Why do indigenous fauna often fail to respond to predator control?
Predicting the outcomes of managing various parts of dryland ecosystems (Fig. 1) can be difficult. However, understanding how these ecosystems are structured, and the strengths of the component interactions, is needed in order to maintain and restore indigenous biodiversity. Grant Norbury and colleagues conducted a large-scale field experiment in Otago to examine the impacts of introduced predatory mammals on local indigenous fauna.
Grant’s team controlled cats, mustelids, and hedgehogs over 400 ha near Alexandra, and over 650 ha near Macraes Flat, by deploying kill-traps continuously at high densities (26 per km2) for more than 3 years. They measured the responses of indigenous lizards and invertebrates (and mice, rabbits and hares) in the area trapped for predators, and compared them to the responses in replicate non-trapped areas. They looked at changes in abundance over time, and compared the way rates of population change depended on their initial density. Intriguingly, the researchers found little evidence of an effect of predator removal on indigenous fauna (e.g. lizards, see Fig. 2).
To explain the lack of responses of indigenous fauna, Grant and his colleagues proposed four alternative hypotheses:
Hypothesis 1: Removal of cats and mustelids leads to more mice, and therefore increased predation or competition with indigenous fauna.
The data did not support this hypothesis: rates of change for mice over the critical winter period were no different between the treated and untreated areas (Fig. 3).
Hypothesis 2: Predator removal leads to increased herbivore populations and therefore depleted habitat for indigenous fauna.
There was no evidence that populations of rabbits and hares increased in response to predator removal, or that vegetation biomass was depleted. For example, pasture growth rates did not differ between predator removal and non-removal sites (Fig. 4).
Hypothesis 3: Indigenous fauna are not limited by predation but are driven primarily by food resources.
There was some support for this hypothesis because fruit abundance was a key predictor of lizard population growth rates in summer, and seed abundance was an even stronger predictor of mouse dynamics (Fig. 5). As areas of grassland are destocked, increases in shrub fruits, grasses, and seeds are likely to affect lizard and mouse populations. Mice are predators and potential competitors with lizards so the net effect of destocking could be detrimental for some lizard species, or be beneficial for species that respond to increased vegetation cover.
Grant and colleagues suggest that species conservation in particularly dry environments may be better served by addressing ‘bottom-up’ processes, such as food and habitat quality, rather than top-down processes such as predation.
Hypothesis 4: Predator removal is poorly done.
The simplest explanation may be that the predator removal was ineffective at maintaining predator abundances at low enough levels. This is because despite intensive trapping, there was no clear evidence of a decline in the abundance of cats and mustelids throughout the 3-year study, presumably because the areas were too small to overcome their constant immigration.
The possibility of poor trapping efficacy of top predators in this experiment means Grant and his team cannot reject the first three hypotheses dealing with lack of top-down regulation. While this is frustrating, the lessons learnt are valuable for ecological restoration. In particular, there are management-scale thresholds that are broadly governed by the mobility of the target pest species, and pest management resources are likely to be wasted unless control operations exceed this area threshold, or small control areas are protected by predator-proof barriers. This study indicates that predator control of less than 650 ha will be insufficient to overcome their immigration. Therefore, control strategies that are cognisant of operational scale and the mobility of the targeted pest species are more likely to achieve successful outcomes.
This work was funded by the Ministry of Science and Innovation (Programmes C09X0505 and C09X0909).
Grant Norbury, James Smith, Andrea Byrom, Roger Pech, Dean Clarke & Guy Forrester
Fig 1. A simplified food web in a typical dryland ecosystem.
Fig 2. Trends in lizard population densities at the Alexandra and Macraes Flat sites with (blue line) and without (red line) predator trapping (means ± 95% CL). Arrows indicate start of trapping.
Fig 3. Daily population growth rates (r) of mice during winter with (blue points) and without (red points) predator trapping, plotted against the density of mice in autumn.
Fig 4. Daily growth rates (r) of pasture with (blue points) and without (red points) predator trapping (pooled for spring, summer and autumn/winter periods), plotted against the initial pasture biomass.
Fig 5. Daily population growth rates (r) of mice between spring and autumn in relation to grass seed abundance at the Alexandra and Macraes Flat sites.
