Large predators are in retreat throughout the world. This is especially true of predator species that threaten (or at least, frighten) people, and that attack our livestock. Such species have suffered intense persecution for centuries, and efforts to eliminate them from the more densely populated parts of the world have mostly succeeded (Prugh et al. 2009). Now, changed attitudes to the conservation of wild predators are allowing, and in some places encouraging, their return to ecosystems. In Europe and North America the recovery of populations of big predators is providing outstanding opportunities to study the ways in which these species shape terrestrial communities. An example is the effects of the removal and then the restoration of wolves Canis lupus in the Yellowstone Ecosystem. Loss of wolves led to irruption of large herbivores (wapiti Cervus canadensis) and decline of woodlands by over-browsing; wolf restoration has brought wapiti back under control and is allowing woodland recovery (Beschta & Ripple 2009).
The wolf-wapiti-plant system is a classic three-level community, which behaves as predicted by Hairston, Smith & Slobodkin (1960) in their ‘green world’ hypothesis: predators protect plant biomass by checking the abundance of herbivores. But species like wolves and big cats were removed from ecosystems that also contained smaller predators. Interactions between top predators and these ‘mesopredators’ provide another pathway by which top predators can structure ecosystems (Souléet al. 1988; Crooks & Soulé 1999). Mesopredators are often generalist carnivores that, in the absence of top-down control, can reach high population densities and impose high predation pressure on a wide range of small prey species. By controlling mesopredators, top predators indirectly protect biodiversity at lower tropic levels from the effects of over-predation (Prugh et al. 2009; Ritchie & Johnson 2009).
The interaction of top predators with mesopredators is analysed in the new paper by Elmhagen et al. (2010), who show how the lynx Lynx lynx population of Finland, now returning under protection from hunting, is in the process of re-organizing the boreal ecosystem. The study is made possible by an extraordinary ecological database created by the ‘Finish Wildlife Triangle Scheme’. This consists of a network of about 900 transects, each about 12 km long, distributed over an area of 200 000 km2 (i.e. most of Finland), that are surveyed by hunters each winter for tracks. Analysis of 17 years of data from this giant monitoring program allows Elmhagen et al. to demonstrate that recolonizing lynx are strengthening top-down control in this ecosystem. Increasing lynx populations suppress fox Vulpes vulpes populations, and this reduces the impact of fox predation on the mountain hare Lepus timidus.
In Finland, both lynx and fox prey on hares. Lynx are larger than foxes, but foxes form a negligible part of the diet of lynx. These species therefore belong to a simple three-level community, if that is defined according to what they eat: lynx and foxes prey on hares, and hares eat plants. Elmhagen et al. analyse the changing structure of this community along a gradient of productivity, from the productive southwest of Finland to the relatively unproductive northeast. They do this because they want to understand how productivity affects the structure of their system, and because Oksanen & Oksanen (2000) have provided clear predictions on the way in which productivity gradients should affect the distribution of biomass among trophic levels in ecosystems with top-down control by predators (Fig. 1). Oksanen & Oksanen’s ‘exploitation ecosystem hypothesis’ argues (for endotherms) that in unproductive environments, herbivore populations will be too sparse to support predators. In the absence of predator control, herbivores hold plant biomass low. Increased productivity allows plants to make more tissue, but this surplus is consumed by herbivores. So as productivity rises, herbivore biomass increases, but standing plant biomass does not. At the point where there are enough herbivores to support a predator, productivity-driven increases in recruitment to the herbivore population are consumed. Predator populations therefore increase with productivity while herbivore populations are held constant, and this allows plant biomass to increase with productivity, creating Hairston et al.’s green world.
Viewing the lynx-fox-hare system as the top part of a simple three-level predator-herbivore-plant community therefore leads to this prediction: the biomass of lynx and fox should increase along productivity gradients, while hare biomass should be constant. But while lynx rarely eat foxes, they do attack and kill them. In Sweden where lynx are also currently increasing, 50% of the deaths recorded in a radio-tracked sample of foxes were due to attacks by lynx, which caused an annual mortality of 14% (Helldin, Liberg & Gloersen 2006). This aggression can be interpreted as a strong form of interference competition for a shared prey, in which lynx pre-empt competition from foxes by killing them. Lynx should therefore have a strong negative effect on fox biomass, and that effect is likely to be strongly amplified if foxes avoid habitats where they are likely to encounter lynx (Ritchie & Johnson 2009). However, because (as Elmhagen et al. show) lynx are on average about 40 times less abundant than foxes, the impact of lynx predation on hare biomass should be less than that of an unconstrained fox population. So, applying the logic of Oksanen & Oksanen (2000) to this case of interference between predators leads to the following predictions (Fig. 1): the biomass of lynx should increase with productivity, while the biomass of foxes should not; the biomass of hares should also increase with productivity, but it should on average be higher than it would be if no lynx were present and hare populations were held constant by fox predation.
Elmhagen et al.’s analysis shows rapid population increase of lynx since the early 1990′s. Lynx and fox biomass were negatively related, while lynx and hare biomass were positively related. By constructing models that account for variation in productivity, Elmhagen et al. show that fox biomass was lower than expected from productivity where lynx biomass was high, while hare biomass was lower than expected where fox biomass was high.
The re-establishment of lynx is not yet complete, and they are still increasing throughout the country. This observation suggests another way in which productivity could affect the lynx-fox-hare system. In unproductive systems, where foxes and hares have naturally low abundances, a smaller population of lynx will be needed to enforce control of foxes. This will be achieved earlier in the process of population growth. Thus, Elmhagen et al. find that in the unproductive northeast of Finland top-down control is currently strongest: as predicted (Fig. 1), lynx and hare biomass increase with productivity within this region, while fox biomass does not. In the productive southwest, where lynx biomass relative to productivity is still low, fox biomass increases with productivity while hare biomass does not. The southwest appears to be a ‘mesopredator-release system’ in which top-down control of mesopredators biomass has not yet been restored; presumably, this will happen over the next few years if lynx are allowed to continue their recovery.
As yet, there has been no general increase of hares in Finland following lynx re-invasion. This is possibly because climate effects – recent warm winters – have disadvantaged hares, and the effect of lynx has been to offset the impact of this and allow them to sustain their populations. Elsewhere, recovery of lynx has allowed increase in the prey of red foxes, such as hare and grouse (Helldin et al. 2006), or of competitors of red foxes such as arctic foxes (Shirley et al. 2009). This effect is likely to be strongest in places where the preferred prey of lynx (small deer) are abundant, because then lynx continue to suppress red foxes but without hunting the same species as foxes.
Elmhagen et al.’s study provides a striking demonstration of the power of large predators to regulate interactions in communities, and control the abundance of other species through cascading effects. Their approach elegantly reveals the dynamics of the effects of a top predator as it returns to an ecosystem, and by focussing on productivity gradients they provide a link to classical ecological theory based on energy flows through distinct trophic levels. They also show that this classical approach is not useful in cases where species have strong antagonistic interactions that do not consist simply of eating and being eaten. Viewing this system simply as a food web is inadequate, and to understand the mechanisms involved and predict its dynamics it will be much more useful to take an approach based on network theory, as recently argued by Ings et al. (2009). This will be especially valuable when we come to tackle more complex systems, in which top predators interact with multiple mesopredator species as well as with their own prey and (indirectly) with the prey of mesopredators, while mesopredators also interfere with one another.
Many conservation biologists now advocate the restoration of top predators as a way to restore stability to ecosystems, and prevent extinction of species vulnerable to the impacts of over-abundant mesopredators (Terborgh et al. 1999; Hayward & Somers 2009). The story of lynx in Finland, as it continues to unfold, will be especially valuable as a demonstration of the results that predator restoration could bring elsewhere.