Institute of Arctic Biology, and Alaska Cooperative Fish and Wildlife Research Unit, University of Alaska, Fairbanks, Alaska 99775, USA

Institute of Arctic Biology, University of Alaska Fairbanks,
Fairbanks, Alaska 99775, USA

Coastal Resources Inc., Vista, California, USA

We use three analytical techniques to examine home-range dynamics of river otters in Prince William Sound, Alaska, USA, from February 1997 to January 1998, and discuss problems with analysis of linear home ranges. River otters inhabiting marine environments where fish were abundant had smaller home ranges than animals living in freshwater systems with fewer prey, whereas otters using multiple salmon runs had larger home ranges than otters in other habitats.

River otters (Lutra canadensis) in Prince William Sound, Alaska, USA, occur at densities of 28-80 animals/100 km of shoreline (Testa et al. 1994), with home ranges that encompass about 20-40 km of shoreline (Bowyer et al. 1995). These otters forage intensively in intertidal and subtidal zones where they prey principally on marine fish (Larsen 1984, Bowyer et al. 1994, Ben-David et al. 1996, 1998). River otters living in this marine environment form large social groups (<18 individuals; Rock et al. 1994, Testa et al. 1994), but little is known about their spatial relationships or social structure.

We tested the hypothesis that home-range dynamics and other spatial relationships of otters would be related to availability and distribution of their primary prey (fish). We present telemetry analyses using three techniques for calculating home-range area because no single method is suitable for all purposes–the technique most appropriate to use depends upon the hypotheses being tested (Harris et al. 1990) and data collected. Estimates of kernel density define a utilization distribution by assessing the probability that an animal will occur at particular points in space. Kernel estimators are nonparametric and can estimate densities of any shape (Seaman and Powell 1996) by supplying a third dimension representing the amount of time an animal spent in any given area (Seaman et al. 1999); thus these methods are useful for examining the internal structure within home ranges, particularly core areas of use that may be important for foraging or den sites. Kernel estimates, however, are sensitive to autocorrelation (Harris et al. 1990) and changes in smoothing parameters (Worton 1995), and may result in over estimation of the area used (Seaman and Powell 1996). Furthermore, kernel estimates were developed for analysis of spatial data occurring in two dimensions and are problematic for analysis of animals with primarily unidimensional patterns of movement such as river otters, which use a narrow aquatic-terrestrial ecotone (Sauer et al. 1999). Minimum convex polygons (MCP) do not have underlying assumptions of distribution, are not affected by autocorrelation (Harris et al. 1990), and are the oldest and most common method for estimating home ranges (White and Garrott 1990, Seaman et al. 1999). The MCP technique uses the outer points in the spatial distribution to define the boundaries of the home range and may contain large areas that are never used, especially for animals that move in unidimensional space.

For purposes of comparison with other studies of river otters (Green et al. 1984, Reid 1994), we present area calculations for MCP, and Adaptive Kernel (ADK) estimates. We also use Geographical Information System (GIS) ArcInfo, (Redlands, California) to calculate kilometers of shoreline within each of these area estimates using a method described by Sauer et al. (1999).

We live-captured river otters using both Hancock and leg-hold traps (Blundell et al. 1999) in spring and early summer 1996-1997. Traps were placed in blind sets (i.e., no bait or lure) on trails at latrine sites and monitored by means of trap transmitters (Telonics, Mesa, Arizona, USA) that signaled when a trap had been sprung. River otters were anesthetized with Telazol (9mg/kg; A. H. Robins, Richmond, Virginia, USA) administered by hand injection for otters captured in Hancock traps and with Telinject darts and a blowgun for otters captured in leg-hold traps. We surgically implanted river otters with telemetry transmitters (Model IMP/400/L, Telonics, Mesa, Arizona) inserted into the peritoneal cavity through an incision made on the right side, posterior to the last rib. Each muscle layer was closed separately with simple-interrupted sutures, and the skin was closed with a continuous subcuticular suture line. As a final precaution, the skin incision was sealed with surgical glue. We implanted 17 otters (12 males, 5 females) in 1996 in the Jackpot Bay area and 8 river otters (5 males, 3 females) in 1997 in this area. Twelve otters (8 males, 4 females) were implanted with radio transmitters in Herring Bay in 1997. All radio tracking in 1996 was conducted from a boat, resulting in only partial home-range information for otters using freshwater systems. For this reason, we only report data collected using aerial tracking. Otters were radio tracked from a small fixed-wing aircraft from February 1997 to January 1998 (n = 29 occasions). Once a telemetered otter was located, Geographic Positioning System (GPS) coordinates were recorded by flying the plane directly over the location and recording latitude and longitude. Additionally, point locations for each otter were plotted on United States Geological Survey (USGS) maps (1:63360 scale) to provide a secondary source of location information in the event of an error in recording GPS locations. When otters were observed engaging in foraging activity, their location, distance from shore, and group size was recorded. All methods used in this research were approved by an Institutional Animal Care and Use Committee at the University of Alaska Fairbanks.

We conducted scuba-diving transects in July of 1996 and 1997 in both study areas to assess fish abundance at otter latrine sites (n = 15 sites/year) and at random sites (n = 15 sites/year). Fish were counted along two 30-m transects/site and categorized into eight family groups and three size classes (<8 cm, 8-15 cm, >15 cm). We also assessed six random and six latrine sites in the freshwater system.

We used CALHOME (Kie et al. 1996) to estimate home ranges (ADK and MCP) and GIS-ArcInfo to calculate shoreline distances within these polygons (Sauer et al. 1999). We arbitrarily selected 50% ADK contours to examine core areas of use.

We obtained sufficient locations (n = 657 locations; 0 = 25 locations per otter) to assess home-range size for 29 river otters (20 males, 9 females) from February 1997 to January 1998. River otters inhabiting our study areas used three general habitat-prey associations: marine, fresh water, and areas with salmon runs. Prey abundance in the marine system did not differ between study areas, so we combined data from both areas to test for differences in sizes of home range between otters using different habitat-prey associations, and differences in home-range size between genders.

River otters inhabiting marine environments, where fish were abundant, had smaller home ranges than otters living in freshwater systems with fewer prey, whereas otters using multiple runs of salmon, which were geographically dispersed, had larger home ranges than otters in either marine or freshwater habitats (Table 1). Shoreline within the freshwater habitat is underrepresented because locations in secondary and tertiary tributaries did not result in creek shoreline being measured near these locations.

Table 1. Differences in home ranges for river otters inhabiting different habitats in Prince William Sound, Alaska, USA, (February 1997 – January 1998), pooling data from both sexes and both study areas (P-values from one-way ANOVA).

Males had significantly larger home ranges for both 95% estimates and core areas (50%) than did females in both marine and freshwater environments (Table 2), but the proportion of the 95% area contained within the core area tended to be greater for females than for males (Table 3).

Table 2. Difference in size of home ranges for female and male otters in Prince William Sound, Alaska, USA, (February 1997 – January 1998), pooling data from all habitats and both study areas (P-values from one-way ANOVA).

Table 3. Proportion (0 50% ADK/0 95% ADK) of the shoreline distance in the entire home range (95% Adaptive Kernel; ADK) represented in the core area of use (50% ADK) for otters in our study areas in Prince William Sound, Alaska, USA, (February 1997 – January 1998).

River otters in Prince William Sound exhibited intersexual overlap of home ranges but intrasexual patterns differed between the genders. Female otters had low spatial overlap and most appeared to have exclusive core areas of use, whereas male otters showed a substantial overlap in home-range areas including overlap of male-group home ranges with those of other male groups and with solitary males (50% Adaptive Kernel; Figure 1).

Each method of home-range analysis showed similar trends although MCP estimates were more conservative in both area and shoreline estimates (Table 1 and Table 2) than ADK estimates. Standard methodologies for calculating home-range areas may not be appropriate for use in river otters because otters use narrow strips of habitat associated with the aquatic-terrestrial ecotone (Figure 2).

Figure 1. Core areas of use (50% Adaptive Kernel) for female (A) and male (B) river otters in Prince William Sound, Alaska, USA (February 1997 - January 1998). In most instances core areas for females did not overlap. Males (both solitary and males in groups) had substantial overlap in core areas.

Figure 2. Adaptive Kernel contours (95% and 50%) and minimum convex polygons (95%) for two otters in Prince William Sound, Alaska, USA (February 1997 - January 1998), showing different patterns of movement. Symbols are the telemetry locations for each otter.

We suggest that the social structure of river otters (males primarily in groups, females mostly solitary) is resource related and has a strong influence on spatial relationships, home-range size, and diet. Spatial organization in solitary Carnivora is believed to be resource related (Sandell 1989), with distribution of females determined by food availability, and male distribution, at least during mating season, dependent upon female dispersion. Female river otters tended to have larger core areas relative to total home-range area (Table 3), suggesting that a larger proportion of their home ranges may be important for foraging. Our sample size for females, however, is small and we may not have a complete representation of space use for this gender.

The sexual dimorphism in our system is not pronounced. Males range from 10-13% larger than females in length:weight ratios, yet their home ranges in all habitats ranged from two to ten times larger than those of females (Table 2). We suggest that larger home ranges for males is more likely related to females as a resource, because males use areas much larger than would be needed to support their metabolic needs (McNab 1963, Sandell 1989). Moreover, a smaller proportion of their total area is contained in the core for males compared with females, the area presumably of greatest importance to individuals. Additionally, we suggest that male otters traveling in groups may be foraging cooperatively. Indeed, the otters using numerous salmon runs were a group of males that traveled together once the salmon started to spawn. These males traveled greater distances than males using other resources, but salmon runs provided rich sources of prey, likely compensating for the distance traveled.

Seaman and Powell (1996) concluded that Adaptive Kernel estimates resulted in overestimation of area, and this may be occurring in our results here (Figure 2). Nonetheless, the comparison of relative size of home ranges between genders and different habitat-prey associations that we present herein still provide valid assessments, and information on spatial relationships is largely independent of analysis technique.

We hypothesize that the distribution of prey, rather than simply abundance of forage, has a substantial effect on spacing behavior of otters. We will further explore this and similar hypotheses by testing for seasonal shifts in home ranges using additional radio telemetry locations and we will investigate seasonal variation in abundance of marine fish at otter latrines and random locations. We also will use microsatellite DNA to examine the effect of genetic-relatedness on spatial relationships of river otters, and perhaps gain some insight into which males (social or solitary) gain reproductive opportunities.

We thank our pilot J. De Creeft, also J. K. Maier for GIS analyses, J. Kie for helpful suggestions, and L. Faro, P. Berry, S. Andersson, C. Durham, and H. Golden for assistance in the field. Funding was provided by the Exxon Valdez Oil Spill Trustees Council, the Institute of Arctic Biology at the University of Alaska Fairbanks (UAF), the Alaska Cooperative Fish and Wildlife Research Unit at UAF, and the Alaska Department of Fish and Game.

The Siberian flying squirrel is a herbivorous, nocturnal, and arboreal rodent living in Eurasian boreal coniferous forests. A total of 53 adult and 22 juvenile flying squirrels were radio tracked in southern Finland during 1996–1998. Each animal was fitted with a radio collar and tracked with a portable receiver. Presented here are the methods used in capturing and tracking together with the home-range size of adults and natal dispersal of juveniles.

The Siberian flying squirrel (Pteromys volans L.) is an inhabitant of coniferous boreal forest from Finland to eastern Siberia and Japan (Ognev 1966, Wilson and Reeder 1993). In Western Europe the flying squirrel occurs only in Finland and in small numbers in the Baltic countries. It is mostly nocturnal and arboreal, roosting and nesting in tree cavities and dreys (nests made of twigs, mosses, and lichens on tree branches). The food of the flying squirrel mainly consists of the leaves of deciduous trees in summer, and catkins of birch and alder supplemented with buds of both coniferous and deciduous trees in autumn and winter (Mäkelä 1996). In autumn it stores catkins in tree or rock cavities and on spruce branches (Sulkava and Sulkava 1996).

The flying squirrel population in Finland has declined during recent decades (Hokkanen et al. 1982). Therefore, in the Red Data Book, the flying squirrel has been classified as a declining species with a need for monitoring its population abundance (Rassi and Väisänen 1987).

This paper describes the radio-tracking methods used for the Siberian flying squirrel and presents results on home-range sizes and natal dispersal. More detailed results on home ranges, movements, and habitat and nest-site use are available elsewhere (Hanski 1998, Hanski et al. 1999, Hanski et al. unpublished manuscript). These are the first studies where the results on the spatial behavior of Siberian flying squirrels outside their dens have been presented. Comparable radio-tracking studies on two closely related species of flying squirrels belonging to the genus Glaucomys have been done in North America (Bendel and Gates 1987, Fridell and Litvaitis 1991, Witt 1992).

Siberian flying squirrels were studied in two areas in southern Finland during 1996–1998 (Figure 1). The main tree species in the study areas are Norway spruce (Picea abies), Scots pine (Pinus sylvestris), birches (Betula pendula and B. pubescens), aspen (Populus tremula) and alders (Alnus incana and A. glutinosa). Study area A was in Iitti, (60° 55’ N, 26° 30’ E) in managed coniferous forests. The terrain is undulating varying from 65-110 m above sea level (a.s.l.). These forests are dominated by spruce, owned by private landowners, and intensively managed. In mature stages spruce forests reach heights of 25-28 m. Forest stands are surrounded by clearcuts, sapling stands, young forests of various ages, and to a lesser extent, by pine bogs. Study area B was in Nuuksio National Park, Espoo (60° 18’ N, 24° 32’ E) and surrounding managed forests. The terrain is rugged with steep cliffs or undulating slopes rising to hilltops and ridges with open pine forest. The height varies from 30-100 m a.s.l. Forests are relatively continuous and are interrupted by a few bogs, lakes, and fields. These forests consist of patches of spruce-dominated, pine-dominated and deciduous- dominated forest types. The mean temperature in both study areas in January is -8°C with 43 mm of precipitation as snow. In July the mean temperature is +17°C with 73 mm of precipitation. Snow cover lasts from mid November to the last half of April. The day length varies from 5.5 h in December to 19 h in June.

We first used bait traps to capture flying squirrels but this failed. Instead, each spring, potential nest sites of flying squirrels were located by searching for aspens with a collection of yellow-brown feces at the base. Once a potential nest was located, flying squirrels were trapped by placing a Perspex trap over the entrance of the nest cavity. When an animal came out it fell down into the trap. The trap was checked 1 h after sunset and all traps were removed 4-6 h after sunset. Captured animals were ear-tagged, sexed, and weighed. Animals in the home-range analyses were adults, i.e. born the previous summer or earlier. The juveniles captured were two or more months old.

Figure 1. Study areas Iitti (A) and Nuuksio (B) National Park in southern Finland.

To each animal we fitted a TW-4 radio collar (230-231 MHz, from Biotrack, UK) weighing 5.5 g. The weight of the tag represented on average 3.7% of the body weight of females and 4.3% of males. The radio tag had a 4-mm wide brass-loop collar with a nut and bolt fastening. Two types of material were used to cover the collars. In 1996 and 1997 the brass collar was covered with soft heat shrink sleeving. This material proved satisfactory in summer when adult flying squirrels are solitary. However, in autumn when several animals may gather in the same cavity for roosting, it appeared that other animals chewed the heat shrink, sometimes leaving only the bare brass collar. In adults the collar still remained around the neck. However, juveniles are smaller in their first summer and when a mother chewed the collar, she enlarged the loop and often the collar slipped off the neck. To prevent chewing, we glued cayanne pepper on the outer part of the collar as suggested by Stuart-Smith and Boutin (1995) and Adams and Campbell (1996), but without any success. In 1998 Biotrack produced a new covering material. The inside part of the brass loop was lined with sticky-back Velcro leaving the outer part of the collar bare brass. This did not seem to have any negative effects on squirrels, but prevented chewing and allowed us to successfully radio tag juveniles. No anaesthesia was used when handling and tagging flying squirrels. The entire handling and tagging procedure took about 15 minutes.

The animals were located once a night up to five times a week from March to December with RX-81 or RX-8910 receivers and two- or three-element Yagi antennas (from Televilt, Sweden). During tracking, we followed the signal with a portable receiver until we were within 15-20 m of the animal. When an approximate position of the squirrel was found, we took bearings from several directions around the site until the animal was located in a single tree, or a small group of trees. The location was marked with a flag and the map coordinates were later obtained with a portable Global Positioning System (GPS) unit. The range of radio signals in optimal conditions was up to 1.5–2 km and the battery life of the collars was 6-7 months.

In addition to nocturnal tracking, we checked the locations of radio-tagged animals in daylight at least once a week to keep track of their nesting and roosting sites and to determine if squirrels were active in daylight. In the home-range calculations, only the fixes when an animal was outside of its den were included. When tracking, the observer did not seem to disturb the animals, because in almost all cases the animal stayed in the tree where it was first located, and when seen, appeared to be undisturbed and continued foraging in the foliage.

Home ranges were analyzed using the Ranges V computer package (Kenward and Hodder 1996). We calculated home-range size by the 100% and 95% minimum convex polygon method (MCP). The 100% MCP we present includes all fixes made for each animal both outside and inside its nest, and the 95% MCP includes only the fixes made when an animal was outside its nest. We estimated the 95% MCP using the arithmetic mean algorithm (Kenward and Hodder 1996). Cluster analysis (Kenward 1987, Kenward and Hodder 1996) was used to define core areas of high activity. The clustering is based on the nearest-neighbor distance of fixes and borders one or more patches that are most frequently used by an animal. The core areas were based on 85% of the fixes made outside the nest. For detailed description of home-range calculation methods, see Hanski 1998, Hanski et al. 1999, and Hanski et al. unpublished manuscript.

Based on these results, radio telemetry seems to be a suitable method for studying the behavior of the Siberian flying squirrel. Radio tagging with the collars as described worked well. Among 53 radio-tagged adults, a radio collar was associated with only one death. One male became stuck at the entrance of a cavity. The cavity could not have been used as a nest cavity because it was too small to enter even without the radio collar. The tagging method also seems to be appropriate for juveniles as soon as they have gained approximately 70-g weight (less than half of the average weight of adult females; Hanski et al. unpublished manuscript).

The maximum range of signals (1.5–2 km) was less than the maximum distance between the most distant points in the home ranges of adult males or less than the maximum distances they moved at night (Hanski et al. unpublished manuscript). However, this did not cause serious problems. If the signal could not be heard at one site, the squirrels could be located by tracking them from the hilltops. The limited signal range was more problematic among juveniles. When dispersing, the juveniles could quickly move 3-4 km from the natal site. The only way to locate them was an extensive search over a large area around the natal home range.

For males the mean size of 100% MCP was 59.9 ± 41.1 ha, the mean of 95% MCP was 39.6 ± 30.6 ha, and the mean of 85% cluster was 5.4 ± 4.5 ha (Table 1). Females had the mean 100% MCP of 8.3 ± 7.3 ha, the mean 95% MCP of 5.7 ± 5.2 ha, and the 85% cluster of 0.9 ± 0.7 ha (Table 1). Variation, especially among males, was large but the results also showed a large intersexual difference in the home-range size. The home-range size of males was significantly larger than that of females at all levels.

Table 1. The mean and range of home-range sizes (ha; 100%, 95% multiple convex polygons [MCP] and 85% cluster) for adult female and male flying squirrels.

The home ranges of Siberian flying squirrels were several times larger than those of the North American species. In the northern flying squirrel (G. sabrinus) the home ranges measured by 95% MCP were on average 3.7 ha (Witt 1992), and in the southern flying squirrel (G. volans) 9.9 ha (Fridell and Litvaitis 1991). The home-range size of both male and female Siberian flying squirrels was also much larger than that of other similar-sized herbivorous mammals (Swihart et al. 1988).

Both sexes concentrated their activities in small core areas. These were clusters including 85% of fixes, and they represented 9 and 11% of the home-range area in males and females, respectively (Table 1). In the core areas the densities of aspen and alder were greater than elsewhere in the home range (Hanski 1998). Also nest sites were more often located within the core areas than in other parts of the home range (Hanski et al. unpublished manuscript). The flying squirrels seemed to concentrate their activities in the parts of the home range where food and nest sites were abundant. Similar core-area use has been found in several other mammal species, including the North American flying squirrels (e. g. Bendel and Gates 1987, Wauters et al. 1994).

Juvenile flying squirrels of the first litter of the year were born in late April and they dispersed from their natal home range in August. The average dispersal distance for juvenile females from the first litter was 1.4 km and for juvenile males 2.4 km (Table 2). Before their final departure from the birth site, juveniles made long nocturnal trips in various directions, but returned to the nest before sunrise. Only one juvenile from the second litter dispersed (Table 2); all others remained in the natal home range for the first winter. It is not known if these juveniles dispersed the next spring, but one male and one female were observed to stay and breed in their natal home range. Currently, the data on natal dispersal in the Siberian flying squirrel are few and the results are preliminary, especially from the young of the second litter.

Table 2. Distances (km) of natal dispersal in the Siberian flying squirrel.

P. Ihalempiä, H. Rockas, V. Selonen, T. Seppä and P. Stevens helped with fieldwork. P. Stevens gave valuable comments on an earlier draft of the manuscript. I thank the private landowners for the permission to use and mark their forests and the Finnish Forest and Park Service for the opportunity to work in the Nuuksio National Park. The study was financially supported by Maj and Tor Nessling Foundation, Academy of Finland, Emil Aaltonen Foundation, Ella and Georg Ehrnrooth Foundation, and the Finnish Forest and Park Service. All this assistance and help is gratefully acknowledged.

Ecological and behavioral data on wild boar population in Lower Saxony, Germany, has been collected since 1998. Fifty-two wild boars were captured in large traps and marked. Twenty-nine animals were marked with ear-tag transmitters. By telemetric observations of wild boar groups, data on their home range, habitat use, and daily and nightly movements were collected. In some cases the effects of the hunting activities (driven hunts) on wild boar movements were investigated. Preliminary results indicate that in spite of high hunting pressure during the driven hunts, and contrary to the general assumptions, all of the escaping wild boar groups remained in their established home range. The largest escape-distance registered was approximately 4 kilometers.

The population density of wild boars in Lower Saxony, Germany, has increased dramatically in the last decade. The resulting overpopulation is now damaging crops and influencing the occurrence and distribution of swine-fever disease. As a result, hunting is being used to reduce the population density in the zones endangered by swine fever. An effective hunting method is the driven hunt, in which, beaters – hunters and dogs – drive the wild boars out of their hiding places. But this hunting method includes the risk of spreading the animals over wide areas, which means a greater risk of infecting other wild boars with the swine fever virus. This study was started in 1998 in order to determine the local movements of wild boars and the effect of this hunting method on wild boar escape movements.

The study area is comprised of dense forest stands located in Lower Saxony, Germany, in the region of Knesebeck-Gifhorn, within the endangered zone of swine-fever disease. Since 1998, ecological and behavioral data has been collected on the wild boar population there. Fifty-two young wild boars were captured in large traps and marked. Twenty-nine of the animals were marked with ear-tag transmitters with whip antennas. By telemetric observations of radio-tagged wild boar, data on home range, habitat use, and daily and nightly movements were collected. In some cases, the effects of the driven hunts on wild boar movements were investigated.

Some of the transmitter ear tags or just the antennas were bitten off by the social partners of the wild boars within a short time. During the hunting season, 19 marked wild boars were shot, most of them within 1 km of the marking place. This indicates that the wild boars are quite sedentary. Only one young male was killed 10 kilometers away. After marking and release from the trap, the wild boars escaped on established routes to distances up to 1.5 km. The home range size of the radio-tagged wild boars was up to 600 ha. In spite of high hunting pressure during the driven hunts, all of the escaping wild boars remained in their established home range, contrary to the general assumptions. The largest escape-distance observed was approximately 4 km. For example, during a driven hunt on November 14, 1998, in spite of high pressure from beaters, the boars remained in their established territory throughout the hunt. However, the following night, the boars moved 2 km south of their home range and remained there for some days.

The study was financially supported by the Ministry of Agriculture, Food, and Forestry of Lower Saxony, Germany.

Canadian Wildlife Service (Quebec Region),
Environment Canada, Sainte-Foy, Canada

Union Québécoise de Réhabilitation des Oiseaux de Proie (UQROP),
Saint-Hyacinthe, Canada

In February and April 1998, we captured seven Barrow’s Goldeneye drakes along the St. Lawrence River, Quebec, and implanted them with satellite transmitters. Following release, five were located on the north shore of the St. Lawrence River in May, 60-140 km inland from the estuary and gulf. They spent 34-50 days at their respective site, presumably with a mate, then departed between May 29 and June 28 and flew 800-1,120 km northward to reach their molting ground. Another male, probably a non-breeder, stayed along the St. Lawrence River until June and then flew 1080 km to Ungava Bay, presumably to molt.

Unlike the Common Goldeneye (Bucephala clangula), which is widely distributed throughout the boreal forests of the world, the Barrow’s Goldeneye (B. islandica) has a fragmented distribution restricted to North America and Iceland (del Hoyo et al. 1992). Most (> 90%) of the world’s population of Barrow's Goldeneyes breeds and overwinters west of the Rocky Mountains, in Canada and the United States. Only two small isolated populations are found elsewhere, one in eastern North America estimated at about 4,000 birds (Savard and Dupuis in press), and one in Iceland estimated at 2,000 individuals (Hagemeijer and Blair 1997). While the distribution and ecology of the western North American and Icelandic populations are well documented (Gardarsson 1978; Einarsson 1988, 1990; Savard 1987, 1988; Savard et al. 1991), very little is known of the eastern North American population. Until recently, its breeding areas were unknown and its molting grounds were a mystery. The only available information on this small population was on its wintering distribution; a few thousand Barrow’s Goldeneyes are known to winter along the St. Lawrence River estuary and gulf, mostly in Quebec (Reed and Bourget 1977).

Hunting pressure, low population numbers, clumped winter distribution, and the possible impact of logging on potential breeding areas have raised concerns about the welfare of the small eastern North American population of Barrow’s Goldeneyes (Savard and Robert 1997, Savard and Dupuis in press). As a response to these concerns, we initiated a study, using satellite telemetry, to locate the breeding and molting grounds of those Barrow’s Goldeneyes that overwinter along the St. Lawrence River, which is a wintering stronghold for the eastern North American population (Reed and Bourget 1977).

We captured seven Barrow’s Goldeneyes along the north shore of the St. Lawrence River estuary, Quebec, during the non-breeding season, using decoys and floating mist nets. Two 18-m nets (127 mesh size) were set over shallow tidal waters using three floating rafts (Burns et al. 1995) in two areas inhabited by many Barrow’s Goldeneyes during winter: Baie-des-Rochers (69°48'W, 47°58'N), where three adult males were captured on February 21-22, 1998, and Mistassini (67°57'W, 49°17'N), where four others were caught on April 7-10, 1998. These males were tracked with satellite transmitters. We chose to track males to maximize our efforts, since they follow females to their breeding area where they spend a few weeks before heading to their molting sites (Eadie et al. in press). Thus by following males, we were hoping to obtain the location of both breeding and molting areas.

Because harnesses appear to affect the behavior of diving ducks (Perry 1981), we opted for the implantation of transmitters, following a technique developed by Korschgen et al. (1996) on mallards (Anas platyrhyncos). This technique involves using aseptic surgery under general anesthesia to fit the transmitter in the abdominal cavity of the duck with the external whip antenna exiting through its back; thus requiring a well qualified veterinarian with at least two assistants. The transmitters were gas sterilized and the surgical instruments were autoclaved prior to the fieldwork. We used a portable anesthetic machine to induce birds with 3.5-4.0% isoflurane delivered in oxygen using a customized face mask. To maintain anesthesia, birds were intubated with an endotracheal tube. During the surgical procedure, they were monitored with an ultrasonic breathing monitor and a Doppler flow detector installed on the base of their tongue. Time between capture and release after surgery ranged between 3 h 40 min and 7 h 45 min. We observed no abnormal behavior following the surgeries.

We used Argos PTT-100 satellite transmitters (Microwave Telemetry, Maryland, USA) weighing about 51 g, representing less than 5% of the captured males’ weight (Table 1). Each L-shaped transmitter contained four lithium batteries and measured 57 mm x 36 mm with a thickness ranging from 7.5 mm to 15 mm from one side to the other. The antenna was 22 cm long and made of multi-coated multistrand stainless steel. Each transmitter was programmed to transmit for 7 hours every 48 hours for the first 56 days, 6 hours every 24 hours for the next 70 days, and 7 hours every 72 hours for the rest of its life. Signals were received by two polar-orbiting National Oceanic and Atmospheric Administration satellites and retransmitted to the Argos global processing center (Landover, Maryland) for analysis.

Of the seven males implanted with transmitters, five went to a breeding site and six to a molting area (Table 1). Birds apparently bred on the Laurentian Highlands, between 60 km and 140 km inland along the north shore of the St. Lawrence River estuary and gulf (Figure 1). Departure dates from the St. Lawrence River estuary (the wintering area) seemed related to spring phenology on the breeding areas, being later as we proceed northeasterly. Arrival on the most southwesterly breeding site was on April 26, as compared to May 11 for the most northeasterly site (Table 1, Figure 1). Length of stay on the breeding grounds ranged from 34 days to 50 days and departure for the molting areas was not synchronized. Arrival on the molting areas ranged from June 9 to July 4, again with an indication of earlier arrival by earlier breeders (Table 1).

Four molting sites were located (Figure 1). The first one was in Hudson Bay between the Belcher Islands and the mainland, in the Salikuit Islands area. Two molting sites were located in Ungava Bay, one about 60 km inland from the mouth of the Leaf River (Rivière aux Feuilles), near Tasiujaq on the west side of the bay, and one on the east side of the bay at the mouth of the Tuttutuuq River, about 100 km east of Kuujjuaq. Finally, one bird was located in an estuary on the Labrador Coast, about 20 km southwest from Nain. The transmitters of two birds lasted until fall migration. Both individuals remained on their molting ground until at least mid-October. The only one with strong radio signals (Bird 23124) during fall migration stayed in Ungava Bay until October 24 and was then located on October 26 on its wintering ground (the St. Lawrence estuary), after having covered 1,200 km in less than two days (Figure 1). The other one (Bird 23123) probably staged in the Lake Mistassini area before reaching its winter quarters, along the St. Lawrence River estuary (Figure 1). Two birds apparently did not breed. One (Bird 23123) stayed on its wintering area until June 5 and then migrated directly to its molting site in Ungava Bay (Figure 1). The other one (Bird 23007) stayed along the St. Lawrence River until June 22, after which its radio went dead (not illustrated on Figure 1).

Satellite telemetry proved highly efficient in locating breeding and molting sites of Barrow's Goldeneyes in eastern North America. The number of molting sites discovered during this study was quite unexpected, especially since the number of tracked individuals was low. Studies in western North America (van de Wetering 1997) and Iceland (Gardarsson 1978) suggested that Barrow’s Goldeneye drakes concentrate in a few molting sites only. However, we identified at least four different molting areas for only six Barrow's Goldeneye drakes, and these were in quite distant regions. Very few Barrow’s Goldeneye molting sites were known in eastern North America prior to our work. Yet, both Barrow’s and Common Goldeneyes have been reported in various inlets of the Labrador coast, especially between Makkovik and Ramah, during their molting period (Daury and Bateman 1996, Savard and Dupuis in press). In 1954, a raft of about 1,500 goldeneyes, of which half may have been Barrow’s, were observed molting at the head of Nain Bay, Labrador (Todd 1963), in the same area to which one individual tracked during this study (Bird 23006) flew after having nested on Quebec’s North Shore. Approximately 750 goldeneyes were again reported there in 1955, all thought to be molters (Daury and Bateman 1996). Exact abundance of molting Barrow’s Goldeneyes along the Labrador coast remains to be determined, although there is some indication that many individuals molt there.

Figure 1. Movements of Barrow’s Goldeneye drakes (n=6) implanted with satellite transmitters along the St. Lawrence River in February and April 1998.

Table 1. Characteristics, location and movements chronology of Barrow’s Goldeneyes captured along the north shore of the St. Lawrence River, Quebec, in 1998.

We were quite surprised to see Barrow’s Goldeneyes stay at their molting location long after they had regained their ability to fly. They apparently spent at least two months there after regaining their flight feathers, according to molting chronology known to occur on the West Coast (Eadie et al. in press). We speculate that these possibly highly productive and secluded arctic areas are ideal to complete their body molt and to feed. Also, staging was quite limited, both in spring and fall; movements between breeding, molting, and wintering sites were usually quite fast and direct. Savard (1985) documented an overnight flight by a pair of Barrow's Goldeneyes from its wintering area near Vancouver to its breeding location in Central British Columbia, suggesting that such quick migration may be the rule.

The breeding locations found during this study confirm earlier observations of pairs on the highlands of the Quebec North Shore (Savard 1996). The Laurentian Highlands appear to be a major breeding area for Barrow’s Goldeneyes wintering along the St. Lawrence River. Interestingly enough, ground surveys conducted in the area used by a male tracked during this study (Bird 23005) led to the sighting of the first Barrow’s Goldeneye brood for eastern North America (Bannon et al. 1998). Preliminary ground surveys conducted in 1998 on a few of the Quebec North Shore areas inhabited by Barrow’s Goldeneyes during the nesting season also stressed the urgency to develop a management plan for the species, as these areas seem to be under heavy logging pressure.

Satellite tracking allowed us to make giant steps forward in our understanding of the ecology of the eastern North American Barrow's Goldeneye. It would have taken decades and a lot of luck to acquire the same information by conventional studies.

This study was funded by the Endangered Species Division of the Canadian Wildlife Service (Quebec Region) and the United States Fish and Wildlife Service. We express our sincere appreciation to Dr. J. Tremblay for assisting Dr. Fitzgerald in implanting the transmitters. We also thank C. Marcotte, A. Bourget, P. Brousseau, G. Cyr, G. Falardeau, S. Plourde, J.-F. Rail, and F. Shaffer for their help during the fieldwork. We extend our thanks to M. Melançon who kindly produced the figure, and to R. C. Cotter who made comments that improved this paper.

We studied the behavior of carrion crows using radio telemetry and behavioral observation in order to determine whether they cause damage in agricultural systems and among the fauna. During the breeding season, the flock is divided in two groups, paired crows and a flock of solitary non-breeders. Outside the breeding season, the paired couples join the flock. Aggressive behavior against other animals, small birds or small game was not observed in either group.

The carrion crow (Corvus corone corone) is a resident bird in Central Europe, present in a wide spectrum of habitats (Bezzel 1982, 1993, 1995; Glutz von Blotzheim 1993; Epple 1996). In general, this euryoecios species prefers open sites with individual or small groups of trees, shrubs, and areas with low ground cover. They generally avoid dense forest but increasingly become adapted to settlements. Their preferred nest sites are generally at a height of approximately 5 m. The tree species vary regionally. Occasionally, nests can be found in rock formations if no suitable trees are available. Adult carrion crows are typical omnivorous, but nestlings are mainly fed food of animal origin. The female produces three to six eggs that are usually laid in April. The breeding duration ranges between 17-22 days, followed by a nestlings stage of 30-35 days. During breeding season, it is possible to observe two different behavioral societies, territorial pairs that are involved in reproduction and flocks of solitary individuals that do not breed. Outside the breeding season the paired crows join the flocks.

The carrion crow has repeatedly been blames for significant damage to agriculture and native fauna, especially small birds and small game (Bezzel 1988, Dick 1995). The main areas of damage have involved destruction of seeds and seedlings, especially when larger flocks invade agricultural systems. In addition, carrion crows have been blamed for puncturing plastic foil used to cover silage, which may cause substantial damage due to the fermentation initiated by the exposure of the stored product to air. Predation of eggs and young birds and juvenile hares and rabbits has also been listed as one of the problems caused by crows (Böhmer 1976, Mäck 1998, Prinzinger and Hund 1981, Rahmann et al. 1988, Richner 1989, Wittenberg 1968).

The Ministry for Environment and Forestry initiated a study to determine the actual dimension of the damage caused by carrion crow. On the basis of these findings, it was to be decided whether regulatory mechanisms (e.g. hunting) are required to control and reduce the population density of the carrion crows in Rhineland Palatinate. Behavioral observation and radio telemetric studies were carried out in order gain new insight into the ecology and behavior of the carrion crow.

Observations took place from fall 1996 to spring 1998. The population of carrion crows was monitored on three sites, each approximately 15 km², near Kaiserslautern, Germany (Figure 1).

The equipment for radio telemetry was supplied by Biotrack Limited (UK). Twenty transmitters were used, Model TW-3, emitting pulses of 20 ms duration at approximately 1 Hz. The frequency band ranged between 150.050 MHz and 150.240 MHz. The frequencies of the 20 individual transmitters were separated by 12.5 kHz, allowing individual identification of each transmitter. The radiation output of the transmitters was limited to 0.2 mW (ERP). The weight of the transmitters was less than 5% of the body weight of the crows, and was shown not to alter or influence the behavior of the animals. The transmitters were attached to the backs of the crows with a teflon ribbon harness (glue-mounted, backpack-style attachment) (Kenward 1987). Mariner-57 receivers were used, which allowed acoustic as well as optical identification of the tagged individual birds. The birds were tracked using two five-element Yagi antennas tuned to the 150 MHz range.

The distribution of different behavioral patterns displayed by paired crows during breeding season is presented in Table 1. The data was derived from 4 different pairs, which were observed for 86 hrs over several days. The behavior of the four individual pairs showed variations in absolute numbers. However, there was similarity in the distribution of the behavior. During breeding season the carrion crows spent most of their time feeding, foraging and resting. The distribution of different biotop types used by paired crows during breeding season is presented in Table 2.

Table 1. Behavioral patterns of paired carrion crows during breeding season.

Table 2. Biotop types used by paired carrion crows during the breeding season.

The observed crows spent 40% of their time in areas with low ground cover (24% in freshly mowed greenlands and 16% in pastures) where they fed on ground arthropods. The mean time spent foraging was approximately 20 minutes, after which the adults returned to the nest (for approximately 4 minutes) to feed the nestlings. Intake of vegetation (grass or grain) could not be observed. Other agricultural habitats such as seeded land, and summer and winter grain field were hardly used. The time spent in shrubs and hedges was mainly used for resting. The distribution of different biotop structures used by paired crows during breeding season is presented in Table 3.

Table 3. Biotop structures used by paired carrion crows during the breeding season.

Most (61%) of the crows’ activity took place on open sites, followed by activity at forest edge (15%) and partly open land (14%). Although carrion crows can be found in very different habitat types, it is known that they prefer open landscape structures. Our results confirm those earlier findings.

Our observations showed that the daily activity of the carrion crows can start as early as 0430 hours (Central European Standard Time), often starting with calling behavior. At around 2145 hours, the activity of the crows began to cease.

The mean home ranges of the paired crows during the breeding season ranged from 215 m² to 369 m². Outside the breeding season the home ranges were significantly larger. It was impossible for us to determine the maximum home range size due to the immense distances that the crows flew and the limited range of the transmitters in uneven habitats.

The behavioral patterns displayed outside the breeding season was much more limited compared to the repertoire shown during the breeding season. Most of the time was spend feeding on open greenlands and grain fields, or resting. However, compared to the breeding season, the crows spent more time in forest areas. No interaction with other animals was observed, unlike during the breeding season.

In conclusion, our observation did not support the belief that carrion crows cause substantial damage. The formation of flocks, reduced availability of food, and the increased mobility of the carrion crows outside the breeding season are obvious observations. These factors combined with the presence of other species of the crow family, e.g. rook (C. frugilegus) and jackdaw (C. monedula), may occasionally cause damage in agricultural systems.

However, we found no indication that the carrion crows cause extensive damage in agricultural systems or among small birds and small game. We conclude from our results that the reduction of the populations of carrion crows is not required.

We would like to thank the Ministry for Environment and Forestry of Rhineland-Palatinate for its financial support of this research project. We are also grateful to our co-workers H.-M. Helb, M. Helb, J. Jeblick, K. Müller, K. Nagel, L.-G. Otto, M. Stremmel and T. Vicinus for their engaged support in the field as well as in the laboratory.