The spatial distribution of animals has consequences for nutrition, predator-prey dynamics, spread of diseases, and population dynamics in general. Animals must establish a home range to secure adequate resources to fuel their energetic needs. Home ranges, therefore, are temporally and spatially dynamic given the changing requirements of an animal and the availability of resources on the landscape. We used data from two populations of bighorn sheep with contrasting population dynamics following pneumonia epizootics and different habitat quality on their summer range to test the hypothesis that the distribution and size of home ranges are influenced by environmental conditions and reproductive status. We used a combination of data from 768 vegetation transects and remotely sensed metrics to index forage quality of consecutive biweekly home ranges for 27 bighorn sheep, June–August 2019–2021. There were population differences in space use that were consistent with resource limitations in the population declining in abundance. Animals in both populations increased the area of their space use through the summer in association with declining forage quality indexed by plant phenology. Furthermore, animals in the Whiskey Mountain population without live offspring used areas more than twice the size of animals with offspring, whereas there were no differences in the area of space use between animals with and without offspring in Jackson. We demonstrated that limitations young offspring impose on space use of a mother may have consequences for animals living where larger home ranges are needed to secure adequate resources—sheep in Whiskey Mountain had to travel 1,000 m from escape terrain to access the same amount of biomass that the Jackson sheep could access directly adjacent to escape terrain. Forage quality and availability influence movement and home range dynamics. In the presence of disease, movement and home range dynamics may influence pathogen transmission and persistence. Thus, forage availability may play an indirect role in population dynamics in the presence of disease, which is another line of evidence for how environmental and nutritional conditions may influence population dynamics of populations coping with disease.

We used data from two populations of Rocky Mountain bighorn sheep within the Greater Yellowstone Ecosystem in northwest Wyoming, USA, 2019–2021 (Figure 1). The Whiskey Mountain population experienced a pneumonia epizootic in 1991 (Ryder et al. 1992) and has since exhibited population decline via low juvenile recruitment (22 juveniles per 100 adult females in winter on average 2019–2021; Wyoming Game and Fish Department 2021a), leaving the population at ~20% of its former population size (upwards of 1,500 animals; Wyoming Game and Fish Department, unpublished data). The Jackson population experienced pneumonia epizootics in 2001 and 2012 but has been able to recover to previous population size (~ 400 animals) and maintain higher juvenile recruitment (38.6 juveniles per 100 adult females; Wyoming Game and Fish Department 2021b).

Study animals were seasonal elevational migrants. Summer ranges were high-elevation (~3,000m) alpine habitats with alpine meadows, talus fields, and rocky outcrops. Winter ranges were mixed conifer, mountain shrub, and grasslands. The climate of both study areas was cool, short summers and long winters with high snowfall, especially at high elevations. The summer range of the Jackson population had higher biomass, species diversity, and available macro- and micronutrients than that of the Whiskey Mountain population (Wagler et al. 2023; Figure 1). See Wagler et al. (2023) for more detail on the study area.

Animal capture and handling

We captured adult female bighorn sheep via helicopter net-gunning (Krausman et al. 1985, Wagler et al. 2022) in the Jackson and Whiskey Mountain populations in late winter 2019–2021. We fit sheep with a GPS collar with hourly fix rates (Vertex Plus, VECTRONIC Aerospace Gmbh, Berlin, Germany). We censored GPS locations that had dilution of precision > 10, or speed > 3.6 km per hour, which were arbitrary thresholds that removed problem locations for our data (D'Eon et al. 2002). We used ultrasonography (5-MHz transducer, Ibex Pro, E.I. Medical Imaging, Loveland, CO, USA) to determine pregnancy status (Stephenson et al. 1995) and fit pregnant animals with a vaginal implant transmitter (VIT, VECTRONIC Aerospace Gmbh, Berlin, Germany) programmed to notify field personnel of a birth event. Once a birth event was detected we captured neonates by hand (Smith et al. 2014) and fit them with an expandable GPS collar (Vertex Mini, VECTRONIC Aerospace Gmbh, Berlin, Germany) with 4-hour fix rates and set to enter mortality mode after no movement for 8 hours. Field personnel investigated the site to confirm mortality when a collar entered mortality mode, allowing us to determine the date of death. All animal handling was approved by the Wyoming Game and Fish Department (Chapter 33-1278), Institutional Animal Care and Use Committee (IACUC; 20180305KM00296), and was in accordance with the guidelines of the American Society of Mammologists (Sikes 2016).

Home range dynamics

To quantify home range size, we created 95% utilization distributions using dynamic Brownian bridge movement models (Horne et al. 2007) with the BBMM package in R (Nielson et al. 2013). Dynamic BBMMs provide an occurrence distribution and were able to capture the directional movement of our study animals without inflating the area of space use. We calculated occurrence distributions for 6 consecutive 2-week periods throughout the summer (1 June–31 Aug) for each animal year. Home range size was the area (km2) of the 95% contour. We used two-week intervals because this was short enough to capture behavioral changes associated with the loss of a juvenile but long enough to accurately describe the space use of each animal. To further assess home range dynamics, we calculated the degree of spatial overlap in home ranges between each consecutive 2-week period for each animal. We used utilization distribution overlap index (Fieberg and Kochanny 2005) for this metric—higher values indicated an animal did not move its home range at all from one period to the next and a 0 indicated that an animal completely relocated its home range.

Assessment of forage quality

We used a combination of vegetation transects and remotely sensed data to assess forage quality within biweekly home ranges. Data from vegetation transects represented core foraging areas and remotely sensed metrics represent the entire home range. For each transect, we calculated six metrics of forage quality (biomass, species diversity, dry matter digestibility [DMD], crude protein [CP], and DMD and CP relative to plant cover). Biomass was indexed with the spatially explicit above-ground biomass estimate for the starting point of each transect (Rangeland Analysis Platform [RAP]; kilograms/hectare; 30-m spatial resolution; yearly temporal resolution) (Robinson et al. 2019). We used the Shannon-Weiner species diversity index (Whittaker 1972) to calculate species diversity of each transect. We calculated two metrics for macronutrients (DMD and CP) to index forage quality. First, we calculated the mean of each macronutrient at each transect weighted by the number of hits per genera, ignoring inorganic hits. This metric reflects the quality of forage at the transect, regardless of how much was available at the transect. To account for how plant cover affects the availability of macronutrients, we calculated a second weighted mean at each transect including a 0 for each inorganic hit. The second metric is an index of macronutrient availability at a transect. See Wagler et al. (2023) for more detail on vegetation transects and quality metrics.

Quality of forage for ungulates is directly linked to phenological stage, which can be indexed with remotely sensed metrics of vegetation biomass, specifically the Normalized Difference Vegetation Index (NDVI) (Hamel et al. 2009). We used the mean and coefficient of variation of the number of days to or from peak instantaneous rate of green up (IRG) to index forage quality at a larger scale (95% 2-week home range) than the transect level metrics (core foraging areas). We calculated IRG (scaled between 0 and 1) with the first derivative of the NDVI curve, which results in a curve that peaks when vegetation growth is most rapid and thus, when forage quality peaks for ungulates (Fryxell 1991, Hebblewhite et al. 2008) (Appendix S1).

To index forage quality for each biweekly home range, we calculated the mean of each forage quality metric (DMD, CP, DMD and CP relative to plant cover, species diversity, and biomass) of each transect that was within the corresponding 2-week home range contour (i.e. any transect contained within a biweekly home range was included regardless of the animal the transect corresponded to). A transect represented all the biweekly home ranges that spatially overlapped it. Because we expect quality of forage for ungulates to be highest during maximum IRG and to decline in quality as it senesces throughout the summer (Bischof et al. 2012, Geremia et al. 2019), we adjusted macronutrient values for each biweekly period (i.e. macronutrient values were different for the same transect at the beginning of June than at the end of August) using equations described in Wagler et al. (2023) and reported in Appendix S1 Tables S1 and S2. This phenology adjustment allowed us to use macronutrient values from every transect for each biweekly home range, regardless of the date the forage samples were collected. Additionally, we calculated the mean and coefficient of variation of the number of days to or from peak IRG for each home range.

To compare landscape structure between the two summer ranges, we fit a linear mixed-effects model with herbaceous biomass (Rangeland Analysis Platform [RAP]; kilograms/hectare; 30-m spatial resolution; yearly temporal resolution) as the response and an interaction between distance to escape terrain (m) and population as the predictor. The biomass and escape terrain values were extracted from each GPS location June–August. We defined escape terrain as areas that had slope > 30 degrees and terrain roughness (the difference between the minimum and maximum value of a cell and its 8 surrounding cells; Wilson et al. 2007) > 70. The model included a random intercept for animal year.