Hunting effort and game vulnerability studies on a small scale: a new technique combining radio-telemetry, GPS and GIS

Authors


Henrik Brøseth (fax + 47 73 80 1401; e-mail Henrik.Broseth@ninatrd.ninaniku.no).

Summary

1. Global positioning systems (GPS) were used to track hunters in an area of central Norway where willow ptarmigan Lagopus lagopus were contemporaneously radio-tracked. A geographical information system (GIS) was then used to study spatiotemporal interactions between hunters and game.

2. Recording hunting activity with the GPS produced accurate and unbiased information about the behaviour and effort of hunters. When GPS tracking nine hunters during 50 hunter-days, data were lost for an estimated time of 30 h 45 min, which constituted about 10% of the total hunting time.

3. Willow ptarmigan hunters walked on average 16·2 km daily at a speed of 2·8 km h–1, and they hunted for 9 h each day, of which almost 6 h was active hunting time. During 50 hunter-days they had 295 h of active hunting, covered a distance of 818 km and harvested 20% of the willow ptarmigan population in the area.

4. The spatial distribution of hunting pressure was strongly dependent on the starting point of the hunters, and areas close to the base cabin were subject to most hunting activity. Areas furthest away, towards the border of the hunting area, experienced little hunting activity.

5. Logistic regression showed that survival probability of ptarmigan was best predicted by distance from the cabin. Shot radio-tagged birds lived closer to the cabin, and had twice as high hunting pressure in their home range, compared with surviving radio-tagged birds.

6. This method of obtaining quantitative data about human effort will have application in other studies when there is a need to quantify and analyse human effort on temporal and spatial scales.

Introduction

Hunting occurs both for subsistence and recreational purposes, and the exploitation of wildlife populations has increased rapidly during the last century (Taylor & Dunstone 1996). In many managed populations hunting affects both population structure and individual behaviour (Hutchings & Harris 1995; Moreira & Macdonald 1996; Solberg et al. 1999). It is important in harvesting management to understand how different harvesting strategies and techniques affect populations (Mangel et al. 1996; Taylor & Dunstone 1996). Recent studies have shown that hunting effort can be spatially uneven among populations (Lyon & Burcham 1998), creating refuge areas in the hunted population (Clayton, Keeling & Milner-Gulland 1997). Heterogeneous hunting effort can affect the impact that hunting has on the exploited population in both space and time (McCullough 1996; FitzGibbon 1998; Milner-Gulland & Mace 1998).

It is important, therefore, to have techniques that provide spatial data on the behaviour and effort of hunters contemporaneously with the assessment of impact on the hunted population. However, unbiased spatial data of hunting effort are rarely available for harvested populations under natural conditions. Now, however, technological advance has led to the increased use of global positioning systems (GPS) and geographical information systems (GIS) in the study of ecological problems in time and space (Haslett 1990; Slonecker & Carter 1990; Haines-Young, Green & Cousins 1993). These techniques offer major advantages, particularly in the assessment of hunting.

GPS can provide continuous tracking of hunters, producing unbiased data about spatiotemporal patterns of hunting effort, much the same way as satellite tracking of vessels is used in fishery management. Compared with subjective methods (e.g. questionnaire: Olsson, Willebrand & Smith 1996; interviews: Steinert, Riffel & White 1994; hunters’ field notes: Marks 1994), the GPS technique obtains information with higher precision, especially at the spatial scale. GPS and GIS therefore constitute excellent tools for studying spatial and temporal aspects of the interaction between hunters and game.

In this study, we used data from GPS tracking of ptarmigan hunters in a mountain area of central Norway to explore how this new technique can be applied to assess hunting effort accurately. We demonstrate, by using data from radio-tagged willow ptarmigan Lagopus lagopus L. in the same area, how this technology can be applied to identify important factors affecting survival probability of the hunted game.

Materials and methods

Study area

Our study was conducted during the period 10–22 September 1997, as part of a larger study assessing the impact of hunting on willow ptarmigan populations. The study area was a 30-km2 tract of private land in Meråker (63°15′ N, 11°35′ E) in central Norway, on the border between the north boreal and the low alpine region. It is dominated by scattered mountain birch Betula pubescens Ehrh. forest intersected with some drier areas and bogs. The scrub layer is dominated by dwarf birch B. nana L., juniper Juniperus communis L. and some Salix spp., whereas in the field layer heather species (Vaccinium myrtillus L., Empetrum nigrum L., V. uliginosum L. and Arctostaphylos alpina L.), sedges and grasses are most common. The higher part of the area features mainly dwarf birch heath and some moraine ridges with lichens and sedges. The terrain is hilly, but generally gently sloping from Nautfjellet (924 m a.s.l.) in the south-west towards the lowest point in the north (460 m a.s.l.). Most of the area is below the timber line, mainly at 600–800 m a.s.l. The climate is maritime with 1000–1500 mm precipitation and about 140 days with mean temperature above 6 °C a year (Moen 1998). The density of willow ptarmigan in the study area was estimated at 22·5 km–2 by line transect sampling with pointing dogs in mid-August 1997 (Pelletier & Krebs 1997; Pedersen et al. 1999).

Recording hunting effort

The hunt was organized as two separate teams of five and four people, respectively, that had exclusive access to hunt the whole study area. Both teams consisted of hunters with long hunting experience, who had hunted in the area for many years. The first team hunted between 10 and 15 September, and the second from 17 to 22 September, all with pointing dogs. There is no road access to the study area and both teams used a cabin located in the middle of the hunting area as base camp.

Data on hunting behaviour and effort were collected with a portable GPS receiver attached to the hunters’ backpacks. We used hand-held, non-differentially corrected 12-channel GPS receivers (GPS 12 and GPS 12XL; © 1998 GARMIN Corporation, Olathe, Kansas) with a capacity to store up to 1024 positions. The receivers were turned on every morning when the hunters left the cabin, and downloaded for data on a computer when they returned every evening. GPS receivers were programmed to take positions with 1-min intervals. From these data we obtained a line track of hunter movements during the day (Fig. 1). All hunters were trained to operate the GPS receiver before the hunt, and were instructed to mark positions of harvested ptarmigan in the same way as we tested point repeatability of the receivers.

Figure 1.

Figure 1.

Recorded line tracks of willow ptarmigan hunters carrying GPS receivers in a 30-km2 private hunting area in central Norway during 50 hunter-days. White circle = base cabin.

Spatial distribution of hunting pressure

The hunting area was converted to a 50 × 50-m grid, based on GPS repeatability, to visualize spatially the hunting pressure in the hunting area. Line tracks with an 80-m buffer, representing hunting activity from a hunter with pointing dog, were overlaid on the hunting area and the number of times each grid cell was affected by hunting activity during the 2 weeks of hunting was calculated. We used an 80-m buffer around the line track based on the probability curve from transect sampling data on the willow ptarmigan population (Pedersen et al. 1999). There is a steep drop in the probability of finding birds situated further than 80 m from the transect line when transect sampling with a pointing dog, and we assumed the same is true when hunting with pointing dogs.

Gps test

To test the horizontal repeatability of the GPS system when plotting a position, we took five consecutive locations in a fixed, easily recognizable, test-point during a 30-min session. To quantify the variation in repeatability, we calculated the Euclidean distance (error) between the five locations and the arithmetic mean centre of locations taken in a session, as a measure of horizontal accuracy (van Diggelen 1998). A total of nine different test-points was distributed in the study area, representing variation in cover of birch canopy and topography, which can affect the signal strength from the satellites and thereby reduce the accuracy of the GPS (D’Eon 1995, 1996; Deckert & Bolstad 1996).

Line track function repeatability of the GPS receivers, used to reveal hunter movements in the hunting area, was tested by walking three straight lines of 200 m, from different angles, through each of the nine different test-points distributed in the area. We used the Euclidean distance from the test-point location (the arithmetic mean centre of five repeated locations) to the nearest point of the line track as the error estimate. All GPS receiver tests were performed during the 2 weeks of the study, and point locations were taken with an ‘average measurement’ function in the receivers for 1–2 min.

Radio-tracking and survival probability

Willow ptarmigan were captured either in March–April by use of spotlight and landing net from snowscooters, or in August by pointing dogs and hand-held nets (Skinner, Snow & Payne 1998). They were fitted with a necklace radio transmitter, either from Biotrack Ltd (Dorset, UK) or Televilt International AB (Lindesberg, Sweden). At the start of the hunting season, 28 radio-tagged willow ptarmigan were resident inside the hunting area. We located radio-tagged birds every morning as long as they were alive during the hunt, by taking several cross-bearings at a distance of 50–100 m. To estimate a home range for birds with six or more telemetry locations during the 2 weeks of hunting (n = 19), we calculated a 90% minimum convex polygon using the Ranges V software program (Kenward & Hodder 1996). For birds with five or less locations (n = 9) shot early in the period we used the arithmetic centre of locations to estimate a home range as a circle with an area equal to the average home range size of birds with six or more locations (30·9 ha).

We analysed survival probability of radio-tagged birds in relation to hunting pressure in their home range and distance from the cabin with logistic regression. We calculated hunting pressure (h km–2) from GPS line tracks within home ranges of individual birds (Fig. 2), and used the arithmetic mean centre in the home range to measure distance from the cabin.

Figure 2.

Figure 2.

Examples of hunting effort within home ranges of individual radio-tagged willow ptarmigan in a private hunting area in central Norway during 50 hunter-days. White circle = base cabin.

Statistical analyses

Data were tested for normality with the Kolmogorov–Smirnov test, and equality of variances with the Levene’s test. Non-parametric tests were applied when deviation from these assumptions was found, and two-tailed tests were used throughout. All statistics, except the G-test, were calculated in SPSS for Windows (Release 8.0; © 1997 SPSS Inc., Chicago, Illinois). ArcView GIS (version 3.1; © 1998 ESRI Inc., Redlands, California) with extension from Hooge & Eichenlaub (1997) was used for data handling and analyses on the spatial scale.

Results

Performance of the gps

In some line tracks from the hunting trips there were gaps with missing data. The most common cause was that the length of the hunt exceeded battery life, but also poor satellite coverage and moisture in the receivers resulted in loss of data. We were able to estimate the time loss by recording when hunters left and returned to the cabin, and comparing this with the data in the receiver memory. We were not able to reconstruct missing spatial data, but we could estimate the distance lost from time and speed records of the hunters. During the 2 weeks of tracking, data were lost for an estimated 30 h 45 min, which constituted about 10% of the total hunting time, but no locations of shot ptarmigan were lost. The horizontal repeatability of the GPS system was high for both point and track data collected in this study. In 95% or more of the cases, point and track data were estimated to be less than 35 m from the position indicated by the GPS receiver (Table 1).

Table 1.  Horizontal repeatability (metres) of non-differentially corrected GPS receivers used to collect positions of shot willow ptarmigan, and track movements and effort of ptarmigan hunters
GPS repeatabilityPoint (n = 45)Track (n = 27)
Mean19·313·6
Median18·311·0
95 percentile35·027·4

Behaviour of hunters and hunting effort

During 50 hunter-days there was a total of 295 h of active hunting, and the nine hunters walked a total distance of 818 km (Table 2). They harvested 135 birds, approximately 20% of the available ptarmigan population. In the first hunting period more than twice as many ptarmigan were taken compared with the second period (97 of 675 and 38 of 578, respectively), although the hunting effort of the two teams was almost identical. Differences in the proportion of young birds in the harvest cannot explain the observed difference (66% and 64%, respectively). One possible explanation is that the first team reduced the ptarmigan density in the area nearest the cabin, so that the second team had to hunt further from the cabin. This is supported by the distribution of harvested birds: the second team shot their ptarmigans further from the cabin than the first team, indicating a depletion near the cabin (z = 2·63, n1 = 38, n2 = 97, P = 0·008). Mean daily hunting distance for the nine hunters was 16·2 km (range = 12·8–20) at an average speed of 2·8 km h–1 (range = 2·6–3·3; Table 3). They hunted for nearly 9 h each day, of which 5 h and 48 min was active hunting. The remainder of the time was used as breaks in the hunt (e.g. lunch, dinner or shorter resting periods), either in the cabin or out in the field.

Table 2.  Hunting effort revealed by nine willow ptarmigan hunters carrying GPS receivers in a 30-km2 private hunting area in central Norway during the first 2 weeks of the hunting season in 1997. Team 1 hunted the first week (10–15 September) and team 2 hunted the second week (17–22 September)
 Number of huntersHunter-daysHunting distance (km)Hunting time (h)Number of ptarmigan harvested
Team 152641314697
Team 242440514938
Total950818295135
Table 3.  Individual hunting effort revealed by nine willow ptarmigan hunters carrying GPS receivers during the hunt
 MeanSD
Hunting speed (km h–1)2·80·2
Hunting distance (km day–1 hunter–1)16·22·7
Hunting time (h day–1 hunter–1)5·81·0
Resting time (h day–1 hunter–1)3·00·5
Harvest (no. day–1 hunter–1)2·71·7

The spatial distribution of hunting pressure showed that 27% of the grid cells in the area were not affected by hunting activity during the hunting period, 28% were affected once, 15% twice, 11% three times, and 19% four or more times (Fig. 3). The hunting area within 2·5 km of the cabin was utilized most intensely, receiving 82% of the hunting pressure. The hunting pressure decreased with increasing distance from the cabin, with the areas towards the border of the hunting area experienced little hunting activity, if any (Fig. 3).

Figure 3.

Figure 3.

Spatial distribution of hunting effort by willow ptarmigan hunters carrying GPS receivers in a private hunting area in central Norway during 50 hunter-days. White circle = base cabin.

Hunting vulnerability of ptarmigan

The spatial relationship between hunting pressure and vulnerability of willow ptarmigan was evident when we examined radio-tagged birds through the hunting period. The hunting pressure within the home range of individual ptarmigan revealed that shot birds experienced twice as much hunting pressure as those surviving, 14·3 vs. 8·1 h km–2 (U = 45, n1 = 15, n2 = 13, P = 0·015). There was a much lower variation in the hunting pressure for shot birds (CV = 0·28) compared with surviving birds (CV = 0·96, F = 5·2, P = 0·03); some birds survived the whole period in spite of high hunting pressure in their home range. Also, surviving birds had home ranges situated further from the base cabin than those that were shot (2091 ± 550 SD and 1222 ± 426 SD m, respectively, t26 = 4·71, P < 0·001). When using a logistic regression analysis with hunting pressure and distance from cabin, the latter predicted survival probability best (z = –9·17 + 5·6 (distance from cabin in km), P < 0·001; Fig. 4). Hunting pressure also predicted survival probability well (z = 1·78 – 0·17 (hunting pressure in home range, h km–2), P < 0·01), as these two variables were correlated.

Figure 4.

Figure 4.

Logistic regression model of survival probability for willow ptarmigan in a private hunting area in central Norway (z = –9·17 + 5·6 (distance from cabin in km)).

Discussion

Existing information on hunting behaviour and effort is limited to results mainly from questionnaires and interviews (Fraser & Sweetapple 1992; Nugent 1992; Steinert, Riffel & White 1994; Olsson, Willebrand & Smith 1996). Tracking hunting activity with the GPS system produces new and valuable information about the behaviour and effort of hunters in time and space, with more accurate and unbiased estimates than studies based on hunter declarations. Combining this technique with information obtained from radio-tagged individuals makes it possible to analyse survival probability of the hunted species in relation to hunting effort on a scale that previously was impossible. Thus, the major advances offered by this technique lie in the possibility of obtaining unbiased spatial data of hunting effort, which can increase our understanding of the impact that hunting has on hunted populations as well as hunter–prey interactions.

Only one previous study, of elk Cervus elaphus nelsoni Bailey hunters in the Rocky Mountains, has recorded hunter behaviour through the use of GPS (Lyon & Burcham 1998). Compared to elk hunters on foot, willow ptarmigan hunters hunted actively for a longer time period each day. Also, elk hunters walked a shorter distance each day, and at a lower speed. This would be expected due to different hunting techniques for different species. Elk hunting involves slow walking with frequent scanning and is often restricted to morning and afternoon sessions when the elk are most active. Ptarmigan hunters, on the other hand, hunt at all times of the day and walk at a higher speed, and they rely on dogs to find or flush birds.

The spatial distribution of hunting pressure was strongly dependent on the starting point of the willow ptarmigan hunters. Generally, areas close to the cabin received most of the hunting activity, whereas areas farthest away, towards the border of the hunting area, experienced little if any hunting activity. However, both topography and habitat configurations seemed to have some modulating effect on this pattern. This predictability has important implications for harvesting management in a hunting area. Game managers will be able to predict the hunting pressure spatially with greater precision, and should be able to manage the harvest more accurately. Improved access to hunting areas, for example through new roads or cabins, will clearly alter the spatial distribution of hunting activity, with subsequent impact on the harvested population. A recent study of elk hunters demonstrated the significance of roads on the spatial distribution of hunting pressure, where much of the hunting activity was predicted by the presence of roads (Lyon & Burcham 1998). In several ungulate populations, improved access to hunting areas has been shown to have a negative influence on population density (Fraser & Sweetapple 1992; Rempel et al. 1997), and depletion of game species can increase with proximity to hunters’ bases (Marks 1994).

Vulnerability to hunting is not solely determined by hunting pressure; factors like habitat cover, fragmentation and topography can have a modulating effect on vulnerability (Foster, Roseberry & Woolf 1997). Individuals in many hunted populations show behavioural responses in relation to these modulating factors (Swenson 1982; Naugle et al. 1997; Verkauteren & Hygnstrom 1998). In brown hares Lepus europaeus Pallas, different hunting methods have resulted in different behavioural responses to hunting through changes in the flushing behaviour (Hutchings & Harris 1995). In other species hunting disturbance has been shown to cause individuals to move to refuge areas where they are inaccessible or more difficult to hunt (deer: Swenson 1982; Verkauteren & Hygnstrom 1998; waterfowl: Madsen 1998; Duncan et al. 1999).

The methods used and the results found in our study should have application to hunting management of other areas and species. Our method of obtaining quantitative data on human effort should also have considerable application to other studies where the effort on temporal and spatial scales, e.g. sampling effort in monitoring, must be documented. Incorporation of digital landscape maps and elevation models in the GIS, together with GPS data on hunting effort and radio-tracking of game will facilitate a more comprehensive understanding of hunter–game interactions.

Acknowledgements

The Norwegian Directorate for Nature Management, the Norwegian Research Council’s programme ‘Use and management of outlying fields’ and the Norwegian Institute for Nature Research provided financial support for our study. We thank H. Albech, K. Albech, D.C. Clausen, S.K. Gamlemsvik, H. Helseth, M. Pinto, J. Roaldseth, M. Saltnes and A. Thorp for their effort and patience in carrying the GPS receiver during the hunt, and Meraker Brug A/S for permission to work on their land. We appreciate field help by T. Lande, A. Moen, G.E. Nylund, I. Rimul, O. Rimul, S.L. Svartaas, B. Wahl and T. Aarvak. Two anonymous referees, J.A. Kålås, H. Steen, J.E. Swenson and T. Aarvak made valuable comments on an earlier version of this manuscript.

Ancillary