Table 2. Generalized linear model analysis of the effect of treatment on the number of sporozoites produced per oocyst and midgut analyzed independently for each experimental temperature.

	20°C (n = 116)			24°C (n = 118)			28°C (n = 42)
factors	Wald X2	d.f.	p	Wald X2	d.f.	p	Wald X2	d.f.	p
sporozoites/oocyst
intercept	2581.88	1	<0.0001	5628.47	1	<0.0001	420.03	1	<0.0001
treatment	9.98	2	0.007	25.86	2	<0.0001	11.45	2	0.003
replicate	0.78	2	0.678	11.13	2	0.004	2.39	2	0.303
centred oocyst intensity	9.79	1	0.002	1.53	1	0.216	12.76	1	<0.0001
treatment x centred oocyst intensity	-	-	-	-	-	-	11.96	2	0.003
sporozoites/midgut
intercept	11912.35	1	<0.0001	8610.63	1	<0.0001	1581.66	1	<0.0001
treatment	2.89	2	0.236	28.50	2	<0.0001	15.14	2	0.001
replicate	0.55	2	0.759	10.45	2	0.005	7.70	2	0.021
centred oocyst intensity	97.52	1	<0.0001	1497.95	1	<0.0001	0.15	1	0.699
treatment x centred oocyst intensity	6.64	2	0.036	-	-	-	20.96	2	<0.0001

Omnibus tests confirmed that each fitted model was significantly different from its null model (sporozoites/oocyst: 20°C - likelihood ratio X²_{1, 7} = 122.09, p < 0.0001; 24°C - likelihood ratio X²_{1, 5} = 309.95, p < 0.0001; 28°C - likelihood ratio X²_{1, 7} = 25.42, p < 0.0001; sporozoites/midgut: 20°C - likelihood ratio X²_{1, 7} = 122.09, p < 0.0001; 24°C - likelihood ratio X²_1,7 = 315.16, p < 0.0001; 28°C - likelihood ratio X²_{1, 7} = 39.98, p < 0.0001). Goodness of fit was assessed by evaluating potential overdispersion through model deviance scores and model residuals. Sporozoite/oocyst data were transformed and fit to normal distributions (20°C - deviance value/d.f. = 1.57; 24°C - deviance value/d.f. = 4.92; 28°C - deviance value/d.f. = 1.13). Sporozoite/midgut: 20°C - gamma distribution, deviance value/d.f. = 1.57; 24°C - transformed data fit to a normal distribution, deviance value/d.f. = 4.92; 28°C - gamma distribution, deviance value/d.f. = 1.13.