Evaluation of Wire Fences for Coyote Control BRUCE C. THOMPSON Abstract Thirty-four electric an nonelectric wire fence configurations were evaluate for eterrent effect to coyotes (Can& l&runs). Tests of fences were conucte using a conitione test regime or live prey to elicit fence-crossing responses from 15 captive coyotes uring 98 exposures to fences. Fence height an mesh size were important factors in controlling jumping over an crawling through, respectively. Overhangs an aprons were necessary to preclue climbing over an crawling uner fences. Electric fences generally were not effective eterrents uner test conitions. Construction of a fence necessary to eter all methos of crossing is escribe. Subsequent fiel tests have verifie the suitability of such a fence to control losses of sheep to coyotes. Uses of wire fences to protect livestock an fiels were escribe in the Unite States (Youngbloo an Cox 1922; Young an Jackson 195 l), Africa (Wooley 1965; Denney 1972), an Australia (Bauer 1964; NIcKnight 1969). McAtee (1939) an Fitzwater (1972) escribe small-scale uses of electric fencing to exclue ungulates an preatory animals. Storer et al. (1938) objectively evaluate the use of electric wires to control bears an ocumente some success, although reactions of bears were variable. Gates et al. (1978) escribe an effective anti-coyote electric fence that completely protecte penne sheep from four coyotes for several weeks. Electric fencing reportely was successful in reucing fox preation on nesting terns (Forster 1975). Lantz (195) evaluate 14 fences as eterrents to captive coyotes, but his ata were insufficient to etermine effectiveness. Jarine ( 19% 19 11) teste two fence configurations use for sheep enclosures in Oregon an Colorao. He state that a 15-cm high V-mesh fence was aequate to exclue coyotes. Stullken an Kirkpatrick (1953) stuie the relationship of mesh size to passage by selecte preatory mammals but exclue coyotes from their tests. Despite the reporte wiesprea use of fences for control of epreations, evaluations of effectiveness have been limite. This research was initiate to test objectively numerous configurations of electric an nonelectric fences uner controlle conitions, thereby permitting jugment of the effectiveness of fences in retaring or preventing crossing by coyotes. The research reporte in this paper was conucte while the author was a grauate research assistant at Oregon State University. Thompson is presently a grauate teaching assistant at Texas A&M University, Department of Willife an Fisheries Sciences, Texas A&M University, College Station 77843. This research was supporte through U.S. Fish an Willife Service Contract No. 14-16-8-834. Oregon Agricultural Experiment Station Technical Paper No. 478. P.A. Vohs an M. Cropsey initiate the project, J.A. Crawfor aie in initial sampling esign, an B.J. Verts supervise the research. B.J. Verts an S.B. Linhart critically reviewe early rafts of the manuscript. Manuscript receive October 5, 1978. Met hos Thirty-four test fences (TF), comprising seven groups (Fig. l), were teste from April 1975 through March 1976. Test fences were esigne to represent moifications of existing fencing or new construction. To represent existing fences, a stanar sheep fence (SHF) an a stanar stock fence (STF) were efine base on information gathere uring an informal survey of fences use by stockmen in Oregon s Willamette Valley. The stanar fences were constructe of 99-cm woven wire, with barbe wires at groun level an 7.6 cm above the woven wire. The STF (TFl) an SHF (TF5) ha 15.2-cm an 3.5-cm stay woven wire, respectively. Post interval was about 4.6 m. Group II inclue five configurations previously use in Oregon an Colorao (Jarine 198-1911; M.G. Cropsey pers. comm.). International 65 an Ho1 Dem Moel 69 fence chargers powere by a -volt wet cell bettery were use to charge electric wires. The woven wire was groune (metal ro burie about 1.3 m) when use in combination with electric wires. Detaile escriptions of all fences were presente previously (Thompson 1976, 1978). Tests of fences were conucte in three phases using 15 wil-caught ault coyotes. The feet of all coyotes caught in steel traps were allowe to heal prior to initiation of tests of fences; no impairment of foot movement persiste after healing. Coyotes were fe about 45 g of chunk-style ry og chow aily, supplemente with car-kille mammals. The coyotes were confine accoring to U.S. Department of Health, Eucation, an Welfare (1972) stanars, an all coyotes survive the confinement perio. Tests of fences were conucte insie a 6.5-ha chain-link fence enclosure locate 17.7 km northwest of Corvallis, Oregon. For Phase 1 tests, 1 coyotes were conitione to traverse a specific route when release from their cages (Thompson 1978, Fig. 1), after which test fences were place in the path of travel. The conitioning process was similar to instrumental conitioning (Thorpe 1963) an was escribe previously (Thompson 1978). Valiation proceures, consisting of exposure to easily crosse fences, were use throughout this phase to ensure that coyotes remaine conitione espite failure to cross fences. Fence groups I to VII were teste sequentially; tests of each group were arrange in a ranomize block esign (RBD). Coyotes were expose to fences for lo- or 15-minute test perios an were observe from an elevate blin. The number of failures to cross each fence were ajuste to a normal istribution using (fi+ x) an analyze using two-way ANOVA to account for variation introuce by possible changes in coyote behavior (Sneecor an Co&ran (1967:325). When F tests were significant, fences were ranke using Duncan s (1955) new multiple range test (NMRT) to permit selection of similarly effective fences for further evaluation. Fences that ranke highest an were statistically similar base on groupe ata were rate base on cost, material availability, ease of construction, an relative effectiveness. The eterrent effects of comer shiels an overhangs were evaluate with one egree-of-freeom Chi-square tests. Use of trae names oes not imply enorsement of proucts. JOURNAL OF RANGE MANAGEMENT32(6), November 1979 457
GROUP I GROUP II n A GROUP III GROUP IV _ 456 JOURNAL OF RANGE MANAGEMENT32(6), November 1979
GROUP, V GROUP VI R GROUP VII, - BARBED I IWOVEN I I --&ELECTRIC CHICKEN. SCALE (cm) Fig. 1. Test fence configurations comprising Groups I through VII. View from coyote sie offences. Numbers ientify test fences as referre to in text. Fences 26 through 34 ha triangular plywoo shiels in corners but not epicte here. Post intervals are not to scale. JOURNAL OF RANGE MANAGEMENT32(6), November 1979 459
Phase 2 involve tests of the top-rate electric an nonelectric fences that were selecte in Phase 1 an a control, the SHF. A tethere live rabbit was use as the stimulus for coyotes to negotiate test fences. Five coyotes, not use in Phase 1, that ha kille live prey after negotiating woven wire fences were use for these tests. For each test, coyotes were eprive of foo for 4 ays, then release in the test area for 3 minutes or until a kill occurre. Tests of fences were arrange in RBD; the number of failures were transforme an analyze as above. Mean separation was accomplishe using least significant ifference (Sneecor an Cochran 1967:272). Differences observe uring Phase 1 an 2 tests were consiere significant at a =.5. In Phase 3, the most effective fence selecte in Phase 2 initially was expose to coyotes in the confinement fencing. Seconly, two.2-m square pens within the large enclosure were fence with the SHF an the test fence, an coyotes were allowe to roam in the large enclosure overnight. Coyotes that kille rabbits in Pen 1 (SHF) were later allowe to roam overnight with a rabbit tethere in Pen 2 (test fence). Results Phase 1 Test fences 2,3,6,7,1,26,28,29, an 3 were significantly greater eterrents to crossing by coyotes than other fences in their respective groups an were retaine for further evaluation (Table 1). There were no significant ifferences in eterrent effect among fences in Groups IV, V, an VII; therefore, all were retaine. Data for 23 fences were groupe for further analysis, an TF 26,28,29,3,31,32,33, an 34 were significantly greater eterrents to coyotes than were other fences. ANOVA was base on ata for seven coyotes that were vali uring tests of all fences. I consiere ata grouping legitimate because there was never significant variation between replications, an coyotes reacte similarly uring all tests. Test fences 28 an 33 were the highest-rate electric an nonelectric fences, respectively; thus they were selecte for testing in Phase 2. An electric fence an a nonelectric fence were chosen to permit further comparison of the effectiveness of the two types of fencing. Throughout all tests, there was a significant positive correlation (r=.87, PcO.1) between the percentage of coyotes that faile to cross fences an the height of fences. A threshol existe at about 168 cm because coyotes frequently hit fences when attempting to jump that height or more. Coyotes crosse fences at corners significantly more often than they crosse elsewhere along the fences uring tests of Group I, Group II, an Group IV. This was attribute to the extent to which coyotes use horizontal corner braces as toe-hols. Coyotes crosse fences at locations other than comers significantly more often in tests of Group VI when comer shiels were ae, an in tests of Group VII when overhangs an/or comer shiels were use. During tests of four types of overhangs, I foun that an effective overhang require no larger than 15.2-cm stay woven wire an extene at least 38 cm from the vertical fence. Larger mesh or less extension permitte coyotes to crawl through or climb over the overhang, respectively. Tests of electric fences were inconclusive, but use of electric wires apparently was ineffective in excluing coyotes. Generally, it was ifficult to position electric wires so they woul ensure that coyotes receive a shock. This ifficulty was emonstrate by infrequent receipt of shocks (13 tests of 466 tests) uring Phase 1 an was not cause by avoiance of electric fences because coyotes were similarly successful in crossing electric an nonelectric fences (Table 1). Difficulty in Table 1. Number of tests, failure rates, an uring phase 1, April to October 1975. Group Fence No. No. of tests rankings of 34 fences Failures of coyotes to cross fences No. % Ranking a I 1 3 8 27 2 3 14 47 3 3 11 37 5 4 1 3b 2b 4 3 8 27 LP 5 3 5 17 ~ II 6 3 7 3 8 3 9 9 3 9 1 3 11 III 11 13 14 15 IV 16 17 18 19 3 3 3 3 3 11 1 3 8 9 1 6b 7b 3-37 C 37 33 V 21 27 1 37 22 27 11 41 23 27 11 41 24 27 11 41 25 27 44 VI 26 27 15 27 27 28 27 14 29 27 15 3 27 15 VII 31 24 16 67 32 24 17 71 33 24 17 71 34 24 17 71 55 44 52 27 28 b 26 b 29 b 3 b 55-55 a New multiple range test (Duncan 1955). Fences unerscore by the same line are not significantly ifferent. b Test fences retaine for further evaluation. c All fences iscare from further evaluation. No ifferences between fences, all retaine for evaluation. aministering electric shocks i not appear attributable to rapi crossings by conitione animals, because coyotes frequently spent 2 to 5 minutes in the test area before crossing electric fences. Apparently coyotes were able to avoi electric wires when crossing, at least with the fences teste. Phase 2 Each fence was teste 15 times an 14, 11, an 6 failures were recore for TF 33,28, an 5 respectively. Test fence 33 was a significantly greater eterrent to coyotes than TF 28 or the SHF, thus TF 33 was selecte for further testing. The single crossing of TF 33 occurre when a small female crawle through after stretching the mesh. Phase 3 During the first part of Phase 3, coyotes were release in the test area overnight eight times. One coyote crosse TF 33 by crawling uner after excavating a epression an stretching wires near a comer. A 45.7-cm woven wire apron was ae to TF 33 an five more overnight tests were conucte. One coyote crosse the test fence uring these tests, apparently by 46 JOURNAL OF RANGE MANAGEMENT32(6), November 1979
crawling through the mesh because no igging was apparent. This coyote also crawle through TF 33 uring Phase 2. During the secon part of Phase 3, two of the five coyotes kille rabbits in Pen 1. These two coyotes were allowe to roam singly in the large enclosure overnight with a rabbit tethere in Pen 2 enclose by TF 33 with apron. One coyote crosse the fence an kille a rabbit uring each of two overnight tests. The other coyote i not cross the fence uring four overnight tests. The coyote that entere Pen 2 was the same animal that ha crosse (crawle through) TF 33 three times previously. Discussion The emonstrate aeptness of coyotes to use a variety of fence crossing methos, the variation in livestock confinement practices, an the high cost of certain fence materials complicate the evelopment of an effective fence. The ability of some coyotes to crawl through unstretche meshes as small as 15.2 x 1.2 cm was a critical consieration because woven wire typically use for fencing livestock has such openings within 15 cm of the bottom, thus permitting easy access to coyotes. I use rabbit an poultry wire to obtain small mesh wire that was relatively low cost an commercially available in quantity. The largest openings in this mesh were 15.2 x 1.2 cm; openings below 5 cm were smaller. Even this mesh i not preclue one coyote from crawling through. I i not use V-mesh for test fences 6 through 9 as originally escribe by Jarine because it was not easily obtaine uring my tests. I believe the V-mesh woul preclue coyotes from crawling through, but excessive cost (>$3 per meter) an limite prouction reuce its utility. Unobstructe comers presente a problem because of the toe-hols an stepping-points available at the fence juncture an on bracing apparatus. Use of very narrow braces situate iagonally an as low as possible may reuce the steppingstone avantage to coyotes. It i not appear that placing wires on one sie of posts versus the other ha any bearing on coyotes ability to cross fences. However, coyotes were wary of anything hanging irectly overhea, thus comer shiels an overhangs tene to keep coyotes away from fences an reuce the probability of crossing. Comer shiels may be an economical an useful aition to fences that coyotes are known to cross. In general, moifying existing fences by aing wires to the posts i not appear to be effective eterrents because coyotes continue to be able to contact the woven wire irectly. Aing outlying electric an/or nonelectric wires near existing fences may provie some eterrent effect ue to avoiance. Electric shock alone may not eter coyotes that are accustome to killing livestock, although recent research inicate otherwise (Gates et al. 1978). New fence construction shoul be given primary consieration for protecting livestock from coyotes because it can be esigne to eter all methos of crossing, an it eliminates initial familiarity to coyotes that is present when existing fences are moifie. A greater egree of flexibility exists with new construction because existing fences are not incorporate; however, costs are higher. Those I teste woul cost $1,2 to $1,525 per km for materials in August 1978. In contrast, aing three to five electric wires woul cost $155 to $245 per km for materials plus $3 to $5 for a fence charger. Aing a single barbe wire woul cost about $7 per km. There is wie variation in costs relative to geographic area, purchase quantities, an type of labor use. Base on my results, an effective fence shoul be at least 168cm high, have meshes no larger than 15.2 x 1.2 cm (preferably smaller), an have an overhang an apron projecting at least 38 cm. The overhang an apron shoul be 15.2-cm stay or smaller woven wire. Aprons of use wire woul eliminate ae cost. Shiels also shoul be ae to further obstruct comers. This fence may not exclue all coyotes as evience by one coyote that crosse it uring tests; however, I believe it woul eter all but the most exceptional coyotes uner fiel conitions. Also, such a fence woul make exiting ifficult for any coyote that crosse it. DeCalesta an Cropsey (1978) fiel teste my recommene esign an obtaine favorable results uring the first year of exposure in two areas subject to coyote epreation problems. These authors also presente a preliminary cost-benefit analysis base on current material costs an the reuction of sheep losses observe uring their stuy. It was suggeste that fencing coul be avantageous at least for enclosing large areas or areas use by sheep uring highly vulnerable perios. Thus, fencing is a promising nonlethal control metho, but iniviual proucers must ultimately ecie whether it woul be practical uner current financial conitions, market fluctuations, an levels of losses. Literature Cite Bauer, F.H. 1964. Queenslan s new ingo fence. Aust. Geogr. 9:244-245. ecalesta, D.S., an M.G. Cropsey. 1978. Fiel test of acoyote-proof fence. Will. Sot. Bull. 6:256-259. Denney, R.N. 1972. Relationships of willife to livestock on some evelope ranches on the Laikipia Plateau, Kenya. J. Range Manage. 25415425. Duncan, D.B. 1955. Multiple range an multiple F tests. Biometrics 11: l-42. Fitzwater, W.D. 1972. Barrier fencing in willife management. Proc. Vertebr. Pest Conf. 5:49-55. Forster, J.A. 1975. Electric fencing to protect sanwich terns against foxes. Biol. Conserv. 7:85. Gates, N.L., J.E. Rich, D.D. Gotel, an C.V. Hulet. 1978. Development an evaluation of anti-coyote electric fencing. J. Range Manage. 3 1: 151-153. Jarine, J.T. 198. Preliminary report on grazing experiments in a coyoteproof pasture U.S. Dep. Agr. Forest Serv. Circ. 156. 32 p. Jarine, J.T. 199. Coyote proof pasture experiment, 198. U.S. Dep. Agr. Forest Serv. Circ. 16. p. Jarine, J.T. 191. The pasturage system for hanling range sheep. U.S. Dep. Agr. Forest Serv. Circ. 178. p. Jarine, J.T. 1911. Coyote-proof inclosures in connection with range lambing grouns. U.S. Dep. Agr. Forest Serv. Bull. 97. 32 p. Lantz, D.E. 195. The relation of coyotes to stock raising in the West. U.S. Dep. Agr. Farmers Bull. 226. 23 p. McAtee, W.L. 1939. The electric fence in willife management. J. Will. Manage. 3:1-13. M&night, T.L. 1969. Barrier fencing for vermin control in Australia. Geogr. Rev. 59:33-347. Sneecor, G.W., an W.G. Cochran. 1967. Statistical Methos. Iowa State Univ. Press, Ames. 593 p. Storer, T.I., G.H. Vansell, an B.D Moses. 1938. Protection of mountain apiaries from bears by use of electric fence. J. Will. Manage. 2:172-178. Stuliken, D.E., an C.M. Kirkpatrick. 1953. A etermination of hole sizes which exclue certain preatory animals. J. Will. Manage. 17:4-8. Thompson, B.C. 1976. Evaluation of wire fences for control of coyote epreations. MS Thesis. Oregon State Univ., Corvallis, 59 p. Thompson, B.C. 1978. Fence crossing behavior exhibite by coyotes. Will. Sot. Bull. 6:14-17. Thorpe, W.H. 1963. Learning an Instinct in Animals. Harvar Press, Cambrige, Mass. 558 p. U.S. Department of Health, Eucation, an Welfare. 1972. Guie for the care an use of laboratory animals. DHEW Publ. No. NIH. 73-23. 56 p. Wooley, F.W. 1965. Game efence barriers. E. Afr. Will. J. 3:89-94. Young, S.P., an H.H.T. Jackson, 1951. The Clever Coyote. The Stackpole Co., Harrisburg, Pa., an The Willife Management Institute, Washington, D.C. 411 p. Youngbloo, B., an A.B. Cox. 1922. An economic stuy of a typical ranching area on the Ewars Plateau of Texas. Texas Agr. Exp. Sta. Bull. 297. 437 p. JOURNAL OF RANGE MANAGEMENT32(6), November 1979 461