Prediction model for sequence variation in the glycoprotein gene of infectious hematopoietic necrosis virus in California, USA

The influence of spatio-temporal factors on genetic variation of infectious hematopoietic necrosis virus (IHNV) is an active area of research. Using host-isolate pairs collected from 1966 to 2004 for 237 IHNV isolates from California and southern Oregon, we examined genetic variation of the mid-G gene of IHNV that could be quantified across times and geographic locations. Information hypothesized to influence genetic variation was environmental and/or fish host demographic factors, viz. location (inland or coastal), year of isolation, habitat (river, lake, or hatchery), the agent factors of subgroup (LI or LII) and serotype (1, 2, or 3), and the host factors of fish age (juvenile or adult), sex (male or female), and season of spawning run (spring, fall, late fall, winter). Inverse distance weighting (IDW) was performed to create isopleth maps of the genetic distances of each subgroup. IDW maps showed that more genetic divergence was predicted for isolates found inland (for both subgroups: LI and LII) than for coastal watershed isolates. A mixed-effect beta regression with a logit link function was used to seek associations between genetic distances and hypothesized explanatory factors. The model that best described genetic distance contained the factors of location, year of isolation, and the interaction between location and year. Our model suggests that genetic distance was greater for isolates collected from 1966 to 2004 at inland locations than for isolates found in coastal watersheds during the same years. The agreement between the IDW and beta regression analyses quantitatively supports our conclusion that, during this time period, more genetic variation existed within subgroup LII in inland watersheds than within coastal LI isolates.