Prevalence of the pathogenic chytrid fungus, Batrachochytrium dendrobatidis, in an endangered population of northern leopard frogs, Rana pipiens

Table 3 Parameter estimates and statistical significance of the factors that explain the prevalence of Bd in R. pipiens.

(A) Maximum Likelihood Estimates
Model	Model No.	df	dev	AIC
pcr ~ method + season + year + stage + season:stage	1	308	165.77	189.77
pcr ~ method + season + year + stage	2	311	172.95	190.95
pcr ~ season + year + stage	3	312	191.10	207.10
pcr ~ method + year + stage	4	313	180.54	194.54
pcr ~ method + season + stage	5	314	181.75	193.75
pcr ~ method + season + year	6	313	183.51	197.51
(B) Likelihood ratio tests
Effect	Comparison	Δ df	Δ dev	p
season:stage	1 vs. 2	3	7.18	0.066
method (swab vs. other methods)	2 vs. 3	1	18.15	<0.001
season (spring, summer, fall)	2 vs. 4	2	7.59	0.022
year (2003, 2005, 2006, 2007)	2 vs. 5	3	8.80	0.032
stage (yoy, juvenile, adult)	2 vs. 6	2	10.56	0.005
(C) Parameter estimates
Factor	Estimate	s.e.	z	p
Intercept	-4.159	1.241	-3.351	<0.001
method: swab	-2.028	0.488	-4.157	<0.001
season: summer	-1.888	1.125	-1.679	0.093
season: fall	0.549	0.707	0.777	0.437
year: 2005	2.099	0.900	2.333	0.020
year: 2006	2.528	1.015	2.490	0.013
year: 2007	2.704	1.059	2.554	0.011
stage: juvenile	1.705	0.816	2.089	0.037
stage: adult	1.976	0.610	3.239	0.001

We used GLM with binomial errors to model the prevalence of Bd in R. pipiens (n = 320 captures) as a function of four factors: tissue-sampling method, season, year, and stage class. (A) Shown are the residual degrees of freedom (df), the residual deviance (dev) and the Akaike information criterion (AIC) for six different models. (B) Shown are the likelihood ratio tests between models, the change in the degrees of freedom (Δ df), change in the residual deviance (Δ dev), and the associated p-value (p). (C) Shown are the estimates of the parameters, the standard errors (s.e.), the associated z-scores (z) and p-values (p).

ISSN: 1472-6785