Human Activities and Ecosystem

Ungulate Mortality due to Fencing and Perceptions of Pasture Fences in Part of the Future Qilianshan National Park

  • Sydney M. GREENFIELD , 1 ,
  • Aliana C. NORRIS 1 ,
  • Joseph P. LAMBERT 1 ,
  • Wu liji 2 ,
  • Se yongjun 2 ,
  • ZHAN Jinqi 1 ,
  • MA Bing 1 ,
  • LI Deng 1 ,
  • SHI Kun , 1, 3, * ,
  • Philip RIORDAN 1, 4
  • 1. Wildlife Institute, School of Ecology and Nature Conservation, Beijing Forestry University, Beijing 100083, China
  • 2. Administration of Yanchiwan National Nature Reserve, Jiuquan, Gansu 736399, China
  • 3. Eco-Bridge Continental, Beijing 100085, China
  • 4. Marwell Wildlife, Thompsons Lane, Winchester, Hampshire, SO21 1JH, United Kingdom
*SHI Kun, E-mail:

Sydney M. GREENFIELD, E-mail:

Received date: 2020-05-14

  Accepted date: 2020-09-01

  Online published: 2021-03-30

Supported by

National Natural Science Foundation of China(31470567)


Fencing is an important part of husbandry for pastoral communities; however, these same fences can have unintended consequences for wildlife populations by restricting movement, reducing connectivity, and causing direct mortality. This paper assesses the current status and effects of fencing present in Yanchiwan National Nature Reserve, soon to be part of the recently proposed Qilianshan National Park. A questionnaire survey was conducted among 70 households to gauge local herders’ perceptions of fences, threats of fencing to native ungulates, and the number of wildlife found entangled in fencing. We found that local communities rely on fencing for livestock management and individuals who had encountered wildlife entangled in fences were more likely to perceive fences as having negative effects. Furthermore, those who perceived fencing as harmful to wildlife were more likely to support the dismantling of fences. On the other hand, families who needed to hire others to tend to their livestock were less likely to support dismantling efforts. However, the best model was only able to account for some of the data variability, suggesting that while perceptions of fences are important, other factors could be influencing support for fence dismantling. Hence, increasing awareness of threats alone may not be enough to generate community support of a fence dismantling program. Therefore, outreach and community collaboration to reduce the impacts of fence alterations upon livestock management will be necessary for a successful fence dismantling program within the new national park. Finally, those surveyed reported finding kiang, argali, and Tibetan gazelle dead in fences, with kiang found more often than the other two. This suggests that these three species may be more vulnerable to fence entanglement and that they are good targets for future studies and dismantling efforts.

Cite this article

Sydney M. GREENFIELD , Aliana C. NORRIS , Joseph P. LAMBERT , Wu liji , Se yongjun , ZHAN Jinqi , MA Bing , LI Deng , SHI Kun , Philip RIORDAN . Ungulate Mortality due to Fencing and Perceptions of Pasture Fences in Part of the Future Qilianshan National Park[J]. Journal of Resources and Ecology, 2021 , 12(1) : 99 -109 . DOI: 10.5814/j.issn.1674-764x.2021.01.010

1 Introduction

Fencing used alongside human development such as roads, highways, and railways, and to enclose pastures act as significant barriers for numerous ungulate species (Epps et al., 2005; Harrington and Conover, 2006; Olsson and Widen, 2008; Kaczensky et al., 2011b). The impact of fencing varies depending on the species, the fencing type, and the animals’ perceptions of and previous experiences with fences (Harrington and Conover, 2006; Riley et al., 2006; Rey et al., 2012). These barriers limit connectivity and fragment popu- lations, which may reduce their genetic diversity over time (Epps et al., 2005; Riley et al., 2006). The loss of genetic diversity in isolated populations leads to reduced re- production, survival, and fitness (Hogg et al., 2006). Fur- thermore, without large connected populations, over time these species may not have the genetic diversity needed to adapt to novel threats, such as introduced diseases (Willi et al., 2006).
Fences also pose an immediate threat to ungulates by limiting their access to important resources, such as water or seasonal pastures, and decreasing their ability to respond to environmental stressors. Dynamic changes in an environment may cause small areas to become critical to a species, as has been seen with kiang (wild ass, Equus kiang) and alpine sedge meadows (Schaller, 2000; Bhatnagar et al., 2006). Additionally, fences not only prevent animals from reaching better areas, but can also limit their ability to escape from poor ones. When species are constrained to small areas, they cannot escape from detrimental environmental changes, such as extreme cold weather or drought, which can decimate restricted populations (Kaczensky et al., 2011a). Therefore, by restricting the movement of wildlife, fences can limit their ability to survive.
For many animals, fences are not a complete barrier, but a dangerous obstacle through which they must maneuver. Animals often attempt to cross fences with varying outcomes depending upon the species, the animals’ age, the type of fencing used, and the season (Harrington and Conover, 2006; Hanophy,2009; Rey et al., 2012; You et al., 2013). Mortality is highest among animals that choose to cross over fencing, as they are often entangled and either injured or killed by the fences. Tall fences or those made of woven-wire panels topped with single-strand barbed wire are especially dangerous (Harrington and Conover, 2006). Fences with wire components are also problematic for large antlered species, such as deer and elk, as their antlers can become entangled (Hanophy, 2009; Paige, 2012).
In western China, fencing and enclosures have increased on rangelands since the introduction of the Rangeland Contract Responsibility System in the mid-1980s. This system transferred the ownership of livestock and grazing land from the state to individual households and collectives, although in practice most land is contracted to individual households (Cao et al., 2013a; Cao et al., 2013b). Currently, nature reserves and parks are being developed in China to protect wildlife and reconnect their habitats (Ministry of the Environment, 2017). Some studies have suggested that conservation and rangeland development policies, which are still being implemented in reserves, are in direct conflict and will make the conservation of wild ungulates more difficult (Fox et al., 2009). A large protected area called Qilianshan National Park has been proposed recently, which will encompass and connect Yanchiwan National Nature Reserve (YNNR) and Qilianshan National Nature Reserve (QNNR), as well as eight other nearby protected areas in Gansu and Qinghai provinces, spanning an estimated area of 50200 km2 (China Forestry Network, 2019). One of the goals of the national park is to reconnect wildlife habitats and populations. The proposed regulations will ban fences inside the core protected zone (estimated to be 27500 km2 or 55% of the national park), while help and guidance are suggested for assisting herders with the removal or renovation of pasture fences in the “general control” zone (22700 km2) (China Forestry Network, 2019). Altogether, the proposal calls for the removal of over 7000 km of fencing. However, understanding the needs of residents will be imperative for reducing the negative impacts of fence removal on the local people, as they will bear the greatest burden of new protections and policies, but may not necessarily benefit the most from the proposed park (Barua et al., 2013). Furthermore, it is important to collaborate with local communities in conservation. When conservationists and governments impose restrictions and limit access to key resources, seemingly without regard to local needs, resentment, resistance, and sabotage can undermine the conservation efforts (Holmes, 2007).
As observed elsewhere (Schaller, 2012), we suspect that in YNNR several ungulate species will be particularly vulnerable to fencing entanglement: Tibetan gazelle, or ‘Goa’, (Procapra picticaudata), kiang, argali (Ovis ammon), and white-lipped deer (Cervus albirostris). Though these species are relatively less studied, similar species have been documented to become entangled in fences, including Przewalski’s gazelle (Procapra przewalskii) (Cao et al., 2013a) and elk (Cervus canadensis) (Harrington and Conover, 2006). Furthermore, certain animal behaviors around fences are known to increase the likelihood of entanglement, such as frequent crossing attempts or a preference for jumping fences (Harrington and Conover, 2006). In YNNR, Tibetan gazelle are often observed ramming against fencing near roads in an attempt to flee approaching vehicles. Kiang behavior also suggests they are more vulnerable to fencing as they are often observed jumping the fences to escape perceived threats or to access grazing areas. Finally, we suspect that argali also have an increased risk, as they migrate seasonally across fenced pastureland from one end of YNNR to the other (Wuliji et al., unpublished data).
Studies on fencing mortality are limited and difficult to conduct (Jakes et al., 2018). Interviews with local people can provide useful initial data in situations where a rapid assessment of the situation is needed, or where more de- tailed data are difficult to collect (Meijaard et al., 2011; Starr et al., 2011; Ziembicki et al., 2013). In this study, we used interviews to make an initial assessment of both local residents’ perceptions of fences within YNNR and the rela- tive frequency of direct mortality for different species of wildlife. Specifically, we aimed to:
$\bullet$ Assess resident herders’ perceptions of fencing, attitudes towards dismantling existing fences, and which factors contributed to those beliefs.
$\bullet$ Determine which species have the most reported mortalities in the fences.
$\bullet$ Determine the timing and regional distribution of reported fence mortalities.
$\bullet$ Give a list of recommendations for future research in this field.
By combining data on local herders’ attitudes with information on fence mortalities, we hope to provide useful information that can be used to inform management and future research decisions in Qilianshan National Park regarding the planned removal of fencing, and to specify ways in which more detailed information should be collected.

2 Methods

2.1 Study area

Yanchiwan National Nature Reserve (YNNR) is located in a mountainous region on the northeastern edge of the Qinghai-Tibetan Plateau in Gansu, China (38°33°-39°10°N and 95°19°-97°13°E; Fig. 1). Encompassing an area of about 13600 km2, the reserve is characterized by a subarctic alpine climate, and ranges in altitude from 2600 m to 5500 m with an average of 4800 m (Liu et al., 2010). The reserve is crossed by three mountain ranges, the Danghe South Mountains, the Yema South Mountains, and the Shule South Mountains in the north. South of the reserve, these mountains connect to the Qilian Mountains which cross QNNR.
Fig. 1 Location of Yanchiwan National Nature Reserve
YNNR provides protected habitat for seven ungulate species, including three national level Class I protected spe- cies (the highest level of protection in China): kiang, wild yak (Bos mutus), and white-lipped deer (Liu et al., 2010; China Forestry Network, 2018). The other four ungulate species are all national level Class II protected: argali, blue sheep, Tibetan gazelle (goa), and goitered gazelle (Gazella subgutturosa). Other species potentially affected by fencing found within YNNR include brown bears (Ursus arctos), snow leopards (Panthera uncia), wolves (Canis lupis), dhole (Cuon alpinus), and Eurasian lynx (Lynx lynx). Of these, the dhole is listed as Endangered by the International Union for the Conservation of Nature (IUCN) Red List, and four species (snow leopards, wild yak, white-lipped deer, and goitered gazelles), are listed as Vulnerable (IUCN, 2019).
According to local park officials, approximately 944 people comprising 343 families reside within YNNR. YNNR is part of the Subei Mongolian Autonomous region and encompasses two townships, Shibaocheng and Yanchiwan. Two villages comprising 129 households are found in Shibaocheng Township, and five villages with 214 households are located in Yanchiwan Township. While the autonomous region is more diverse, according to local Chinese National Forestry and Grassland Administration (NFGA) officials, residents of YNNR are almost entirely ethnically Mongolian and have an average annual income of 5432 yuan (767.33 USD) (Liu et al., 2010). Within YNNR, the primary source of income is transhumant pastoralism and residents typically own a variety of livestock including sheep, cashmere goats, yak, horses, camels, cows, and donkeys. Residents maintain one or more small homes for seasonal grazing, as well as occasionally using yurts (gers) to follow their livestock in the summer. Seasonal grazing areas have been designed and assigned to each village to reduce overgrazing, however, at the moment these assignments may be more of just a guideline and they may not be strictly enforced. These seasonal areas are a mixture of unfenced and fenced regions of various shapes and sizes. Currently, YNNR is not open to tourism and therefore development and vehicular traffic is limited.

2.2 Fence type used within the reserve

According to local NFGA officials, pasture fences utilized throughout the reserve are made of woven wire panels with a single strand of barbed wire on top. The standard fence height used in YNNR and surrounding regions is 1.5 m, and the distance between all wires, including the barbed wire and the woven panel, is 0.2 m. During our surveys in the reserve and other unrelated studies in the grasslands, we found no evidence refuting this claim. According to YNNR officials, most fences were erected before the creation of the reserve in 1982.

2.3 Questionnaires

Seventy households in Yanchiwan Township (representing 33% of the township) were surveyed in 2017. While no one declined to answer the questionnaire, one was subsequently excluded due to concerns about response consistency and accuracy. A quota system based upon the population of the five villages was created to ensure a representative sample (Daniel, 2011). Numbers for the population at the time were provided by the local NFGA. Interviews were conducted in Chinese, with onsite translation to Mongolian where neces- sary. Respondents were asked which species they had found in the past that were dead or entangled in fences, how many times entangled wildlife were seen, in which habitat types, and during what time period. For the latter, respondents were asked to specify the month (when possible), or if they could not remember, the season. When asked about habitat types, participants listed broad categories such as grassland, mountains, or desert. We did not verify the respondents’ accuracy in assigning habitat types through maps or by visi- ting the locations. Respondents were then presented with open-ended questions asking them to describe the benefits and drawbacks they experienced from the use of fences, their perceptions of the impact of fencing upon local wild- life, and finally if they supported the dismantling of fences. The specific questions from the survey can be found in Ap- pendix 1.
Appendix 1 Questions asked during interviews about fencing, their perceived effect, and support for dismantling fences within YNNR
No. of
Chinese English translation
1 有没有见过动物被缠在或者绞死在围栏上? Have you ever seen wildlife dead or injured in fences?
2 牧场围栏对您的生活工作是否有好处?解决了您哪些问题? Are ranch fences good for your life and work? What problems did you solve?
3 牧场围栏是否对您的生活工作有负面影响?具体在哪些方面? Does the pasture fence have a negative impact on your life and work? In what ways?
4 您认为牧场围栏是否会对野生动物造成影响?正面还是负面的? Do you think the pasture fence will affect wildlife? Positive or negative?
5 您是否建议为保护野生动物拆除牧场围栏? Do you recommend removing fences from pastures to protect wildlife?

2.4 Analyses

All analyses were done in R (R Development Core Team, 2019). We used binomial logistic regression to explore po- tential relationships between support for fence dismantling and perceptions of the effects of fencing on wildlife and husbandry factors. Generalized Additive Models (GAMs), using the gam function in the mgcv package (Wood, 2019), were used as they can fit either linear effects, similar to a Generalized Linear Model, or non-linear effects, as needed (Zuur et al., 2007). The Un-Biased Risk Estimator (UBRE) method was used to choose the smoothing for each model, and Aikaike Information Criterion with small sample size correction (AICc) was used to determine the best model, where variables remained in the model if overall they yielded an AIC value of at least 2 less than a simpler model (Zuur et al., 2007). The analysis for dismantling support and perceptions of fences was limited to 63 households, exclud- ing the six households with uncertain or non-responses.
In addition to perceived effects of fences upon wildlife and the number of wildlife encountered in fences, we also included numbers of different types of livestock owned by the respondents, some measures of livestock management, and some socioeconomic variables. As many of these variables are correlated, such as the number of livestock and income, to avoid collinearity, not all variables were combined into one model. The summary of the variables and models tested can be found in Table 1 and Appendix 2, respectively.
Table 1 Predictor variables used to model support for dismantling fences and perceptions of the effects of fencing upon wildlife
Type of variable Variable name Short name Description
Number of small stock small stock Number of sheep and goats owned
Household variables Number of yak yak Number of yak owned
Number of total livestock total livestock Number of yak, cows, horses, camels, donkeys, sheep, and goats owned
Income income Annual income reported
Age age Age of the respondent
Sheep in shed sheep shed The proportion of the year the household had a shed or walled pen to hold their sheep and goats, where each season with a pen was equal to 0.25, with a max value of 1 for the whole year
Hires others hires others The proportion of the year a household hired others to look after their livestock, where each season hired is equal to 0.25, with a max value of 1 for the whole year
Wildlife variables
Positive or negative impact of fences on wildlife pos. neg. Ranges from -1 to 1, with -1 being negative, 0 being no effect, and 1 being a positive effect
Number wildlife observed in fences number observed Number of wildlife reportedly seen entangled in fences. Ranged from 0 to 2
Appendix 2 Formulas for GAM models for support for dismantling fences in YNNR
Model Formula AICc ΔAICc R2 (adjusted) Deviance explained (%)
Null Dismantle fence ~1 50.02 0.00 0.00 0.00
1 Dismantle fence ~ number observed + hires others 50.42 -0.40 0.03 8.2
2 Dismantle fence ~ number observed + sheep shed 48.63 1.39 0.08 17.4
3 Dismantle fence ~ number observed + small stock 59.23 -9.21 0.08 16.5
4 Dismantle fence ~ number observed + yak 51.93 -1.91 0.002 5.07
5 Dismantle fence ~ number observed + total livestock 52.15 -2.13 0.01 4.62
6 Dismantle fence ~ number observed + age 51.55 -1.53 0.02 5.87
7 Dismantle fence ~ number observed + income 51.57 -1.55 0.02 5.62
8 Dismantle fence ~ pos. neg. + hires others 45.67 4.35 0.18 22.4
9 Dismantle fence ~ pos. neg. + sheep shed 47.17 2.85 0.16 23.9
10 Dismantle fence ~ pos. neg. + small stock 48.03 1.99 0.15 22.4
11 Dismantle fence ~ pos. neg. + yak 49.38 0.64 0.13 14.5
12 Dismantle fence ~ pos. neg. + total livestock 49.46 0.56 0.16 27.2
13 Dismantle fence ~ pos. neg. + age 48.47 1.55 0.12 16.4
14 Dismantle fence ~ pos. neg. + income 49.30 0.72 0.10 14.7
15 Dismantle fence ~ number observed 50.02 0.00 0.02 4.7
16 Dismantle fence ~ pos. neg. 47.50 2.52 0.12 13.7
17 Dismantle fence ~ hires others 50.91 -0.89 0.0002 2.6
18 Dismantle fence ~ sheep shed 48.75 1.27 0.03 11.5
19 Dismantle fence ~ small stock 49.78 0.24 0.04 10.0
20 Dismantle fence ~ yak 51.59 -1.57 -0.01 1.2
21 Dismantle fence ~ total stock 52.16 -2.14 -0.02 0.003
22 Dismantle fence ~ age 51.60 -1.58 -0.01 1.2
23 Dismantle fence ~ income 51.93 -1.91 -0.01 0.5

3 Results

3.1 Perceptions of fences

When asked what benefits they received from fencing, all participants could list at least one benefit, with most offer- ing several. The benefits that participants reported receiving from fences were organized into nine different categories. Out of 69 respondents, 50 (72%) mentioned better small stock management, 44 (64%) said clearer grazing bounda- ries, 25 (36%) said overall better livestock management, 21 (30%) said better large stock management, 12 (17%) said excluding wildlife from pastures, 10 (14%) said reducing disputes, 9 (13%) said excluding other families, 7(10%) said protecting the grassland, and 2 (3%) said livestock pro- tection (Fig. 2). When asked about the negative impacts, 68 responded, with 59 (86%) saying fences had no negative effects and 9 (13%) saying fences had some negative effects.
The reported negative effects were grouped into five categories. Four people (6%) said fences were an obstacle or required detours, 3 people (4%) mentioned that they required maintenance, 2 people (3%) said fences can kill livestock, 1 person (1%) mentioned that fences can trap wildlife, and 1 person (1%) said that fencing can lead to overgrazing.
Fig. 2 Benefits of pasture fences as reported by resident herders
When asked about the impact of fencing on wildlife, 67 people responded, with 41 respondents (61%) saying fences had no effect on wildlife, 21 (31%) said fences had a negative effect, 4 (6%) said they did not know, and 1 (1%) said fences had a positive effect on wildlife. As only a single person reported a positive effect of fences upon wildlife, we chose to run a simple binomial logistic GAM, excluding the positive value for this analysis. The GAM models revealed that the respondents who reported finding ungulates in fencing were more likely to report fences as having a negative effect upon wildlife, with an R2 of 0.16 and explaining 13.6% of the deviance (Fig. 3). Some participants offered examples of negative effects, which were grouped into five categories. Eleven respondents (17%) said the fences inhibited animal movements or migration, 8 (12%) said wildlife had adapted to the fences, 6 (9%) felt the fences limited wildlife access to resources such as food or water, and 4 (6%) said the effects were minimal as they had no or few wildlife in their area. Finally, when asked if they recommended that the fencing be dismantled to protect wildlife, of the 67 who responded, 55 (82%) did not recommend dismantling the fences, 8 (12%) did recommend it, and 4 (6%) were unsure.
Fig. 3 Response curve for the perceived effects of fencing on wildlife against the number of wildlife found entangled in fences

Note: The perceived effect of fencing on wildlife is expressed as binary, where 1 is no effect and 0 is a negative effect. The solid line shows the response to the independent variable and the shaded region shows the 95% confidence interval.

3.2 Modeling support for dismantling the fences

The best GAM model retained the variables for perceptions of the effects of fences upon wildlife and whether or not a household hired others to help them. Households which perceived fences as having a negative or positive effect upon wildlife were more likely to support the dismantling of fences than those which felt fences had a neutral effect. Households which reported hiring others to look after livestock were less likely to support the dismantling of the fences, in which the greater the proportion of the year that others were hired, the less likely a household was to recommend dismantling the fences (Fig. 4). The best model explained 23.4% of the deviance and had an adjusted R2 of 0.18. A summary of the statistics for the top models is available in Table 2, and a list of all models tested and their summaries can be found in Appendix 2.
Fig. 4 Response curves for willingness to dismantle the fences to the two variables retained in the best model

Note: Willingness to dismantle is binary, where 1 indicates support and 0 indicates did not support dismantling the fences. (a) Effect of fences on wildlife is on a scale of -1 to 1, where -1 is a negative effect, 0 is no effect, and 1 is a positive effect of fences upon wildlife. (b) Hires others is the portion of the year a herder hired someone to look after their livestock. In both (a) and (b) the solid line shows the response of the dependent variable to one independent variable when the response to the other is held constant at its median, and the shaded region shows the 95% confidence interval.

Table 2 The three best GAM models and the null model for comparison
Number Formula AICc ΔAICc R2 (adjusted) Deviance explained (%)
Null Dis. fence ~1 50.02 0.00 0 0
8 Dis. Fence ~ pos. neg. + hires others 45.67 4.35 0.18 22.4
9 Dis. Fence ~ pos. neg. + sheep shed 47.17 2.85 0.16 23.9
16 Dis. Fence ~ pos. neg. 47.50 2.52 0.12 13.7

3.3 Species killed by fences and the timing and regional distribution of fence mortalities

Of the 69 households, 20 (29%) reported finding a total of 26 ungulates and a single passerine entangled and dead in the fences; none of the wildlife found entangled were still alive. Of the ungulates found, 15 were kiang, six were argali, four were Tibetan gazelle, and one was an unknown wild sheep species.

3.4 Timing and regional distribution of fence mortalities

Of the 26 ungulates found, 4 were found in the mountains, 6 in the desert, 15 in the grasslands, and 1 was not known. When asked about the timing of fence entanglements, respondents reported that one animal was found in the spring, one in the summer, three in the fall, and ten in the winter. The respondents could not remember the exact season in which the other 11 were found (Fig. 6).
Fig. 5 The total numbers of each animal species found dead in the fences by interviewees within Yanchiwan National Nature Reserve
Fig. 6 The number of each species reportedly found by residents in YNNR by season

4 Discussion

4.1 Local perceptions of the effects of fencing and attitudes towards dismantling the fences

Our results indicate that the local people benefit from fences in terms of husbandry and management and may be seriously impacted by the indiscriminate removal of the fences. Residents reported many benefits they received from fencing, with most focusing on improved livestock management and clarified boundaries. Only 13% felt that the fences had any drawbacks, focusing primarily on inconvenient detours. While about a third of residents felt the fences had a negative effect upon wildlife, nearly two-thirds believed that the fences did not affect wildlife at all, as wildlife had adapted to the presence of the fences. A GAM analysis indicated that finding wildlife entangled in fences did increase the perception that fences had a negative effect upon wildlife. During interviews, the most noted effects on wildlife were limitations to animal movements and migration as well as restrictions on access to resources such as food and water.
Despite a third of households believing that the fences harmed wildlife, only 13% of people recommended dismantling the fences to help wildlife. While the best GAM model indicated that those who perceived negative effects of the fences upon wildlife were more likely to support fence dismantling, the best model only accounted for some of the variability in support of fence dismantling and it explained less than a quarter of the deviance. Hiring others to care for livestock during the year was also an important variable, and proved to be better than any of the other tested measures of livestock numbers or husbandry tested. Hiring others to care for livestock is not likely a direct cause of support, but may be an indicator that the households’ capacity to manage their livestock and/or husbandry methods are driving dependency on fences, and thereby influencing the support for dismantling them. While the model also indicated that those who perceived positive effects on wildlife were more likely to support dismantling the fences, this is based on a single household, and further research is needed to draw accurate conclusions.
To increase the support for dismantling the fences and to have a successful dismantling and adaptation program, our results suggest officials should carefully collaborate with residents to reduce the impacts upon their husbandry before taking action. Education will be necessary to help residents understand the mortality risks, long term effects of habitat fragmentation upon local wildlife, and the need to dismantle and adapt the fences for the wildlife within the protected areas. However, while the perceptions of the impacts of fences appear to have a degree of influence on support for dismantling, since they are not completely driving dismantling recommendations, outreach alone may not be enough to change the perceptions. Thus, active collaboration for win-win scenarios and decreasing the impact upon herders will be essential for gaining local support.

4.2 Species with greatest mortalities in the fences

Of the seven ungulates present in YNNR, the argali, the kiang, and the Tibetan gazelle were found entangled and dead in fences. Of the three species, the kiang was found more often than either Tibetan gazelle or argali. Though only 26 animals were reported, it is likely that many more animals are killed, but were not found due to scavenging animals (Harrington and Conover, 2006). This is especially true in the seasonally rested grazing areas in which fences remain standing even when unused by herders. Since we currently lack up-to-date estimates of ungulate populations or their distributions in YNNR, we cannot determine if one species is more vulnerable than another or if the fences are having a significant impact upon the wildlife populations. However, this rapid assessment does raise concerns and indicates that the vulnerability of kiang, argali, and Tibetan gazelle to fence entanglement needs to be studied further.
While more research is necessary, the behaviors of these species around the fences may be putting them at a greater risk. Kiang and Tibetan gazelle are nervous around vehicles and frequently attempt to cross fences to reach open grasslands. For Kiang, it should be noted that this sort of nervousness around humans and their vehicles has also been reported in Asiatic wild ass (Equus humionus) in Mongolia. Kaczensky et al. (2010) found that in 11 observations, Asiatic wild asses were easily disturbed by humans and their activities, fleeing from vital waterholes when humans approached 100% of the time. By contrast, kiang ignored wolves one out of the two times they approached, and were nearly indifferent to other ungulates.
Finally, while argali spend most of their time in the mountains and are less likely to be disturbed by people, they may be crossing fences to access resources and escape harsh environments. It is likely that argali seasonally migrate across YNNR and are believed to be intolerant to deep snow (Wingard et al., 2011), as has been seen in bighorn sheep (Ovis canadensis) (Jalkotzy, 2000). All of the reported argali entanglements occurred in the winter and only one argali was found in mountain pasture fences. This indicates the argali may be descending to lower altitudes to access better forage, forcing them to cross fences and increasing their risk of entanglement. Future research should focus on estimating and monitoring the populations of these species, as well as tracking their movements and behaviors, to help predict the impacts of fencing in more detail.
Though we did not assess the placement of fences in this study, we noticed that fences lining access roads and the main road have inadvertently created corridors that funnel animals onto the main road of the park, but do not allow animals on the main road to escape for some distance. It is possible that strategically placing access roads and crossing points for wildlife could alleviate some of this risk. When planning the removal, alteration, or construction of fencing, careful attention should be given to fences in proximity to existing roads, water, and other areas frequented by ungulates. This will be especially important as the planned tourism within the new Qilianshan National Park will lead to an increase in vehicles, which could result in even greater risk.

4.3 Determining the timing and regional distribution of fence mortalities

The data collected for this study were insufficient to determine if one season had a significantly greater number of fence mortalities. However, there is a definite increase in the mortalities found in winter for kiang and argali. Like argali, for kiang this could be explained by foraging behavior. In other regions of the Qinghai-Tibetan Plateau, herders and kiang compete for vital winter forage in alpine sedge meadows, and the kiang are perceived as serious competi- tion by the herders (Bhatnagar et al., 2006). It is possible that before the creation of the nature reserve and implementation of the protections for wildlife, valuable pastureland was fenced off to exclude the wildlife. In the winter, ungu- lates may be at the greatest risk as they are forced to cross dangerous fences to access necessary resources, however, further research is needed to confirm if this is true in YNNR. Fencing mortality needs to be prevented, particularly when it occurs within a nature reserve. One of the key goals of the Qilianshan National Park is to remove and alter the fences. Our results suggest future studies should focus on the impacts upon kiang, argali, and Tibetan gazelle. However, a lack of fencing mortalities for other species should not be interpreted as meaning that fencing has no impact upon them. Rather, the fencing present in YNNR could be acting as an impermeable barrier for other species, and future studies are also needed to examine the impact of fences on migration and gene flow within and outside of the protected area.

4.4 Type and extent of fence placement within the nature reserve

The fencing found within YNNR is identified in the literature and guides as one of the most dangerous types of fencing for wildlife (Harrington and Conover, 2006; Hanophy, 2009; Paige, 2012). Though data is limited, the barbed wire on these fences frequently snags animals jumping the fence, but when they attempt to kick free, their legs become tangled in the stiffer woven panels (Fig. 7). Where wildlife are concerned, it is generally recommended to avoid woven- wire panel fencing with strung wire, and when wire must be used, to avoid barbed wire on the top and bottom wires to reduce the risk of injury (Harrington and Conover, 2006; Hanophy, 2009; Paige, 2012). The dimensions of the fencing reportedly used in YNNR are also of concern. At 1.5 m, this is well above the maximum range (0.85 m to 1.07 m) recommended for fencing by wildlife guides and the literature (Harrington and Conover, 2006; Hanophy, 2009; Paige, 2012; Rey et al., 2012). Previous studies have found that more than 70% of mortalities occurred on fences over 1 m in height, and they were especially lethal for juvenile animals (Harrington and Conover, 2006). According to local officials, this is the standard fencing type utilized in the region within and outside YNNR, and is similar to what has been reported in nearby regions (You et al., 2013); therefore, it is also potentially used in other parts of the proposed Qilianshan National Park. Further work is needed to confirm the dimensions and materials used throughout these protected areas, and efforts should be made to remove or alter the fences where necessary. However, since fence effects are species-specific, research will also be needed to determine if alterations are effective in reducing the risk to wildlife and increasing connectivity for local wildlife populations.
Fig. 7 An adult kiang entangled in pasture fencing in YNNR Note: Photo credit to Buyin SAI.
There are a few key areas we suggest the Qilianshan National Park fence dismantling projects should focus on. Firstly, fences in rested grazing areas should be opened to allow the free passage of wildlife. Increased monitoring of opened rangelands will be necessary to monitor for degradation and to assuage residents’ fears that wildlife may overgraze the rested pastures. Secondly, access roads already in place between pastures should be moved to align with one another and provide animals with quick passage to open habitat. Thirdly, the fences, particularly those with frequent wildlife crossings, should be changed into more wildlife-friendly designs, such as those without wires or other designs as suggested in wildlife-friendly fencing guides, and kept to heights of 1.07 m or less (Harrington and Conover, 2006; Hanophy, 2009; Paige, 2012). As these recommendations are based upon North American species, any changes should be tested before and after implementation to ensure that they are effective for the species found within this region.
Finally, in order to achieve accurate estimations of the risk to wildlife caused by the fences, further studies are needed to estimate wildlife populations, determine their seasonal movements and resource demands, and document their behavior around fences. Accurate maps of fencing, data on the height and materials of different sections, and the locations of frequent wildlife crossings and entanglements will also be necessary.

5 Conclusions

This study found that the herders of this part of YNNR, soon to be incorporated into QNP, believe fences are a benefit to their livestock husbandry and most do not support the removal of the fences. Local herders reported encountering three species entangled in fencing, kiang, argali, and Tibetan Gazelle. Those who reported finding wildlife caught in fences were more likely to believe fences were having a negative impact on wildlife, and those who believed fences were having a negative impact on wildlife were more likely to recommend pasture fence removal. This indicates that education and outreach may increase support, if local people are made more aware of the potential negative impacts on wildlife. However, those who reported needing to hire others to look after their livestock at some point in the year were less likely to support dismantling efforts. This may indicate that a herder’s ability to manage their own stock is affecting support. Therefore, the removal of fences within YNNR is likely to be met with resistance by the residents of YNNR and the proposed national park may affect their ability to manage their livestock. However, previous studies suggest this is an important goal which is worth pursuing to protect the wildlife of YNNR and other nearby protected areas. Government efforts to remove and adapt pasture fences within the proposed park should be done in collaboration with local communities and focus on areas of frequent animal sightings and migration routes, particularly for kiang, argali, and Tibetan gazelle.

Ethical Statement

All interviews were conducted with adult participants only and with full consent, expressed in the subject’s native language (Chinese or Mongolian). No personally attributable details or information were requested, collected, stored or distributed. The methodology was assessed and approved by ethical review committees at each of the collaborating institutions or organizations.
Barua M, Bhagwat S A, Jadhav S. 2013. The hidden dimensions of human-wildlife conflict: Health impacts, opportunity and transaction costs. Biological Conservation, 157:309-316.


Bhatnagar Y V, Wangchuk R, Prins H H T, et al. 2006. Perceived conflicts between pastoralism and conservation of the kiang Equus kiang in the Ladakh trans-Himalaya, India. Environmental Management, 38(6):934-941.


Cao J, Yeh E T, Holden N M, et al. 2013a. The roles of overgrazing, climate change and policy as drivers of degradation of China’s grasslands. Nomadic Peoples, 17(2):82-101.


Cao J, Yeh E T, Holden N M, et al. 2013b. The effects of enclosures and land-use contracts on rangeland degradation on the Qinghai-Tibetan Plateau. Journal of Arid Environments, 97:3-8.


China Forestry Network. 2018. List of national protected wildlife. Viewed 3 Apr 2019.(in Chinese)

China Forestry Network. 2019. Qilianshan National Park master plan. Viewed 25 May 2020.(in Chinese)

Daniel J. 2011. Choosing the type of nonprobability sampling. In: Sampling essentials: Practical guidelines for making sampling choices. London, UK: SAGE Publications, 103-105.

Epps C W, Palsbøll P J, Wehausen J D, et al. 2005. Highways block gene flow and cause a rapid decline in genetic diversity of desert bighorn sheep. Ecology Letters, 8(10):1029-1038.


Fox J L, Dhondup K, Dorji T. 2009. Tibetan antelope Pantholops hodgsonii conservation and new rangeland management policies in the western Chang Tang Nature Reserve, Tibet: Is fencing creating an impasse? Oryx, 43(2):183-190.


Hanophy W. 2009. Fencing with wildlife in mind. Colorado Division of Wildlife.

Harrington J L, Conover M R. 2006. Characteristics of ungulate behavior and mortality associated with wire fences. Wildlife Society Bulletin, 34(5):1295-1305.


Hogg J T, Forbes S H, Steele B M, et al. 2006. Genetic rescue of an insular population of large mammals. Proceedings of the Royal Society B: Biological Sciences, 273(1593):1491-1499.


Holmes G. 2007. Protection, politics and protest: Understanding resistance to conservation. Conservation and Society, 5(2):184-201.

IUCN (International Union for Conservation of Nature). 2019. The IUCN red list of threatened species. Viewed 29 Jan 2020.

Jakes A F, Jones P F, Paige L C, et al. 2018. A fence runs through it: A call for greater attention to the influence of fences on wildlife and ecosystems. Biological Conservation, 227:310-318.


Jalkotzy M. 2000. Habitat selection by bighorn ewes on three winter ranges in the east Kootenays. Report prepared for Columbia Basin fish and wildlife compensation program. Arc Wildlife Services Ltd. and East Kootenay Wildlife Association.

Kaczensky P, Dresley V, Vetter D, et al. 2010. Water use of Asiatic wild asses in the Mongolian Gobi. Exploration into the Biological Resources of Mongolia (Halle/Saale, Germany), 11:291-298.

Kaczensky P, Ganbataar O, Altansukh N, et al. 2011a. The danger of having all your eggs in one basket—Winter crash of the re-introduced Przewalski’s Horses in the Mongolian Gobi. Plos One, 6(12):e28057. DOI: 10.1371/journal.pone.0028057.


Kaczensky P, Kuehn R, Lhagvasuren B, et al. 2011b. Connectivity of the Asiatic wild ass population in the Mongolian Gobi. Biological Conservation, 144(2):920-929.


Liu N F, Zhang H C, Dou Z G. 2010. Synthetical scientific investigation on Yanchiwan National Reserve in Gansu Province. Lanzhou, China: Lanzhou University Press. (in Chinese)

McDonald J H. 2014. Multiple tests. In: Handbook of Biological Statistics (3rd ed.). Balitmore, Maryland: Sparky House Publishing, 254-260.

Meijaard E, Mengersen K, Buchori D, et al. 2011. Why don’t we ask? A complementary method for assessing the status of Great Apes. Plos One, 6(3):e18008. DOI: 10.1371/journal.pone.0018008.


Ministry of the Environment. 2017. Report on the state of the ecology and environment in China 2017. Beijing. Retrieved from

Olsson M P O, Widen P. 2008. Effects of highway fencing and wildlife crossings on moose Alces alces movements and space use in southwestern Sweden. Wildlife Biology, 14(1):111-118.


Paige C. 2012. A landowner’s guide to wildlife friendly fences (2nd ed.). Private Land Technical Assistance Program. Helena, USA: Montana Fish, Wildlife & Parks.

R Development Core Team. 2019. R: A language and environment for statistical computing. R Foundation for Statistical Computing.

Rey A, Novaro A J, Guichón M L. 2012. Guanaco (Lama guanicoe) mortality by entanglement in wire fences. Journal for Nature Conservation, 20(5):280-283.


Riley S P D, Pollinger J P, Sauvajot R M, et al. 2006. A southern California freeway is a physical and social barrier to gene flow in carnivores. Molecular Ecology, 15(7):1733-1741.


Schaller G B. 2000. Kiang (Tibetan Wild Ass). In: Wildlife of the Tibetan steppe. Chicago, USA: University of Chicago Press, 165-167.

Schaller G B. 2012. Tibet wild: A naturalist’s journeys on the roof of the world. Washington DC, USA: Island Press, 154-155.

Starr C, Nekaris K A I, Streicher U, et al. 2011. Field surveys of the Vulnerable pygmy slow loris Nycticebus pygmaeus using local knowledge in Mondulkiri Province, Cambodia. Oryx, 45(1):135-142.


Willi Y, Van Buskirk J, Hoffmann A A. 2006. Limits to the adaptive potential of small populations. Annual Review of Ecology, Evolution, and Systematics, 37(1):433-458.


Wingard G J, Harris R B, Pletscher D H, et al. 2011. Argali food habits and dietary overlap with domestic livestock in Ikh Nart Nature Reserve, Mongolia. Journal of Arid Environments, 75(2):138-145.


Wood S. 2019. MGCV: Mixed GAM computation vehicle with GCV/AIC/ REML smoothness estimation.

You Z, Jiang Z, Li C, et al. 2013. Impacts of grassland fence on the behavior and habitat area of the critically endangered Przewalski’s gazelle around the Qinghai Lake. Chinese Science Bulletin, 58(18):2262-2268.


Ziembicki M R, Woinarski J C Z, Mackey B. 2013. Evaluating the status of species using indigenous knowledge: Novel evidence for major native mammal declines in northern Australia. Biological Conservation, 157:78-92.


Zuur A F, Ieno E N, Smith G M. 2007. Analysing ecological data. New York, USA: Springer.