Effects of thinning on scatter-hoarding by rodents in temperate forest

Deforestation and thinning are human activities that can destabilize the forest ecological system and, consequently, impact significantly on habitat and behavior of forest-dwelling animals. This hypothesis was tested in Yugong in the Mount Taihangshan area by comparing the tracks of tagged seeds of Armeniaca sibirica . in sites of unthinned and thinned forests. Our results showed that: (i) the diversity of vegetation and rodents drastical ly reduced in sites with thinned forests, compared to unthinned sites; (ii) the amount of both removed and scat -ter-hoarded seeds significantly declined in sites with thinned forests, compared with the unthinned sites; (iii) there was no significant difference observed in the distance of seed dispersal between the thinned and unthinned areas; and (iv) the thinning did not show a significant change to the model of cache size. These results suggest ed that the thinning of forests negatively influenced the species richness and food-hoarding behavior of rodents. In addition, the results indicated that the weakened scattered-hoarding might be disadvantageous to seedling re cruitment and forest restoration.


INTRODUCTION
Habitat destruction and fragmentation are the major reasons for species loss (Pimm & Raven 2000). Categorically, cutting of trees in forests is a human activity that dramatically alters forest habitat (Nepstad et al. 1997). This activity, mostly referred to as deforestation, is known to have a negative influence on the diversity of local plants and animal species (Uhl et al. 1991;Verissimo et al. 1992;Nepstad et al. 1999). Forest birds and squirrels are easily disturbed and harmed by deforestation of their habitats (Wilson & Wilson 1975).
The term "thinning", as related to forestry activity, is defined as selective deforestation for timber and/ or non-timber products (Smith David 1986;Nepstad et al. 1999;Yu et al. 2006;Chen et al. 2008), involving the selective removal of large tradable trees for specific purposes (Asner et al. 2005). In comparison, the deforestation is also selective-trees are selectively logged per unit area, however, these logged forests are usually highly degenerated (Uhl et al. 1991;Johns et al. 1996). Nevertheless, deforestation or thinning are associated with intense altering of both the relative species abundance and the vegetation coverage (Asner et al. 2005(Asner et al. , 2006Bleher et al. 2006;Farwig et al. 2006;Yu et al. 2006;Chen et al. 2008). This consequently results in habitat fragmentation, which leads to decreased food availability for forest-dwelling animals (Kirika et al. 2008;Poulsen et al. 2011). Eventually, this contributes to the food-hoarding behavior of animals (Schleuning et al. 2011;Lai et al. 2014).
In the recent past, similar studies on forest thinning have attracted controversy, with some published studies claiming that thinning enhances seed dispersal by seed eating animals (Farwig et al. 2006;Schleuning et al. 2011); other studies have strongly argued that it reduces seed dispersal (Kirika et al. 2008;Gutiérrez-Granados 2011). Furthermore, numerous studies have suggested that thinning would have no effect on seed dispersal rates but can decrease the dispersal distance (Wright et al. 2000;Gutiérrez-Granados 2011). Despite all these arguments, however, the overwhelming consensus is that habitat disturbance severely influences the composition of native rodent species (Kirika et al. 2008) and their food-hoarding behavior (Farwig et al. 2006;Kirika et al. 2008;Wang et al. 2011). It is also the general consensus that habitat disturbance impairs the effects of seed dispersal, seed germination and seedling recruitment (Markl et al. 2012).
Seleced logging often severely impacts habitats, species and animal-mediated ecosystem processes in tropical forest (Nepstad et al. 1999). However, the consequences of deforestation and thinning, as associated to seed dispersal and scatter-hoarding by rodents, are rarely explored in temperate forests. Initially, we had hypothesized that thinning could change habitat and plant species formation and, subsequently, disturb the rodents' hoarding behavior. Furthermore, we had predicted that the dispersal rate, seed fate, dispersal distance and cache size might also be altered by the thinning operation. It was imperative in this study to evaluate the ecological influence of thinning on the food-hoarding behavior of rodents and to compare these parameters to the diversity of native plants and rodents. There was a need to conduct this study to unravel some existing controversy on forest thinning and rodent food-hoarding behavior in the temperate forest. This study can also contribute to enriching our knowledge of both animal protection and forest management.
Two experimental plots (each at least 200 × 300 m) were set up as follows: unthinned forest (no logging program and 1 km away from logging area) and thinned forest.

Investigations on species abundance of rodents and plants
The species composition of rodents was surveyed with wire live traps (30 × 13 × 12 cm) in May 2012 and 2013. In total, 80 traps baited with peanuts (Arachis hypogaea) were set up in each sample plot. Four transects were built at 25-m intervals, and 20 traps were set up at 5-m intervals along each transect. We checked the trap success at 0600 hours and 1900 hours, and recorded the species and sex of captured animals. Captured animals were marked by toe clipping and released at the capture site. Trapping operation was carried out for 3 consecutive days after the seed dispersal experiment.
We further investigated both the plant diversity and abundance in unthinned and thinned forests in May 2012 and 2013. To achieve this, 3 quadrats spaced at more than 30 m and 10 × 10 m in size were established along the terrain in each sample plot. The species and the amount of arbor and shrub were recorded.
During the same period of May 2012 and 2013, we also investigated the abundance of seeds in the soil in unthinned and thinned forests. This involved the setting of 15 quadrats (1 × 1 m), spaced at more than 5 m in each quadrat along the terrain. We subsequently sur-veyed and recorded the species and amount of seeds found in leaf litter, humus and soil layers up to 10-cm depth.

Seed manipulation and tracking
We collected mature seeds of A. sibirica in the fruiting period and dried them in the shade. Seeds with similar shape and features were selected and tagged with white plastic sheets at a size of 2.5 × 3.5 cm and weighing less than 0.3 g. A tiny hole was drilled into the edge of the husk, with kernels left intact without damage. The tag was numbered and tied through the drilled hole to the seed, with a thin steel wire (10 cm in length) (Xiao et al. 2006).
In May 2012 and 2013, 2 parallel transects spaced at 25-m intervals were set up in each sample plot. In each transect, 6 seed sites with spacing of 25-m intervals were demarcated. Then 50 marked seeds were released in each seed site, making a total of 12 × 50 seeds released in each sample plot. We examined the seeds every day to observe whether seeds were removed or not, and recorded the seed fate, cache size, burial depth and dispersal distance accordingly.

Data analysis
Data were analyzed with SPSS for Windows 16.0. The survival analysis of Cox regression was used to examine the influence of thinning on the seed dispersal rate each year, with RS, RC, A, AA, S, SA, SS and SEA as covariates. Generalized linear mixed models were used to analyze the influence of thinning on seed fate (EI, ER, SH, AS, R, M), dispersal distance and cache size. The thinning per year was designated as a fixed effect, and RS, RC, A, AA, S, SA, SS and SEA were used for random effects. All data were shown with mean ± SEM, and the significant statistical level was set at α = 0.05.

Species abundance of rodents and plants
In unthinned forests, 3 species of rodents, A. peninsulae, S. davidianus and N. confucianus, were live captured in both 2012 and 2013, with a total trap success rate of 6.67% in 2012 and 5.33% in 2013 (Table 1). In thinned forest, only A. peninsulae was trapped in 2012, with a total trap success rate of 2.67%, whereas the species of A. peninsulae and N. confucianus were captured in 2013, with a total trap success rate of 2.67% (Table 1).
In unthinned forests, the main tree species recorded during the 2 years was Chinese cork oak (Q. variabilis) with a small number of black locust (R. pseudoacacia), while the vegetation of shrub mainly consisting of Vitex negundo var. heterophylla, C. coggygria and R. xanthine. In thinned forests, only Q. variabilis and V. negundo were found in 2012, while Q. variabilis, V. negundo, C. coggygria and R. xanthine were recorded in 2013 (Table 1). During the 2 years under review, there was a tree density of 0.48/m 2 in the unthinned forests and 0.21/m 2 in the thinned forests; the density of shrubs was up to 0.22/m 2 in the unthinned forests in both years, while the thinned forests had the density of shrubs of 0.06/m 2 in 2012 and 0.15/m 2 in 2013 (Table 1).
In 2012, 2 species of seeds, Q. variabilis and A. sibirica with a seed abundance of 3.80/m 2 , were found in unthinned forests. However, only Q. variabilis seeds were found in thinned forests during the same year, with a seed abundance of 1.80/m 2 (Table 1). In 2013, seeds of Q. variabilis, A. sibirica and A. davidiana were found in unthinned forests, with an abundance of 2.20/m 2 . However, only Q. variabilis seeds were found in thinned forests, with an abundance of 1.40/m 2 ( Table 1).

Dynamics of seeds dispersal
In 2012, at the end of the survey, 50% of seeds remaining in situ in thinned forests, while only 0.33% of seeds were left in unthinned forests. The 2-day median survival time of seeds released in unthinned forests was found to be significantly lower than the 10 days in thinned forests (χ 2 = 302.570, df = 1, P < 0.001) (Fig. 1a). In 2013, 3.33% of seeds remaining in situ in thinned forests at the end of the survey, while no balance of seeds was observed in unthinned forests. Similarly, the 2013 median survival time of seeds in unthinned forests was 7 days: significantly lower than the 14 days in thinned forests (χ 2 = 560.087, df = 1, P < 0.001) (Fig. 1b).
The seed dispersal rate showed significant difference between 2012 and 2013 (Wald = 185.345, df = 1, P < 0.001). Thinning displayed a significant influence on the seed dispersal rate (Wald = 220.374, df = 1, P < 0.001), but RS, RC, A, AA, S, SA, SS and SEA did not have significant effects on it.

Seed fate
The proportions of EI, ER and SH were consistently lower in thinned forests than that in unthinned forests in the 2 years (Fig. 2). In contrast, the proportion of AS and M showed some levels of inconsistency in both thinned and unthinned forests over 2 years (Fig. 2).

Dispersal distance of seeds
In 2012, the mean dispersal distance in thinned forests (11.71 m ± 0.71) was further than in unthinned forests (7.40 m ± 0.48) (Fig. 3a). In contrast, the median dispersal distance in the unthinned forests (6.68 m ±  (Fig. 3b). A significant difference was shown on dispersal distance between 2012 and 2013 (t = 6.612, df = 624, P < 0.001); however, there was no significant influence observed on dispersal distance between thinned and unthinned forest (t = −1.737, df = 624, P = 0.083). In addition, the dispersal distance was not significantly affected by the categories of RS, RC, A, AA, S, SA, SS and SEA.

Cache size
Rodents scatter hoarded 1, 2 or more than 3 seeds in one cache site. In both years 2012 and 2013, the cache sites in thinned forests were predominantly found with 1 seed, while more 2 or ≥3 seeds cache sites were found in unthinned forests (Fig. 4). During the years under study, results showed significant difference in the 1 seed cache site (t = −2.885, df = 45, P = 0.004) and in the >3 seeds cache site (t = 3.143, df = 45, P = 0.003), but no significant difference in the 2 seeds cache site (t = 0.526, df = 45, P = 0.599).

DISCUSSION
The results from our study showed that thinning negatively influenced the abundance of species diversity on vegetation, rodents and soil seeds. Similarly, the seed dispersal rate and seed fate were significantly influenced by thinning, which was in accordance with our hypothesis and predictions.

Effects of thinning on vegetation and rodents
Our study supports the fact that forest thinning and deforestation have a negative influence on ecological structures, which consequently impairs the animal-mediated ecosystem processes, resulting in a reduction of food abundance for forest-dwelling animals (Farwig et al. 2006;Broadbent et al. 2008;Kirika et al. 2008). In addition, this human disturbance is significantly associated with the decrease in plant diversity and quantity Farwig et al. 2006;Kirika et al. 2008). Therefore, the present study does not concur with the research results that moderate thinning does not influence the diversity of plants and rodents (Edwards et al. 2011;Schleuning et al. 2011). The reasons for rodent reduction may be attributed to the decrease in vegetation density, which increases the predation risk (Kirika et al. 2008;Markl et al. 2012), and/or the habitat destruction leading to the migration of rodents.

Influence of thinning on seeds dispersal
The finding that thinning might be conducive to increasing seed dispersal (Farwig et al. 2006;Neuschulz et al. 2011;Schleuning et al. 2011) conflicts with the re-sults of our study where we found that seeds dispersed more rapidly in the unthinned forests. The reasons might be that disturbance from thinning decreases rodent species number and density due to migration, which significantly impacts on seed dispersal (Wong & Jones 1985;DeMattia et al. 2004;Li & Zhang 2007); furthermore, the reduction in coverage by trees and shrubs directly reduces animal activities (Lima & Dill 1990;Perea et al. 2011;Steele et al. 2015), and the seed dispersal by rodents is positively correlated with the coverage of vegetation (Wada 1993;Perea et al. 2011).

Difference in seed fate with thinning
Our results showed that rodents ate and scatter hoarded less seeds in thinned forest than in unthinned forest. It has been confirmed that thinning can reduce the species and relative density of rodents (Wang et al. 2011), and seed fate is closely related to the rodent species and quantity (Li & Zhang 2007;Lu & Zhang 2008;Chang & Zhang 2011). For example, different species of rodents usually adopt different hoarding strategies (Fleming & Brown 1975;Lu & Zhang 2008;Chang & Zhang 2011), and rodents with different body sizes usually exhibit different hoarding behaviors (Vieira et al. 2003). Meanwhile, the composition and structure of forests changes following thinning, and the reduction of canopy coverage leads to increases in the predation risk when rodents forage and hoard food (Perea et al. 2011). The thinning eventually has an adverse impact on rodents' feeding and hoarding activities (Leaver 2004;Sivy et al. 2011).

Variation in dispersal distance with thinning
Previous studies have shown that dispersal distance of seeds can be influenced by seed size and nutrition (Xiao et al. 2005), predation risk (Lima & Valone 1986;Leaver 2004), seed production (Vander Wall 2002;Jansen et al. 2004) and human disturbance (Kirika et al. 2008;Gutiérrez-Granados 2011). Our study indicated that thinning did not have significant influence on dispersal distance, a finding that concurs with some other published studies (e.g. Markl et al. 2012). However, Wright et al. (2000) and Gutiérrez-Granados (2011) demonstrated that dispersal distance of seeds was short in thinned areas. We suspect that this dispute comes from the complexity of influencing factors on dispersal distance. Further studies, involving long-term sampling strategies, should be carried out to verify these parameters.

Effect of thinning on cache size
Our study showed that rodents scatter hoarded 1, 2 or more than 3 seeds at each site, and this is consistent with some previous studies (Ma et al. 2010;Wang et al. 2013). Thinning increased the proportion of cache sites with one seed but this was not significant. The cache size was always influenced by a variety of internal and external factors, such as the body size of hoarders and seed size (Vander Wall 1990;Vieira et al. 2003). These results may be attributed to a long-term adaptation by rodents to the environment.

CONCLUSIONS
Thinning can strongly affect the species richness and community composition of vegetation and rodents in temperate forests. Thinning weakens the seed dispersal and scatter-hoarding behavior of rodents, and, hence, disrupts animal-mediated ecosystem processes. Furthermore, this typical dispersal can be disadvantageous to seedling recruitment and forest restoration.

ACKNOWLEDGMENTS
We thank the staff of State-owned Yugong Forest Farm for their help with field work. This study was funded by the National Basic Research Program of China (No. 2007CB109106), the Key Research Programs in Colleges and Universities of Henan Province (No. 16A180039) and the Postdoctoral Research Foundation of Zhengzhou University.