Figure 1.
Locations of spotlight transects within the major pastoral regions of the South Island of New Zealand.
Figure 2.
Median and 95% credible intervals (CIs) for parameters in the observation and Gompertz models: cats (filled circles); rabbits (hollow circles).
For each species the parameters in the observation model are β1, β2, which respectively describe the linear and quadratic effects of transect length on detection, and γ, the observation error. Parameters in the Gompertz model comprise seasonal intercepts, ab, anb (for the breeding (b) and non-breeding seasons (nb)), a coefficient for density dependence, b, and the process error, σ. For cats, the Gompertz model also included seasonal coefficients for the log of rabbit abundance in the previous sampling season, cb, cnb. Parameters with 95% CIs overlapping the horizontal lines are non-significant.
Figure 3.
Predicted detection probability for cats and rabbits in relation to transect length.
Detection probability for cats (a) increased significantly with increasing transect length up to 25 km. Detection probability for rabbits (b) was not significantly associated with transect length. The black line is the posterior mean across transects and seasons.
Figure 4.
Estimated density (individuals per transect kilometre) of cats and rabbits in the breeding and non-breeding seasons for cats (b and nb respectively).
Estimates of cat (a) and rabbit (b) abundance were higher during the non-breeding seasons. The thin grey lines are the mean density estimates for each spotlight transect (n = 66), while the thick black line is the mean across transects.
Figure 5.
Predicted log(cat abundance) in relation to log(rabbit abundance) in the previous season.
Graphs show (a) the non-breeding season (March–August) and (b) the breeding season for cats (September–February). t = time.