A stochastic computer simulation of island group colonisation by Rattus norvegicus in small near shore island systems : specifically Tia island and the Boat group

Abstract

Rats have a devastating effect of the biota of islands. In New Zealand rats are responsible for the extinction or reduction of many species. In recent times an important conservation strategy in New Zealand has been to clear offshore islands of rats so that they can be used as havens for species vulnerable to rat predation. One of the main limitations in the application of this strategy is our ability to predict how long an island will stay rat free once it is cleared. In this study stochastic simulation models focusing on Rattus norvegicus (Norway rat) were developed to help managers better predict the outcome of rat eradications. A sensitivity analysis was carried out to determine what parameters the outcome of the model was most sensitive to. Modelled experiments on Tia Island and the Boat Group were performed. The modelled experiment on Tai Island asked; whether or not Tia Island could be rid of rats either by repeated eradication or permanent poisoning operations, and do age biased mortality rates have an effect on suppressing populations of R. norvegicus on islands? The modelled experiment on the Boat Group asked the questions; can stopping the colonisation of one patch in a small metapopulation have consequences for other patches in that metapopulation, is the model strong enough to make precise predictions about how long each island in the group will take to be invaded for the first time, what is the predicted invasion pattern of the Boat Group, and is a 'rescue effect' likely to be operating. Parameter and demographic stochasticity were built into the model so that they could affect the amount of uncertainty around the answers to the questions stated above. The demographic part of the model was most sensitive to the juvenile mortality parameter me and the adult mortality parameter Mic The 'time to invasion' predicted by the combined model was most sensitive to the distance of the target island, the tendency of migrants to swim straight, and the female fertility parameter Bn₀. In the modelled experiment on Tia Island raising both adult and pre-adult mortality significantly reduced the population. Raising pre-adult mortality was more effective at reducing the population than raising adult mortality. Keeping Tia Island rat free by repeated eradication was predicted to be unfeasible because invasion was too frequent. The predictions of the model became less affected by the uncertainty built into the model as mortality rates increased. The modelled experiments on the Boat Group revealed that Kundy Island was the main sources of invasion for the whole group, with Rat Island acting as a secondary source. Stopping the invasion of these two islands greatly reduced the chances of invasion for the other islands in the group. There was no indication of islands being used as stepping-stones. The predictions of the model were too imprecise to be useful to managers. The rescue effect occurred on the islets Ka and Ra. Kb showed no effect and Ca showed a large effect but in the opposite direction than predicted by the rescue effect.