Thermal acclimation to warmer temperatures can protect host populations from both further heat stress and the potential invasion of pathogens

Abstract

Phenotypic plasticity in response to shifts in temperature, known as thermal acclimation, is an essential component of the ability of a species to cope with environmental change. Not only does this process potentially improve an individual's thermal tolerance, it will also act simultaneously on various fitness related traits that determine whether a population increases or decreases in size. In light of global change, thermal acclimation therefore has consequences for population persistence that extend beyond simply coping with heat stress. This particularly important when we consider the additional threat of parasitism associated with global change, as the ability of a pathogen to invade a host population depends on both its capacity to proliferate within a host and spread between hosts, and thus the supply of new susceptible hosts in a population. Here, we use the host Daphnia magna and its bacterial pathogen Pasteuria ramosa to investigate how thermal acclimation may impact various aspects of host and pathogen performance at the scale of both an individual and the population. We independently test the effect of maternal thermal acclimation and direct thermal acclimation on host thermal tolerance, measured as knockdown times, as well as host fecundity and lifespan, and pathogen infection success and spore production. We find that direct thermal acclimation enhances host thermal tolerance and intrinsic rates of population growth, despite a decline observed for host fecundity and lifespan. Pathogens, on the other hand, faired consistently worse at warmer temperatures at the within-host scale, and also in their potential to invade a host population. Our results suggest that hosts could benefit more from warming than their pathogens, but highlight that considering both within- and between-host thermal performance, including thermal tolerance and fitness traits, is needed to fully appreciate how increasing thermal variability will impact host and pathogen populations.

Acclimation to warmer temperatures can protect host populations from both further heat stress and the potential invasion of pathogens

“…All analyses (Hector et al., 2024) were conducted in R (v. 3.6.2; R Development Core Team, http://www.r-project.com). For all traits, maternal acclimation temperature (2 levels: 20 or 25°C), focal acclimation temperature (2 levels: 20 or 25°C), pathogen treatment (4 levels: pathogen genotype C1, C14 and C20, or uninfected controls) and their interactions were fitted as fixed effects and analysed via an analysis of variance (ANOVA Type III; car package: Fox & Weisberg, 2019).…”

Can pharmaceutical pollution alter the spread of infectious disease? A case study using fluoxetine

“…For transmission, direct estimates of spore loads at host death ( ω ) were available for each infected individual and we estimated the rate at which infections end owing to host death ( D ) as the inverse of host lifespan (1/days). The environmental transmission rate ( β ) was estimated using the numbers of infected and uninfected individuals from each treatment, where the probability of remaining uninfected ( P ) depends on the density of pathogen spores ( Z , 25 000 per 20 ml) and the length of the infection period ( t , 4 days), such that P = e − β Zt (following [ 54 , 58 ]). For R 0 , we estimated the per capita mortality rate ( μ ) as the inverse of host lifespan (1/days) for the unexposed (susceptible) hosts, and then host birth rate ( b ) as the sum of the intrinsic rate of increase ( r , estimated directly, see above) and μ for the unexposed hosts ( b = r + μ , [ 50 , 54 , 58 ]).…”

“…For all derived traits we used the Stan modelling language [ 59 ] as implanted via the CmdStanR package v. 0.5.2 [ 60 ]) in R, to calculate Bayesian posterior distribution estimates for each metric in turn (see also [ 54 , 58 ]). We used the default sampling settings of Stan and semi-informative priors that follow the appropriate distributions for each trait.…”

“…For transmission, direct estimates of spore loads at host death (ω) were available for each infected individual and we estimated the rate at which infections end owing to host death (D) as the inverse of host lifespan (1/days). The environmental transmission rate (β) was estimated using the numbers of infected and uninfected individuals from each treatment, where the probability of remaining uninfected (P) depends on the density of pathogen spores (Z, 25 000 per 20 ml) and the length of the infection period (t, 4 days), such that P = e −βZt (following [54,58]).…”

“…For R 0 , we estimated the per capita mortality rate (μ) as the inverse of host lifespan (1/days) for the unexposed (susceptible) hosts, and then host birth rate (b) as the sum of the intrinsic rate of increase (r, estimated directly, see above) and μ for the unexposed hosts (b = r + μ, [50,54,58]). For all derived traits we used the Stan modelling language [59] as implanted via the CmdStanR package v. 0.5.2 [60]) in R, to calculate Bayesian posterior distribution estimates for each metric in turn (see also [54,58]). We used the default sampling settings of Stan and semi-informative priors that follow the appropriate distributions for each trait.…”