Use of exposure history to identify patterns of immunity to pneumonia in bighorn sheep (Ovis canadensis)

Abstract

Individual host immune responses to infectious agents drive epidemic behavior and are therefore central to understanding and controlling infectious diseases. However, important features of individual immune responses, such as the strength and longevity of immunity, can be challenging to characterize, particularly if they cannot be replicated or controlled in captive environments. Our research on bighorn sheep pneumonia elucidates how individual bighorn sheep respond to infection with pneumonia pathogens by examining the relationship between exposure history and survival in situ. Pneumonia is a poorly understood disease that has impeded the recovery of bighorn sheep (Ovis canadensis) following their widespread extirpation in the 1900s. We analyzed the effects of pneumonia-exposure history on survival of 388 radio-collared adults and 753 ewe-lamb pairs. Results from Cox proportional hazards models suggested that surviving ewes develop protective immunity after exposure, but previous exposure in ewes does not protect their lambs during pneumonia outbreaks. Paradoxically, multiple exposures of ewes to pneumonia were associated with diminished survival of their offspring during pneumonia outbreaks. Although there was support for waning and boosting immunity in ewes, models with consistent immunizing exposure were similarly supported. Translocated animals that had not previously been exposed were more likely to die of pneumonia than residents. These results suggest that pneumonia in bighorn sheep can lead to aging populations of immune adults with limited recruitment. Recovery is unlikely to be enhanced by translocating naïve healthy animals into or near populations infected with pneumonia pathogens.

Figure 1. Pneumonia dynamics in Hells Canyon.

A. Population estimates and pneumonia dynamics of monitored bighorn sheep populations in Hells Canyon (Idaho, Oregon, and Washington, U.S.A), 1994–2010. Colored lines represent the three most intensively monitored populations: Redbird (RB), Wenaha (WE), and Black Butte (BB). Opaque circles represent years with lamb pneumonia outbreaks (detected or suspected; see Cassirer et al. [20]), open circles represent years when no lamb pneumonia outbreak was detected or suspected. Grey lines represent population size estimates for all other populations monitored in Hells Canyon. The population estimates for Black Butte include the removal of 72 bighorn sheep during the 1995 epidemic . B. Estimated population growth rate, r (natural log of population size in year t divided by population size in year t-1) during the year in which pneumonia invaded the population (invasion = 0) and in post invasion years with pneumonia mortalities in adults or lambs (invasion = 1). The invasion year r for Black Butte incorporates the removal of 72 bighorn sheep . C. Years in which adult pneumonia mortality was detected in the three populations depicted in A: Redbird (RB), Wenaha (WE), and Black Butte (BB).

Figure 2. Study area and pneumonia history calculation.

A. Study area: we report data from 388 radio-collared adult bighorn sheep and 753 ewe-lamb pairs within 12 of these 16 populations in Hells Canyon (WA = Washington, ID = Idaho, OR = Oregon; see Fig. S1 for populations' names and pneumonia histories). B. How individual pneumonia histories were constructed. The bighorn sheep in panel B was collared (II) at age 6 and died (III) at age 11 (in the middle of the biological year). Age was estimated at capture (II) or by incisor cementum analysis after death (III). Based on its population's pneumonia history (red indicates years with pneumonia mortality, green indicates years when pneumonia was not detected; see Fig. S1), this animal experienced 8 pneumonia exposures (Count = 8). The time since last exposure (Lag) was 0 when it died.

Figure 3. Cox proportional hazards model-estimated relative risks of mortality of ewes, rams, and lambs after accounting for differences in sex and age.

Risk of dying of pneumonia relative to naïve individuals given: an indicator for any previous pneumonia exposure (Indicator), the number of years since the last exposure event (Lag of 1, 2 or > = 3), increasing number of exposure events (Count), and translocation status (Trans). Bottom panel: model-estimated risk of dying for lambs during outbreaks of pneumonia given ewe covariates. Coefficient estimates from univariate models are in grey and coefficient estimates from multivariate models are in black. Risk values are drawn from proportional hazards model coefficient estimates, with the point estimate denoted by the box or vertical line, and horizontal lines extending to the 95% confidence limits.

Figure 4. Relationship between cumulative past exposure and risk of dying.

a. Ewe risk of dying from pneumonia as a function of the number (count) of previous exposures (relative to unexposed ewes). A relative risk of 1 on the y-axis represents no effect of count on the risk of dying of pneumonia. The shaded area represents the 95% confidence bounds on the hazard ratio associated with continuously increasing number of previous exposures (in a model fit on an age-scale that included a fixed effect for translocation status). The error bars bound the 95% confidence range for the uniquely estimated hazard ratios associated with each value of count. The line connects the estimated median risk of dying relative to the risk for previously unexposed ewes. Count values above six are grouped within the count = 6 category. b. Lambs' risk of dying in a pneumonia epidemic given the number of times the lamb's mother was exposed. The shaded area represents the 95% confidence bounds on the hazard ratio associated with continuously increasing number of previous exposures (count) of the ewe (in a model fit on lamb age-scale that included a fixed effect for ewe's translocation status). The error bars bound the 95% confidence range for the uniquely estimated hazard ratios associated with each number of exposures for ewes. The line connects the estimated median risk of lamb mortality. c. A conceptual diagram illustrating how individual frailty may drive the apparent relationship in part a. Frailties decrease as number of exposures (or age) increase because the weak (at the upper tail of the distribution) die first, and cannot be observed in later years, leaving an increasing proportion of strong individuals.