Table 1.

Description of statistical approaches used

Analytical approach	n	Outcome variable	Fixed effects	Random effect(s)
Linear regression model	192	Duration of the heartworm transmission season (months)	Year (2000–2015) Urban zone (1–4) Latitude	NA
Binomial generalized linear mixed model	315a	Infection (yes/no)	Year (2001–2016; no heartworm data were collected in 2006 and 2007) Age class (pup (6–12 months), juvenile, adult) Sex Urban zone (1–4) Proportion of adults tested each year (as an offset)	Site Animal ID
Binomial generalized linear mixed modelb	146	Infection (yes/no)	Year Age class Resident status (resident vs. transient)c Proportion low developed urban land in home range Proportion medium developed urban land in home range Proportion mosquito habitat in home range Proportion of adults tested each year (as an offset) Age class * proportion low developed Age class * proportion medium developed Age class * proportion mosquito habitat	Site

^aSixteen of the coyotes were captured more than once

^bFour models were run using this model structure and composition: (1) for residents and transients using MCP; (2) for residents only using MCP; (3) for residents and transients using a-LoCoH; and (4) for residents only using a-LoCoH

^cVariable was included only when both resident and transient coyotes were analyzed