New Zealand Journal of Ecology (2007) 31(1): 47- 59

Estimating density of ship rats in New Zealand forests by capture- mark-recapture trapping

Research Article
Deborah J. Wilson 1,*
Murray G. Efford 1,3
Samantha J. Brown 2
John F. Williamson 2
Gary J. McElrea 1
  1. Landcare Research, Private Bag 930, Dunedin 9054, New Zealand
  2. Landcare Research, Private Bag 052, Palmerston North 4442, New Zealand
  3. Current address: Department of Zoology, University of Otago, PO Box 56, Dunedin 9054, New Zealand
*  Corresponding author

We developed a capture-mark-recapture protocol for measuring the population density (D) of ship rats (Rattus rattus) in forest. Either mesh cage traps or Elliott box traps were set at each of six sites (48 traps per site for 5 nights) in the Orongorongo Valley on two occasions in autumn 2003. Cage traps only were set at three sites in autumn 2004. Rats were caught much more readily in cage traps than in Elliott traps and none were recaptured in Elliott traps. Additional food, bedding and trap covers reduced mortality and interference with traps. To estimate density we fitted a spatial detection model; this method avoids the need to estimate effective trapping area. Estimates were based on both a model assuming equal capture probability (0) and a model incorporating temporal and individual variation (th). Our target for precision was CV() ≤ 20%, but when data were pooled from multiple sites with cage traps, CV(th) was ~30%. Estimated density of rats (th) was 5 ha-1in 2003 and 9 ha-1 in 2004; these estimates did not differ significantly. The overall capture index in 2004 was 3 rats per 00 corrected trap-nights on snap-trap lines set after live trapping. House mice were caught in both types of live trap, but at rates high enough for density estimation only where Elliott traps were used.

Field estimates of detection functions for rats captured with cage traps allowed us to simulate the performance of alternative trapping systems. We predict that a 64-trap layout at three sites with five trapping occasions would yield acceptable precision of th (20–23%) at the observed rat densities. Our use of th was conservative; slightly higher precision may be achieved by assuming constant trappability ( 0), and future work may justify this assumption.