Table 3.

Zero-altered gamma model identifying predictors for the weight of the kidney perirenal fat of the polecats. The model has two parts, one (a) using a gamma generalised linear model to analyse the non-zero positive continuous data and a second part (b) using a logistic regression to describe the probability of kidney fat being observed. Results are presented for the most parsimonious model identified after model selection. In the initial model, we included sex, age (in months), abundance of T. acutum, abundance of S. nasicola, month of sampling and snout-vent length as fixed factors, while including the year of sampling and major landscape unit of origin as random effects (random intercepts). We only included two-way interactions.

(a) Generalised linear model coefficients (gamma with log link)
Coefficients	Estimate	s.e.	z value	p value
(Intercept)	−1.609	0.561	−2.870	0.004
S. nasicola abundance	0.005	0.003	1.662	0.097
Sex–Male	−0.211	0.105	−2.004	0.045
Month of sampling	−0.015	0.009	−1.748	0.080
Snout-vent length	0.009	0.001	6.494	<0.001
(b) Zero hurdle model coefficients (binomial with log link)
Coefficients	Estimate	s.e.	z value	p value
(Intercept)	6.702	1.915	3.501	<0.001
T. acutum abundance	−0.032	0.014	−2.188	0.029
S. nasicola abundance	−0.077	0.024	−3.215	0.001
Sex–Male	1.367	0.401	3.412	<0.001
Snout-vent length	−0.022	0.005	−4.350	<0.001
Random effects	Variance
Year	0.023
Major landscape unit	0.017
Residual	0.430
marginal R2: 0.155
conditional R2: 0.227
Dispersion estimate: 0.430