Habitat use and social mixing between groups of resident and augmented bighorn sheep

Monitoring dispersal, habitat use, and social mixing of released ungulates is crucial for successful translocation and species conservation. We monitored 127 female bighorn sheep (Ovis canadensis) released in three populations from 2000 to 2009 to investigate if augmented bighorns expanded and shifted seasonal ranges, used different habitat compared with resident females, and if animals mixed socially. Augmented bighorns in all populations expanded range use compared with residents by shifting utilization distributions. Size of utilization distributions, however, were smaller for augmented females compared with residents in all areas except one. Overlap of seasonal utilization distributions between augmented and resident bighorns and use of slope and elevation differed across populations. In two populations, differences in size and overlap of seasonal utilization distributions and use of slope and elevation supported the hypothesis that habitat use of bighorns in their source area influenced their habitat use after release. Mixing between resident and augmented adult females occurred on average during only 21% of sightings and was similar across populations. Our results clarify how augmented bighorns mix with resident animals and how habitat use is modified following augmentations. Such information is needed to improve bighorn sheep augmentations and can be applied to augmentations of other ungulates.

field methods. All historical populations of bighorn sheep were extirpated from our study areas by the 1930s [52][53][54] ; therefore, from 2000 to 2007, 157 first resident and then augmented bighorn sheep were released to Mount Timpanogos (n = 82), Rock Canyon (n = 32), and Mount Nebo (n = 43) ( Table 1). Of those resident and augmented bighorns, Ninety-four females were equipped with VHF radio collars at the time of release (Table 1), and 12 additional females received telemetry collars periodically throughout the study 50 Fig. 2). All our study areas were historical bighorn sheep habitat and identified as suitable release sites by the Utah Division of Wildlife Resources. Wildlife biologists from the Utah Division of Wildlife Resources followed established protocols when handling, translocating, and attaching radio-transmitting collars and ear tags to bighorns 55 . Across all years of our study, mean (±SD) population sizes (i.e., all sex and age classes) for these areas were as follows: Mount Timpanogos  (Table 1) 50 , because survival rates for those 14 females released from 2001 to 2002 were low 50 , and similar clumping of ungulates released in subsequent years occurred in another study 14 . Doing such also allowed use to compare resident animals with augmented animals that were released in the same year (2007) in all populations, which helped control for environmental factors that could have influenced movements and space use (e.g., snow depth). We also considered animals released in Rock Canyon (2001) and on Mount Nebo (2004) as resident animals, and those released in 2007 as augmented bighorns (Table 1) 50 .
To differentiate resident from augmented individuals, all bighorns released in all study areas in 2007 were marked with two, colored ear tags. Therefore, after releases in 2007, all augmented females had either a   49,50,56 . We also observed groups of uncollared bighorns in those same areas and years. We watched groups of bighorns an average of 24 times each month. When we observed a group, we noted sex of individuals and group size and composition 42,51 . We considered animals to be a part of the same group if they were ≤50 m from one another, or if they appeared to be aware of the presence of other sheep and moved as a cohesive unit 51,57 . We only used sightings that contained ≥1 adult, female bighorn sheep for analysis. In all areas, those sightings could include young, yearlings, and males, as long as ≥1 adult female was present in the group.

Data analyses.
To produce 95% utilization distributions and quantify social mixing, we defined bighorn groups as follows: resident groups were ≥1 adult, female bighorn sheep from original releases with any other bighorn sheep (i.e., females, young, yearlings, males) from original releases; augmented groups were ≥1 adult, female bighorn sheep from 2007 augmentations with any other bighorn sheep from 2007 augmentations. Mixed groups consisted of ≥1 adult, resident female with ≥1 adult, augmented female with any other bighorn sheep from any release. When groups of resident and augmented bighorn sheep were mixed, we assigned a location for each group (resident and augmented) for that sighting in our spatial analyses. During our study, two augmented females released in Rock Canyon in 2007 crossed a highway that was considered a barrier to major movements of bighorn sheep and occupied areas on Mount Timpanogos 49 . We censored locations from those females in our analyses, because their inclusion artificially overestimated 95% utilization distributions for augmented bighorns in Rock Canyon.
We calculated size (km 2 ) of seasonal and annual 95% kernel utilization distributions 47,58,59 by study area and group (resident or augmented) using direct plug-in methodology with select bandwidth 60,61 in R 62 , and only produced utilization distributions for areas or seasons with ≥18 locations. We determined seasons by plotting precipitation against temperature, which had been done previously for these areas 48 . The following four seasons were evident for our study areas: spring (March-May), summer (June-September), autumn (October), and winter (November-February) 48 . Because October was a transitional month 48 , and sample sizes were <18 for all study areas during that month, we did not produce seasonal utilization distributions for autumn. Autumn locations, however, were included in annual utilization distributions.
We quantified the degree of overlap in seasonal and annual utilization distributions by study area and group (resident or augmented) using the utilization distribution overlap index (UDOI) 46,47,63 . UDOI is a three dimensional index that considers the volumetric overlap between utilization distributions. UDOI values typically range from 0 to 1, with 0 indicating no overlap of distributions and 1 indicating 100% overlap of uniform distributions. UDOI can be >1 when utilization distributions are non-uniformly distributed and have a high degree of overlap 46 . We used the KernSmooth package 60 for kernel smoothing and density estimation and the adehabitat package 63,64 for utilization distribution analyses and mapping in R 62 . We also compared utilization distributions for resident animals before and after augmentation to assess any changes in distributions for resident animals after augmentations. We used the 2-sample Z-test for proportions 65 , which allowed sampling with replacement, to investigate if the proportion of mixed female groups was different among populations. We also used a Bonferroni correction to adjust alpha for our comparisons; that adjusted alpha was 0.02.
We compared use of slope, elevation, and ruggedness for each observation between groups of resident and augmented bighorn sheep 42,57,66,67 using data from a United States Geological Survey 30-m digital elevation model. Slope was calculated with the slope tool in the Spatial Analyst Tools extension in ArcGIS. We quantified elevation at the site of each location by extracting that value from the digital elevation model. Ruggedness was calculated using the Vector Ruggedness Measure Tool in the Terrain Tools extension in ArcGIS 67 . That tool measures terrain ruggedness as the variation in three-dimensional orientation of grid cells within a neighborhood. Vector Ruggedness Measure values can range from 0 (no terrain variation) to 1 (complete terrain variation). Typical values for natural terrains range between 0 and about 0.4 67 . We used a 95% confidence interval of the difference of means to test for differences in bighorn use of slope, elevation, and ruggedness between groups within study areas 68 .

Results
Mount timpanogos. We used 1,613 sightings to produce seasonal and annual 95% utilization distributions for groups of female bighorn sheep on Mount Timpanogos (Table 2). Augmented female bighorn sheep had smaller annual utilization distributions compared with that of residents (Table 2). On average, size of 95% seasonal utilization distributions for resident bighorn sheep were 2 times larger than estimates of utilization distributions for augmented females (SD = 0.47, range = 1.5 to 2.7 times larger, Table 2), with the largest area used by all bighorns in summer, and the smallest area used by all females in spring and winter. Augmented female bighorn sheep from Colorado had smaller utilization distributions, except in summer, than utilization distributions of resident and Montana bighorns ( Table 2).
Annual 95% utilization distributions for resident bighorn sheep before compared with after augmentation exhibited a high degree of overlap on Mount Timpanogos (UDOI = 1.09), indicating little change in range use of resident animals before compared with after augmentations. After augmentation of bighorn sheep in 2007, overlap of seasonal and annual 95% utilization distributions were substantial between augmented females from Montana and residents (Table 3, Fig. 2). Overlap of annual and seasonal 95% utilization distributions were markedly lower, however, for augmented females from Colorado compared with resident females, with the lowest seasonal overlap occurring in winter and summer (Table 3, Fig. 2). On average, augmented Montana bighorns used areas 68 m higher in elevation compared with resident animals; augmented Colorado bighorns on average used areas 333 m higher in elevation than resident animals (Table 4). No difference occurred with use of slope or (2019) 9:14984 | https://doi.org/10.1038/s41598-019-51370-y www.nature.com/scientificreports www.nature.com/scientificreports/ ruggedness between resident and augmented animals on Mount Timpanogos (Table 4). Overall, augmentations in this area expanded range use by 5.2 km 2 (Fig. 2).

Rock canyon.
We used 708 sightings to produce seasonal and annual 95% utilization distributions for groups of female bighorn sheep in Rock Canyon. Augmented female bighorn sheep had a larger annual utilization distribution compared with that of residents (Table 2). On average, size of 95% seasonal utilization distributions for resident bighorn sheep were 1.6 times smaller than estimates for augmented females (SD = 0.88, range = 1.3 to 2.1 times smaller, Table 2), with the largest area used by all females in summer, and the smallest area used by resident females in spring.
Annual 95% utilization distributions for resident bighorn sheep before compared with after augmentation exhibited a high degree of overlap in Rock Canyon (UDOI = 1.00), indicating little change in seasonal range use of resident animals before compared with after augmentations. After augmentation of bighorn sheep in 2007, overlap of seasonal and annual 95% utilization distributions occurred between augmented and resident females, with the least overlap occurring in spring (Table 3, Fig. 2). A difference in use of slope occurred, with resident animals using areas of steeper slope than augmented females (Table 4). No difference occurred with use of elevation and ruggedness between resident and augmented females in Rock Canyon (Table 4). Augmentation expanded range use by 3.9 km 2 (Fig. 2).

Mount nebo.
We used 402 sightings to produce seasonal and annual 95% utilization distributions for groups of female bighorn sheep on Mount Nebo (Table 2). Augmented female bighorn sheep had a smaller annual utilization distribution compared with that of residents (Table 2). On average, size of 95% seasonal utilization distributions for resident bighorn sheep were 2.2 times larger than estimates for augmented females (SD = 1.50, range = 1.2 to 3.9 times larger, Table 2), with the largest area used by all females in summer, and the smallest area used by augmented females in spring.
Annual 95% utilization distributions for resident bighorn sheep before compared with after augmentation exhibited low overlap on Mount Nebo (UDOI = 0.44), indicating that resident females changed seasonal range use after augmentation. That low overlap was likely because of 7 of 8 collared resident females dying of disease within 11 months after augmentation. After augmentation of bighorn sheep in 2007, overlap of seasonal and annual 95% utilization distributions between augmented and resident females was minimal, with the least overlap occurring in spring (Table 3, Fig. 2). On average, augmented females used areas 260 m higher in elevation compared with resident animals (Table 4). No difference occurred with use of slope or ruggedness between resident and augmented females on Mount Nebo (Table 4). Augmented females expanded range use by 6.5 km 2 (Fig. 2).  Fig. 3). On Mount Timpanogos when groups were mixed, resident bighorns mixed with Montana bighorns 64% of sightings; while resident bighorns mixed with Colorado animals 6% of sightings. Montana and Colorado bighorns mixed 22% of sightings, and all three groups mixed 8% of the sightings.

Discussion
Similarities among augmentations. Augmented bighorn sheep in all populations expanded range use by shifting utilization distributions after augmentation regardless of how long resident bighorns had been in the area. Similar range expansion has been documented in other populations 42,69 . That shifting of utilization distributions of augmented adult females is important to wildlife conservationists who want to expand areas used by existing populations, which can help with connectivity and population persistence 39 www.nature.com/scientificreports www.nature.com/scientificreports/ and spring. Those outcomes likely occurred because winter range was limited, resulting in animals moving to lower elevations during winter, and after winter subsequently moving to smaller areas for birthing. Additionally, the season with the largest size of utilization distributions for augmented animals was summer. Therefore, releasing bighorn sheep in areas with connectivity to large areas of underutilized summer habitat has the potential to expand range use.
Augmented animals can learn information by socializing with others 25,29,73 , and understanding social relations by animals can improve wildlife conservation 40,74 . Comparatively little work, however, has been done on the behavior of released animals 24,35,36 , and is needed to improve translocations 34 . Indeed, animal augmentations provide unique opportunities for understanding social interactions 34 . In our study, mixing between resident and augmented adult females occurred infrequently. Intriguingly, the frequency of group mixing across populations was similar, indicating a consistency in the lack of social mixing of augmented bighorn sheep with residents on our study sites. Low levels of social interaction between females in an historical population and augmented females have been documented anecdotally in one population of bighorn sheep 69 . Female bighorn sheep from historical populations may recognize individuals from their own group and may not readily join other groups even if their range use overlaps 75 . Additionally, low social interactions occurred between juvenile (<3 years old) translocated bighorn sheep and bighorns from an historical population in nursery groups in summer in one population 37 . Those low social interactions between juvenile augmented bighorn sheep and animals from that historical population affected time spent foraging and body mass of augmented individuals, as well as possibly translocation   Augmented Montana bighorns on Mount Timpanogos and Mount Nebo. Resident female bighorn sheep used seasonal areas that were on average almost twice as large as seasonal estimates for augmented females from Montana. That difference in use of larger seasonal areas may have been because resident bighorns had been in those areas for 7 (Mount Timpanogos) and 2 years (Mount Nebo) prior to release of augmented bighorn sheep, and had sufficient time to become more familiar with their environment and thereby use more habitat 25 . Further, overlap www.nature.com/scientificreports www.nature.com/scientificreports/ of utilization distributions for resident females on Mount Timpanogos did not change after augmentations, indicating that augmentations had little effect on utilization distributions of resident animals. Overlap of seasonal utilization distributions were substantial, however, between resident and augmented bighorn sheep on Mount Timpanogos, indicating that augmented bighorns used similar, smaller areas than those used by resident bighorn sheep. Additionally, social mixing was highest between those two groups, which was most likely driven by the significant overlap in utilization distributions between resident and augmented, Montana bighorn sheep on Mount Timpanogos. Similar to Colorado bighorns on Mount Timpanogos, augmented bighorns on Mount Nebo used areas of higher elevation, and seasonal overlap in habitat use was low, indicating that augmented animals expanded range use to areas of higher elevation, especially in spring.

Differences among augmentations.
Augmented Colorado bighorns on Mount Timpanogos. As with augmented bighorns from Montana, resident female bighorn sheep used seasonal areas that were on average almost twice as large as seasonal estimates for augmented females from Colorado. Again, that difference in use of large seasonal areas by resident animals likely was because resident bighorns had been in that area for 7 years prior to release of augmented animals 25 . Additionally, overlap of seasonal utilization distributions was low between resident and Colorado bighorn sheep. That lack of overlap was caused by augmented bighorns using areas of much higher elevation in summer and moving to different areas than resident bighorns in winter, which also resulted in the lowest social mixing between those groups. Indeed, bighorn sheep from Colorado came from a source herd where many of those animals use year-round habitat at high elevations (>3,000 m).
Addressing a priori questions in reintroduction biology allows biologists to use scientific evidence to apply best management practices in the field 26,35 . Biologist from the Utah Division of Wildlife Resources released bighorns from Colorado on Mount Timpanogos to improve the use of high-elevation habitat. That release did expand range use to areas of high elevation, and our data support the hypothesis that habitat use of a source population can influence habitat use by animals after release 30 . Indeed, Mount Timpanogos offered a unique opportunity to compare augmentations from different source habitats; such replication often is lacking in studies of reintroduction biology 30,35 .
Rock canyon. Unlike resident female bighorn sheep on Mount Timpanogos and Mount Nebo, resident females in Rock Canyon used seasonal areas that were on average 1.6 times smaller than areas used by augmented female bighorns. That reduction in area used by resident females likely occurred because those animals moved to small rocky areas of steeper slope to give birth that overlapped little with augmented animals, similar to what occurred in other areas. Resident bighorn sheep also used small winter ranges compared with augmented animals. In addition, birthing and wintering areas for resident animals were near the release site, indicating a lack of dispersal. Also, overlap of utilization distributions for resident females did not change after augmentations, again indicating that augmentations had little effect on utilization distributions of resident animals. Resident bighorn sheep in Rock Canyon came from a mine site near Cadomin, Canada. Bighorns in that area have small seasonal ranges on the reclaimed mine site; therefore, we hypothesize that limited movements of resident animals in Rock Canyon at their capture site influenced habitat use by those animals after release. Our results from resident females in Rock Canyon add support for the hypothesis that habitat use of a source population can influence habitat use by animals after release 30,56 . Additional research should test if habitat use by augmented animals and resident animals becomes similar with longer time after release.
Density-dependent factors. Population size was low and similar in our three study areas. Additionally, female bighorns in our study populations experienced high pregnancy rates indicating that those populations were not likely influenced by density-dependent factors [48][49][50] . As documented in all of our study populations, augmented animals expanded range use by shifting utilization distributions after augmentation. That expansion could have occurred because resident bighorn sheep were at low densities and had not occupied all available habitat in each study area. Testing ideas about density dependence while monitoring released animals is needed 34 . For example, small populations, or low densities, may be less attractive than larger ones for released animals 36 , thus potentially influencing habitat use and social mixing. We hypothesize that in areas with resident bighorn sheep at higher population densities than ours, that range expansion by augmented animals may not be as pronounced as what we observed. Further, we predict that social interactions would occur more often when augmented bighorn sheep are released into areas with resident bighorn sheep at high population densities 34 . Future research can test those ideas.
Social interactions and learning. Habitat use and social mixing of bighorn sheep can be influenced by learning that is shaped by predator avoidance and feeding efficiency 41,76,77 . This learning can influence habitat use and social mixing of younger animals especially, as they interact with and follow older individuals 41,78,79 , or as they form groups to avoid predation 41,76,77 . Furthermore, movement to and use of the seasonal distribution of high-quality forage by translocated bighorn sheep appears to be culturally transmitted across decades, with knowledge acquired of such areas of forage increasing over time since release 40 . Bighorns we studied that were considered resident (animals released from 2000 to 2004) may have still been familiarizing themselves with habitats in our study areas 40 ; therefore, we caution about comparing our results to those of translocations of bighorn sheep into established historical populations. In our study areas, we hypothesize that habitat use and social mixing between resident and augmented adult females will increase as migration traditions are inherited over generations 40  www.nature.com/scientificreports www.nature.com/scientificreports/ to supplement declining herds 38,39,42 . Despite those efforts, success rate of translocated populations of bighorns is low 38,42,80 ; indeed, suitable habitat and connectivity between augmented and resident animals is crucial for successful bighorn sheep augmentations 32 . We were able to quantify habitat use and social mixing of augmented and resident bighorn sheep across three study populations, that replication was crucial for the generalization of findings 24,35,45 . We documented range expansion by all augmented bighorn sheep as they shifted utilization distributions. Size of utilization distributions, however, were smaller for augmented females compared with residents in two of three areas. Differences in size and overlap of seasonal utilization distributions and use of slope and elevation in two populations supported the hypothesis of habitat use of animals in their source areas transferring to release areas after augmentation 30 . Finally, mixing between resident and augmented adult female bighorn sheep occurred infrequently, and was similar across populations. Our results provide important information on how augmented bighorn sheep mix with resident animals and how use of habitat changes after releases. Such information is needed to improve augmentations of bighorn sheep, and can be applied to releases of other open-habitat social ungulates as well.