Human–Wildlife Interactions 10(2):268–282, Fall 2016 Effi cacy of non-lead ammunition for culling elk at Theodore Roosevelt National Park B(cid:725)(cid:714)(cid:724)(cid:718) E. M(cid:716)C(cid:714)(cid:727)(cid:727), Resource Management, Wildlife, Theodore Roosevelt National Park, 315 Second Avenue, P.O. Box 7, Medora, ND 58645, USA [email protected] W(cid:722)(cid:725)(cid:725)(cid:722)(cid:714)(cid:726) W(cid:721)(cid:722)(cid:733)(cid:736)(cid:728)(cid:731)(cid:733)(cid:721), Resource Management, Theodore Roosevelt National Park, 315 Second Avenue, P.O. Box 7, Medora, ND 58645, USA R(cid:728)(cid:715)(cid:718)(cid:731)(cid:733) A. N(cid:718)(cid:736)(cid:726)(cid:714)(cid:727), Department of Biology, University of North Dakota, 10 Cornell Street, Stop 9019, Grand Forks, ND 58202, USA Abstract: During 2010 to 2013, park staff and public volunteers culled 983 elk (Cervus elaphus) from Theodore Roosevelt National Park (United States) utilizing non-lead rifl e ammunition as part of a sanctioned herd management operation. Because there is little empirical evidence available on the performance of non-lead ammunition, staff recorded information on tools and techniques relevant to the scenarios under which elk were culled and the outcome of each engagement. We also conducted a fi ring range experiment to evaluate the precision of non- lead ammunition used in park fi rearms. Specifi c objectives were to identify program factors predicting effi cient destruction of elk with non-lead ammunition and to evaluate the precision of non-lead ammunition in National Park Service (NPS) fi rearms to facilitate accurate shot placement. To address these objectives, we conducted multivariate ordinal regression analyses of 13 variables, including bullet type, marksman type, shot distance, initial shot impact location, number of shots fi red, and need for a killing shot, as predictors of distance traveled by elk after being shot. Among 921 elk removals evaluated, mean shot distance was 182 meters, and the median and mode of distance traveled were 46 m and 0 m, respectively. Multivariate analyses revealed that shots to the head and neck were most eff ective, followed by those striking the shoulder and chest. Heavier bullets should be used whenever practical. Mean group size for non-lead ammunition fi red through NPS fi rearms was 50 mm at 91 m, with minimum and maximum group sizes of 18.8 and 98.6 mm, respectively. We found that non-lead ammunition provided the necessary precision for accurate shot placement in spot and stalk hunting conditions and that these bullets typically accomplished instantaneous or near-instantaneous incapacitation of elk whenever vital areas of the body were impacted. We conclude that non-lead bullets are eff ective for wildlife management and hunting scenarios. Key words: ammunition, bullets, effi cacy, elk, lead-free, management, non-lead, NPS Elk (Cervus elaphus) were extirpated from agencies, nonprofi t organizations, and Native southwest North Dakota during the late 1800s American tribes. However, in 2002, when the (Seabloom 2011; Figure 1). In 1985, 47 elk were population was again approaching its upper reintroduced to the South Unit of Theodore limit, a moratorium was placed on translocation Roosevelt National Park (TRNP), joining bison of elk due to concerns regarding chronic wasting (Bison bison), pronghorn (Antilocapra americana), disease (CWD; Natl. Park Serv. 2010a). and bighorn sheep (Ovis canadensis) as restored In the absence of an authorized reduction elements of the native biotic community tool, and with the elk population growing (Harmon 1986). With low natural mortality unchecked, the National Park Service formally and few natural predators, management initiated the development of an Environmental intervention to prevent overpopulation and Impact Statement in 2004 to identify alternative resource degradation was anticipated. A forage management methods for elk population allocation study was conducted during the control. Ultimately, direct reduction with 1980s and early 1990s to determine population fi rearms was selected as the primary tool in management objectives for elk and other large the Record of Decision (Natl. Park Serv. 2010b). ungulates inhabiting the park (Irby et al. 2002). It was stipulated that for the initial reduction When elk numbers exceeded the maximum phase, a period expected to take 4–5 years, determined population objective of 400 park staff would lead teams of volunteers to individuals in 1993 and 2000, helicopter-assisted decrease the population below the established roundups were conducted and live elk were upper limit. removed and transferred to other government Volunteer shooters were required to comply Non-lead ammunition for culling elk • McCann et al. 269 Figure 1. Elk in the badlands of North Dakota at Theodore Roosevelt National Park. (Photo courtesy of R. and L. Honeyman) with directives related to federal fi rearms citing the added expense, a limited variety of laws and use of non-lead ammunition in loadings, and reputed accuracy and performance national parks. Meat (quarters, backstraps, problems (Friend et al. 2009, Knott et al. 2009, and tenderloins) was recovered from culled Caudell et al. 2012). Issues of expense and animals, packed out of backcountry locations variety are indisputable, as non-lead projectiles to a centrally located facility, then distributed are typically loaded in premium ammunition to project volunteers, state food pantries across products, and the vast array of lead projectile North Dakota, and various Native American off erings largely outnumbers non-lead options organizations. Brain stem and lymphatic tissue (Knott et al. 2009, Thomas 2013). However, were collected from all animals and screened there is litt le empirical evidence pertaining to for CWD. Remaining portions of carcasses were precision and performance of non-lead rifl e left to recycle into the environment. ammunition in fi eld conditions (Caudell et al. Non-lead ammunition was required for use 2012, Caudell 2013). by staff and volunteers at TRNP to comply with Gremse et al. (2014) compared lead and agency policy and to prevent exposure of lead non-lead ammunition performance when by meat recipients and animals scavenging fi red into ballistic soap and reported similar carcasses (Ross-Winslow and Teel 2011). terminal ballistic behavior. In Germany, Secondary consumption of lead projectiles has comparison of thoracic wound channels in 34 been implicated in mortality of birds of prey large ungulates produced by lead and non-lead and several game and non-game species (Fisher projectiles revealed no detectable diff erence et al. 2006, Hunt et al. 2006, Pauli and Buskirk between bullet types (Trinogga et al. 2013). In 2007, Rogers 2010). Further, lead fragments the United Kingdom, researchers conducted pose a human health risk where wild game qualitative analyses, where shot outcomes were harvested with lead ammunition is consumed ranked by professional marksmen, to evaluate (Hunt et al. 2009, Kosnett 2009). However, the ammunition performance in terms of perceived general public and some wildlife professionals accuracy and animal incapacitation. Analysis have resisted the use of non-lead ammunition, of approximately150 deer of multiple species, 270 Human–Wildlife Interactions 10(2) removed using lead and non-lead ammunition, Our specifi c objectives are to identify program resulted in equal rankings for accuracy and factors predicting effi cient destruction of elk slightly higher scores for lethality of lead with non-lead ammunition and to evaluate bullets (Knott et al. 2009). However, the authors the precision of non-lead ammunition in NPS concluded that diff erences in killing power fi rearms to facilitate accurate shot placement. were marginal (Knott et al. 2009). Killing power has been described as levels of Methods incapacitation, ranging from instantaneous (i.e., Elk reduction operations were conducted a shot stopping the animal immediately), to in the South Unit of Theodore Roosevelt near instantaneous (stopping the animal within National Park, which encompasses 18,756 ha 1–30 seconds), and rapid (stopping the animal in the badlands of southwest North Dakota within 30 seconds to 5 minutes; Caudell et al. (centroid coordinates: 46°57’12” N, 103°27’33” 2012, Caudell 2013). The level of incapacitation W). Park staff led teams of volunteers in fi eld achieved in any situation depends on the activities to cull elk during the fi rst 2 years of location of the bullet strike (i.e., tissue and the reduction (during October 2010 to January organs impacted), the mass and shape of the 2011 and October to December 2011) to lower bullet, velocity, depth of penetration, and other herd numbers to <400 animals (Figure 2). NPS factors relevant to fi eld conditions (Fackler employees alone conducted maintenance-level et al. 1988, MacPherson 2005, Caudell 2013). removal operations subsequently (during Therefore, it may be challenging to identify October to December 2012 and November the specifi c cause for ultimate incapacitation to December 2013). Volunteers provided outcomes without close examination of fi eld their own shooting equipment and supplies, tools and the scenarios under which they representing a variety of fi rearm makes, are employed. However, where fi eld data models, calibers, and ammunition types. describing tools and techniques for destruction However, volunteers primarily used 3 brands of hundreds of animals are available, much of projectiles: (1) Barnes Bullets Inc. (Mona, may be learned through objective analysis. The UT), (2) Hornady Manufacturing Company elk reduction program at TRNP provides one (Grand Island, NE), and (3) Nosler Inc. (Bend, such scenario. OR; Table 1). Prior to fi eld work, the park Here, we present a quantitative analysis of required volunteers to score 3 of 5 shots within elk reduction program data and a fi ring range a 200 mm diameter circle at 183 m to qualify experiment aimed at evaluating the effi cacy rifl es and ammunition for use on the project. of non-lead ammunition to serve as a basis We then recorded the chambering, caliber, from which fi eld procedures may be refi ned. type of bullet, and bullet weight used by each individual, and we subsequently categorized rifl e chamberings as “standard” or “magnum” velocity for our analysis (Table 1). With few exceptions, park staff used rifl es chambered in .308 Winchester, fi ring 150 grain and 165 grain Barnes TSX bullets. We used marked animals fi tt ed with combination radio-telemetry and satellite tracking collars to locate groups of elk. When elk were located, team leaders att empted to maneuver marksmen to within 183 m of targeted animals, though longer shots were Figure 2. Staff member (right) and volunteer (left) engaged in elk occasionally taken when distances reduction activities at Theodore Roosevelt National Park during 2010. (Photo courtesy of J. Powers, Natl. Park Serv.) could not be reduced and Non-lead ammunition for culling elk • McCann et al. 271 Table 1. Fourteen variables collected during elk reduction activities at Theodore Roosevelt National Park from 2010 to 2013; TSX® = triple shock X; TTSX® = tipped triple shock X; GMX® = gilding metal expanding; and ETIP® = energy-expansion-cavity tipped. Variable Type Description Distance traveled Ordinal/categorical 0, 46, 91, 183, 274, 366+ m Year Categorical Years 1, 2, 3, and 4 Marksman type Categorical Volunteer or staff a No. elk in group Categorical 1, 2–10, 11–25, 26–50, 51–100, >100 Shot distance Continuous Estimated: using laser range fi nder; 27–366 m No. shots fi red Continuous Shots fi red at elk; not striking elk Killing shot required Categorical Yes or no Initial shot impact Categorical Location of fi rst bullet strike: head, neck, shoulder, chest, abdomen, leg Rifl e caliber Categorical 0.257, 0.277, 0.284, 0.308, and 0.338 Velocity rank Categorical Standard or magnum judged ad hoc by belting and case capacity Bullet type Categorical Barnes (TXS, TTSX), Hornady (GMX), and Nosler (ETIP) Bullet weight Continuous 115–225 grains Sex of elk Categorical Male or female Age of elk Categorical Calf (<1 yr), yearling (>1 yr), adult (≥2 yr); estimated by size and dentition a In few instances, staff returning as volunteers during later project years were still considered “staff ” in statistical analyses. where environmental conditions (e.g., wind the crania and mandible; (2) neck: shots striking speed) allowed. Typically, 1 marksman was cranial to the shoulder but caudal to the head; designated the primary shooter, and the team (3) shoulder: bullet impacts to the upper front leader provided a follow-up shot if necessary. leg, scapula, spine, and associated musculature Marksmen would engage 1 animal at a time, cranial to the exposed rib cage; (4) chest: shots targeting the shoulder and chest, and confi rm penetrating the rib cage, sternum, and spine, that the elk was down before fi ring at a second cranial to the abdomen; (5) abdomen: bullet animal in the group. strikes to the musculature, organs, and bones of the body caudal to the chest; and (6) leg: Identifi cation of factors contributing shots striking distal portions of limbs extending to effi cient kills below the abdomen and chest. Finally, we For each elk dispatched, we recorded animal included project year as a predictor variable to and herd demographics, marksman type (i.e., account for variance associated with changes in volunteer or staff ), initial shot distance, number operational intensity and animal behavior over of shots fi red, distance traveled after shot, time as a work-environment factor. In total, we and initial shot impact location (Table 1). We recorded 14 variables relevant to the scenarios measured initial shot distance to the nearest under which elk were culled and the outcome meter with Nikon Rifl e Hunter 550, laser of each engagement (Table 1). rangefi nders (Nikon, Inc., Melville, NY), and we Quantitative analyses. We conducted statistical recorded distance traveled by elk as an ordinal analyses using Statistica (Statsoft, Inc., Tulsa, variable ranging from 0 to 366+ m, based on OK) and program R (R Foundation for Statistical manual estimates, visually judged and paced on Computing, Vienna, Austria). We identifi ed foot (Table 1). We partitioned initial shot impact “distance traveled” as our response variable, into 6 anatomical categories: (1) head: shots to because distance moved by animals after bullet 272 Human–Wildlife Interactions 10(2) Figure 3. A sample of 6 non-lead bullets recovered from elk carcasses at Theodore Roosevelt National Park, viewed from the side (at bottom) and above (at top), demonstrating commonly observed controlled expansion (2 bullets at left) versus less frequent intermediate expansion (center right), occasional fragmentation (center left and second from right), and rare instances of non-expansion and bullet tumbling (at right). (Photo courtesy of Blake McCann) impact has been previously identifi ed as a signifi cant relationships and identify candidate measure of shot effi cacy, and time associated predictor variables for multivariate analysis. with animal movements is directly related to Next, we performed an ordinal regression of levels of incapacitation (Ruth and Simmons all candidate predictors on distance traveled 1999, Maiden 2009, Caudell 2013). The top under all eff ects and best subsets parameters, speed of elk has been reported as 72 km per employing model building using the Akaike hour on fl at land, though 48 km per hour is a Information Criterion (AIC). We then conducted more typical pace, and we expect that speed model-averaged, multi-model inference using would be reduced where steep slopes and thick package MuMIn in program R to elucidate the vegetation must be navigated (Willoughby most parsimonious regression model through 1974, Ballard 2012). Given an ad hoc speed of 24 evaluation of sample size corrected AIC scores to 32 km per hour, to account for complexity of across all models with weights summing to terrain at TRNP, an elk could traverse 200–267 ≥95%. Finally, we calculated descriptive statistics m within 30 seconds. Therefore, our response to examine relationships further explaining variable categories generally correspond to independent eff ects of categorical predictor previously defi ned levels of incapacitation as variables. follows: instantaneous (0 m), near instantaneous Precision of non-lead ammunition (46, 91, 183 m), and rapid (274 and 366+ m; Caudell et al. 2012, Caudell 2013). We performed a shooting test of 8 rifl es We performed multiple exploratory analyses and factory-loaded ammunition used for elk to detect trends in the dataset and to identify reduction operations at TRNP. We conducted potential areas of covariance and interaction the test at a fi ring range near Belfi eld, North among predictors. We conducted Spearman Dakota on October 18, 2013. Area weather rank correlation of continuous and ordinal records indicate that ambient temperature variables, correspondence analyses between was between 0.6-8.3°C, wind speed was 21–40 categorical variables, and Analysis of Variance km per hour from the west-northwest, and (ANOVA) of continuous variables by categorical barometric pressure was 762 mm of mercury variable groupings. We then evaluated variance (<htt p://www.wunderground.com>, September in distance traveled in a series of logit-based, 3, 2015, unpublished data). The fi ring range was univariate ordinal regression analyses to detect situated with targets positioned to the north Non-lead ammunition for culling elk • McCann et al. 273 Table 2. Univariate regression results of 13 individual predictor NY) to actuate the fi ring variables on "distance traveled," representing elk reduction data mechanism. collected at Theodore Roosevelt National Park during 2010 to 2013. For each rifl e, we fi red Signifi cance at α = 0.05 used to screen variables for multiple regression analysis. a fouling round and then performed a 5-round Eff ect df Wald stat. P Included in multiple course of fi re at 91 m, regression analysis? aligning the crosshairs All intercepts ≤0.001*** at target center between shots. We allowed rifl e Year 3 83.12 ≤0.001*** Yes barrels to cool to perceived Marksman 1 0.05 0.82 No ambient temperature and type then completed a second No. elk 5 5.82 0.32 No 5-round course of fi re with Killing shot 1 16.64 ≤0.001*** Yes each rifl e. We recorded the required velocity of each round using Initial shot 5 103.97 ≤0.001*** Yes a Beta Master chronograph impact (Shooting Chrony Inc., Velocity rank 1 4.13 0.04* Yes Amherst, NY). We then Bullet type 3 19.0001 0.0003** Yes measured the maximum Rifl e caliber 4 5.58 0.23 No spread of bullet strikes to the nearest 0.03 mm from Bullet weight 1 5.35 0.02* Yes the visually estimated center Sex of elk 1 0.53 0.47 No of bullet holes with a dial Age of elk 2 0.06 0.97 No caliper for each rifl e across Shot distance 1 7.21 0.007** Yes both courses of fi re. Finally, we performed a Student’s No. shots 1 74.72 ≤0.001*** Yes fi red t-test to detect variation in patt ern size and velocity *P ≤ 0.05 **P ≤ 0.01 between courses of fi re and ***P ≤ 0.001 conducted linear regression of mean velocity of 5-shot of the shooter, which presented a substantial strings on patt ern size to test for variation in crosswind during our evaluation period. precision associated with diff erential charging of Firearms were Remington Model 700 rifl es ammunition. (Remington Arms Company, Ilion, NY), equipped with sporter-weight barrels, and Results chambered in .308 Winchester. All rifl es were Effi ciency of kills fi tt ed with Nikon Monarch (model #8421) 4–16 Prior to initiation of reduction activities in variable-power scopes with 42 mm objective November 2010, the elk herd was estimated lenses (Nikon, Inc., Melville, NY). However, at approximately 1,200 individuals (Natl. 2 rifl es were equipped with aftermarket, Park Serv., unpublished data). During 2010 to adjustable triggers, and one was glass bedded. 2013, park staff and volunteers shot 1,000 elk. To maintain consistency of loaded rounds, Among these, 983 animals were recovered we used a single case of Federal Premium® and 17 were wounded and lost, resulting in a Ammunition loaded with 150-grain Barnes TSX 98.3% recovery rate overall. Notably, 279 elk bullets for all courses of fi re. We cleaned the were dispatched with a single shot. Dozens of bore of each rifl e and ensured that actions were bullets were recovered from elk carcasses over clear of debris and in good working condition the course of the program, with most exhibiting prior to the test. We then mounted each rifl e in deep, controlled expansion (Figure 3). All elk a Hyskore Model 30185 machine rest (Hyskore, tested negative for CWD, and 67,170 kg of meat Huntington Station, NY) and fi tt ed a hydraulic were distributed to qualifi ed recipients. trigger release (Hyskore, Huntington Station, After fi ltering records to remove cases 274 Human–Wildlife Interactions 10(2) Table 3. All eff ects (top) and best subset (bott om) multiple regression analysis results for elk reduction data collected at Theodore Roosevelt National Park during 2010–2013. Model Eff ect df Wald stat. P All eff ects: Intercept 5 621.00 ≤0.001*** Shot distance 1 0.09 0.77 No. shots 1 33.15 ≤0.001*** Bullet Weight 1 2.57 0.11 Year 3 67.91 ≤0.001*** Killing shot required 1 5.17 0.02* Initial shot impact 5 88.60 ≤0.001*** Velocity rank 1 0.75 0.39 Brand acronym (bullet type) 3 4.04 0.26 Best subset: Intercept 5 621.83 ≤0.001*** No. shots 1 36.16 ≤0.001*** Bullet weight 1 4.12 0.04* Year 3 88.04 ≤0.001*** Killing shot required 1 5.18 0.02* Initial shot impact 5 88.95 ≤0.001*** *P ≤ 0.05 **P ≤ 0.01 ***P ≤ 0.001 Table 4. Covariate relative importance values derived from analysis of all models with Akaike Information Criterion (AIC) weights summing to 95% or greater (n = 22) in model-averaging, multi-model inference analysis of eight variables pertaining to elk reduction activities at Theodore Roosevelt National Park during 2010 to 2013. Presented are the weights of models including each variable, divided by the sum of the AIC weights of all averaged models, with a highest possible score of 1.00. Variable Importance value No. models containing variable Initial shot impact 1.00 22 No. shots fi red 1.00 22 Year 1.00 22 Killing shot required 0.85 16 Bullet weight 0.72 12 Velocity rank 0.27 9 Shot distance 0.26 9 Brand acronym 0.10 10 Non-lead ammunition for culling elk • McCann et al. 275 missing data (n = 29), records with inconsistent bullet weight was of lower importance (Table information (n = 5), and instances where 4; Appendix B). chamberings or projectile types were represented Range evaluation of ammunition <5 times (n = 28), we retained 921 cases for analysis. Among these, mean shot distance was Muzzle velocity was successfully recorded 182 m, and the median and mode of distance for 78 of 80 shots taken during two 5-shot traveled were 46 m and 0 m, respectively. courses of fi re with 8 rifl es. Mean velocity was Volunteers were the primary shooter for 502 851 m per second, and mean group size was cases, whereas park staff shot 419. 50 mm, with minimum and maximum group Exploratory analyses of fi eld data revealed sizes varying 2-fold from the mean (Table 6). covariation among multiple continuous and Between courses of fi re, mean group size did categorical variables, suggesting considerable not vary (t = 0.13, P = 0.90), though muzzle 14 potential for interaction among predictors (results velocity was signifi cantly diff erent (t = -3.01, 76 not presented). Eight of the 13 independent P = 0.004), with mean velocity 5 m per second variables were signifi cant individual predictors faster for the second course of fi re. However, (Table 2). Multivariate ordinal regression of mean muzzle velocity was not a signifi cant these 8 variables revealed that “initial shot predictor of 5-shot group size (R2 = 0.1801, F 1,14 impact,” “killing shot required,” “number of = 3.08, P = 0.10). shots,” and “year” were signifi cant predictors of the effi ciency of kills in terms of distanced Discussion traveled (Table 3). Model building with best Program factors contributing to effi cient culling of elk subsets supported the inclusion of these four variables and “bullet weight” in the most The successful removal of 983 elk at TRNP parsimonious regression model (Table 3). Model with minimal wounding loss suggests that the averaging analyses confi rmed the importance tools and techniques employed were highly of all 5 aforementioned variables in predicting eff ective. Our structured approach to fi eld distance traveled (Table 4; Appendix A). operations, with a team leader directing the Parameter estimates of the regression actions of reduction team members, generally indicated that diff erences in levels of eff ect resulted in effi cient kills. Program policies, such for initial shot impact between marginal shots as range qualifi cation of rifl es and ammunition, in the legs and abdomen, shots striking the targeting of the shoulder and chest, stalking to shoulder and chest, and those shots impacting within 183 m, and directing the fi re of multiple the neck are driving the signifi cance of this marksmen on individual elk maximized variable for predicting distance traveled recovery of animals. Individual eff orts of (Appendix B). Descriptive statistics for initial staff and volunteers to coordinate actions and shot impact generally supported a hierarchy facilitate scenarios leading to success should not of eff ectiveness for shot placement, with short be overlooked when considering the effi cacy of distances resulting from head and neck shots, non-lead bullets. intermediate distances following chest and Regardless of bullet design, shot placement shoulder wounds, and the longest distances is an important factor for eff ective destruction resulting from abdomen and leg impacts (Table of animals (Ruth and Simmons 1999, Caudell et 5). Animals requiring a killing shot traveled al. 2013). Intentional targeting of the shoulder farther than those that did not, and number of and chest by marksmen often resulted in near shots was positively correlated with distance instantaneous incapacitation of elk during our traveled (r = 0.27, P < 0.05). The distance program, similar to that observed by Ruth and traveled by elk after being shot was diff erent Simmons (1999) for whitetail deer (Odocoileus for year 1 and all other years (Table 5; Appendix virginianus). However, our analyses indicate B). Finally, the weight of bullets was negatively that shots to the neck are more likely to result in correlated with distance traveled (r = -0.07, instantaneous incapacitation of animals. Ruth P < 0.05), suggesting that heavier bullets may be and Simmons (1999) also observed that shots more eff ective at incapacitating elk. However, to the neck immediately immobilized animals, when compared to other signifi cant factors, but the neck of whitetail deer was considered 276 Human–Wildlife Interactions 10(2) a problematic target due to its small size. In range at which marksmen can deliver accurate contrast, the relatively large size of an elk’s neck shots depends on individual skill level and may present a reasonable target for precision knowledge of ballistics (DeMuth 1966, Vaughn kills at close or intermediate ranges. Notably, 2000, Litz 2011). Signifi cant covariance among at TRNP, 75% of shots to the neck resulted in shot distance and number of shots fi red (r = instantaneous incapacitation, versus 40% for 0.18, P < 0.05) and between shot distance and shots to the chest and 46% for shots striking initial shot impact (F = 2.83, P = 0.02) indicates 1,5 the shoulder. This disparity in bullet strike that shot distance has impinged directly on outcomes is likely due to the concentration of the variance of other important factors in the vascular and central nervous system organs model. Thus, we infer indirectly that closer in the neck (Caudell et al. 2012). Knott et al. shots will be more eff ective at dispatching elk. (2009) reported that sika deer (Cervus nippon) The relationship for number of shots fi red shot in the chest with non-lead ammunition and the need for a killing shot with longer had to be tracked and concluded that head distances traveled by elk is easily understood. shots were most practical for that study site. At Wounded animals were fi red upon until they TRNP, the few recorded shots to the head of elk were incapacitated or out of sight and were were mostly unintentional, but also generally the most likely to require a killing shot. The resulted in instantaneous incapacitation. signifi cance of year on distance traveled may be However, the small size of the crania made explained in part by heavy snowfall during the the head an impractical target for dynamic fi rst year of the project. During October 2010 to fi eld conditions at our site, where animals January 2011, snow drifted and accumulated were typically engaged at 183 m and required to heights of >1 m in some locations at TRNP, shooter precision at that distance was a 200-mm possibly causing diffi culties for elk to traverse patt ern for 60% of shots. Further, shots to the the landscape. Additionally, large herds of head with non-lead ammunition often result elk existed in the park at the beginning of in bullet pass-through, which could result in the project, and they had not previously injury to other animals or ricochets (Caudell et experienced hunting pressure within the park al. 2012). Thus, we infer that the shoulder is the boundary. Therefore, elk were naïve to removal best aiming point for future operations at the activities and faced environmental diffi culties park because it provides the greatest margin of when att empting to escape during the fi rst error. Shots striking the shoulder will typically year, resulting in short distances traveled. result in quick kills, whereas shots fl ying wide Conversely, distance traveled by elk during will impact the heart and lungs caudally or subsequent years may have been increased the neck cranially. In all cases instantaneous by conditioning of radio-collared animals to or near-instantaneous incapacitation may be removal operations (Bender et al. 1999). expected. Though our analysis has provided useful Mass, velocity, and bullet confi guration insights, several factors were not addressed have been identifi ed as critical factors in the in our investigation due to the physical wounding potential of projectiles (Bellamy and challenges and complexity of operations Zajtchuk 1999, MacPherson 2005, Caudell et required to accomplish management objectives. al. 2013). Further, the importance of matching Specifi cally, (1) we did not att empt to evaluate bullet type with animal type and shooting wound channel pathology, (2) we did not scenarios is well understood (DeMuth 1966, consistently record the number of shots Caudell et al. 2009, Litz 2011). Therefore, it striking elk or the impact locations for bullets makes sense that bullet weight was included subsequent to the initial shot, and (3) we did in our top regression model. Though an ideal not record precise measurements or spatial bullet weight or confi guration cannot be gleaned variables (i.e., direction, slope, elevation from our analysis, a general recommendation is gradients) pertaining to movement of elk after to use heavier bullets for elk whenever practical being shot. Evaluation of wound channels in terms of availability and accuracy in the rifl es and the number and order of bullet impacts utilized. Additionally, accuracy and velocity would allow for determination of the series are inversely related to distance, and the of vital organs aff ected and identifi cation of Non-lead ammunition for culling elk • McCann et al. 277 Table 5. Descriptive statistics for response variable "distance traveled" (m) across signifi cant categori- cal predictor variables in the most parsimonious regression model obtained from elk reduction data collected during 2010 to 2013. Grouping Level n Mean SD Median Mode Mode Min Max variable frequency Initial shot impact Abdomen 71 79.2 80.6 46 91 25 0 366 Chest 332 36.2 38.8 46 46 167 0 274 Head 5 9.1 20.4 0 0 4 0 46 Leg 18 147.3 138.9 91 46/91/366a 4 0 366 Neck 67 15.7 32.4 0 0 50 0 183 Shoulder 428 36.1 51.5 46 0 195 0 366 Year One 371 27.5 52.3 0 0 228 0 366 Two 438 50.1 58.7 46 46 224 0 366 Three 67 38.2 37.0 46 46 30 0 183 Four 45 47.7 56.8 46 46 25 0 274 Killing shot required Yes 186 62.2 85.1 46 0 68 0 366 No 735 34.4 43.8 46 0 324 0 366 a Denotes that multiple modes were observed. central nervous system strikes that would help In most cases, rifl es and non-lead ammunition explain ultimate outcomes in terms of levels used on our project yielded comparable of incapacitation (Caudell et al. 2012, Caudell levels of precision. In fact, when our worst 2013, Trinogga et al. 2013). Field necropsies performing rifl e was excluded, mean minimum might also provide information regarding group size was 35.8 mm and mean maximum bullet and bone fragmentation, which may be group size was 51.1 mm at 91 m, falling well an important factor in killing power (DeMuth within the range of precision described for 1966, Fackler et al. 1984, Cruz-Martinez et al. most sporting arms fi ring lead-based projectiles 2015; Figure 3). Spatial information, such as GPS (Vaughn 2000). Range evaluations of non-lead track logs retracing the course traveled by elk, ammunition performed elsewhere generally would provide more accurate and informative support this fi nding (C. Batha and P. Lehman, metrics as response variables for multivariate Wisconsin Department of Natural Resources, analyses. Future eff orts to understand the unpublished data). The Barnes Ballistics Lab effi cacy of bullets in context of fi eld operations reported mean 5-shot group size of 12.7 mm should be designed to bett er measure these and 17.8 mm at 91 m for their TTSX bullets fi red variables. through rifl es chambered in 7mm Winchester Short Magnum and .300 Weatherby Magnum, Utility of non-lead ammunition for respectively (Barnes Bullets Inc., unpublished precise shot placement data). Therefore, the accuracy potential for In a meta-analysis of accuracy tests for non-lead ammunition is clearly suffi cient for sporting arms, Vaughn (2000) reported that the dispatching large ungulates in typical hunting typical maximum group size for 5 shot strings of scenarios. fi re for hunting cartridges was roughly 50.8 mm Variable accuracy of rifl es may be explained and the typical mean group size was 38.1 mm. by a number of factors, including diff erences