Virus shedding kinetics and unconventional virulence tradeoffs

Abstract

Tradeoff theory, which postulates that virulence provides both transmission costs and benefits for pathogens, has become widely adopted by the scientific community. Although theoretical literature exploring virulence-tradeoffs is vast, empirical studies validating various assumptions still remain sparse. In particular, truncation of transmission duration as a cost of virulence has been difficult to quantify with robust controlled in vivo studies. We sought to fill this knowledge gap by investigating how transmission rate and duration were associated with virulence for infectious hematopoietic necrosis virus (IHNV) in rainbow trout (Oncorhynchus mykiss). Using host mortality to quantify virulence and viral shedding to quantify transmission, we found that IHNV did not conform to classical tradeoff theory. More virulent genotypes of the virus were found to have longer transmission durations due to lower recovery rates of infected hosts, but the relationship was not saturating as assumed by tradeoff theory. Furthermore, the impact of host mortality on limiting transmission duration was minimal and greatly outweighed by recovery. Transmission rate differences between high and low virulence genotypes were also small and inconsistent. Ultimately, more virulent genotypes were found to have the overall fitness advantage, and there was no apparent constraint on the evolution of increased virulence for IHNV. However, using a mathematical model parameterized with experimental data, it was found that host culling resurrected the virulence tradeoff and provided low virulence genotypes with the advantage. Human-induced or natural culling, as well as host population fragmentation, may be some of the mechanisms by which virulence diversity is maintained in nature. This work highlights the importance of considering non-classical virulence tradeoffs.

Fig 1. Cumulative survival of fish exposed to IHNV.

Panels show Kaplan-Meier survival curves for fish exposed to IHNV genotypes HV (red), LV (blue), MER95 (green), LR80 (purple), mixed infections (black, 1:1 ratio of two genotypes in experiment), or a mock negative control (orange) in experiments 1–3 (A-C, respectively). Mixed infections were not assessed in experiment 3. Error bars show 95% confidence interval based on cumulative hazard, derived from Kaplan-Meier analyses. All treatments began with 20 fish, with the exception of mock control which had 8 fish. Mortalities were recorded daily for the entire course of the experiment (30–35 days).

Fig 2. Number of fish shedding virus.

Panels show number of remaining live fish shedding detectable quantities of IHNV RNA for genotypes HV (red), LV (blue), MER95 (green), LR80 (purple), in single (solid lines) or mixed infections (dashed lines), in experiments 1–3 (A-C, respectively). Mixed infections were not assessed in experiment 3. There were 20 total fish per treatment group. Symbols indicate time points where water samples were taken for shedding detection, measured by qPCR. Fish were excluded from the day after death onwards.

Fig 3. Virus shedding intensity.

Panels show the mean amount of IHNV [log10(virus RNA copies/ml H₂O)] shed per day for genotypes HV (red squares), LV (blue circles), LR80 (purple triangles), and MER95 (green diamonds), in experiments 1 (A), 2 (B), and 3 (C). Points represent days samples were taken to quantify viral shedding in single (closed symbols and solid lines) and mixed infections (open symbols and dotted lines). Error bars show +/- 1 standard error of the mean. Only live fish shedding detectable virus into the water were included in the figure. A missing point indicates no virus shedding was detected by qPCR for all fish in a given treatment. Points not connected by a line indicate virus dropped below detection at sampling time points in-between. Points without standard error bars indicated fewer than 2 fish were positive for virus. Number of fish included in the mean values for each data point are the number of live fish shedding at the same time point in Fig 2.

Fig 4. Total virus shed.

Panels show mean total virus shed [log10(total virus RNA copies/ml H₂O+1)] for the peak (days 0–4, filed bars) and post peak periods (days 5 onwards, open bars) for genotypes HV (red), LV (blue), LR80 (purple), and MER95 (green), in experiments 1 (A), 2 (B), and 3 (C). Single infections are shown on the panel edges and mixed infections together in the panel center, to denote that viral quantities for the two genotypes within an experiment in mixed infections came from the same fish. Error bars show +/- 1 standard error of mean. Shedding from all fish is included in the mean up to one day after fish death. For experiment 2, days 7 and 16 were excluded from total because only single infection samples were taken. Letters above bars (a and b) denote significant differences (P<0.5, see text) between bars within panels and infection periods. Statistical comparisons were not made across panels (experiments) or infection periods (peak vs. post-peak).

Fig 5. Association between virulence and fitness.

Panels show association between (A) transmission rate (peak viral load shed [log10(virus RNA copies/ml H₂0)]), (B) recovery rate (1/duration of shedding (days)) and (C) total transmission potential (total virus shed [log10(virus RNA copies/ml H₂0)]), with virulence (1/day fish died). Each data point represents an individual fish infected with a given IHNV genotype in a given experiment (see in figure legend). Only fish exposed to a single virus genotype, which shed detectable virus, are shown. Mixed infections are excluded due to potential confounding effects of genotype interactions on virulence. The solid line represents the best fit ((A) Y = 5.607+/-0.057, (B) Y = 0.187 +/- 0.014, (C)Y = 5. 758 +/-0.056) to the data +/- 1 standard error (SE) (grey area). The line demonstrates a uniform, non-significant association in all cases. As such, virulence explained none of the variance in the viral fitness parameters on an individual fish basis.

Fig 6. Model output of IHNV transmission at a population level.

Plots show incidence (solid line) and cumulative total (dashed line) number of fish (Log + 1) infected with either the high virulence (HV–red) or low virulence (LV–blue) IHNV genotype, through time (days). Data are derived from an SIRD (susceptible, infectious, recovered, dead) model (see methods Eqs 1–4) parameterized from experimental data (see Table 1). For panels A and B, populations of fish are simultaneously infected with both genotypes HV and LV, and thus both lines represent the same fish population. For panels C and D, populations of fish are singly infected with each genotype separately, and thus each line represents a different fish population. For panels B and D, culling is implemented when mortality reaches 30% of the initial fish population size. Culling involves removal of all fish from the population, regardless of infection status (i.e. set S, I, R, and D to zero). Culling events are denoted by * on the figure. State variables are initially set to S = 200,000, infected HV = 10, infected LV = 10, R = 0, D = 0; to reflect conditions in a raceway on a standard trout farm (Table 1).