Wild red wolf Canis rufus poaching risk

The reintroduced red wolf population in northeastern North Carolina declined to 7 known wolves by October 2020. Poaching (illegal killing) is the major component of verified anthropogenic mortality in this and many other carnivore populations, but it is still not well understood. Poaching is often underestimated, partly as a result of cryptic poaching, when poachers conceal evidence. Cryptic poaching inhibits our understanding of the causes and consequences of anthropogenic mortality which is important to conservation as it can inform us about future population patterns within changing political and human landscapes. We estimate risk for marked adult red wolves of 5 causes of death (COD: legal, nonhuman, unknown, vehicle and poached) and disappearance, describe variation in COD in relation to hunting season, and compare time to disappearance or death. We include unknown fates in our risk estimates. We found that anthropogenic COD accounted for 0.724 – 0.787, including cryptic and reported poaching estimated at 0.510 – 0.635 of 508 marked animals. Risk of poaching and disappearance was significantly higher during hunting season. Mean time from collaring until nonhuman COD averaged 376 days longer than time until reported poached and 642 days longer than time until disappearance. Our estimates of risk differed from prior published estimates, as expected by accounting for unknown fates explicitly. We quantify the effects on risk for three scenarios for disappearances, which span conservative to most likely COD. Implementing proven practices that prevent poaching or hasten successful reintroduction may reverse the decline to extinction in the wild of this critically endangered population. Our findings add to a growing literature on endangered species protections and enhancing the science used to measure poaching worldwide.

135 Problems such as these can be addressed with models that allow multiple sources of data to 136 inform estimates of variables, including unobservable variables like cryptic poaching [16,28,44].
137 For this study we define unknown fates as those animals that are radio-collared but become lost-138 to-contact and unmonitored. Most studies involving mortality data make assumptions that 139 unknown fates resemble known fates [16]. When wolves are legally killed, they are always 140 included in known fates and in calculated mortality risk. Treves et al. [16] tested the hypotheses 141 in populations where cryptic poaching occurs, unknown fates will not accurately estimate known 142 fates causing important losses of information producing systematic error. When they corrected 143 estimates of mortality risk of four endangered wolf populations by excluding legal killing from 144 unknown fates, their estimates of poaching risk were higher than government estimates, which 145 assumed known fates would be representative of unknown fates. For example, when estimates 146 for relative risk from other human causes included unknown fates, government reported risk for 147 red wolves was 0.26 -0.40 lower than corrected estimates [16]. By accounting for all marked 148 animals, m (unknown fates ) + n (known fates), and estimating cryptic poaching, we extract more 149 information for mortality risk than traditional methods of censoring those marked animals that 150 disappeared [16]. We will adapt the above methods to test the hypothesis that censoring or 151 ignoring marked wolves that disappear under-estimates poaching and loses essential information 152 to produce a systematic bias in conclusions about the NENC red wolves. 241 instead, we estimate risk more accurately by taking into account how many of m to reallocate 242 from the unknown COD and FU in Table 2 to our three classes of imperfectly reported CODs 243 (vehicle, poaching, nonhuman). The reallocation step is an estimation procedure with several 244 possible scenarios that apportion different amounts of m to cryptic poaching, vehicle, or 245 nonhuman, depending on assumptions that we explain next (Fig 2).   (Fig 2). Therefore, we had to recalculate the risk posed by the 30 cases of legal killing 269 (Table 1) with the denominator, n + m, to estimate the risk of legal killing for all collared red 270 wolves at 5.9% (Table 3). Without this correction, risk of legal killing in Table 1 would be 271 overestimated as 9.6%. When one over-estimates the risk of legal killing, one underestimates all 272 other imperfectly reported CODs because the total must sum to 100%.

278
The unknown fate collared wolves in m contain only imperfectly reported CODs but an 297 However, we assumed non-zero cryptic poaching in the NENC red wolf population because we 298 had prior information. At least 23 reported poaching incidents in n show evidence of attempted 299 and failed cryptic poaching (tampering or damage to the transmitter did not cause its failure, so 300 the USFWS recovered the collar or carcass even though the poacher tried to conceal it). We 301 categorized these 23 as failed cryptic poaching based on the following circumstances; only a 302 damaged collar was found, the dead wolf was found with a damaged collar, or the dead wolf was 303 discovered in a suspicious location (e.g., dumped in a canal, near a beagle that had also been shot 304 also found in that canal, and one was in the same location as another shot wolf). Damaged collars 305 included obvious human tampering such as bullet holes and knife cuts. These 23 deaths include 306 12 instances categorized by the USFWS as "suspected foul play" and 11 with suspicious 307 circumstances recorded in field notes [47]. Therefore, we inferred that a scenario with zero 308 cryptic poaching was so unlikely as to be discarded. We defined a range of plausible cryptic 309 poaching with the following scenario building.  Fig 5). There were no 403 significant differences in vehicle, nonhuman, or legal COD. 415 wolves between 20 -40 days, and only one at 9 days after collaring. Time to disappearance and 416 time to death from legal and vehicle were very similar (Fig 6). 570 Implementing proven practices that prevent poaching or hasten successful reintroduction can 571 reverse the trend of a decreasing NENC red wolf population and once again allow red wolves to 572 thrive, not only in NENC but additional future reintroduction sites.
573 Acknowledgments 574 We thank the USFWS for access to their red wolf database and numerous conversations 575 about the reintroduction program in NENC. We thank Dr. Joseph Hinton from the University of 576 Georgia who provided not only access to his field notes and data but also his guidance and 577 expertise regarding the red wolf program from a research perspective. Thank you to the 578 Carnivore Coexistence Lab at the University of Wisconsin Madison, especially Dr. Francisco J.
579 Santiago-Avila, for their statistical assistance and brainstorming sessions.