DEVELOPING AND IMPLEMENTING EFFECTIVE BLACK BEAR EXCLUSION FENCES TO PROTECT MOBILE APIARIES By Tammy E. Otto A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Fisheries and Wildlife – Master of Science 2013 ABSTRACT DEVELOPING AND IMPLEMENTING EFFECTIVE BLACK BEAR EXCLUSION FENCES TO PROTECT MOBILE APIARIES By Tammy E. Otto Growing and expanding American black bear (Ursus americanus) populations and increased societal demand for honey bee pollination services have resulted in conflict between bears and beekeepers. Developing devices to protect mobile beehives from black bears is important for reducing economic losses to beekeepers and for conflict mitigation. I tested the effectiveness of 4 portable electric fence designs for excluding black bears from bait sites in northern Michigan, 2010. I determined the effectiveness of each fence design by observing bear behavior obtained from 24-hr video surveillance. From > 433 minutes of bear-fence interactions (BFI), I recorded 168 BFIs in 73 visits by an estimated 15 bears. The only fence design deemed 100% effective consisted of 3 polytape strands charged with ≈ 5,000 V and spaced 0.58, 0.39, and 0.23 m from the ground. To help ensure that different bears interacted with the fence designs, I identified individual bears using several physical characteristics and unique behavior, when observed. I evaluated the repeatability of this methodology by determining the inter-rater reliability of students trained in my bear identification methods. Each student was given a 20-question test consisting of paired photographs or video recordings of unique or different bears collected during the study. Fleiss’ kappa for the photographic tests was 0.49 (n = 35, z = 53.1, p < 0.001) for all students and 0.78 (z = 11, p < 0.001) for the top 5, indicating moderate and substantial agreement, respectively. Fleiss’ kappa for the video test was 0.52 (n = 5, z = 7.3, p < 0.001) indicating moderate agreement. Results indicate that my techniques for identifying individual bears from photos and videos were generally repeatable. ACKNOWLEDGEMENTS I would like to thank The Berryman Institute, Safari Club International - Michigan Involvement Committee, Michigan Beekeepers Association, Michigan State University (MSU) Graduate School, Todd Broad and the Beaver Lake Hunt Club, Michigan Department of Natural Resources (MDNR), specifically Dan Moran, Jennifer Kleitch, Bruce Baker, and the Atlanta, MI Field Office, Steve Fitzner from Fitzner Fencing, and the Department of Fisheries & Wildlife Graduate Student Organization for funding and support for this project. I would also like to thank MSU Department of Fisheries and Wildlife (FW), Dr. Mike Jones, and my committee: my advisor Dr. Gary Roloff (MSU), Jordan Burroughs (MSU), Dr. Kelly Millenbah (MSU), and Dr. Dwayne Etter (MDNR), my field assistant Marie Stevenson, as well as the MSU FW graduate student community. I have been fortunate to have met and worked with so many amazing people! To my husband, Clint, many thanks for your help and care. May we continue to forge ahead together and forever find peace in nature, frogs on low branches over ponds, and birds on nests. To our beloved girls, Coralyn and Aurelia, my earliest memories of you will always be one with this project. Thank you for taking it easy on me when we were in the field together. To Dad, thank you for your invaluable contributions to my work. And Mom, thank you for all of your hard work and dedication and for making so much possible for me. Much love. iii TABLE OF CONTENTS LIST OF TABLES .............................................................................................................. v LIST OF FIGURES ........................................................................................................... vi 1. Introduction ......................................................................................................... 1 2. Methods............................................................................................................... 4 2.1. Study Area ........................................................................................... 4 2.2. Pre-baiting and Baiting ........................................................................ 5 2.3. Bear Identification ............................................................................... 6 2.4. Fence Testing ....................................................................................... 8 2.5. Digital Video Recorder (DVR) System ............................................... 9 2.6. Voltage Readings and Battery Longevity .......................................... 10 3. Results ............................................................................................................... 12 3.1. Pre-baiting and Baiting ...................................................................... 12 3.2. Bear Identification ............................................................................. 12 3.3. Fence Testing ..................................................................................... 14 3.4. Voltage Readings and Battery Longevity .......................................... 15 4. Discussion ......................................................................................................... 17 4.1. Conclusions ........................................................................................ 25 Appendix ............................................................................................................... 29 Literature Cited ..................................................................................................... 46 iv LIST OF TABLES Table 1. Pre-baiting, nights to bear detection at pre-baiting sites, and total nights of baiting used to test the efficacy of portable electrical fences for excluding black bears from bait sites, Beaver Lake Hunt Club, Lachine, Michigan, 2010. Fence Design A was tested at 2 sites over 8 nights, Design B at 2 sites over 4 nights, Design C at 3 sites over 6 nights, and Design D at 2 sites over 5 nights. Site numbers 7, 9, and 11 were not used to test fences and were omitted from this table..…………………….. .......................................30 Table 2. Relative strength of concordance associated with Fleiss’ kappa ...........................31 Table 3. Equipment list for black bear exclusion fences. Materials were not necessarily purchased from the supplier(s) listed. Prices are approximate and based on those advertised by the suppliers listed (2012) ................................32 Table 4. Bear activity and fence performance, per fence design, during fence testing at the Beaver Lake Hunt Club, Lachine, MI, 2010 ....................................34 Table 5. Product specifications for DVR cameras, AVTECH Model: KPC139D ..............37 Table 6. Longevity and mean ± (1 SE) wire voltage of two 12 V deep-cycle marine/recreational vehicle batteries subjected to different power management scenarios from June – November, 2011. Tested batteries were used to power the electric fence energizer during the 2010 field season. A test was complete once the battery had insufficient charge to power the fence energizer or 14 days had passed ..................................................38 Table 7. Bear-fence-interaction (BFI) and other bear activity, per bear, during fence testing at the Beaver Lake Hunt Club, Lachine, MI, 2010 ....................................39 v LIST OF FIGURES Figure 1. Physical characteristics that were used to identify individual bears. A and B portray lines of equal length based on an obvious morphometric characteristic (e.g., distance from belly to top of foot (A), distance between the ears (B)) were used to differentiate variation in body size among individual bears. When observed, tags and pelage markings were useful (C). Facial markings, shapes, and colors were also useful at times (D) ..........................................................................................................................40 Figure 2. Example slides from the bear photo identification questionnaire administered to undergraduate students (n=35) that were members of the Fisheries and Wildlife Club at Michigan State University ....................................41 Figure 3. Fence designs that were tested for excluding black bears (Ursus americanus) in the Beaver Lake Hunt Club, Lachine, MI, June – August, 2010. All fence designs used the same wire type: 1.3 cm (0.5 in.) white polytape ..................................................................................................................42 Figure 4. This illustration depicts one fence design (Design D) used to evaluate the efficacy of portable, electrical fences for excluding black bears. In this example, the bear has been excluded from a bee yard. Fundamental fence materials are labeled: step-in post (A), fiberglass rod (B), polytape (C), fence energizer (D), double-insulated wire (E), and grounding rod with clamp (F). ...............................................................................................................43 Figure 5. The DVR housing consisted of a water-tight, aluminum, military-grade foot-locker, a custom-welded steel frame with anchor rods, and chains ...............44 Figure 6. Pearson’s product moment correlation (Sokal and Rohlf 1981) between voltage meter readings from the Kencove Dual Fault Finder and Volt Meter and the Fi-Shock Digital Voltage Reader during fence testing ...................45 vi 1. Introduction Crop pollination by honey bees (Apis mellifera) is critical to the agriculture industry (Morse and Calderone 2000, United States Environmental Protection Agency 2008, Fish and Wildlife Service 2009). Global declines in feral honey bee populations have resulted in considerable demand for commercial pollination (vanEngelsdorp and Meixner 2010). According to the USDA Farm Service Agency (FSA), Michigan is a leader in the production of apples (Malus sp.), cucumbers (Cucumis sp.), blueberries (Vaccinium sp.), and cherries (Prunus sp.); each of which requires ≥81% pollination by commercial bees (Morse and Calderone 2000, Burgett et al. 2010). Efficacy of pollination depends on the stability and social structure of bee colonies (vanEngelsdorp and Meixner 2010). In many parts of the United States black bears are a growing nuisance to beekeepers because they are attracted to bee colonies as a food source (Maehr 1984, Caron and Bowman 2004, Maryland Department of Natural Resources 2004). Although bears might only seek bee larvae in late summer (Meadows et al. 1998), they are active from spring to late fall and are likely to investigate human derivatives when natural foods are less available (Masterson 2006). Furthermore, once a bear locates a food source, it is likely to return (Hygnstrom and Craven 1996, Masterson 2006). A single bear encounter with an unprotected apiary can result in hive damage and colony loss, often imposing considerable negative economic consequences for the beekeeper (Maehr and Brady 1982, Jonker et al. 1998) and creating a problem that potentially lasts throughout the pollination season (Clark et al. 2005). One area of concern is the northwestern Lower Peninsula of Michigan where pollinator crops, particularly cherries, are important to the local economy (Michigan Land Use Institute (MLUI) 2009). This region encompasses seven counties (Antrim, Benzie, Grand Traverse, 1 Kalkaska, Leelanau, and Wexford) that contribute nearly 25% to the annual fruit yield in Michigan. Nearly 60% of the cherries produced in Michigan are grown in this area (MLUI 2009). Black bear density increased in this region from 1990 through 2005, but more recently the population has stabilized (Michigan Department of Natural Resources (MDNR) Unpublished data). The potential for apiarist-bear conflict is a black bear management concern so in 2004 and 2008 the MDNR expanded bear hunting opportunities in this region in an attempt to reduce conflicts between bears and beekeepers (pers. comm. D. Etter, MDNR, Lansing, MI). Although a number of different techniques have been tested as a means to reduce bear damage (e.g., repellents and toxicants, scare devices, capture and removal, shooting, chasing with dogs, raised platforms (as mentioned in Hygnstrom and Craven 1996)), electric fencing appears to be the most effective means to exclude bears from apiaries (Meadows et al. 1998, Caron and Bowman 2004, Clark et al. 2005), yet fencing is < 100% effective. Data suggest that this may result from poor fence design, setup, and maintenance (Huygens and Hayashi 1999). Anecdotal evidence in Michigan also indicates varying effectiveness (pers. comm., Larry Hilbert, Michigan Beekeepers Association (MBA) District 5 Representative; Dr. Roger Hoopingarner, Professor Emeritus, President, MBA), presumably due to the same reasons. Variability in fence effectiveness may be attributed in part to fundamental characteristics of portable versus permanent electric fencing. Portable fences are often the better option for commercial beekeepers because they are meant to protect hives placed in orchards and fields only during the flowering season (Burgett et al. 2010). These fences consist of lighter, less sturdy materials (e.g., step-in and fiberglass posts, polywire or polytape) that aid in easy transport and setup. In contrast, permanent fences consist of sturdier materials (e.g., wood or metal posts, hi-tensile steel or aluminum wire) that better withstand rough and frequent 2 tampering (by bears and humans) and exposure to the elements (e.g., solar radiation, weather). Permanent fence lasts longer and likely requires less-frequent maintenance than portable fence not only because of differences in construction materials, but also because the fencing materials are not subjected to repeated assembly and disassembly. Although areas experiencing consistent bear damage and large, permanent attractants (e.g., landfills, camps, feed storage sheds) will benefit most from more expensive, permanent electric fencing, less expensive portable polytape fences are usually sufficient to prevent short-term bear damage (Hygnstrom and Craven 1996). Several studies, dating back to 1938 (Storer, et al. 1938), have tested the effectiveness of electric fences for excluding black bears. Few have used video surveillance to support their conclusions (e.g., United States Department of Agriculture (USDA) Forest Service 2007), and to date information on bear behavior around electric fences is lacking (McKillop and Sibly 1988). Understanding bear behavior could prove useful for improving fence design. The goal of this project was to quantify bear behavior in proximity to portable fences and use that information to identify an effective, portable fence design for use by beekeepers. My criteria for effectiveness included relative ease of installation and removal, low cost, and ability to exclude bears. I provide quantitative evidence in support of using portable electrical fencing to exclude bears from small-scale attractants and offer insights into proper construction and maintenance techniques. 3 2. Methods 2.1. Study Area This study occurred in the northeastern Lower Peninsula (LP) of Michigan. Recent capturemark-recapture estimates of bear abundance in the LP indicates a declining bear population (pers. comm. D. Etter, MDNR, Lansing, MI); however, the study area supports some of the highest quality bear habitat in the LP (Carter et. al 2010) and bear density in this region remains high (pers. comm. Dwayne Etter, MDNR). Twelve potential fence testing sites were identified in Alpena County near the town of Lachine on property owned by Beaver Lake Hunt Club (BLHC). Historically, the Lachine area was dominated by white and red pine (Pinus spp.) forest and cedar (Thuja occidentalis) swamp (Barnes and Wagner 2004). Economic development in this region was initially based on fur trapping and logging until establishment of dairy and crop farming in the 1950’s (Green Township Planning Commission 2007). Beaver Lake Hunt Club 2 encompasses 17 km (4,200 acres) of what is now northern hardwood forest and forested wetlands (Acer, Betula, Tilia, Prunus, Pinus, Thuja, and Abies spp.), ranges in elevation from 231.6 – 304.8 m (760 – 1,000 ft.), and borders other privately-owned hunt club properties (e.g., Turtle Lake Hunt Club, Doctor’s Hunt Club). Hunt club properties are actively managed for popular game species, primarily white-tailed deer (Odocoileus virginianus). Management includes planting food plots, and manipulating vegetation to provide cover and mast-producing 1 trees. Average annual precipitation ranges between 68.6 and 76.2 cm (27 – 30 in.) (USDA Natural Resources Conservation Service (NRCS) 2006). Average annual temperature ranges from 5° – 9° C (41° – 47° F) with a freeze period of approximately 145 days (USDA NRCS 1 Although black bears are opportunistic feeders with a liberal palate, they will feed primarily on nuts, berries, grass shoots, leafy forbs, and insect larvae during the summer months (Masterson 2006). 4 2006). Fence testing sites were selected in consultation with BLHC staff and members who identified areas of high bear activity and favorable habitat (e.g., forested wetlands). Testing sites varied in distance from one another (𝑥̅ = 2.47 km (1.54 mi.), range 0.7 – 6.0 km (0.44 – 3.73 mi.)). Data collection took place in 2010, from June 30 to August 11. 2.2. Pre-baiting and Baiting I started pre-baiting potential test sites on June 24, 2010 (Table 1). I used 4.4 – 13.2 L (1-3 gal.) of bait per day to attract black bears to potential testing sites; the amount used on any given day depended on how much bait remained from the day before. Specific bait items included combinations of bread, cookies, trail mix, Circus Peanuts (Spangler Candy Company, Bryan, Ohio), cinnamon-chocolate chips, vanilla icing, blueberry pie filling, honey, bacon, sardines, and fryer grease. I could not use grain, fruit, or vegetables for bait because my research was conducted inside the high-risk bovine tuberculosis (Mycobacterium bovis) zone in Michigan. In this zone, the MDNR banned the use of bait that is likely to attract cervids to reduce the risk of disease spread (MDNR 2009, Michigan Department of Natural Resources and Environment 2010). At each potential test site, I used a motion-triggered, infrared, digital trail camera (Cuddeback® Excite 2.0 Megapixel Digital Trail Camera, Non Typical Inc., Green Bay, WI) to determine the presence of bears at the bait. These cameras were placed in trees at a height of approximately 0.9 – 1.2 m (3 – 4 ft.), a distance from the bait of approximately 1.8 – 2.7 m (6 – 9 ft.), and were angled slightly downward. If a bear was detected by the camera during this prebaiting phase, the site was considered active and used to test a fence design. If no bears appeared at the site within 10 days of pre-baiting (with periodic refreshing of the bait) I considered the site 5 inactive and it was removed from the candidate sites for fence testing. During fence testing, the same baiting mix and regime were used to lure bears into interacting with the fence. 2.3. Bear Identification Individual bear identification is difficult from photos and video, especially at night, unless obvious markings are evident or bears differ substantially in appearance. Ideally, fence effectiveness should be evaluated based on individual bears. In my study, different fence testing sites could be visited by the same bear so I developed a process for identifying individuals. I distinguished individual bears by evaluating (in combination) several physical characteristics (Fig. 1), the presence or absence of cubs, and behavior. These included: estimated height at shoulder (i.e., when bears were observed within 8 cm (3 in.) of a step-in post, I estimated shoulder height using the same 6 gradations that occurred on all fence posts), scars and/or pelage markings (e.g., white patches, bald spots; Fig. 1C), coat condition (e.g., thin vs. thick, short vs. long), body appearance (e.g., slender vs. stocky, straight-, sway-, or hump-backed while walking), muzzle shape and color, distance between the ears (i.e., larger bears will often have a noticeably larger space between their ears), relative ear-to-head size (i.e., ears will appear larger on smaller bears (Fig. 1B), and repetitive behavior (e.g., stretching up or shaking the same nearby tree, digging, charging, pounding). As part of an unrelated research project, bears in the vicinity were also ear-tagged and branded further enhancing my abilities for individual identification. My project was consistent with methods approved by the Institutional Animal Use and Care Committee at Michigan State University (#05/10-051-00). To evaluate the repeatability of my bear identification methods, I conducted two tests. First, I asked 35 undergraduate students at Michigan State University to scrutinize 20 pair of black bear photos that were collected via trail cameras during pre-baiting in the study area. The 6 cameras frequently captured multiple images of the same bear within minutes because most bears fed from the bait for a prolonged period. As a result, I had access to an assortment of imagery for which bear identification was reasonably certain. Of the 1,390 images captured of bears via trail cameras, I was able to populate a questionnaire with bears of presumably known identity. Testing was performed with the approval of the Institutional Review Board (IRB) at Michigan State University (# x11-909e; i039448). Prior to administering the test, I trained students in the bear identification techniques that I used (Fig. 1). Students judged whether each photo pair was the “same”, “probably same”, “probably different”, or “different” bear(s) after viewing the images simultaneously for 45 seconds (e.g., Fig. 2). I accepted responses of “probably same” or “probably different” when the correct answers were “same” or “different”, respectively. The top 5 students correctly identifying variation between individual bears during the photo test were invited to complete a second questionnaire that compared 20 pair of video-clips. Video-clips were extracted from the video I collected during fence testing. Like the photo test, each pair of video-clips was positioned side-by-side and played simultaneously. Video-clip pairs varied in duration from 1.4 – 2.5 minutes. Students were not subjected to a time limit for this test and were free to pause, rewind, and fast-forward as needed. I used Fleiss’ kappa (Fleiss 1971) to determine concordance, or inter-rater reliability, among students. Fleiss’ kappa is a generalization of the kappa statistic (Fleiss 1971) developed by Cohen (1960). Cohen’s kappa has the advantage over percentage agreements in that it takes into account the probability that 2 raters, answering randomly, could appear to agree (Cohen 1960, Wood 2007). Unlike Cohen’s kappa, Fleiss’ kappa coefficients represent the degree of concordance among multiple raters (Fleiss 1971). I calculated Fleiss’ kappa first for all 35 students of the photo test, and again for the top-five 7 performers, for both the photo and video tests. Although the interpretation of the kappa statistic is somewhat flexible, Fleiss’ kappa can be interpreted using a commonly cited scale (Landis and Koch 1977; Table 2). I estimated Fleiss’ kappa using the package irr in the program R (Gamer et al. 2012), version 4.14.1 (R Development Core Team 2010). 2.4. Fence Testing 2 I evaluated four exclusion fence designs (Fig. 3). Each fence cost approximately $250-300 (in 2010) and all were comprised of the same components: white electric polytape, 1.3 cm (0.5 in.) wide; fiberglass corner fence rods, 1.8 m (6 ft.) long and 0.04 m (1.5 in.) dia.; plastic step-in fence posts, 1.1 m (43 in.) long; a portable, battery-powered fence energizer (Kencove Portable Battery Energizer; Kencove Farm and Fence Supplies, Blairsville, PA)); galvanized grounding rods, 1.2 m (4 ft.) long and 1.3 cm (0.5 in.) dia.; brass grounding rod clamps, 1.3 cm (0.5 in.); and double-insulated electric wire rated at ~20,000V to connect the energizer to the fence and 3 grounding rods (Fig. 4) . The energizer was powered by a 12 V deep-cycle marine/recreational vehicle battery and fence voltage was maintained at approximately 5,000 V, consistent with other studies (e.g., Hygnstrom 1994, Hygnstrom and Craven 1996, Montana Fish, Wildlife, & Parks (MFWP) 2010). The order in which fence designs were tested was randomly assigned, and the first test site was randomly assigned. Subsequent test sites on BLHC property were selected based on their distance from the last active test site (i.e., the test site located farthest from the previous test site was used). Separating the sites by farthest distance increased the likelihood of different 2 The cost per fence can be reduced by approximately $100 with the purchase of a less-powerful fence energizer (e.g., 0.25 Joule). A 0.25 Joule battery-powered fence energizer, for example, costs approximately $90 and can easily power 2.4 km (1.5 mi.) of charged wire. 3 For a complete list of fencing materials and cost, see Table 3. 8 bears visiting the fences; however distances were too short to ensure that unique bears visited different fence testing sites. I recorded data on each bear-fence interaction (BFI) including date, duration of BFI, whether the bear touched the fence, whether the bear received a shock and how the bear responded to the fence and/or shock, whether the bear broke the plane of the fence, whether the bear breached the fence, whether the bait was accessed by the bear, as well as several physical characteristics of each bear that aided in individual identification. “Breaking the plane” of the fence is defined as an event when a bear extends any part of its body through the vertical plane of the fence. A “breach” is defined as a bear breaking the plane of the fence by extending the entirety of at least one leg through the vertical plane of the fence. I defined a BFI as an event when a bear came within 3 m (10 ft.) of the fence and showed interest in either the fence or bait. A bear “visit” is defined as any time a bear was detected by the video cameras at a test site. A bear showed interest in the fence or bait by directly approaching it with a clear line of sight, and with ears and nose concentrated on the test site. Bears often circled fences repeatedly, sometimes approaching within 3 m of the fence multiple times as they moved. As a result, a bear could accrue multiple BFIs in one visit (it was also possible for a bear to accrue no BFIs during a visit.). In this way, BFIs were tallied and linked to individual bears. My goal was to document ≥ 10 BFIs with a minimum of 3 bears per test site before moving to a different site and fence design. However, if three nights passed with fewer than 10 BFIs from 3 bears, I moved testing to another site. I determined fence effectiveness, per design, by dividing the number of breach events where bears accessed bait by the number of BFIs, then multiplying the result by 100 to find x. I then subtracted x from 100 to find percent effectiveness (i.e., the percentage of BFIs that did not result in bait access); Table 4). 9 2.5. Digital Video Recorder (DVR) System Bear-fence interactions were recorded using a high-resolution, motion-sensing, anti-vibration DVR system (Model: GV-LX4C2V; GeoVision, Neihu District, Taipei 114, Taiwan). The weatherproof, infrared security cameras (Model: KPC139D; AVTECH Corporation, Nankang, Taipei 115, Taiwan; see Table 5) captured video in color during the day, black and white at night, and were capable of recording in complete darkness (0 lux). Each camera (3 at each site) was attached to a wooden base, secured at a height of 2.4 – 3.1 m (8 – 10 ft.) in a tree (DBH ≥ approximately 15 cm (6 in.)) with cords and cable ties, and aimed directly at the bait from a distance that ensured the entirety of the fence could be viewed through each camera. Video data were reviewed on site once every 18 to 24 hours of fence deployment. During these field viewings, I tallied the number of BFIs and approximated the number of individual bears; information necessary to determine when testing at a site was complete. The DVR system was powered by a 12 V deep-cycle marine/recreational vehicle battery that was replaced every one to three days by a battery with a full charge. These batteries were used only to power the DVR system and were not switched with those used to power the fence energizer. The DVR system and its power source were housed in an aluminum military-grade locker to protect the equipment from bears and weather (Fig. 5). This strong, waterproof case was secured to a tree using a custom-welded steel frame and chains (Fig. 5). 2.6. Voltage Readings and Battery Longevity I used two hand-held voltage readers in 2010 during field testing of fence designs. The first, a Kencove Dual Fault Finder and Digital Volt Meter (Kencove Farm and Fence Supplies, Blairsville, PA), was a dual-purpose unit that measured voltage and could detect ground-faults in the electrical circuitry of the fence. It was difficult to achieve consistent voltage readings with 10 this unit because the shape of the connector plates did not facilitate contact with the stainless steel threads woven into the polytape. As a result, I purchased a second unit (Fi-Shock Digital Voltage Reader; Fi-Shock Animal Containment Systems, Lititz, PA) that was equipped with a grounding peg and allowed improved connection with the stainless steel threads. I collected voltage readings from every tier of the fence at the corners nearest and farthest from the energizer each time I visited the test site, both before the energizer was switched off and again after it was switched back on. Once the Fi-Shock unit was acquired, I collected paired readings from the Kencove and Fi-Shock units. Pearson’s product moment correlation (Sokal and Rohlf 1981) was used to test for correlation between voltage readings taken by both units. The FiShock unit was also used to read fence voltage during the battery longevity tests in 2011. Longevity of the 12 V batteries used to power the fence energizer was measured from June - November 2011. A fence identical to Design D (Fig. 3, 4) was erected and electrified using the same Kencove Portable Battery Energizer used in 2010. I conducted five tests (Table 6); the energizer ran continuously for Tests 1 and 2, and was switched off and on daily for Tests 3 through 5. I included the latter tests to simulate a scenario where a beekeeper may need access to hives and thereby periodically disrupt the power supply to the fence. A test was complete when either the battery died or 14 days had passed. 11 3. Results 3.1. Pre-baiting and Baiting Pre-baiting occurred at 9 candidate testing sites (Table 1). On average, a site was pre-baited for 3.1 ± 0.2 nights (1 SE) before a bear was detected on camera (n=7) or based on sign (e.g., a wide pathway led to the bait area; the vegetated ground where the bait had been placed had been worked over thoroughly; all bait was gone and a swath of bare earth remained) (n=1; Table 1). Bears were observed at 8 sites; only Site 2 showed no evidence of bear activity after 10 days (Table 1). Once bears were detected at a site, baiting continued until a fence was established and testing was complete at the site. Baiting (i.e., pre-baiting and baiting during fence testing) occurred an average of 9.9 ± 1.2 nights at each testing site (Table 1). Variation in total baiting nights among sites was caused by 2 factors: 1) the fence was being tested on another site and hence was not available for immediate deployment, and 2) because pre-baiting occurred simultaneously at multiple sites, fence deployment had to be staggered once bears were detected. 3.2. Bear Identification The methods I used to identify individual bears from photos and videography were generally repeatable among trained observers, suggesting that other photo interpreters were consistently seeing the same differences among bears that I was detecting. The average percentage of answers (35 students, 20 questions) for the photo identification of individual bears that were consistent with my determinations was 81 ± 2% (range 60 – 95). The top-five performing students provided answers consistent with my determinations 94 ± 1% (range 90 – 95%) of the time. These students agreed completely and correctly on 15 questions, agreed 80% on 4 questions, and 60% on 1 question. The mean score for the video identification test (top five 12 students) was lower overall at 84 ± 4% (range 75 – 95). Here, students agreed completely and correctly on 10 questions, 80% on 6 questions, 60% on 2 questions, and 40% on 2 questions. Fleiss’ kappa for the photo identification test was 0.49 (z = 53.1, p < 0.001), indicating moderate agreement (Table 2) and was not due to chance (i.e., p < 0.001). Fleiss’ kappa for the top-5 performing students in photo identification was 0.78 (z = 11, p < 0.001), indicating substantial agreement (Table 2). The video test was administered to the same five students several weeks after having completed the photo test and without an instructor-led tutorial. Kappa was lower (Fleiss’ kappa = 0.52, z = 7.3, p < 0.001) but still indicated moderate agreement (Table 2). Given that videography contains more identification information than still photos, my results suggest that instructor-led tutorials are an important component for consistently identifying bears among various observers. My most conservative estimate of individual bears in this study was 15 (Table 7). I differentiated sows with birth-year cubs using camera footage, but I counted each familial group as a single bear. I counted families as one bear because cubs were impossible to differentiate as they moved into and beyond camera range. These movements happened frequently and independently. Counting families as a single entity resulted in a lower total number of BFIs (because only one bear was required within the 3 m BFI zone to begin the clock on a single BFI). If one cub entered the BFI zone and another joined it, the first could leave the zone and the clock would continue to run until the second cub departed. Alternatively, the total number of BFIs could have been inflated because movements by multiple bears represent those of a single individual. This was less likely to occur than the former scenario, however, since at least one bear typically remained in the BFI zone while the family visited the testing site. Only two familial groups visited testing sites, both of which were used to test fence design C (Table 7). 13 Additionally, the cumulative duration of BFIs from familial units was comparable to single bear visits, hence I do not believe this nuance in data collection resulted in biased results. 3.3. Fence Testing Of the 4 fence designs I tested, 3 were tested at 2 sites (Table 1). Fence design C was tested at 3 sites because of low bear activity at site 3. Seventy-three bear visits were recorded (Table 4). Total bear visits ranged from 12 to 24 per site (Table 4). The number of BFIs per bear visit was similar among the 4 fence designs: Design A had 2.2 BFIs per visit, Design B 2.5, Design C 2.2, and Design D 2.5, indicating that each fence design received comparable bear attention. Conservatively, I estimated that 3 to 6 individual bears interacted with each fence design (Table 4). On average, bear fence interactions lasted < 5 minutes (Table 4). Total duration of BFIs for Design A (246.17 minutes) exceeded that of any other fence design (Table 4). Design A failed to exclude bears from the bait and hence, the time spent consuming bait inflated the BFI total and average (Table 4; feeding accounted for 41.3% (101.63 minutes) of the total BFI duration). Similarly, bears spent 43.3% (37.45 minutes) of the total BFI duration (86.45 minutes) consuming bait for the design C fence (Table 4). Bears visited fences multiple times (range 2.0 – 6.0) and I recorded multiple BFIs from each bear (range 5.0 – 13.0). The frequency in which bears either touched or broke the plane of the fence was greatest for Designs A and C (Table 4). The frequency of BFIs that resulted in a bear either breaking the plane or touching the fence was 0.21, 0.17, 0.35, and 0.16, for Designs A, B, C, and D, respectively, suggesting that fence designs A, B, and D received comparable bear attention during testing and that Design C received approximately twice the attention. Breach events occurred on all fence designs, but the proportion of BFIs that resulted in a breach varied among 14 designs (Table 4). Design D had the lowest proportion (2.6%) of BFIs that resulted in a bear breaching the fence. All breach events that occurred for Designs A and C resulted in bait access, while none of the breach events that occurred for Designs B and D resulted in bait access. Fence Design B had two breach events without bait access. During one of these breaches, a large bear entered the fence between the top and bottom polytape strands. It received a shock on its back left leg, along the inner thigh. The bear responded by turning and leaping to avoid colliding with a step-in post. Because the top polytape strand was over its back as it had entered the fence, the strand pulled free from the corner post as the bear fled. Although this breach event did not result in bait access, fence damage occurred making the bait vulnerable to future bear visits. The other breach events did not result in fence damage; both bears attempted to step over the top polytape strand and received a shock on the upper inside of a front leg (a desirable location for bears to feel the shock). Designs B and D were considered 100% effective because bears did not access the bait during the breach events. Given that each fence design received a comparable amount of bear attention, that all designs were breached but 2 kept bears away from the bait, and that Design D allowed the fewest breach events per BFI where no fence damage occurred, I designated Design D as the preferred fence design for excluding black bears from bait. 3.4. Voltage Readings and Battery Longevity The total number of voltage readings collected in 2010 was 216. The mean for all readings was 4,690 ± 29.5 volts. Forty paired voltage readings were collected over 8 days using the Kencove and the Fi-Shock units as a way to potentially calibrate readings taken singly by the Kencove unit earlier in the season. The mean for readings taken by these units during this period was 4,640 ± 34.4 volts (Kencove) and 4,876 ± 46.0 volts (Fi-Shock). The collection of voltage readings was not correlated (correlation coefficient = -0.08, F-statistic = 0.24, p-value = 0.62) 15 indicating that the Kencove measurements could not be predicted by the Fi-Shock measurements (Fig. 6). Regardless of the voltage meter used, my data suggest that I consistently provided approximately 200 to 400 volts less than my goal of 5,000 volts during fence testing. This shortcoming can be attributed to a variety of conditions including changing soil moisture (influencing conduction of electricity through the grounding rod), varying atmospheric conditions, and measurement error through use of the voltage meter. In the fence energizer battery tests, the battery charge was sufficient to maintain a mean output of 5,056 ± 15 volts for 14 days during the continuous power tests (Table 6). Results varied somewhat for Tests 3-5, where during Test 3, the battery did not have sufficient charge to power the fence energizer on Day 4 (mean 5,133 ± 33 volts; Table 6), thus ending the trial. During Tests 4 and 5, the battery charge was sufficient to power the fence energizer for 14 days while maintaining a mean output of 5,062 ± 32 volts. Three tests remained functional beyond 14 days, but I recommend that maintenance intervals on portable, electric exclusion fencing should not exceed 2 weeks. Fence maintenance is equally as important to fence effectiveness as proper fence design and set up (Conover 2001, Hoare 2003, Honda et al. 2009). 16 4. Discussion The goal of my project was to identify a portable electric fence that was 100% effective at 2 2 excluding black bears from a relatively small area (i.e., 13.4 m (144 ft )). When an electric fence fails to exclude bears, fence design, set up, and maintenance are often questioned, regardless of how motivated the bear was to access the area or the level of experience bears have in negotiating the fence (i.e., if a fence did not prevent damage, it was considered inadequate). I found that two of the designs I tested (Designs B and D) were 100% effective at preventing bait access. Other studies (e.g., Gates et al. 1978, Reidy et al. 2008, Tolhurst et al. 2008, Honda et al. 2009) evaluated electric fence effectiveness for different species (i.e., coyotes (Canis latrans), feral pigs (Sus scrofa), badgers (Meles meles), raccoon dogs (Nyctereutes procyonoides), masked palm civets (Paguma larvata)), using the same criterion for evaluation (access to attractant). All agreed that a properly designed and maintained fence is an important tool for the prevention of damage caused by wildlife. Reidy et al. (2008) observed that juvenile pigs successfully breached fencing more frequently because of their small size. I made a similar observation; small bears (≤ 0.5 m (20 in.) at the shoulder) successfully breached fences more frequently than did larger bears. Behavioral differences were also apparent between young bears (birth year) and large, presumably older bears. After receiving a shock from the fence, birth-year cubs often seemed bewildered potentially not making the connection between cause (touching the fence) and effect (shock). These young bears often attempted to breach the fence again just moments after the initial shock. Large bears (≥ 0.7 m (28 in.) at the shoulder), alternatively, were rarely seen again at the fences after receiving a shock. My results suggest that electrical fencing may prove useful in conditioning older bears to avoid localized areas, provided that the motivation to seek a reward (i.e., the bait) does not override the risk associated with receiving a shock. Bears likely 17 exhibit varying motivations for challenging a fence based on the urgency to acquire the perceived reward. Older bears may possess a more extensive cognitive map of various food sources within their environment and hence more readily abandon a protected reward. Their size 4 also allows them to travel greater distances more efficiently . Yet, when natural food is scarce, as would occur in years of low mast production, it is likely that bears of any age would challenge electric fences more frequently and with greater vigor, particularly in areas of relatively high bear density (Garshelis 1989, Hygnstrom 1994). Of the 3 breach events I observed that did not result in bait access, 2 may not have occurred had a fourth tier of polytape been present above the top-most strand on Designs B and D (Fig. 3). Other studies citing high effectiveness with portable electric fencing at excluding black bears (U. americanus or U. thibetanus) (e.g., Storer et al. 1938, Huygens and Hayashi 1999, Creel 2007) used fences with 4 tiers, where the top strand was situated between 0.91 – 1.02 m (35.83 – 40.16 in.) above the ground. Storer et al. did not report the number of bears repelled by a four-tier fence, but they listed the design as “satisfactory”. Huygens and Hayashi used bear tracks as evidence for bear activity near fences and reported 100% effectiveness after 7 sets of tracks approached but stopped at the fence. Creel identified 10 bears, 3 of which had circumvented the electric fence (i.e., 70% exclusion effectiveness). Additionally, Masterson (2006) recommended a minimum of 4 tiers for black bear exclusion fencing. In my study, fence damage did not occur during breach events because the polytape was taut across each of the corner posts, was reinforced by step-in posts along its length, came into contact with each bear on a relatively hairless part of its body, and because each bear fled 4 Besides the tendency of males to occupy larger territories than females, it is possible that territory size is a reflection of food quality and/or availability, but it may also reflect intrinsic variation among individuals (Masterson 2006). 18 without dragging its leg through the fence. The third breach event, although it did not result in bait access, did result in fence damage making the fence potentially ineffective should another bear try to access the bait. The breach event that resulted in damage underscores the need for frequent visits to electrical fence enclosures to ensure that the fence is functioning properly. Bear behavior observed most frequently during this study was consistent with that of other studies (e.g., Storer et al. 1938, Huygens and Hayashi 1999, Creel 2007). Prior to a BFI, bears cautiously approached the fence circling from a distance to presumably investigate the fence and the test site while remaining vigilant of their surroundings (e.g., nose up and sniffing the air, looking around). During a BFI, bears appeared to remain cautious and inquisitive, circling frequently, and sniffing the fence and fence energizer. If they received a shock from the fence, most bears ran away. After a BFI, some bears continued circling the test site, while others moved beyond the reach of the cameras, presumably having left the area. Though bears were observed occasionally digging around the fence, no bear focused consistently in a single location or dug down beyond approximately 2.5 cm (1 in.). Although some species excluded by electric fences are known to dig under the fence (e.g., rats (Rattus spp.), ferrets (Mustela furo), rabbits (Oryctolagus cuniculus cuniculus), coyotes), it is not a tendency of bears (Storer et al. 1938). I found only one other account of a bear that attempted to dig under an electric fence (USDA Forest Service 2007). Yet, another source suggests burying chain link mesh around an electric exclusion fence intended for black bears, “when necessary” (Washington Department of Fish and Wildlife (WDFW) 2007). I also observed bears charging the fence, standing up on hind legs, pounding the ground outside the fence, running in bursts past the fence, and shaking and trampling small trees and underbrush near the fence. With the exception of standing up on hind legs (which could be interpreted as either inquisitive or a stress behavior, depending on ear 19 position (Herrero and Higgins 2003, Masterson 2006)), I interpreted these as stress or more specifically “threat” behaviors, consistent with other studies (i.e., Jordan 1976, Herrero and Higgins 2003). In these instances, bears would often stop immediately following a display to look at the fence with ears turned forward and scanning, as if to check whether the fence had responded. All of these behaviors were more likely to occur on return visits, after a bear had received at least one shock, and regardless of whether the bear had previously accessed the bait (recall that some of these bears had been feeding at the bait location prior to fence testing). Creel (2007) found that once a bear accessed bait, it returned more frequently to test sites compared to bears that had not accessed bait; an example of cognitive mapping ability that is supported by numerous accounts (e.g., Beckmann et al. 2004, Clark et al. 2005, Masterson 2006, Leigh 2007, WDFW 2007) and is one which provides incentive to install fencing proactively as a means to decrease the risk of bear damage. Creel (2007) also found that a bear could access bait once and be deterred by the same electric fence on subsequent visits. I recorded 168 BFIs in 73 visits by 15 bears. The estimated number of bears was 5 determined after the field portion of my study ended by scrutinizing gross morphological and behavioral variation between bears. A similar technique was used by Creel (2007) to identify black bears, although the author had captured and marked bears. It was discovered in the Creel (2007) study however, that neither the color nor numbers printed on ear tags could reliably be determined via infrared video camera at night. Hence, Creel (2007) used morphological 5 Because I was not able to determine a conservative estimate of bears during the time of my study, the test for Design D violated study design protocol in that only 2 individual bears were determined to have visited the fence at one site before it was moved to a subsequent testing site. This does not affect calculations used for my results or involving bear visits, BFIs, or fence effectiveness. Furthermore, the fence was visited by the minimum number of bears required by my study design (3), and the duration and number of BFIs, total nights tested, and breach attempts for this design are comparable to those of other designs (Table 4). 20 characteristics and behavior, coupled with the presence or absence of tags, to identify individual bears. Incidentally, the number of individual bears identified (10) by Creel (2007) was similar to that of my study, as was the duration of fence testing (21 days) and the approximate size of the 2 2 study area (< 25 km (9.7 mi )). I tested the repeatability of my methods for identifying bears with encouraging results. An acceptable means by which to improve kappa scores would potentially be to train test participants more thoroughly, particularly on scrutinizing video data. One of the video-test students was involved with another project that also examined video data collected during my study. This student agreed consistently with my determinations 90% of the time, whereas other students, with the exception of one who scored 95%, agreed consistently with my determinations only 75-80% of the time. After completing the test, students were asked if they had reviewed the bear-identification study materials provided. Those who scored ≤ 80% were also those who did not thoroughly review the materials prior to completing the video questionnaire. Furthermore, I did not incorporate imagery taken directly from video into the study materials and it may have been beneficial to do so. Yet, Fleiss’ kappa for the video test does indicate moderate agreement and can be used in support of exploring these methods further. The construction of portable electric fences described in this project was relatively intuitive. It took approximately 1 hour for 1 person to set up a fence. I compiled information from multiple sources (cited below) and from my own experience, which I believe will be useful to individuals seeking to use electric fencing for deterring bear damage to localized areas: • Electric fences work because electricity is less stable in an open, or incomplete, circuit. When an animal (which is in contact with the soil) touches a charged wire, it allows the flow of electricity to move from the fence, to the animal, into the soil, and to the 21 grounding rod (which is connected to the fence energizer by a heavily insulated wire; Fig. 4) thereby closing, or completing, the circuit (User Manual 2008, part # 0077368.001 Rev 8, Zareba Systems, Inc., Woodstream Corporation, Lititz, PA; MFWP 2010). • Polytape (wires) fence can be arranged as a continuous loop or it can be made to “dead end”, creating two fence “arms” extending from the energizer (Zareba Model A200LI User Manual, Zareba Systems, Inc., Woodstream Corporation, Lititz, PA.). Check the user manual for your fence energizer to determine the optimal design. • The voltage output along a fence is weakest at the farthest location on the fence from the energizer. It is from this point that voltage readings should be taken (pers. comm. Steve Fitzner, Fitzner Fencing, Wolverine, MI). • Every polytape or wire should be charged, or “hot” (MFWP 2010). • Do not choose a fence energizer based on its voltage output rating alone, though keeping in the range of 3,500 – 7,000 volts is important (Hygnstrom 1994, Hygnstrom and Craven 1996, Masterson 2006, MFWP 2010). Look for the Joule rating and purchase an 6 energizer with the capacity you need (Living with Wildlife Foundation (LWWF) 2009) . In general, you will need 0.5 Joule for every 4.8 km (3 mi.) of charged wire (i.e., A 1.6 km (1 mi.) fence with 3 charged wires = 4.8 km (3 mi.) of charged wire) (University of California Cooperative Extension (UCCE) 1991). Battery-powered fence energizers can be purchased in a variety of Joule ratings, some as low as 0.25 Joule. 6 7 A rating of 0.7 Joules is recommended for bears (LWWF 2009, MFWP 2010). 7 For safety reasons, do not purchase an energizer with a substantially (e.g., ≥ +0.5 Joule) higher Joule rating than is necessary. The fence energizer used in this study was rated much higher than was needed. It was equipped with an adjustable output, an unusual feature necessary to maintain steady fence voltage amid changing soil-moisture conditions (i.e., grounding conditions). When in operation, the dial on the fence energizer I used did not exceed one-quarter turn. 22 • Purchase a low-impedance fence energizer. Because of differences in circuitry and transformers within the unit, these energizers better resist voltage leakage (relative to high-impedance fence energizers) and are thus able to maintain high energy levels on the fence despite contact from vegetation (UCCE 1991; Hallman Fence Systems, www.hallman.ca/impedance.htm, Winnipeg, Manitoba, Canada). Low-impedance energizers are a safer unit, offer more convenience, and are readily available. • Place the fence energizer inside the fence perimeter. This will serve to protect it from bears. If the energizer is not weather-proof, provide housing. Such housing may also serve as camouflage to protect the energizer from theft or vandalism. • Grounding is critical to fence effectiveness. When soil is dry, it acts as an insulator for the grounding rod. When the grounding rod is insulated, the electrical circuit – at worst – will not be closed when an animal touches the fence (i.e., the animal will not receive a shock). Grounding can be improved with additional grounding rods or by wetting the soil. One potential solution to improve grounding is to attach sheets of chicken wire to the original grounding rod, laying them flat on the soil along (but not under or touching) the base of the fence (Hygnstrom and Craven 1996, MFWP 2010). • The intensity of a shock delivered from an electric fence can be increased as grounding rods are added (User Manual; part # 0077368.001 Rev 8, Zareba Systems, Inc., Woodstream Corporation, Lititz, PA.). • Purchase a polytape that is of high color contrast to the environment in which it will be used. Bears have similar vision to our own, and can see in color (Masterson 2006). White polytape was highly visible at all of my test sites. High visibility can improve the 23 effectiveness of your fence. White polytape may also resist sun damage better than a darker color (pers. comm. Steve Fitzner, Fitzner Fencing, Wolverine, MI). • The northern LP of Michigan experiences frequent cloud cover throughout the year. An 18 – 20 watt solar panel is recommended in these conditions to adequately power a 2 Joule fence energizer continuously throughout a 24-hour period (pers. comm. Kencove Sales Representative, Kencove Farm and Fence Supplies, Blairsville, PA). • Place warning signs in several places along your fence. Wildlife may not be the only visitor (Kencove Farm and Fence Supplies, Blairsville, PA; Zareba Systems, Inc., Woodstream Corporation, Lititz, PA.). • Angle each of the corner posts slightly outward from the center of the area you plan to protect with your fence. This will help ensure that the polytape tiers remain taut. • Bears frequently stuck their noses through the gaps between polytape tiers while investigating the fences, and could easily have fit their heads through the gaps. This should be considered when positioning bee hives within the fence. • A standard maintenance check should occur bi-weekly to evaluate: 1. Fence voltage (use a voltage reader designed for electric fence, preferably one with a grounding peg; voltage readings should be 3,500-7,000V), 2. Battery charge (turn the fence energizer off and disconnect it from the battery before tampering with the battery. Battery should be re-charged or replaced on a regular basis – I recommend once every 14 days. Use a standard volt meter to check battery charge: A 12-V battery should read ≥ 12V, and a 6-V battery should read ≥ 6V. Check that the battery terminals are free of corrosion.), 24 3. Fence energizer (ensure that all wire connections associated with the fence energizer are clean and tight), 4. Polytape (polytape should be taught and straight, and free of debris and vegetation. There 8 should be no snapping heard along the fence as the fence energizer pulses. ), 5. Fence posts (should be securely in place), 6. Grounding (create a short in the fence by leaning 1 – 2 steel posts on the live fence as far from the grounding rod as possible (not > 30 m (98.4 ft.)). Take a reading at the grounding rod using the fence voltage reader; it should read 0-200V. If voltage is detected > 200V, grounding should be improved (Fi-Shock Animal Containment Systems, Woodstream Corporation, Lititz, PA). Add a second grounding rod 3 m (10 ft.) from the first, or lay a sheet of chicken wire flat on the soil along, but not under or touching, the fence. Connect these to the original grounding rod with double-insulated wire rated at ~20,000V. If more than two grounding rods are required, connect each subsequent rod to its predecessor so they appear in a line.). 4.1. Conclusions Human-bear conflict is an on-going concern for wildlife managers that can negatively affect the views of stakeholders towards wildlife management agencies. Education focused on ways to reduce human-bear conflict is improving and becoming more common, but in areas along the boundaries of growing bear populations, understanding of bear habits, habitat, and ecology is often lacking, as well as tolerance to bear activity. Bear damage can be extremely expensive for individual beekeepers, and of those who have experienced it, their attitudes toward bears are 8 Snapping (or cracking) sounds signify a short (or fault) in the fence. Shorts can reduce the number of volts delivered to a bear when it touches the fence because voltage is flowing away from the fence in more than one place (i.e., There is shared distribution of voltage.). 25 likely negative. My results suggest that properly designed, erected, and maintained portable electric fences effectively deter bear access to an attractant. Even the least effective design deterred bear access to bait during 86.4% of all bear visits. Standard wooden pallets (approximately 1.0 x 1.2 m (40 x 48 in.)) are often used as mobile platforms for bee hives and it is common for beekeepers to move entire pallets of bee hives into croplands (pers. comm., Dr. Roger Hoopingarner, Professor Emeritus, President, MBA; Fig. 4). Cost estimates derived from Dadant & Sons, Inc. (www.dadant.com, Hamilton, IL) and Pigeon Mountain Trading Company (www.pigeonmountaintrading.com, Lafayette, GA) in 2012 reveal that the start-up cost for a single bee hive with bee colony is approximately $260. Each additional hive with colony costs approximately $180. Typically, four hives (i.e., bee colonies) are placed onto a single pallet and 1 – 4 pallets are placed in one location in an agricultural area (pers. comm., Dr. Roger 9 Hoopingarner, Professor Emeritus, President, MBA). Hence, the value of a single working apiary is $800 – $2,960, excluding lost income from honey production (if applicable). Although the fences I tested are cost-effective (see Table 3), beekeepers frequently live or work out-ofstate while their bees perform services for land owners. In these cases, arrangements must be made to ensure that electric fences receive regular maintenance. Other precautionary measures can be taken to reduce the risk of bear damage, including hive placement away from riparian areas (especially riparian corridors), positioning hives at least 0.6 m (2 ft.) from the electric fence, and maintaining the area around the fence clear of trees and debris that bears could potentially topple onto the fence. 9 The number and configuration of hives is also dependent upon the crop being pollinated as well as the area of the orchard or field (pers. comm., Dr. Roger Hoopingarner, Professor Emeritus, President, MBA). 26 The bear identification methods used in this study may be of interest to wildlife managers when the costs of marking or sampling (e.g., genetic analyses) bears are of concern. Although I found that trained observers could consistently interpret identification cues from photos and videography, a more thorough evaluation of these methods is required. Specifically, the evaluation should be conducted on bears with known identity, include a greater sample size, and involve more extensive training of test participants. The fences tested in this study might also be useful against other nuisance mammals if the number of tiers is increased and the distance between each tier is decreased. Few North American mammals are known to seek bee hives as a source of food. Bears and striped skunks (Mephitis mephitis) are known pests (Hygnstrom and Craven 1996), while raccoons (Procyon lotor), Virginia opossums (Didelphis virginiana), and mice (Suborder Myomorpha) may also be of nuisance to apiarists (Caron 2000), although mice are primarily an issue during winter (Morland 1938, Caron 2000). Although no skunks were seen visiting our test sites, raccoons, opossums, and mice were frequently observed breaching exclusion fences by moving underneath the lowest fence tier. It may be possible to exclude raccoons and opossums with the fences used in this study, but the step-in posts I used may be inappropriate because of a structural design that limits how low polytape can be placed. Electric fences designed by Honda et al. (2011) were successful at excluding medium-sized mammals (e.g., raccoon dog (Nyctereutes procyonoides), Japanese hare (Lepus brachyurus), Japanese marten (Martes melampus)). These fences, which were also used to exclude larger species (e.g., Asiatic black bear, sika deer (Cervus nippon), wild boar (Sus scrofa)), were comprised of 9 tiers of metal wire, where the distance from the ground to each successive tier was 50 mm (2 in.), 200 mm (8 in.), and 420 mm (≈16.5 in.), increasing by 220 mm (≈8.5 in.) increments thereafter. Honda et al. (2011) also fastened a sheet of nylon bird 27 netting (45 mm (1.75 in.) mesh) along the length of the fence to prevent animals from jumping through the gaps between tiers. A similar fence comprised of polytape would require testing to determine its effectiveness against smaller mammals. 28 APPENDIX 29 Table 1. Pre-baiting, nights to bear detection at pre-baiting sites, and total nights of baiting used to test the efficacy of portable electrical fences for excluding black bears from bait sites, Beaver Lake Hunt Club, Lachine, Michigan, 2010. Fence Design A was tested at 2 sites over 8 nights, Design B at 2 sites over 4 nights, Design C at 3 sites over 6 nights, and Design D at 2 sites over 5 nights. Site numbers 7, 9, and 11 were not used to test fences and were omitted from this table. Site Number Began baiting Bears detected 1 June 24 June 29 Nights baited prior to bear detection 3 2 July 21 - - 3 July 12 July 16 4 July 26 5 Total nights baited (includes fence testing) 15 Fence Designs Tested Number of Nights Tested A 3 10 - - 4 9 C 2 July 29 3 8 D 3 August 3 August 6 3 8 B 3 6 June 17 June 21 2 17 A and B 5 (A), 1 (B) 8 July 28 July 31 or August 2 3 8 D 2 10 July 21 July 24 3 8 C 2 12 July 18 July 24 4 6 C 2 a b a There were no bears detected at Site 2. b No camera on site, but bear activity was evident (e.g., a wide pathway led to the bait area, the vegetated ground where the bait had been placed had been worked over thoroughly: all bait was gone and a swath of bare earth remained). 30 Table 2. Relative strength of concordance associated with Fleiss’ kappa (Landis and Koch 1977). Kappa <0 0.01 – 0.20 0.21 – 0.40 0.41 – 0.60 0.61 – 0.80 0.81 – 0.99 Agreement Less than chance agreement Slight agreement Fair agreement Moderate agreement Substantial agreement Almost perfect agreement 31 Table 3: Equipment list for black bear exclusion fences. Materials were not necessarily purchased from the supplier(s) listed. Prices are approximate and based on those advertised by the suppliers listed (2012). Material Cost Supplier Polytape $1.44/m ($0.44/ft.) Amazon, Kencove, Rural King, Zareba Battery $75/each Walmart Step-in Posts Fiberglass rod (“Fiber Rod”) $3.45/each $4.27/m ($1.30/ft.) Fi-Shock, Kencove, Rural King, Tractor Supply Co., Zareba Kencove Grounding rod $4.79/m ($1.46/ft.) Kencove Fence energizer $90+/each Kencove, Zareba Double-insulated wire $20/roll Amazon, Kencove, Tractor Supply Co. Brass clamps Volt meter $1.25/each $10 – $35/each Kencove ACE Hardware, The Deer Shock Depot, Fi-Shock Wire ties* $0.13/m ($0.04/ft.) Any hardware store Post pounder $30/each Home Depot, Lowe’s, Tractor Supply Co. *Wire ties can be cut from a spool of ≈14 gauge aluminum wire. ACE Hardware: ACE Hardware Corporation, Oak Brook, IL, www.acehardware.com. Amazon: Amazon.com, Inc., Seattle, WA, www.amazon.com. Fi-Shock: Fi-Shock Animal Containment Systems, Woodstream Corporation, Lititz, PA, www.fishock.com. Home Depot: The Home Depot, Atlanta, GA, www.homedepot.com. Kencove: Kencove Farm and Fence Supplies, Blairsville, PA, www.kencove.com Lowe’s: Lowe’s, Mooresville, NC, www.lowes.com. Rural King: Rural King Supply, Mattoon, IL, www.ruralking.com. The Deer Shock Depot: The Deer-Shock Depot, Sandwich, MA, www.electric-deer-fence.com. Tractor Supply Co.: Tractor Supply Company, Brentwood, TN, www.tractorsupply.com. 32 Table 3 (cont’d) Walmart: Walmart, Bentonville, AR, www.walmart.com. Zareba: Zareba Systems, Inc., Woodstream Corporation, Lititz, PA, www.zarebasystems.com. 33 Table 4. Bear activity and fence performance, per fence design, during fence testing at the Beaver Lake Hunt Club, Lachine, MI, 2010. Fence Design Bear Visits Detection Nights* BFIs Total Duration of BFIs (min.) A 24 7 of 8 52 246.17 4.73±1.39 B 12 4 of 4 30 30.05 1.00±0.15 C 22 5 of 6 48 86.45 1.80±0.42 D 15 4 of 5 38 70.52 1.86±0.28 *The number of nights in which a bear was detected by cameras at the fence. 34 Mean Duration of BFIs (min.) Table 4 (cont’d) Break plane of or touch † fence Breach without Bait Accessed Breach with Bait Accessed Fence Design Unique Bears Visits/Bear, BFIs/Bear A 4 [6.0] [13.0] 11 (21%) 0 3 B 6 [2.0] [5.0] 5 (17%) 2 0 C 4 [5.5] [12.0] 17 (35%) 0 3 D 3 [5.0] [12.7] 6 (16%) 1 0 †The number and percentage of BFI’s that resulted in a bear either breaking the plane of the fence or touching the fence. 35 Table 4 (cont’d) Fence Design BFIs with Breach (%) Fence § Effectiveness (%) A 5.8 94.2 B 6.7 100.0 C 6.3 93.8 D 2.6 100.0 § Percent fence effectiveness: (breach with bait accessed/BFIs * 100 = x) then (100 - x = % fence effectiveness). 36 Table 5: Product specifications for DVR cameras, AVTECH Model: KPC139D. Pick up Element: 1/3 in. Color CCD image sensor S/N Rate: More than 48dB (AGC off) White Balance: ATW Number of pixels: 512(H)x492(V ) Electronic Shutter: 1/60 (1/50) to 1/100,000 sec. IP Rating: IP67 Resolution: 520TV Lines Lens: F3.6mm/F2.0 Video Output: 1.0 Vp-p composite, 75Ω Min. Illumination: 0.2Lux/F2.0/0 Lux (IR on) Lens Angle: 92.6° Power Source (±10%): DC12V IR LED: 35 units IRIS Mode: AES Dimensions (mm): 183.69(L)x85.24 (W)x150(H) Effective Range: 25 m AGC: Auto Weight (gw): 430 37 Table 6. Longevity and mean ± (1 SE) wire voltage of two 12 V deep-cycle marine/recreational vehicle batteries subjected to different power management scenarios from June – November, 2011. Tested batteries were used to power the electric fence energizer during the 2010 field season. A test was complete once the battery had insufficient charge to power the fence energizer or 14 days had passed. Test 1 2 Test Type Continuous power Continuous power a Test Duration (days) Wire Voltage (V) 14 5,050 (17) 14 5,062 (26) 3 3 5,133 (33) 4 Switched off and on daily 14 5,022 (26) 5 a Switched off and on daily Switched off and on daily 14 5,102 (41) One voltage reading was taken daily for Tests 1-2. Two voltage readings (“before switched off” and “after switched on”) were taken daily for Tests 3-5. The average ground voltage reading was 0 V. 38 Table 7. Bear-fence-interaction (BFI) and other bear activity, per bear, during fence testing at the Beaver Lake Hunt Club, Lachine, MI, 2010. Bear ID Total BFIs Cumulative duration BFIs (min.) Mean duration BFIs (min.) BFIs touch fence/break plane (%) Number of sites visited Fence designs interacted with†† Total number of days BFIs observed 1* 20 99.15 4.96±2.97 20.00 2 A 2 3 1.80 0.60±0.13 33.33 1 B 3 10 36.07 3.61±0.99 20.00 1 A and B 4 2 1.40 0.70±0.50 50.00 1 B 5* 2 29.02 14.51±14.16 100.00 1 A 6* 22 83.13 3.78±1.54 18.18 1 A 7 2 1.05 0.53±0.01 0.00 1 C 8† 11 9.62 0.87±0.32 45.45 1 C 9*† 17 44.05 2.59±1.05 47.06 1 C 10 29 53.63 1.85±0.36 24.14 2 C and D 11 5 7.47 1.49±0.52 0.00 1 D 12 22 41.15 1.87±0.38 13.64 1 D 13 17 17.88 1.05±0.18 0.00 1 B 14 2 0.67 0.33±0.02 0.00 1 B 15 4 7.10 1.78±0.69 50.00 1 B *A bear that fed on bait. These individuals fed at length which increased the cumulative and mean duration of BFIs. †Denotes a family of bears: One sow with cubs. ††See Fig. 3 for description of fence designs. 39 5 1 2 1 1 3 1 2 4 5 1 1 2 2 1 Figure 1. Physical characteristics that were used to identify individual bears. A and B portray lines of equal length based on an obvious morphometric characteristic (e.g., distance from belly to top of foot (A), distance between the ears (B)) were used to differentiate variation in body size among individual bears. When observed, tags and pelage markings were useful (C). Facial markings, shapes, and colors were also useful at times (D). For interpretation of the references to color in this and all other figures, the reader is referred to the electronic version of this thesis. A. B. C. D. 40 Figure 2. Examples slides from the bear photo identification questionnaire administered to undergraduate students (n=35) that were members of the Fisheries and Wildlife Club at Michigan State University. 14. These bears are: 17. These bears are: A. Not the same B. Probably not the same C. Probably the same D. The same A. Not the same B. Probably not the same C. Probably the same D. The same 41 Figure 3. Fence designs that were tested for excluding black bears (Ursus americanus) in the Beaver Lake Hunt Club, Lachine, MI, June – August, 2010. All fence designs used the same wire type: 1.3 cm (0.5 in.) white polytape. 42 Figure 4. This illustration depicts one fence design (Design D) used to evaluate the efficacy of portable, electrical fences for excluding black bears. In this example, the bear has been excluded from a bee yard. Fundamental fence materials are labeled: step-in post (A), fiberglass rod (B), polytape (C), fence energizer (D), double-insulated wire (E), and grounding rod with clamp (F). 43 Figure 5. The DVR housing consisted of a water-tight, aluminum, militarygrade foot-locker, a custom-welded steel frame with anchor rods, and chains. 44 Figure 6. Pearson’s product moment correlation (Sokal and Rohlf 1981) between voltage meter readings from the Kencove Dual Fault Finder and Volt Meter and the Fi-Shock Digital Voltage Reader during fence testing. 5500 y = -0.0821x + 4632.9 ² R = 0.0063 Kencove Readings (V) 5000 4500 4000 3500 3000 4000 4200 4400 4600 4800 5000 Fi-Shock Readings (V) 45 5200 5400 5600 LITERATURE CITED 46 LITERATURE CITED Barnes, B. V., and W. H. Wagner Jr. 2004. Michigan Trees: A Guide to the Trees of the Great Lakes Region. University of Michigan Press, Ann Arbor, USA. Beckmann, J. P., C. W. Lackey, and J. Berger. 2004. Evaluation of deterrent techniques and dogs to alter behavior of ‘‘nuisance’’ black bears. Wildlife Society Bulletin 32:1141– 1146. Burgett, M., S. Daberkow, R. Rucker, W. Thurman. 2010. U.S. Pollination Markets: Recent Changes and Historical Perspective. American Bee Journal 150(1):35-41. Caron, D.M. 2000. Pests of honey bees. MAAREC Publication 4.3, Mid-Atlantic Apicultural Research & Extension Consortium, University of Delaware, Newark, DE. Caron, D. M., and J. L. Bowman. 2004. Bears and bees. MAAREC Publication 4.10, MidAtlantic Apicultural Research & Extension Consortium, University of Delaware, Newark, DE. Carter, N.H., D.G. Brown, D.R. Etter, and L.G. Visser. 2010. Predicting black bear habitat suitability in Michigan’s northern Lower Peninsula. Ursus 21:57-71. Clark, J. D., S. Dobey, D. V. Masters, B. K. Scheick, M. R. Pelton, and M. E. Sunquist. 2005. American black bears and bee yard depredation at Okefenokee Swamp, Georgia. Ursus 16(2):234-244. Cohen, J. 1960. A coefficient of agreement for nominal scales. Educational and Psychological Measurement 20(1):37-46. Conover, M. 2001. Resolving Human-Wildlife Conflicts: The Science of Wildlife Damage Management. Lewis Publishers, Boca Raton. Creel, E.M. 2007. Effectiveness of deterrents on black bears (Ursus americanus) to anthropogenic attractants in urban-wildlife interfaces. Thesis, Humboldt State University, Arcata, CA, USA. Retrieved from http://humboldtdspace.calstate.edu/xmlui/handle/2148/296 on April 2, 2012. Fish and Wildlife Service (FWS). November 3, 2009. Pollinators. Retrieved from http://www.fws.gov/Pollinators/ on April 6, 2010. Fleiss, J.L. 1971. Measuring nominal scale agreement among many raters. Psychological Bulletin 76(5):378-382. 47 Gamer, M., J. Lemon, I. Fellows, and P. Singh. 2012. Package ‘irr’; Various coefficients of interrater reliability and agreement. Retrieved from http://www.rproject.org/web/packages/irr/irr.pdf on September 28, 2012. Garshelis, D.L. 1989. Nuisance bear activity and management in Minnesota. Pp. 163-168 in M.Bromley, editor. Bear-People Conflicts: Proceeding of a Symposium on Management Strategies. Northwest Territories Department of Renewable Resources, April 6-10, 1987, Yellowknife, Canada. Gates, N.L., J.E. Rich, D.D. Godtel, and V. Hulet. 1978. Development and evaluation of anticoyote electric fencing. Journal of Range Management 31(2):151-153. Green Township Planning Commission. 2007. Green Township, Alpena County Land Use Master Plan. Retrieved from http://www.nemcog.org/downloads/green_township_master_plan_2007.pdf on July 12, 2012. Herrero, S. and A. Higgins. 2003. Human injuries inflicted by bears in Alberta: 1960-98. Ursus, 14(1):44-54. Hoare, R. 2003. Fencing and other barriers against problem elephants. AfESG Website HEC Section,Technical Brief Series. Retrieved from IUCN African Elephant Specialist Group, http://www.iucn.org/afesg on November 9, 2012. Honda, T., Y. Miyagawa, H. Ueda, and M. Inoue. 2009. Effectiveness of newly-designed electric fences in reducing crop damage by medium and large mammals. Mammal Study 34(1):13-17. Honda, T., H. Kuwata, S. Yamasaki, Y. Miyagawa. 2011. A low-cost, low-labor-intensity electric fence effective against wild boar, sika deer, Japanese macaque and medium-sized mammals. Mammal Study, 36(2):113-117. Huygens, O.C., and H. Hayashi. 1999. Using electric fences to reduce Asiatic black bear depredation in Nagano prefecture, central Japan. Wildlife Society Bulletin 27(4):959964. Hygnstrom, S.E. 1994. Black bears. Pp. C5-C15 in Prevention and Control of Wildlife Damage. 1994. Editors, S.E. Hygnstrom, R.M. Timm, and G.E. Larson. University of Nebraska-Lincoln. 2 vols. Hygnstrom, S., and S. Craven. 1996. Bear damage and abatement in Wisconsin. University of Wisconsin-Extension, R-10-96-2M-.75, Madison, WI. 6pp. Jonker, S. A., J. A. Parkhurst, R. Field, and T. K. Fuller. 1998. Black bear depredation on agricultural commodities in Massachusetts. Wildlife Society Bulletin 26:318-324. 48 Jordan, R.H. 1976. Threat behavior of the black bear (Ursus americanus). Bears: Their Biology and Management, Vol. 3, A selection of the papers from the Third International Conference on Bear Research and Management, Binghamton, New York, U.S.A., and Moscow, U.S.S.R., June 1974. IUCN Publications New Series no. 40, pp. 57-63. Landis, J.R., G.G. Koch. 1977. The measurement of observer agreement for categorical data. Biometrics 33:159-74. Leigh, J. 2007. Effects of aversive conditioning on behavior of nuisance Louisiana black bears. Thesis, Louisiana State University, Baton Rouge, USA. Living with Wildlife Foundation, Inc. (LWWF) 2009. Living with Predators Resource Guide – Practical electric fencing resource guide: controlling predators. Living with Predators Project, Condon, Montana. Retrieved from www.lwwf.org/electric_fence_2009.pdf on September 9, 2011. 70pp. Maehr, D. S., and J. R. Brady. 1982. Florida black bear-beekeeper conflict: 1981 beekeeper survey. American Bee Journal 122:372-375. Maehr, D. S. 1984. Black bear depredation on bee yards in Florida. Eastern Wildlife Damage Control Conference 1:133-135. Maryland Department of Natural Resources. September 28, 2004. Black Bears and Beekeeping in Maryland. Retrieved from http://dnr.maryland.gov/wildlife/hunt_trap/blackbear/bbandbees.asp on March 31, 2010. Masterson, L. 2006. Living with Bears: A Practical Guide to Bear Country. Masonville, CO: PixyJack Press, LLC. McKillop, I. G. and R. M. Sibly. 1988. Animal behaviour at electric fences and the implications for management. Mammal Review 18(2):91-103. Meadows, L. E., W. F. Andelt, and T. I. Beck. 1998. Managing bear damage to bee hives. Colorado State University Cooperative Extension Report No. 6.5619, Fort Collins, CO. Michigan Department of Natural Resources. 2008. A review of bear management in Michigan. Michigan Department of Natural Resources – Wildlife Division, Lansing, MI. Michigan Department of Natural Resources. 2009. 2009 Michigan Hunting and Trapping Guide. Retrieved from http://www.michigan.gov/dnr/0,1607,7-15310371_14724-148682--,00.html on May 15, 2010. Michigan Department of Natural Resources and Environment. 2010. 2010 Michigan Bear Hunting Digest. Retrieved from http://www.michigan.gov/documents/dnre/WLD_2010_Bear_Digest_Web_318904_7.pdf on May 13, 2010. 49 Michigan Land Use Institute (MLUI). 2009. Northwest Michigan’s farm factor: Economic impacts, challenges, and opportunities. Retrieved from http://www.mlui.org/downloads/AgWhitePaperFinal.pdf on August 3, 2012. Montana Fish, Wildlife & Parks (MFWP). 2010. Bears and electric fencing: a starter’s guide for using electric fencing to deter bears. Helena, Montana. Retrieved from http://fwpiis.mt.gov/content/getItem.aspx?id=48893 on September 9, 2011. 11pp. Morland, D.M.T. 1938. Recent investigations into bee-keeping at Rothamsted. Journal of the Royal Society of Arts 86(4451):394-404. Morse, R. A. and N. W. Calderone. 2000. The value of honey bees as pollinators of U.S. crops in 2000. Retrieved from http://www.utahcountybeekeepers.org/Other%20Files/Information%20Articles/Value%2 0of%20Honey%20Bees%20as%20Pollinators%20-%202000%20Report.pdf on April 6, 2010. R Development Core Team. 2010. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. Retrieved from http://www.Rproject.org. Reidy, M.M., T.A. Campbell, D.G. Hewitt. 2008. Evaluation of electric fencing to inhibit feral pig movements. Journal of Wildlife Management 72(4):1012-1018. Sokal, R.R., F.J. Rohlf. 1981. Biometry – The Principles and Practice of Statistics in Biological Research. San Francisco: W.H. Freemann. Storer, T. I., G. H. Vansell, and B. D. Moses. 1938. Protection of mountain apiaries from bears by use of electric fence. Journal of Wildlife Management 2:172-178. Tolhurst, B.A., A.I. Ward, R.J. Delahay, A. MacMaster, T.J. Roper. 2008. The behavioral responses of badgers (Meles meles) to exclusion from farm buildings using an electric fence. Applied Animal Behaviour Science 113:224-235. United States Department of Agriculture (USDA) Natural Resources Conservation Service. 2006. Land Resource Regions and Major Land Resource Areas of the United States, the Caribbean, and the Pacific Basin. United States Department of Agriculture Handbook. 296(296). USDA Forest Service. 2007. Specification for portable electric fence systems as potential a alternative methods for food storage. USDA Forest Service, Technology and Development Program, Missoula, Montana, 0723-2305P-MTDC. 10pp. United States Environmental Protection Agency. October 2008. Pesticide Issues in the Works. Retrieved from http://www.epa.gov/pesticides/about/intheworks/honeybee.htm on April 6, 2010. 50 University of California Cooperative Extension (UCCE). 1991. Electric fence energizers. Napa and Solano Counties U.C.C.E., Livestock/Range Management Program, Napa, California, and Fairfield, California, Livestock & Range report no. 913. 4 pp. vanEngelsdorp, D. and M. D. Meixner. 2010. A historical review of managed honey bee populations in Europe and the United States and the factors that may affect them. Journal of Invertebrate Pathology 103:S80-S95. Washington Department of Fish and Wildlife (WDFW). 2007. Living with wildlife: Black bears. Washington Department of Fish and Wildlife, Wildlife Program, Olympia, Washington. 7 pp. Retrieved from http://wdfw.wa.gov/living/bear.pdf on September 3, 2012. Wood, J.M. 2007. Understanding and computing Cohen’s Kappa: A tutorial. WebPsychEmpiricist. Retrieved from http://wpe.info/papers_table.html on November 15, 2012. 51