- Email: [email protected]
Modelling browsing of deer on beech and birch in northern Germany
Forest Ecology and Management 358 (2015) 212–221
Contents lists available at ScienceDirect
Forest Ecology and Management journal homepage: www.elsevier.com/locate/foreco
Modelling browsing of deer on beech and birch in northern Germany Maria Bobrowski ⇑, Benjamin Gillich, Caroline Stolter Department of Animal Ecology and Conservation, University of Hamburg, Hamburg, Germany
a r t i c l e
i n f o
Article history: Received 26 January 2015 Received in revised form 24 August 2015 Accepted 26 August 2015
Keywords: Large ungulates Browsing damage Feeding habitat selection Vegetation composition Fagus sylvatica Betula spp.
a b s t r a c t Food selection by cervids is of high interest for forest management as they tend to browse on trees, causing severe damage and shifts in the regeneration capacity of certain preferred tree species. This behaviour could contradict forestry management strategies as deer may browse on target species, e.g. beech or oak, which are of high economic value for forestry and nature conservation. The choice of feeding habitat is influenced by a variety of different parameters, ranging from forest characteristics to human impacts. Therefore, we focused on factors influencing browsing of roe deer (Capreolus capreolus) and red deer (Cervus elaphus) on two different plant species: beech (Fagus sylvatica), which is an important species for forestry, and birch (Betula spp.), which may be an alternative food resource to beech. We measured browsing damage in relation to, variables such as tree density, tree height, ground cover vegetation, possible human impacts and landscape features in the Lüneburger Heide, Germany, for our modelling approach. As expected, our results revealed that browsing on beech is positively influenced by the availability of young beech and surprisingly by birch trees, whereas browsing on birch was influenced by the availability of birch trees only. Furthermore, the occurrence of blueberry (Vaccinium myrtillus) was positively and significantly related to browsing damage on both plant species, but a negative relationship was found for old Norway spruce stands (Picea abies). Surprisingly, other tree species, landscape features and human impacts had no significant influence on the feeding damage of either tree species. Our results indicate that the availability of stands exclusively consisting of birch (preferably young birch) might decrease the proportion of browsed beeches in neighbouring stands, or even favour the selection of birch over beech for food, which would consequently lead to increased future beech regeneration. Our modelling approach is applicable to forest management, providing an overview of interactions between browsing by cervids and young tree regeneration processes. Ó 2015 Elsevier B.V. All rights reserved.
1. Introduction The regeneration of natural woodland has become a major priority of sustainable forestry management globally over the last few decades and this is especially true in Europe (Zerbe, 2002; NMELV, 2011). Nowadays, nature conversion issues in Germany and other north-western European countries are focused on changing contemporary, monocultural coniferous forest stands into mixed stands with indigenous, deciduous tree species (Kuiters and Slim, 2002; Scherer-Lorenzen et al., 2005; Heinze et al., 2011). In our study area, the aim of modern and sustainable forest management is therefore the alteration of these coniferous monocultures into birch-oak (Betulo-Quercetum) and beech-oak (Fago-Quercetum) forest communities (Kuiters and Slim, 2002). Browsing by large herbivores on tree seedlings and saplings is considered a problem for forest regeneration worldwide ⇑ Corresponding author. E-mail address: [email protected] (M. Bobrowski). http://dx.doi.org/10.1016/j.foreco.2015.08.031 0378-1127/Ó 2015 Elsevier B.V. All rights reserved.
(Chouinard and Filion, 2005; Côté et al., 2004; Moser et al., 2006). Browsing pressure by cervids is considered an influential factor in the success of forest regeneration and has been well investigated (Gill, 1992a; Reimoser and Gossow, 1996; Kuiters et al., 1996). Many studies assume that continuous, selective browsing by cervids has a retarding effect on shrub and tree growth and negatively affects natural regeneration in forests throughout Europe (Gill, 1992a; Kuiters and Slim, 2002; Moser et al., 2006). Cervids can influence vegetation regeneration in various ways (Gill, 1992a): e.g. browsing may inhibit sapling growth rate and survival, leading to a modification of species composition in the understory and tree layer (Edenius et al., 2002). Furthermore, feeding behaviour may change tree morphology (e.g. enhanced or simplified branching pattern, compensation or slowed growth, e.g. Bergström and Danell, 1987; Bergquist et al., 2003; Hester et al., 2004; Stolter, 2008; Myking et al., 2013), resulting in reduced timber quality (Vila et al., 2003). The diet selection of cervids is influenced by food quality and quantity (Hagemoen and Reimers, 2002; Dussault et al., 2005;
M. Bobrowski et al. / Forest Ecology and Management 358 (2015) 212–221
Stolter et al., 2005, 2013; Ferretti et al., 2008; Bjørneraas et al., 2012). However, food quality is related to vegetation composition, which depends not only on abiotic factors (e.g. soil parameters, climate), but also on forest management plans (Morrison et al., 2006). Factors like the spatial arrangement of plant species (Bergman et al., 2005), cover and visibility (Mysterud and Østbye, 1999) and human impact (e.g. hunting strategies and distance to human settlements (Dussault et al., 2005; Newton et al., 2011)) may have an important role in forage selection by cervids. Therefore, the probability that a particular tree will be browsed may not only depend on its own characteristics, such as chemistry and morphology (Stolter et al., 2013), but also on the relative quality and abundance of surrounding vegetation (e.g. herbs and shrubs) (Gill, 1992b; Verheyden-Tixier et al., 1998) and the frequency of disturbance (Cross, 1998). Bee et al. (2009) showed that browsing on individual plant species is influenced by the availability of alternative forage plants in the surrounding environment. They suggested that browsing by deer on a focal plant is simply related to the lower palatability of the surrounding vegetation (Bee et al., 2009). Furthermore, plants with a lower nutritional quality e.g. lower concentration of plant secondary metabolites often received less damage than those of higher quality (e.g. Stolter et al., 2013). Many studies suggest that the availability of diverse food resources is a key factor for the amount of feeding damage on trees (Welch et al., 1991; Gill, 1992b; Vila et al., 2003; Moser et al., 2006). In this study we chose beech (Fagus sylvatica) as the most important forest regeneration species in our study area and birch (Betula spp.) as a species which might be of interest as an alternative food source because it is a common food plant for cervids, but is irrelevant for forestry. Improved knowledge about factors influencing browsing behaviour of both deer species on these trees is likely beneficial for the development of forest regeneration management strategies and could help minimise browsing damage on forest target species like F. sylvatica. Additionally, it may help identify management alternatives, such as enhancing the browsing damage on non-target species like Betula spp. Therefore, we addressed the following research questions:
213
2. Materials and methods 2.1. Study sites The study was conducted at Wilsede, Forestry Department Sellhorn (5390 N, 9590 E, 100 m above sea level), located southeast in the conservation area ‘‘Lüneburger Heide‘‘, Lower Saxony, Germany. The study area is characterised by a yearly precipitation of 800 mm and a mean annual temperature of 8C. Most of the 1240 ha study site is forested; the vegetation composition is dominated by coniferous trees, planted in monocultures over 200 years ago. Dominant overstory tree species are Scots pine (Pinus sylvestris), Norway spruce (Picea abies), Douglas fir (Pseudotsuga menziesii) and European larch (Larix decidua). The existing deciduous forest stands, such as beech (Fagus sylvatica), birch (Betula spp.) and oak (Quercus spp.), are relatively small and often mixed with other deciduous trees (alder buckthorn (Frangula alnus), hackberry (Prunus padus), holly (Ilex aquifolium), rowan (Sorbus aucuparia) and willow (Salix spp.)). Dominant ground cover vegetation species include blueberry (Vaccinium myrtillus), blackberry (Rubus fruticosus), raspberry (Rubus idaeus) and fern (Pteridium aquilinum). The nature conservation area ‘‘Lüneburger Heide‘‘ has ideal local site conditions; different forest succession stages (including early succession stages) and free-ranging red deer (C. elaphus) can be found here. This is relatively unique in Germany because red deer distribution ranges are usually restricted to declared ‘‘red deer areas”. The dominant large herbivores in the site are roe deer (C. capreolus) at an estimated density of 20 individuals per 100 ha. Red deer density is estimated at 0.2 individuals/ha and the sex ratio is 1:1 (K. Sierk, pers. comm.). At the time of the study we had no information on large predator presence in the study area, although wolves have been observed in other parts of the Lüneburger Heide since 2007.
2.2. Data collection (1) Which tree species is utilised the most by cervids in relation to tree abundance and height? (2) What are the factors influencing browsing on F. sylvatica and Betula spp. saplings in relation to surrounding vegetation, geographical features and human impact? (3) Are there differences in the influence of these factors on browsing between F. sylvatica and Betula spp.? In detail, we hypothesised that: (a) Deciduous trees are more used than coniferous trees by red deer (Cervus elaphus) and roe deer (Capreolus capreolus), due to their higher palatability and digestibility (i.e. plant tissues tend to contain lower amounts of secondary metabolites) (Bryant and Kuropat, 1980). (b) Smaller trees are more frequently browsed than larger trees as they are more accessible. (c) Presence of ground cover vegetation such as blueberry (Vaccinium myrtillus), blackberry (Rubus fruticosus), raspberry (Rubus idaeus) favours browsing on young deciduous trees, as they provide additional food resources, which might attract animals. (d) Geographical features influence feeding habitat selection by cervids. For example, human settlements, different road types (paths, hiking trails, roads for forestry, country roads and highways) and raised hides have a negative impact on the feeding habitat choice of the animals, while mineral licks have a positive impact.
Fieldwork was conducted between February and March 2012. A systematic design was used for data collection. Using ET Vector Grid in ArcView (Version 3.2; 1999), the position of the sampling grid was randomly selected and the sampling plots were regularly distributed 100 m apart (Fig. 1). Each intersection point represented a 2 2 m sampling plot. A total of 1309 plots were investigated covering 5236 m2. Within each sampling plot (1) tree species, (2) tree height, (3) browsing damage and height and (4) ground cover vegetation were recorded. We restricted our recordings to the ten most common tree species (Stolter, unpublished data). To avoid identification errors of winter-dormant trees, Betula pubescens and Betula pendula were not separated and were treated as Betula spp. Rubus fructicosus and Rubus idaeus were also not seperated and were treated as Rubus spp. Tree height categories (HC) were recorded to determine vegetation composition and occurrence of tree browsing since the animals forage preferentially on tree species up to 150 cm (C. elaphus) and 75 cm (C. capreolus) in height (Heroldova et al., 2003). All trees were classified at height categories: HC1: 40–100 cm, HC2: 101–150 cm and HC3: >150 cm. Tree heights under 40 cm were commonly browsed by rodents and lagomorphs and were therefore excluded. We estimated feeding damage in each 2 2 m plot by counting browsed species in each of the three height categories. We recorded overall browsing damage and divided trees into two further categories: browsed (b) and unbrowsed (ub). This categorisation was less subjective than assessing browsing scores for tree species and was conducted to minimise observer bias.
214
M. Bobrowski et al. / Forest Ecology and Management 358 (2015) 212–221
Fig. 1. Sampling grid of the study area in Forestry Department Sellhorn (5390 N, 9590 E, 100 m a.s.l.), ‘‘Lüneburger Heide‘‘, Lower Saxony, Germany (dots indicate sampling plots, N = 1309).
Furthermore, this method was less time-consuming than quantifying browsing damage by counting twigs (e.g. Stolter, 2008). We used the function ‘‘near table” in ArcGIS (Version 10.0; 2011) to calculate distances to human settlements and different road types (paths, hiking trails, roads for forestry, country roads and highways), as well as to raised hides (N = 176) and mineral licks (N = 46), to analyse their impacts on the browsing damage caused by deer. 2.3. Statistical analyses All statistical analyses were performed in SPSS Version 21 (2012). Data were tested for deviation from normality using the Kolmogorov–Smirnov test. The comparison of plots with browsing (N1 = 291) and plots without browsing (N0 = 1018) was performed with a Mann–Whitney-U-Test. Food species preference of cervids was evaluated using Chi2 statistics, in order to determine whether there was a significant difference between the expected frequencies and the observed frequencies of browsing between tree species of HC1. Expected values were calculated considering the differences in availability of the different plant species. A Generalized Linear Model (GzLM) was used to analyse the importance of different factors on feeding habitat selection of roe and red deer. Because of the distribution of the data (high number of zero values) we used Tweedie-distribution for modelling. Browsing damage on F. sylvatica (target species for forestry) and Betula spp.
(additional food resource) in HC1 were used as dependent variables. We included vegetation characteristics, browsing damage on all tree species and geographical features as potential covariates (Table 1). To reduce the number of variables we pre-examined our covariates for relationships with our dependent variables using Spearman correlations and scatter plots (Zuur et al., 2010; see also Stolter et al., 2013 for details). Thus, only variables that showed a significant relationship to damage on Betula spp. (HC1) or F. sylvatica (HC1) were used for the development of the models (Table 1). Other variables which had a low abundance (occurrence on less than five sampling plots) were not considered for the development of the models. However, due to the multi-collinearity of the data we used factor analyses (principle component analyses, PCA, for extraction) to group variables into principal components (PCs; Figs. 6 and 8). We used the Anderson-Rubin-Method to extract PCs which ensured the independence of the extracted variables (PCs). Instead of a high number of simple explanatory variables we used the PCs for the development of the GzLM. By using this statistical method the number of variables was again reduced without losing information. We are aware that this type of statistical analysis makes it difficult to determine the impact of a single measured factor. However, because most variables were multiple correlated and the calculated PCs were often reasonably combined, we decided to use PCA, which gave us the ability to include all relevant measured variables.
215
M. Bobrowski et al. / Forest Ecology and Management 358 (2015) 212–221
Table 1 Measured landscape features and vegetation characteristics. Spearman correlations between browsing damage on Fagus sylvatica, Betula spp. (both HC 1) and explanatory variables are given, with ⁄⁄⁄p < 0.001, ⁄⁄p < 0.01, ⁄p < 0.05. Variables which were included in the model development are indicated by ‘‘x”. Tree species were measured in three different height categories (HC1-3) and categorised into browsed (b) and unbrowsed (ub) specimens. Measured variables
Variables used for model development
Variables used for model development
Spearman’s coefficient
Spearman’s coefficient
Landscape features Distance to settlement (m) Distance to paths and hiking trails (m) Distance to roads for forestry (m) Distance to country roads (m) Distance to raised hides (m) Distance to mineral licks (m)
0.032 0.017 0.032 0.065⁄ 0.028 0.048
Ground cover vegetation Rubus spp. Vaccinum myrtillus
0.061⁄ 0.145⁄⁄
Tree species Betula spp.
Fagus sylvatica
Ilex aquifolium Ilex aquifolium Prunus padus Quercus spp. Sorbus aucuparia Larix decidua Picea abies Pinus sylvestris Pseudotsuga menziesii
HC HC HC HC HC HC HC HC HC HC HC HC HC HC HC HC HC HC HC HC HC
1 2 2 3 3 1 2 2 3 3 1 1 1 2 1 1 1 1 3 1 1
(ub) (ub) (b) (ub) (b) (ub) (ub) (b) (ub) (b) (ub) (b) (ub) (ub) (ub) (ub) (ub) (b) (ub) (ub) (ub)
0.055⁄ 0.030 0.030 0.029 0.030 0.898⁄⁄ 0.308⁄⁄ 0.317⁄⁄ 0.092⁄⁄ 0.098⁄⁄ 0.016 0.020 0.051 0.042 0.028 0.005 0.104⁄⁄ 0.072⁄⁄ 0.098⁄⁄ 0.084⁄⁄ 0.106⁄⁄
Modelling beech damage
X
X X
X X X X X
X X X X X
0.040 0.077⁄⁄ 0.014 0.012 0.003 0.064⁄
Modelling birch damage
X
X
0.090⁄⁄ 0.138⁄⁄
X X
0.938⁄⁄ 0.357⁄⁄ 0.357⁄⁄ 0.188⁄⁄ 0.190⁄⁄ 0.039 0.013 0.018 0.062⁄ 0.045 0.057⁄ 0.051⁄ 0.098⁄⁄ 0.076⁄⁄ 0.028 0.016 0.022 0.036 0.074⁄⁄ 0.023 0.005
X X X X X
X X X X X
X
3. Results 2990 trees were recorded on 1062 out of 1309 sample plots. The number of trees per plot ranged from 0 to 43. Out of all recorded trees 55.45% were coniferous and 44.55% were deciduous species. The most abundant conifer was P. abies (39.93%) and the most abundant deciduous species was F. sylvatica (23.44%). In order to assess the structure of the study area, all recorded trees were classified according to the tree height categories. HC3 trees had the highest relative proportion at 55.35%, followed by HC1 (32.81%) and HC2 (11.84%) (Fig. 2). Of all recorded trees, P. abies HC3 was clearly the most abundant conifer (21.10%) and F. sylvatica HC3 was the most abundant broadleaf (12.37%). Our results show that out of all documented trees, 24.01% showed signs of browsing damage, which represents 27.40% of all sample plots with trees. 52.33% of the deciduous trees were browsed, whereas only 1.27% of conifers showed browsing damage by cervids. Height categorisation of browsed trees provided variable results: HC1 trees had the highest proportion of browsing damage (54.18%), while HC2 trees (19.64%) and HC3 trees (26.18%) had less browsing damage. 97.07% of all browsed trees were deciduous, whereas coniferous trees showed a negligible amount of browsing damage (2.93%). Maps of the spatial distribution of browsing damage and the damage on different height categories are given as Supplementary material. The examination of deciduous tree density in each height category revealed significant differences for all three HCs (Fig. 3). Plots with browsed trees had a higher density of HC1 deciduous trees (Mann–Whitney-U-Test: Z = 23.418; p 6 0.001) than unbrowsed plots. Deciduous HC2 trees (Z = 17.069; p 6 0.001) and deciduous
Fig. 2. Percentage of recorded trees in each height category (HC1 = light grey, HC2 = dark grey, HC3 = black; N = 2990).
HC3 trees (Z = 11.819; p 6 0.001) were also denser on plots where browsing was recorded. The comparison of the density of coniferous trees per height category between plots with and without browsing showed significant differences for all three height categories (Fig. 4). There were higher densities of HC1 conifers (Z = 6.337; p 6 0.001) and HC2 conifers (Z = 4.029; p 6 0.001) on plots with browsed trees.
216
M. Bobrowski et al. / Forest Ecology and Management 358 (2015) 212–221
Fig. 3. Density of deciduous trees according to height categories on plots with (N1 = 291) and without browsed trees (N0 = 1018). Significant differences are marked with ⁄⁄⁄p 6 0.001. Fig. 5. Percentage of browsed trees for each species in each height category (HC1 = light grey, HC2 = dark grey, HC3 = black; N = 718).
However, HC3 coniferous trees were more frequent on plots without browsed trees (Z = 4.759; p 6 0.001). The results of the Chi2 test revealed significant differences between expected and observed values of browsed HC1 trees (X2 = 391.661; df = 7; p 6 0.001; N = 389). For most species the expected value was lower than the observed value (F. sylvaticaobs = 94; F. sylvaticaexp = 39; Betula spp.obs = 199; Betula spp.exp = 95; Quercus spp.obs = 32; Quercus spp.exp = 16; P. padusobs = 15; P. padusexp = 5; S. aucupariaobs = 21; S. aucupariaexp = 18). However, for F. alnus (obs = 3; exp = 3.), I. aquifolium (obs = 9; exp = 12) and coniferous trees (obs = 16; exp = 200) the expected values were equal or higher than the observed values. All HC1 and HC2 deciduous tree species (except I. aquifolium) were more likely to be browsed than HC3 trees (Fig. 5): 54.78% of F. sylvatica showed signs of damage from herbivory, the highest proportion of which were in HC1 (51.82%). 57.60% of all Betula spp. were browsed, mostly in HC1 (57.67%). 3.1. Models for browsing damage on young F. sylvatica (HC1) The pre-examination resulted in significant relationships between 13 variables and browsing damage on F. sylvatica in HC1 (Table 1). These variables were grouped by using PCA to avoid multi-collinearity (Fig. 6) and the extracted factors were used as
covariates in the GzLM (Table 3). No other measured variables (e.g. distance to settlements, feeding damage on I. aquifolium, Table 1) were included in our model because they did not correspond with our statistical procedure described above. The results of the model (Table 2) revealed that PC1 (occurrence F. sylvatica in HC1 ub, HC2 ub and HC3 ub) had the most impact on feeding damage of F. sylvatica, followed by PC3 (occurrence of Betula spp. in HC1 ub, V. myrtillus and the occurrence of P. abies in HC3 (ub), which was negatively related to the feeding damage on F. sylvatica). PC2 (F. sylvatica in HC2 b and HC3 b), PC6 (P. abies in HC1 b and P. sylvestris ub) and PC 5 (Rubus spp.) were less important and PC4 (distance to country roads and L. decidua in HC1 ub) was not significant at all. The findings of our models reveal the importance of F. sylvatica and Betula spp. Specifically, the number of browsed F. sylvatica HC 1 depends on the number of F. sylvatica HC 1 per plot. We found more browsed F. sylvatica HC1, the higher the number of F. sylvatica HC1 per plot (Fig. 7A, rs = 0.898, p 6 0.001). The positive influence of Betula spp. on F. sylvatica trees is made evident by Fig. 7B, which shows that high number of Betula spp. of HC1 lead to high numbers of browsed F. sylvatica per plot (rs = 0.055, p 6 0.05). 3.2. Models for browsing on young Betula spp. (HC1)
Fig. 4. Density of coniferous trees according to height categories on plots with (N1 = 291) and without browsed trees (N0 = 1018). Significant differences are marked with ⁄⁄⁄p 6 0.001.
The pre-examination found significant relationships between 15 variables and browsing damage on Betula spp. in HC1 (Table 1). These were grouped by using PCA to avoid multi-collinearity (Fig. 8) and these factors were used as covariates in GzLM (Table 3). As described above, all other measured variables were excluded from our model. As for the F. sylvatica model, the results of the model for browsing damage on small Betula spp. trees (Table 3) revealed that PC1 (occurrence of Betula spp. in HC1 ub and HC2 b and ub) had the highest impact on feeding damage of Betula spp. in HC1, followed by PC6 (occurrence of V. myrtillus and the occurrence of P. abies in HC3 ub, which is negatively related to the feeding damage on Betula spp.) and PC4 (occurrence of Betula spp. HC3 b, ub and Rubus spp.). The results for the other PCs were not significant (occurrence of P. padus HC1 b, ub; I. aquifolium HC1 b, ub; F. sylvatica HC3 ub and the distance to hiking trails and mineral licks). The results of our models reveal the importance of young Betula spp. Here, the number of browsed Betula spp. HC 1 is positively
M. Bobrowski et al. / Forest Ecology and Management 358 (2015) 212–221
217
Fig. 6. Results of the principal component (PC) analysis summarising explanatory variables for modelling browsing damage on Fagus sylvatica. Results for the PC (eigenvalue, explained variance) as well as factor loadings are given. Only PCs with an eigenvalue >1 are shown. HC = height category, b = browsed, ub = unbrowsed.
Table 2 GzLM for browsing damage on Fagus sylvatica.
***
Browsing on beech HC1
AIC
Omnibus test
Model Constant term PC 1 PC 2 PC 3 PC 4 PC 5 PC 6
849.69
144.70***
Model effects Wald statistics 318.22*** 82.29*** 20.04*** 29.19*** 0.531 13.84*** 17.00***
p < 0.001.
related to the number of Betula spp. HC 1 per plot (Fig. 9; rs = 0.938, p 6 0.001). 4. Discussion Feeding decisions by large, wild herbivores can have an enormous impact on vegetation. In the case of cervids, their feeding decisions might lead to economic losses caused by browsing damage on young trees, which are important for plant recruitment. Therefore, studies like ours, which aimed to identify factors influencing feeding decisions, are likely to be of high value for forest management. Owing to the high population of cervids, especially C. capreolus, in our study area and the aim of forestry to convert coniferous forest into deciduous forest, the recruitment of young trees, particularly of focus species like Fagus and Quercus, are of interest for forest management. We found deer preferred deciduous to coniferous trees as they were far more frequently browsed, although less frequently in availability. Young F. sylvatica and Betula spp. (i.e. lower than 100 cm) were browsed intensively, which contradicts other published findings where Quercus spp. was more intensively used (e.g. Duncan et al., 1998b; Götmark et al., 2005; Chevrier et al., 2012). As browsing on individual plants is influenced by the surrounding environment (Bee et al., 2009), the extremely high browsing damage on F. sylvatica might be related to the absence of other more palatable species (Moser et al., 2006), or the presence of a more palatable tree species (e.g. Betula spp.) in the neighbourhood, which may initially attract deer (Gill, 1992b). Therefore, browsing on individual plants could be density dependent. However, our results show both are true: rare
species (e.g. P. padus) are extensively used in addition to F. sylvatica, the most common species, which is also highly preferred. We focused on two tree species our study: one of large economic value but subject to browsing (F. sylvatica), and one deciduous species with lower economic value (Betula spp.), but with a high potential to serve as a forage plant for cervids, as well as being of potentially high interest as a supplementary food plant for forest management use. To evaluate factors which might influence browsing on these two species we included biotic and abiotic factors into our models. Overall, our study shows that vegetation composition is most important in the selection of a feeding habitat. Not surprisingly, number of browsed trees for both species was positively influenced by the availability of the equivalent species in the lower height classes, i.e. smaller trees up to 100 cm tall. However, and in contrast to the models for browsing on small Betula spp., the results for F. sylvatica also indicate that the occurrence of young Betula spp. (HC1) had a positive impact on browsing pressure on young F. sylvatica in HC1. In contrast, the model for browsing on young Betula spp. (HC1) revealed a solely positive influence of the occurrence of Betula spp. in the lower height classes (HC1 and HC2). For browsing damage on both plant species our results revealed a positive impact of V. myrtillus and a negative impact of the occurrence of old and tall P. abies (HC3). This leads to the assumption that old coniferous tree stands do not serve as feeding habitats, whereas more open young forest stands with a prominent ground cover vegetation of V. myrtillus does. Surprisingly, the models for both tree species show that browsing on F. sylvatica and Betula spp. by cervids is not affected by factors such as landscape features or human disturbance. Instead, feeding decisions are driven by tree species composition, their age (i.e. height class), abundance and the surrounding vegetation. This finding is in agreement with other studies on cervids (e.g. Cervus elaphus and Alces alces; Bee et al., 2009; Stolter, 2015). Our results show that coniferous trees are not a food resource, irrespective of height class, in the investigated area in winter. One explanation for this is food quality. Whereas most deciduous trees are of high digestibility, combined with a moderate amount of secondary plant metabolites, some coniferous trees are known for their constitutive defence and low digestibility (Stolter et al., 2009). For example, P. abies is very common in the study area but often only used when preferred food resources decline (Kamler and Homolka, 2011). Another explanation for the avoidance of coniferous trees is that the diet selection of cervids depends particularly on the surrounding vegetation, i.e. by the relative palatability of surrounding plant species (Gill, 1992a). Therefore,
218
M. Bobrowski et al. / Forest Ecology and Management 358 (2015) 212–221
Fig. 7. Bivariate relationships of browsing on Fagus sylvatica and Betula spp.: (A) Number of Fagus sylvatica HC1 and number of browsed Fagus sylvatica HC1 per plot. (B) Number of Betula spp. HC1 and number of browsed Fagus sylvatica HC1 per plot. (NFagus sylvatica = 239, Nbrowsed = 199, Nmax per plot = 12 and Betula spp.: NBetula spp. = 99, Nbrowsed = 94, Nmax per plot = 11).
Fig. 8. Results of the principal component (PC) analysis summarising explanatory variables for modelling browsing damage on Betula spp. Results for the PC (eigenvalue, explained variance) as well as factor loadings are given. Only PCs with an eigenvalue >1 are shown. HC = height category, b = browsed, ub = unbrowsed.
the low utilisation rate of coniferous trees (e.g. P. abies) indicates a sufficient availability of food of higher quality in the study area, even at the end of winter. The choice of certain food species over others needs to be taken in context with overall food plant availability, which makes the categorisation of ‘‘good” and ‘‘bad” food plants difficult. A food plant of ‘‘low quality” might be relatively ‘‘good” if the surrounding plants are of lower quality for deer (e.g. Sauvé and Côté, 2007).
Deciduous trees were preferentially browsed in deciduous stands with a high abundance of young trees (HC1) and the presence of ground cover vegetation such as V. myrtillus. Such shrubs co-occur with young trees because the absence of a closed canopy provides sufficient light availability for growth. V. myrtillus is known as a good additional food resource for C. elaphus (Morellet and Guibert, 1999), but when the ground cover vegetation is sparse or absent (e.g. because of snow cover), deer shift their diets and
219
M. Bobrowski et al. / Forest Ecology and Management 358 (2015) 212–221 Table 3 GzLM for browsing damage on Betula spp.
* ** ***
Browsing on birch HC1
AIC
Omnibus test
Model Constant term PC 1 PC 2 PC 3 PC 4 PC 5 PC 6 PC 7
401.44
96.40***
Model effects Wald statistics 239.031*** 55.665*** 0.978 0.100 4.563* 1.839 10.096** 1.598
p < 0.05. p < 0.01. p < 0.001.
select even more for young trees, as they are valuable in winter (Heinze et al., 2011). In Germany a common forest management practice is the removal of single stems instead of clear-cutting. This results in a high diversity of age classes, even in small areas, and a large component of mature forest with the almost total loss of early succession stages. Whereas the removal of single stems creates small patches facilitating the regrowth of deciduous trees, which can serve as natural tree nurseries. However, these patches are usually used intensively by cervids because they are the main available food resource in a mature forest stand, and this leads to an enormous conflict between deer food selection and forest management goals. High C. capreolus abundance results in increased browsing on F. sylvatica and other deciduous trees (Heinze et al., 2011). It is assumed that under low deer densities, deciduous trees other than F. sylvatica are preferred and F. sylvatica is even avoided (Heinze et al., 2011). Therefore, incentives to strengthen the regeneration of less economically-valuable supplementary food plant species, such as Betula, Populus or Salix, may reduce browsing pressure on target species such as F. sylvatica or Quercus spp. when they occur in stands without target species (e.g. F. sylvatica), to lure large herbivores away from nurseries or patches with trees of high economical value. This might be realised in small scale early succession stands, which will also provide supplementary food by the occurrence of palatable ground vegetation (e.g. V. myrtillus). These tree species may be beneficial for both managing deer populations and forestry owing to their high ability to survive browsing damage, or even respond to browsing damage with compensatory growth, which in turn might result in a positive feed-back loop for herbivores (e.g. Stolter, 2008 on willows (Salix phylicifolia)). Whereas some deciduous tree species (e.g. Betula pendula and Sorbus aucuparia) are able to tolerate browsing damage to a certain
extent (Hester et al., 2004), other tree species, especially conifers, react to browsing with high lateral branching, slowed growth (Picea abies and Picea sitchensis; Bergquist et al., 2003; Duncan et al., 1998a) and even plant mortality (e.g. P. sylvestris). There are differences in plant responses among species, e.g. F. sylvatica seems to have a lower mortality rate compared to B. pendula and Q. robur (Van Hees et al., 1996). However, the extent of the plant response seems to be related to the severity of the damage (Persson et al., 2005), soil productivity (Edenius et al., 1995; Danell et al., 1997) and prevailing light conditions (Long et al., 2007). Even though contemporary research addresses the importance of deer population density reduction for successful long-term forest regeneration (Welch et al., 1991; Cederlund et al., 1998; Heinze et al., 2011), browsing is crucial for deciduous tree regrowth and high tree species diversity. Therefore, forest management strategies need to incorporate this information to prevent target species from being browsed. C. capreolus and C. elaphus clearly prefer to browse on young deciduous trees so the creation of small areas with early succession stages, which naturally consist of light-demanding species such as Betula, Populus and Salix and abundant understory vegetation (e.g. V. myrtillus, R. fruticosus and R. idaeus), which might lead to the diversion of browsing from targeted species. This could actually decrease the proportion of browsed F. sylvatica in other stands, or even favour the selection of Betula spp. over beech for food, which would consequently lead to increased future F. sylvatica regeneration. Another possibility would be the avoidance of high palatable species (e.g. Betula spp.) in direct neighbourhood to beech or the plantation of low palatable species (e.g. P. abies) in direct neighbourhood to Betula spp. However, this has to be tested. Our findings revealed that more palatable trees (e.g. Betula spp.) may attract animals to feed on less palatable plants (e.g. F. sylvatica), if they are growing in direct neighbourhood. However, they can possibly act as a sink, reducing browsing damage on the focal plant (Gill, 1992a and references therein), if those species are not growing in close vicinity. The influence of vegetation composition on forage selection refers to neighbouring or associational effects (Barbosa et al., 2009). To minimise costs (e.g. planting and fencing trees) this goal could be achieved by establishing compensation areas containing an adequate alternative food supply for cervids, with an extensive ground cover vegetation and trees of low economic value for forestry. 5. Conclusions Our findings provide an overview of the interactions between cervid browsing and young tree regeneration processes and will therefore be useful for forest management. The extent and success
Fig. 9. Number of Betula spp. HC1 and number of browsed Betula spp. HC1 per plot (NBetula
spp.
= 99, Nbrowsed = 94, Nmax
per plot
= 11, Nplots = 42).
220
M. Bobrowski et al. / Forest Ecology and Management 358 (2015) 212–221
of changes in forest regeneration strategies depends largely on forest administration and the impacts of such changes could be substantial. The results of this study improve current knowledge of diet selection of cervids through the investigation of the extent to which the browsing of a target tree species is affected by neighbouring tree species, the ground cover vegetation, human impact and landscape features. In order to analyse the complexities of feeding habitat selection by cervids, we recommend further studies examine factors influencing browsing decisions based on seasonal fluctuations in forage quality and availability, which will ameliorate the regeneration of targeted species. Acknowledgements We thank Hans-Herrmann Engelke and Knut Sierk for their extensive advice and the Forstamt Sellhorn for allowing us to work in the area. We thank the University of Hamburg for funding the fieldwork. We are grateful to James Turner for language editing. We also acknowledge useful comments by two anonymous reviewers. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.foreco.2015.08. 031. These data include Google maps of the most important areas described in this article. References Barbosa, P., Hines, J., Martinson, H., Szczepaniec, A., Szendrei, Z., 2009. Associational resistance and associational susceptibility: having right or wrong neighbors. Annu. Rev. Ecol. Evol. http://dx.doi.org/10.1146/annurev.ecolsys.110308. 120242. Bee, J.N., Tanentzap, A.J., Lee, W.G., Lavers, R.B., Mark, A.F., Mills, J.A., Coomes, D.A., 2009. The benefits of being in a bad neighbourhood: plant community composition influences red deer foraging decisions. Oikos. http://dx.doi.org/ 10.1111/j.1600-0706.2008.16756.x. Bergman, M., Iason, G.R., Hester, A.J., 2005. Feeding patterns by roe deer and rabbits on pine, willow and birch in relation to spatial arrangement. Oikos. http://dx. doi.org/10.1111/j.0030-1299.2005.13794.x. Bergquist, J., Bergström, R., Zakharenka, A., 2003. Responses of young Norway spruce (Picea abies) to winter browsing by roe deer (Capreolus capreolus): effects on height growth and stem morphology. Scand. J. Forest Res. http://dx.doi.org/ 10.1080/0282758031005431. Bjørneraas, K., Herfindal, I., Solberg, E.J., Sæther, B.E., van Moorter, B., Rolandsen, C. M., 2012. Habitat quality influences population distribution, individual space use and functional responses in habitat selection by a large herbivore. Oecologia. http://dx.doi.org/10.1007/s00442-011-2072-3. Bergström, R., Danell, K., 1987. Effects of simulated winter browsing by moose on morphology and biomass of two birch species. J. Ecol. 75, 533–544. Bryant, J.P., Kuropat, P.J., 1980. Selection of winter forage by subarctic browsing vertebrates: the role of plant chemistry. Ann. Rev. Ecol. Sys. http://dx.doi.org/ 10.1146/annurev.es.11.110180.001401. Cederlund, G., Bergqvist, J., Kjellander, P., Gill, R.M.A., Gaillard, J.M., Boisaubert, B., Ballon, P., Duncan, P., 1998. Managing roe deer and their impact on the environment: maximising the net benefits to society. In: Andersen, R., Duncan, P., Linnell, J.D.C. (Eds.), The European Roe Deer: The Biology of Success. Scand UP, Oslo, pp. 337–372. Chevrier, T., Said, S., Hamard, O., Saint-Andrieux, J., Gaillard, J., 2012. The oak browsing index correlates linearly with roe deer density: a new indicator for deer management? Eur. J. Wildl. Res. http://dx.doi.org/10.1007/s10344-0110535-9. Chouinard, A., Filion, L., 2005. Detrimental effects of white tailed deer browsing on balsam fir growth and recruitment in a second-growth stand on Anticosti Island, Quebec. Ecoscience. http://dx.doi.org/10.2980/i1195-6860-12-4-506.1. Côté, D., Rooney, T.P., Temblay, J.P., Dussault, C., Waller, D.M., 2004. Ecological impacts of deer overabundance. Annu. Rev. Ecol. Evol. Syst. http://dx.doi.org/ 10.1146/annurev.ecolsys.35.021103.105725. Cross, P.C., 1998. The Impacts of White-Tailed Deer and Disturbance on Sapling Establishment and Forest Succession in Virginia Undergraduate Thesis. University of Virginia, Charlottesville, Virginia. Danell, K., Haukioja, E., Huss-Danell, K., 1997. Morphological and chemical responses of mountain birch leaves and shoots to winter browsing along a gradient of plant productivity. Ecoscience 4, 296–303. Duncan, A.J., Hartley, S.E., Iason, G.R., 1998a. The effect of previous browsing damage on the morphology and chemical composition of Sitka spruce (Picea
sitchensis) saplings and on their subsequent susceptibility to browsing by red deer (Cervus elaphus). For. Ecol. Manage. http://dx.doi.org/10.1016/S0378-1127 (97)00177-1. Duncan, P., Tixier, H., Hofmann, R.R., Lechner-Doll, M., 1998b. Feeding strategies and the physiology of digestion in roe deer. In: Andersen, R., Duncan, P., Linnell, J.D. C. (Eds.), The European roe Deer: the Biology of Success. Scand UP, Oslo, pp. 91– 116. Dussault, C., Ouellet, J.P., Courtois, R., Huot, J., Breton, L., Jolicoeur, H., 2005. Linking moose habitat selection to limiting factors. Ecography. http://dx.doi.org/ 10.1111/j.2005.0906-7590.04263.x. Edenius, L., Danell, K., Nyquist, H., 1995. Effects of simulated moose browsing on growth, mortality, and fecundity in Scots pine: relations to plant productivity. Can. J. Forest Res. 25 (4), 529–535. Edenius, L., Bergman, M., Ericsson, G., Danell, K., 2002. The role of moose as a disturbance factor in managed boreal forests. Silva Fenn. 36, 57–67. ESRI, 1999. ArcView 3.2: Environmental Systems Research Institute. ArcView GIS: Release 3.2. Redlands, CA: Environmental Systems Research Institute, 1992– 1999. ESRI, 2011. ArcGIS Desktop: Release 10. Redlands, CA: Environmental Systems Research Institute. Ferretti, F., Sforzi, A., Lovari, S., 2008. Intolerance amongst deer species at feeding: roe deer are uneasy banqueters. Behav. Processes. http://dx.doi.org/10.1016/j. beproc.2008.02.008. Gill RMA, 1992a. A Review of Damage by Mammals in North Temperate Forests. 1. Deer. Forestry. http://dx.doi.org/10.1093/forestry/65.2.145. Gill RMA, 1992b. A review of damage by mammals in north temperate forests. 3. Impact on trees and forests. Forestry. http://dx.doi.org/10.1093/forestry/65.4. 363-a. Götmark, F., Paltto, H., Nordén, B., Götmark, E., 2005. Evaluating partial cutting in broadleaved temperate forest under strong experimental control: short-term effects on herbaceous plants. For. Ecol. Manage. http://dx.doi.org/10.1016/ j.foreco.2005.03.052. Hagemoen, R.I.M., Reimers, E., 2002. Reindeer summer activity pattern in relation to weather and insect harassment. J. Anim. Ecol. http://dx.doi.org/10.1046/j.13652656.2002.00654.x. Heinze, E., Boch, S., Fischer, M., Hessenmöller, D., Klenk, B., Müller, J., Prati, D., Schulze, E.D., Seele, C., Socher, S., Halle, S., 2011. Habitat use of large ungulates in northeastern Germany in relation to forest management. For. Ecol. Manage. http://dx.doi.org/10.1016/j.foreco.2010.10.022. Heroldova, M., Homolka, M., Kamler, J.V., 2003. Breakage of rowan caused by red deer – an important factor for Sorbeto-Piceetum stand regeneration? For. Ecol. Manage. http://dx.doi.org/10.1016/S0378-1127(03)00122-1. Hester, A.J., Millard, P., Baillie, G.J., Wendler, R., 2004. How does timing of browsing affect above- and below- ground growth of Betula pendula, Pinus sylvestris and Sorbus aucuparia? Oikos. http://dx.doi.org/10.1111/j.0030-1299.2004.12605.x. Kamler, J., Homolka, M., 2011. Needles in faeces: an index of quality of wild ungulate winter diet. Folia Zool. 60 (1), 63–69. Kuiters, A.T., Mohren, G.M.J., Van Wieren, S.E., 1996. Ungulates in temperate forest ecosystems. For. Ecol. Manage. http://dx.doi.org/10.1016/S0378-1127(96) 03876-5. Kuiters, A.T., Slim, P.A., 2002. Regeneration of mixed deciduous forest in a Dutch forest heathland, following a reduction of ungulate densities. Biol. Conserv. http://dx.doi.org/10.1016/S0006-3207(01)00204-x. Long, Z.T., Pendergast, T.H., Carson, W.P., 2007. The impact of deer on relationships between tree growth and mortality in an old-growth beech-maple forest. For. Ecol. Manage. http://dx.doi.org/10.1016/j.foreco.2007.06.034. Morrison, ML., Marcot, BG., Mannan, RW., 2006. Wildlife-Habitat Relationships: Concepts and Applications. Island Press, Washington, DC. Morellet, N., Guibert, B., 1999. Spatial heterogeneity of winter forest resources used by deer. For. Ecol. Manage. http://dx.doi.org/10.1016/S0378-1127(99)00007-9. Moser, B., Schütz, M., Hindenlang, K.E., 2006. Importance of alternative food resources for browsing by roe deer on deciduous trees: the role of food availability and species quality. For. Ecol. Manage. http://dx.doi.org/10.1016/ j.foreco.2006.01.045. Myking, T., Solberg, EJ., Austrheim, G., Speed, JD., Bøhler, F., Astrup, R., Eriksen, R., 2013. Browsing of sallow (Salix caprea L.) and rowan (Sorbus aucuparia L.) in the context of life history strategies: a literature review. Eur. J. For. Res. http://dx. doi.org/10.2980/20-3-3619. Mysterud, A., Østbye, E., 1999. Cover as a habitat element for temperate ungulates: effects on habitat selection and demography. Wildl. Soc. Bul. 27, 385–394. Newton, A.C., Echeverría, C., Cantarello, E., Bolados, G., 2011. Projecting impacts of human disturbances to inform conservation planning and management in a dryland forest landscape. Biol. Cons. http://dx.doi.org/10.1016/j.biocon. 2011.03.026. NMELV (Niedersächsisches Ministerium für Ernährung, Landwirtschaft und Verbraucherschutz), 2011. Langfristige Ökologische Wald-Entwicklung in Niedersachsen (LÖWE), Niedersächsische Landesforsten. Persson, I.L., Danell, K., Bergström, R., 2005. Different moose densities and accompanied changes in tree morphology and browse production. Ecol. Appl. http://dx.doi.org/10.1890/04-0499. Reimoser, F., Gossow, H., 1996. Impact of ungulates on forest vegetation and its dependence on the silvicultural system. For. Ecol. Manage. http://dx.doi.org/ 10.1016/S0378-1127(96)03816-9. Sauvé, D.G., Côté, S.D., 2007. Winter forage selection in white-tailed deer at high density: balsam fir is the best of a bad choice. J. Wildlife Manage. http://dx.doi. org/10.2193/2006-056.
M. Bobrowski et al. / Forest Ecology and Management 358 (2015) 212–221 Scherer-Lorenzen, M., Körner, C., Schulze, ED., 2005. Forest diversity and function. Ecol Studies, vol. 176. Springer-Verlag, Berlin Heidelberg. SPSS Version 21, 2012. SPSS IBM, New York, USA. Stolter, C., Ball, J.P., Julkunen-Tiitto, R., Lieberei, R., Ganzhorn, J.U., 2005. Winter browsing of moose (Alces alces) on two different willow species: food selection in relation to plant chemistry and plant response. Can. J. Zool. http://dx.doi.org/ 10.1139/z05-077. Stolter, C., 2008. Intra-individual plant response due to moose browsing: feed-back loops and impacts on multiple consumers. Ecol. Monogr. http://dx.doi.org/ 10.1890/07-0401.1. Stolter, C., Niemelä, P., Varvikko, T., Julkunen-Tiitto, R., Ball, J.P., Vanhatalo, A., Danell, K., Ganzhorn, J.U., 2009. Comparison of secondary chemistry and digestibility of three different boreal coniferous trees. Basic Appl. Ecol. http:// dx.doi.org/10.1016/j.baae.2007.12.001. Stolter, C., Ball, J.P., Julkunen-Tiitto, R., 2013. Seasonal differences in the relative importance of specific phenolics and twig morphology result in contrasting patterns of foraging by a generalist herbivore. Can. J. Zool. http://dx.doi.org/ 10.1139/cjz-2012-0270. Stolter, C., 2015. More Than Meets the Eye: Patch Quality Explain Differences in Utilization by a Large Herbivore, the Moose, manuscript in preparation.
221
Van Hees, A.F.M., Kuiters, A.T., Slim, P.A., 1996. Growth and development of silver birch, pedunculate oak and beech as affected by deer browsing. For. Ecol. Manage. 88 (1), 55–63. Verheyden-Tixier, H., Duncan, P., Ballon, P., Guillon, N., 1998. Selection of hardwood saplings by European roe deer: effects of variation in the availability of palatable species and of understory vegetation. Rev. Ecol. Terre Vie. 53, 245– 253. Vila, B., Torre, F., Martin, J.L., Guibal, F., 2003. Response of young Tsuga heterophylla to deer browsing: developing tools to assess deer impact on forest dynamics. Trees Struct. Funct. http://dx.doi.org/10.1007/s00468-003-0272-0. Welch, D., Staines, B.W., Scott, D., French, D.D., Catt, D.C., 1991. Leader browsing by red and roe deer on young sitka spruce trees in western Scotland. I. Damage rates and the influence of habitat factors. Forestry. http://dx.doi.org/ 10.1093/forestry/64.1.61. Zerbe, S., 2002. Restoration of natural broad-leaved woodlands in Central Europe on sites with coniferous forest plantations. For. Ecol. Manage. http://dx.doi.org/ 10.1016/S0378-1127(01)00686-7. Zuur, A.F., Ieno, E.N., Elphick, C.S., 2010. A protocol for data exploration to avoid common statistical problems. Methods Ecol. Evol. http://dx.doi.org/10.1111/ j.2041-210X.2009.00001.x.
Contents lists available at ScienceDirect
Forest Ecology and Management journal homepage: www.elsevier.com/locate/foreco
Modelling browsing of deer on beech and birch in northern Germany Maria Bobrowski ⇑, Benjamin Gillich, Caroline Stolter Department of Animal Ecology and Conservation, University of Hamburg, Hamburg, Germany
a r t i c l e
i n f o
Article history: Received 26 January 2015 Received in revised form 24 August 2015 Accepted 26 August 2015
Keywords: Large ungulates Browsing damage Feeding habitat selection Vegetation composition Fagus sylvatica Betula spp.
a b s t r a c t Food selection by cervids is of high interest for forest management as they tend to browse on trees, causing severe damage and shifts in the regeneration capacity of certain preferred tree species. This behaviour could contradict forestry management strategies as deer may browse on target species, e.g. beech or oak, which are of high economic value for forestry and nature conservation. The choice of feeding habitat is influenced by a variety of different parameters, ranging from forest characteristics to human impacts. Therefore, we focused on factors influencing browsing of roe deer (Capreolus capreolus) and red deer (Cervus elaphus) on two different plant species: beech (Fagus sylvatica), which is an important species for forestry, and birch (Betula spp.), which may be an alternative food resource to beech. We measured browsing damage in relation to, variables such as tree density, tree height, ground cover vegetation, possible human impacts and landscape features in the Lüneburger Heide, Germany, for our modelling approach. As expected, our results revealed that browsing on beech is positively influenced by the availability of young beech and surprisingly by birch trees, whereas browsing on birch was influenced by the availability of birch trees only. Furthermore, the occurrence of blueberry (Vaccinium myrtillus) was positively and significantly related to browsing damage on both plant species, but a negative relationship was found for old Norway spruce stands (Picea abies). Surprisingly, other tree species, landscape features and human impacts had no significant influence on the feeding damage of either tree species. Our results indicate that the availability of stands exclusively consisting of birch (preferably young birch) might decrease the proportion of browsed beeches in neighbouring stands, or even favour the selection of birch over beech for food, which would consequently lead to increased future beech regeneration. Our modelling approach is applicable to forest management, providing an overview of interactions between browsing by cervids and young tree regeneration processes. Ó 2015 Elsevier B.V. All rights reserved.
1. Introduction The regeneration of natural woodland has become a major priority of sustainable forestry management globally over the last few decades and this is especially true in Europe (Zerbe, 2002; NMELV, 2011). Nowadays, nature conversion issues in Germany and other north-western European countries are focused on changing contemporary, monocultural coniferous forest stands into mixed stands with indigenous, deciduous tree species (Kuiters and Slim, 2002; Scherer-Lorenzen et al., 2005; Heinze et al., 2011). In our study area, the aim of modern and sustainable forest management is therefore the alteration of these coniferous monocultures into birch-oak (Betulo-Quercetum) and beech-oak (Fago-Quercetum) forest communities (Kuiters and Slim, 2002). Browsing by large herbivores on tree seedlings and saplings is considered a problem for forest regeneration worldwide ⇑ Corresponding author. E-mail address: [email protected] (M. Bobrowski). http://dx.doi.org/10.1016/j.foreco.2015.08.031 0378-1127/Ó 2015 Elsevier B.V. All rights reserved.
(Chouinard and Filion, 2005; Côté et al., 2004; Moser et al., 2006). Browsing pressure by cervids is considered an influential factor in the success of forest regeneration and has been well investigated (Gill, 1992a; Reimoser and Gossow, 1996; Kuiters et al., 1996). Many studies assume that continuous, selective browsing by cervids has a retarding effect on shrub and tree growth and negatively affects natural regeneration in forests throughout Europe (Gill, 1992a; Kuiters and Slim, 2002; Moser et al., 2006). Cervids can influence vegetation regeneration in various ways (Gill, 1992a): e.g. browsing may inhibit sapling growth rate and survival, leading to a modification of species composition in the understory and tree layer (Edenius et al., 2002). Furthermore, feeding behaviour may change tree morphology (e.g. enhanced or simplified branching pattern, compensation or slowed growth, e.g. Bergström and Danell, 1987; Bergquist et al., 2003; Hester et al., 2004; Stolter, 2008; Myking et al., 2013), resulting in reduced timber quality (Vila et al., 2003). The diet selection of cervids is influenced by food quality and quantity (Hagemoen and Reimers, 2002; Dussault et al., 2005;
M. Bobrowski et al. / Forest Ecology and Management 358 (2015) 212–221
Stolter et al., 2005, 2013; Ferretti et al., 2008; Bjørneraas et al., 2012). However, food quality is related to vegetation composition, which depends not only on abiotic factors (e.g. soil parameters, climate), but also on forest management plans (Morrison et al., 2006). Factors like the spatial arrangement of plant species (Bergman et al., 2005), cover and visibility (Mysterud and Østbye, 1999) and human impact (e.g. hunting strategies and distance to human settlements (Dussault et al., 2005; Newton et al., 2011)) may have an important role in forage selection by cervids. Therefore, the probability that a particular tree will be browsed may not only depend on its own characteristics, such as chemistry and morphology (Stolter et al., 2013), but also on the relative quality and abundance of surrounding vegetation (e.g. herbs and shrubs) (Gill, 1992b; Verheyden-Tixier et al., 1998) and the frequency of disturbance (Cross, 1998). Bee et al. (2009) showed that browsing on individual plant species is influenced by the availability of alternative forage plants in the surrounding environment. They suggested that browsing by deer on a focal plant is simply related to the lower palatability of the surrounding vegetation (Bee et al., 2009). Furthermore, plants with a lower nutritional quality e.g. lower concentration of plant secondary metabolites often received less damage than those of higher quality (e.g. Stolter et al., 2013). Many studies suggest that the availability of diverse food resources is a key factor for the amount of feeding damage on trees (Welch et al., 1991; Gill, 1992b; Vila et al., 2003; Moser et al., 2006). In this study we chose beech (Fagus sylvatica) as the most important forest regeneration species in our study area and birch (Betula spp.) as a species which might be of interest as an alternative food source because it is a common food plant for cervids, but is irrelevant for forestry. Improved knowledge about factors influencing browsing behaviour of both deer species on these trees is likely beneficial for the development of forest regeneration management strategies and could help minimise browsing damage on forest target species like F. sylvatica. Additionally, it may help identify management alternatives, such as enhancing the browsing damage on non-target species like Betula spp. Therefore, we addressed the following research questions:
213
2. Materials and methods 2.1. Study sites The study was conducted at Wilsede, Forestry Department Sellhorn (5390 N, 9590 E, 100 m above sea level), located southeast in the conservation area ‘‘Lüneburger Heide‘‘, Lower Saxony, Germany. The study area is characterised by a yearly precipitation of 800 mm and a mean annual temperature of 8C. Most of the 1240 ha study site is forested; the vegetation composition is dominated by coniferous trees, planted in monocultures over 200 years ago. Dominant overstory tree species are Scots pine (Pinus sylvestris), Norway spruce (Picea abies), Douglas fir (Pseudotsuga menziesii) and European larch (Larix decidua). The existing deciduous forest stands, such as beech (Fagus sylvatica), birch (Betula spp.) and oak (Quercus spp.), are relatively small and often mixed with other deciduous trees (alder buckthorn (Frangula alnus), hackberry (Prunus padus), holly (Ilex aquifolium), rowan (Sorbus aucuparia) and willow (Salix spp.)). Dominant ground cover vegetation species include blueberry (Vaccinium myrtillus), blackberry (Rubus fruticosus), raspberry (Rubus idaeus) and fern (Pteridium aquilinum). The nature conservation area ‘‘Lüneburger Heide‘‘ has ideal local site conditions; different forest succession stages (including early succession stages) and free-ranging red deer (C. elaphus) can be found here. This is relatively unique in Germany because red deer distribution ranges are usually restricted to declared ‘‘red deer areas”. The dominant large herbivores in the site are roe deer (C. capreolus) at an estimated density of 20 individuals per 100 ha. Red deer density is estimated at 0.2 individuals/ha and the sex ratio is 1:1 (K. Sierk, pers. comm.). At the time of the study we had no information on large predator presence in the study area, although wolves have been observed in other parts of the Lüneburger Heide since 2007.
2.2. Data collection (1) Which tree species is utilised the most by cervids in relation to tree abundance and height? (2) What are the factors influencing browsing on F. sylvatica and Betula spp. saplings in relation to surrounding vegetation, geographical features and human impact? (3) Are there differences in the influence of these factors on browsing between F. sylvatica and Betula spp.? In detail, we hypothesised that: (a) Deciduous trees are more used than coniferous trees by red deer (Cervus elaphus) and roe deer (Capreolus capreolus), due to their higher palatability and digestibility (i.e. plant tissues tend to contain lower amounts of secondary metabolites) (Bryant and Kuropat, 1980). (b) Smaller trees are more frequently browsed than larger trees as they are more accessible. (c) Presence of ground cover vegetation such as blueberry (Vaccinium myrtillus), blackberry (Rubus fruticosus), raspberry (Rubus idaeus) favours browsing on young deciduous trees, as they provide additional food resources, which might attract animals. (d) Geographical features influence feeding habitat selection by cervids. For example, human settlements, different road types (paths, hiking trails, roads for forestry, country roads and highways) and raised hides have a negative impact on the feeding habitat choice of the animals, while mineral licks have a positive impact.
Fieldwork was conducted between February and March 2012. A systematic design was used for data collection. Using ET Vector Grid in ArcView (Version 3.2; 1999), the position of the sampling grid was randomly selected and the sampling plots were regularly distributed 100 m apart (Fig. 1). Each intersection point represented a 2 2 m sampling plot. A total of 1309 plots were investigated covering 5236 m2. Within each sampling plot (1) tree species, (2) tree height, (3) browsing damage and height and (4) ground cover vegetation were recorded. We restricted our recordings to the ten most common tree species (Stolter, unpublished data). To avoid identification errors of winter-dormant trees, Betula pubescens and Betula pendula were not separated and were treated as Betula spp. Rubus fructicosus and Rubus idaeus were also not seperated and were treated as Rubus spp. Tree height categories (HC) were recorded to determine vegetation composition and occurrence of tree browsing since the animals forage preferentially on tree species up to 150 cm (C. elaphus) and 75 cm (C. capreolus) in height (Heroldova et al., 2003). All trees were classified at height categories: HC1: 40–100 cm, HC2: 101–150 cm and HC3: >150 cm. Tree heights under 40 cm were commonly browsed by rodents and lagomorphs and were therefore excluded. We estimated feeding damage in each 2 2 m plot by counting browsed species in each of the three height categories. We recorded overall browsing damage and divided trees into two further categories: browsed (b) and unbrowsed (ub). This categorisation was less subjective than assessing browsing scores for tree species and was conducted to minimise observer bias.
214
M. Bobrowski et al. / Forest Ecology and Management 358 (2015) 212–221
Fig. 1. Sampling grid of the study area in Forestry Department Sellhorn (5390 N, 9590 E, 100 m a.s.l.), ‘‘Lüneburger Heide‘‘, Lower Saxony, Germany (dots indicate sampling plots, N = 1309).
Furthermore, this method was less time-consuming than quantifying browsing damage by counting twigs (e.g. Stolter, 2008). We used the function ‘‘near table” in ArcGIS (Version 10.0; 2011) to calculate distances to human settlements and different road types (paths, hiking trails, roads for forestry, country roads and highways), as well as to raised hides (N = 176) and mineral licks (N = 46), to analyse their impacts on the browsing damage caused by deer. 2.3. Statistical analyses All statistical analyses were performed in SPSS Version 21 (2012). Data were tested for deviation from normality using the Kolmogorov–Smirnov test. The comparison of plots with browsing (N1 = 291) and plots without browsing (N0 = 1018) was performed with a Mann–Whitney-U-Test. Food species preference of cervids was evaluated using Chi2 statistics, in order to determine whether there was a significant difference between the expected frequencies and the observed frequencies of browsing between tree species of HC1. Expected values were calculated considering the differences in availability of the different plant species. A Generalized Linear Model (GzLM) was used to analyse the importance of different factors on feeding habitat selection of roe and red deer. Because of the distribution of the data (high number of zero values) we used Tweedie-distribution for modelling. Browsing damage on F. sylvatica (target species for forestry) and Betula spp.
(additional food resource) in HC1 were used as dependent variables. We included vegetation characteristics, browsing damage on all tree species and geographical features as potential covariates (Table 1). To reduce the number of variables we pre-examined our covariates for relationships with our dependent variables using Spearman correlations and scatter plots (Zuur et al., 2010; see also Stolter et al., 2013 for details). Thus, only variables that showed a significant relationship to damage on Betula spp. (HC1) or F. sylvatica (HC1) were used for the development of the models (Table 1). Other variables which had a low abundance (occurrence on less than five sampling plots) were not considered for the development of the models. However, due to the multi-collinearity of the data we used factor analyses (principle component analyses, PCA, for extraction) to group variables into principal components (PCs; Figs. 6 and 8). We used the Anderson-Rubin-Method to extract PCs which ensured the independence of the extracted variables (PCs). Instead of a high number of simple explanatory variables we used the PCs for the development of the GzLM. By using this statistical method the number of variables was again reduced without losing information. We are aware that this type of statistical analysis makes it difficult to determine the impact of a single measured factor. However, because most variables were multiple correlated and the calculated PCs were often reasonably combined, we decided to use PCA, which gave us the ability to include all relevant measured variables.
215
M. Bobrowski et al. / Forest Ecology and Management 358 (2015) 212–221
Table 1 Measured landscape features and vegetation characteristics. Spearman correlations between browsing damage on Fagus sylvatica, Betula spp. (both HC 1) and explanatory variables are given, with ⁄⁄⁄p < 0.001, ⁄⁄p < 0.01, ⁄p < 0.05. Variables which were included in the model development are indicated by ‘‘x”. Tree species were measured in three different height categories (HC1-3) and categorised into browsed (b) and unbrowsed (ub) specimens. Measured variables
Variables used for model development
Variables used for model development
Spearman’s coefficient
Spearman’s coefficient
Landscape features Distance to settlement (m) Distance to paths and hiking trails (m) Distance to roads for forestry (m) Distance to country roads (m) Distance to raised hides (m) Distance to mineral licks (m)
0.032 0.017 0.032 0.065⁄ 0.028 0.048
Ground cover vegetation Rubus spp. Vaccinum myrtillus
0.061⁄ 0.145⁄⁄
Tree species Betula spp.
Fagus sylvatica
Ilex aquifolium Ilex aquifolium Prunus padus Quercus spp. Sorbus aucuparia Larix decidua Picea abies Pinus sylvestris Pseudotsuga menziesii
HC HC HC HC HC HC HC HC HC HC HC HC HC HC HC HC HC HC HC HC HC
1 2 2 3 3 1 2 2 3 3 1 1 1 2 1 1 1 1 3 1 1
(ub) (ub) (b) (ub) (b) (ub) (ub) (b) (ub) (b) (ub) (b) (ub) (ub) (ub) (ub) (ub) (b) (ub) (ub) (ub)
0.055⁄ 0.030 0.030 0.029 0.030 0.898⁄⁄ 0.308⁄⁄ 0.317⁄⁄ 0.092⁄⁄ 0.098⁄⁄ 0.016 0.020 0.051 0.042 0.028 0.005 0.104⁄⁄ 0.072⁄⁄ 0.098⁄⁄ 0.084⁄⁄ 0.106⁄⁄
Modelling beech damage
X
X X
X X X X X
X X X X X
0.040 0.077⁄⁄ 0.014 0.012 0.003 0.064⁄
Modelling birch damage
X
X
0.090⁄⁄ 0.138⁄⁄
X X
0.938⁄⁄ 0.357⁄⁄ 0.357⁄⁄ 0.188⁄⁄ 0.190⁄⁄ 0.039 0.013 0.018 0.062⁄ 0.045 0.057⁄ 0.051⁄ 0.098⁄⁄ 0.076⁄⁄ 0.028 0.016 0.022 0.036 0.074⁄⁄ 0.023 0.005
X X X X X
X X X X X
X
3. Results 2990 trees were recorded on 1062 out of 1309 sample plots. The number of trees per plot ranged from 0 to 43. Out of all recorded trees 55.45% were coniferous and 44.55% were deciduous species. The most abundant conifer was P. abies (39.93%) and the most abundant deciduous species was F. sylvatica (23.44%). In order to assess the structure of the study area, all recorded trees were classified according to the tree height categories. HC3 trees had the highest relative proportion at 55.35%, followed by HC1 (32.81%) and HC2 (11.84%) (Fig. 2). Of all recorded trees, P. abies HC3 was clearly the most abundant conifer (21.10%) and F. sylvatica HC3 was the most abundant broadleaf (12.37%). Our results show that out of all documented trees, 24.01% showed signs of browsing damage, which represents 27.40% of all sample plots with trees. 52.33% of the deciduous trees were browsed, whereas only 1.27% of conifers showed browsing damage by cervids. Height categorisation of browsed trees provided variable results: HC1 trees had the highest proportion of browsing damage (54.18%), while HC2 trees (19.64%) and HC3 trees (26.18%) had less browsing damage. 97.07% of all browsed trees were deciduous, whereas coniferous trees showed a negligible amount of browsing damage (2.93%). Maps of the spatial distribution of browsing damage and the damage on different height categories are given as Supplementary material. The examination of deciduous tree density in each height category revealed significant differences for all three HCs (Fig. 3). Plots with browsed trees had a higher density of HC1 deciduous trees (Mann–Whitney-U-Test: Z = 23.418; p 6 0.001) than unbrowsed plots. Deciduous HC2 trees (Z = 17.069; p 6 0.001) and deciduous
Fig. 2. Percentage of recorded trees in each height category (HC1 = light grey, HC2 = dark grey, HC3 = black; N = 2990).
HC3 trees (Z = 11.819; p 6 0.001) were also denser on plots where browsing was recorded. The comparison of the density of coniferous trees per height category between plots with and without browsing showed significant differences for all three height categories (Fig. 4). There were higher densities of HC1 conifers (Z = 6.337; p 6 0.001) and HC2 conifers (Z = 4.029; p 6 0.001) on plots with browsed trees.
216
M. Bobrowski et al. / Forest Ecology and Management 358 (2015) 212–221
Fig. 3. Density of deciduous trees according to height categories on plots with (N1 = 291) and without browsed trees (N0 = 1018). Significant differences are marked with ⁄⁄⁄p 6 0.001. Fig. 5. Percentage of browsed trees for each species in each height category (HC1 = light grey, HC2 = dark grey, HC3 = black; N = 718).
However, HC3 coniferous trees were more frequent on plots without browsed trees (Z = 4.759; p 6 0.001). The results of the Chi2 test revealed significant differences between expected and observed values of browsed HC1 trees (X2 = 391.661; df = 7; p 6 0.001; N = 389). For most species the expected value was lower than the observed value (F. sylvaticaobs = 94; F. sylvaticaexp = 39; Betula spp.obs = 199; Betula spp.exp = 95; Quercus spp.obs = 32; Quercus spp.exp = 16; P. padusobs = 15; P. padusexp = 5; S. aucupariaobs = 21; S. aucupariaexp = 18). However, for F. alnus (obs = 3; exp = 3.), I. aquifolium (obs = 9; exp = 12) and coniferous trees (obs = 16; exp = 200) the expected values were equal or higher than the observed values. All HC1 and HC2 deciduous tree species (except I. aquifolium) were more likely to be browsed than HC3 trees (Fig. 5): 54.78% of F. sylvatica showed signs of damage from herbivory, the highest proportion of which were in HC1 (51.82%). 57.60% of all Betula spp. were browsed, mostly in HC1 (57.67%). 3.1. Models for browsing damage on young F. sylvatica (HC1) The pre-examination resulted in significant relationships between 13 variables and browsing damage on F. sylvatica in HC1 (Table 1). These variables were grouped by using PCA to avoid multi-collinearity (Fig. 6) and the extracted factors were used as
covariates in the GzLM (Table 3). No other measured variables (e.g. distance to settlements, feeding damage on I. aquifolium, Table 1) were included in our model because they did not correspond with our statistical procedure described above. The results of the model (Table 2) revealed that PC1 (occurrence F. sylvatica in HC1 ub, HC2 ub and HC3 ub) had the most impact on feeding damage of F. sylvatica, followed by PC3 (occurrence of Betula spp. in HC1 ub, V. myrtillus and the occurrence of P. abies in HC3 (ub), which was negatively related to the feeding damage on F. sylvatica). PC2 (F. sylvatica in HC2 b and HC3 b), PC6 (P. abies in HC1 b and P. sylvestris ub) and PC 5 (Rubus spp.) were less important and PC4 (distance to country roads and L. decidua in HC1 ub) was not significant at all. The findings of our models reveal the importance of F. sylvatica and Betula spp. Specifically, the number of browsed F. sylvatica HC 1 depends on the number of F. sylvatica HC 1 per plot. We found more browsed F. sylvatica HC1, the higher the number of F. sylvatica HC1 per plot (Fig. 7A, rs = 0.898, p 6 0.001). The positive influence of Betula spp. on F. sylvatica trees is made evident by Fig. 7B, which shows that high number of Betula spp. of HC1 lead to high numbers of browsed F. sylvatica per plot (rs = 0.055, p 6 0.05). 3.2. Models for browsing on young Betula spp. (HC1)
Fig. 4. Density of coniferous trees according to height categories on plots with (N1 = 291) and without browsed trees (N0 = 1018). Significant differences are marked with ⁄⁄⁄p 6 0.001.
The pre-examination found significant relationships between 15 variables and browsing damage on Betula spp. in HC1 (Table 1). These were grouped by using PCA to avoid multi-collinearity (Fig. 8) and these factors were used as covariates in GzLM (Table 3). As described above, all other measured variables were excluded from our model. As for the F. sylvatica model, the results of the model for browsing damage on small Betula spp. trees (Table 3) revealed that PC1 (occurrence of Betula spp. in HC1 ub and HC2 b and ub) had the highest impact on feeding damage of Betula spp. in HC1, followed by PC6 (occurrence of V. myrtillus and the occurrence of P. abies in HC3 ub, which is negatively related to the feeding damage on Betula spp.) and PC4 (occurrence of Betula spp. HC3 b, ub and Rubus spp.). The results for the other PCs were not significant (occurrence of P. padus HC1 b, ub; I. aquifolium HC1 b, ub; F. sylvatica HC3 ub and the distance to hiking trails and mineral licks). The results of our models reveal the importance of young Betula spp. Here, the number of browsed Betula spp. HC 1 is positively
M. Bobrowski et al. / Forest Ecology and Management 358 (2015) 212–221
217
Fig. 6. Results of the principal component (PC) analysis summarising explanatory variables for modelling browsing damage on Fagus sylvatica. Results for the PC (eigenvalue, explained variance) as well as factor loadings are given. Only PCs with an eigenvalue >1 are shown. HC = height category, b = browsed, ub = unbrowsed.
Table 2 GzLM for browsing damage on Fagus sylvatica.
***
Browsing on beech HC1
AIC
Omnibus test
Model Constant term PC 1 PC 2 PC 3 PC 4 PC 5 PC 6
849.69
144.70***
Model effects Wald statistics 318.22*** 82.29*** 20.04*** 29.19*** 0.531 13.84*** 17.00***
p < 0.001.
related to the number of Betula spp. HC 1 per plot (Fig. 9; rs = 0.938, p 6 0.001). 4. Discussion Feeding decisions by large, wild herbivores can have an enormous impact on vegetation. In the case of cervids, their feeding decisions might lead to economic losses caused by browsing damage on young trees, which are important for plant recruitment. Therefore, studies like ours, which aimed to identify factors influencing feeding decisions, are likely to be of high value for forest management. Owing to the high population of cervids, especially C. capreolus, in our study area and the aim of forestry to convert coniferous forest into deciduous forest, the recruitment of young trees, particularly of focus species like Fagus and Quercus, are of interest for forest management. We found deer preferred deciduous to coniferous trees as they were far more frequently browsed, although less frequently in availability. Young F. sylvatica and Betula spp. (i.e. lower than 100 cm) were browsed intensively, which contradicts other published findings where Quercus spp. was more intensively used (e.g. Duncan et al., 1998b; Götmark et al., 2005; Chevrier et al., 2012). As browsing on individual plants is influenced by the surrounding environment (Bee et al., 2009), the extremely high browsing damage on F. sylvatica might be related to the absence of other more palatable species (Moser et al., 2006), or the presence of a more palatable tree species (e.g. Betula spp.) in the neighbourhood, which may initially attract deer (Gill, 1992b). Therefore, browsing on individual plants could be density dependent. However, our results show both are true: rare
species (e.g. P. padus) are extensively used in addition to F. sylvatica, the most common species, which is also highly preferred. We focused on two tree species our study: one of large economic value but subject to browsing (F. sylvatica), and one deciduous species with lower economic value (Betula spp.), but with a high potential to serve as a forage plant for cervids, as well as being of potentially high interest as a supplementary food plant for forest management use. To evaluate factors which might influence browsing on these two species we included biotic and abiotic factors into our models. Overall, our study shows that vegetation composition is most important in the selection of a feeding habitat. Not surprisingly, number of browsed trees for both species was positively influenced by the availability of the equivalent species in the lower height classes, i.e. smaller trees up to 100 cm tall. However, and in contrast to the models for browsing on small Betula spp., the results for F. sylvatica also indicate that the occurrence of young Betula spp. (HC1) had a positive impact on browsing pressure on young F. sylvatica in HC1. In contrast, the model for browsing on young Betula spp. (HC1) revealed a solely positive influence of the occurrence of Betula spp. in the lower height classes (HC1 and HC2). For browsing damage on both plant species our results revealed a positive impact of V. myrtillus and a negative impact of the occurrence of old and tall P. abies (HC3). This leads to the assumption that old coniferous tree stands do not serve as feeding habitats, whereas more open young forest stands with a prominent ground cover vegetation of V. myrtillus does. Surprisingly, the models for both tree species show that browsing on F. sylvatica and Betula spp. by cervids is not affected by factors such as landscape features or human disturbance. Instead, feeding decisions are driven by tree species composition, their age (i.e. height class), abundance and the surrounding vegetation. This finding is in agreement with other studies on cervids (e.g. Cervus elaphus and Alces alces; Bee et al., 2009; Stolter, 2015). Our results show that coniferous trees are not a food resource, irrespective of height class, in the investigated area in winter. One explanation for this is food quality. Whereas most deciduous trees are of high digestibility, combined with a moderate amount of secondary plant metabolites, some coniferous trees are known for their constitutive defence and low digestibility (Stolter et al., 2009). For example, P. abies is very common in the study area but often only used when preferred food resources decline (Kamler and Homolka, 2011). Another explanation for the avoidance of coniferous trees is that the diet selection of cervids depends particularly on the surrounding vegetation, i.e. by the relative palatability of surrounding plant species (Gill, 1992a). Therefore,
218
M. Bobrowski et al. / Forest Ecology and Management 358 (2015) 212–221
Fig. 7. Bivariate relationships of browsing on Fagus sylvatica and Betula spp.: (A) Number of Fagus sylvatica HC1 and number of browsed Fagus sylvatica HC1 per plot. (B) Number of Betula spp. HC1 and number of browsed Fagus sylvatica HC1 per plot. (NFagus sylvatica = 239, Nbrowsed = 199, Nmax per plot = 12 and Betula spp.: NBetula spp. = 99, Nbrowsed = 94, Nmax per plot = 11).
Fig. 8. Results of the principal component (PC) analysis summarising explanatory variables for modelling browsing damage on Betula spp. Results for the PC (eigenvalue, explained variance) as well as factor loadings are given. Only PCs with an eigenvalue >1 are shown. HC = height category, b = browsed, ub = unbrowsed.
the low utilisation rate of coniferous trees (e.g. P. abies) indicates a sufficient availability of food of higher quality in the study area, even at the end of winter. The choice of certain food species over others needs to be taken in context with overall food plant availability, which makes the categorisation of ‘‘good” and ‘‘bad” food plants difficult. A food plant of ‘‘low quality” might be relatively ‘‘good” if the surrounding plants are of lower quality for deer (e.g. Sauvé and Côté, 2007).
Deciduous trees were preferentially browsed in deciduous stands with a high abundance of young trees (HC1) and the presence of ground cover vegetation such as V. myrtillus. Such shrubs co-occur with young trees because the absence of a closed canopy provides sufficient light availability for growth. V. myrtillus is known as a good additional food resource for C. elaphus (Morellet and Guibert, 1999), but when the ground cover vegetation is sparse or absent (e.g. because of snow cover), deer shift their diets and
219
M. Bobrowski et al. / Forest Ecology and Management 358 (2015) 212–221 Table 3 GzLM for browsing damage on Betula spp.
* ** ***
Browsing on birch HC1
AIC
Omnibus test
Model Constant term PC 1 PC 2 PC 3 PC 4 PC 5 PC 6 PC 7
401.44
96.40***
Model effects Wald statistics 239.031*** 55.665*** 0.978 0.100 4.563* 1.839 10.096** 1.598
p < 0.05. p < 0.01. p < 0.001.
select even more for young trees, as they are valuable in winter (Heinze et al., 2011). In Germany a common forest management practice is the removal of single stems instead of clear-cutting. This results in a high diversity of age classes, even in small areas, and a large component of mature forest with the almost total loss of early succession stages. Whereas the removal of single stems creates small patches facilitating the regrowth of deciduous trees, which can serve as natural tree nurseries. However, these patches are usually used intensively by cervids because they are the main available food resource in a mature forest stand, and this leads to an enormous conflict between deer food selection and forest management goals. High C. capreolus abundance results in increased browsing on F. sylvatica and other deciduous trees (Heinze et al., 2011). It is assumed that under low deer densities, deciduous trees other than F. sylvatica are preferred and F. sylvatica is even avoided (Heinze et al., 2011). Therefore, incentives to strengthen the regeneration of less economically-valuable supplementary food plant species, such as Betula, Populus or Salix, may reduce browsing pressure on target species such as F. sylvatica or Quercus spp. when they occur in stands without target species (e.g. F. sylvatica), to lure large herbivores away from nurseries or patches with trees of high economical value. This might be realised in small scale early succession stands, which will also provide supplementary food by the occurrence of palatable ground vegetation (e.g. V. myrtillus). These tree species may be beneficial for both managing deer populations and forestry owing to their high ability to survive browsing damage, or even respond to browsing damage with compensatory growth, which in turn might result in a positive feed-back loop for herbivores (e.g. Stolter, 2008 on willows (Salix phylicifolia)). Whereas some deciduous tree species (e.g. Betula pendula and Sorbus aucuparia) are able to tolerate browsing damage to a certain
extent (Hester et al., 2004), other tree species, especially conifers, react to browsing with high lateral branching, slowed growth (Picea abies and Picea sitchensis; Bergquist et al., 2003; Duncan et al., 1998a) and even plant mortality (e.g. P. sylvestris). There are differences in plant responses among species, e.g. F. sylvatica seems to have a lower mortality rate compared to B. pendula and Q. robur (Van Hees et al., 1996). However, the extent of the plant response seems to be related to the severity of the damage (Persson et al., 2005), soil productivity (Edenius et al., 1995; Danell et al., 1997) and prevailing light conditions (Long et al., 2007). Even though contemporary research addresses the importance of deer population density reduction for successful long-term forest regeneration (Welch et al., 1991; Cederlund et al., 1998; Heinze et al., 2011), browsing is crucial for deciduous tree regrowth and high tree species diversity. Therefore, forest management strategies need to incorporate this information to prevent target species from being browsed. C. capreolus and C. elaphus clearly prefer to browse on young deciduous trees so the creation of small areas with early succession stages, which naturally consist of light-demanding species such as Betula, Populus and Salix and abundant understory vegetation (e.g. V. myrtillus, R. fruticosus and R. idaeus), which might lead to the diversion of browsing from targeted species. This could actually decrease the proportion of browsed F. sylvatica in other stands, or even favour the selection of Betula spp. over beech for food, which would consequently lead to increased future F. sylvatica regeneration. Another possibility would be the avoidance of high palatable species (e.g. Betula spp.) in direct neighbourhood to beech or the plantation of low palatable species (e.g. P. abies) in direct neighbourhood to Betula spp. However, this has to be tested. Our findings revealed that more palatable trees (e.g. Betula spp.) may attract animals to feed on less palatable plants (e.g. F. sylvatica), if they are growing in direct neighbourhood. However, they can possibly act as a sink, reducing browsing damage on the focal plant (Gill, 1992a and references therein), if those species are not growing in close vicinity. The influence of vegetation composition on forage selection refers to neighbouring or associational effects (Barbosa et al., 2009). To minimise costs (e.g. planting and fencing trees) this goal could be achieved by establishing compensation areas containing an adequate alternative food supply for cervids, with an extensive ground cover vegetation and trees of low economic value for forestry. 5. Conclusions Our findings provide an overview of the interactions between cervid browsing and young tree regeneration processes and will therefore be useful for forest management. The extent and success
Fig. 9. Number of Betula spp. HC1 and number of browsed Betula spp. HC1 per plot (NBetula
spp.
= 99, Nbrowsed = 94, Nmax
per plot
= 11, Nplots = 42).
220
M. Bobrowski et al. / Forest Ecology and Management 358 (2015) 212–221
of changes in forest regeneration strategies depends largely on forest administration and the impacts of such changes could be substantial. The results of this study improve current knowledge of diet selection of cervids through the investigation of the extent to which the browsing of a target tree species is affected by neighbouring tree species, the ground cover vegetation, human impact and landscape features. In order to analyse the complexities of feeding habitat selection by cervids, we recommend further studies examine factors influencing browsing decisions based on seasonal fluctuations in forage quality and availability, which will ameliorate the regeneration of targeted species. Acknowledgements We thank Hans-Herrmann Engelke and Knut Sierk for their extensive advice and the Forstamt Sellhorn for allowing us to work in the area. We thank the University of Hamburg for funding the fieldwork. We are grateful to James Turner for language editing. We also acknowledge useful comments by two anonymous reviewers. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.foreco.2015.08. 031. These data include Google maps of the most important areas described in this article. References Barbosa, P., Hines, J., Martinson, H., Szczepaniec, A., Szendrei, Z., 2009. Associational resistance and associational susceptibility: having right or wrong neighbors. Annu. Rev. Ecol. Evol. http://dx.doi.org/10.1146/annurev.ecolsys.110308. 120242. Bee, J.N., Tanentzap, A.J., Lee, W.G., Lavers, R.B., Mark, A.F., Mills, J.A., Coomes, D.A., 2009. The benefits of being in a bad neighbourhood: plant community composition influences red deer foraging decisions. Oikos. http://dx.doi.org/ 10.1111/j.1600-0706.2008.16756.x. Bergman, M., Iason, G.R., Hester, A.J., 2005. Feeding patterns by roe deer and rabbits on pine, willow and birch in relation to spatial arrangement. Oikos. http://dx. doi.org/10.1111/j.0030-1299.2005.13794.x. Bergquist, J., Bergström, R., Zakharenka, A., 2003. Responses of young Norway spruce (Picea abies) to winter browsing by roe deer (Capreolus capreolus): effects on height growth and stem morphology. Scand. J. Forest Res. http://dx.doi.org/ 10.1080/0282758031005431. Bjørneraas, K., Herfindal, I., Solberg, E.J., Sæther, B.E., van Moorter, B., Rolandsen, C. M., 2012. Habitat quality influences population distribution, individual space use and functional responses in habitat selection by a large herbivore. Oecologia. http://dx.doi.org/10.1007/s00442-011-2072-3. Bergström, R., Danell, K., 1987. Effects of simulated winter browsing by moose on morphology and biomass of two birch species. J. Ecol. 75, 533–544. Bryant, J.P., Kuropat, P.J., 1980. Selection of winter forage by subarctic browsing vertebrates: the role of plant chemistry. Ann. Rev. Ecol. Sys. http://dx.doi.org/ 10.1146/annurev.es.11.110180.001401. Cederlund, G., Bergqvist, J., Kjellander, P., Gill, R.M.A., Gaillard, J.M., Boisaubert, B., Ballon, P., Duncan, P., 1998. Managing roe deer and their impact on the environment: maximising the net benefits to society. In: Andersen, R., Duncan, P., Linnell, J.D.C. (Eds.), The European Roe Deer: The Biology of Success. Scand UP, Oslo, pp. 337–372. Chevrier, T., Said, S., Hamard, O., Saint-Andrieux, J., Gaillard, J., 2012. The oak browsing index correlates linearly with roe deer density: a new indicator for deer management? Eur. J. Wildl. Res. http://dx.doi.org/10.1007/s10344-0110535-9. Chouinard, A., Filion, L., 2005. Detrimental effects of white tailed deer browsing on balsam fir growth and recruitment in a second-growth stand on Anticosti Island, Quebec. Ecoscience. http://dx.doi.org/10.2980/i1195-6860-12-4-506.1. Côté, D., Rooney, T.P., Temblay, J.P., Dussault, C., Waller, D.M., 2004. Ecological impacts of deer overabundance. Annu. Rev. Ecol. Evol. Syst. http://dx.doi.org/ 10.1146/annurev.ecolsys.35.021103.105725. Cross, P.C., 1998. The Impacts of White-Tailed Deer and Disturbance on Sapling Establishment and Forest Succession in Virginia Undergraduate Thesis. University of Virginia, Charlottesville, Virginia. Danell, K., Haukioja, E., Huss-Danell, K., 1997. Morphological and chemical responses of mountain birch leaves and shoots to winter browsing along a gradient of plant productivity. Ecoscience 4, 296–303. Duncan, A.J., Hartley, S.E., Iason, G.R., 1998a. The effect of previous browsing damage on the morphology and chemical composition of Sitka spruce (Picea
sitchensis) saplings and on their subsequent susceptibility to browsing by red deer (Cervus elaphus). For. Ecol. Manage. http://dx.doi.org/10.1016/S0378-1127 (97)00177-1. Duncan, P., Tixier, H., Hofmann, R.R., Lechner-Doll, M., 1998b. Feeding strategies and the physiology of digestion in roe deer. In: Andersen, R., Duncan, P., Linnell, J.D. C. (Eds.), The European roe Deer: the Biology of Success. Scand UP, Oslo, pp. 91– 116. Dussault, C., Ouellet, J.P., Courtois, R., Huot, J., Breton, L., Jolicoeur, H., 2005. Linking moose habitat selection to limiting factors. Ecography. http://dx.doi.org/ 10.1111/j.2005.0906-7590.04263.x. Edenius, L., Danell, K., Nyquist, H., 1995. Effects of simulated moose browsing on growth, mortality, and fecundity in Scots pine: relations to plant productivity. Can. J. Forest Res. 25 (4), 529–535. Edenius, L., Bergman, M., Ericsson, G., Danell, K., 2002. The role of moose as a disturbance factor in managed boreal forests. Silva Fenn. 36, 57–67. ESRI, 1999. ArcView 3.2: Environmental Systems Research Institute. ArcView GIS: Release 3.2. Redlands, CA: Environmental Systems Research Institute, 1992– 1999. ESRI, 2011. ArcGIS Desktop: Release 10. Redlands, CA: Environmental Systems Research Institute. Ferretti, F., Sforzi, A., Lovari, S., 2008. Intolerance amongst deer species at feeding: roe deer are uneasy banqueters. Behav. Processes. http://dx.doi.org/10.1016/j. beproc.2008.02.008. Gill RMA, 1992a. A Review of Damage by Mammals in North Temperate Forests. 1. Deer. Forestry. http://dx.doi.org/10.1093/forestry/65.2.145. Gill RMA, 1992b. A review of damage by mammals in north temperate forests. 3. Impact on trees and forests. Forestry. http://dx.doi.org/10.1093/forestry/65.4. 363-a. Götmark, F., Paltto, H., Nordén, B., Götmark, E., 2005. Evaluating partial cutting in broadleaved temperate forest under strong experimental control: short-term effects on herbaceous plants. For. Ecol. Manage. http://dx.doi.org/10.1016/ j.foreco.2005.03.052. Hagemoen, R.I.M., Reimers, E., 2002. Reindeer summer activity pattern in relation to weather and insect harassment. J. Anim. Ecol. http://dx.doi.org/10.1046/j.13652656.2002.00654.x. Heinze, E., Boch, S., Fischer, M., Hessenmöller, D., Klenk, B., Müller, J., Prati, D., Schulze, E.D., Seele, C., Socher, S., Halle, S., 2011. Habitat use of large ungulates in northeastern Germany in relation to forest management. For. Ecol. Manage. http://dx.doi.org/10.1016/j.foreco.2010.10.022. Heroldova, M., Homolka, M., Kamler, J.V., 2003. Breakage of rowan caused by red deer – an important factor for Sorbeto-Piceetum stand regeneration? For. Ecol. Manage. http://dx.doi.org/10.1016/S0378-1127(03)00122-1. Hester, A.J., Millard, P., Baillie, G.J., Wendler, R., 2004. How does timing of browsing affect above- and below- ground growth of Betula pendula, Pinus sylvestris and Sorbus aucuparia? Oikos. http://dx.doi.org/10.1111/j.0030-1299.2004.12605.x. Kamler, J., Homolka, M., 2011. Needles in faeces: an index of quality of wild ungulate winter diet. Folia Zool. 60 (1), 63–69. Kuiters, A.T., Mohren, G.M.J., Van Wieren, S.E., 1996. Ungulates in temperate forest ecosystems. For. Ecol. Manage. http://dx.doi.org/10.1016/S0378-1127(96) 03876-5. Kuiters, A.T., Slim, P.A., 2002. Regeneration of mixed deciduous forest in a Dutch forest heathland, following a reduction of ungulate densities. Biol. Conserv. http://dx.doi.org/10.1016/S0006-3207(01)00204-x. Long, Z.T., Pendergast, T.H., Carson, W.P., 2007. The impact of deer on relationships between tree growth and mortality in an old-growth beech-maple forest. For. Ecol. Manage. http://dx.doi.org/10.1016/j.foreco.2007.06.034. Morrison, ML., Marcot, BG., Mannan, RW., 2006. Wildlife-Habitat Relationships: Concepts and Applications. Island Press, Washington, DC. Morellet, N., Guibert, B., 1999. Spatial heterogeneity of winter forest resources used by deer. For. Ecol. Manage. http://dx.doi.org/10.1016/S0378-1127(99)00007-9. Moser, B., Schütz, M., Hindenlang, K.E., 2006. Importance of alternative food resources for browsing by roe deer on deciduous trees: the role of food availability and species quality. For. Ecol. Manage. http://dx.doi.org/10.1016/ j.foreco.2006.01.045. Myking, T., Solberg, EJ., Austrheim, G., Speed, JD., Bøhler, F., Astrup, R., Eriksen, R., 2013. Browsing of sallow (Salix caprea L.) and rowan (Sorbus aucuparia L.) in the context of life history strategies: a literature review. Eur. J. For. Res. http://dx. doi.org/10.2980/20-3-3619. Mysterud, A., Østbye, E., 1999. Cover as a habitat element for temperate ungulates: effects on habitat selection and demography. Wildl. Soc. Bul. 27, 385–394. Newton, A.C., Echeverría, C., Cantarello, E., Bolados, G., 2011. Projecting impacts of human disturbances to inform conservation planning and management in a dryland forest landscape. Biol. Cons. http://dx.doi.org/10.1016/j.biocon. 2011.03.026. NMELV (Niedersächsisches Ministerium für Ernährung, Landwirtschaft und Verbraucherschutz), 2011. Langfristige Ökologische Wald-Entwicklung in Niedersachsen (LÖWE), Niedersächsische Landesforsten. Persson, I.L., Danell, K., Bergström, R., 2005. Different moose densities and accompanied changes in tree morphology and browse production. Ecol. Appl. http://dx.doi.org/10.1890/04-0499. Reimoser, F., Gossow, H., 1996. Impact of ungulates on forest vegetation and its dependence on the silvicultural system. For. Ecol. Manage. http://dx.doi.org/ 10.1016/S0378-1127(96)03816-9. Sauvé, D.G., Côté, S.D., 2007. Winter forage selection in white-tailed deer at high density: balsam fir is the best of a bad choice. J. Wildlife Manage. http://dx.doi. org/10.2193/2006-056.
M. Bobrowski et al. / Forest Ecology and Management 358 (2015) 212–221 Scherer-Lorenzen, M., Körner, C., Schulze, ED., 2005. Forest diversity and function. Ecol Studies, vol. 176. Springer-Verlag, Berlin Heidelberg. SPSS Version 21, 2012. SPSS IBM, New York, USA. Stolter, C., Ball, J.P., Julkunen-Tiitto, R., Lieberei, R., Ganzhorn, J.U., 2005. Winter browsing of moose (Alces alces) on two different willow species: food selection in relation to plant chemistry and plant response. Can. J. Zool. http://dx.doi.org/ 10.1139/z05-077. Stolter, C., 2008. Intra-individual plant response due to moose browsing: feed-back loops and impacts on multiple consumers. Ecol. Monogr. http://dx.doi.org/ 10.1890/07-0401.1. Stolter, C., Niemelä, P., Varvikko, T., Julkunen-Tiitto, R., Ball, J.P., Vanhatalo, A., Danell, K., Ganzhorn, J.U., 2009. Comparison of secondary chemistry and digestibility of three different boreal coniferous trees. Basic Appl. Ecol. http:// dx.doi.org/10.1016/j.baae.2007.12.001. Stolter, C., Ball, J.P., Julkunen-Tiitto, R., 2013. Seasonal differences in the relative importance of specific phenolics and twig morphology result in contrasting patterns of foraging by a generalist herbivore. Can. J. Zool. http://dx.doi.org/ 10.1139/cjz-2012-0270. Stolter, C., 2015. More Than Meets the Eye: Patch Quality Explain Differences in Utilization by a Large Herbivore, the Moose, manuscript in preparation.
221
Van Hees, A.F.M., Kuiters, A.T., Slim, P.A., 1996. Growth and development of silver birch, pedunculate oak and beech as affected by deer browsing. For. Ecol. Manage. 88 (1), 55–63. Verheyden-Tixier, H., Duncan, P., Ballon, P., Guillon, N., 1998. Selection of hardwood saplings by European roe deer: effects of variation in the availability of palatable species and of understory vegetation. Rev. Ecol. Terre Vie. 53, 245– 253. Vila, B., Torre, F., Martin, J.L., Guibal, F., 2003. Response of young Tsuga heterophylla to deer browsing: developing tools to assess deer impact on forest dynamics. Trees Struct. Funct. http://dx.doi.org/10.1007/s00468-003-0272-0. Welch, D., Staines, B.W., Scott, D., French, D.D., Catt, D.C., 1991. Leader browsing by red and roe deer on young sitka spruce trees in western Scotland. I. Damage rates and the influence of habitat factors. Forestry. http://dx.doi.org/ 10.1093/forestry/64.1.61. Zerbe, S., 2002. Restoration of natural broad-leaved woodlands in Central Europe on sites with coniferous forest plantations. For. Ecol. Manage. http://dx.doi.org/ 10.1016/S0378-1127(01)00686-7. Zuur, A.F., Ieno, E.N., Elphick, C.S., 2010. A protocol for data exploration to avoid common statistical problems. Methods Ecol. Evol. http://dx.doi.org/10.1111/ j.2041-210X.2009.00001.x.