Relative impact of browsing by red deer on mixed coniferous and broad-leaved seedlings—An enclosure-based experiment

Forest Ecology and Management 222 (2006) 302–313 www.elsevier.com/locate/foreco

Relative impact of browsing by red deer on mixed coniferous and broad-leaved seedlings—An enclosure-based experiment D. Pe´pin a,*, P.-C. Renaud a,b, Y. Boscardin c, M. Goulard d, C. Mallet b, F. Anglard b, P. Ballon c a

Inra, Comportement et Ecologie de la Faune Sauvage, Boıˆte Postale 52627, 31326 Castanet-Tolosan, Cedex, France b Inra, Unite´ de Recherche Sur Les Herbivores, Theix, 63122 Saint-Gene`s-Champanelle, France c Cemagref, Unite´ de Recherche Ecosyste`mes Forestiers et Paysages, Domaine des Barres, 45290 Nogent Sur Vernisson, France d Inra, Biome´trie et Intelligence Artificielle, Boıˆte Postale 52627, 31326 Castanet-Tolosan, Cedex, France Received 11 October 2004; received in revised form 27 September 2005; accepted 7 October 2005

Abstract Browsing by mammalian herbivores may help regulate the balance between woody and herbaceous vegetation where agricultural land has been abandoned. Until now, the potential of controlled grazing as a research tool has not yet been fully utilized. In this study, we used an enclosure-based experiment to assess the impact of browsing by red deer on the fate of seedlings on a woodland edge. Species, size, number of lateral shoots, and relative location of seedlings were recorded in 1 m2 plots (n = 57) distributed among four vegetation types. From autumn 2000 to winter 2002, after the stay of three tame hinds in the enclosure for 5-day periods in each season (a year-round deer density of 15 hinds/km2), we checked for loss or browsing damage of seedlings. From autumn 2000 to summer 2001, we additionally recorded the diurnal time spent feeding by the hinds during days 1, 3, and 5, and we simulated their food intake by hand during days 2 and 4. Mortality rates for willow Salix caprea and ash Fraxinus excelsior was higher than for silver fir Abies alba seedlings. Losses due to dehydration, which peaked under coniferous, were counterbalanced by growth of new seedlings. Loss due to browsing was uncommon. Deer impact on coniferous seedlings was insignificant all year long. More broad-leaved seedlings were browsed in spring and summer than in winter. Leader browsing increased with height, most of the taller seedlings occuring in herbaceous vegetation, especially wild cherry Prunus avium. Previously damaged seedlings suffered significantly higher leader browsing than previously undamaged ones. Behavioural observations showed that old pasture was selected by foraging deer all year-round. In autumn, spring and summer, due to quick depletion, the relative contribution of seedlings to the total food intake peaked during day 1, most (>75%) concerning wild cherry. Our results suggest that deer browsing on woody seedlings has more to do with species preference, height structure, growth, and seasonal phenology of seedlings than with induced mortality. Choice of previously damaged seedlings and non-random habitat use by deer can prove useful by minimizing natural regeneration of seedlings close to pastures. Further work is required on long term controlled browsing experiments to better understand natural regeneration of mixed woodland species. # 2005 Elsevier B.V. All rights reserved. Keywords: Deer browsing impacts; Enclosure; Natural regeneration; Seedling; Wild cherry; Woodland

1. Introduction Grazing by livestock and/or wild ungulates have a major impact in determining the nature of ground vegetation and many other components of ecosystems (Baines et al., 1994; deCalesta, 1994; Seagle, 2003). Even if deer populations are rapidly rising in Europe during the last decades, considerable local and regional density variation now occurs (Kuiters et al.,

* Corresponding author. Tel.: +33 5 61 28 51 31; fax: +33 5 61 28 55 00. E-mail address: [email protected] (D. Pe´pin). 0378-1127/$ – see front matter # 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.foreco.2005.10.034

1996; Mouron and Boiseaubert, 1997; Putman and Moore, 1998). Severe damage to forest regeneration only occurs in some areas and is often little related to the size of populations (Senn and Suter, 2003). Impact of deer herbivory on heathlands, grasslands, wetlands and woodlands can be encouraged to help arrest invasion of scrub and trees (Bullock and Pakeman, 1997; Putman and Moore, 1998), to retard woodland successional development (Gill and Beardall, 2001), to change ground vegetation (Morecroft et al., 2001) and to cause a marked decline of species richness of productive grassland (Virtanen et al., 2002). Consequently both livestock and wild ungulates can be used as a tool for landscape management, for instance to reduce

D. Pe´pin et al. / Forest Ecology and Management 222 (2006) 302–313

brush cover or control shrub encroachment (Kosco and Bartolome, 1983; Sharrow, 1989; Augustine and McNaughton, 2004), to counteract grass-encroachment when natural forest regeneration is reduced (Kooijman and van der Meulen, 1996; Kooijman and Smit, 2001; Smit and Kooijman, 2001), to help maintaining landscape diversity for woodland conservation purposes (Bartolome´ et al., 2000), or to conserve plant species diversity in naturally-regenerated forest (Brockway and Lewis, 2003). In a review of damage by mammals in North temperate forests, Gill (1992a,b) examines the numerous factors influencing temporal and spatial variation in deer impact on trees and on forest regeneration. Structural changes created by deer may broadly be through browsing on: (i) seedlings which limits stem density, (ii) leading shoots which limits height growth, and (iii) lateral (side) shoots which reduces foliage density. The incidence of damage in each species varies considerably, the tree species most sensitive to browsing changing by season in the same area (see for instance Miller and Cummins, 1998) and from one area to another, due to the presence and the relative numbers of other surrounding shrubs or tree species (Eiberle and Bucher, 1989; Gomez et al., 2001), the social influences on foraging (see Galef and Giraldeau, 2001 for review) or the hunting impact on habitat use (Martin and Baltzinger, 2002). As pointed out by Gill and Beardall (2001), since deer tend to shift their feeding to side shoots as trees grow beyond reach, the relative importance of these effects depends upon to some extent on stand age and type. Seasonal changes in habitat use by deer during specific periods could be also directly linked to changes in browsing rates (Heuze´, 2002); for instance while the use increased during autumn and winter in the area studied by Castleberry et al. (2000), the deer stay in the region studied by Homolka and Heroldova (2003) only from spring to autumn and tend to move to lower altitudes during the period of continuous snow cover winter. The above connected factors explain why Hester et al. (2000) advocate in favour of the use of controlled grazing as a research and management tool, especially on exclosure-based systems, because their potential has not yet been fully utilized. Experimental systems include deer-proof fenced plots (Scott et al., 2000; Virtanen et al., 2002; Kuiters and Slim, 2002) sometimes associated with selective systems to differentiate between herbivory impact due to roe Capreolus capreolus and red deer Cervus elaphus (Maillard, 1989; Saint-Andrieux and Klein, 2002) or to species other than deer (Cadenasso and Pickett, 2000; Castleberry et al., 2000). Other studies could be also based on simulated browsing by mechanical clipping (Krefting et al., 1966; Canham et al., 1994; Harmer, 1999, 2001), on cafeteria tests (Chevallier-Redor et al., 2001; Renaud et al., 2003) and even on density regulation of ungulate populations (Kuiters and Slim, 2002). However, until now, controlled grazing by large herbivores in enclosure-based systems has not often been used to gather data on the impact of forest ungulates on their environment (Bergstro¨m and Edenius (2003), except in a few cases of simulated stable densities (Tilghman, 1989; deCalesta, 1994; Scott et al., 2000). Consequently, an understanding of the factors influencing the susceptibility of leader and lateral shoots

303

of seedlings to browsing by large herbivores is important in forestry and land use pratices. The great advantage of working with a known density and social structure of deer acting within a deer-proof fenced area is to have precise knowledge of plant– ungulate interactions. Here, we conducted a controlled browsing experiment using tame red deer hinds which were maintained seasonally within a mixed vegetative 1 ha enclosure. In order to work with a constant, stable and rather high deer density all year-round, i.e. about 15 individuals/km2, three hinds were put in this enclosure for 5-day periods during each season from autumn 2000 to winter 2002. Seasonal precise measurements of both the forage availability and red deer diet composition were carried out within this enclosure to analyse the relationship between the choices made on the first day by the hinds and the available vegetation (Dumont et al., 2005). Edible vegetation biomass peaked at 3.8 t of dry matter/ha during both summer and autumn, and decreased to 1.6 t of dry matter/ha in winter. The hinds were highly selective, with 30–40% of their diet on broad-leaved trees and seedlings during the growing season. Conifer species were almost never used. During the growing season, the other main food items were shrub (i.e. 12–21% of dog-rose Rosa canina, with in summer 31% of broom Sarothamnus scoparius), and forbs and legumes (i.e. 10– 25% of Plantago lanceolata, Gallium spp., Taxacum densleonis, Vicia cracca and Trifolium repens). In winter, red deer consumed grasses (76%) and brambles Rubus fructicosus (14%), but very few broad-leaved trees and seedlings (<1%) (Dumont et al., 2005). The aims of this browsing experiment were, at each season, to estimate the relative impact of hinds on broad-leaved and coniferous seedlings, including subsequent leader damage and height increase, and to quantify the relative part of these seedlings in their total food consumption. In a first step, we assess how the seedling stock within the enclosure is distributed, both taking into account their relative density and their size distribution. Due to biokinetic factors related to shoulder size, foliage height of saplings was found to affect the feeding preferences of red deer (Renaud et al., 2003). Consequently, we predict that deer impact should increase with seedling height. Because of the seasonal phenology of plants (dormancy in winter versus new growth in spring and summer), we predict that the rate of deer damage will be low in winter and should increase during the growing season. When leader damage occurs, we predict that final height should be reduced. Because of the highly selective consumption of broad-leaved seedlings throughout the growing season on the first day in the 1 ha enclosure (Dumont et al., 2005), we predict that, due to quick depletion, the relative contribution of seedlings to the total food intake of deer should drastically decrease on days 3 and 5. 2. Materials and methods 2.1. Experimental study site and records of seedling data We established a 1 ha deer-proof fenced enclosure at the Institut National de la Recherche Agronomique of Theix,

304

D. Pe´pin et al. / Forest Ecology and Management 222 (2006) 302–313

France (458420 N, 038300 E). This enclosure included two constrasted zones separated by a low variation of slope: the upper part (2/3) was an old pasture previously used as wintering area for tame red deer and recently invaded by some woodland species, and the lower part (third remaining) was a mixed coniferous and broad-leaved woodland. On the basis of vegetation structure and composition, we marked the limits of homogeneous patchs with wood stakes. We used a real-time differential hand-held GPS (GPS PRO XR, Trimble Navigation Ltd.) to locate each stake, and used SIG Arc View to identify the corresponding map coordinates. Finally, the patchs were pooled into vegetation types to estimate their respective area. Grasses, forbs and legumes were dominant (28.5%), following by coniferous trees (23%, i.e. mainly silver fir Abies alba and Norway spruce Picea abies), by broad-leaved trees (20%, i.e. mainly common oak Quercus robur, wild cherry Prunus avium, common hazel Corylus avellana, common birch Betula alba, willow Salix caprea, chestnut tree Castanea vulgaris, common beech Fagus sylvatica, and rowan Sorbus aucuparia), and by broom (15%, located in the upper part of the enclosure). The remaining vegetation types included fern Pteris aquilina (9%, located in the lower part of the enclosure), bramble (3%) and shrub (1.5%, i.e. mainly blackthorn Prunus spinosa and dogrose). We placed on site a total of 57 plots of 1 m2 distributed among both the upper (n = 22) and the lower (n = 35) parts of the enclosure. We allocated plots within four vegetation types, i.e. within herbaceaous areas (12 and 6, respectively), within coniferous trees (8 and 13, respectively), within broad-leaved trees (2 and 10, respectively), and within fern (six plots in the lower part). Broom, bramble, and shrub were not sampled because we supposed seedling establishment impossible (dense vegetation on ground). Before hinds were introduced in the enclosure in October 2000 and 2001, we registered within each plot the species, the size (up to a maximum of 140 cm), the relative location within the plot, and the number of lateral shoots (only in October 2000) of each seedling. Then, as soon as the hinds left the enclosure which was used for a consecutive 5-day period at each season, we again recorded the species and relative location of seedlings. For each seedling loss, we distinguished natural mortality (due to dehydration) from other reasons. For the remaining seedlings, we noted the contingent consumption of shoots (leader, and/or number of lateral shoots) by hinds. 2.2. Records of behavioural data Before each seasonal field session, eight free-ranging tame adult hinds were first kept together and introduced for a 3-week period in a neighbouring 3 ha enclosure to carry out common operating procedures of behavioural data recordings among observers, and to ensure familiarity with natural vegetation and rumen acclimatation (Duncan et al., 1998, 2001). Three individuals were then selected (Dumont et al., 2005). To simulate a year-round deer density of about 15 hinds/km2, they were introduced in the experimental 1 ha enclosure for a 5-day period at each season, i.e. 16–20 October 2000 (autumn 1), 19–

23 February 2001 (winter 1), 14–18 May 2001 (spring), 9–13 July 2001 (summer), 22–26 October 2001 (autumn 2), and 11– 15 February 2002 (winter 2). During the first four seasons, i.e. from October 2000 to July 2001, from sunset to sunrise during days 1, 3, and 5, three observers on foot staying within 3 m of each hind were continuously recorded into a portable tape recorder the start and end of each inactive (resting, standing) or active (foraging, moving) bout, as well as the number of used patches and the food items ingested. Technical problems occurred in the morning of the 14th May 2001 during which the records of behaviour of hinds no. 01 and 13 were lost for about 3 h and a half in each case. During days 2 and 4, using the 3 ha enclosure, the feeding intake of each hind for all different food items eaten during at least 1 min was estimated by each observer for their animal (Dumont et al., 2005). Bite rates on the different plant species were counted for 1 min at several occasions, and bite weight was estimated by taking grab samples simulating the eating behaviour of the animals (Parker et al., 1993; Agreil and Meuret, 2004; Dumont et al., 2005). 2.3. Data analysis On the basis of the first autumnal seedling records within the 57 plots, we estimated the total abundance of each species and their size distribution by pooling them into 1–10, 11–20, 21–40, and 41+ cm height classes. We used a two-way ANOVA with a fully factorial model to assess the effects of the location of plots among vegetation types and the two parts of the pasture on the mean density of coniferous and broad-leaved seedlings. To compare the above size distribution of coniferous and broadleaved seedlings within the two parts of the enclosure, we first used global x2-tests. Then, to compare their mean height, we also used a two-way ANOVA on the log-transformed data to better approximate a normal distribution. For species with enough data (n > 20), we used exponential regression to assess the relationship between the size and the number of lateral shoots. Following each seasonal 5-day stay of hinds in the enclosure, we checked the mortality rate of seedlings and tested the possible influence of four vegetation types, of four seedling height classes, and of species on seedling state. To compare the global mortality rate among species from autumn 1 to autumn 2, we first distinguished silver fir from pooled data concerning Norway spruce and Scots pine Pinus sylvestris (other coniferous seedlings, n = 18), and among broad-leaved, ash Fraxinus excelsior, wild cherry, willow, and rowan from pooled data concerning white beam Sorbus aria, common birch, chestnut tree, common oak, field maple Acer campestris, common beech, and common hazel (other broad-leaved seedlings, n = 79). The seedlings are grouped within the 57 plots and certain species tend to occur on plots within certain areas of the enclosure. To account for this problem, we examined the probability that a seedling died by fitting at each season the remaining individual seedlings to a generalized linear mixed model with a binomial error term and a logit link function (McCullagh and Nelder, 1989; Breslow and Clayton,

D. Pe´pin et al. / Forest Ecology and Management 222 (2006) 302–313

1993; Wolfinger and O’Connel, 1993). In this mixed model, plots were included as random effects, while vegetation types, seedling height classes and species were included as fixed effects and tested for their significance. We used the R package glmmML (R Development Core Team, 2004). As above, we also used generalized linear mixed models to examine at each season the probability that a seedling had its leading shoot browsed. Because deer impact was particularly focused on wild cherry seedlings (Dumont et al., 2005), we separated wild cherry data from all other broad-leaved species to analyze seedling shoot consumption by hinds (globally, or by taking into account height classes). Only using the broad-leaved seedling stock which survived to the end of summer record, i.e. n = 460, we estimated by x2-tests the effect of previous leader and lateral browsing damage on the subsequent susceptibility to leader browsing by hinds. We also estimated the effect of leader browsing on annual growth of broad-leaved seedlings between October 2000 and 2001 by using linear functions relating the start and end heights of seedlings which were not leader browsed or which were leader browsed at least once. To quantify the relative contribution of coniferous, wild cherry, and all other broad-leaved seedlings to the total food intake by hinds, we used the behavioural information regarding the feeding behaviour of hinds on various food items. Data recorded during the same day, i.e. days 1, 3 or 5, were gathered. The seasonal quantity (g of dry matter content) of food items consumed by hinds was calculated for each individual from the product of the corresponding average mouthful weight (after 2 days in a drying store at 80 8C), speed of ingestion (number of

305

mouthfuls per minute) and time spent feeding (Dumont et al., 2005). Having corrected food intake of hinds no. 01 and 13 because of loss of data for about 3 h and a half during day 1 in spring, we used Kruskal–Wallis tests to compare the total daily food intake of three hinds between seasons. Then, the seedling intake (without distinguishing apical and/or lateral shoots from other foliage parts) was separated from all other food items to estimate the relative contribution of seedlings within each diurnal diet at each season. 3. Results 3.1. Availability of seedlings and number of lateral shoots as related to height classes Before the introduction of hinds in the 1 ha enclosure, the first autumnal records on the 57 plots of 1 m2 concerned a total of 1006 seedlings (Table 1). Among the broad-leaved seedlings, most (54.4%) records concerned ash, while among coniferous the great majority (96.3%) was for silver fir. The median height of both ash and silver fir was short, i.e. 6 cm (first column of Table 1). The average density of broad-leaved and coniferous seedlings did not differ between vegetation types (F = 0.50, d.f. = 7, p = 0.831), but there were significant differences between the two parts of the enclosure (F = 8.39, d.f. = 3, p < 0.001). More broad-leaved seedlings (80.2%) than coniferous seedlings were found in the old pasture (mean density of 15.1 and 3.7 seedlings/ m2, respectively), while on the contrary more coniferous seedlings (68.2%) than broad-leaved seedlings were found in

Table 1 Seedling availability among the 57 plots of 1 m2 in autumn 2000, and corresponding disappearance after the stay of hinds in the 1 ha enclosure for a 5-day period from autumn 2000 to summer 2001, and just before the stay of hinds in autumn 2001 Seedling species

Median height (cm)

Broad-leaved seedlings Ash Beech Birch Chestnut tree Common hazel Field maple Oak Rowan White beam Wild cherry Willow

6 16.5 9 25 17 9.5 13.5 8 16 15 7

Sub-total Coniferous seedlings Norway spruce Scots pine Silver fir Sub-total Total

Initial sample size

283 4 6 23 5 4 36 71 1 66 21 520

13 52 6

Seasonal mortality (due to dehydration) Autumn 2000 6

Winter 2000 6

Spring 2001 14

Summer 2001 5 (2)

Autumn 2001 46 (43)

9 (5)

1 4 (3)

5 (5) 4

6 (6) 8 (7)

63 0 6 3 2 0 4 17 0 9 2

23 (12)

69 (63)

105

1 3 (3) 1 (1)

1

3

10

1 1

1 (1)

2

1

10

17 (1)

Number of new seedlings

16 2 468

2

6 (2)

13 (5)

31 (22)

3 0 29

486

2

6 (2)

13 (5)

31 (22)

32

12

23 (3)

46 (17)

100 (85)

137

1006

10

Mortality due to dehydration is shown between brackets. Number of new seedlings found in autumn 2001 is given in the last column.

306

D. Pe´pin et al. / Forest Ecology and Management 222 (2006) 302–313

Fig. 1. Relative distribution of coniferous and broad-leaved seedlings within various height classes, as related to the two considered zones of the 1 ha enclosure. The corresponding total sample size is given in each case at the bottom of histograms.

the woodland (mean density of 11.5 and 5.4 seedlings/m2, respectively) (Fig. 1). Consequently, there was a clear spatial heterogeneity in the seedling stock within the enclosure. The sampling design within the two parts of the enclosure did not include the same relative proportion of starting height classes among coniferous and broad-leaved seedlings (x2 = 84.04, d.f. = 9, p < 0.001) (Fig. 1). A great majority (81.1%) of coniferous seedlings were found over the whole enclosure within the smallest class (1–10 cm); only a small proportion (1.4%) was >40 cm tall. On the contrary, significant changes of starting height structure occurred among broad-leaved seedlings between old pasture and woodland plots, with more 11–40 cm seedlings in the woodland than in the old pasture (34.6% versus 14.8%), and less 1–10 and 41+ cm seedlings in the woodland than in the old pasture (61.7% versus 73.8%, and 3.7% versus 11.4%, respectively). On average, the broad-leaved seedlings had much greater starting size than coniferous seedlings in both the old pasture (8.3 cm versus 4.6 cm) and the woodland (8.9 cm versus 5.6 cm) (F = 22.40, d.f. = 3, p < 0.001). In autumn, irrespective of their location, the broad-leaved seedlings of 1–10 cm height rarely had lateral shoots. From 11– 20 cm height, more lateral shoots were found for the coniferous than for the broad-leaved seedlings. Then, structural differences also occurred for a given size of seedlings among the broadleaved species, with for instance more lateral shoots in chestnut

Fig. 2. The total number of shoots of silver fir, chestnut tree, and wild cherry seedlings in relation to their starting height in October 2000, before the stay of hinds in the 1 ha enclosure.

tree than in wild cherry (see Fig. 2). For the species with at least a sample size of 20, we found that all adjusted exponential equations (y = eax) linking the size (x) to the total number of shoots (y) of seedlings were highly significant. The values of the parameter a were respectively 0.0954 for silver fir (r2 = 0.660), 0.0399 for common oak (r2 = 0.663), 0.0356 for chestnut tree (r2 = 0.811), 0.0227 for rowan (r2 = 0.247), 0.0201 for ash (r2 = 0.701), and 0.0151 for wild cherry (r2 = 0.595). No lateral shoots were found on willow seedlings. From these equations, we can deduce that the average total number of shoots for seedlings of 50 cm height was 118 for silver fir, seven for common oak, six for chestnut tree, three for rowan and ash, and only two for wild cherry. 3.2. Seedling mortality and growth of new seedlings Seedling mortality regularly increased from autumn 1 to autumn 2. More than half occurred between summer and autumn 2, and were especially due to dehydration (Table 1). After the 5day stay of hinds within the enclosure in autumn 2, the additional seedling mortality concerned seven ash, onewillow and one silver fir. The corresponding additional seedling mortality in winter 2 included eight ash, one rowan and five silver fir. Significant differences in the seedling mortality occurred in relation to the species classes in all seasons ( p ranging from 0.047 in spring to <0.001 in autumn 1, summer, and autumn 2), and also in relation to the vegetetation types in autumn 1 ( p = 0.02), winter ( p = 0.04), spring ( p = 0.047), and summer ( p = 0.004), and in relation to the height classes in spring ( p = 0.01), winter ( p = 0.009), summer ( p < 0.001), and autumn 2 ( p < 0.001). Globally, higher mortality rates occurred for willow (71.4%), especially in herbaceaous areas, and ash (27.2%) seedlings while silver fir and other coniferous seedlings showed lower mortality rates in all the four vegetation types (11.1% and 0.0%, respectively) (Fig. 3). The smallest starting height class (1– 10 cm) of seedlings mostly contributed to this decline. Among the known mortality causes (64.4%), dehydration concerned 105 seedlings, a great part (57%) dying in the

Fig. 3. Global mortality rate of various categories of seedlings from autumn 2000 to 2001 (prior to the introduction of hinds in the enclosure in both cases), after the stay of hinds during 5-day periods at each season. The starting sample size is given in each case. Other broad-leaved seedlings refered to the gathering of white beam, common beech, common birch, chestnut tree, common hazel, field maple, and common oak, while other coniferous refered to Norway spruce and Scots pine.

D. Pe´pin et al. / Forest Ecology and Management 222 (2006) 302–313

coniferous vegetation type. Thirteen seedlings were lost due to a red fox Vulpes vulpes den. Four seedlings previously leader browsed only once disappeared (three rowan 3, 5 and 14 cm tall in summer, and one silver fir 8 cm tall in spring), while two other seedlings previously trampled disappeared in autumn 2 (one rowan 30 cm tall and one willow 14 cm tall). Prior to the stay of three hinds in the enclosure in autumn 2, 137 new seedlings (105 broad-leaved and 32 coniferous) were found (see last column of Table 1). Consequently, the natural mortality of older seedlings (76 broad-leaved and 29 coniferous) due to dehydration was more or less counterbalanced by the growth of these new seedlings (x2 = 0.573, d.f. = 1, p = 0.449). 3.3. Seasonal pattern in lateral and leader shoot browsing Browsing of lateral shoots was especially registered in autumn 1 and winter 1 among the 41+ cm starting height class of seedlings, with a greater impact on available broad-leaved shoots (11.7% and 2.8%, respectively) than on coniferous ones (0.1% and 0.7%, respectively). In spite of the recruitment of new shoots in spring and summer (not quantified in the present study), the remove of lateral shoots remained uncommon among coniferous seedlings (14 and 2 cases, respectively), and increased with the starting height classes among broad-leaved seedlings (Fig. 4). Significant differences in leader shoot browsing occurred in all seasons in relation to both the species classes ( p ranging from 0.015 in winter to <0.001 in autumn 1 and summer), and the height classes ( p < 0.001 in all cases), and also in relation to the vegetation types in autumn 1 ( p < 0.001), spring ( p = 0.013), and summer ( p = 0.008). Leader browsing of coniferous seedlings remained uncommon (only three shoots were removed, one case in winter 1, the two others in spring). Leader browsing of broad-leaved seedlings changed between seasons, with few cases during autumn 1 (2.7%) and winter 1 (0.4%) and a significant increase in spring (16.3%) and summer (10.4%), more especially in herbaceaous areas (Fig. 5). Low rates of leader browsing again occurred in autumn 2 (2.6%) and

307

winter 2 (1.0%). Globally, we found 1.6%, 6.9%, 27.6%, and 54.8% of apical shoots removed respectively for 1–10, 11–20, 21–40, and 41+ cm height classes. 3.4. Effect of previous browsing damage on the subsequent susceptibility to leader browsing From the 460 broad-leaved seedlings which survived until the end of summer, we found than the rate of leader damaged two or three times by hinds was significantly greater for wild cherry seedlings than for all other pooled broad-leaved seedlings (36.1% versus 2.5%, respectively) (x2 = 92.07, d.f. = 1, p < 0.001). The subsequent susceptibility to leader browsing was particularly found in the starting 41+ cm height class, and concerned more wild cherry seedlings than other broad-leaved seedlings (Fig. 6). All these seedlings were only found in the old pasture, mainly in herbaceaous areas. Among the broad-leaved seedlings which suffered browsing damage in autumn 1 and winter 1, due to leader browsing (n = 16) or to lateral browsing (n = 6), half were leader browsed by hinds in spring. Of all the remaining previously undamaged seedlings, hinds removed 14.8% of leader shoots (x2 = 18.78, d.f. = 1, p < 0.001). In summer, 35.2% of seedlings which were previously leader browsed (n = 85) or had lateral shoots previously removed (n = 6) suffered leader browsing, while only 4.3% previously undamaged broad-leaved seedlings were leader browsed (x2 = 74.23, d.f. = 1, p < 0.001). Hinds removed significantly more leader shoots from previously undamaged seedlings in spring than in summer (x2 = 24.47, d.f. = 1, p < 0.001). This seasonal variation could be explained by a greater impact of hinds in spring on previously undamaged wild cherry (45.5%) than on previously undamaged other broad-leaved seedlings (only 10.4%) (x2 = 46.65, d.f. = 1, p < 0.001). 3.5. Effect of leader browsing on annual growth of broadleaved seedlings The impact of leader browsing on height increase from October 2000 to next October varied between seedling species. Indeed, comparing the annual growth of unbrowsed seedlings, the effect of leader browsing was generally to stop height increase. That concerned ash, common oak, rowan, and especially wild cherry seedlings leader browsed at least once. However, although some willow seedlings were leader browsed once, compensatory growth occurred (Fig. 7). 3.6. Relative contribution of seedlings to the total food intake by hinds

Fig. 4. Seasonal incidence of lateral shoot browsing on broad-leaved seedlings in relation to starting height classes, after the stay of hinds during 5-day periods at each season.

Hinds spent from 82.4% in winter to 95.5% in autumn of their total diurnal foraging time within the old pasture which represents 2/3 of the total size of the enclosure. The estimated total daily food intake (g of dry matter content) by the three hinds during the diurnal field observations ranged from 4225 to 6846 g (mean = 5677 g, median = 5589 g), and did not

308

D. Pe´pin et al. / Forest Ecology and Management 222 (2006) 302–313

Fig. 5. Seasonal incidence of leader browsing on seedlings in relation to their respective location within the four vegetation types, i.e. broad-leaved trees (dotted), herbaceous areas (black), coniferous trees (herring-bone), and fern (hatched), after the stay of hinds during 5-day periods at each season. Other c. refers to other coniferous seedlings, while other b. refers to other broad-leaved seedlings.

significantly change between seasons (H = 0.69, d.f. = 3, p = 0.875) or from day to day (H = 1.85, d.f. = 2, p = 0.397). Globally, in autumn, spring, and summer, the relative contribution of seedlings to the daily food intake peaked during day 1 (10.2%, 15.1%, and 13.8%, respectively, see Table 2). Then, this relative part drastically decreased during day 3 (ranging from 1.4% to 4.0%) and day 5 (ranging from 0.3% to 2.0%). In all cases, coniferous seedlings were not eaten by hinds. Among the broad-leaved seedlings, the wild cherry ones were more selected. By contrast, the intake of seedlings was very uncommon in winter, a few coniferous contributing to the diet of days 3 and 5 (Table 2). 4. Discussion Higher mortality rates were found among broad-leaved seedlings for willow and ash, while a lower mortality rate was noted for coniferous seedlings. Moreover, broad-leaved seedlings suffered markedly high degrees of natural mortality due to dehydration within the coniferous vegetation type, contributing to maintainance of the spatial heterogeneity.

Fig. 6. Comparison of leader browsing rate of wild cherry (a) and of all other broad-leaved seedlings (b) in relation to starting height classes, after the stay of hinds during 5-day periods at each season, from October 2000 to July 2001. The corresponding sample size is given in each case at the top of histograms.

However, we found that such natural mortality is counterbalanced by the growth of new seedlings. By contrast, the influence of previous seasonal leader browsing and trampling on mortality of seedlings was uncommon. Contrary to the belief among foresters that browsing by deer is the driving force behind inadequate regeneration failure (Senn and Suter, 2003), the mortality of seedlings may be consequently much more affected by competition among plant species than by deer pressure (Putman and Moore, 1998; Gill and Beardall, 2001). In our enclosure-based experiment, simulating a year-round deer density of 15 hinds/km2 which is rather high for the geographical region of interest, the leader browsing of coniferous seedlings by tame hinds was uncommon all yearround. On the contrary, many studies conducted in open forests with the presence of several ungulate populations together has shown that coniferous trees may suffer high damage rates in winter (Maillard, 1989; Maizeret and Ballon, 1990; Picard et al., 1994; Ammer, 1996; Saint-Andrieux et al., 1999; SaintAndrieux and Klein, 2002). Using a selective system to differentiate the relative part of roe deer and red deer on damage of silver fir terminal shoots, Maillard (1989), Saint-Andrieux et al. (1999) and Saint-Andrieux and Klein (2002) demonstrated that impact of roe deer is higher than those of red deer. Saint-Andrieux et al. (1999) noted that damage just stoped with the opening of silver fir buds. However, some exceptions to this general winter browsing pattern on coniferous species occur: leader browsing was for instance mostly confined to spring new growth on Sitka spruce Picea sitchensis (Staines and Welch, 1984; Welch et al., 1991). One reason for the very low impact of hinds on coniferous seedlings may be probably related to the high occurrence of the 1–10 cm height class in both parts of the enclosure, the rate of leader browsing by deer being closely related to height (see the next paragraph). In cases where the vegetation was higher than the seedling, the physical protection afforted by the vegetation ¨ rlander, 1998; Pietrzyseemed to dominate (Bergquist and O kowski et al., 2003). On the other hand, Palmer and Truscott

D. Pe´pin et al. / Forest Ecology and Management 222 (2006) 302–313

309

Fig. 7. Linear relationships between the initial height (x) in October 2000 and the ending height (y) in October 2001 for ash, chestnut tree, common oak, rowan, and wild cherry seedlings which were not leader browsed (-, fine line), or at least leader browsed once (o, heavy line). The sample size are 184, 12, 25, 39, 21, and 22, 8, 7, 12, 33, respectively. For the willow seedlings, the linear regression included two specimens unbrowsed and four specimens leader browsed once (one case in autumn, and the three others in spring). Constants of all linear regressions were not significant different from 0.

(2003) indicated that coniferous seedlings of a given height were more likely to be browsed by deer if they were above the ground vegetation than at the same height as the ground vegetation, apparently indicating that the visibility of a seedling to deer is important. For leader browsing of broad-leaved seedlings by hinds, in accordance with other studies (see Gill, 1992a for a review), our results show that there was a clear seasonal pattern with a low damage rate in winter when available seedlings were in the period of dormancy, and higher impacts in spring and summer due to new growth and/or compensatory growth following damage. A similar seasonal pattern was found for young broadleaved plantations browsed by fallow deer Dama dama (Moore et al., 2000): browsing was generally related to the extent of budding of individual trees, with trees that were in leaf being more prone to damage than those without leaves erupted. The leading shoots (and upper leaves) are known to be actively

selected by deer because they are usually the most actively growing parts of young trees, and consequently show quick recovery to damage (Price, 1991; Bowyer and Bowyer, 1997; ¨ rlander, 1998; Gill and Beardall, 2001). Bergquist and O We found that browsing impact by deer regularly increased with height, more than half of broad-leaved seedlings of 41+ cm height class being leader browsed. From our plot sampling, we also found that all the taller seedlings were located in the old pasture part, most of them (especially wild cherry) occurring in the herbaceous vegetation. Some other studies also related browsing impact with tree height. For instance, young Sitka spruce trees from 30 to 60 cm tall were found to suffer the most damage, while few trees taller than 80 cm were leader browsing (Staines and Welch, 1984; Welch et al., 1991). Depending on the height of snow cover in winter in mountain conditions, and the body size of ungulates present (Senn and Suter, 2003), the most affected height classes are

D. Pe´pin et al. / Forest Ecology and Management 222 (2006) 302–313

310

Table 2 Total quantity (g of dry matter content) and relative proportion (%) of seedlings ingested by the three red deer hinds as estimated from field observations carried out during daytime in each season within the 1 ha enclosure Season

Day

Dry matter content of seedlings

Relative proportion of seedlings (%)

Quantity (g)

Part in the diurnal diet (%)

Wild cherry

Other broad-leaved

Coniferous

Autumn

1 3 5

429.9 73.5 22.8

10.18 1.39 0.33

78.5 70.4 60.7

21.5 29.6 39.3

0.0 0.0 0.0

Winter

1 3 5

1.8 5.2 5.4

0.03 0.08 0.10

27.8 30.1 41.7

62.2 0.0 8.8

0.0 69.9 49.5

Spring

1 3 5

640.7 149.4 123.4

15.13 2.70 1.97

97.7 94.6 94.4

2.3 5.4 5.6

0.0 0.0 0.0

Summer

1 3 5

850.0 276.6 76.7

13.80 4.04

98.0 77.4 47.0

2.0 22.6 53.0

0.0 0.0 0.0

those between 80 and 140 cm for rowan, but between 60 and 100 cm for Norway spruce (Motta, 2003). In cafeteria tests, for tame red deer having shoulder height ranging from 80 to 105 cm, preferred bunches of willow were at 85 and 115 cm (Renaud et al., 2003). Site factors such as altitude, composition of surrounding vegetation and spatial structure of plant communities can strongly influence the intensity of browsing damage (Miller et al., 1982; Reimoser and Gossow, 1996; Gomez et al., 2001; Takada et al., 2002; Rao et al., 2003). Welch et al. (1991) found that interior plots suffered less damage than near-edge or edge plots, because edge plots had greater deer usage as measured by pellet-group counts. On the contrary, Andren and Angelstam (1993) found that moose damage was neither significantly related to stand size nor to distance to the edge. Previously damaged broad-leaved seedlings (leader or lateral shoots) were found to suffer significantly higher leader browsing damage than previously undamaged seedlings, the susceptibility being greater for wild cherry than for other species. This subsequent susceptibility to leader browsing could be due to induced changes in morphology, biomass and nutrient characteristics in browsed plants. In a review on moose Alces alces—plant interactions, Danell et al. (1994) concluded that there is no single coherent response of woody plants to herbivory by moose, and that the most logical response by the herbivores should be to avoid browsed plants. However, they also pointed out that some studies have shown that there is a higher probability for a browsed tree than for an unbrowsed tree to be browsed 1 year later, and that the reason for this preference may be related to the induced changes in morphology. Welch et al. (1991, 1992), Bowyer and Bowyer (1997), Moore et al. (2000), and Palmer and Truscott (2003) indicated that young coniferous or broad-leaved trees damaged by deer or moose during one year have been found to be preferentially re-browsed during subsequent years. For coniferous species, susceptibility to leader browsing was related to some degree of compensatory growth, with the presence of

either multiple or new leaders (i.e. <1-year-old). Using experimental conditions where variation due to factors such as geographical location and associated vegetation is minimised, Duncan et al. (1998) tested the preference by red deer of either browsed or left undamaged Sitka spruce saplings in a subsequent year. They found that chemical composition was not affected by browsing, and that biomass removed, bites taken, and time spent at trees were greater on previously undamaged saplings than on damaged saplings, all these three variables being however positively related to the height of each tree. In our study conditions, the impact of leader browsing on broad-leaved seedlings was to stop annual height increase of ash, chestnut tree, common oak, and wild cherry, the only exception being for the very high compensatory growth of willow. Miller et al. (1982) indicated that rowan was more tolerant to damage than silver birch Betula pendula, and also than Scots pine and juniper Juniperus communis. However, Van Hees et al. (1996) found that browsing reduced height and changed above-ground biomass distribution (more branch biomass and less leaf biomass) of beech, pedunculate oak and silver birch saplings, the browsing-induced mortality being lower for beech than for the two other species. Bowyer and Bowyer (1997) demonstrated that willow Salix glauca browsed by moose resulted in leaders of regrowth that were larger than unbrowsed stems, moose obtaining more food for the same effort by feeding on stems they had browsed in previous years. From our field observations, we clearly showed that tame hinds did not forage randomly within the enclosure. Of the total diurnal time spent feeding, they spent a minimum of 82.4% in winter and a maximum of 95.5% in autumn in the old pasture, though the size of this zone made up only 2/3 of the enclosure. Comparing the edible dry matter biomass on the enclosure (from 1.6 t/ha in winter to 3.6 t/ha in summer) (Dumont et al., 2005), the diurnal consumption of vegetation by the three hinds remained very low (from 30 kg in winter to 27 kg in spring). Consequently, the estimated total daily food intake did not

D. Pe´pin et al. / Forest Ecology and Management 222 (2006) 302–313

significantly change between seasons, or from day to day. However, the relative part of broad-leaved seedlings in the diurnal diet decreased during autumn, spring and summer from day 1 to 5, the wild cherry being the dominant browsed species on all of the 5 days of these three browsing seasons. In winter, the diurnal diet only included a very low abundance of seedlings, with occasional coniferous seedlings. The records on seedling data within the plots mainly concerned leader damage by hinds (the recruitment of new lateral shoots in spring and summer was not quantified in the present study), while the behavioural field observations also took into account the browsing impact on all leaves. The seedlings were in the period of dormancy from autumn to winter, and consequently those which were previously damaged in autumn (the taller) did not have new growth. By contrast, because hinds only remained within the enclosure for 5-day periods, the seedlings damaged in spring and summer may have the possibility of new compensatory growth. So, the height increase of both previously undamaged and previously damaged seedlings to leader browsing act together to offer a great availability of suitable seedlings to leader browsing by hinds.

311

natural regeneration of seedlings from the impact of deer in areas with herbaceaous vegetation type, it is not necessary to act beforehand but to wait for their establishment because the younger seedlings are not likely to be browsed if they are smaller than the ground vegetation. Further work is required on long term controlled browsing experiments to better understand natural regeneration of mixed woodland species. Acknowledgments A thesis grant from the Institut National de la Recherche Agronomique was provided to one of us (PCR). Marcel Verdier, the Redon red deer farmer, kindly offered us the opportunity to carry out this study. We thank Cyril Agreil, Bertrand Dumont, Michel Meuret, Nicolas Morellet, and He´le`ne VerheydenTixier for invaluable assistance in the collection of behavioural data or for their expertise. Two anonymous referees provided very helpful comments on earlier drafts of the manuscript. Peter Winterton and Mark Hewison kindly improved the English. References

5. Conclusion Ungulate herbivory is traditionally known to interfere with growth and survival of saplings in young forest plantations, but browsing patterns are seldom understood when considering natural regeneration. Since several factors drive the dynamics of seedling community in a complex interplay which varies highly in both space and time, the relative loss by browsing must be accurately quantified by means of longitudinal surveys of seedlings and compared to other potential mortality parameters on each woody species (Senn and Suter, 2003). And even when seedling loss by browsing is relatively high in comparison with other causes, the mortality of trees could be less important than the number that survive for the operational stand regeneration (Reimoser et al., 1999). The current controlled browsing experiment clearly indicates that loss by browsing only concerns the establishment of a very few number of seedlings with short size, while repeated leader browsing of upper broad-leaved seedlings located within herbaceous vegetation is not followed by the immediate mortality of those damaged seedlings. This enclosure-based experiment suggests that deer browsing on woody seedlings has more to do with species preference, height structure, growth, and seasonal phenology of seedlings than with induced mortality. However, both the selection of pasture by deer for foraging, and the choice of previously damaged seedlings can prove useful by minimizing natural regeneration of woodland species bordering recently abandoned pastures. From a landscape management point of view, if free ranging deer stay all year-round in such forest edge, an acknowledged fact could be that as soon as broad-leaved seedlings are above the ground vegetation, they had a great chance to be leader browsed. In this circumstance, the compensatory growth of previously damaged seedlings may be systematically stopped. In addition, in order to protect the

Agreil, C., Meuret, M., 2004. An improved method for quantifying intake and ingestive behaviour of ruminants in diverse and variable habitats using direct observation. Small Ruminant Res. 54, 99–113. Ammer, C., 1996. Impact of ungulates on structure and dynamics of natural regeneration of mixed mountain forests in the Bavarian Alps. For. Ecol. Manage. 88, 43–53. Andren, H., Angelstam, P., 1993. Moose browsing on Scots pine in relation to stand size and distance to forest edge. J. Appl. Ecol. 30, 133–142. Augustine, D.J., McNaughton, S.J., 2004. Regulation of shrub dynamics by native browsing ungulates on East African rangeland. J. Appl. Ecol. 41, 45– 58. Baines, D., Sage, R.B., Baines, N.M., 1994. The implications of red deer grazing to ground vegetation and invertebrate communities of Scottish native pinewoods. J. Appl. Ecol. 31, 776–783. Bartolome´, J., Franch, J., Plaixats, J., Seligman, N.G., 2000. Grazing alone is not enough to maintain landscape diversity in the Montseny Biosphere Reserve. Agr. Ecosyst. Environ. 77, 267–273. ¨ rlander, G., 1998. Browsing damage by roe deer on Norway Bergquist, J., O spruce seedlings planted on clearcuts of different ages. 2. Effect of seedling vigour. Forest Ecol. Manage. 105, 283–293. Bergstro¨m, R., Edenius, L., 2003. From twigs to landscapes—methods for studying ecological effects of forest ungulates. J. Nat. Conserv. 10, 203– 211. Bowyer, J.W., Bowyer, R.T., 1997. Effects of previous browsing on the selection of willow stems by Alaskan moose. Alces 33, 11–18. Breslow, N.E., Clayton, D.G., 1993. Approximate inference in generalized mixed models. J. Am. Statist. Assoc. 88, 9–24. Brockway, D.G., Lewis, C.E., 2003. Influence of deer, cattle grazing and timber harvest on plant species diversity in a longleaf pine bluestem ecosystem. Forest Ecol. Manage. 175, 49–69. Bullock, J.M., Pakeman, R.J., 1997. Grazing of lowland heath in England: management methods and their effects on heathland vegetation. Biol. Conserv. 79, 1–13. Cadenasso, M.L., Pickett, S.T.A., 2000. Linking forest edge structure to edge function: mediation of herbivore damage. J. Ecol. 88, 31–44. Canham, C.D., McAninch, J.B., Wood, D.M., 1994. Effects of the frequency, timing, and intensity of simulated browsing on growth and mortality of tree seedlings. Can. J. Forest Res. 24, 817–825. Castleberry, S.B., Ford, W.M., Miller, K.V., Smith, W.P., 2000. Influences of herbivory and canopy size on forest regeneration in a southern bottomland hardwood forest. Forest Ecol. Manage. 131, 57–64.

312

D. Pe´pin et al. / Forest Ecology and Management 222 (2006) 302–313

Chevallier-Redor, N., Verheyden-Tixier, H., Verdier, M., Dumont, B., 2001. Foraging behaviour of red deer Cervus elaphus as a function of the relative availability of two tree species. Anim. Res. 50, 57–65. Danell, K., Bergstro¨m, R., Edenius, L., 1994. Effects of large mammalian browsers on architecture, biomass, and nutrients of woody plants. J. Mammal. 75, 833–844. deCalesta, D.S., 1994. Effect of white-tailed deer on songbirds within managed forests in Pennsylvania. J. Wildl. Manage. 58, 711–718. Dumont, B., Renaud, P.C., Morellet, N., Mallet, C., Anglard, F., VerheydenTixier, H., 2005. Seasonal variations of red deer selectivity on a mixed forest edge. Anim. Res. 54, 369–381. Duncan, A.J., Hartley, S.E., Iason, G.R., 1998. 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. 103, 57–67. Duncan, A.J., Hartley, S.E., Thurlow, M., Young, S., Staines, B., 2001. Clonal variation in monoterpene concentrations in Sitka spruce (Picea sitchensis) saplings and its effect on their susceptibility to browsing damage by red deer (Cervus elaphus). Forest Ecol. Manage. 148, 259–269. Eiberle, K., Bucher, H., 1989. Interdependenzen zwischen dem Verbiss verschiedener Baumarten in einem Plenterwaldgebiet. Z. Jagdwiss. 35, 235–244. Galef, B.G., Giraldeau, L.-A., 2001. Social influences on foraging in vertebrates: causal mechanisms and adaptive functions. Anim. Behav. 61, 3–15. Gill, R.M.A., 1992a. A review of damage by mammals in North temperate forests: 1 deer. Forestry 65, 145–169. Gill, R.M.A., 1992b. A review of damage by mammals in North temperate forests: 3 impact on trees and forests. Forestry 65, 363–388. Gill, R.M.A., Beardall, V., 2001. The impact of deer on woodlands: the effects of browsing and seed dispersal on vegetation structure and composition. Forestry 74, 209–218. Gomez, J.M., Hodar, J.A., Zamora, R., Castro, J., Garcia, D., 2001. Ungulate damage on Scots pines in Mediterranean environments: effects of association with shrubs. Can. J. Bot. 79, 739–746. Harmer, R., 1999. Survival and new shoot production by artificially browsed seedlings of ash, beech, oak and sycamore grown under different levels of shade. Forest Ecol. Manage. 116, 39–50. Harmer, R., 2001. The effect of plant competition and simulated summer browsing by deer on tree regeneration. J. Appl. Ecol. 38, 1094– 1103. Hester, A.J., Edenius, L., Buttensch1n, R.M., Kuiters, A.T., 2000. Interactions between forests and herbivores: the role of controlled grazing experiments. Forestry 73, 381–391. Heuze´, P., 2002. Impact a` moyen terme des grands herbivores sauvages sur le renouvellement de la heˆtraie-sapinie`re des Hautes Vosges. Ph.D. Thesis, Univ. Metz, Metz, France. Homolka, M., Heroldova, M., 2003. Impact of large herbivores on mountain forest stands in the Beskydy Mountains. Forest Ecol. Manage. 181, 119– 129. Kooijman, A.M., van der Meulen, F., 1996. Grazing as a control against ‘grassencroachment’ in dry dune grasslands in The Netherlands. Landsc. Urban Plan. 34, 323–333. Kooijman, A.M., Smit, A., 2001. Grazing as a measure to reduce nutrient avaibility and plant productivity in acid dune grasslands and pine forests in The Netherlands. Ecol. Eng. 17, 63–77. Kosco, B.H., Bartolome, J.W., 1983. Effects of cattle and deer on regenerating mixed conifer clearcuts. J. Range Manage. 36, 265–268. Krefting, L.W., Stenlund, M.H., Seemel, R.K., 1966. Effect of simulated and natural deer browsing on mountain maple. J. Wildl. Manage. 30, 481–488. Kuiters, A.T., Mohren, G.M.J., Van Wieren, S.E., 1996. Ungulates in temperate forest ecosystems. Forest Ecol. Manage. 88, 1–5. 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. 105, 65–74.

Maillard, D., 1989. Impact des cervide´s sur la re´ge´ne´ration naturelle du sapin pectine´ (Abies alba) dans les Vosges du Nord. Bilan d’une premie`re anne´e d’expe´rimentation. Bull. Mens. (Office national chasse) 141, 14–17. Maizeret, C., Ballon, P., 1990. Analyse du de´terminisme des de´gaˆts de cervide´s (Cervus elaphus, Capreolus capreolus) sur le pin maritime dans les landes de Gascogne. Gibier Faune Sauvage 7, 275–291. Martin, J.L., Baltzinger, C., 2002. Interaction among deer browsing, hunting, and tree regeneration. Can. J. Forest Res. 32, 1254–1264. McCullagh, P., Nelder, J.A., 1989. Generalized Linear Models, 2nd ed. Chapman and Hall, London, U.K., 532 p. Miller, G.R., Cummins, R.P., 1998. Browsing by red deer on naturally regenerated birch and juniper saplings on wintering ground at Glen Feshie. Scott. Forest. 52, 138–145. Miller, G.R., Kinnaird, J.W., Cummins, R.P., 1982. Liability of saplings to browsing on a red deer range in the Scottish Highlands. J. Appl. Ecol. 19, 941–951. Moore, N.P., Hart, J.D., Langton, S.D., 2000. Browsing by fallow deer (Dama dama) in young broad-leaved plantations: seasonality, and the effects of previous browsing and bud eruption. Forestry 73, 437–445. Morecroft, M.D., Taylor, M.E., Ellwood, S.A., Quinn, S.A., 2001. Impacts of deer herbivory on ground vegetation at Wytham Woods, central England. Forestry 74, 251–257. Motta, R., 2003. Ungulate impact on rowan (Sorbus aucuparia L.) and Norway spruce (Piceas abies (L.) Karst.) height structure in mountain forests in the eastern Italian Alps. Forest Ecol. Manage. 181, 139–150. Mouron, D., Boiseaubert, B., 1997. La situation du cerf e´laphe en France Mise a` jour de l’inventaire zooge´ographique des massifs a` cerf e´laphe. Bull. Mens. (Office national chasse) 218, 18–21. Palmer, S.C.F., Truscott, A.M., 2003. Browsing by deer on naturally regenerating Scots pine (Pinus sylvestris L.) and its effects on sapling growth. Forest Ecol. Manage. 182, 31–47. Parker, K.L., Gillingham, M.P., Hanley, T.A., 1993. An accurate technique for estimating forage intake of tractable animals. Can. J. Zool. 71, 1462–1465. Picard, J.-F., Ballon, Ph., Colin, G., Frochot, H., 1994. Incidence des populations de cervide´s sur la re´ge´ne´ration du sapin dans les Vosges. Rev. For. Fr. XLVI, 137–151. Pietrzykowski, E., McArthur, C., Fitzgerald, H., Goodwin, A.N., 2003. Influence of patch characteristics on browsing of tree seedlings by mammalian herbivores. J. Appl. Ecol. 40, 458–469. Price, P.W., 1991. The plant vigor hypothesis and herbivore attack. OIKOS 62, 244–251. Putman, R.J., Moore, N.P., 1998. Impact of deer in lowland Britain on agriculture, forestry and conservation habitats. Mamm. Rev. 28, 141–164. Rao, S.J., Iason, G.R., Hulbert, I.A.R., Elston, D.A., Racey, P.A., 2003. The effect of sapling density, heather height and season on browsing by mountain hares on birch. J. Appl. Ecol. 40, 626–638. R Development Core Team, 2004. R: a language and environment for statistical computing. R foundation for Statistical Computing, Vienna, Austria. http:// www.R-project.org. Reimoser, F., Armstrong, H., Suchant, R., 1999. Measuring forest damage of ungulates: what should be considered. Forest Ecol. Manage. 120, 47–58. Reimoser, F., Gossow, H., 1996. Impact of ungulates on forest vegetation and its dependence on the silvicultural system. Forest Ecol. Manage. 88, 107–119. Renaud, P.C., Verheyden-Tixier, H., Dumont, B., 2003. Damage to saplings by red deer (Cervus elaphus): effect of foliage height and structure. Forest Ecol. Manage. 181, 31–37. Saint-Andrieux, C., Wilmart, D., Bernard, S., 1999. Impacts du cerf et du chevreuil sur la re´ge´ne´ration naturelle de sapin pectine´. Bull. Mens. (Office national chasse) 247, 12–21. Saint-Andrieux, C., Klein, F., 2002. Roe and red deer diet in relation to damage to silver fir. In: Thomaidis, C., Kypridemos, N. (Eds.), Proceedings of the XXIVth Congress on International Union Game Biology. ThessalonikiGreece, September 20–24, 1999, (Hunting Fed. Macedonia-Thrace, s.d.), pp. 409–422. Scott, D., Welch, D., Thurlow, M., Elston, D.A., 2000. Regeneration of Pinus sylvestris in a natural pinewood in NE Scotland following

D. Pe´pin et al. / Forest Ecology and Management 222 (2006) 302–313 reducing in grazing by Cervus elaphus. Forest Ecol. Manage. 130, 199– 211. Seagle, S.W., 2003. Can ungulates foraging in a multiple-use landscape alter forest nitrogen budgets? OIKOS 103, 230–234. Senn, J., Suter, W., 2003. Ungulate browsing on silver fir (Abies alba) in the Swiss Alps: beliefs in search of supporting data. Forest Ecol. Manage. 181, 151–164. Sharrow, S.H., 1989. Sheep grazing as a silvicultural tool to suppress brush. J. Range Manage. 42, 2–4. Smit, A., Kooijman, A.M., 2001. Impact of grazing on the input of organic matter and nutrients to the soil in a grass-encroached Scots pine forest. Forest Ecol. Manage. 142, 99–107. Staines, B., Welch, D., 1984. Habitat selection and impact of red (Cervus elaphus L.) and roe (Capreolus capreolus L.) deer in a Sitka spruce plantation. In: Proceedings of the Royal Society of Edinburgh, Section B: Biological Sciences, vol. 82B. pp. 303–319. Takada, M., Asada, M., Miyashita, T., 2002. Cross-habitat foraging by sika deer influences plant community structure in a forest-grassland landscape. Oecologia 133, 389–394.

313

Tilghman, N.G., 1989. Impacts of white-tailed deer on forest regeneration in Northwestern Pennsylvania. J. Wildl. Manage. 53, 524–532. 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. Forest Ecol. Manage. 88, 55–63. Virtanen, R., Edwards, G.R., Crawley, M.J., 2002. Red deer management and vegetation on the Isle of Rum. J. Appl. Ecol. 39, 572–583. 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 64, 61–82. Welch, D., Staines, B.W., Scott, D., French, D.D., 1992. Leader browsing by red and roe deer on young Sitka spruce trees in Western Scotland. II. Effects on growth and tree form. Forestry 65, 309–330. Wolfinger, R., O’Connel, M., 1993. Generalized linear mixed models: a pseudolikelihood approach. J. Statist. Comput. Simul. 48, 233–243.