Introduction

Bark stripping (Saint-Andrieux et al. 2009; Borkowski and Ukalski 2012; Arhipova et al. 2015; Yen et al. 2015) and browsing (Kupferschmid and Bugmann 2008; Hidding et al. 2012; Kupferschmid et al. 2015) by deer can strongly affect tree stand structure and species composition. Bark stripping has strong negative effects on xylem water conductivity (Welch et al. 1997). It also increases a tree’s susceptibility to fungal infection (Arhipova et al. 2015), which has considerable impacts on tree growth and survival. After bark stripping, physical and physiological damage to the trees continues; some trees suffer repeated stripping (Akashi and Nakashizuka 1999; Vospernik 2006; Welch and Scott 2017). Several surveys are needed to clarify the effects of repeated bark stripping, but few such studies have been done (Akashi and Nakashizuka 1999; Arhipova et al. 2015; Welch and Scott 2017), nor has any study examined the longevity of trees after repeated bark stripping.

The management of the effects of deer in forest ecosystems requires quantification of the effect of bark stripping in relation to other factors (e.g., competition) that influence tree survival. Although stripping intensity may govern the fate of stripped trees (e.g., stripped area; Akashi and Nakashizuka 1999), tree survival is also influenced by competition (e.g., shading and crowding) with neighboring trees (e.g., Coates et al. 2013; Nagaike and Takanose 2014; Fraver et al. 2014). Thus, describing which factors most impact survival is necessary. However, no study has simultaneously examined the effects of bark stripping and competition on the dynamics of forest trees. Such a comprehensive approach to identify the factors most affecting tree mortality is required for planned management of deer effects in forest stands.

In Japan, populations of sika deer, Cervus nippon, have increased rapidly since the 1990s (Takatsuki 2009; Kaji et al. 2010), resulting in negative impacts on forest trees due to deer browsing (Akashi et al. 2015) and changes in forest understory species (Suzuki et al. 2013) and grasslands (Nagaike 2012; Nagaike et al. 2014; Ohashi et al. 2014). Further, the impacts of bark stripping on trees in plantations (Nagaike and Hayashi 2003; Jiang et al. 2005) and natural forests (Takeuchi et al. 2011; Iijima and Nagaike 2015a) in Japan are serious. However, few studies exist that examined tree longevity and survival after bark stripping and that identify the factors most affecting the survival of stripped trees. To identify the effects of bark stripping, we need repeated surveys that consider multiple factors; however, until now, no such study had been conducted.

The aim of this study was to (1) examine how bark stripping intensity affects the longevity and survival of trees and (2) to determine the effects of bark stripping and competition from neighboring trees on the survival of Abies veitchii, a dominant tree species in subalpine coniferous forests in central Japan.

Materials and methods

Study site

The study was conducted in a subalpine zone on the northern slope of Mt. Fuji, located in central Japan (2100 m above sea level [a.s.l.]; 35°22′N, 138°41′E). At the nearest meteorological station (Kawaguchiko; 860 m a.s.l.), mean annual precipitation was approximately 1600 mm and temperature was 10.6 °C. Snow cover in the study site was usually 50 cm from December to April.

Most of the area is covered by old-growth forests, typically dominated by A. veitchii, Abies mariesii, and Tsuga diversifolia (Ohsawa 1984; Sugita and Nagaike 2005). Some sections of the forest were impacted by a road constructed around 50 years ago, while other parts retained their old-growth condition (Nagaike 2003). The estimated density of C. nippon in this area increased sharply from 1.4/km2 in 2005 to 55.1/km2 in 2012 (Iijima et al. 2013).

Field study

In 1999, a 50 × 140 m (0.7 ha) plot was set in the old-growth subalpine coniferous forest. The long side of the plot was perpendicular to the road (Nagaike 2003). Many saplings grew near the road and beneath canopy gaps in the forest interior, where strong sunlight penetrated (Nagaike 2003). The slope of the plot was relatively gentle (approximately 15°), and the topography was simple. The plot was divided into 280 grids of 5 × 5 m. In 1999, all trees in the grids that were ≥ 2 m tall were tagged and identified to the species level. Their girth was measured at breast height. These trees and new recruits were counted and measured in 2003, 2005, 2007, 2012, and 2017. To maintain the same duration of bark stripping for all stripped trees, I focused on trees that were first stripped in 2005. Tree survey data collected in 2005, 2007, 2012, and 2017 were examined. At each survey, a cause of death was assigned to each dead tree [i.e., uprooted, snapped, standing dead, toppled by other trees (Holzwarth et al. 2013), or bark stripped].

In the study stands, coniferous tree species had been stripped but not heavily browsed. Similarly, other studies in this area did not report damage from browsing (Jiang et al. 2005; Takeuchi et al. 2011), in contrast to reports from studies conducted in Europe (e.g., Gill 1992; Kupferschmid et al. 2014, 2015; Nagel et al. 2015). In western Japan, bark stripping was the dominant cause of mortality of A. veitchii (Tsujino et al. 2013). Thus, I focused on bark stripping as the main damage caused by sika deer. The degree of deer bark stripping visible on each tree was recorded in 2005, 2007, 2012, and 2017, recorded as the proportion of the tree circumference that had been stripped, using increments of 10%. This proportion is referred to hereafter as the bark stripping ratio (SR). A 100% SR indicated that deer had stripped all bark from the circumference at a chosen height from the ground. Our analyses focused on A. veitchii because this species was the most impacted by deer in 2005 (see Results and Fig. 2). Because most trees suffered repeated stripping (see Results), the maximum bark stripping ratio (MSR) of each tree was determined throughout the study period.

Analysis

I used the asymptotic Wilcoxon signed rank test to compare the diameter at breast height (DBH) of stripped and nonstripped trees. Tree competition (i.e., shading and crowding) was analyzed using asymmetric and symmetric competition models, in which crowding and shading in the asymmetric model were expressed as the cumulative basal area (CBAa for the asymmetric model) of a specific range of trees that had DBH values greater than that of the focal tree. For the symmetric model, the CBA (CBAs for the symmetric model) of all trees other than the focal tree was used for the analysis (Masaki et al. 2006; Nagaike and Takanose 2014). The CBAa and CBAs values in 2005 for every bark-stripped tree were calculated in each grid. These values were strongly correlated in each of the grids (p < 0.001, r2 = 0.971; Pearson’s correlation analysis). Thus, for this study, we used only the CBAs value as a competition index (CI).

Factors (i.e., SR, MSR, and CI) affecting the longevity and survival of A. veitchii trees after bark stripping were identified using a recursive partitioning and regression tree analysis (Therneau et al. 2015). This analysis applies a binary recursive partitioning approach to split the data set into subsets based on explanatory variables chosen to minimize the deviance in the response variables in each of the resulting subsets. Longevity after bark stripping in 2005 and after attaining MSR, and the survival of stripped trees from 2005 to 2017, were the response variables of each focal tree. These variables were applied separately. We applied the DBH in 2005, the CI where the tree was located in the grid (a measure of the intensity of the competition in 2005), the SR in 2005, and the MSR in their survived period, as explanatory variables for each focal tree.

All statistical analyses were performed using R 3.4.1 software (R Core Team 2017) with the rpart package (Therneau et al. 2015) for the regression tree analysis, using default settings.

Results

Changes in stand structure and bark stripping

In the study plot, T. diversifolia and A. mariesii were the dominant components of the basal area (BA) and stem density (Fig. 1). Abies veitchii had a lower BA and stem density, indicating the presence of many small trees of this species. Tree densities declined from 2005 to 2017, particularly for A. veitchii, while DBH and BA increased. Between 2005 and 2017, the majority of dead A. veitchii that had not been stripped were assigned as standing dead (99%), followed by snapped trees (1%). Thus, no severe disturbance event (e.g., a typhoon or insect outbreak) occurred during the study period.

Fig. 1
figure 1

Changes in stand structure variables

Full size image

Bark stripping peaked in 2005. At this time, the density of newly stripped A. veitchii was the highest of all species (Fig. 2). The DBH values of bark-stripped trees were significantly lower than those of non-bark-stripped trees (Table 1).

Fig. 2
figure 2

Changes in stem density in newly stripped trees

Full size image
Table 1 Comparison of the diameter at breast height between bark stripped and non-stripped trees of Abies veitchii
Full size table

The survivorship of trees stripped in 2005 varied in SR (Fig. 3). Trees with SR > 80% in 2005 had lower survivorship compared to trees with a lower SR. The survivorship of non-stripped trees (0%) was similar to trees with an SR < 70% (Fig. 4). In 2005, most of the bark-stripped trees were repeatedly stripped; consequently, the MSR for most trees during the study period was larger than the SR recorded in 2005 (Fig. 4).

Fig. 3
figure 3

The survival ratio and bark stripping ratio (i.e., the proportion of the tree circumference that had been bark stripped, in 10% increments) in 2005 for Abies veitchii

Full size image
Fig. 4
figure 4

The relationship between the bark stripping ratio in 2005 and the maximum bark stripping ratio during the study period (2005–2017) for A. veitchii

Full size image

Effects on longevity and survivorship after bark stripping

For clarifying the factors affecting the longevity of trees after bark stripping in 2005, the first node of the regression tree analysis was divided at 65% of the SR in 2005 (Fig. 5a), indicating that trees with an SR > 65% in 2005 had shorter longevity. The second and third nodes of the regression tree analysis were divided by tree size and competition. Similarly, for factors affecting the longevity of trees after attaining MSR, the first node was divided at 65% of the MSR (Fig. 5b). Because DBH and CI were not selected as causal factors for the analysis of the MSR, the effects of size and competition on longevity after bark stripping were not as strong. After bark stripping in 2005, the factors affecting survival from 2015 to 2017 were mainly impacted by MSR, implying that trees with > 85% of the MSR tended to die (Fig. 6). SR was not selected in this analysis.

Fig. 5
figure 5

Factors affecting the longevity of trees after bark stripping in 2005 (a) and after the maximum bark had been stripped (b). SR: bark stripped ratio in 2005; MSR: maximum bark stripping ratio during the study period (2005–2017); DBH05: diameter at breast height in 2005; CI: competition index calculated from a symmetric competition model

Full size image
Fig. 6
figure 6

Factors affecting survival of bark-stripped A. veitchii from 2005 to 2017

Full size image

Discussion

Effects of repeated bark stripping

Here, as in previous studies (Akashi and Nakashizuka 1999; Welch and Scott 2017), many trees suffered repeated bark stripping. Thus, to interpret the effects of stripping accurately, it is important to evaluate the intensity of bark stripping that occurred during the study. This study used SR in the initial stage and MSR throughout the study to evaluate the stripping effects. After stripping in 2005, the factors affecting longevity were SR, DBH, and CI (Fig. 5a); however, only MSR and SR had the greatest impact on longevity after 2005 (i.e., when MSR was used) (Fig. 5b). Moreover, the effects on survivorship were dominated by MSR and not SR in 2005 (i.e., bark stripping at a point in time; Fig. 6). Thus, MSR showed good sensitivity for indicating the effects of bark stripping on longevity rather than SR in 2005 (Fig. 5). Many reports have surveyed the incidence of bark stripping (Welch and Scott 2017). Thus, continuous surveys and the use of MSR in analyses are important to help identify the impact of bark stripping.

Factors affecting survivorship

During the study, the survivorship of non-bark-stripped trees was approximately 70% (Fig. 3). Tree deaths were largely attributed to competition. Trees that perished were assigned as standing dead. The survivorship of trees with an SR in 2005 of < 70% was not strongly affected by bark stripping. However, an SR in 2005 of 80% or more showed quite strong effects. In bark stripping studies, the stripping intensity for each tree was estimated as one of three types: presence or absence (Nagaike and Hayashi 2003; Iijima and Nagaike 2015a), length and width of stripped area (Welch et al. 1997; Ueda et al. 2002; Arhipova et al. 2015), or percentage of the trunk circumference that was stripped (e.g., 25%, 50%, and 75%) (Akashi and Nakashizuka 1999; Takeuchi et al. 2011; Shibata 2007; Yokoyama et al. 2001). This study assessed the stripped percentage of the trunk circumference in 10% increments to express intensity at the individual tree level.

Full bark stripping of a tree by deer (i.e., 100% SR) is equivalent to girdling (Shibata 2007). Girdling can disrupt xylem water conductivity and sometimes rapidly kill a tree (Ueda et al. 2014). The survivorship rate decreased with bark stripping and was especially low when the MSR exceeded 85%. Yokoyama et al. (2001) found that 50% of A. veitchii trees survived 100% stripping, but survival after their survey ended was not clarified because they checked the effect of bark stripping on only one occasion. Studies of longevity and survivorship using 10% increments to indicate the stripped percentage of the trunk circumference is thus necessary.

Large wounds on stripped trees can promote fungal infections, water deficiency, and physical frailty. The wound damage inflicted by moose (Alces alces) and red deer (Cervus elaphus) on Pinus contorta increased over time and was strongly correlated with the extent of decay caused by fungal infections (Arhipova et al. 2015). Thus, high SR and MSR values could contribute to decreased longevity and low survivorship after bark stripping.

Bark-stripped trees were smaller than non-stripped trees (Table 1) as also noted in other studies (Nagaike and Hayashi 2003; Akashi and Terazawa 2005; Vospernik 2006; Welch and Scott 2017). The smaller trees were more susceptible to shading and crowding effects from larger trees (e.g., Fraver et al. 2014). Bark stripping had greater effects on survivorship than on competition, even though the bark-stripped trees were significantly smaller (Table 1 and Fig. 6). In Picea sitchensis plantations, large stripped trees survived better than the small stripped trees (Welch and Scott 1998). Here, the second most important factor affecting survival was the size of the trees stripped, indicating that the larger the tree, the greater the odds of survival (Fig. 6).

Welch et al. (1997) concluded that bark stripping in P. sitchensis is not a large threat to survivorship because healing of the damaged portions reduces mortality. For A. veitchii, damage does not heal well and instead becomes enlarged due to repeated bark stripping (Fig. 4). Thus, differences in survivorship among species after bark stripping may depend on wound-healing capabilities (Welch et al. 1997), the size of the deer population, and the intensity of their activity. Moreover, besides bark stripping, differences among species characteristics may also play a role in survivorship because long-lived species are often physically tougher than short-lived, opportunistic species. In Japan, Abies species are generally shorter-lived than Picea species (Kanzaki 1984; Miyadokoro et al. 2003). Thus, A. veitchii survivorship would be susceptible to bark stripping. Deer have preferences among the tree species they select for bark stripping (Iijima and Nagaike 2015b). Survivorship rates after bark stripping will depend on the taxa selected by the deer because tree species differ in their responses to this form of damage (Akashi and Nakashizuka 1999).

Conclusions

The survivorship of bark stripped A. veitchii trees was most strongly influenced by the maximum bark stripping ratio. Thus, it is important to assess the MSR to predict the impact of bark stripping. Although light bark stripping did not have a strong influence on survival of trees after 12 years, the mortality of heavily stripped stems greatly increased. Thus, management to prevent bark stripping should contribute to maintaining species composition in forests subjected to selective deer damage. In the absence of such management practices, selective disappearance of tree species is predicted (Schuldt et al. 2015; Bradshaw and Waller 2016) and may have a long-term impact on forest regeneration and succession (Akashi and Nakashizuka 1999; Yen et al. 2015).