Time-to-event models for wild animals generally model exposure of individuals to natural conditions that may affect the risk of mortality and disappearance. Most models neglect to consider seasons of high human activity that may affect such risks, or interactions between endpoint hazards (reflected in incidences) that may illuminate ecology. For many large carnivores, which suffer from low natural mortality yet are also subject to high risk of anthropogenic mortality and poaching, seasons of anthropogenic activity may be as important as natural ones in mediating cause-specific mortality and disappearance.
Importantly, such anthropogenic seasons of higher mortality need not be specific to the animals being studied, especially if the species is controversial and much mortality illegal: our anthropogenic seasons consist of state hunting and hounding seasons for species other than wolves (i.e., deer or bear hunting, and hounding; not wolf hunting), but that mediate human activity on the landscape during those seasons. Our results support the hypothesis that increases in poaching risk during hunting seasons may be attributable to the surge of individuals with inclination to poach on the landscape14,18,29. Alternatively, it could also suggest enhanced criminal activity of a few poachers during the same periods. We temper this increase in poaching risk by establishing snow cover as a major environmental factor strongly associated with poaching. Moreover, our time-to-event analyses illuminate how to evaluate the effects that such anthropogenic seasons may have on risk of mortality and disappearance of monitored animals throughout their lifetime, and how considering such seasons may elucidate the mechanisms behind anthropogenic mortality and disappearance.
Additionally, our analysis period precedes and completely excludes any established public wolf hunting seasons. Hence, our modeled anthropogenic seasons represent the periods of most relevant anthropogenic activity for wolves, as hypothesized by other studies14,29,33 and suggested by social science studies on inclinations to poach self-reported by both deer hunters and bear hunters, as well as acceptance of poaching by hunters and farmers30,31,32.
Our analyses show increases in the hazard of disappearances of collared wolves (LTF) relative to the baseline period (which excludes environmental and anthropogenic risks) for all seasons. The highest hazard of LTF occurs during the snow season, whereas increases in hazard are lower (and similar) for the two seasons that included hounding and hunting. LTF may experience changes in hazard due to changes in the hazard of any/all of its components: migration, collar failure, or cryptic poaching.
Constant and steep increases in LTF hazard throughout a wolf’s lifetime suggests mechanisms other than migration regulating LTF hazard, given migration for adults is most frequent by yearlings and younger adults, around 1.5 to 2.2 years34,35,36. Moreover, only migration out of state would end monitoring, not routine extraterritorial movements of radio-collared wolves. That our seasonal LTF curves depict the cumulative hazards more than doubling beyond those t generally associated with dispersal (~ t < 500, given wolves were collared as adults), and that such hazards remain high throughout a wolf’s lifetime relative to other endpoints, suggests mechanisms behind LTF hazard that are additional to migration out of state. If migration had been the driving mechanism behind LTF hazard, we would also expect higher increases in hazard (more similar to the snow season) during other periods also associated with increased dispersal for adults, such as Oct–Nov36 within the hunt/hound and hunt/hound/snow seasons. Instead, during the latter seasons we observe smaller increases in LTF hazards, again suggesting mechanisms other than long-range movements out of state raising LTF hazard.
Although our study is unable to evaluate the contribution of collar failure to LTF hazard, we note that average and max time to LTF (t = 497, 2330 respectively) was similar to that of other anthropogenic endpoints (legal, t = 472, 2357; poached, t = 477, 2303) and much shorter than for other endpoints (collision, t = 590, 2235; natural, t = 655, 3051; unknown, t = 773, 2999) or censored observations (t = 882, 2833), which implicates causes other than battery or collar failure14.
As for the cryptic poaching component of the LTF hazard, the mechanism is consistent with the observed steep increase in hazard of LTF throughout a wolf’s lifetime and seasons (contrary to the natural hazard), and with the similarities in time to endpoint between LTF and other anthropogenic, intentional killing (i.e., legal killing and reported poached).
The lower increases in LTF hazard during the hunt/hound and hunt/hound/snow seasons relative to the snow season show different patterns to that of the reported poached endpoint. We hypothesize the much higher relative cumulative hazard of the LTF endpoint for all seasons except hunt/hound/snow (for which reported poached is highest) may suggests a rate of cryptic poaching that increases not only due to more cryptic poaching activity than baseline during periods of more anthropogenic activity (hunt/hound and hunt/hound/snow seasons), but also due to decreased detection of poaching on the landscape given environmental conditions during the snow season33. This reduced detection of cryptic poaching which increases LTF hazard during the snow season does not translate to the hunt/hound/snow season (despite similar environmental conditions) due to a surge of individuals on the landscape that result in not only more, but detectable poaching, therefore increasing the reported poached rather than the LTF hazard. This seems to resemble the pattern reported in Santiago-Ávila et al.18 of an increase in the hazard of reported poached relative to that of LTF during a census period in which dozens of civilian wolf-trackers went out in snow months to count wolves. Therefore, search effort and visibility due to landscape conditions are important variables to consider when designing anti-poaching interventions.
The hazard of reported poached more than doubles during the snow season relative to the baseline season, and doubles again during the hunt/hound/snow season, during which wolves are simultaneously exposed to environmental and anthropogenic conditions. The reported poached cumulative hazard during the hunt/hound/snow season is the highest of any across endpoint-seasons. These results implicate snow cover as a major factor mediating poaching activity (much lower hazard during snowless seasons), potentially by increasing wolf track detection. To those conditions, the hunt/hound/snow season may add more potential poachers or their increased killing, particularly during the (firearm) deer season, which more than doubles the snow season reported poached hazard. An important observation is that despite a decrease in incidence of LTF that season, in fact the LTF hazard increases, which points to this seasonal decrease in LTF incidence being an effect of the substantial increase in reported poaching hazard; i.e., the much higher rate of reported poached decreases LTF incidence despite an increased hazard of LTF. Therefore, we conclude that the reporting and documentation of poaching is improved when there are more people on the landscape, and worsened when there are fewer and snow cover is high.
For all anthropogenic and environmental seasons modelled, the natural endpoint shows an initial higher hazard but with a decrease in its seasonal hazard over time relative to baseline (i.e., non-proportional effects). The natural hazard is in general lowest during the hunt/hound season. For the hunt/hound/snow and snow seasons, the natural hazard is substantially lower than the LTF or reported poached endpoints. Moreover, the deceleration in the increase in natural hazards relative to the baseline period is suggestive of wolves learning to mitigate some seasonal natural hazards over their lifetime (e.g., intraspecific strife, starvation). We do not observe such a pattern with the LTF or reported poached endpoints, for which increases in hazard continue unabated over time. The difference in patterns between natural and anthropogenic endpoints suggests wolves may have difficulty and limited success in mitigating the hazard of anthropogenic killing, which is also by far the highest hazard overall. We also note that the natural hazard is lower than that for reported poached during the snow season, despite the marginally higher natural incidence, suggesting the latter could be an effect of the interaction of the natural hazard with lowered hazards from other, less prevalent endpoints (e.g., unknown, legal). The higher hazard of poaching (cryptic, through LTF, and reported) relative to other endpoints makes any possible interactions (compensatory or depensatory) among the other hazards (e.g., between natural death and legal killing) seem marginal and possibly influenced by (correlated to) fluctuations in the hazard of poaching. Hence, we caution researchers looking for compensatory or depensatory mechanisms to account for the role of poaching, including its cryptic component, first and foremost.
Our results also indicate different seasonal patterns of hazard for our natural and unknown endpoints, which suggests they should be analyzed separately (contra29). Failure to do so would inflate estimates of anthropogenic mortality and exaggerate the sustainability of lethal management programs that base predictions on estimates of human-caused mortality (e.g.37). Results for endpoints of lower prevalence, such as unknown, collisions, and (to a lesser extent) legal killing when implemented as in Wisconsin (by government agents removing suspected predators of livestock primarily), should be considered preliminary given their respective lower numbers of events per modeled covariate than those recommended to ensure accurate estimation38,39.
The increase in hazard of reported poached and LTF during the hunt/hound/snow season makes this season the deadliest for wolves throughout most of their adult lives (see Supplementary Material Fig. 3). The high hazards of LTF and reported poached, which are higher than all other endpoints for most seasons (hunt/hound, hunt/hound/snow and snow) and throughout t, also confirm poaching as by far the highest mortality hazard for collared adult wolves in Wisconsin throughout their lifetimes14,18.
Furthermore, given attitudes toward wolves became more negative among relevant demographics after wolf hunts were implemented in Wisconsin in 201232, the general hazard of poaching (cryptic and reported, for all seasons) may have increased relative to our study period (when wolf hunts were not legal) despite possibly resulting in a relatively lower incidence due to the magnitude of the increase in legal killing (e.g., Wisconsin February 2021 wolf hunt40). Moreover, the ‘facilitated poaching’ hypothesis suggests further increases in poaching after permitting wolf hunting, trapping, and hounding (2012–2014, 2021–) relative to only permitting selected legal killing (our study period)17,18,25. Such an effect of public wolf-hunts would hypothetically be mediated by a policy signal that further devalues wolves or suggests overabundance.
We are not aware of effective efforts by the WDNR to mitigate poaching hazard, neither through increased enforcement nor through public education initiatives. Rather, WDNR efforts have been focused on ‘tolerance hunting’ through reducing protections, despite multiple lines of evidence pointing to such actions not decreasing and potentially increasing total (cryptic and reported) poaching hazard14,18,25,31,32. In other jurisdictions, such ‘tolerance killing’ is viewed skeptically as a management tool both scientifically and legally13,41,42,43. Our results underscore the need for increased protections and anti-poaching interventions to improve the wellbeing of wolves and their populations, and to reduce illegal exploitation of the public trust.
Source: Ecology - nature.com
