Northern Prairie Wildlife Research Center
Global Positioning System (GPS) tracking of animals is the latest major development in wildlife telemetry. It uses a GPS receiver in an animal collar to calculate and record the animal's location, time, and date at programmed intervals, based on signals received from a special set of satellites.
In 1973, the United States Department of Defense (DoD) began developing a Global Positioning System primarily to provide 24-hour, complete global satellite coverage for military purposes. In 1993 the GPS reached initial operational capacity when the 24th satellite was in place (Rodgers et al. 1996; Tomkiewicz 1996). Each satellite contains an almanac of all the other satellite positions, its current position, and the exact time.
With satellite telemetry, in contrast to GPS telemetry, the animal's PTT is a transmitter sending information to the satellite receivers which relay this information to a recording center on Earth. With GPS telemetry, a different set of satellites function as transmitters, while the animal's telemetry unit acts as a receiver. The signal information is used by the animal's telemetry unit to calculate its location based on current positions of satellites and the time taken for the signal sent from each satellite to reach the animal's receiving unit. These location data are then stored by the animal's unit for later unit retrieval or remote downloading.
At least four GPS satellites are always in view from any position on Earth, with each satellite orbiting approximately every 12 hours. This allows for 3D-position acquisition based on the four variables (latitude, longitude, altitude, and time/receiver clock bias). When line-of-sight to a particular satellite is obstructed due to, for example surrounding topography, a 2D position can be calculated using three satellites and three variables (latitude, longitude, and time/receiver clock bias). Altitude in a 2D fix is automatically calculated by either using the last known altitude from a 3D fix or by averaging a subset of the recent known altitudes (Rodgers et al. 1996).
Originally, the DoD incorporated intentional error into GPS signals received by civilian users for reasons of national security. This incorporation of intentional error with GPS data was called "Selective Availability" (SA). Under SA, accuracy was still very good, within 100 m 95% of the time (U.S. Department of Defense 1984).
However, using differential correction (DGPS), civilian users could correct for the intentional error (Moen et al. 1997) and also eliminate much of the error caused by ionospheric and tropospheric delays, with a resulting accuracy of 5 m, and with very expensive sophisticated correction equipment, within centimeters (Rodgers et al. 1996). In May 2000, the SA policy was abandoned by U.S. authorities, affording standard wildlife GPS units the approximate accuracy of differentially-corrected units under SA (Dussault et al. 2001).
In 1994, Lotek Engineering, Inc. introduced the first animal-based GPS location system, the GPS_1000. Since its introduction, size has been reduced, longevity increased, and data storage and retrieval have improved. Today's standard collar consists of a GPS receiver and antenna, a VHF beacon system (for location backup and system verification), data handling and control hardware, and power supply (Rodgers et al. 1996).
Originally, the Lotek GPS_1000 collar weighed 1.8 kg and was too heavy for mid-sized mammals (Rodgers et al. 1996). A second generation of Lotek's original collar, the GPS_2000, is small enough for large cats, deer, wolves, and bears. Similarly, Telemetry Solutions offers GPS "Simplex" collars that weigh as little as 600 g, appropriate for an animal the size of a mountain lion or wolf. For these animals, the GPS Simplex collars are advantageous because they allow remote data downloading, whereas the Lotek GPS_2000 requires collar retrieval for data acquisition.
The GPS-Simplex is powered by two batteries, one for the GPS receiver, data storage, VHF beacon, report transmission, etc. and the other for the VHF beacon after the first battery has expired. When the collar is using the back-up battery, the pulse rate of the VHF beacon changes to alert the researcher that the first battery has expired and GPS fixes are no longer being taken. Once the collar switches to the back-up battery, the VHF beacon runs for approximately 6 more months during which time the researcher can try to retrieve the collar (see below). Telemetry Solutions also markets GPS backpack units as light as 70 g and minimalist GPS collars weighing 120 g.
General weight reduction has also indirectly affected longevity of GPS collars (Tomkiewicz 1996). For example, suppose a researcher previously used a 1,600-g collar on a large bear. If a 1,200 g collar became available for the same animal, 400 g would be available as "extra" weight for an increase in battery size and therefore, longevity. Longevity has also been increased by duty-cycling the VHF beacon to turn off when not needed (e.g. at night or for longer periods such as 3 months of hibernation).
The greatest drain on GPS collar batteries occurs when the system searches for satellite signals to acquire a location fix. The search time is critical to collar longevity. In many areas, location acquisition requires 2-4 minutes because of cover and topography. However, Televilt has produced a GPS collar (POSREC-Science) that can obtain a fix in 10 seconds under ideal circumstances. Further advances in decreasing the signal-acquisition time will greatly increase battery life.
Other recent advances in GPS telemetry include new features such as an indicator of time-in-mortality, a mechanism to automatically drop off the collar (for data acquisition and/or collar re-use), field-replaceable batteries, temperature and activity sensors, and remote two-way communication. These features help minimize researcher invasiveness to the animal by reducing animal handling time and by condensing the means for various studies into one data-collection device.
For example, using an automatic drop-off mechanism, the researcher does not need to recapture the animal to retrieve the collar. Remote two-way communication also minimizes animal contact since the communication link can be used in some models to reprogram scheduling of the fixes and other parameters of the unit without having to recapture the animal and retrieve the collar.
Field-replaceable batteries mean only one recapture instead of two. Units with batteries that are not field-replaceable must be returned to the company for refurbishing which requires a minimum of two more captures after the initial deployment, i.e. one to regain the collar with the expired GPS battery and another to place the newly refurbished collar back on the animal. Since field-replaceable batteries can be changed so much more quickly without having to ship them to a company, they also minimize lost opportunities for data acquisition while the GPS unit is not functioning.
Features such as time-in-mortality and temperature and activity sensors allow researchers to combine location-data-collecting projects with other physiological and ecological studies that may have previously required separate investigations. For example, researchers can use the GPS unit to temporally correlate the animal's activity (i.e. moving, resting, feeding) to ambient temperature while still obtaining location data.
Three main methods of data storage and retrieval are used in GPS telemetry: 1) on-board storage for later collar retrieval and subsequent downloading, 2) remote downloading to a portable receiver, 3) remote relaying through the Argos satellite system. There are advantages and disadvantages to each type of data storage and retrieval.
GPS Data Stored On Board. Collars with only store-on-board capabilities minimize researcher effort and invasiveness to the animal (only one handling required) since the collar is simply attached to the animal and later retrieved after an automatic or remotely triggered drop-off mechanism has released the collar from the animal. The data are then simply downloaded all at once from the collar (Merrill et al. 1998). Another advantage of the store-on-board collars is their relatively smaller size. Store-on-board collars contain comparatively smaller circuitry and are less complex than other types of GPS collars and thus can carry heavier (longer lasting) batteries for the same overall collar weight (Tomkiewicz 1996).
Since the store-on-board collars are less complex, they require less hardware (e.g. special field receivers) so are less expensive. Also, collars with remote or automatic break-away or drop-off mechanisms are advantageous because the retrieved collar can be sent back to the manufacturer for refurbishing and later reused resulting in increased cost savings (Merrill et al. 1998). ATS, Lotek, Telemetry Solutions, and Telonics all offer collars with store-on-board capabilities.
The main disadvantage when using a store-on-board-only GPS unit is data loss. If a GPS collar fails to release, all the data are lost unless the animal can be recaptured (Merrill et al. 1998). Also, since there are no intermediate data reports, the unit could malfunction and not collect data or may collect data at the wrong intervals. Some units contain VHF beacons that alert the researcher to the status of the last location-attempt. Nevertheless, the beacon only indicates that the unit appears to be functioning properly; it does not transmit any data and therefore, if the collar is not retrieved, all the data are lost (Merrill et al. 1998).
GPS Data Downloaded To A Portable Receiver. The second method of data retrieval ensures that at least partial data recovery will occur even if the collar malfunctions and fails to release from the animal. This method allows remote downloading directly to the researcher throughout the study. The collar is programmed to transmit data through a VHF signal (some systems use FM-relay devices or a UHF modem) to the researcher's receiver.
Researchers can receive daily reports repeatable up to 5 times per day, or as infrequently as once per week. This timely retrieval of data allows biologists to supplement the location information with field data. For example, if location data from a carnivore indicates that the animal spent much time in a concentrated area, that may indicate the location of a kill. The researcher can then try to find the kill using a hand-held GPS navigation unit.
Interpretation of the GPS reports can also alert the researcher to a malfunctioning GPS unit or suggest changes in programming for more optimal data collection. With two-way communication, sampling regimens can be remotely altered if initial data reports indicate another location-acquisition routine may be more appropriate.
A vital feature with this type of GPS unit is long-term data retention following remote data transmission. Units that follow data transmission with a complete memory sweep are undesirable because often reception of the transmitted reports may not always be successful (Zimmerman et al. 2001). While intermittent reports are valuable in allowing data analysis throughout the study, long-term, on-board data storage completes the picture by allowing the researcher to fill in any blanks when the collar is retrieved. Telonics and Telemetry Solutions both offer collars with remote data downloading for large animals, but at present only Telemetry Solutions markets these collars for small-to-medium-sized mammals.
Some disadvantages to this method include the relative increase in complexity, and therefore, weight and expense of both the animal's telemetry unit and the receiving equipment. Apart from the added cost of the equipment itself, it takes additional labor to retrieve the intermediate data reports. To retrieve the reports, the researcher must be within VHF receiving range, 5-10 km ground to ground, within 15-20 km air to ground, or for UHF, 15 km line-of-sight (Rodgers et al. 1996).
GPS Data Relayed by Satellite. The third main method of data retrieval and storage for GPS telemetry uses the Argos satellite system to relay the intermittent data reports. Thus the researcher needs neither to be in the field to collect the data reports, nor to maintain special receivers or other additional equipment. Lotek specializes in these types of GPS collars.
Disadvantages include the added bulk and weight of the animal's telemetry unit since transmitting data to satellites takes more power. This added weight limits the size of animal that can tolerate this type of GPS unit. In addition, the researcher must also pay Argos to relay data information through its satellites. Furthermore, to remotely change sampling schedules and report frequency, one must purchase a separate portable receiver/interrogator, adding expense.
Global Positioning System (GPS) tracking allows the researcher to obtain data on animal location in all weather as frequently as every minute or as infrequently as once per week with potential accuracy of within 5 m (Moen et al. 1996). While GPS units afford increased accuracy, their longevity is much less than that of conventional VHF units. VHF units for wolf-sized animals usually last about 4 years, whereas current GPS units rarely last longer than 1 year. GPS tracking is also expensive (see below).
However, per data point or for large, expensive studies, the costs of GPS tracking can be cheaper than for conventional VHF radio-tracking (see below). This is because for a given unit of researcher labor, GPS radio-tracking can gather many more location data. On the other hand, the types of data points differ. With GPS data, the points are usually serially correlated, whereas with standard radio-tracking they often are not, depending on their time intervals. In addition, biases in the data must be considered because of differential interference of various habitat types with the receivability of the GPS signal (Merrill 2002).
Furthermore, studies based on GPS tracking frequently use fewer individual animals because of the expense per GPS unit (Otis and White 1999). If the animals themselves are considered the study unit, this reduced sample size can cause data-analysis problems when generalizing about a population (White and Garrott 1990; Rodgers et al. 1996; Otis and White 1999).
Since GPS telemetry for wildlife is relatively new, most studies have involved testing the reliability and accuracy of the equipment in varying environments and applications. Performance of various GPS collars have been tested for moose (Rempel et al. 1995; Moen et al. 1996; Rodgers et al. 1996; Dussault et al. 1999), white-tailed deer (Merrill et al. 1998; Bowman et al. 2000), and wolves (Merrill et al. 1998; Merrill and Mech 2000; Merrill 2002). The collars have functioned well, especially the most recent versions, which can be placed on an animal when it is most easily captured and can be programmed to begin duty cycling some months later (Nelson and Mech submitted).
No doubt, tests of GPS technology for wildlife will continue since new products are still rapidly forthcoming. For example, recent weight decreases have made remote-data-downloading GPS collars available for use on wolf-sized animals. Furthermore, with the establishment of baseline accuracies and statistically appropriate research applications, along with increased awareness of the potential for highly accurate data, increasing numbers of studies using GPS telemetry can be expected. Also, the cost should eventually decline to a more affordable level. Improvements such as these will hasten the use of GPS for a greater range of species.
A single GPS collar usually ranges from $3,000 to $4,500, about 10 times that of a VHF collar for mid-sized mammals (Merrill et al. 1998). An example start up package for one animal fitted with a remote-data-downloading GPS collar costs about $10,500. This includes a receiver (about $5,000), software with supporting cables (about $2,000), and a collar with a drop-off mechanism and one extra battery for field replacement (about $3,500). The cost of additional animals fitted with GPS collars is much less than the first in that the same receiver can be used for many collars. It is also important to note that GPS collars are reusable, with only drop-off mechanism ($275) and battery ($187) needing replacement.
Although GPS systems cost much more than VHF systems, this does not necessarily mean they are less economical. When cost/location is considered, as opposed to cost/animal, GPS collars can be the cheaper alternative and also save personnel costs since the study may be less labor intensive.
For instance, after examining multiple options including VHF and satellite telemetry, Rodgers et al. (1996) found that GPS-based telemetry was the most economical and logistically feasible method to track 60 moose located monthly with a subset of 20 moose located 35-50 times during three periods of intensive monitoring (early winter, late winter, and spring-summer-fall). On the other hand, for such studies as mortality investigations, the much longer life of VHF transmitters must be considered.
Each telemetry system has its advantages and disadvantages. Within each system there are also options to specifically tailor the telemetry packages to the researcher's unique needs. However, some generalizations apply when deciding which type of telemetry is most appropriate for a particular study (Merrill and Mech 2000, Merrill 2002).
If funding for a study is low or if a large number of animals are to be studied for long periods, VHF telemetry is the only option. Furthermore, VHF units can be used on virtually any animal whereas satellite and GPS telemetry units are often heavier and thus limited to medium-to-large mammals, except that solar-powered units can be used on birds. Another advantage of VHF units is their long history of use. Therefore, they are generally more reliable than the newer technology in GPS units.
However, VHF telemetry is generally more labor-intensive and less accurate. The costs of increased labor and transportation and the researcher's flexibility about data quality must be considered. While VHF is not as accurate as GPS telemetry, it can be combined with direct observations (following homing in on the animal) for finer-scale studies (Mech 1980).
Although satellite telemetry is more expensive than VHF tracking, in some cases it may be the only option, for example, for far ranging species such as transoceanic migratory birds or offshore marine mammals (Rempel et al. 1995). Satellite telemetry, as with conventional VHF telemetry, is not usually an appropriate method for fine-scale (25-250 m) habitat studies (Rempel et al. 1995).
GPS telemetry is the most accurate form of tracking apart from visual confirmation of the animal's location, so GPS telemetry can be used with reasonable confidence for relatively fine-scale habitat studies (Moen et al. 1996; Rodgers et al. 1996). Additionally, GPS affords one benefit that visual confirmation may not. GPS units accurately locate an animal without the researcher's immediate presence. This means less researcher-introduced disturbance and therefore, potentially lower probability of unnatural animal behavior. This translates into less biased data.
A principal advantage of GPS units is the number of locations acquired per animal. For example in a 30-day period, 2,880 locations per animal can be acquired with a GPS unit programmed for 15-minute fixes. Additionally, GPS can be used in all types of weather all year round (Moen et al. 1996).
GPS telemetry is not without its drawbacks, though, cost being chief among them. When costs are prohibitive, researchers often compromise sound statistical sampling methods such that their results are based on a small number of animals carrying transmitters (Rodgers et al. 1996). Another disadvantage, when using GPS units is that they generally do not last longer than 1-1½ years.
Regardless of which telemetry system is selected, potential effects on an animal's normal behavior must be considered whenever an animal is handled or instrumented (Cochran 1972; Marks and Marks 1987; Vaughan and Morgan 1992). It is to the researcher's advantage to minimize these effects since the goal of radio-tracking is to obtain data most closely reflecting the animals' natural behaviors.
Adverse effects from capturing and radio-tagging an animal can range from short to long-term and from apparently tolerable to severe or fatal (Birgham 1989). Whether specific effects are important in a study depends upon the objectives of the study (White and Garrott 1990). Many of the usual deviant behaviors last only 1-2 weeks. Therefore, some workers recommend that data should not be considered reliable until after at least 1 week of acclimation to the radio-tag (White and Garrott 1990).
Conceivably radio signals themselves could have some ill effect on the animal wearing a collar. However the ERP from VHF transmitters is so low that this possibility seems highly unlikely. Although the radiated power of PTTs is several orders higher than for conventional animal-tracking transmitters, there have been no findings of detrimental effects to the animal during 360 mS transmissions from the PTT (Taillade 1992).
An animal could display behavioral and energetic deviations as a result of the capture, handling, transmitter package, or presence of the radio-tracker. No data can actually prove that radio-tagging or radio-tracking an animal has no adverse effect. Rather, results can only show that no effect was detected when tested using the specified statistical power (White and Garrott 1990).
Generally, if a study animal maintains its weight, successfully mates, establishes and defends a territory, and/or produces offspring, and otherwise appears to look and behave normally, researchers consider the effects to be of minimal impact (Mech 1983, White and Garrott 1990). Researchers interested in informally monitoring adverse effects can watch for signs in their recaptured animals such as consistent weight loss from the first capture throughout the study (which would suggest movements are hindered and that the animal is likely more susceptible to predation), chaffing or hair loss under the collar, etc. (Mech 1983).
Experiments designed to detect adverse effects from radio-tagging and radio-tracking have focused mainly on birds. Most birds are relatively light and depend upon flying for survival. Therefore, one might expect that negative effects from the transmitter's weight and attachment packages would be easier to detect on a bird than on a large mammal, for example. Researchers studying the effects of radio-tagging on birds have been primarily concerned with the transmitter-to-body weight ratio (White and Garrott 1990).
While the "rule of thumb" for complete transmitter packages dictates that they weigh no more than 3-5% of the total body weight (Cochran 1980), subsequent studies have shown the importance of species-specific considerations (Caccamise and Hedin 1985; Aldridge and Brigham 1988; Anderka and Angehrn 1992). Some studies have examined not only the impact of the transmitter pack's weight and bulk, but also the effects of capture, handling, and tracking (Hill and Talent 1990; Taylor 1991).
Investigations such as the above have resulted in greater awareness of the adverse effects on the animal's behavioral patterns, survival, and reproductive success that a researcher might expect when conducting a radio-telemetry study. Direct effects of the transmitter packages themselves can include antennas and attachment packages becoming snagged in vegetation (Dunstan 1977; Jackson et al. 1977), animals themselves becoming entangled in loose collars or harnesses (Schladweiler and Tester 1972; Hirons and Owen 1982; Hines and Zwickel 1985), chaffing or feather loss (Bartholomew 1967; Hessler et al. 1970; Corner and Pearson 1972; Greenwood and Sargeant 1973; Perry 1981; Kenward 1982; Wywialowski and Knowlton 1983; Hines and Zwickel 1985; Jackson et al. 1985), electrocution in birds fitted with whip antennas while perched on wires (Dunstan 1977), and increased drag when swimming (Wilson and Wilson 1989), lifting, or flying (Pennycuick 1975; Pennycuick and Fuller 1987; Obrecht et al. 1988; Pennycuick 1989; Pennycuick et al. 1989).
Other avian investigations into the effects of radio-tagging have documented increased time spent in comfort activities such as preening (Greenwood and Sargeant 1973; Gilmer et al. 1974; Siegfried et al. 1977), attempts to remove the transmitter package (Perry 1981; Hooge 1991), weight loss (Perry 1981), abandonment of brood (Horton and Causey 1984), reduced time spent in flight (Gessaman et al. 1991), increased metabolism and energetic output (Gessaman et al. 1991), avoidance of water (Greenwood and Sargeant 1973), decreased courtship activity (Ramakka 1972), decreased feeding activity (Boag 1972), decreased clutch survival (Amlaner et al. 1978), lower survival rates in initially low-weight birds (Johnson and Berner 1980), greater susceptibility to predation (Sargeant et al. 1973; Siegfried et al. 1977; Erikstad 1979), and decreased reproductive success (Massey et al. 1988; Paton et al. 1991; Foster et al. 1992).
The above studies documenting adverse effects on avian reproduction are contrasted by those conducted by Boag et al. (1973), Johnson (1971), Amlaner et al. (1978), Kalas et al. (1989), Sodhi et al. (1991), and Taylor (1991) who found no significant effects on the reproductive success of red grouse, ring-necked pheasants, herring gulls, great snipe, merlins, or barn owls respectively. While the spotted owl study by Foster et al. (1992) found decreased reproductive success by birds fitted with a backpack harness, the studies also documented no difference in mean mass of spotted owls before and after wearing the radio transmitters. The researchers suggested the use of tail-mounted transmitters to avoid the apparent reproductive success problems caused by the larger backpack transmitters.
Kenward's (1977) study detected no difference in weight loss or dispersal behavior between goshawks fitted with transmitters and those fitted with leg bands. He also found comparable hourly feeding rates for sparrow hawks before being fitted with a transmitter and afterward (Kenward 1978).
Although the majority of radio-tagged mammals are large predators or ungulates, most studies on the impacts of radio-tagging on mammals have concerned smaller mammals such as black-tailed jack rabbits, meadow voles, and lemmings (White and Garrott 1990). Instrumented small mammals have shown impaired movements (Banks et al. 1975), decreased digging ability (Corner and Pearson 1972), and decreased survival (Webster and Brooks 1980; Wywialowski and Knowlton 1983).
Aldridge and Brigham (1988) found that the 5%-body-weight "rule" caused adverse effects in maneuverability when applied to bats within a certain size range. That study showed that adverse impacts from the weight of transmitter packages should be examined not only uniquely for each species but also for size variations within each species if no such data from closely related species exist.
Larger mammal radio-tagging impact studies have been limited but include river (Melquist and Hornocher 1979) and sea otters (Garshelis and Siniff 1983), mule (Goldberg and Haas 1978; Wenger and Springer 1981; Garrott et al. 1985) and white-tailed deer (Nelson and Mech 1981; Clute and Ozoga 1983), caribou (Pank et al. 1985), and mountain lions (Garcelon 1977).
Two studies involving the impacts of radio-tagging white-tailed deer noted adverse effects. Nelson and Mech (1981) conducted a mortality study with white-tailed deer fitted with collars taped with yellow (for ease in aerial spotting). They learned that the bright collars also allowed hunters to more easily see the deer and, therefore, the mortality data were biased with respect to hunting pressures. Clute and Ozoga (1983) noted that during cold spells white-tailed deer fawns with expandable collars collected heavy ice on their necks, probably from water splashed up when they crossed streams.
While both these studies found detrimental impacts, Hamilton (1976) saw no weight loss in leopards before and after fitting with transmitters. Gorman et al. (1992) recorded no adverse effects on African wild dogs when fitted with a PTT weighing just under 900 g. Neither did Creel et al. (1997) find any adverse effects as measured by stress hormones in African wild dogs following typical anesthesia and radio-collaring.
Garrott et al. (1985) examined the hypothesis that mule deer fawns fitted with collars, as opposed to ear-tag transmitters, suffered greater predation rates due to collars providing coyotes a more secure grip on the fawn's neck. The authors found comparable survival rates between fawns with collars and those with ear-tag transmitters. Although their study included 91 animals, subsequent analysis (White and Garrott 1990) revealed that the sample size was short of the 96 animals required to detect a 30% difference in predation rates with 80% certainty. This example highlights the importance of study design with respect to statistical power when investigating impacts of radio-tagging and tracking.
Surgically implanting transmitters usually involves additional trauma to an animal and possibly requires recapture(s) to administer follow-up care (Morris 1980). However, several studies found no lasting negative impacts from the implant or the surgical procedure. Klugman and Fuller (1990) documented that implantation had no discernable effect on feeding, foraging, alert, and walking behaviors in sandhill cranes, and cranes dominant before surgery remained so afterwards. Van Vuren (1989) found similar pregnancy rates and mean litter sizes between female yellow-bellied marmots surgically implanted with transmitters and females not implanted. Reid et al. (1986) also concluded that implants did not affect the reproductive cycle in river otters; copulation, embryonic and fetal development, and lactation behaviors were all normal in surgically implanted river otters.
In slight contrast to the above, 3 of 10 beavers implanted developed adhesions between the transmitting capsule and peritoneal tissues (Guynn et al. 1987). One beaver died from intestinal obstruction as a result. The other 9 beavers were necropsied and had no indications of harmful pathology. The authors suggested encasing the implant in a layer of omentum during surgery to avoid complications caused by adhesion.