey often respond to predatory pressure by developing behaviours or morphologies that reduce the chance of encounters with predators or enhance the chance of escape once detected (Lima & Dill 1990). In contrast, prey in communities with novel alien predators are likely to be predator-naive and to lack specific avoidance behaviours. Such naiveté would facilitate greater hunting efficiency in predators and lead to greater suppressive impacts on naive prey compared with native predators hunting prey with which they have long coexisted. It is possible that the frequent historical biotic interchanges between the contiguous continents of Eurasia, Africa and the Americas cause them to share similar terrestrial predator archetypes, which may render prey less naive towards novel predators introduced from the same continents (Cox & Lima 2006). Recent research shows that Australia has also possessed a rich assemblage of marsupial carnivores from Miocene to recent times (Wroe et al. 2004) and that the native fauna may not be as predator naive as previously thought. However, Australia never had placental carnivores until they were introduced by humans, and it may be that these novel predators use tracking and hunting tactics that differ from those of their marsupial counterparts, to which native prey have little or no defence (Croft & Eisenberg 2006).

Alien predators have long been presumed to have greater impacts in island ecosystems when compared with the mainland ecosystems, for reasons including prey naiveté, yet we found no support for such phenomena. In these analyses, Australia was considered mainland; but despite its large size, Australia's island-like characteristics, such as geographical isolation and diversity of endemic species, may also have contributed to the profound impacts of alien predators there. Many native Australian prey populations are now restricted to small, island-like refugia within large tracts of unsuitable, disturbed land owing to the widespread alteration and fragmentation of habitat, and natural adaptations to stochastic environmental conditions (Morton 1990; Letnic & Dickman 2006). Compared with real island ecosystems, the large land mass of Australia may have provided more room for such refugia, which may possibly have saved some native prey populations from otherwise inevitable extinction. Although the large-scale effects of alien predators may be blunted by such refuge areas, the island-like characteristics of the refugia make them especially vulnerable to alien predators, which could be expected to have very depressive effects on local populations of native prey where predator activity is intense. Indeed, one of the largest observed effect sizes, 4.80, followed the removal of the red fox from rock outcrops containing remnant colonies of rock wallabies (Petrogale lateralis; Kinnear et al. 1998).

The concept of filter effects suggests that remnant species which survive the initial alien invasion are resistant to their ongoing impacts (Pimm et al. 1995). For example, it has recently been reported that extinction events of island birds can be related to the numbers of introduced predator species, but no relationship was found between the numbers of exotic predators and the current risk of extinction for remnant populations of surviving species (Blackburn et al. 2004; but see Blackburn et al. 2005; Didham et al. 2005). Our results challenge this notion and show that once released from the impact of alien predators, remnant bird and mammal populations undergo greater population increases than those experiencing predation from native predators. This indicates that alien predators can impose intense population suppression, which also puts prey at greater inherent risk of extinction from stochastic forces. Our results also contradict suggestions that alien predators merely compensate for native predators which themselves have been lost from the system (de Vos et al. 1956; Long 2003). Instead, the intense impacts may be compounded if additive to other sources of mortality, and could be expected to further destabilize community processes.