Table 1.

Results of the three generalised linear mixed models testing the effects of different predictors (seasonal temperature change, habitat type, host role in the life cycle, and host taxon) on seasonal change in trematode infection levels

Predictor	Estimate	Standard error	95% confidence intervals	t-value
Prevalence ($\mathrm{\Delta }P$) in first intermediate hosts (N = 278)
(Intercept)	− 1.203	2.924	− 6.705, 4.312	0.411
Seasonal temperature change ($\mathrm{\Delta }T$)	0.635	0.282	0.098, 1.168	2.255
Habitat type: pond	− 2.281	1.801	− 5.736, 1.113	1.267
Habitat type: river	− 0.832	2.133	− 4.883, 3.203	0.390
Prevalence ($\mathrm{\Delta }P$) in second intermediate and definitive hosts (N = 231)
(Intercept)	3.347	22.597	− 40.492, 44.876	0.148
Seasonal temperature change ($\mathrm{\Delta }T$)	− 0.945	0.771	− 2.397, 0.475	1.226
Habitat type: pond	12.625	8.165	− 2.729, 28.064	1.546
Habitat type: river	5.860	4.667	− 2.855, 14.673	1.255
Host life cycle role: second intermediate	2.110	5.610	− 11.606, 15.032	0.376
Host taxon: crustacean	− 4.987	22.947	− 47.436, 37.454	0.217
Host taxon: fish	− 2.309	20.350	− 40.056, 37.457	0.113
Host taxon: leech	8.047	25.165	− 37.811, 54.899	0.319
Host taxon: mollusc	0.513	20.601	− 37.546, 38.868	0.025
Abundance ($\mathrm{\Delta }A$) in second intermediate and definitive hosts (N = 251)
(Intercept)	6.806	25.402	− 42.371, 56.477	0.268
Seasonal temperature change ($\mathrm{\Delta }T$)	0.424	0.816	− 1.217, 1.928	0.519
Habitat type: pond	− 2.370	9.610	− 20.491, 16.130	0.247
Habitat type: river	5.310	4.521	− 3.250, 14.241	1.174
Host life cycle role: second intermediate	− 12.973	8.845	− 31.933, 9.079	1.467
Host taxon: crustacean	2.787	25.191	− 44.671, 48.851	0.111
Host taxon: fish	− 18.749	22.238	− 60.781, 24.608	0.843
Host taxon: leech	− 12.553	26.250	− 63.013, 38.892	0.478
Host taxon: mollusc	1.409	24.510	− 44.456, 46.250	0.057

For categorical predictors, the reference level included in the intercept is ‘lake’ for Habitat type, and ‘amphibian’ for Host taxon; alternate analyses with other reference levels yielded similar results

Significant effects (95% confidence intervals not overlapping zero) are shown in bold