An emerging viral pathogen truncates population age structure in a European amphibian and may reduce population viability

Infectious diseases can alter the demography of their host populations, reducing their viability even in the absence of mass mortality. Amphibians are the most threatened group of vertebrates globally, and emerging infectious diseases play a large role in their continued population declines. Viruses belonging to the genus Ranavirus are responsible for one of the deadliest and most widespread of these diseases. To date, no work has used individual level data to investigate how ranaviruses affect population demographic structure. We used skeletochronology and morphology to evaluate the impact of ranaviruses on the age structure of populations of the European common frog (Rana temporaria) in the UK. We compared ecologically similar populations that differed most notably in their historical presence or absence of ranavirosis (the acute syndrome caused by ranavirus infection). Our results suggest that ranavirosis may truncate the age structure of R. temporaria populations. One potential explanation for such a shift might be increased adult mortality and subsequent shifts in the life history of younger age classes that increase reproductive output earlier in life. Additionally, we constructed population projection models which indicated that such increased adult mortality could heighten the vulnerability of frog populations to stochastic environmental challenges.


Introduction Statistical Analyses
Body size by age and age at sexual maturity. 145 We conducted all statistical modelling in R (R Core Team, 2014). We used linear mixed effects 146 regression (lmer) models, implemented in the package lme4 (Bates et al., 2015), fitted with a 147 Gaussian error structure, and a stepwise simplification procedure to investigate the impact of 148 population ranavirosis history on the body size (SVL) of R. temporaria. Age, ranaviral disease 149 history of the source population (a two level binary factor) and their interaction were fitted as fixed 150 effects. We controlled for variation in SVL between sampled populations by the inclusion of 151 population of origin as a random effect, applied to the intercepts (Table 1). Since male and female 152 frogs grow at different rates (Gibbons, 1983;Ryser, 1996;Miaud, Guyétant & Elmberg, 1999), 153 the datasets of each sex were analysed separately.
154 A separate lmer model was fitted to investigate the impact of ranavirosis history status of the source 155 population on age at sexual maturity. In the full model, age at maturity was fitted as the response 156 variable, ranaviral disease history of source population as a fixed effect and source population as 157 a random effect. As female R. temporaria mature later than males (Gibbons 1983, Miaud et al. 158 1999, the datasets of each sex were analysed separately. Influence of ranavirus on population age structure 159 The impact of disease status on the age structure of R. temporaria populations was investigated 160 using a Bayesian ordinal mixed effects model in the package MCMCglmm (Hadfield, 2010). We 161 fitted age class as an ordinal response variable (9 discrete classes, ages 2 -10 years), disease status 162 of the source population as the fixed effect, and source population as a random effect. We used 163 uninformative priors for both the random effect (G) and residual variance (R) structures, but fixed 165 2018). The model was run for a total of 600,000 iterations with a burn-in period of 100,000 166 iterations and a thinning rate of 500, giving a final sample of 1,000 draws from the posterior 167 distributions. We assessed model convergence using the Gelman-Rubin (G-R) statistic calculated 168 from three independent chains initiated with overdispersed starting values. All G-R values were 169 <1.05, indicating convergence. Mean probability of membership and associated 95% credible 170 intervals for each age class were calculated from the linear predictor, for each of the two disease 171 history groups. Age structure plots (Fig. 2) suggest that observed changes were similar for both 172 sexes, so the dataset was not split by sex.

Population Matrix Modelling
173 To investigate how changes in population age structure, as well as scenarios that can bring about 174 such changes, can impact the dynamics and stability of R. temporaria populations, we constructed 175 population matrix models. Comprehensive methodologies of our matrix modelling can be found 176 in our supplementary methods section (Methods S1). However, in brief, we created hypothetical 177 R. temporaria populations of 150 sexually mature female animals and projected these populations 178 20 years into the future based on two matrices which represented potential vital rates at ranavirosis-179 positive R. temporaria population (increasing annual mortality in each adult age class) and a 180 disease-free R. temporaria population (uniform adult mortality). We populated these matrices 277 Concurrently, we found that a significantly higher number of 2 -5 year old frogs were captured at 278 disease-free populations than at populations with a history of ranavirosis. Life history theory 279 predicts that the first compensatory response to high adult mortality should come in the survival 280 rates of lower age classes (Stearns 1992). We lack any data on the immature age classes present at 281 our study populations and the snapshot nature of our study means we cannot draw inference on 288 we hypothesised that R. temporaria originating from populations where ranavirosis causes 289 increased adult mortality would reach sexual maturity at an earlier age than frogs from disease-290 free populations. We also hypothesised that a trade-off in the allocation of resources to early 291 reproduction, away from growth, would cause frogs from ranavirosis-positive populations to attain 292 a lower body size per age than those from disease-free populations. However, we found that a 297 The age at which R. temporaria mature is intrinsically linked to attaining a minimum body length 298 needed to successfully reproduce (Ryser 1996, Miaud et al. 1999). The age at which this length is 299 attained has been shown to be heavily influenced by several environmental factors such as photo-300 period and altitude (Miaud et al. 1999). All frogs in our study were found to mature at either 2, 3 301 or 4 years of age and this is consistent with the findings of other similar studies on R. temporaria 302 (Gibbons 1983, Ryser 1996, Miaud et al. 1999). In fact, no previous study has found male or 303 female R. temporaria to reach maturity younger than 2 years of age and post-sexual maturity there 304 is little detectable trade-off between growth and fecundity in response to sub-prime environments 305 (Lardner & Loman, 2003). This evidence suggests that the life history strategy of R. temporaria 306 may already be optimised to generate maximum reproductive fitness in light of other 307 environmental factors and that scope for further plasticity in traits such as age at sexual maturity 308 and subsequent growth rate in response to diseases may be minimal.
309 Given the apparent lack of compensatory change in the onset of sexual maturity an alternative 310 explanation for the fact that we encountered 2-and 3-year-old breeding frogs at ranavirosis- Manuscript to be reviewed

Figure 3(on next page)
Differences in posterior probabilities of belonging to a given age class between population groups of varying disease history.
The mean difference in the posterior probabilities of belonging to a given age class by population ranavirosis history. Values > 0 indicate that an age class is more likely to be observed in a ranavirosis-positive population and < 0 a disease-free population. An age with 95% (2.5% -97.5%) credible intervals that do not span zero suggests that influence of disease history on that age class is significantly supported by our model. This is the case for all classes other than age 6 which although found to be observed more often in disease-free populations has credible intervals spanning 0.  Manuscript to be reviewed  Manuscript to be reviewed