Movements of Domestic Sheep in the Presence of Livestock Guardian Dogs1

Volume 30, 2015 - July

B. L. Webber1,6,9, K. T. Weber1, P. E. Clark2,7, C. A. Moffet3,8, D. P. Ames4, J. B. Taylor3,7, D. E. Johnson5 and J. G. Kie11

1 GIS Training and Research Center, Idaho State University, Pocatello, ID 83209

2 USDA Agricultural Research Service, Northwest Watershed Research Center, Boise, ID 83712

3 USDA, Agricultural Research Service, Range Sheep Production Efficiency Research Unit, Dubois, ID 83423

4 Department of Civil and Environmental Engineering, Brigham Young University, Provo, UT 84602

5 Department of Rangeland Ecology & Management, Oregon State University, Corvallis, OR 97331

6 Research was funded in part by: USDA CSREES NRI Managed Ecosystems grant 2009-35403-05379 titled: Development of Planning and Monitoring Elements of an Adaptive Management System for Wolf-Livestock Relations; and USDA NIFA AFRI Managed Ecosystems grant 10335257, titled: Establishment of an Adaptive Management System and its Components to Address Effects of Gray Wolf Reintroduction on Ecosystem Services.

7 The use of trade, firm, or corporation names in this publication is for the information and convenience of the reader. Such use does not constitute an official endorsement or approval by the United States Department of Agriculture or the Agricultural Research Service of any product or service to the exclusion of others that may be suitable.

8 Current affiliation: USDA, Agricultural Research Service, Southern Plains Research Station, Woodward, OK 73801

9 Correspondence and current affiliation: USDA-APHIS-Wildlife Services, 4700 River Road, Unit 87, Riverdale, MD 20737-1232, USA; Phone: 1(443)521-3017; E-mail: [email protected]

Acknowledgements The authors wish to thank all those who helped with this study, the USDA ARS Range Sheep Production Efficiency Research Unit, and the Idaho State University (ISU) GeoSpatial Club for volunteering time and resources. Furthermore, the authors wish to thank Darci Han- son (president of the ISU GeoSpatial Club) and Dr. Fang Chen for their time and work on this project.

Summary group, 12 to 18 ewes were randomly out LGD present (8,210 ± 571 m vs. selected to be fitted with GPS tracking 6,797 ± 538 m, respectively; P = 0.04). Livestock guardian dogs (LGD) are collars, which were programmed to col- No other differences were detected. This one of the most effective methods avail- lect and record the ewe’s location and study demonstrated that ewes grazing able to reduce depredation on livestock. velocity at 1-s intervals. In random order, with accompanying LGD will travel The purpose of this study was to deter- each group was assigned to graze with two greater daily distances compared with mine if the presence of LGD changes LGD present for a 2-d trial period and ewes grazing without LGD accompani- grazing behavior of domestic sheep in an then graze without LGD present for a 2- ment. As a result of traveling greater dis- environment where predators are com- d trial period or vice versa. A LGD Pres- tances, ewes may also be exposed to more mon. Western white-face ewes (n = 560) ence × Day of Trial interaction was and varied foraging opportunities. with attending lambs were used. Ewes detected (P < 0.05). On Day 2 of the were 32 d and 45 d postpartum and famil- trial, ewes grazing with LGD present Key Words: Behavior, GIS, GPS, iar with LGD. Ewes were divided into traveled farther than ewes grazing with- Guardian Dog, Predator, Sheep four groups (n = 140). Within each

©2015, Sheep & Goat Research Journal Sheep & Goat Research Journal, Volume 30, 2015 - July 18 Introduction Range Sheep Production Efficiency center of the 2 × 2 grid. Topography is Research Unit (RSPER) Headquarters gently rolling with slopes ranging from 0 Increasing predator populations on near Dubois, Idaho. At the time of the percent to 20 percent and averaging rangelands in the United States have study, the RSPER maintained approxi- about 4 percent. The plant community resulted in a concomitant increase in mately 3,000 adult sheep (Rambouillet, in each pasture is dominated by three-tip livestock depredation (USDA, 2000, Targhee, Columbia, Polypay, Suffolk, sagebrush (Artemisia tripartita) and blue- 2001, 2010, and 2011). The National and crossbreeds) with additional attend- bunch wheatgrass (Psuedoroegenaria spi- Agricultural Statistics Service reported ing young (Dr. J. B. Taylor, personal cata). Annual forage production was livestock depredation resulted in losses communication). Throughout the year similar among the four pastures and esti- of $16.5 million in 1999 for the sheep sheep are grazed, with attending LGD, mated at 144 animal unit months and goat industries and $51.6 million in on various sagebrush steppe and forested (AUM) for each pasture. Except for the 2000 for the cattle industry. Over the rangelands, and are exposed to predation year of the study, the pastures were man- last 10 years, annual costs related to threats by black bears (Ursus ameri- aged similar to adjacent pastures and in depredation almost doubled, with eco- canus), mountain lions (Puma concolor), accordance with RSPER grazing man- nomic losses estimated at $20.5 million grey wolves (Canis lupus), and coyotes agement plan. in 2009 for sheep and $98.5 million in (Canis latrans) that frequent the grazing 2010 for cattle. While direct death and areas (Kozlowski, 2009). Animals, Treatment Assignment, injury losses of livestock to depredation The study area (259 ha) was located and Data Collection is of primary importance for producers, at approximately 1,670 m elevation on The use of GPS technology and stress induced in livestock exposed to the RSPER headquarters property near radio telemetry (Shivik et al., 1996) to depredation threat is also a substantial Dubois, Idaho (lat. 44°13’24”, long. map and analyze animal activities is concern. Stress may adversely impact 112°11’03”). This area was surrounded common (Morehouse, 2011), and results livestock health and productivity, and cross-fenced with 2-m high preda- from Johnson and Ganskopp (2008) including stress-reduced livestock wean- tor-proof fencing, forming four 65-ha demonstrated a positive relationship ing weight, decrease in overall animal pastures in a 2 × 2-grid, resulting in 4 between the frequency of positional data health, and increased proportions of pastures free of predators (Figure 1). collection and the accuracy of animal unusable meat (Grandin, 1989 and Sheep watering and mineral-supplement activity measurements. Subtle changes 1997). Chronic exposure of livestock to locations for each pasture were near the in activity may need more frequency in depredation threat, consequently, may adversely impact ranch profitability. Livestock producers throughout the Figure 1. Map of four pastures (65 ha each), located at the Range Sheep United States have invested in livestock Production Efficiency Research Unit near Dubois, ID. Pastures were enclosed protection strategies to mitigate eco- with predator-proof fencing and used for sheep behavioral trials investigating the nomic losses due to predation (Berger effects of livestock guardian dog (LGD) presence on the daily distance traveled 2006; Rashford et al., 2010). by sheep grazing sagebrush steppe rangelands. Water/mineral supplement and dog For sheep (Ovis aries) producers, live- food locations are marked. stock guardian dogs (LGD) are often the most effective and affordable method for substantially reducing predation (Black, 1981; Coppinger et al., 1983; Andelt, 1992, 1999; USDA, 1994; Rigg, 2001; Hansen et al., 2002; Marker et al., 2005). Not addressed in studies investigating the utility of LGD as predator deterrents for sheep flocks, was an analysis of sheep behavior (e.g., movement) in the presence of LGD. Such analysis can be informative from both an animal behavior and livestock production perspective. The objective of this study was to evalu- ate whether the presence of LGD affected daily distance traveled or percent time spent traveling by domestic sheep that were grazing sagebrush steppe rangelands. Materials and Methods

Study Area The study was conducted at the

19 Sheep & Goat Research Journal, Volume 30, 2015 - July ©2015, Sheep & Goat Research Journal sampling intervals; therefore, we used 1- s sampling intervals. Table 1. Livestock guardian dog (LGD) treatment and pasture assignments for The study flock consisted of 560 sheep groups in trials conducted at the Range Sheep Production Efficiency mature ewes with suckling lambs Research Unit near Dubois, ID. (Targhee, Columbia, Polypay, or crossbreeds of these breeds), which is about Test Start End Sheep 19 percent of the RSPER adult popula- Period Trial Date Date Group Pasture1 LGDs tion. All procedures relating to sheep 1 1 4/29/2010 4/30/2010 1 30D Present care, handling, and well being was 1 1 4/29/2010 4/30/2010 2 30A Absent reviewed and approved by the RSPER 1 2 5/2/2010 5/3/2010 1 30C Absent Institutional Animal Care and Use 1 2 5/2/2010 5/3/2010 2 30B Present Committee. Ewes were between 32 d 2 3 5/6/2010 5/7/2010 3 30A Present and 45 d postpartum when placed in the 2 3 5/6/2010 5/7/2010 4 30D Absent study pastures. Ewes were experienced 2 4 5/9/2010 5/10/2010 3 30B Absent with LGD, and prior to the study, ewes 2 4 5/9/2010 5/10/2010 4 30C Present had been continuously managed with them throughout most of the year. Ewes 1 Refer to Figure 1 for map of pasture layout. and attending lambs were randomly assigned to 4 groups (Groups 1, 2, 3, and 4) of 140 ewes each. Groups were stud- positional accuracy for collar type #2 was placed in the diagonally-opposed pasture ied in specific trial periods called Trials ± 3.56 m at 95 percent CI. All GPS col- (30A) (LGD present). Trial 4 began on 1, 2, 3, and 4, which are described in lars were programmed to collect and 9 May, 2010 with Group 3 in the south- more detail below and presented in record the collared ewe’s location and west pasture (30C) and Group 4 in the Table 1. Groups 1 and 2 were used during velocity at 1-s intervals during the trials. diagonally-opposed pasture (30B). The Trials 1 and 2 and Groups 3 and 4 were The date and time of each location LGD were placed with Group 3 in pas- used during Trials 3 and 4. During each record, along with additional quality ture 30C; Group 4 did not have LGD trial, groups were placed in diagonally parameters such as the number of GPS present for this final trial. Consequently, adjacent pastures, 1 group with 2 LGD satellites used to calculate the location, the experiment was repeated during 2 (an Akbash and an Akbash/Great Pyre- were also recorded. test periods in which Period 1 included nees cross) and 1 group without LGD. Each trial was 2 d (48 h) in duration Trials 1 and 2 and involved sheep At the end of each trial, groups were (Table 1) and was immediately preceded Groups 1 and 2, while Period 2 included moved to opposing pastures, and the by a 12-h pretrial period during which Trials 3 and 4 and involved sheep LGD were placed with the previously ewes and attending young were moved Groups 3 and 4. The LGD treatment unaccompanied group of sheep. Follow- into the trial pastures and allowed to assignments were reversed between ing Trial 2, ewes and lambs were explore and acclimate to the pasture Period 1 and Period 2 to allow separation removed from the pastures and the environment. Trial 1, involving Groups of LGD and Period effects. experiment was replicated with the 1 and 2 began on 29 April, 2010. Group remaining 2 groups of sheep, Groups 3 1 was accompanied by 2 LGD and placed Data Processing and Analyses and 4, utilizing the same LGD. Through- in the southeast pasture (30D; Figure 1), Data were downloaded from the out the course of this study, ewes were Group 2 was not accompanied by LGD GPS collars and imported into a spread- provided ad libitum access to water and and placed in the diagonally-opposed sheet for error checking. Errors caused by mineral supplements. LGD were also pasture (30A) to minimize or eliminate GPS low-battery conditions, power provided ad libitum access to water interaction between groups. Trial 2 interrupts, signal loss, and multi-path (shared with the sheep) and dog food. began on 2 May, 2010 involving Groups interference effects were detected and A random selection of ewes from 1 and 2. Before the trial and pretrial removed using the following procedure. each group were fitted with GPS track- acclimation period, Group 1 was moved First, an initial screening was conducted ing collars, (n = 12, 18, 12, and 18, to the southwest pasture (30C) and to identify and remove corrupted data, respectively for Groups 1, 2, 3, and 4). Group 2 was moved to the diagonally- which were readily recognized as strings The somewhat disproportional sampling opposed pasture (30B). The LGD were of random characters instead of numeric design was due to logistical difficulties moved from pasture 30D and placed in positional data. Second, geospatial con- experienced as ewes were collared and pasture 30B with sheep Group 2. At the sistency testing was applied by importing placed in pastures. The average age of end of Trial 2, the GPS collars from the screened data as point features into a collared ewes was 2.6 yr (SE = 0.10). sheep Groups 1 and 2 were removed, GIS and projected into Idaho Transverse Two GPS collar types were distributed GPS data were downloaded to a com- Mercator (IDTM) NAD83 coordinate between each set of Groups and later puter, collar batteries were replaced, and system. These point vectors were over- determined to be distinguishable by their the collars were then placed on ewes in laid on a GIS layer representing the level of positional accuracy. Horizontal Groups 3 and 4. Trial 3 began on 6 May, boundaries of the study site pastures. positional accuracy for collar type #1 was 2010, with Group 3 in the southeast pas- Points falling outside the perimeter of ± 4.45 m at 95 percent CI. Horizontal ture (30D) (LGD absent), and Group 4 the study pasture and at a distance

©2015, Sheep & Goat Research Journal Sheep & Goat Research Journal, Volume 30, 2015 - July 20 greater than the GPS horizontal accu- ence on daily distance traveled and the Collar Type × Day of Trial, Collar Type racy for that particular collar type were percentage time spent moving by domes- × LGD, and Collar Type × Day of Trial tagged as erroneous and removed from tic sheep was analyzed using a mixed lin- × LGD were not significant (P = 0.84, the data set. On average only 0.27 per- ear model procedure (Baayen et al., 0.89, and 0.99, respectively). The effect cent of the points from each 2-d (48-h) 2008; Littell et al., 1998; Singer, 1998) of Collar Type was more of a function of trial were removed due to this error. The within SAS PROC MIXED (SAS Soft- technology accuracy/sensitivity rather data from within the study pastures were ware v. 9.2; SAS Inst. Inc. Cary, N.C.; than an effect on ewe-grazing behavior classified into stationary (< 0.09 m/s) or SAS, 2011). Both the daily distance- in the presence or absence of LGD. non-stationary (moving). Speed meas- traveled model and the percentage-time- Least squares means for distance urements used were determined from spent-moving models included LGD traveled are presented in Table 2. The outputs calculated by the GPS units, presence, Day of Trial, Collar Type, and effect of LGD and LGD × Day of Trial using Doppler shift and positional all their interactions as fixed effects and was significant (P = 0.04), but effect of change calculations (Townshend et al., Period as a random effect. Individual Day of Trial was not significant (P = 2008). The third error-removal process sheep were considered the sample units 0.97). Ewes that were grazing with LGD used GIS to convert the time-stamped or subjects in both models. The interac- present traveled a greater distance than positions or point features from each tion of Trial and Day of Trial was used as ewes grazing without LGD present (P = sheep into a line feature representing the the repeated measure term in both mod- 0.04). When considering the interaction, movement path of the sheep. Because all els. In both models, Shapiro-Wilk tests the results further indicated that this dif- GPS data contain some amount of posi- indicated the model residuals met nor- ference was mainly a function of Day 2 of tional error, the GPS positions collected mality assumptions. Mean separations the trial. Lack of Day of Trial effect indi- from a completely stationary collared were accomplished using a Tukey- cated that simply being in the pasture animal will tend to wander about rather Kramer adjustment to account for from Day 1 to Day 2 did not determine than all the positions falling on the sin- unequal sample sizes. All differences distance traveled. These results addressed gle, true stationary location of the ani- reported in this article were significant our original question, “Does the presence mal. To minimize the number of these at P < 0.05. of LDG affect daily distance traveled by erroneous positions, each line was sim- grazing domestic sheep?” plified by removing line vertices that Results and Discussion were within 1 m of the preceding vertex. Percentage of Time Traveling This distance threshold value was Distance Traveled Livestock guardian dog presence (P selected because it was well within the = 0.32), Day of Trial (P = 0.49), Collar known accuracy of the GPS chipsets and All data used in the analysis met the Type (P = 0.07), and corresponding thereby removed erroneous positions, Shapiro-Wilk’s test of normality (P > interactions (LDG × Day of Trial, P = while preserving actual movement 0.05). Collar Type was included in the 0.78; LDG × Collar Type, P = 0.93; Day observations. The intended result of this model to account for variation of the 2 of Trial × Collar Type, P = 0.78; and line simplification procedure was to types of collars used. Collar Type was LDG × Day of Trial × Collar Type, P = remove most of the positions or vertices found to be significant (P <0.05), while 0.39) did not significantly affect per- of stationary animals except for the initial position, when the animal first became stationary. Removal of these stationary positions, which were extra- Table 2 Least squares means1 (SE) of distance traveled (m) by ewes2 that were neous, would thus prevent the GPS error grazing rangeland in the absence or presence of livestock guardian dogs3 associated with these positions from (LGD). inflating the recorded movement budget and daily travel distance for the animal. Day of Trial The length of each simplified line was LGD 1 2 Day 1 vs.2 (P-value) recorded as the daily travel distance, in Absent 7,515 m (495) 6,797 m (538)y 0.49 meters, for each collared sheep. Present 7,517 m (465) 8,210 m (517)x 0.44 Preparation of data for percentage Absent vs. Present time spent in each speed class (station- (P-value) 1.00 0.04 ary or non-stationary) required a final error-removal step, whereby data a, b Unlike superscripts within respective row indicate that means were different describing unrealistic speeds (> 9 m/s) (α = 0.05). were identified and removed. The maxi- x, y Unlike superscripts within respective column indicate that means were mum number of points removed at this different (α = 0.05). step was < 200 (0.23%) per daily collec- 1 The effect of the LGD × Day of Trial interaction was significant (P < 0.05). tion period (n = 86,400). 2 Ewes were accompanied by suckling young. Breeds with groups were Targhee, Statistical Analyses Columbia, Polypay, or western whiteface crossbreeds. 3 Livestock guard dog breeds were Akbash and an Akbash/Great Pyrenees cross. The treatment effect of LGD pres-

21 Sheep & Goat Research Journal, Volume 30, 2015 - July ©2015, Sheep & Goat Research Journal centage of daily time that ewes spent (distance traveled) has been altered by Andelt, W. F. 1999. Relative Effective- traveling vs. remaining stationary. Ewes the presence of LGD. ness of Guarding-Dog Breeds to accompanied by LGDs spent a mean of Deter Predation on Domestic Sheep 27.6 percent of the daily time traveling, Conclusion in Colorado. Wildl. Soc. Bull. which was similar to the mean (25.9 per- 27(3):706-714. cent) for ewes grazing without LGDs Animals grazing in areas with high- Baayen, R. H., D. J. Davidson, and D. M. present. Since LGDs are not herding predator populations may continually be Bates. 2008. Mixed-effects modeling dogs this result was anticipated. A placed under acute or chronic stress by with crossed random effects for sub- related study (Jensen et al. 2013) sug- either direct predation attempts (e.g., jects and items. J. Mem. Lang. gests the presence of LGDs affects over- pursuit events) or fear memories of preda- 59(4):390-412 all flock fidelity, where LGD presence tion (Grandin, 1998). This stressful state Berger, K. M. 2006. Carnivore-Livestock may decrease individual vigilance activ- may have a negative impact on the over- Conflicts: Effects of Subsidized all health of livestock, including reduc- ity and allow broader pasture utilization. Predator Control and Economic tions in weight gain, increased disease Jensen’s study demonstrated flock associ- Correlates on the Sheep Industry. susceptibility, and lowered reproductive ation decreased with the presence of Conserv. Biol. 20(3):751-761. success. The end result is an economic LGDs suggesting a more dispersed pat- Black, H. L. 1981. Navajo Sheep and loss to the rancher and the livestock tern of movement. Goat Guarding Dogs: A New World These results suggested that the pres- industry. Solution to the Coyote Problem. ence of LGD influenced the daily dis- This study demonstrated that ewes Ranglands. 3(6):235-237. tance traveled by ewes grazing sagebrush grazing with accompanying LGD will Coppinger, R., J. Lorenz, J. Glendinning, steppe rangelands. Ewes traveled farther travel greater daily distances compared and P. Pinardi. 1983. Attentiveness when accompanied by LGDs than with- with ewes grazing without LGD accom- out LGD accompaniment. There are paniment. As a result of traveling greater of Guarding Dogs for Reducing Pre- many factors that could help explain this distances, ewes may also be exposed to dation on Domestic Sheep. J. Range observation: 1. sheep without LGD may more and varied foraging opportunities. Manage. 36(3):275-279. remain near areas previously proven safe The observed changes in movement Grandin, T. 1989. Behavioral principles from predation; 2. sheep without LGD behavior may result in more effective use of livestock handling (with 1999 may be trying to remain in close proxim- of pasture resources. The more effective updates on vision and hearing in ity to other sheep (Sibbald et al. 2008) as use of pasture may result in the increase cattle and pigs). The Professional a safety mechanism (e.g., herd effect); in the net rate of nutrient intake, which Animal Scientist. December and/or 3. sheep with LGD may be more could also lead to increase health of the 1989:1-11. mobile, as they spend less time being animals. While it is unknown if the ani- Grandin, T. 1997. Assessment of stress attentive to danger and spend more time mal utilized the varied foraging opportu- during handling and transport. J. grazing and moving. nities presented, this study offers insight Anim. Sci. 75:249-257. Exploring ewe-movement speed into domestic animal interactions that Grandin, T. 1998. Review: reducing showed no difference in the percent time may also help direct future studies. handling stress improves both pro- ewes spent stationary or traveling rela- Research by Grandin (1989, 1997, ductivity and welfare. The Profes- tive to the presence of LGD. This is and 1998) and Coppinger (1983) has sional Animal Scientist. 14:1-10. important to note, because while sheep changed the way livestock are managed. Hansen, I., T. Staaland, and A. Ringso. with LGD traveled farther, they did not This study offers another step toward 2002. Patrolling with Livestock spend significantly more time travelling. improving the health of domestic live- Guard Dogs: A Potential Method to This in turn suggested sheep with LGDs stock, as well as increased awareness of Reduce Predation on Sheep. Acta the benefits of LGD. If the presence of tended to move at higher velocities than Agric. Scand. A. 52:43-48 LGD is shown to increase weight gains, sheep without LGDs. However, the data Jensen, D., K. T. Weber, and J. B. Taylor. improve animal health, and increase developed for this study cannot support 2013. Association of Domestic such granularity. Nevertheless, our find- lamb weaning weights, then the use of LGD will carry increased economic Sheep Flocks in the Presence of ings support the hypothesis that sheep Livestock Guardian Dogs. URL = with LGD spend less time being vigilant importance to the livestock industry. http://giscenter.isu.edu/research/Tec for predators and more time moving, While currently only speculation, these hpg/LGD/pdf/SheepAssociation05. although a more in-depth study needs to questions should be investigated in future pdf visited 9-April-2015. 7pp. be done to determine animal activity studies.. budgets. Consequently, the presence of Johnson, D. D. and D. C. Ganskopp. LGD may offer more than just protec- Literature Cited 2008. GPS collar sampling fre- tion for domestic livestock. Their pres- quency: effects on measures of ence may result in less restricted move- Andelt, W. F. 1992. Effectiveness of resource use. Rangeland Ecol. ment and decreased stress. While this Livestock Guarding Dogs for Reduc- Manag. 61(2):226-231. study cannot show any direct positive ing Predation on Domestic Sheep. impact on the general health of domestic Wildl. Soc. Bull. 20(1):55-62. sheep it does show that sheep behavior

©2015, Sheep & Goat Research Journal Sheep & Goat Research Journal, Volume 30, 2015 - July 22 Kozlowski, S. 2009. Draft U.S. Sheep Rigg, R. 2001. Livestock guarding dogs: USDA. 2000. National Agricultural Experiment Station Grazing and their current use worldwide. Statistics Service: Sheep and Goats Associated Activities Project 2009 IUCN/SSC Canid Specialist Group Predator Loss. http://usda.mannlib. Biological Assessment and Wildlife Occasional (Online) Paper No. 1. cornell.edu/MannUsda/viewDocu- Report. http://www.ars.usda.gov SAS Institute. 2009. SAS/STAT(R) 9.2 mentInfo.do?documentID=1627 /SP2UserFiles/Place/53640000/USS user’s guide, second edition. Cary, (Accessed 13 April 2014) ESPROJECT/20091123_ARS- NC, USA: SAS Institute, Inc. USDA. 2001. National Agricultural BA_wl_report-draft.pdf Shivik, J. A., M. M. Jaeger, and R. H. Statistics Service: Cattle Predator Littell, R. C., P. R. Henry, and C. B. Barrett. 1996. Coyote Movements Loss. http://usda.mannlib.cornell. Ammerman. 1998. Statistical analy- in Relation to the Spatial Distribu- edu/MannUsda/viewDocument sis of repeated measures data using tion of Sheep. J Wildl. Manage. Info.do?documentID=1427 SAS procedures. J. Anim. Sci. 60(2):422-430. (Accessed 13 April 2014) 76:1216-1231. Sibbald, A. M., S. P. Oom, R. J. Hooper, USDA. 2010. National Agricultural Marker, L. L., A. J. Dickman, and D. W. and R. M. Anderson. 2008. Effects Statistics Service: Sheep and Goats MacDonald. 2005. Perceived Effec- of social behavior on the spatial dis- Predator Loss. http://usda.mannlib. tiveness of Livestock-Guarding tribution of sheep grazing a complex cornell.edu/MannUsda/viewDocu- Dogs Placed on Namibian Farms. J. vegetation mosaic. Appl. Anim. mentInfo.do?documentID=1628 Range Manage. 58(4):329-336. Behav. Sci.. 115:149-159. (Accessed 13 April 2014) Morehouse, A. T. and M.S. Boyce. 2011. Singer, J. D. 1998. Using SAS PROC USDA. 2011. National Agricultural Venison to beef: seasonal changes in MIXED to fit multilevel models, Statistics Service: Cattle Predator wolf diet composition in a livestock hierarchical models, and individual Loss. http://usda.mannlib.cornell. grazing landscape. Front. Ecol. Env- growth models. J. Educ. Behav. Stat. edu/MannUsda/viewDocument iron. 9:440-445. 23(4):323-355. Info.do?documentID=1625 Rashford, B. S., T. Foulke, and D. T. Tay- Townshend, A. D., C. J. Worringham, (Accessed 13 April 2014) lor. 2010. Ranch-Level Economic and I. B. Stewart. 2008. Assessment Impacts of Predation in a Range of Speed and Position during Livestock System. Rangelands. Human Locomotion Using Nondif- 32(3):21-26. ferential GPS. Med. Sci. Sports Exerce. 40(1):124-132. USDA Animal and Plant Health Inspection Service. 1994. A Produc- ers Guide to Preventing Predation of Livestock. USDA Agriculture Information Bulletin Number 650.