Alien predators are widely considered to be more harmful to prey populations than native predators. To evaluate this expectation, we conducted a meta-analysis of the responses of vertebrate prey in 45 replicated and 35 unreplicated field experiments in which the population densities of mammalian and avian predators had been manipulated. Our results showed that predator origin (native versus alien) had a highly significant effect on prey responses, with alien predators having an impact double that of native predators. Also the interaction between location (mainland versus island) and predator origin was significant, revealing the strongest effects with alien predators in mainland areas. Although both these results were mainly influenced by the huge impact of alien predators on the Australian mainland compared with their impact elsewhere, the results demonstrate that introduced predators can impose more intense suppression on remnant populations of native species and hold them further from their predator-free densities than do native predators preying upon coexisting prey.

The four-year cycle of microtine rodents in boreal and arctic regions was first described in 1924 (ref. 1). Competing hypotheses on the mechanisms underlying the small mammal cycle have been extensively tested, but so far the sustained rodent oscillations are unexplained. Here we use two mutually supportive approaches to investigate this question. First, building on studies of the interaction between rodents and their mustelid predators, we construct a predator-prey model with seasonality. Second, we use a new technique of nonlinear analysis to examine empirical time-series data, and compare them with the model dynamics. The model parameterized with field data predicts dynamics that closely resemble the observed dynamics of boreal rodent populations. Both the predicted and observed dynamics are chaotic, albeit with a statistically significant periodic component. Our results suggest that the multiannual oscillations of rodent populations in Fennoscandia are due to delayed density dependence imposed by mustelid predators, and are chaotic.

We studied numerical and functional responses of breeding European Kestrels (EK) (Falco tinnunculus), Short—eared Owls (SO) (Asio flammeus), and Long—eared Owls (LO) (Asio otus) during 1977—1987 in 47 km2 of farmland in western Finland. The pooled mean yearly breeding density varied from 0.1 to 2.4 pairs/km2. The number of nesting EKs (range 2—46 pairs), SOs (0—49), and LOs (0—19) fluctuated in close accordance with the spring density of Microtus (M. agrestis and M. epiroticus) voles. The mean yearly number of fledglings produced per pair ranged from 0.4 to 3.8 and, for each species, was positively correlated with spring density of Microtus voles. Due to their high degree of mobility, EKs, SOs, and LOs were able to track the population fluctuations of their microtine prey without time lags. An increase in microtine densities caused a rapid immigration into the study area and a decrease caused a rapid emigration from the area. Microtus voles were the most important prey group by mass in the diet of each species. Water voles, bank voles, shrews, and small birds were the most frequent alternate prey. The spring density of Microtus spp. was positively correlated with the percentage of these voles in the diet of EK, SO, and LO. The pooled functional response curve of these three raptor species to the fluctuating densities of Microtus spp. was close to linear, indicating that consumption rates are independent of vole densities. Breeding EKs, SOs, and LOs seemed to take a larger proportion of voles available in peak years than in low ones.

We studied responses of stoats and least weasels to fluctuating vole abundances during seven winters in western Finland. Density indices of mustelids were derived from snow-tracking, diet composition from scat samples, and vole abundances from snap-trapping. Predation rate was estimated by the ratio of voles to mustelids and by the vole kill rate by predators (density of predator x percentage of voles in the diet). We tested the following four predictions of the hypothesis that small mustelids cause the low phase of the microtine cycle. (1) The densities of predators should lag well behind the prey abundances, as time lags tend to have destabilizing effects. The densities of stoats fluctuated in accordance with the vole abundances, whereas the spring densities of least weasels tracked the vole abundances with a half-year lag and the autumn densities with a 1-year lag. (2) Predators should not shift to alternative prey with declining vole densities. The yearly proportion of Microtus voles (the staple prey) in the diet of stoats varied widely (range 16-82%) and was positively correlated with the winter abundance of these voles. In contrast, the same proportion in the food of least weasels was independent of the vole abundance. (3) The ratio of voles to small mustelids should be smallest in poor vole years and largest in good ones. This was also observed. (4) Vole densities from autumn to spring should decrease more in those winters when vole kill rates are high than when they are low. The data on least weasels agreed with this prediction. Our results from least weasels were consistent with the predictions of the hypothesis, but stoats behaved like "semi-generalist" predators. Accordingly, declines and lows in the microtine cycle may be due to least weasel predation, but other extrinsic factors may also contribute to crashes.

Summary 1.We investigated the hypothesis that cyclic lemming populations indirectly affect arcticnesting greater snow geese ( Anser caerulescens atlanticus L.) through the behavioural and numerical responses of shared predators. 2. The study took place on Bylot Island in the Canadian High Arctic during two lemming cycles. We recorded changes in foraging behaviour and activity rate of arctic foxes, parasitic jaegers, glaucous gulls and common ravens in a goose colony during one lemming cycle and we monitored denning activity of foxes for 7 years. We also evaluated the total response of predators (i.e. number of eggs depredated). 3. Arctic foxes were more successful in attacking lemmings than goose nests because predators were constrained by goose nest defence. Predators increased their foraging effort on goose eggs following a lemming decline. 4. Activity rates in the goose colony varied 3·5-fold in arctic foxes and 4·8-fold in parasitic jaegers, and were highest 2 and 3 years after the lemming peak, respectively. The breeding output of arctic foxes appeared to be driven primarily by lemming numbers. 5. Predators consumed 19-88% of the annual goose nesting production and egg predation intensity varied 2·7-fold, being lowest during peak lemming years. Arctic foxes and parasitic jaegers were the key predators generating marked annual variation in egg predation. 6. Our study provides strong support for short-term, positive indirect effects and long-term, negative indirect effects of lemming populations on arctic-nesting geese. The outcome between these opposing indirect effects is probably an apparent competition between rodents and many terrestrial arctic-nesting birds.

No abstract

The hypothesis that the regular multiannual population oscillations of boreal and arctic small rodents (voles and lemmings) are driven by predation is as old as the scientific study of rodent cycles itself. Subsequently, for several decades, the predation hypothesis fell into disrepute, possibly because the views about predation and rodent dynamics were too simplistic. Here we review the work that has been done on the predation hypothesis primarily in Fennoscandia over the past decade. Models of predator–prey interaction have been constructed for the least weasel (Mustela nivalis) and the field vole (Microtus agrestis), which are considered to be the key specialist predator and the key prey species in the multispecies communities in the boreal forest region in Fennoscandia. The basic model has been parameterized with independent field data, and it predicts well the main features of the observed dynamics. An extension of the model also including generalist and nomadic avian predators predicts correctly the well‐documented and striking geographic gradient in rodent oscillations in Fennoscandia, with the amplitude and cycle period decreasing from north to south. These geographic changes are attributed to the observed latitudinal change in the density of generalist and nomadic predators, which are expected to have a stabilizing effect on rodent dynamics. We review the other observational, modeling, and experimental results bearing on the predation hypothesis and conclude that it accounts well for the broad patterns in rodent oscillations in Fennoscandia. We discuss the application of the predation hypothesis to other regions in the northern hemisphere. The predation hypothesis does not make predictions about multiannual and latitudinal changes in body size, behavior, and demography of rodents, which may have some population‐dynamic consequences. With the current evidence, however, we consider it unlikely that the phenotypic and genotypic composition of populations would be instrumental for generating the broad patterns in rodent oscillations.

Quantifying the relative impacts of top‐down vs. bottom‐up control of ecosystems remains a controversial issue, with debate often focusing on the perennial question of how predators affect prey densities. To assess predator impacts, we performed a worldwide meta‐analysis of field experiments in which the densities of terrestrial vertebrate predators were manipulated and the responses of their terrestrial vertebrate prey were measured. Our results show that predation indeed limits prey populations, as prey densities change substantially after predator manipulations. The main determinant of the result of an experiment was the efficiency of predator manipulation. Positive impacts of predator manipulation appeared to increase with duration of the experiment for non‐cyclic prey, while the opposite was true for cyclic prey. In addition, predator manipulation showed a large positive impact on cyclic prey at low prey densities, but had no obvious impact at peak prey densities. As prey population densities generally respond predictably to predator manipulations, we suggest that control of introduced vertebrate predators can be used to effectively conserve and manage native wildlife. However, care should be taken when controlling native predators, especially apex species, owing to their importance as strong interactors and the biodiversity value of their habitats. We discuss gaps in our knowledge of predator–prey relationships and methodological issues related to manipulation experiments. An important guideline for future studies is that adequate monitoring of predator numbers before and during the experiment is the only way to ensure that observed responses in prey populations are actually caused by changes in predation impacts.