Introduction

Recent introductions of elk (Cervus elaphus Linnaeus 1758) in Mexico (e.g., ^{McKinney & Villalobos, 2014}) necessitate better understanding of elk-livestock disease interactions, both to increase likelihood of successful introductions and to minimize conflicts with Mexico’s cattle industry. Productivity (i.e., production and survival of calves) of elk populations is declining in several areas of the USA (^{Noyes et al., 2002}; ^{Rearden, 2005}; ^{Piasecke, 2006}). Disease is a potential contributing factor to decreased elk productivity, particularly where they are sympatric with cattle, but aside from malnutrition (^{Cook et al., 2004}; ^{Bender & Cook, 2005}) and brucellosis (^{Cheville et al., 1998}; ^{Thorne, 2001}) has received little evaluation. Elk are sympatric with cattle throughout most of their range, and are potential hosts for a variety of diseases that may affect elk and cattle (^{Thorne et al., 2002}). Of these, the most important are diseases that affect the reproductive output of each species, as these have the greatest potential to impact recreational or economic returns from either population.

Several diseases can affect pregnancy, cause abortion, and influence calf survival in elk and cattle, particularly brucellosis, leptospirosis, infectious bovine rhinotracheitis (IBR), bovine viral diarrhea (BVD), and neosporosis (^{Thorne et al., 2002}), and these diseases are included in most bovine abortion profiles (i.e., exposure assessments). Each of these can interfere with reproductive function, primarily by causing abortions (^{Kahrs, 1981}; ^{Van Campen et al., 2001}; ^{Thorne et al., 2002}; ^{Baszler, 2003}; ^{Cook et al., 2004}), although some can also cause fetal malformation, stillbirth, and nonviable neonates, among many other manifestations (e.g., ^{Van Campen et al., 2001}). Through these impacts, these diseases can potentially compromise individual reproduction, and thus decrease reproductive output of the population. Even diseases that are usually rare in wild elk populations and that occur in only specific local areas, such as brucellosis and leptospirosis, can be a significant concern because of potential impacts to cattle (^{Bender & Hall, 1996}; ^{Thorne et al., 2002}; ^{Peel et al., 2010}; ^{Milián-Suazo et al., 2016}). However, the cattle industry is advantaged in that vaccines are available for most reproductive diseases (^{Castro, 2001}; ^{Thorne, 2001}; ^{Leighton & Kuiken, 2001}; ^{Van Campen et al., 2001}; ^{Segura-Correa et al., 2016}), although vaccines may always be effective (^{Xue et al., 2011}). In contrast, vaccination of free-ranging wildlife is largely impossible, so shared diseases are likely to disproportionately affect elk. Increasing public demand for “natural” (i.e., unvaccinated) beef, however, is increasing the number of vaccine-free cattle operations worldwide.

Our goal was to test whether exposure as indicated by positive serology to causative agents of diseases associated with reproductive failure in cattle affected productivity of elk, specifically pregnancy and survival of calves to weaning. Positive serology indicates presence of antibodies to an agent, which includes previous exposure or past infection, not necessarily active infection (^{Calisher & Taylor, 1993}). However, high seroprevalence or longitudinal persistence in positive serology can indicate disease presence (^{Calisher & Taylor, 1993}; ^{Bender et al., 2003}), and thus serological surveys are commonly used to evaluate the potential presence of, and risk factors associated with, disease in populations (e.g., ^{Bender et al., 2003}; ^{Milián-Suazo et al., 2016}; ^{Segura-Correa et al., 2016}). Therefore, we assessed seroprevalence to causative agents of bovine reproductive diseases in multiple elk populations throughout the western USA. We compared serological prevalence with previously published data, and modeled exposure effects on pregnancy and preweaning calf survival of elk. We also identify disease risks for both elk and cattle associated with introductions or translocations of elk in Mexico.

Materials and methods

Study populations. Our study populations covered a variety of locations throughout the United States (Table 1). Chaco Culture National Historic Park (CC) is located in northwestern New Mexico (approximately 36º 00’ N, 108º 00’ W). This site is a desert grass and shrubland with scattered pinyon (Pinus edulis)-juniper (Juniperus spp.) woodlands. Fort Riley is a 403 km2 military training facility located in the Flint Hills of northeastern Kansas (approximately 39º 06’ N, 96º 48’ W). The area is primarily rolling tallgrass prairie of big bluestem (Andropogon gerardii) and other tallgrass natives with scattered wooded areas along riparian corridors and lowlands, interspersed with agricultural fields and wildlife plantings. Rocky Mountain National Park (RMNP) covers 1,076 km2 in the Rocky Mountain Front Range of northcentral Colorado (approximately 40º 23’ N, 105º 38’ W). The site consists primarily of montane forest interspersed with grassland, shrublands, and open tundra occur at higher elevations. Lincoln National Forest (LNF) is located in the Sacramento Mountains of southcentral New Mexico (approximately 32º 51’ N, 105º 44’ W). This study area was primarily semiarid woodland and montane forest interspersed with small grassy meadows at high elevations. The Forks study site was located in the coastal hills of western Washington state (approximately 47º 54’ N, 124º 35’ W). Land-use in this area is primarily industrial tree farms of Douglas-fir (Pseudotsuga menziesii) and western hemlock (Tsuga heterophylla). The Valles Caldera National Preserve was located in the Jemez Mountains of northcentral New Mexico (approximately 35º 55’ N, 106º 31’ W). This study area consists of high elevation mesic montane grasslands and mixed conifer forest.

Capture. We captured cow elk ≥ 1.5 years old in autumn (November) and late-winter (March–April). Elk were darted from a Bell 206B Jet Ranger helicopter (or from vehicles along roads in RMNP) using carfentanil citrate and xylazine hydrochloride (3.6 mg carfentanil + 100 mg xylazine/elk) as sedatives, and blindfolded to reduce stress and prevent eye injury (^{Kreeger, 1996}; ^{Bender, 2015}). We also treated each elk with penicillin, vitamin E/selenium, vitamin B, and an 8-way Clostridium bacterin to reduce physiological stress and trauma of capture. Captured elk were aged to yearling or adult using presence or absence of deciduous teeth (^{Quimby & Gaab, 1957}). Immobilants were antagonized with 300 mg naltrexone (half intravenous and half subcutaneous) and 800 mg tolazoline (delivered intravenously) (^{Kreeger, 1996}; ^{Bender, 2015}).

Disease screening. We obtained whole blood samples for the bovine abortion profile and pregnancy testing from immobilized elk through jugular venal puncture. Whole blood samples were transferred to serology tubes, which were spun (4,500 rpm; 8–10 min) to separate serum shortly after collection. Serum samples were then frozen until analysis.

We determined pregnancy status from pregnancy-specific placental protein B (PSPB) (BioTracking, Moscow, Idaho, USA). Elk from which autumn PSPB results were uncertain were corroborated using serum progesterone (Colorado State University Endocrinology Lab, Fort Collins, Colorado, USA). Progesterone levels of ≥ 1.0 ng/ml and ≤ 93% binding of elk antiserum to PSPB (^{Noyes et al., 1997}; ^{Bender et al., 2002}) indicated pregnancy. We determined lactation status for cows by checking the udder for milk, which indicated survival of a calf to within ≤ 3–11 days (^{Bender et al., 2002}). We could not determine lactation status from spring captures because most calves are weaned by this time (^{Johnson, 1951}).

A bovine abortion profile was performed on serum samples from individual elk to detect exposure to causative agents of profiled diseases (New Mexico Department of Agriculture Veterinary Diagnostic Laboratory, Albuquerque, New Mexico, USA; Washington Animal Disease Diagnostic Lab, Pullman, Washington, USA). Serology included the card test for brucellosis (Brucella abortus; ^{Alton et al., 1988}), virus neutralization for BVD (bovine viral diarrhea virus [BVDV]) and IBR (bovine herpesvirus 1 [BHV-1] (^{Carbrey et al. 1971}), the microscopic agglutination test for leptospirosis (including Leptospira interrogans serovars pomona, hardjo, grippo-typhosa, ictero-hemorrhagiae, bratislava, canicola) (^{Gouchenour et al., 1958}), and an enzyme-linked immunosorbent assay (ELISA) for neosporosis (Neospora caninum; Shares et al., 2001). Serology was considered negative at <1:4 for BVDV and BHV-1, <1:100 for Leptospira serovars, no agglutination for B. abortus, and ELISA values <30% for N. caninum.

Data analysis. We compared seroprevalence to causative agents among populations using Fisher’s exact tests (^{Zar, 1996}). We used hierarchical logistic regression to model the dichotomous outcomes of pregnancy and lactation (i.e., pregnant/not pregnant, lactating/not lactating) at the individual level as a function of population and whether each cow elk was exposed to a particular agent or not (^{Hosmer & Lemeshow, 1989}; ^{Kuss, 2004}). If the serological result from either autumn or spring during pregnancy was positive, we classed these as positive exposure for pregnancy modeling. For lactation modeling, we used only serological results from the autumn after the calf was born, i.e., when the cow was lactating. For analyses of lactation, we excluded yearling elk because they are never lactating (^{Raedeke et al., 2002}).

Results

Among populations, we tested 177–194 cow elk for pregnancy and exposure to disease causative agents, and 107–122 cow elk for lactation status and exposure to causative agents. Seroprevalence to causative agents varied among populations (Fisher’s exact P < 0.01) with the exception of B. abortus, for which we did not detect exposure in any population (Table 2).

Pregnancy averaged 0.84 (SE = 0.04; range 0.73–0.96) and lactation averaged 0.50 (SE = 0.06; range = 0.40–0.67) among populations. Pregnancy varied by population in all contrasts (P < 0.016) but not by exposure to any agent (P > 0.213) (Table 3). Proportion of cow elk lactating in autumn did not vary among populations (P > 0.247) nor by exposure to any agent (P > 0.281) (Table 3). One cow that was definitively pregnant when tested in autumn was found to be not pregnant when subsequently recaptured and retested again in late winter. She was negative for all screened causative agents.

Discussion

The primary risk factors associated with transmission of most bovine reproductive diseases among elk include high elk densities and co-occurrence of elk and cattle (^{Thorne, 2001}; ^{Thorne et al., 2002}). Seroprevalence of screened agents in our populations was reflective of the range of exposure seen in elk throughout North America (Table 2), highlighting the potential for exposure of elk to reproductive diseases of cattle (and vice versa) wherever elk and cattle co-occur. Despite high seroprevalence to certain agents, however, exposure was not related to pregnancy or preweaning calf survival in our study populations (Table 3), even though several of these populations showed relatively low pregnancy rates and calf survival (Table 1; ^{Piasecke, 2006}). Although our index of calf survival (lactation status) assessed only preweaning survival (^{Bender et al., 2002}) and not survival of a calf to recruitment, once a calf has survived to weaning it has passed the peak of juvenile mortality and will most likely survive to reproductive age (^{Guinness et al., 1978}; ^{Taber et al., 1982}; ^{Clutton-Brock et al., 1988}). These results, as well as the negative exposure result for the 1 cow that lost its fetus, indicates that past or current exposure to common reproductive diseases of cattle likely has negligible effects on population productivity of elk. The exception to this is brucellosis, which can cause significant declines in elk productivity where endemic in North America (^{Cheville et al., 1998}; ^{Thorne, 2001}).

Exposure to cattle reproductive disease causative agents is relatively widespread in Mexico, both in dairy (^{Milián-Suazo et al., 2016}) and beef (^{Segura-Correa et al., 2016}) cattle. For example, ^{Milián-Suazo et al. (2016)} recently surveyed multiple dairy operations throughout Mexico and found seroprevalence of 4–15% (depending upon test used), 37%, 79%, and 73% to agents for brucellosis, neosporosis, BVD, and IBR, respectively. ^{Segura-Correa et al. (2016)} also recently surveyed beef operations in Tamaulipas and found seroprevalence of 48% and 68% for agents of BVD and IBR, respectively. Certain risk factors for exposure were common to both studies, and included herd size and introduction of new cattle to the herd. High seroprevalence in elk to several of these disease causative agents suggests that introduction or co-occurrence of elk may be an additional risk factor for unvaccinated herds in Mexico.

For example, IBR was the only disease for which the causative agent (BHV-1) showed exposure in all elk populations (Table 2). While potentially affecting a variety of systems in cattle, IBR is primarily of concern because of the potential to cause abortions regardless of the severity of disease or whether the disease is present in respiratory or ocular form (^{Fraser & Mays, 1986}). Because wild ruminants frequently do not display clinical signs of IBR infection, the disease is primarily considered a concern only for sympatric cattle (^{Castro, 2001}). Our data supports this conclusion; despite a wide range of exposure (4–43%), hierarchical logistical analysis indicated that probability of pregnancy and calf survival to weaning were both unrelated to exposure to BHV-1. However, because exposure to BHV-1 was seen in all tested elk populations, it has the highest likelihood of the agents we surveyed of being present in elk and potentially transferred to cattle. While seroprevalence to BHV-1 is high and widespread in cattle in Mexico (^{Milián-Suazo et al., 2016}, ^{Segura-Correa et al., 2016}), it appears less common in areas near the central and western USA border (^{Milián-Suazo et al., 2016}) where introductions of elk are most likely.

Similarly, reproductive diseases can be transmitted to elk from cattle. Of greatest concern in Mexico would be brucellosis, as it is widely distributed in Mexico (^{Peel et al., 2010}; ^{Milián-Suazo et al., 2016}), has demonstrated negative impacts on elk productivity, and once infected elk can serve as a reservoir for the disease, increasing the difficulty of erradication programs (^{Cheville et al., 1998}; ^{Thorne, 2001}). Presence of brucellosis in cattle thus can compromise the success of elk introductions or translocations. Similarly, elk that are translocated from areas where brucellosis is endemic in Mexico to areas that are free of brucellosis should be tested for the presence of B. abortus, and translocations should not proceed if brucellosis is present in elk. Such movements may complicate ongoing efforts to eradicate brucellosis in Mexico (^{Milián-Suazo et al., 2016}).

Last, as previously noted, positive serology indicates antibody presence and thus exposure to a disease causative agent, not necessarily active infection (^{Calisher & Taylor, 1993}). Consequently, the lack of effect of screened bovine reproductive diseases on elk productivity that we observed may have been due to past exposure or past infection, and not current infection. However, longitudinal persistence in positive serology is indicative of disease presence (^{Calisher & Taylor, 1993}; ^{Bender et al., 2003}), and several of our study populations (i.e., CC, LNF) have shown long-term persistence of positive serology. Moreover, high seroprevalence where vaccination is not present suggests that the actual prevalence of the disease is high (^{Milián-Suazo et al., 2016}). For example, ^{Morales et al. (2001)} found that cattle herds in Mexico with higher seroprevalence for neosporosis had a greater number of abortions; seroprevalence was 72% for herds with >13% abortions, but 36% for herds with <12% abortions. Thus, while a lack of observed effect of exposure to bovine reproductive diseases may have been due to lack of active infection, high longitudinal seroprevalence in many populations suggests that it is likely that the causative agents in the bovine abortion screen did not impact pregnancy or calf survival in free-ranging elk. Again, the exception to this would be brucellosis, which was not present in our study populations as it is endemic only to the greater Yellowstone area in North America (^{Cheville et al., 1998}).