Maintenance of host DNA integrity in field-preserved mosquito (Diptera: Culicidae) blood meals for identification by DNA barcoding

Table 1 The best-fit logistic regression model predicting amplification success for four host DNA preservation methods

	Wald χ ²	df	P
Intercept	20.4	1	< 0.001
Preservation method	24.8	3	< 0.001
Post-feeding time	28.5	1	< 0.001
Storage time	10.9	1	< 0.001
Preservation method × post-feeding time	10.9	3	0.012
Preservation method × storage time	5.9	3	0.12

The dependent variable in this analysis was amplification success, coded so that 0 = no amplification, and 1 = amplification. The inclusion of the selected variables important in predicting PCR amplification success was based on comparisons of the Akaike information criterion. The significance of individual variables in predicting amplification success was determined by Wald Chi-squared tests at the 95 % confidence interval. Model fit was assessed with the Hosmer-Lemeshow goodness-of-fit test (χ ² = 6.81, df = 8, P = 0.557), and likelihood ratio test (χ ² = 256.1, df = 5, P < 0.001). The best-fit model was identified as logit(amplification success) = 2.608–2.146 cold–2.297 desiccation + 2.208 filter paper–0.012 storage–0.089 post-feeding–0.085 cold*storage–0.004 desiccation*storage + 0.006 filter paper*storage + 0.044 cold*post-feeding + 0.041 desiccation*post-feeding–0.04 filter paper*post-feeding. In the model, preservation method is a categorical variable, dummy-coded with ethanol as the reference level

ISSN: 1756-3305