Effects of nitrogen addition and mowing on rodent damage in an Inner Mongolian steppe

Abstract Rodent damage is a serious threat to sustainable management of grassland. The effects of nitrogen (N) deposition and grassland management on rodent damage have been scarcely studied. Here, we reported the effects of 2 years of N addition and mowing on burrow density and damage area of Citellus dauricus in a semiarid steppe in Inner Mongolia. N addition significantly aggravated, while mowing alleviated rodent damage in the grassland under study. Burrow density and damage area increased 2.8‐fold and 4.7‐fold, in N addition plots compared to the ambient N addition treatment, respectively. Conversely, mowing decreased burrow density and damage area by 75.9% and 14.5%, respectively, compared to no mowing plots. Observed changes in rodent damage were mainly due to variations in plant community cover, height, and aboveground net primary productivity. Our findings demonstrate that N addition and mowing can affect the rodent density and activity in grassland, suggesting that the effects of a changing atmospheric composition and land use on rodent damage must be considered in order to achieve better grassland management.

tons of annual biomass loss caused by rodents (Zhang, Zhong, & Fan, 2003). According to the National Grassland Monitoring Report of China in 2016, the rodent-infested area occupies 28.07 × 10 6 ha, covering 7.1% of the total pasture area (Liu, 2017).
Previous studies have documented that rodent damage can be affected by factors influencing food sources and rodent predators.
Indeed, most previous studies of rodent damage in grasslands have focused on the effects of grazing, climate warming, and changes in annual precipitation regime. For instance, rodent survival rates are reportedly higher under heavy grazed management (Harris, 2010;Li et al., 2016;Zhang, Pech, et al., 2003;Zhou, Zhao, Tang, Gu, & Zhou, 2005). Climate warming has aggravated rodent damage through promoting clonal growth of Potentilla anserina in an alpine meadow (Li et al., 2011). Abnormal annual precipitation could inhibit rodent outbreaks in the grassland (Zhang, Pech, et al., 2003). However, responses of rodent damage to changes in other environmental factors or land management practices, such as N deposition and mowing, have been rarely studied.
Increasing atmospheric N deposition is one of the most disturbance factors related to global change. The N deposition rate has increased 2-3 times since the industrial revolution, and is predicted to continuously increase during this century Penuelas et al., 2013). Given that N is one of the main factors limiting plant growth (Lebauer & Treseder, 2008;Liu et al., 2017), N deposition can stimulate plant growth (Chang et al., 2016;Xia & Wan, 2008), which in turn affects animal communities (Cebrian, Kielland, & Finstad, 2008;Wimp, Murphy, Finke, Huberty, & Denno, 2010).
Previous studies have reported that N addition can alter life history strategies (Parsons, Hellgren, Jorgensen, Leslie, & Benton, 2005) and food selection of the rodents (Yi, Li, Zhang, Zhang, & Wang, 2016), but the effect of nitrogen addition on rodent density or on the extent of damage they cause has not been documented.
Additionally, mowing can expose rodents to their predators, thereby reducing rodent damage (White, Horskins, & Wilson, 1998). Mowing reduced the population density of Microtus, but it increased that of Peromyscus maniculatus, in a tallgrass prairie in North America (Lemen & Clausen, 1984). It also increased the population size of Sigmodon hispidus and Microtus ochrogaster, but did not affect populations of Peromyscus leucopus and P. maniculatus, in an old grassland in eastern Kansas (Slade & Crain, 2006). Thus, the impacts of mowing on rodent damage remain controversial. Furthermore, N addition and mowing have interactive effects on plant growth , which may affect rodent damage.
However, to the best of our knowledge, interactive effects of N addition and mowing on rodent damage have not been reported.
Therefore, we conducted an in situ experiment involving N addition and mowing in a semiarid steppe in Duolun County in the Xilingol grassland, and investigated burrow density and percentage of rodent damage area, and plant community parameters including cover, height, aboveground net primary productivity, Shannon-Wiener diversity index, Simpson dominance index, and evenness in the experiment.
We hypothesized that N addition would aggravate, whereas mowing would alleviate rodent damage, since N addition increases, while mowing decreases aboveground biomass, which is the food source used by rodents. Furthermore, with the interactive effect between N addition and mowing on plant growth in mind , we hypothesized that N and mowing would interactively affect rodent damage.

| Description of the study site
This study was performed in a temperate steppe in an ecotone of agriculture-animal husbandry in Duolun County (42°02′N, 116°17′E, 1,324 m a.s.l.), in the Xilingol grassland, Inner Mongolia, China. The Xilingol grassland is an important pasture area and ecotone between agriculture and animal husbandry in China. It covers 2.0 × 10 5 km 2 (Han, Owens, Wu, & Huang, 2009) and supports a human population of 2.1 × 10 5 (Shiyomi et al., 2011). It is estimated that 64% of the Xilingol grassland is degraded; thus, forage quality has drastically decreased in the region (Han et al., 2009). Rodent damage is one of the most important factors underlying degradation in the Xilingol grassland. The rodent damage area amounts to 2-3 × 10 4 km 2 , causing severe economic losses (Zhang, Yang, Wang, Cai, & Qiao, 2014). The N deposition rate has reached 9.24 kg N ha −1 year −1 and steadily increases at a rate of 0.42 kg N ha −1 year −1 in this area Lü & Tian, 2007). Mowing is the only management approach for the local herdsmen to store forage in the nongrowing season (Cao et al., 2009).
However, the effects of N addition and/or mowing on rodent damage in this area have not yet been investigated. The mean annual temperature in the steppe is approximately 2.1°C, with temperature ranging from −17.5°C in January to 18.9°C in July. Mean annual precipitation is approximately 383 mm, 90% of which falls in the growing reason from May to October. The soil is chestnut-type (Calcic Kastanozems). The  (Zhang & Ye, 1988) and Pest Management (Guo et al., 2005), and then released. Only ground squirrel (C. dauricus) was trapped during the experimental period. C. dauricus are the common rodents in the steppe. They are morphologically and behaviorally similar to rats. Their diets are the green parts of most plant species in the steppe, including A. frigida, S. krylovii, and M. ruthenica (Liu, Wang, Wang, Han, et al., 2013). Ground squirrels live underneath the ground, in burrow systems, and push loosened soil to the surface, thus destroying the grassland by burying the plants (Liu, Wang, Wang, Han, et al., 2013). Their population density was estimated to have reached 6.5 heads per ha in the Inner Mongolian typical grassland (Liu, Wang, Wang, Zhang, et al., 2013).

| Experimental design
The experiment was laid out in a factorial design with two factors, each with two levels: N addition (ambient N deposition and ambient plus 10 g N m −2 year −1 ) and mowing (no mowing, mowing). Thus, there were four treatments, including (1)

| Soil temperature and moisture
Soil temperature at the depth of 10 cm was measured three times a month, with a thermocouple probe  Lincoln, NE, USA) in 2014. Soil temperature in each plot was measured at three random points in each measurement. Soil moisture at a depth of 10 cm was measured four times a month, with portable time domain reflectometer equipment (TDR200; Soil Moisture Equipment Corp., Santa Barbara, CA, USA). As in soil temperature measurement, soil moisture was probed randomly at three points in each plot. Neither soil temperature nor moisture was affected by N addition, mowing, or their interaction (Table 1, Figure 2).

| Ground squirrel burrow density and damage area percentage
Burrow density in the steppe remained at about 20 ha −1 from 2010 to 2013; however, it increased to 375 ± 190 ha −1 in 2014, ranging from 60 ha −1 to 1,500 ha −1 (observed by Liu and Ma). Nonetheless, no burrow was found before 2014 within the experimental site (Observed by Liu and Ma). Burrow density and damage area percentage were evaluated in 2014 following the methods used in a previous study in a Tibetan alpine meadow (Li et al., 2011). The number of the burrow entrances ( Figure 1b) was counted in late July, when burrow number peaks. The loosened soil-buried area in each plot was divided into many rectangles and triangles. The total area of all the rectangles and triangles was added together and then divided by the total plot area (16 m 2 ) to calculate damage area percentage.

| Plant community survey
All community parameters were monitored at peak plant biomass in early August 2014. One 1 × 1 m 2 quadrat was established at each plot in 2005, and a 1 × 1 m 2 steel frame with 10 × 10 cm 2 grid squares was installed above the canopy of each quadrat to facilitate cover estimation. Species cover for each species in all the grids was visually estimated, and community cover was calculated by the sum of all species covers. The number of individuals of each species in the quadrat was counted as the density at the species level, and the densities of all species in the quadrat were added together as the community plant density. Species richness was recorded as the number of plant species in the quadrats. Plant height of each species was randomly measured on F I G U R E 1 Photographs of rodent and mousetrap (a); entrance to a burrow (b) in the experimental site TA B L E 1 Effects of mowing (M), N addition (N), and their interaction on soil temperature, soil moisture, plant community cover, density, species richness, community height, ANPP, Shannon-Wiener index, dominance, evenness, burrow density, and damage area, as analyzed by two-way ANOVA ^, p < 0.1; *, p < 0.05; **, p < 0.001  (Liu, Mu, Niklas, Li, & Sun, 2012).
After estimations of cover, density, species richness, and height were completed, aboveground living biomass of each species within the quadrat was clipped and oven-dried to constant weight at 65°C.
Total dry biomass for each plot was calculated as the aboveground net primary productivity (ANPP). Shannon-Wiener diversity index, Simpson dominance index and evenness of every plot were calculated based on biomass data of plant community in each quadrat.

| Ground squirrel damage
Nitrogen addition significantly enhanced burrow density by 2.8fold (p < .05, Figure 3a), while mowing significantly decreased it by 75.9% (relative difference, Table 1, Figure 3a). Additionally, we ob- Damage area increased 4.7 times by N addition (p < .05, Figure 3b), whereas it decreased by only 14.5% due to mowing (absolute difference, Table 1, Figure 3b). Furthermore, significant interactions were found between N addition and mowing on damage area percentage (Table 1). N addition enhanced the damage area percentage by 23.4% in the nonmown plots (absolute difference, post hoc, p < .05, Figure 3b), but only increased it by 7.1% in the mown plots (post hoc, p < .05, Figure 3b). On the other hand, mowing reduced the damage area percentage by 6.3% in the ambient N deposition plots (absolute difference, post hoc, p < .05, Figure 3b) and decreased it by 22.6% in the N addition plots (absolute difference, post hoc, p < .05, Figure 3b).

| Plant community
Nitrogen addition significantly elevated community cover by 10.4% (absolute difference, Table 1, Figure 4a). In contrast, mowing did not alter community cover (  Figure 4f-h). However, a significant interactive effect by nitrogen addition and mowing on the Shannon-Wiener diversity index was detected (Table 1). N addition significantly reduced the diversity index by 0.35 in the mown plots, but not in the nonmown plots ( Figure 4f).
Damage area percentage significantly increased with increasing community cover (R 2 = .440, p = .001, Figure 5d) and community height (R 2 = .714, p < .001, Figure 5e); additionally, it marginally increased with increasing ANPP (R 2 = .182, p = .060, Figure 5f).  Zhang, Zhong, & Fan, 2003). Thus, we may conclude that the steppe suffered serious rodent damage. There may be two reasons for the higher density of C. dauricus observed in the study. First, the steppe is dominated by A. frigida, which occupied 30%-50% of biomass in the community (Yang et al., 2011), and happens to be the main diet for C. dauricus (Guo et al., 2005;Liu, Wang, Wang, Zhang, et al., 2013). Second, the study site is adjacent to a cropping area cultivated with potatoes, which provides abundant food for the rodents. Additionally, previous studies have reported that C. dauricus is a common rodent species in the steppe, but is less destructive than other rodent species, such as Brandt's vole (Lasiopodomys brandtii), the Chinese striped hamster (Cricetulus Barabensis), and the Mongolian gerbils (Meriones unguiculatus) (Liu, Wang, Wang, Han, et al., 2013;Liu, Wang, Wang, Zhang, et al., 2013;Zhang, Zhong, & Fan, 2003). Thus, C. dauricus is not considered as the dominant rodent pest in the Inner Mongolian grassland (Harris, 2010;Zhang, Zhong, & Fan, 2003). Nevertheless, our results suggested that C. dauricus can cause drastic disturbance in the grassland, and therefore, the damage induced by this species should be considered in the ecotone of agriculture-animal husbandry.

| D ISCUSS I ON
Multiple previous studies suggested that the abundance of herbivores increased with N addition (Bowdish & Stiling, 1998;Wimp et al., 2010); however, those results mainly concerned the effects of N addition on the abundance of insects. Our study showed that N addition increased both, burrow density and damage area. These observations support our hypothesis and confirmed that outbreaks of small mammals may result from N addition, suggesting that N deposition-induced rodent damage must be considered in the future. There may be at least three potential causes for the outbreak of C. dauricus in the present study. First, N addition stimulated plant growth, thereby increasing the food resources for herbivores (Nijssen, Wallisdevries, & Siepel, 2017;Ritchie, 2000). In this aspect, although N addition did not significantly increase ANPP in the present study, ANPP showed an increasing trend (Figure 4e). Second, a taller plant community and a higher provided increased opportunity for the rodents to hide and escape from predators, mainly night owls and hawks (Andrey, Humbert, Pernollet, & Arlettaz, 2014;Woodcock & Pywell, 2009). Third, as we did not record data on feeding behavior by C. dauricus, we cannot exclude effects of N addition on food quality and intake selectivity by the rodents, as has been reported in a previous study (Yi et al., 2016).
Our study showed that mowing reduced both burrow density and damage percentage of C. dauricus, which further supports our hypothesis. These results are consistent with previous studies by Lemen and Clausen (1984), and by Edge, Wolff, and Carey (1995); however, the phenomenon disagrees with results observed by Jacob (2003), and by Slade and Crain (2006). In contrast to the positive effects of N addition, mowing reduced the ANPP (Figure 4e), consequently reducing the food supply, which in turn resulted in the reduction in rodent population density. Moreover, mowing may also decrease the standing litter height and alter the microclimate, which in turn would alter the plant growth strategy and decrease plant height (Klimesova, Latzel, de Bello, & van Groenendael, 2008;Liu et al., 2017). Actually, plant community height in the mown plots was marginally lower than that in the nonmown plots (Figure 4d), which we attributed to changes in growth strategy, considering the unchanged microclimate in our study. A shorter plant community height increases the risk of exposure to predators (Andrey et al., 2014;Ritchie, 2000;Woodcock & Pywell, 2009); therefore, rodents may choose taller plant communities to dig their burrows.
The results indicate that mowing may be an effective method for controlling rodent damage in grasslands.
In addition, our observation that N addition-induced outbreak of rodent damage was alleviated by mowing in the steppe supports our hypothesis about the interaction between N addition and mowing on rodent damage. The reason for this may be attributed the interactive effect between N addition and mowing on plant growth, which has been reported in a previous study . Actually, the F I G U R E 5 Correlation between burrow density and community cover (a), community height (b), and ANPP (c); correlation between damage area percentage and community cover (d), community height (e), and ANPP (f). Data points in each scatter diagram represent the values in experimental plot. Data of burrow density was lg-transformed in the figure plant community was much shorter in the nitrogen addition plots than in plots where no N was added (Figure 4d). Given that the effect of N addition on rodent damage was largely attributed to the increase in plant community height, mowing should partially suppress the effect of N addition on rodent damage, as, in fact, was recorded.

| CON CLUS ION
Our results showed that both N addition and mowing significantly altered rodent damage caused by the variation caused in plant community parameters. As rodents are the dominant pest in the northern grasslands of China, variation in rodent damage (e.g., burrow density and damage area) may result in changes in plant community structure and soil texture and structure, thereby affecting grassland productivity. Our study suggests that mowing can alleviate rodent damage and indicates that the impact of the changing atmospheric composition on rodent damage can be controlled by proper land-use management in the semiarid grassland. Youth Training Foundation of Henan University (yqpy20140031).

CO N FLI C T O F I NTE R E S T
None declared.