Patterns of livestock depredation and Human–wildlife conflict in Misgar valley of Hunza, Pakistan

Throughout the world, livestock predation by mammalian carnivores causes significant economic losses to poor farmers, and leads to human–wildlife conflicts. These conflicts result in a negative attitude towards carnivore conservation and often trigger retaliatory killing. In northern Pakistan, we investigated livestock depredation by large carnivores between 2014 and 2019, and subsequent Human–wildlife conflict, through questionnaire-based surveys (n = 100 households). We used a semi-structured questionnaire to collect data on livestock population, depredation patterns, predation count, and conservation approaches. We found a statistically significant increasing pattern of predation with influential factors such as age, gender, occupation, education of respondents, population of predators, threats index for predators and conservation efforts. Some 310 livestock heads with an average of 51 animals per year out of the total 9273 heads were killed by predators, and among them 168 (54%) were attributed to the wolf and 142 (45.8%) to snow leopard. Major threats to carnivores in the area included retaliatory killing, habitat destruction and climate change. Incentivization against depredation losses, guarded grazing and construction of predator-proof corral may reduce Human–wildlife conflict and both livelihood and predator can be safeguarded in the study area.

www.nature.com/scientificreports/ Though there are records of frequent predation cases in the remote areas of Karakoram but still there is lack of appropriate documented data about the scale and scope of Human-wildlife conflict and its consequences. Being one of the buffer zone valley of Khunjerab National Park and a potential wildlife corridor between Pakistan, China and Tajikistan, Misgar hosts a variety of predators which make the valley vulnerable to livestock depredation. The average household income of communities residing in Misgar is less than US$ 500 per year and in a such a scenario, loss of one or two livestock heads per family means a lot for the affected herding family and that in most of the cases ends in retaliatory killing. Therefore, in the current study, we investigated the scale and scope of Human-wildlife conflict in Misgar valley so that we could be able to inform decision makers about site specific conservation strategies.

Material and methods
Study area. Misgar is in the northwestern region of district Hunza, Gilgit-Baltistan, Pakistan, adjoining to the Khunjerab National Park (KNP) towards the north-east. Misgar Valley lies at 36° 46′ 34′′ N and 74° 45′ 56′′ E, at 8000 ft. above sea level. The population of the valley is 3000 people and it shares a border with both China and Afganistan 20 . The vast meadows of the valley provide grazing grounds to a large herd of livestock and wild ungulates, and offer important habitat for predators such as wolves, snow leopards, and red fox.
Data collection. For data collection, questionnaires were designed for household level (HH) data and HH level interviews and focus group discussion (FGDs) were conducted as earlier used by various researchers 19,21 . Out of 270 households, 100 (37%) households from the predation hotspots like Kilike, Murkushi and Dardee areas of Misgar Valley (Fig. 1) were selected randomly for data collection on scale and scope of livestock predation between 2014 and 2019. The interviews were conducted following an informed consent of the respondents and overall the study plan was approved by the graduate research management committee of the Karakoram International University. In addition to this, all methods were performed in accordance with the relevant guidelines and regulations.
Data analysis. Data collected on different parameters of Human-wildlife conflict was analyzed in R software (R Development Core Team, 2021), and poisson regression 22,23 was run to evaluate effects of socio-ecological covariates on livestock predation: Here β ′ s is the regression coefficient and ǫ is the model residual. Models were run separately for the snow leopard and wolf predation counts using the poisson regression model. We considered more than 22 different factors that have direct or indirect impact on predation dynamics, which included demographics, seasonal information, government policies, ecosystem, education, occupation, economic situation etc. Snow leopard and wolf predation counts were separately modelled through poisson regression. Moreover, the stepwise model section algorithm helped in designing the parsimonious model, significantly identified the influential factors and showed the best statistical performance.
Results indicates age of respondent, occupation of respondent, gender of respondent, education of respondent, annual income of respondent, estimated income from livestock, reasons of selling live stoke ( Fig. 3A-G), protective measures, number of wolf, reasons of attack on livestock, wildlife threats ( Fig. 4A-H) and season, measure to save livestock, compensation for livestock lost, income earned as compensation, wolf and snow leopard status appeared influential factors in explaining the variation in snow leopard wolf a predation ( Fig. 5A-H). Reference to adult age of respondent, the wolf predation is 0.0795 times less likely for aged respondents (p-value 0.002) and the wolf predation is 3.5 times more likely for young respondent (p-value < 0.01). Reference to basic education, the wolf predation is 0.032 times less likely for respondents having higher education (p-value < 0.001) and the wolf predation is 0.011 times less likely for illiterates (p-value < 0.001). Reference to farmers the wolf predation is 0.056 times less likely for government employees (p-value < 0.001) and is 0.034 times less likely for private employee (p-value < 0.001). The wolf predation decreases for male respondents by 0.039 times (p-value < 0.001). The wolf predation decreases for high income respondents by 0.001 times (p-value < 0.001). Reference to guarded dog the wolf predation is 6.23 times more likely with guarded grazing and is 0.43 times less likely with roof corrals (p-value < 0.001). The wolf predation decreases with high number of wolfs by 0.017 times (p-value = 0.016). The wolf predation decreases with laws of wild life by 0.003 times (p-value < 0.001). Reference to laws for conservation the wolf predation decreases by 0.036 times with government rules (p-value < 0.001). Reference to exposed to predators the wolf predation increases by 2.34 times with favorite food (p-value < 0.001). Reference to autumn the wolf predation decreases by 0.006 times in winter (p-value < 0.001). Reference to the middle income the wolf log(Predation Counts) = β 0 + β i Demeographic factor + β j seasonal factor + β k government policiy factor + β k ecosystem factor + β k economic factor + ǫ. www.nature.com/scientificreports/ predation decreases by 0.04 times with the high income (p-value < 0.004). Reference to absent wolf status the wolf predation decreases by 0.193 times with common wolf status (p-value < 0.001) ( Tables 3, 4). About 54% of the respondents claimed that illegal hunting is the major threat followed by illegal hunting and habitat destruction with 32% and climate change is threatening about 14% of the predator's population (Tables 3, 4, Fig. 4H).
When respondents were asked about the snow leopard status in their area, 89% of the respondents were of the view that it is common while 9% thought that it is rare and 2% of the respondents claimed it to be absent in their area. Similarly, when respondents were asked about the status of grey wolf, 85% said that it common, 13% claimed it as a rare species while 2% claimed it absent in their area (Fig. 5G,H).   Perceptions towards carnivores. About 63% of respondents perceived snow leopard to be the most dangerous animal while 37% declared same for the grey wolf (Fig. 4B). Concerning reasons behind huge depredation losses in the valley, majority people (53%) attributed to loss of natural prey, some (24%) believed livestock is more exposed to the predators and easier to capture, while rest (23%) of the respondents claimed livestock to be the preferred prey (Fig. 4G). Despite ongoing human-predator conflicts in the area, about half (49%) respondents considered predators to be no important species of the area (Fig. 4A).

Detection and seasonal patterns of predators.
When asked about seasonality of the predation cases about 22% of respondents claimed that predation occurs in winter, 21% of the respondents claimed summer season as a threat, 17% said more predation cases occur in autumn season, 15% claimed spring season as a threat, 13% of the respondents are of the view that most of the predation cases occur in both spring and season and 12% of the respondents said that majority of the predation cases are recorded in summer, autumn and spring seasons (Fig. 5A).
Challenges for conservation. About 69% of respondents perceived that there are low threats to snow leopard while 5% are of them opinion that the threats faced by the snow leopard are of medium nature and 26% of the respondents declared the threats faced by the snow leopard to be of severe nature. When asked about the nature of threats faced by wolf, 69% of the respondents are of the view that the threats faced by wolf are not severe while13% perceived the threats to be of medium nature and 30% respondents are of the view that the threats faced by the wolf are of severe nature (Fig. 4E,F). In response to the effectiveness of grazing patterns about 49% of the respondents perceived predators proof corrals as the topmost conservation strategy while 32% considered guard dogs as a beneficial conservation strategy and 19% perceived guarded grazing as a conservation strategy (Tables 3,4, Fig. 5B). These protection measures are important because, majority shed (80%) in pasture are unprotected (Fig. 5C). Measure suggested by respondents to maintain wildlife population in their area included developing relevant rules and regulations (50% respondents), engagement of local community in conservation programs (37%), and enforcement existing wildlife laws, rules, and regulations (13%) (Fig. 5D). Majority herders (80%) did not any compensation for livestock losses, few (15%) which did, they received partial compensation (PKR 1000-6000). Only 5% respondents received substantial compensation (Fig. 5E,F).

Discussion
Patterns of livestock predation. In highly natural resources dependent areas around the globe, when a predator kills livestock, the herding communities retaliate 8,24 . Same is the case with Misgarities who rely mostly on livestock as one of the major livelihood options. The present study indicates that livestock numbers have increased between 2104 and 2019 with a record percent increase of 37.64% in the year 2019 (Table 2) possibly due decreasing number of livestock in other areas bordering Misgar like the Khunjerab Villagers, which are in agreement with the Parks and Wildlife Department of Gilgit-Baltistan that requires them to decrease the livestock number in order to qualify for conservation incentives in the form of tourist entry fee and income from sustainable trophy hunting programme. Thus, there is an increasing demand for meat from the adjacent communities and Misgarities are taking it as an opportunity. The current study also shows an increasing livestock depredation rates with 3.34% higher depredation rate as compared to other cases reported around the world 25,26 .   www.nature.com/scientificreports/ www.nature.com/scientificreports/ carnivores in this area. However, this risk could be mitigated through sound conservation strategies that include compensation for the livestock losses.

Perception towards carnivores.
In the Karakoram-Pamir mountains, predation of livestock is one of the major issues with varying intensity. The spatial variations can be attributed to predators and prey abundance, herd size and herding practices and predator type. Snow leopard, wolf, and lynx are thought to be dangerous predators, while the brown bear is reported as less fatal to livestock in the region 8 . Similar perceptions were noticed during the current study. People regard snow leopard and wolf as dangerous animals, lynx was not recorded in the area, while brown bear was considered less fatal. Wild carnivores are known to prey selectively upon different livestock species to optimize their foraging strategy. The foraging behaviors accounts for the size of the predator and the size of their prey, prey preference, and abundance 29 . Prey preference observed during the current study could be explained under framework of optimal foraging. For example, we found that most of the sheep were preyed by wolf, while goats, cows and yaks were hunted by snow leopards. Being larger and stringer animal, snow leopards are probably going after larger preys. Din et al., reported average sighting of snow leopard (1.6 + 0.15) and wolf (3.9 ± 0.32) in the Pamir region, which overlaps with our study area. In concurrence with that, majority of respondents (82%) during the current study claimed 1-5 encounters with snow leopards, and remaining 18% had observed snow leopards more than 5 times. Similarly, 79% respondents sighted grey wolf 20 to 30 times, and 21% respondents sighted more than 30 times in their life (Fig. 4C,D). Higher encounter for wolves was possibly due to their wider distribution and group living behavior. Seasonality in predation is common phenomenon that widely documented for large carnivores 7,21,23 . For example, Li et al. reported predation in two seasons only; autumn (37.2%) and summer (36%). In Misgar, Autumn was a tie of lowest predation, while highest predations were recorded in winter, followed by spring and summer.

Conservation implications.
Human-wildlife conflict is one of the emerging challenge towards conservation of predators, specifically in underdeveloped countries 30 . Other interlinked threats include, persecution by herders, human population growth, reduction in wild prey-base and illegal hunting 31 . The present study documents that wildlife in Misgar is facing similar threats in greater intensity. Tested approaches towards mitigation of Human-wildlife conflict include improved animal husbandry, predator-proof corals, guarded grazing, habitat restoration, and promotion of co-existence with predators 21 . Community-based livestock insurance schemes along with awareness program can promote co-existence 32 . We believe that the aforementioned mitigation measures are appropriate for the Misgar valley, these should be initiated by the Government of Gilgit-Baltistan without further delay, in order to safeguard future of unique biodiversity of this pristine habitat. In current study, we have noted that the responded who were compensated for the losses have relatively positive attitude towards the predators and therefore they were not involved in retaliatory killings. Extreme climate events leading to disasters, soil erosion, drought, and ecosystem imbalance are also threatening the existence of wildife 33 in the area, thus need attention of policy makers.

Conclusion
The current study indicates that both the livelihoods of herder communities and population of the carnivores are at the stake due to Human-wildlife conflict and other associated threats in Misgar. Therefore, a robust Human-wildlife conflict management program at the state level, needs to be realized that addresses issues related to depredation losses through mitigation measures and provision of safety nets and promotes acceptance of predation through environmental education in the area.