Population size assessments

2.6.1

Mark-recapture operations

To estimate how much the size of each population was reduced following density reduction, mark- recapture operations were carried out at the beginning (‘pre-reduction mark-recapture event’) and the end (‘post-reduction mark-recapture event’) of each study. Each of these events involved live- capture cage trapping possums for four to five consecutive nights, and recording which individuals were recaptured and which were new individuals. Fifty traps were used (resulting in a density of three traps per hectare), which were distributed evenly throughout each of the sites, to ensure that all individuals had access to a trap and therefore did not differ in their probability of capture (Otis et al., 1978). Trap types, locations and lures were all consistent between both mark-recapture events at each site. Density assessments at Site 3 were undertaken within the 15-hectare study area, not the entire 60-hectare fragment. The home ranges of the study animals were all within this 15-hectare study area and therefore it was considered that assessing only this area would provide the most accurate estimate of density.

The behaviour of individuals in the post-reduction mark-recapture events may have been influenced by their trapping experiences in the pre-reduction mark-recapture events. For example, individuals may have become ‘trap-shy’ after the first event and therefore not caught in the second event. However, research has shown that possums are not intrinsically un-trappable and that previous capture is unlikely to have a long-lasting effect on the trappability of individuals (Morgan et al., 2007). As such, it was considered that analysing each trapping event independently did not bias these estimates.

The resulting data were analysed in the computer program ‘Program MARK’ (Version 6.1), which gave an estimate of population sizes (Cooch and White, 2011; Lettink and Armstrong, 2003; Pryde, 2003). This program is the most comprehensive and widely used software for the analysis of

marked individuals (Cooch and White, 2011). Generally, the proportion of marked animals will be high when population size is low and the proportion will be low when population size is large (Begon et al., 1996). Due to the short timeframe for each mark-recapture operation (less than one week), it was assumed that the population was closed, remaining generally constant in both size and composition, and not subject to animals leaving and entering the population through births, deaths, emigration and immigration (Cooch and White, 2011; Lettink and Armstrong, 2003; Otis et al., 1978; Pryde, 2003).

Four different pre-defined Closed Population Estimation models were run for each mark- recapture event, to account for potential biases in estimates due to variations in the behaviour of individual possums (

Table 2.3). The four models for each event were then ranked by Program MARK using Akaike Information Criteria (AIC), whereby the weights of all models in the set sum to one and the model with the highest Akaike Weight is considered to have the best fit for the available data (Burnham and Anderson, 2002). Population estimates were derived through model averaging, to account for potential relevant information in models ranked below the best model (Anderson, 2008; Burnham and Anderson, 2002). Models were constructed for the total population, each age class (adult and juvenile) and each sex (male and female). Density reduction was subsequently estimated by calculating the percent difference between the pre-reduction population estimate and the post- reduction estimate. This work was peer-reviewed by a biometrician, Dr Darryl MacKenzie of Proteus Wildlife Research Consultants. Detailed results of the population size and density reduction estimates are presented in ‘Appendix B: Population size and density reduction estimates’.

There is another capture-recapture program that estimates the density of possums directly from live-capture trapping (Efford, 2004). This program was developed to account for ‘edge effects’, whereby animals move on and off a trapping grid due to random spacing of traps and the home range size of individuals (Efford et al., 2005). Program MARK was chosen over this program, however, as the traps in this study were generally placed in grids and the entire study area was trapped; therefore, edge effects likely did not occur. Furthermore, spatial variation was of less interest in this study, with the change after density reduction being the main focus. It was also more important to account for behavioural changes in the trapping probabilities of individuals, which Program MARK addressed.

Table 2.3 List of models used to estimate population size in Program MARK (Otis et al., 1978; White, 2010)

Model Description

M (o) Equal capture probabilities among individuals (null model)

M (b) Behavioural response after initial capture (capture probabilities vary with trap _{experience – i.e., “trap-happy” or “trap-shy” individuals)} M (t) Time-specific variation in trapping probabilities (capture probabilities vary through time, _{e.g., due to weather changes)} M (tb) Both time-specific variation and behavioural response after initial capture

2.6.2

WaxTag® surveys

WaxTag® surveys were also carried out at Site 1 as another method to estimate changes in possum abundance following density reduction (NPCA, 2008). WaxTags® were stapled to trees along survey lines to allow the identification of species, such as possums, through bite marks on the wax. A Bite Mark Index (BMI) was then calculated based on the percentage of WaxTags® that received possum interference (NPCA, 2008). Monitoring was conducted at the beginning of the experiment (pre- reduction) and then immediately following density reduction (post-reduction). However, regardless of the removal of 79 individuals (approximately 50% of the population), this index was not substantially different between events (pre-reduction BMI: 86%; post-reduction BMI: 79%). This was most likely due to contagion, whereby individual possums move along the whole survey line biting a large proportion of WaxTags®, causing an overestimation of population densities (Thomas et al., 2007). This technique was therefore not deemed a useful indicator of density change and was not employed at the following two study sites.