Puumala Orthohantavirus Infection Does Not Affect the Trapping Success of Its Reservoir Host

Abstract

Pathogens might affect behavior of infected reservoir hosts and hence their trappability, which could bias population estimates of pathogen prevalence. In this study, we used snap-trapping data on Puumala orthohantavirus (PUUV)-infected (n = 1619) and noninfected (n = 6940) bank voles (Myodes glareolus) from five vole cycles, normally representing increase, peak, and decline phase, to evaluate if infection status affected trapping success. If PUUV infection, as previously suggested, increases activity and/or mobility, we would expect a higher proportion of infected than noninfected specimens in the first trapping night. However, the proportion of PUUV-infected voles did not differ across the three trapping nights. We conclude that PUUV infection did not affect trapping success, confirming snap trapping as an appropriate trapping method for studies on PUUV prevalence and likely other orthohantaviruses.

Introduction

Pathogens might affect behavior-like activity of infected reservoir hosts (Escutenaire et al. 2002, Lopes et al. 2016). Infection with fatal pathogens might result in reduced activity (e.g., movement rates, extent, timing, duration, and frequency of movements) of diseased animals (Friend 2006), whereas a symptomatic but nonfatal pathogen might stimulate activity/mobility (Webster 1994). In contrast, asymptomatic pathogens should per definition not affect the host.

Orthohantavirus infections in rodents have been generally reported as asymptomatic. However, Puumala orhtohantavirus (PUUV) infection affects survival and maturation (Tersago et al. 2012) in its reservoir host the bank vole (Myodes glareolus) and might cause odor-induced behavioral changes (Hughes et al. 2014). In addition, PUUV-infected juvenile and subadult male bank voles show longer movement distance than uninfected ones (Escutenaire et al. 2002).

Knowledge on behavioral response of a reservoir host to an infection is important, as host behavior may have consequences for the appropriateness of trapping methods that aim to gain comparable estimates of pathogen prevalence in animal populations. In this study, we tested if trapping success (number of trapped specimens per trapping night standardized for number of traps) is higher for PUUV-infected voles than for noninfected bank voles. If the hypothesis is confirmed, interpretation of PUUV prevalence data based on snap trapping should be done with caution since pathogen prevalence would be overestimated.

Materials and Methods

To test our hypothesis, we used previously published data on PUUV-infected and noninfected bank voles snap-trapped in spring and autumn 1979–1986 and spring and autumn 2003–2013 starting in autumn 2003 (Niklasson et al. 1995, Khalil et al. 2016). The data spans five complete vole cycles (1980–1982, 1983–1985, 2003–2005, 2006–2008, and 2009–2012), four being 3-year cycles with increase, peak, and decline phase, whereas the last was a 4-year cycle that additionally included a pronounced low phase. See Supplementary Box S1 for definition of cycle phases.

We tested our hypothesis on the whole data set (1619 PUUV-infected and 6940 noninfected bank voles) and on a subset only including susceptible (not protected by maternal antibodies; body mass ≥14.4 g) bank voles (1400 PUUV-infected and 5005 noninfected bank voles). Bank voles were trapped for three consecutive nights during each trapping session (spring and autumn each year; see Hörnfeldt 1994 for details on trapping methods).

In total, the data comprised 105 cumulated trap-nights ([3 years × 1 season × 3 trapping nights] + [16 years × 2 seasons × 3 trapping nights] = 105). We used a generalized linear model with binomial error structure to test whether prevalence differed among trapping nights. The response variable was PUUV prevalence (number of PUUV-infected bank voles divided by the total number of bank voles) weighted for sample size (total number of infected and uninfected bank voles), while controlling for season, phase of population cycle, and cycle. We pretested for potential interaction among trapping night and the three variables season, phase, and cycle. Neither term was significant and we excluded those interactions from the model.

Trapping of animals was approved by the Swedish Environmental Protection Agency (latest permission: NV-412-4009-10) and the Animal Ethics Committee in Umeå (latest permissions: Dnr A 61-11), and all applicable institutional and national guidelines for use of animals were followed.

Results

Trapping success was highest in the first trapping night and lowest in the last night for all phases (Supplementary Fig. S1). The proportion of PUUV-infected bank voles trapped the first, second, and third trapping night, respectively, did not differ from that of PUUV-noninfected bank voles while phase and season were significant (Table 1; see also Supplementary Fig. S1).

Table 1.

Summary of Best Generalized Linear Models Predicting Puumala Orthohantavirus Prevalence (Number of Infected Bank Voles Divided by Total Number of Bank Voles) in Bank Voles Considering Season (Spring and Autumn), Phase of Vole Cycle (Increase, Peak, Decline, and Low) and Trapping Night (1–3) for Whole Data Set (1979–1986 and 2003–2013, n = 105 Cumulated Trap-Nights) and for Subset (Including Same Time Period, But Only Voles Being Susceptible, i.e., Weight ≥14.4 g)

Predictors	Whole data set			Subset
	OR	CI	P	OR	CI	p
Intercept	0.43	0.37–0.51	<0.001	0.43	0.37–0.51	<0.001
Season: autumn	0.35	0.31–0.40	<0.001	0.35	0.31–0.40	<0.001
Phase: increase	0.95	0.79–1.15	0.626	0.95	0.78–1.14	0.579
Phase: peak	1.36	1.16–1.60	<0.001	1.36	1.16–1.60	<0.001
Phase: low	0.43	0.18–1.00	0.051	0.36	0.11–1.19	0.094
Trapping night: 2	1.07	0.94–1.22	0.324	1.07	0.94–1.22	0.320
Trapping night: 3	1.04	0.90–1.20	0.562	1.04	0.90–1.20	0.569

Significant p values are shown in bold.

CI, confidence interval; OR, odds ratio.

Discussion

Based on our results, we refute the hypothesis that PUUV-infected bank voles show higher trapping success than noninfected bank voles. The probability to enter a snap trap (assuming equal attractiveness of traps) likely increases with increased rodent activity (e.g., distance moved per time unit) and overall distance moved. Both these behaviors can, isolated or in synergy, be driven by weather conditions, food availability, social interactions (including competition), and/or space availability. Since these factors are density dependent, activity and movement-related behavior varies among the phases of vole cycles.

We would, therefore, expect movement and activity to vary among low, increase, peak, and decline phase. Such behavior should be visible in higher trapping success of PUUV-infected than noninfected bank voles (sensu Escutenaire et al. 2002). However, this was not observed in our study, which indicates that at least short distance movement, the most common movement in bank voles, is not affected by PUUV infection. The identified importance of season and cycle phase for PUUV prevalence likely reflects the number of susceptible bank voles (juveniles vs. adults) in different seasons and phases (see also Khalil et al. 2016). Overall, age structure did, however, not affect our results (cf. whole data set vs. subset in Table 1).

When live trapping voles, trapping frequency (number of times a specimen is trapped in the same trap) is highest for short and lowest for long distance movement (Krebs and Myers 1974). Movement over short distances likely reflects foraging activity, whereas in bank voles, movement over longer distances for dispersal is more common in mating males and in subadults trying to establish a territory.

Using live-trapping, in bank voles, the distance moved by PUUV-infected specimens is larger in infected juveniles and subadults than in noninfected ones (Escutenaire et al. 2002). This result does not contradict our finding of equal trapping success of PUUV-infected and noninfected bank voles. In our study, movements performed by voles before being snap-trapped likely (sensu Krebs and Myers 1974) reflect short distance (foraging) movements, and overall (disregarding age and sex), PUUV-infected and noninfected bank voles in Escutenaire et al. (2002) moved equal distances.

Wild house mice (Mus musculus domesticus) injected with lipopolysaccharide, a bacterial product that results in immune-challenged specimens, show reduced social connectivity (Lopes et al. 2016). Infection with Hantaan orthohantavirus causes a fatal infection in Mus musculus (Wichmann et al. 2002), whereas in Peromyscus maculatus, Sin Nombre orthohantavirus causes some pathological effects (Netski et al. 1999). Whether behavioral changes caused by pathogens are reflected in changed trapping success or not, obviously are species and pathogen dependent.

Our result of equal trapping success of PUUV-infected and noninfected bank voles is of high relevance for studies on PUUV prevalence. We show that snap trapping generates comparable estimates of PUUV prevalence in bank voles, confirming snap trapping as appropriate trapping method for studies on prevalence of PUUV and potentially also other orthohantavirus. The latter has, however, to be investigated considering, for example, (1) fatal infection of Hantaan orthohantavirus in house mouse (Wichmann et al. 2002) and (2) the involvement of different reservoir host species in the occurrence of other orthohantaviruses (e.g., Dobrava–Belgrade orthohantavirus with Apodemus spp. as reservoir hosts).

Conclusion

Snap trapping generates comparable estimates of PUUV prevalence in bank voles.