Use of Global Positioning System Tracking to Assess Landscape Distribution in Extensive Small Ruminant Operations

ABSTRACT Landscape distribution and grazing management of small ruminants are becoming more important issues as the sheep and goat industry continues to grow. The objective of this study was to evaluate spatial movement patterns of sheep and goats in Queensland, Australia using Global Positioning System (GPS) tracking to determine daily distance traveled, distance traveled from water, activity levels, and the influence of ambient temperature on these metrics. Optimized hotspot analysis was also performed to determine areas of increased small ruminant presence across the landscape. GPS positions were recorded at 10-min intervals over the course of several months. Sheep and goats traveled an average distance from water of 0.6–1.1 km, with a maximum of 2.4 km. Daily distance traveled of sheep and goats averaged 6–9 km·d–1, with a maximum of 11 km·d–1. Average daily activity ranged between 42% and 47%. The overall trend of the datum showed that increased temperature led to decreased daily distance traveled, distance traveled from water, and activity of sheep and goats. The hotspot analyses showed a higher concentration of sheep and goats near water sources. Understanding small ruminant distribution in extensive rangeland pastures helps producers implement management regimes to improve efficacy and sustainability of land use and production.


Introduction
As small ruminant populations increase ( Mazinani and Rude 2020 ) and sheep and goat products become more popular and desired ( Montossi et al. 2013 ;Mazhangara et al. 2019 ;Miller and Lu 2019 ), management of small ruminant grazing becomes more important.Understanding how livestock utilize the landscape can help inform management decisions, such as where to place water sources.Daily distance traveled and distance traveled from water are key factors for describing landscape distribution and utilization by small ruminants.Some studies have explored spatial movements of sheep and goats on rangelands, but results were highly variable.For example, Askins and Turner (1972) found the average daily distance traveled by goats was 4.5 km, while Swain et al. (1986) found the average to be 14.3 km.Distances traveled from water are also quite variable; Daws and Squires (1974) reported the average distance to water of sheep to be 4 km, while Fensham and Fairfax (2008) reported the average to be 2.5 km.The variability between these studies could be due to differences in measurement techniques (e.g., Global Positioning System [GPS] tracking vs. visual observations), as well as pasture size, water availability, and terrain.
The majority of research assessing small ruminant landscape distribution has relied on observation, whether through solely human observation ( Askins and Turner 1972 ;Lynch 1974 ), human observation and aerial photographs ( Squires 1974 ), photographs of marked sheep in an experimental laneway ( Squires and Wilson 1971 ), or estimating the distance of each step and counting steps the animal takes ( Swain et al. 1986 ;Animut et al. 2007 ).GPS and geographical information system technology has allowed for more consistent and precise tracking of livestock movement.Yet far fewer studies assess sheep and goat landscape distribution than cattle distribution ( Wade et al. 2023 ), which could prove to be problematic as managers must understand livestock spatial distribution to develop sustainable grazing plans ( Bailey 2004 ).
In Australia, the feral goat population has increased dramatically since 1997, with the 2010 population of feral goats estimated to be 3.3 million ( Pople and Froese 2012 ).Many producers have https://doi.org/10.1016/j.rama.2024.01.010 1550-7424/© 2024 The Authors.Published by Elsevier Inc. on behalf of The Society for Range Management.This is an open access article under the CC BY license ( http://creativecommons.org/licenses/by/4.0/ ) taken advantage of this increase in feral goat numbers and have domesticated these animals for production ( Pople and Froese 2012 ).This study was conceived to provide more information on both sheep and goat landscape utilization to livestock producers and land managers in Australia.In addition, results from this study could be useful to small ruminant producers grazing extensive rangeland areas around the world because of the lack of research in this area ( Wade et al. 2023 ).The objective of this study was to use GPS tracking data to describe spatial behavior of small ruminants grazing rangeland pastures using metrics such as daily distance traveled, distance traveled from water, and daily activity patterns.We explored the relationship between these metrics and ambient temperature, as well as assessed the location of highly utilized sections of the landscape, otherwise known as hotspots.

Location and animals
This study used sheep and goat tracking data from three properties in the semiarid landscape of Longreach, Queensland, Australia.Mean annual rainfall in Longreach is 450 mm, which falls predominantly in the summer.Average maximum temperature in the summer is between 36 °C and 37 °C, while the average maximum temperature in winter is between 23 °C and 26 °C.The average minimum temperature in summer is between 22 °C and 23 °C, and the average minimum in winter is between 7 °C and 8 °C.All properties were in the Mitchell Grass Downs land type, characterized by gentle undulating terrain of grass plains with patches of gidgee ( Acacia cambagei R.T. Baker) wooded plains and ephemeral waterways lined with eucalyptus trees and gidgee.Common grasses in the area include mitchell grass ( Astrebla spp.), buffel grass ( Pennisetum ciliare [L.] Link), flinders grass ( Iseilema sp.), and star grass ( Panicum decompositum R. Br.).Common forbs include roly poly (Salsola kali), black roly poly (Sclerolaena muricata), goathead burr (Sclerolaena bicornis), and pigweed (Portulaca oleracea).
The first property, Camden Park, is 7 365 ha.Due to wild dog predation of livestock and competition for resources by the large population of kangaroos in the area, a 2-m-tall exclusion fence lines the perimeter of the property.Two deployments of GPS tracking collars were used on the Camden Park property.Collared animals were randomly selected from the flock/herd, which contained between 100 and 300 animals.In the first deployment, 9 Dorper sheep and 8 Rangeland goats were tracked in a 4 120-ha pasture, which contained three water dams and an artesian well (bore) creating a stream.The Rangeland goats at Camden Park were captured feral female goats crossed with Boer bucks.The sheep and goats grazed in the same pasture.The first study period at Camden Park began 11 August, 2019 and ended 6 December, 2019 for goats and 31 January, 2020 for sheep.This was during the spring and summer months in Australia.The average maximum temperature was 36 °C, and average minimum temperature was 20 °C for the study period.The second deployment tracked 12 Dorper sheep and 12 Rangeland goats that cograzed a pasture of 1 102 ha with three water dams.This second study period started 5 February, 2021 and ended 27 April, 2021 for goats and 31 May, 2021 for sheep.This was during the summer and fall in Australia.Average maximum temperature was 33 °C, and average minimum temperature was 18 °C for the study period.Even though these deployments were conducted during a drought, forage availability at Camden Park was not considered limiting as stocking rates were very low (14 ha/animal unit month [AUM] in 2019, 6 ha/AUM in 2021).Partway through the second deployment on Camden Park, five of the collared goats were moved to a neighboring property, Beaconsfield, for the remainder of the study, from 2 May, 2021 to 6 June, 2021.
At Beaconsfield, goats had access to 7 778 ha of rangeland with 17 water sources from dams, troughs, or bores.The perimeter of the Beaconsfield property is lined with exclusion fencing.
The third property, Fairfield, also located in Longreach, Queensland is 16 963 ha.This property did not have exclusion fencing at the time of this study.We placed GPS tracking collars on 11 Merino sheep (randomly selected from a larger flock of approximately 800 sheep).The sheep were originally placed in a 2 266-ha pasture with one water dam and one bore (well) that created a stream through the pasture.Two weeks into the study, a large flood came through the area and knocked down several fences within the Fairfield property and a fence between Fairfield and the neighboring property.Following the flood, sheep had access to a much larger area and additional water sources, including the standing water from the flood.Therefore, daily distance traveled and activity measures were based on the entire study period from 17 December, 2020 to 25 March, 2021, but distance traveled from water was only calculated for preflood data from 17 December, 2020 to 3 January, 2021 when the flood occurred.Average maximum temperature during the full study period was 36.5 °C, and average minimum temperature was 23.6 °C.Forage availability at Fairfield was considered limiting at the commencement of the study.Sheep were supplemented with hand-fed flava beans every other day for the first week of tracking at Fairfield.No further supplementation occurred.The stocking rate for this property was 3 ha/AUM, a higher rate than Camden Park.The Fairfield stocking rate was considered typical for commercial sheep operations in the local area.

GPS tracking collars and data analysis
Each animal was fitted with a collar that contained an i-gotU GT-600 GPS unit (Mobile Action i-gotU GT-600 USB GPS Travel Logger, eXpansys Australia).The GPS positions were not differentially corrected, and the mean location error of the i-gotU device was < 10 m ( Morris and Conner 2017 ).For the Fairfield property and the first deployment on Camden Park, the GPS unit recorded a location every 10 minutes.For the second deployment on Camden Park and successively Beaconsfield, GPS units recorded a location every 5 min.All data were downloaded, and 5-min tracking data were resampled to 10 minutes in order to standardize the data from all properties.All GPS data were processed using the 'animaltracker' package in R ( R Core Team 2020 ) ( Champion and Sukianto 2020 ).This process removed erroneous GPS points based on excessive velocity.Locations associated with a velocity > 84 m/min, a normal walking rate for cattle ( Chapinal et al. 2009 ), were deleted.In addition, all GPS points outside of the pasture boundary were removed.The boundary included a 5-m buffer outside the fence to help account for GPS error.In order to calculate distance traveled and distance to water, the GPS points were projected from latitude and longitude coordinates to the Universal Transverse Mercator (UTM) coordinate system, WGS_1984_UTM_Zone_55S in ArcMap.Collars that did not record data for the entire study period were not used.To ensure accurate representation of sheep and goats on expansive rangeland pastures, all days where the animals were moved between pastures or penned were removed from the analyses.
The distance between sequential coordinate points were calculated using the Pythagorean Theorem and UTM coordinates and summed in 24-h periods to determine the daily distance traveled per animal.Because we could only estimate the straight-line distance between recorded positions rather than the actual path, our estimates of daily distance traveled are less than if we recorded positions more frequently ( Johnson and Ganskopp 2008 ).The rate of animal travel between each recorded location was calculated, and an activity threshold of 0.03 m •s −1 was determined after consulting the diurnal activity graphs of animal speed.Velocities > 0.03 m •s −1 were considered active, and those less than this thresh- old were classified as resting (nonactive), similar to the methods of Putfarken et al. (2008) .Using the Euclidian distance tool in the spatial analyst toolbox in ArcGIS (v.10.7.1 Esri), the distance between every GPS position and the nearest water source was calculated to determine the distance to water for each animal.To describe the overall movement patterns, the reported average, median, and standard deviation values of daily distance traveled, distance to water, and percent activity are based on the daily data of all GPS collars for each property.However, the minimum and maximum values for each metric were reported as the average of the monthly extremes for each collar on each property.The optimized hotspot analysis tool in ArcGIS uses the Getis-Ord Gi * statistic ( Getis and Ord 1992 ;Ord and Getis 1995 ) to determine whether the distribution of animal locations across an area is random or clustered.The Getis-Ord Gi * statistic used in this study ( Esri Ar-cGIS n.d. ) is expressed as: where x j is the attribute value for feature j , w i.j is the spatial weight between feature i and j, n is the total number of features and: (3) We set the optimized hotspot analysis to count incidents within hexagon polygons.Otherwise, default analysis parameters were used.Areas of significant GPS clustering (based on the z-score and P value from the optimized hotspot analysis) were considered hotspots and areas of significantly lower GPS clustering were cold spots.We conducted hotspot analyses on all tracking data, as well as only tracking data considered active, based on the definition given earlier.The results of the hotspot analyses were nearly the same for both data sets; therefore, we did not present the hotspot analyses with only active tracking data for brevity.
Maximum and minimum ambient temperature for each day were collected from the nearest weather station, located at Longreach Airport (15-40 km away from the study sites depending on the property).Maximum daily temperature was used as an independent variable with daily distance traveled, average distance traveled from water, maximum distance traveled from water, and percent activity as dependent variables.Linear and quadratic regression analyses were conducted using repeated measures and the PROC MIXED function of SAS ( Littell et al. 2006 ).Animal was considered the subject, and the covariance of repeated records was modeled using the autoregressive 1 feature, which is appropriate for continuous independent variables.Residuals from the analyses were tested for normality.Log or square root transformations of the dependent variable were used in analyses if residuals did not appear to be normally distributed.If the regression analysis results from the untransformed and transformed data were similar, we presented the results from the untransformed data for ease of interpretation of the coefficients.Separate analyses were conducted for sheep and goats at each deployment or property.

Sheep
The maximum daily distance traveled for the seven Dorper sheep at Camden Park during 2019 averaged 10.1 km, while the overall average daily distance traveled was 6.0 km ( Table 1 ).During 2021, Dorper sheep at Camden Park averaged 9.3 km for the maximum daily distance traveled, and the overall average was 6.0 km •d −1 (see Table 1 ).The eight Merino sheep tracked at the Fairfield property averaged 14.7 km for the maximum daily distance traveled, and the overall average daily distance traveled was 8.3 km (see Table 1 ).

Goats
During 2019 at Camden Park, Rangeland goats averaged 8.3 km for the maximum daily distance traveled, with an overall average of 5.8 km •d −1 (see Table 1 ).The average maximum daily distance traveled for the goats at Camden Park in 2021 was 11.3 km, with an overall average of 7.1 km •d −1 .Of the three goats that were moved to Beaconsfield partway through the 2021 deployment, the average maximum daily distance traveled was 16.2 km with an overall average of 9.6 km (see Table 1 ).Blue cold spot areas represent avoided areas with 90% confidence ( P < 0.10), 95% confidence ( P < 0.05), and 99% confidence ( P < 0.01).Polygons where livestock were not tracked are not shaded with color.

Sheep
Dorper sheep at Camden Park in 2019 traveled an average of 0.65 km from water, and the average maximum distance traveled from water was 2.4 km.In 2021, Dorper sheep at Camden Park traveled from water an average of 0.67 km, and the average maximum distance to water was 1.28 km (see Table 1 ).The Fairfield Merino sheep traveled an average distance from water of 1.1 km and an average maximum distance from water of 2.3 km during the period before the flood (see Table 1 ).The distance sheep traveled from water at Fairfield was not calculated after the flood because readily available water could not be mapped.

Goats
The goats at Camden Park traveled an average of 0.6 km from water in 2019, and the average maximum distance traveled from water was 2.0 km (see Table 1 ).In 2021, goats at Camden Park traveled an average of 0.67 km from water, and the average maximum distance from water was 1.58 km.The goats on the Beaconsfield property had access to 17 working water sources but only utilized 8 of them.We measured distance to water based on all 17 water points and found the average distance to water was 0.89 km, with an average maximum distance of 2.0 km (see Table 1 ).We also measured distance to water based on the 8 utilized water sources and the average distance from water was 0.96 km, and the average maximum distance traveled from water was 3.3 km.tioning System locations with 90% confidence ( P < 0.10), 95% confidence ( P < 0.05), and 99% confidence ( P < 0.01).Blue cold spot areas represent avoided areas with 90% confidence ( P < 0.10), 95% confidence ( P < 0.05), and 99% confidence ( P < 0.01).Polygons where livestock were not tracked are not shaded with color.

Activity level
The average percent of time sheep and goats were considered active each day ranged from 42% to 47% across all properties (see Table 1 ).The most variable levels of activity were at the Fairfield property where sheep were active 23.9% to 60.9% of the time.The remaining properties showed sheep and goat activity ranged from 30% to 58% active.

Hotspot analyses
On the basis of optimized hotspot analyses, the 2019 Camden Park deployment showed higher presence of sheep and goats ( P < 0.01) in areas around water sources ( Fig. 1 ).The hotspot areas not directly next to a water source contained small clumps of trees, dry dirt tanks, and/or dry creek beds.Similar to 2019, sheep and goat presence at Camden Park in 2021 was higher ( P < 0.01) in areas immediately around water sources ( Fig. 2 ).Hotspots were also located along the eastern fence line of the 2021 pasture and along dry riparian areas.No cold spots were detected for either sheep or goats in 2019 and 2021 at Camden Park.The Fairfield property had interesting results as the flood increased water availability across the property.Congregations of sheep (hotspots) were detected ( P < 0.01) across the middle of the property where the flood apparently filled portions of the previously dry large waterway ( Fig. 3 a).Similar to the Camden Park 2021 deployment, sheep congregated on the eastern side of the pasture on Fairfield, shown by the large hotspot along the eastern fence line of the middle pasture (see Fig. 3 a).No cold spots were detected on the Fairfield property.
Hotspots were found around some of the water sources and in areas of high tree cover or along dry riparian areas/waterways ( P < 0.01) at Beaconsfield (see Fig. 3 b).We found cold spots for the goats on Beaconsfield, indicating a lower ( P < 0.05) goat presence in certain areas of the pasture (see Fig. 3 b).

Sheep
In 2019, daily distance traveled by Dorper sheep decreased ( P = 0.009) by 33 m for every degree ( °C) increase in maximum daily temperature ( Table 2 ).The day-to-day changes in daily distance traveled and maximum temperature over the course of the study period ( Fig. 4 ) also suggest that sheep traveled less during hot weather.Activity of the sheep decreased as temperature increased ( P < 0.001) (see Table 2 ).The maximum daily temperature had a quadratic relationship ( P ≤ 0.01) with both average and maximum distance to water of Dorper sheep.As the temperature increased from 20 °C to 33 °C sheep traveled farther from water, but as the temperature continued to increase from 33 °C to 45 °C, the sheep remained closer to water (see Table 2 ).In 2021, when maximum temperature increased, the average distance traveled from water by sheep at Camden Park decreased ( P = 0.03, see Table 2 ).We did not detect ( P > 0.05) a relationship between temperature and activity or temperature and daily distance traveled for the sheep at Camden Park during 2021.We did not detect ( P > 0.05) a relationship between maximum daily temperature and daily distance traveled or distance from water for the Fairfield Merino sheep.However, activity of Merino sheep increased ( P = 0.02) as maximum daily temperature increased, until the temperature reached 37 °C when the activity decreased ( P = 0.04) with increased temperature (see Table 2 ).

Goats
Daily distance traveled was quadratically related ( P < 0.001) to maximum daily temperature for the goats at Camden Park in 2019 ( Table 3 ).As the temperature increased from 20 °C to 34 °C the daily distance traveled increased by 581 m with each degree of temperature.When the maximum temperature continued to increase from 34 °C to 43 °C, the daily distance traveled decreased ( P < 0.001) (see Table 3 ).The relationship between maximum temperature and goat activity in 2019 increased until 35 °C ( P = 0.002) and then decreased at higher maximum temperatures (see Table 3 ).Average and maximum distance to water were negatively related with maximum temperature for goats in 2019 (see Table 3 ).As maximum temperature increased, average distance of the goats from water decreased linearly ( P < 0.001) by 21 m.As the maximum temperature increased, daily distance traveled by goats at Camden Park during 2021 decreased ( P = 0.01) by 83 m per degree (see Table 3 ).Goats at Camden Park during 2021 remained closer to water at higher temperatures.As maximum daily temperature increased, the average distance from water decreased ( P < 0.001, see Table 3 ).Activity of Rangeland goats during the 2021 Camden Park deployment decreased ( P < 0.001) with increased temperature (see Table 3 ).We did not detect a relationship ( P > 0.05) between maximum temperature and maximum distance traveled from water for the Camden Park goats during 2021.We did not detect a relationship ( P > 0.05) between maximum temperature and daily distance traveled at the Beaconsfield property.We ran analyses for distance traveled from water based on all available water sources, as well as just the water sources the goats were known to use at the Beaconsfield property.We did not detect a linear or quadratic relationship between temperature and distance traveled from water based on all water sources.The trend shown for the used water sources appeared   similar to the quadratic relationships observed at the other properties, as the temperature increased beyond 28 °C average distance traveled from water decreased ( P = 0.02) (see Table 3 ).Maximum temperature and activity of the goats on Beaconsfield were related ( P = 0.005), as temperature increased activity increased (see Table 3 ).

Discussion
Studies evaluating daily distance traveled by sheep have reported values from 1.29 km •d −1 ( Gipson et al. 2012 ) to 12.5 km •d −1 ( Squires and Wilson 1971 ).Although Gipson et al. (2012) used GPS tracking collars to determine the daily distance traveled of both sheep and goats, the study pasture was only 1.3 ha, which likely resulted in smaller values compared with the extensive rangeland pastures of this current study.Baum (2021) tracked sheep in mountainous terrain using GPS collars and found the daily distance traveled was 6.7 to 7.1 km •d −1 in spring and summer, similar to our findings despite the difference in topography.We found the daily distance traveled by goats ranged from 5.8 to 9.6 km •d −1 across all properties.The high end of this range was from the Beaconsfield deployment that allowed access to a larger pasture area and was a novel property to the goats at the time of the study.A recent GPS tracking study reported daily distance traveled by goats in a Mediterranean forest ranged from 5.62 km to 7.55 km •d −1 depending on season ( Chebli et al. 2022 ).
We found sheep and goat distribution to be greatly influenced by water location, with all properties showing significant hotspots surrounding water sources.The increased small ruminant presence surrounding water is understandable given the average distance from water for the sheep and goats barely exceeded 1 km and was often about 0.6 km for the majority of the deployments (see Table 1 ).Past studies have shown an average distance from water between 0.8 km to 3.2 km for sheep ( Squires and Wilson 1971 ;Squires 1974 ;Fensham and Fairfax 2008 ).These studies did not quantify distances based on GPS tracking but instead through observation in an experimental laneway ( Squires and Wilson 1971 ), or human surveillance and aerial photographs ( Squires 1974 ;Fensham and Fairfax 2008 ).The lower average distance to water found in our study was likely due to more precise tracking using GPS collars with locations recorded at more frequent intervals throughout the day.Across all properties, the maximum distance that sheep and goats traveled from water was between 1.28 km and 2.38 km, with the second deployment on Camden Park having the lowest maximum values.The pasture size during the 2021 deployment at Camden Park was smaller than the other deployments, which likely limited the maximum distance sheep and goats could travel from water.Other studies have found higher maximum distances to water than observed in our study.Daws and Squires (1974) reported 5.6 km for sheep and Russell et al. (2011) reported 4 km for goats.Yet again these studies did not use GPS tracking and instead relied on visual observations whether in laneways or across landscapes.Visual observations are less accurate than GPS tracking, and observers have the potential to impact animal behavior ( Anderson et al. 2013 ).
All properties in our study showed higher sheep and goat presence around water sources and riparian areas, whether containing water or not.Karabatsos (2020) studied mountain sheep in Canada and found hotspots were situated close to water sources as well, mainly near rivers.Although not quantified through hotspot analysis, other studies have shown that sheep preferred areas near water facilities ( Orr 1980 ), and goats exhibited high use around creeks and lakes ( Freudenberger and Barber 1999 ).The affinity for waterways shown by the sheep and goats in our study could be due to the higher tree cover and vegetation abundance often associated with riparian areas ( Pollock et al. 1998 ;Sabo et al. 2005 ).While studying riparian areas in Patagonia, Utrilla et al. (2006) found sheep had higher utilization of green herbaceous forage in very moist areas of the paddock.In addition, the time our study animals spent near water was not just resting, because the hotspot analyses that excluded data considered inactive were similar to that with all data.Our study indicates that water, associated riparian vegetation, and shade were some of the main drivers of sheep and goat spatial distribution.In addition to water availability and riparian vegetation, terrain is often a major factor in livestock grazing distribution ( Bailey 2004 ) but was not a factor in this study because the topography was gentle with few undulations.Future studies of small ruminant spatial distribution and grazing patterns should examine the preferences for certain land types, as well as forage quality and quantity.
For both deployments of sheep and goats on Camden Park, the western portion of the study area had obvious sections of unutilized land.The western sides, specifically northwestern, of both Camden Park pastures were the sections farthest from water.The lack of water in the unutilized parts of the pasture is the most likely explanation for the sheep and goat movement patterns.Available forage in the northwestern part of the pasture was not limiting.The low stocking rate of Camden Park likely also influenced distribution as the sheep and goats did not need to travel as far to find forage and meet their nutrient requirements.In addition, higher use in the east may be a tendency of sheep to graze into the prevailing wind ( Scott and Sutherland 1981 ).Specifically, Orr (1980) found sheep to graze into the prevailing east wind in Central western Queensland, resulting in higher utilization of the eastern areas of the pasture.Longreach has an eastern prevailing wind, a trend largely apparent in the 2021 Camden Park deployment, as the hotspots are situated along the eastern fence line.When we first started the study, the cooperating producers told us that wind played a major role in sheep distribution.They expected sheep would spend considerable time on the eastern part of the pastures.Our tracking data supported the producers' prediction based on their anecdotal observations, as well as studies such as Orr (1980) .The influence of wind on goats has not been studied extensively, but our data show goats did not explore the western side of pastures as much as the eastern side, implying that goats may graze into the eastern prevailing wind as well.Additional research on the effect of wind on small ruminant distribution, especially goats, is needed.
Before the flood, Fairfield Merino sheep had access to the three paddocks in the middle of the property.The sheep only remained outside of the three original paddocks for a month, before being herded back into the appropriate area.Therefore, we see higher sheep presence in the middle paddocks because the sheep spent the most time there.Across the Fairfield property, hotspot areas were mainly located around the large waterway running through the property, and more GPS locations were on the eastern side of the paddock.This implies the Fairfield Merino sheep grazed into the prevailing eastern wind and had higher utilization in areas around water sources, including riparian areas, congruent with the findings at Camden Park.
Goats are known to prefer browse species to forage (grasses and forbs) species ( Rutter 2002 ;Chebli et al. 2020 ) and spend more time in areas of woody vegetation ( Landsberg and Stol 1996 ;Freudenberger and Barber 1999 ).This tendency was showcased in the hotspot among the patch of Gidgee trees in the middle of the Beaconsfield property (see Fig. 3 b).Similarly in 2019, goats at Camden Park used the whole Gidgee tree patch in the southwestern corner of the property while a large part of the Gidgee patch was not used by sheep (see Fig. 1 ).
Several studies have found that sheep ( Squires 1974 ;Leu et al. 2021 ;Thomas et al. 2008 ) and goats ( Lu 1989 ;Sharma et al. 1998 ;Bojkovski et al. 2014 ) travel and graze less and rest more dur-ing periods of high temperatures.In addition, Lynch (1974) found the distance sheep traveled from water was lower during higher temperatures (35 °C vs. 10 °C).At Camden Park, we found an overarching trend of decreased daily distance traveled, distance traveled from water, and activity during periods of high temperatures.This trend was not always apparent at lower temperatures, but as temperature exceeded a threshold, sheep and goats traveled less, stayed closer to water, and were less active.The increase in daily distance traveled, distance traveled from water, or activity at lower temperatures could indicate sheep and goats increased grazing activity during cooler periods of the day to compensate for decreased grazing during high heat periods of the day.For example, Leu et al. (2021) found that in a heatwave ( > 40 °C), sheep increased distance traveled and activity before dawn and after dusk (lower temperatures) in response to decreased distance traveled and activity during midday (higher temperatures).
Similar to findings on Camden Park, activity of sheep on Fairfield increased with maximum temperature until temperature reached and exceeded 39 °C, after which activity decreased.Maximum daily temperature did not affect daily distance traveled or distance from water on Fairfield; perhaps abundance of water after flooding was a factor.Tracking at the Beaconsfield property was during autumn when temperatures were cooler and a negative relationship between daily distance traveled and temperature would not be expected.More data and further analysis are needed to fully explain the relationships with temperature on the Fairfield and Beaconsfield properties.

Implications
Spatial distribution of small ruminants is clearly affected by the location of water sources.Affinity for riparian areas, whether containing water or not, was also a driver of sheep and goat movements.To further assess the influence of vegetation on small ruminant landscape distribution, more research is needed.Our study also found that daily distance traveled, distance traveled from water, and activity of sheep and goats decreased at high temperatures at Camden Park.Both sheep and goats were attracted to areas with trees, but goats more so than sheep.More research is needed to assess if the attraction to areas with trees is a result of small ruminants seeking shade or browse.Small ruminant landscape utilization patterns described in this study will provide valuable information to livestock producers and land managers regarding development of watering locations, pasture size, and other grazing management issues (e.g., strategic supplement placement).On the basis of the findings of this study, water points should be no farther than 4 km apart to allow for optimum landscape utilization.During periods of hot weather, it may be beneficial to have water points even closer to each other to encourage continued grazing because sheep and goat movement decreased with high temperatures.

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Figure 4 .
Figure 4. Daily distance traveled by sheep versus maximum daily temperature at Camden Park located near Longreach, Australia during the 2019 study period.

Table 1
Spatial distribution metrics of sheep and goats for each property each year at Longreach, Australia.
DDT indicates daily distance traveled (m/day); DW, distance traveled from water (m); PA, percent activity (%); SD, standard deviation; Collars, number of tracked animals used for analysis; Days, number of days of tracking data used for analysis.Median, Mean, and SD are based on metrics from each day.Minimum and Maximum values are based on values from each month.

Table 2
Relationships between maximum (MaxTemp) temperature and spatial movement metrics of sheep at Longreach, Australia.

Table 3
Relationships between maximum (MaxTemp) temperature and spatial movement metrics of goats at Longreach, Australia.
ADW indicates average distance traveled from water (m); ADW (used), average distance traveled from the used water sources on Beaconsfield (m); DDT, daily distance traveled (m/d); MDW, maximum distance traveled from water (m); PA, percent activity (%); SE, standard error.