Effects of reflective warning markers on wildlife

Study area and camera trapping

Cangshan Mountain, located in Dali Prefecture of Yunnan Province, China (Fig. 1A), is in the Cangshan Mountain and Erhai Lake Nature Reserve. Previous surveys have shown Cangshan Mountain is rich in native flora and fauna (Shen, 1998; Zhao et al., 2017). The study site is located between Mocan and Heilong streams (25°36′−25°40′N, 100°06′−100°11′E, altitude range 2,550–2,800 m, Fig. 1A). Vegetation at the site has a ground cover of 40%–80% and is dominated by 15–20 m tall Armand pine (Pinus armandii). Human activities are strictly controlled in this nature reserve and almost no artificial light enters the study area.

To avoid impacts caused by environmental heterogeneity, five sub-regions were selected on a single slope of a stream divide. From April 9, 2017 to July 23, 2017, 30 camera traps were set up in study stations. Within each sub-region, three stations included RWMs (RWM group), and another three stations were without RWMs (control group) (Fig. 1A). The distances between the nearest two camera traps were 100–200 m. The RWMs were installed 4 m in front of the camera traps and 0.3 m aboveground (Fig. 1C). The size of the RWMs were 20 cm × 5 cm and were designed with 10 cm × 5 cm red colored strips and 10 cm × 5 cm white colored strips. Both strips reflect light but are not self-luminous (Fig. 1B).

Ltl Acorn® 6310MC cameras were used and set at medium sensitivity for motion, with two photos and a 10 s video captured in rapid succession upon each trigger. Camera installation and basic photographing methods followed specifications proposed by Xiao et al. (2014). All cameras made by Zhuhai Ltl Acorn Electronics Co., LTD in Zhuhai City of Guangdong province, China.

Data processing and statistical analysis

When the camera traps were collected, all memory cards and data were processed using software CTIMCS1.1 (a software for manually sorting camera trap photographs that was developed independently by the Institute of Eastern-Himalaya Biodiversity Research, Dali university, http://www.eastern-himalaya.com.cn/contents/3/990.html). Animals on the photos were identified to the species level with the time and date being noted. The identification of mammal species was accomplished using Smith et al. (2010) and bird species were identified using MacKinnon et al. (2000). Nocturnal mammals such as rodents and bats were excluded from the analysis because they could not be accurately identified on the photographs captured at night. O’Brien, Kinnaird & Wibisono (2003) used several independent photographs to estimate species abundance and activities. We used the same method to uniquely define each photograph in a series consecutive photos taken of the same or different species, or of the same individuals taken during 30 min. Nighttime was defined as lasting from 19:30 to 6:30, and daytime lasted from 6:30 to 19:30.

To assess if RWMs significantly changed the diversity and abundance of species, we used a general linear mixed model, with the number of species and independent photographs captured separately on each camera as the dependent variables (Bolker et al., 2009). We considered the sub-regions as random-effects, and the presence or absence of RWM as the fix-effects. We conducted a covariance analysis to test if the slope of the independent acumination line between the two groups was significantly different.

To estimate the species composition between the groups, we used the Jaccard index (Jaccard, 2010), which is a community similarity index calculated as q = c∕(a + b − c), where q is the Jaccard index, a and b are the number of species in each group, and c is the number of common species in both groups. Indices between: 0≦q≦0.25 were considered extremely dissimilar, 0.25 < q≦0.50 were moderately dissimilar, 0.50 < q≦0.75 were moderately similar, and 0.75 < q≦ 1 were extremely similar. All analyses were completed with Software R (R Core Team, 2018).

Number of species and independent photos

The camera traps captured 13 (RWM group) and 21 (control group) species. The linear mixed model analysis showed that the cameras with RWMs recorded significantly higher species number than those without (2.60 ± 1.99 vs. 4.93 ± 2.99, t =  − 2.70, p = 0.012, Appendix S2).

Cameras traps with RWMs captured 146 independent photographs, whereas the control group had 284 photographs (Fig. 2B). Although not significantly different, cameras with RWMs captured fewer independent photographs than did cameras without RWMs (10.00 ± 13.52 vs. 20.13 ± 21.30, t = 1.66, p = 0.130, Appendix S2). Stations with RWMs had a lower number of independent photos than did the control group in all five sub-sampling regions, indicating these regions had a lower animal density (Fig. 3). The slopes of the number of independent photographs by dates for the two groups were significantly different (df = 152, t = 20.60, p < 0.001).

Composition of community species

Among the 25 species of animals captured by camera traps, seven of the mammalian species and one avian species was recorded in both treatments (Fig. 4). The Jaccard index of species for both treatments was 32%, which was classified as moderately dissimilar. However, the bird species composition was extremely dissimilar between two groups (Jaccard = 7.6%), and the mammal species composition was moderately similar (Jaccard = 58.3%).

More species and independent photographs were captured in the daytime than in the nighttime for both groups (Fig. 5). Eleven bird species and four mammal species were only captured in the daytime, and one bird species and five mammal species were only captured at night. Two mammal species, the long-tailed goral (Naemorhedus caudatus) and leopard cats (Prionailurus bengalensis), were captured during daytime and nighttime in both groups (Fig. 6).

Among the species captured in the daytime, long-tailed thrushes (Zoothera dixoni) and moustached laughingthrushes (Garrulx cineraceus) only appeared in areas with RWMs. Ten other bird species only appeared in the areas without RWMs. One bird species, the Lady Amherst’s pheasant (Chrvsolophus amherstiae), and four mammal species were captured in both groups (Fig. 6).

Tawny owls (Strix aluco), Yunnan hares (Lepus comus), and Malayan porcupines (Hystrix brachyura) were only captured at night in the stations with RWMs, whereas masked palm civets (Paguma larvata) and red muntjacs (Muntiacus muntjak) only appeared in areas without RWMs.

Effects of RWMs on wildlife during nighttime

Although the number of species during nighttime differed slightly between the camera groups with and groups without RWMs, the number of independent photographs in the stations with RWMs was nearly 50% lower than the control group. The differences might be related to the different density or activities pattern of the species recorded in both groups. Of the eight species captured at night, tawny owls, Yunnan hares, and Malayan porcupines were only found in the areas with RWMs. Tawny owls are a nocturnal predatory species that preys mainly on small rodents. The short-eared owl (Asio flammeus) is another owl that has been found to have better predation efficiency at night. Consequently its prey, deer mice (Peromyscus maniculatus), have decreased their foraging behavior and other activities to avoid detection in highlighted environments (Navara & Nelson, 2007). Therefore, the materials might be attractive to the owls because they might use light reflected by RWMs during hunting.

The three species that only appeared in the areas without RWMs were primarily prey species in an ecosystem. Their predation avoidance abilities may keep them from appearing in places with RWMs that may assist predators. Leopard cats and wild boars appeared in both groups, perhaps because they are highly adaptable and live around human residential areas. The leopard cats hunt near villages and wild boars search for food in farmlands (Rajaratnam et al., 2007; Schley & Roper, 2003).

Effects of RWMs on wildlife in the daytime

During the daytime, there were considerable differences in the number of species and independent photographs captured between the two groups. The stations with RWMs had nearly 30% fewer species and 50% fewer independent photographs. The daytime differences in species composition were probably caused by the different community composition of birds. Seven babblers captured, of which six were in the control group and only one was found the in RWM group. White-bellied pheasants were the only species that showed no avoidance response in the RWM groups during daytime. Red colored RWMs can increase visual stimulation in a manner similar to the alert color of insects (Stevens, 2007; Iniesta, Ratton & Guerra, 2017). A bird-repelling device has invented and functions effectively by using a red material that reflects light in an intermittent pattern (Doty III, Turkewitz & Byers, 2003).