Twenty infrared cameras were installed in the study area (Figure 1). The cameras were placed 30–90 cm above the ground. The height can not only capture the entire saplings but also obtain a full shot of the ungulates browsing them. Camera parameter settings were as follows: photo (three photos) + video (15s). After the camera was installed, the weeds in front of the shooting area were removed to prevent misshooting and blocking the lens (Zhao et al., 2020). The monitoring period lasted 2.5 years.
FIGURE 1 Location of the Muling National Nature Reserve in northeast China
2.2.3 Environment variable data
According to research needs, environmental variables are divided into four categories:
(1) Land-use types: Scrub, evergreen coniferous forest, deciduous coniferous forest, mixed coniferous forest, deciduous broad-leaved forest, and farmland, were obtained using Landsat 8 remote sensing image interpretation.
(2) Anthropogenic disturbance factors: Rural roads and forest trails were obtained by vectorization in ArcGIS 10.8 software using forest topographic maps.
(3) Topographic factors: Elevation, aspect, and slope direction were obtained by mask extraction of a 30m resolution digital elevation model (DEM) using the spatial analysis module of the ArcGIS 10.8 software.
(4) Water sources: Streams were obtained by vectorization in the ArcGIS 10.8 software using forested topographic maps.
2.2.4 Data analysis
2.2.4.1 Foraging rhythms of the wapiti and Siberian roe deer
Taking the average time of sunrise and sunset of the local year as the demarcation, the day was divided into day and night, with reference to the local climate. May to October was classified as the warm season, and November to April was classified as the cool season. The daily and annual foraging rhythms were analyzed in this study by using the kernel density estimation method (Chen et al., 2019) with the overlap package and activity package of R. Delta4 values were selected for calculations when both the sample sizes compared were ≥75, and the Delta1 values were selected when the size was <75. In addition, the degree of temporal overlap between the activities of the two species was classified as high overlap when Delta >0.75, medium overlap when Delta<0.50<Delta<0.75, and low overlap when Delta<0.50 (Monterroso et al., 2014).
The foraging patterns of the wapiti and Siberian roe deer on Japanese yew saplings were analyzed by calculating the relative abundance index for the day and night (D RAI), month (M RAI), and season (S RAI) (Liu et al., 2022). The formulas were as follows:D RAI=(D ij/N i)100,M RAI=(M ij/N i)100, andS RAI=(S ij/N i)100.D ij denotes the number of independent foraging occurrences of species i in the time period j (both day and night),M ij denotes species i in the time period j (annual), and S ij denotes species i in the time period j (cold and warm seasons).
2.2.4.2 Foraging habitats suitability prediction
The systematic resampling method was used to analyze the points to avoid spatial autocorrelation caused by too close a distance between them, to standardize the data of the environmental variables, and to conduct Spearman correlation analysis with SPSS 19.0. The selected points and environmental variables were used to predict suitable foraging habitats; 75% of the points were used for the construction of the model, 25% were used for validation, and the model was run for 10 cycles. This was evaluated using the cutting method and its comprehensive contribution. The evaluation criteria used were as follows: AUC range from 0.5 to 0.6 was considered as failing, from 0.6 to 0.7 was poor, from 0.7 to 0.8 was fair, from 0.8 to 0.9 was good, and from 0.9 to 1.0 was excellent. All operations were performed using MaxEnt 3.3.
The suitable habitats for the Japanese yew and ungulates were reclassified using ArcGIS 10.8. The average value of the maximum training sensitivity and specificity after 10 modes of operation was used as the threshold for the distribution of suitable foraging habitats (Li, 2022). The habitats of Japanese yew were divided into two levels—0–0.153 were unsuitable habitats, and 0.153–1 were suitable. The foraging habitats of the ungulates were divided into three levels—0–0.55 were unsuitable foraging areas, 0.55–0.75 were generally suitable and 0.75–1 were suitable.
2.2.4.3 Foraging characteristics of the two ungulates and their effects on saplings
Taking Japanese yew as the center, the study area was extended in eight directions—east, west, south, north, northeast, southeast, northwest, and southwest. The sapling points were recorded along with the information on whether foraging occurred or not within 50m in each direction. A total of 71 points were collected and plotted using the fmsb package in R. The characteristics of ungulates foraging on saplings were summarized using infrared camera data. The t-test and Kolmogorov-Smirnov test were used to analyze the following: (1) whether the monthly relative abundance index of the two ungulates foraging for young trees was different; (2) whether there was a difference in the distance between the two ungulates foraging on the young tree and the mother tree; and (3) whether the growth height of the main branches, the number of new branches, and the growth amount of lateral branches after foraging by the two ungulates were different. All operations were performed using R 4.2.1.
RESULTS3.1 Foraging time strategies of the two ungulatesThe diurnal foraging activity index of the wapiti was higher than that of the Siberian roe deer (D RAI=92.3%), and the nocturnal feeding activity index of the Siberian roe deer was also higher. There were two peaks in the diurnal activity rhythms of the wapiti foraging on saplings, which were from 7:00 to 8:00 and from 14:00 to 15:30, with the overall shape of an ”M,” and the size of the peaks varied. There were three peaks for the roe deer, from 6:30 to 9:30, 15:30 to 18:30, and 24:00 to 2:00, with the entire peak being wavy. In terms of the degree of overlap, the highest overlap indices were observed at 7:00 and 15:00, and temporally the degree of foraging was moderate (Dhat1=0.67) (Figure 2).