Résumé
Nous avons mesuré l’influence de la densité de population sur les déplacements et le budget d’activité de cerfs de Virginie se trouvant à différentes densités en milieux contrôlés et naturels. Le budget d’activité, les déplacements et la biomasse de plantes disponibles ne variaient généralement pas selon les densités contrôlées (7.5 et 15 cerfs/km²). Cependant, nous avons trouvé des différences interannuelles reliées à l’augmentation de l’abondance de végétation après coupe. En effet, suite à l’augmentation de l’abondance de végétation dans les enclos à densités contrôlées, la durée des périodes passées en activité diminuait et les cerfs augmentaient le nombre de périodes passées en activité par jour. Au cours de l’été, lorsque la végétation a augmenté en abondance, les cerfs adultes à 7.5 cerfs/km² diminuaient la proportion du temps passé en activité par jour mais pas à 15 cerfs/km². Au début de l’été en milieu naturel (>20 cerfs/km²), la végétation est peu abondante et les cerfs ont diminué le temps passé en activité possiblement afin d’augmenter le temps de rumination d’une végétation de moins bonne qualité.
Abstract
Foraging decisions in herbivores may be affected by population density as this factor is frequently related to changing availability of preferred plant species, net abundance of biomass and variation in intra-specific competition. We studied the effect of population density on white-tailed deer movements and activity budgets using a controlled-density experiment. The controlled densities (7.5 deer/km² and 15 deer/km²) were obtained by placing 3 deer in 2 enclosures of different sizes (20 ha and 40 ha) where forest was partially harvested in 2001. We repeated the experiment with the same two densities in 3 different locations. Summer activity budgets and movements of yearlings and adults were quantified by VHF telemetry the first, second and third year after the onset of the controlled-density experiment. During one year, we also measured the activity budget of 4 adults in an unfenced area at a density of >20 deer/km². In each enclosure, biomass of 13 of the most abundant and preferred plant species was measured in 40 plots every year.
Adults were less active than yearlings at 7.5 but not at 15 deer/km². Otherwise, movements, activity budgets and available biomass were similar between 7.5 deer/km² and 15 deer/km². However, the available biomass increased through years and activity budgets varied accordingly. As biomass increased, deer increased the number of daily activity bouts, which also became shorter. With increasing biomass, it seemingly took less time for deer to obtain sufficient forage to enter a rumination bout. Adult deer decreased time spent active throughout summer, but only at 7.5 deer/km². However, adults in unfenced cutblocks were less active at the beginning of the summer than deer at 7.5 deer/km² and increased time spent active through summer. When vegetation was less abundant, such as in early summer, deer at 7.5 deer/km² seemed to spend more time active gathering vegetation. In unfenced areas at high density, forage was even less available, free-ranging deer in early summer had to increase processing time to extract available energy, and thus decrease the proportion of time spent active. This study demonstrates that relationships between density and foraging behavior are complex and that controlled-density experiments may help to understand the behavior of herbivores in relation to available resources.
Introduction
Summer is a critical season for herbivores in temperate and boreal regions to restore body condition and build up body reserves for over-winter survival (Putman et al. 1996, Lesage et al. 2001). Maximising energy intake requires that herbivores utilize the most profitable plants in terms of energy content and foraging time necessary to crop them (Bunnell and Gillingham 1985). Herbivores have access to less forage as population density increases (Healy et al. 1997, Côté et al. 2004), hence, foraging strategies may become even more critical at high density. Deer have become the dominant herbivores in most ecosystems of North America and Europe and they have recently reached historic high densities over large areas (Côté et al. 2004). The impacts of deer on ecosystem functioning are far reaching (Côté et al. 2004) and the influence of population density on deer foraging behavior therefore needs to be assessed.
Deer may change their behavior in response to changing availability of preferred plant species, net abundance of forage biomass or social competition in relation to population density. Many studies have obtained conflicting results regarding the influence of forage quality and availability on herbivore behavior (Henriksen et al. 2003). Usually, herbivore density is negatively related to forage availability and ungulates at high density may be constrained to remain active for longer periods in order to ingest enough forage (Trudell and White 1981, Moncorps et al. 1997). Deer may also increase search movements and foraging time with increasing density to consume the most rewarding plants and plant parts (Wickstrom et al. 1984, Bartmann et al. 1992). However, as forage is distributed in spatially separated patches, movements and increased foraging time may also entail higher time and energy costs (Murray 1991). Alternatively, deer may also respond to forage depletion by foraging less selectively to reduce movement costs (Gates and Hudson 1983), and their diel active time would then remain unaffected (Kohlmann and Risenhoover 1994). By feeding on lower quality vegetation, i.e. plants with a high content of structural compounds that reduce plant digestibility (Bryant and Kuropat 1980, Palo 1985, Côté 1998), the rate of passage of forage from the rumen to the lower digestive tract should slow down and rumination time may then increase (Van Soest 1982, Spalinger et al. 1986).
Studies have shown that yearlings spend more time active than adults do as is expected given their smaller mass and relatively smaller digestive system, higher metabolic rate and growth energetic demands (Bunnell and Gillingham 1985, Côté et al. 1997, Shi et al. 2003).
As energetic demands increase allometrically (W0.75) with body size, larger individuals
need less energy per unit of mass than smaller individuals (Illius and Gordon 1987), and because gut size increases linearly with body size and turnover time declines, this allows larger individuals to extract more energy from lower quality forage than smaller individuals (Demment and Van Soest 1985). Active time thus tends to decrease with increasing body size (Moncorps et al. 1997, Ruckstuhl 1997, Mysterud 1998, Pérez-Barbería and Gordon
1999, Jeschke and Tollrian 2005). Given their different use of resources, an increase in
population density should thus affect differently juveniles and adults. For example, fawn survival is more affected by density than adult survival (Jorgenson et al. 1997).
As vegetation increases in abundance during the growing season, it also becomes more lignified and its protein content decreases (Hanley 1984). Time spent active could thus differ between periods of plant growth and periods of plant senescence (Gates and Hudson 1983). Diel-patterns of activities have also been largely documented and studies have shown that deer usually synchronize activity bouts with dawn and dusk (Kammermeyer and Marchinton 1977, Beier and McCullough 1990) and shift their activity periods to times when weather conditions are most favourable for thermoregulation (Beier and McCullough 1990).
Controlled-density experiments have been used as a research tool for many years to study the foraging behavior of domestic animals and its use is now strongly encouraged for wild ungulates (Hester et al. 2000, Gordon et al. 2004). Most studies have investigated the effects of browser densities on vegetation abundance and diversity (Tilghman 1989, Hester et al. 2000); however, controlled-density experiments may also be very useful to understand how browsers modify their behavior at different population densities. Our general objective was to assess how the daily and summer activity patterns of yearling and adult white-tailed deer vary in relation to population density. We predicted that yearling and adult deer at high density would be more active and have higher movement rates than those at lower densities
because of the increased time necessary to gather forage at high density. Alternatively, yearling and adult deer could increase time spent inactive at high density to process vegetation that is more fibrous. Total active and inactive times would then remain similar in all densities but the length of inactive bouts would increase.
Study area
Anticosti Island (49° 28’ N, 63° 00’ W) is located in the Gulf of St. Lawrence, Québec, Canada and covers 7,943 km2. Forests are naturally dominated by balsam fir (Abies balsamea), white spruce (Picea glauca) and black spruce (P. mariana). White birch (Betula papyrifera) and trembling aspen (Populus tremoloides) are irregularly found on the island. Around 200 deer were introduced on the island at the turn of the 19th century. The population spread and grew rapidly because of the absence of predators and the presence of natural and human disturbances that created openings favourable to deer. Today, deer densities of >20 deer/km² are found in many areas on the island (Potvin and Breton 2005). Forest composition has been strongly modified by selective browsing, deciduous browse species have almost disappeared and balsam fir stands are now being replaced by white spruce stands (Potvin et al. 2003). The climate on Anticosti is typically maritime and characterized by long and milder winters than on the continent (Huot 1982). Mean temperatures are -12°C in January and 15°C in July and an average of 406 cm of snow and 630 mm of rain falls every year on the island (Environment Canada 1993).
Methods
Experimental design
Our experimental design is made of 4 blocks, three of which (A, B, C) consist in two enclosures of different sizes (20 ha and 40 ha) in which we introduced deer. The last block (T) is an unfenced area where density was estimated at >20 deer/km². We located all blocks in balsam fir-dominated forests that were partially cut in the summer of 2001. Two of the blocks (B, C) were localised in the center of the island (Jupiter River area); the other two were located 130 km away in the western part of the island (A, T). Between 30 and 40% of residual forest patches of different sizes (0.19–21.6 ha) were left uncut in the enclosures.
Water was easily accessible to deer in streams or artificial water holes in every enclosure. To assess the effects of population density on activity and movement of deer, 2 controlled densities were established in blocks A, B and C. Controlled densities were 7.5 deer/km2 (LDE; 40 ha enclosures with 3 deer) and 15 deer/km2 (HDE; 20 ha enclosures with 3 deer). We chose these deer densities to cover a gradient that would include white-tailed deer density levels proposed for sustainable tree regeneration (7 deer/km²; Tilghman 1989, deCalesta and Stout 1997), the estimated density on Anticosti Island at the beginning of the experiment (15.6 deer/km²; Rochette et al. 2003) and the local estimated in situ density level in management areas adjacent to experimental blocks (>20 deer/km²). Our set up is part of a larger study trying to determine which deer densities are compatible with forest regeneration (Tremblay et al. in prep.).
Deer captures
We used different methods to capture deer: dart guns (Pneu-dart Inc, Williamsport, Pennsylvania, USA), netguns (Coda Enterprises Inc., Mesa, Arizona, USA) shot from a helicopter, Stephenson box traps and cannon nets baited with cattle feed and balsam fir twigs. In June or early July of the first, second and third year after the onset of the controlled-density experiment, deer were released in the study enclosures. On the second year after the onset of the controlled-density experiment, we captured and released 4 adult females fitted with VHF collars in the unfenced cutblocks (Table 2-1). The Animal Care and Use Committee of Université Laval, Québec, Canada (Reference number 2005–008) approved all capture methods.
All deer were fitted with VHF collars equipped with sto-2a variable-pulse activity sensors (LMRT series, Lotek Engineering, Newmarket, Ontario, Canada). One adult male and 1 yearling male lost their collars and 1 yearling male had a malfunctioning collar. We verified reproductive status of adult females by direct observation at capture and at the end of summer. As only 2 reproductive females were monitored, we did not include reproductive status as a variable in the analyses but verified if activity and movements were comparable to the other females.
Table 2–1. Characteristics of white-tailed deer used in an experiment on the effects of population density on deer activity budgets on Anticosti Island, Québec.
a Number of years since the onset of the controlled-density experiment.
b The number of days for which activity budgets of radio-collared deer were monitored during each month and each year of the study.
c These females were observed with a fawn.
Deer no. Yeara Block Density (deer/km²) Age class Sex Number of days monitoredb
July August September
1 1 A 7.5 Adult Male 21 19 2 Yearling Male 21 19 3 Yearling Female 21 9 4 15 Yearling Female 21 19 5 Yearling Female 21 19 6 Yearling Female 20 19 7 2 A 7.5 Adult Femalec 15 22 13 8 Yearling Female 14 22 13 9 Adult Female 15 22 13 10 15 Adult Female 14 22 13 11 Adult Male 15 22 13 12 Yearling Male 13 2 B 7.5 Yearling Female 14 Yearling Male 15 Adult Female 16 15 Adult Female 17 Yearling Female 18 Adult Femalec 19 2 C 7.5 Yearling Female 15 7 5 20 Adult Female 14 8 5 21 Yearling Male 14 7 5 22 15 Adult Male 15 8 5 23 Yearling Female 13 8 5 24 Adult Male 25 2 T >20 Adult Female 0 2 5 26 Adult Female 9 8 8 27 Adult Female 9 6 4 28 Adult Female 9 9 8 29 3 A 7.5 Yearling Male 30 Adult Male 31 Yearling Male 32 15 Yearling Female 33 Yearling Male 34 Adult Male 35 3 C 7.5 Yearling Female 8 13 1 36 Yearling Female 0 12 1 37 Adult Male 7 10 38 15 Adult Male 13 29 1 39 Yearling Male 13 31 2 40 Yearling Male 13 30 2
Forage abundance
To assess plant biomass available in the enclosures, we randomly placed 20 sampling points in cuts and 20 points under forest cover in each experimental unit. At each sampling point, percent of plant cover was estimated in 2 1-m2 plots randomly located in a 10×10 m quadrat centered on the sampling point. We quantified biomass of the following species: Abies balsamea, Betula papyrifera, Cirsium spp., Coptis groenlandica, Cornus canadensis, Epilobium angustifolium, Grass sp., Hieracium sp., Maianthemum canadense, Picea glauca, Rubus idaeus, Rubus pubescens and Trientalis borealis. Plant biomass was assessed using regressions between percent plant cover and mass of dried plants (Bonham 1989). Number of samples needed for regressions was estimated empirically by plotting regression coefficients with number of samples until an asymptote was reached (Frontier 1983). We did not assess forage abundance in the experimental unit located outside the enclosures but considered it comparable to the forage abundance of the experimental units during the first year of the experiment.
Movements
In July and August 2002, the first year of the controlled-density experiment, we radiotracked 6 deer in block A and in the second year after the onset of the experiment, 18 deer were tracked in 3 blocks (A, B, C). Deer were located with receivers (SRX-400 version W9, Lotek Engineering, Newmarket, Ontario, Canada and TR-2 scanner/receiver, Telonics, Meza, Arizona, USA), a unidirectional yagi antenna and a compass. Telemetry stations were positioned with a GPS Garmin (Garmin international, Olathe, Kansas, USA; precision of <5 m) on forest roads adjacent to the enclosures. To limit human disturbance, stations were generally located more than 100 m away from the enclosures. At least 3 azimuths differing by a minimum of 30º were obtained by moving between stations with a vehicle (White and Garrott 1990). To reduce location error, positioning had to be completed within 15 minutes (White and Garrott 1990). Periods of the day were evenly sampled by separating them into 3 periods of 8 hours (8h00–16h00, 16h00–0h00 and 0h00–8h00). These 8-hour periods were rotated between 2 observers and between groups of
enclosures every 3 days to evenly sample the complete 24 hours of a day and all the enclosures.
LOAS software (Location of a Signal; Version 2.07, Ecological Software Solutions, Schwägalpstrasse, Urnäsch, Switzerland) was used to estimate positions and error polygons. Error polygons were calculated with “Andrews” estimator. All locations were plotted with LOAS on maps and were assigned Universal Transverse Mercator (UTM) coordinates. The average error from plotted to actual locations was determined by placing VHF collars at known locations throughout the enclosures and was estimated at 107 m (SE = 88 m; n = 88 trials). We deleted telemetry locations with error polygons greater than 0.1 ha. After processing, 2,916 usable locations amongst the 3,251 recorded were obtained. The minimum movement rate was estimated as the linear distance between two successive deer-locations separated by less than 3 hours divided by the time elapsed between these 2 locations. One deer (deer #3 in Table 2-1) was also fitted with a GPS collar to verify the influence of positioning error on movement rate estimation. The distance moved per hour was equivalent between VHF and GPS collars (VHF: x = 241.1 ± 12.6 m/hour; n = 196 movements; GPS: x = 247.4 ± 7.8 m/hour; n = 944 movements; F1,1138 = 0.12; P = 0.73). The distance moved was slightly related to the time interval between two locations (r = 0.11; P < 0.01; n = 2,219 movements) and the mean time interval between positions was 1h38 (SE = 27 min., n = 2,219 movements).
Activity budgets
Variable-pulse activity sensors of VHF collars use mercury switches that add pulses to the base pulse rate of the collar each time the switch is triggered. The number of pulses above the base pulse rate indicates the degree of animal activity during the period when the pulses were counted (Type STO-2A, Lotek Engineering, Newmarket, Ontario, Canada). Transmitter signals of activity sensors were received and recorded in a SRX-400 version W9 receiver-datalogger (Lotek Engineering, Newmarket, Ontario, Canada) connected to a multidirectional antenna, a 12 V battery, and a solar panel.
The receiver was programmed to measure the duration between 2 pulses for 65 consecutive pulses, record mean pulse rate and then automatically switch to scan another transmitter. The time needed to record 65 pulses was thus dependent on pulse rate. As the SRX receiver scanned one transmitter at a time and because 6 individuals were monitored each day, a measure of pulse rate for each deer was obtained approximately every 6 minutes. Data were downloaded in a portable computer with the help of Winhost software (version 1.0.0.1, Lotek Engineering, Newmarket, Ontario, Canada).
Validation studies of activity sensors have obtained mixed results in overall reliability (74 to 98% accuracy) and have demonstrated that it is necessary to validate methods used to measure activity budgets with direct animal observations (Gillingham and Bunnell 1985, Beier and McCullough 1988, Relyea et al. 1994). We conducted our own validation study with direct observations of deer in small enclosures (Coulombe et al. 2006). By combining the information of 3 successive scans, we correctly assessed 87% of all activity bouts (Coulombe et al. 2006). An inactive bout began when at least 3 inactive scans were observed. To return to an active bout, at least 3 active scans had to be observed. Activity data did not allow us to differentiate amongst different active or inactive behaviors, e.g. resting could not be differentiated from ruminating or eating from moving. However, as it was shown for the Odocoileus genus, activity periods not corresponding to feeding activities (e.g. vigilance, social interactions) represent only 5 to 15% of time spent active (Beier and McCullough 1990, Gillingham et al. 1997) and thus are not an important part of the activity budget. Additionally, a decrease in plant quality is generally related to an increase in rumination time, and simultaneously to an increase in time spent inactive (Mysterud 1998, Pérez-Barbería and Gordon 1999). Time spent inactive is thus an indication of rumination time.
At the onset of the controlled-density experiment, in July and August 2002, one block (A) with 2 densities was available for the study of activity budgets (Table 2-1). The second year, in July, August and September 2003, 2 blocks (A, C) were studied. The third year, in July, August and September 2004, we monitored activity budgets of deer in 1 block (C). Another unit (T) was located in an unfenced area where density was estimated by an aerial
survey to >20 deer/km2. We monitored the activity budget of these free-ranging deer from July to September 2003. To analyse time budgets, we used the proportion of time spent active, the length of active and inactive bouts and the number of activity bouts per day.
Analyses
We first tested if plant biomass available to deer differed between densities (7.5 deer/km², 15 deer/km²), strata (clear-cuts, forests) and years since the onset of the controlled-density experiment, using an analysis of variance with block as a random factor. We then contrasted the mean distance moved and the proportion of time spent active between densities, periods of the day (dawn: 1h30 before sunrise to 1h30 after sunrise, day: 1h30 after sunrise to 1h30 before sunset, dusk: 1h30 before sunset to 1h30 after sunset and night: 1h30 after sunset to 1h30 before sunrise) and week with block and year as random factors. We also used week2 in certain analyses (because plant quality first increases but then decreases throughout the summer) but if the quadratic term did not significantly explain extra variability, it was removed. Periods of the day and weeks were repeated for each deer,