COS 68-9
Modeling stochastic within-host pathogen growth to predict host growth and life history tradeoffs

Wednesday, August 12, 2015: 10:50 AM
326, Baltimore Convention Center
Arietta Fleming-Davies, Ecology & Evolution, University of Chicago
David James Páez, Ecology and Evolution, University of Chicago, Chicago, IL
Greg Dwyer, Department of Ecology and Evolution, University of Chicago, Chicago, IL

Pathogen exposure often affects host growth, as hosts must allocate resources to immune response that might otherwise be used for growth.  These effects can also influence host population dynamics. Prior work on gypsy moth (Lymantria dispar) larvae suggests that their growth responds nonlinearly to baculovirus exposure.  While larvae exhibit decreased pupal size at high virus doses, there is also a surprising increase in size at low doses, relative to controls.  We sought to understand the mechanism by which this nonlinear growth response occurs, by collecting experimental growth trajectory data for infected larvae and then fitting a within-host model.  Specifically, we asked whether the nonlinear growth effect could arise simply from stochastic within-host growth of virus and host cells.  We infected 550 fourth-instar larvae with three doses of baculovirus (300, 1000 or 6000 infectious particles), or a control (water only).  Larvae were from 12 full-sib families, in order to account for host genetic background effects.  After virus exposure, each larva was raised individually in the laboratory and weighed daily until death or pupation.  Using the daily weights and infection status, we fit a stochastic within-host model, modeling populations of host cells, immune cells, free virus, and protein-coated virus particles (occlusion bodies). 


Our results suggest that pathogen exposure has a high initial cost to host growth rate, but that infected hosts are able to catch up to controls or even overcompensate, given sufficient time to maturity.  Growth during the instar (developmental stage) at which virus exposure occurred was lower for exposed survivors than for controls (Male growth rates: 0.19 control versus 0.17 for exposed survivors; Female growth rates: 0.23 control versus 0.18 exposed survivors; all growth rates are slopes from regressions of log (size in g) versus age in days).  However, in the next instar, differences between exposed and control individuals disappeared for males, and female exposed larvae actually grew more quickly than controls. We modeled within-host growth of virus and host cells as a partially observed Markov process, with parameters fit to daily weights data (R package pomp).  Preliminary modeling results suggest that nonlinear effects of virus exposure on size at maturity can result from stochastic within-host pathogen growth, and do not require an additional mechanism.  Considering within-host pathogen growth as a stochastic process could help explain a range of pathogen effects on the life history traits of their hosts, as well as host tradeoffs such as resistance versus fecundity.