Table 1.

ΔWAIC values for all 10 candidate models, for both virus prevalence and seroprevalence in three different host taxon sets. Models that fit the data most satisfactorily (with a ΔWAIC ≤ 2) are italicized. Models are ranked based on overall performance, starting with models that tended to perform best in describing virus prevalence and seroprevalences across all three taxon sets. Generally, models that included both random effects (phylogeny and species), as well as a full or reduced set of fixed-effect, explanatory variables (age, eco-region, season and year) performed best in explaining the variation across all taxon sets, for both virus prevalence and seroprevalence.

model description					response ΔWAIC for prevalence in
predictors			random effects		all species		Anatidae		Scolopacidae
all	reduced	none	species	phylogeny	virus	serology	virusc	serologyc	virusa,b	serologyb
X			X	X	3.7	0.6	1.3	0.0	0.2	1.7
X				X	2.5	0.7	0.4	0.3	1.2	2.7
	X		X	X	1.5	0.0	3.3	6.9	0.2	0.0
	X			X	0.0	1.1	2.3	6.3	1.1	1.4
X			X		9.6	0.9	0.7	0.2	0.1	1.4
	X		X		7.2	0.7	3.4	6.8	0.0	0.5
		X	X	X	141.6	459.9	40.8	68.9	90.7	405.2
X					111.4	1529.3	0.0	135.2	32.0	396.3
	X				140.0	1529.1	2.1	149.6	30.1	445.3
		X			335.7	2367.3	44.0	242.4	169.8	830.2

^aThe models did not contain year as an explanatory variable, given poor model convergence when included.

^bThe models did not contain the arid eco-region due to low sample size.

^cThe models did not contain years 2014, 2015 and 2019, or the tropical eco-region due to low sample size.