Bighorn sheep associations: understanding tradeoffs of sociality and implications for disease transmission

Introduction

Understanding the underlying mechanisms of social associations can help address questions in evolutionary, behavioral, and infectious disease ecology (Pinter-Wollman et al., 2014; Janousek et al., 2021). Sociality, defined as consistent associations between individuals, can benefit individuals, populations, or species in a challenging environment (Hinde, 1976; Krebs & Davies, 1997; Krause & Ruxton, 2002). Benefits include greater sharing of information, better body condition, increased survival, and higher mating success (Krebs & Davies, 1997; Krause & Ruxton, 2002). Mechanisms for these fitness benefits include improved access to and defense of a resource (space, food, or offspring), increased collective vigilance for predators (the many eyes hypothesis), and dilution of risk of predation with increasing group size (the dilution effect) (Delm, 1990; Rieucau & Martin, 2008) and/or dilution of risk of disease by gaining collective immunity (Plowright et al., 2013). Sociality, however, can also be costly through competition for limited resources, risk of injuries through interactions with aggressive individuals, and transmission and persistence of parasites and infectious diseases (Anderson et al., 1986; Altizer et al., 2003). A balance between these costs, benefits, and selective pressures in a “landscape of peril” ultimately shapes sociality (Doherty & Ruehle, 2020).

Trade-offs between costs and benefits can operate concurrently and at multiple scales (Hinde, 1976; Sih, Hanser & McHugh, 2009; Van der Wal et al., 2016). At the population level, associations can reflect the connectivity among individuals across the landscape and can influence metapopulation dynamics. At a finer scale, the strengths of associations between individuals can reflect dyad and individual characteristics such as personality, genetic relatedness (which can increase indirect fitness), behavior phenomenon such as homophily (i.e., a tendency to interact with individuals of similar age, sex, body size, or social status; McPherson, Smith-Lovin & Cook, 2001), and other intrinsic factors (Hamilton, 1964; Wolf & Weissing, 2012). Finally, at the finest scale, locations of associations can reflect time- and habitat-specific characteristics and other extrinsic factors that relate to access to resources and survival of the individual (Travis, Slobodchikoff & Keim, 1995; Sterck, Watts & Van Schaik, 1997; Naud et al., 2016).

Although the importance of the finest scale, the spatial context of associations, is often acknowledged for understanding factors that shape interactions and the types of behavior they represent (Trillmich & Wolf, 2008; Jacoby & Freeman, 2016), it remains missing from analyses of many social wildlife species (Pinter-Wollman, Fiore & Theraulaz, 2017; Albery et al., 2021). The few studies that have incorporated spatial data have demonstrated the influence of nest structure on the collective foraging performance of ants (Pinter-Wollman, 2015) and the importance of interactions that occur at the periphery of home ranges, which allow individuals to defend their territory (Giuggioli, Potts & Harris, 2011; Spiegel et al., 2018). Fortunately, these data are readily available through proximity-based metrics derived from biologging techniques such as global positioning system (GPS) devices that allow for simultaneous tracking of multiple individuals. Through these proximity-based metrics, GPS devices allow researchers to disentangle an individual’s social behavior with conspecifics from its spatio-temporal distribution in relation to habitat (Webber & Van der Wal, 2018).

Bighorn sheep (Ovis canadensis), hereafter bighorn, are a model system to examine social behavior because extensive research exists on habitat selection and disease ecology of bighorn. Although considered secure at global and state scales, local extirpation of herds, mostly due to disease, has occurred throughout their range (Flesch et al., 2022). Bighorn, as social mountain ungulates and habitat specialists, face significant evolutionary pressures from three main factors: competition, predation, and disease. Strong dominance hierarchies within bighorn populations are established and reinforced with intra-sexual competition and may determine reproductive success (Geist, 1971). Males compete with other males through intense dominance interactions before the breeding season and aggressive interactions during the breeding season (Hogg, 1987; Shackleton, Shank & Wikeem, 1999). Predation, especially by mountain lions (Puma concolor), accounts for a large portion of bighorn mortality for both male and female bighorn of all age classes (Ross, Jalkotzy & Festa-Bianchet, 1997; Hayes et al., 2000; Rominger et al., 2004; McKinney, Smith & De Vos Jr, 2006). As such, bighorn habitat selection is heavily influenced by variables related to survival, such as distance to escape terrain and canopy cover (DeCesare & Pletscher, 2006). As with most species, multiple diseases can influence bighorn health including gastrointestinal parasites (e.g., Ostertagia spp., Nematodirus spp., Cooperia spp., Becklund & Senger, 1967) and pathogens including Mycoplasma ovipneumoniae, Pasteurellaceae (e.g., Mannheimia haemolytica, Bibersteinia trehalosi, Pasteurella multocida), and Anaplasma ovis (Tibbitts et al., 1992; Besser et al., 2013; Butler et al., 2018). Many of these pathogens have been investigated as the causal agent for respiratory disease, which is transmitted by direct contact and often cited as a significant factor limiting the distribution and abundance of bighorn (Cassirer et al., 2017). Respiratory disease has been implicated in die-offs of up to 90% of exposed bighorn populations (Spraker et al., 1984; Coggins, 1988; Festa-Bianchet, 1988; Besser et al., 2012). Because respiratory disease can have such high mortality rates, disease exposure and disease status of bighorn have been monitored extensively and can provide coarse fitness consequences to sociality. Although considerable research exists on social interactions, habitat selection, and disease ecology of bighorn separately, few studies simultaneously examine the interactions between these selective pressures and few studies address the mechanisms through which bighorn use risky areas where predation may be more likely or the mechanisms by which disease spreads (but see Manlove et al., 2014; Cassirer et al., 2017; Manlove et al., 2017). By studying the associations and the absence of associations between bighorn in the context of multiple selective pressures, we can better understand disease dynamics and conserve wild bighorn populations.

We evaluated associations between male and female bighorn at 3 scales, asking (1) where are the divisions among subpopulations, (2) which intrinsic factors influence frequency of associations between dyads, and (3) when do direct contacts occur and which extrinsic factors influence direct contact structure? We predicted that population structure based on indirect contacts would parallel the structure derived from surveys for diseases transmitted through direct contact (De la Fuente et al., 2007; Ott et al., 2009). For intrinsic factors that influence frequency of direct associations between dyads, we predicted that direct associations would be strongest between genetically related female–female dyads that had high space-use overlap and weakest between unrelated male–female dyads that had low space-use overlap. For timing of contacts, we predicted that more direct contacts would occur when bighorn are most active (i.e., during the day), during the winter for same sex dyads, and during the mating season for male–female dyads. For direct contact locations, we formed multiple hypotheses based on two categories (Table 1): the need for resources and the need for grouping to avoid predation (i.e., survival). For example, we hypothesized that contact locations would be influenced more by survival variables than by resource variables for all dyad types. We predicted that direct contacts would occur more frequently in areas with fewer resources and higher predation risk.

Materials & Methods

Study area

We conducted this study in the Waterton-Glacier International Peace Park and the Blackfeet Indian Reservation (Fig. 1). This area comprises Glacier National Park in Montana, United States (∼400,000 ha), which is managed by the National Park Service, Waterton Lakes National Parks in Alberta, Canada (50,000 ha), which is managed by Parks Canada, and the western edge of the Blackfeet Indian Reservation (780,000 ha), which is managed by the Blackfeet Nation. Glaciations in the Pleistocene formed major topographical features, including many glacial lakes, moraines, and cirques, and left active glaciers (Thornberry, 2004). The Continental Divide of the Rocky Mountains runs north to south through the park, and greatly influences precipitation; elevations range from 960 to 3,171 m, and valleys east of the divide feature extensive open grassy slopes. Conifer forest (46.08%), grass (14.90%), scrub/shrub (14.85%), barren (7.92%), and deciduous forest (7.71%) are dominant land cover types in the overall study area, which we delineated based on a 25 km buffer around all bighorn GPS locations (Crown Managers Partnership, https://www.sciencebase.gov/catalog/item/51102e04e4b048b5cead853b).

Although no formal population estimates of bighorn in the park exist, bighorn are considered a species of management concern in Glacier National Park (MJ Biel, pers. obs., 2023), and historical counts suggest that ∼500 bighorn comprise this putative metapopulation (Flesch et al., 2022). No hunting occurs in the park, but the Blackfeet Nation manages a limited hunt for bighorn outside the park boundary, and a few poaching incidents occurred inside the park boundary during this study in the northern part of Glacier National Park. Mountain lions are likely the primary predator of bighorn adults in this area, but other predators such as grizzly bears (Ursus arctos), black bears (U. americanus), and eagles (Aquila chrysaetos) have been documented near kills in the park (MJ Biel, pers. obs., 2022) and gray wolves (Canis lupus), coyotes (Canis latrans), wolverines (Gulo gulo), are also present (Sawyer & Lindzey, 2002).

Prior research on pathogens of bighorn in the Waterton-Glacier International Peace Park identified two divisions in the bighorn population related to two natural geographic features. Specifically, De la Fuente et al. (2007) identified a division at the St. Mary Lake Valley by investigating exposure to Anaplasma ovis, and Ott et al. (2009) identified a division at the Belly River by investigating Bibersteinia trehalosi (formerly Pasteurella trehalosi) DNA sequences. Both pathogens were previously linked to respiratory disease in bighorn (Jaworski, Hunter & Ward, 1998; Rudolph et al., 2003; Weiser et al., 2003; Ott et al., 2009).

Results

During 2002–2012, we recorded a total of 168,380 GPS locations (n_M = 84,886, n_F = 83,494) from 97 bighorn (54 males, 43 females) (Table S1, Graves et al., 2023). We identified 138,469 possible indirect contacts using the 10 m distance threshold (n_F−F = 47,470, n_M−M = 40,943, n_M−F = 50,056) and 1,259,882 possible indirect contacts using the 25 m distance threshold (n_F−F = 443,426, n_M−M = 372,500, n_M−F = 443,956). We identified 5,324 direct contacts (n_F−F = 2,098, n_M−M = 2,716, n_M−F = 510) between 252 dyads and censored 30 dyads that were within the same group (16 female–female dyads with n_F−F = 870 contacts, 12 male-male dyads with n_M−M = 1,492 contacts, 2 male–female dyads with n_M−F = 2 contacts) as indicated by correlated movement during any part of the year. We analyzed 2,960 direct contacts (n_F−F = 1,228, n_M−M =1,224, n_M−F = 508) between 222 dyads among 558 possible dyads to explore how dyad characteristics influenced direct contact rates. After we removed locations with missing covariate data and censored dyads that had fewer than five direct contacts, we analyzed 2,091 direct contact locations (n_F−F = 956, n_M−M = 895, n_M−F = 240) between 109 dyads (37 female–female, 45 male-male, and 27 male–female dyads) to investigate where contacts occurred.

Where are the divisions among subpopulations?

The indirect contact network structure revealed that this population of bighorn was structured into two groups separated by the St. Mary Lake Valley, a forested valley running east–west, and that the northern group could be split into two subgroups separated by Waterton Lake (Fig. 2). We refer to these three groups (from north to south) as the West Waterton subpopulation, the North Glacier subpopulation, and the South Glacier population.

The West Waterton subpopulation and the North Glacier subpopulation were connected by one female bighorn that made a long-distance movement from west of Waterton Lake to east of Waterton Lake and returned thereafter. We did not record any direct contacts between this female and any other collared bighorn using the shortest distance criterion of 25 m, but we recorded direct contacts with up to three other female bighorn in the east Waterton area using the longer distance criteria of 50 and 100 m. Direct contact network analysis also revealed that there may be another split within the North Glacier subpopulation across the Belly River, a relatively broad open valley.

When we compared indirect network structure with male and female social groups that were previously classified based on spatial use pattern, we found that greater proportions of indirect contacts occurred among opposite sex social groups than same sex social groups (V = 24, P = 0.041, Fig. S3). Specifically, male groups indirectly contacted more female groups than other male groups (V = 0, P = 0.035).

When do contacts occur?

Contacts occurred most frequently during March for female–female dyads, during August for male-male dyads, and during November, December, and January for male–female dyads (Fig. 3). Few contacts occurred between female–female dyads in July and male–female dyads during the summer (June - October). We detected more contacts when more collared bighorn occurred in the general area (β_bighorn,MM = 0.81 ± 0.07, β_bighorn,FF = 0.66 ± 0.06, β_bighorn,MF = 0.69 ± 0.08) and when bighorn moved more slowly (β_speed,MM = −0.14 ± 0.05, β_speed,FF = 0.07 ± 0.03, β_speed,MF = −0.23 ± 0.08; Fig. S4). Contacts also occurred more frequently at night than during crepuscular and daylight hours for female–female and male–female dyads.

Contact locations vs. general habitat use

Female–female, male-male, and male–female dyads contact location ranks were inversely related to habitat use location ranks (female-only dyads: r_s = −0.61, p < 0.0001; male-only dyads: r_s = −0.76, p < 0.0001; male–female dyads: r_s = −0.56, p < 0.0001; Fig. 4 and Fig. S9, Tosa & Graves, 2023). The most pronounced differences between contact locations and habitat locations across dyad types related to survival variables: canopy cover, distance to escape terrain, and ruggedness (Figs. S7 and S10). Differences in female–female contact and habitat use probabilities were also driven by opposite responses to snow water equivalent (SWE): female bighorn generally selected against areas with high SWE, but when they were in areas with high SWE, they were more likely to contact other female bighorn (Table 1, Figs. S7 and S10). Interestingly, male–female dyads were less likely to contact each other close to mineral licks, even though both sexes generally selected to be closer to mineral licks.

Discussion

Our results show that contact analysis is a valuable method for understanding animal behavior and fitness tradeoffs of sociality. With close proximity to conspecifics, bighorn appear to gain an advantage in a challenging environment that they would not have if they were solitary, despite risks of resource competition and disease. Although spatial proximity does not necessarily equate to physical contact between two individuals (Tosa, Schauber & Nielsen, 2015) and rates of disease transmission may vary by individual or dyad characteristics (Manlove et al., 2017), we demonstrate that the distance criteria we used appears to serve as an appropriate surrogate for direct physical contacts that are necessary for disease transmission given the consistency between breaks in the population identified through the indirect contact network and divisions in previous disease status.

At the landscape scale, we demonstrated that mapping indirect and direct contact network structure can reveal information similar to that gained through disease exposure and genetic analyses. Although we had a limited number of collars deployed each year (maximum of 13), we identified 1 distinct barrier (St Mary’s Lake Valley) and 2 other partial barriers (Belly River and Upper Waterton Lake) that naturally separate 3–4 subpopulations of bighorn in this system (Fig. 1). Two of these possible barriers (St. Mary’s Lake Valley and the Belly River) were consistent with divisions identified by other methods that focused on exposure to diseases such as Anaplasma ovis and Bibersteinia trehalosi that require direct contact (De la Fuente et al., 2007; Ott et al., 2009). Moreover, analyses on the genomic structure of this population confirmed limited long-term movement across the St. Mary Lake Valley (Flesch et al., 2020; Graves & Flesch, 2020). As spatial proximity is a necessary component of the mechanisms for genetic mixing and disease transmission, this demonstrates contact analyses reflect the most frequent mixing. Consequently, these three possible barriers that we identified represent important locations for disease management, should respiratory disease infect any section of the population (Manlove et al., 2014).

Connections between the east and west Waterton subpopulations were surprising, given the connections were formed by female–female dyads. This was unexpected because males typically have larger home ranges, have been observed making long distance movements during the rut, and therefore are more likely to contact other bighorn (Hogg, 2000; DeCesare & Pletscher, 2006). We also found that social groups of bighorn were more likely to contact social groups of the opposite sex, which highlights the importance of mating season for disease transmission (Fig. S3). Because contacts between the West Waterton and East Waterton subgroups occurred within 4 days (25 –29 July 2006), this may represent a single exploratory movement by the female from the West Waterton subgroup to the east pre- or post-dispersal. Limited research currently informs the drivers and frequencies of these rare long-distance movements although they have consequences to disease and genetic mixing.

Concerning factors that influenced the associations between dyads of bighorn, female–female dyads with greater space-use overlap during the summer season had the greatest probability of contact (Table 2). Relatedness was not a significant indicator of contact, suggesting that kinship was not necessary for associations between individuals from separate social groups and that sociality may be beneficial for survival of the individual (direct fitness) instead of increased fitness through related family members (indirect fitness) (Hamilton, 1964). This is consistent with bighorn in Alberta, Canada, where kinship appeared to play a limited role in the social system (Festa-Bianchet, 1991). The other factors, homophily of sex and space-use overlap, indicated that bighorn sociality is structured based on sex and proximity during the summer as described by Geist (1971) and others (Festa-Bianchet, 1991; Ruckstuhl, 1998).

Regarding the effect of landscape and vegetation variables, we found that bighorn habitat use was highly influenced by survival variables and less so by resource variables. Similar to other studies, this suggests that predation has strong selective pressures on locations of bighorn and provides evidence that bighorn operate in a “landscape of fear” (Laundré, Hernández & Altendorf, 2001). The disturbance variables that we explored had effects opposite to what we expected on bighorn space use (i.e., bighorn selected for disturbance variables instead of selecting against disturbance variables). This may reflect the placement of trails, roads, and helicopter routes in less steep areas that facilitate movement for both humans and bighorn, that provide visitors with opportunities to view wildlife and to recreate, or poor alignment of the map of helicopter routes with actual routes. Alternatively, the relative effects of these disturbances could be minimal given the levels of disturbance while bighorn were collared compared to survival and resource variables or bighorn may be habituated to anthropogenic disturbances inside this highly protected area.

The predicted probabilities of contact given habitat use varied greatly from the habitat use models and suggest bighorn are responding to the tradeoffs of being social (Fig. 4). For example, contact probabilities given habitat use were lower when nutritional quality of forage, indicated by IRG, was high (Fig. S10), which suggests that the benefits of social behavior are outweighed by competition for valuable resources. On the other hand, when resources were scarce (e.g., low solar radiation index for male-male and male–female dyads or high snow water equivalent for female–female dyads), contact probabilities were higher, which suggests that the benefits of social behavior (i.e., increased thermoregulation efficiency) outweighed the disadvantage of sharing other resources (Arnold, 1990). The benefits of sociality were more pronounced in the predicted probabilities of contact for survival variables than for the resource variables. In areas where bighorn were vulnerable to predation, contacts were more likely. For example, contacts between female–female dyads were more likely to occur where canopy cover was higher, and contacts between all dyad types were more likely to occur when bighorn were farther from escape terrain (Fig. S10). This supports hypotheses for the evolution of sociality based on anti-predator benefits such as the many eyes hypothesis (Lima, 1995; Rieucau & Martin, 2008), whereby animals can increase the probability of detecting a predator, and the dilution of risk hypothesis (Pulliam, 1973; Foster & Treherne, 1981), whereby animals decrease their risk of predation by splitting the risk between the animals in the social group. Alternatively, these areas of low-ranking habitat may coincide with the edges of social groups, where multiple social groups are most likely to overlap.

The distance to mineral licks variable reveals interesting behavior of bighorn (Fig. S6). Although mineral licks were important to both male and female bighorn as indicated through the habitat use analysis, the contact analysis reveals that the two sexes rarely use this resource at the same time (Figs. S7 and S10). This is consistent with observations of other mountain ungulates such as mountain goats (Oreamnos americanus) where males generally use mineral licks earlier in the year followed by females and family groups starting in June (Poole, Bachmann & Teske, 2010). The importance of mineral licks for male and female habitat use and for female–female and male-male contacts identifies valuable areas for potential bighorn population monitoring. There are only a handful of known mineral licks in the Glacier-Waterton International Peace Park (MJ Biel, pers. obs., 2022); therefore, this naturally limited resource creates ideal locations for monitoring populations of bighorn and other mountain ungulates. Mineral licks provide mountain ungulates access to elements including sodium, calcium, potassium, and magnesium that can be necessary to regulate body fluids such as blood (Hebert & Cowan, 1971), mediate the effects of secondary compounds in forage (Ayotte et al., 2006; Ayotte, Parker & Gillingham, 2008), and meet the physiological demands of lactation and other functions critical to the health of populations (Ayotte et al., 2006). As such, many animals make repeated visits to these mineral licks over their lifetime and often travel long distances (Rice, 2010). These areas, however, may also increase their risk of predation (Côté & Beaudoin, 1997; Sarmento & Berger, 2017) and increase rates of pathogen transmission through direct and indirect contact. Fomites, artificial and supplemental feeding sites, and watering holes that aggregate wildlife in concentrated areas can increase disease transmission rates in other systems (Cross et al., 2010; Paull et al., 2012; Sorensen, Van Beest & Brook, 2014).

Much of our analyses support previous findings concerning bighorn social systems and social structure. Analyses of the correlation of movement between bighorn revealed that bighorn behave in a fission–fusion social system. Dyads moved together during winter and spring but split apart during the summer (Fig. S2). We also documented variability in contact rates by dyad type and seasonality in the timing of contacts (Fig. 3). Female–female contacts were highest in March when vegetation starts to green-up and food resources are spatially restricted (Fig. S11). Female-male contacts were highest during the mating season through early winter (November–January), which corresponds to times of the year when movement probabilities of males were greatest (Borg et al., 2017). In contrast, male–female contacts were lowest during the summer when females separate from males to raise offspring (lambs). During the winter, when food resources are scarce, low ambient temperatures may force bighorn to huddle together for thermoregulation. Moreover, our analyses revealed that contacts were most likely when bighorn were moving slowly. This suggests that contacts occur during resting and foraging bouts instead of during large, directed movements in unfamiliar areas (e.g., at a mineral lick vs. movements toward a mineral lick).

Conclusion

Our study assesses social behavior of a wild, native, and large population of bighorn at the population, group, dyad, and individual scales. Using this unique 10-year GPS dataset of both male and female bighorn, we assessed social behavior among three dyad types–male-only, female-only, and male–female dyads–and modeled when, where, and with whom bighorn interact at close distances. Patterns of association were influenced by intrinsic dyad characteristics and by extrinsic characteristics such as the instantaneous rate of green-up, snow water equivalent, terrain ruggedness, and distance to escape terrain, representing resource availability and predation risk: same sex dyads had higher contact frequencies and bighorn in areas with lower resource availability and higher predation risk had higher conditional probabilities of contact.

Although we documented factors that influenced where and when contacts occurred between adult bighorn in this study, the initial purpose of this study was to create a bighorn habitat use map. Due to logistical constraints, all animal captures were conducted with ground-based darting, and it is likely that the bighorn represented in this study are biased toward those that use areas that are accessible to humans. During this study, the number of direct contacts may have been overestimated because we used a distance criterion as a proxy for direct contact. In other words, it is possible that individuals were within 25 m, but did not get close enough for transmission of respiratory disease (Tosa, Schauber & Nielsen, 2015). Given the large bighorn metapopulation and large spatial extent of the study, however, it is more reasonable to assume that the number and locations of direct contacts between bighorn were underestimated. Although we were unable to incorporate disease status of individuals and the effects that disease status might have on habitat use or dominance position (Edmunds et al., 2018) during this study, this study provides a valuable starting point to consider the connections between the selective pressures of disease, predation, and competition. Future studies could investigate changes in bighorn space use due to disease status to explore these relationships further.

Another consideration for this study is that much of the landscape is dynamic in this area. As snowpack, conifer encroachment, vegetation composition, and forage phenology change, the when and where of contacts could likewise change (Berman et al., 2020). This region is also susceptible to wildfires, and climate change models predict that wildfires will become more frequent and more severe (Barbero et al., 2015; Westerling, 2016). Most recently, the Reynolds wildfire at the St. Mary Lake division has significantly changed the vegetation in this area. As such, the divisions that we identified may no longer consist of the same barriers and populations may become more connected. Given the status of this metapopulation as one of only two large metapopulations in Montana, further research could evaluate changes in contact rates, especially across the St. Mary Lake barrier. This could help inform managers of management options should respiratory disease enter one of these subpopulations. Overall, our findings suggest that contact analyses that incorporate information about individuals and space can be a valuable method for understanding fitness tradeoffs of sociality and information on the who, when, and where of bighorn disease transmission potential at population, group, dyad, and individual scales.

Supplemental Information