Proximity Logger Units

The proximity logger units deployed on foxes on Miguel were purchased from Sirtrack Limited (Hawkes Bay, New Zealand). Sirtrack was the only manufacturer producing proximity loggers sufficiently small in size and mass to meet the 4% of body weight limit for telemetry devices placed on island foxes. The units purchased were model E2C-171-A proximity loggers. These units consisted of a UHF transceiver bundled with a VHF transmitter in an acrylic housing and mounted on a machine belting collar (Fig. 1c). The entire unit weighed 60 g. The proximity logger units used the UHF transceiver to communicate with other units. Specifically, when 2 units came within a specified distance of each other, both units detected and identified the other unit (i.e., a “contact”), and each recorded the duration of the contact. The maximum distance for a contact and the separation time required to end a contact was programmed by the user. The contact information was stored within the unit, and units had to be retrieved in order to download the stored data. The VHF transmitter permitted instantaneous tracking of the unit. Battery life for the units was estimated at 276 days, and each unit cost $499.

The units were easy to program and all were programmed identically. NPS staff chose a 30-m proximity for indicating a contact between foxes. This distance was set by adjusting the “UHF coefficient,” and the appropriate coefficient was determined based on tests with the units placed on a saline bag to simulate a fox's body. Thus, a contact was recorded if 2 foxes wearing units came within approximately 30 m of each other. The contact duration terminated once the animals moved apart and the 2 units were not in contact for >30 s. A base station was also deployed to monitor fox presence at a specific location. This station was programmed to detect foxes within a distance of about 60 m, and the contact was terminated if the fox moved >60 m from the station for >30 s. Approximately every 2 weeks, data from the base station were downloaded and the station was moved. Prior to deployment on foxes, each proximity logger was tested to ensure that it detected other units and that its VHF transmitter was operating properly.

To deploy the proximity logger units, island foxes on Miguel were captured in live traps by NPS staff using the same methods described above for the GPS unit deployment on Rosa. Selected foxes were then fitted with a proximity logger unit and released at the capture site. Field testing consisted of obtaining general locations on the foxes approximately once every 2 weeks by tracking the VHF signals. At the end of the field-testing period, live traps were set in the areas used by each fox in an effort to recapture foxes and remove the proximity logger units. Data were then downloaded from the units to assess their performance under field conditions.

Table 1.

Parameters evaluated to assess the performance and utility of GPS and proximity logger units for island fox monitoring and research.

A variety of parameters were assessed to evaluate the performance of the proximity logger units and their potential utility in island fox monitoring and research (Table 1). As with the GPS unit evaluations, some parameters were quantitatively measured and some were qualitatively evaluated; and we tried to collect as much information as possible to assist in evaluating unit performance and utility.

Each proximity logger unit recorded data independently. Therefore, 2 units ideally recorded the same number of contacts with each other and the duration of these contacts should have been similar. The number of contacts recorded and the total duration of contacts between unit dyads were examined to determine how closely they matched.

GPS Units

In November 2007, 20 Televilt GPS units were deployed on foxes on Catalina: 5 males and 5 females in both burned and unburned areas. Unit-to-fox mass ratios ranged from 2.48% to 3.06% for females and 2.06% to 2.77% for males (Table 2). Tracking the foxes was difficult. The UHF transmitter activated only once a week for a relatively short (3-h) window; and due to a miscommunication with the manufacturer, that window was from 23:00 to 02:00. Although foxes tend to be active at this time and more easily detected, the window was not a convenient time to conduct fieldwork. Furthermore, darkness at this time complicated attempts to track the one collar that malfunctioned and automatically dropped off too early. Indeed, that collar was never recovered, and this issue led to the decision to attempt to recapture all collared foxes to recover the units prior to drop-off. A root cause of this issue was the fact that the units had to be programmed at the factory and not by the user. The UHF transmitter did emit a strong signal that could be heard up to distances of 1000 m. However, the gain on the signal could not be lowered, and therefore it was very difficult to obtain a signal direction when in close proximity to the unit. This effect contributed to the difficulty in locating the dropped collar.

Table 2.

Biological data and details on GPS and proximity logger units worn by island foxes during 4 projects to evaluate the efficacy of the units.

continued

The 20 foxes wore the units for 187–214 days (Table 2). One fox was found dead toward the end of the study, but the carcass was too decomposed to determine the cause of death and whether the GPS unit may have contributed to the death. Of the remaining 19 foxes, 18 were recaptured and the units were removed prior to the scheduled automatic drop-off date. The 19 units recovered were monitored, and the automatic drop-off mechanism activated properly on 14 of these. Data were recovered from 17 of the 19 units. The 2 units that could not be downloaded were sent back to the manufacturer, who also was unable to recover any data. It is unclear whether the 2 units failed to collect data or whether data were collected but could not be accessed. These 2 units also were among the 5 for which the automatic drop-off mechanism failed. Of the 17 units from which data were recovered, the maximum number of locations collected was 197 and the average number was 118 (Table 2). The number of locations obtained relative to the expected number was generally very low and averaged 6.1% (range 1.7%–10.2%). Recaptured foxes did not exhibit any injuries associated with the GPS units. All of the foxes lost weight (range 0.10–1.00 kg) while wearing the units.

For the Rosa project, 16 GPS units were purchased from Telemetry Solutions. These included 9 “store-on-board” (SOB) units and 7 “remote download” (RD) units. The software required for programming the units was continually being updated, which caused some problems. However, once the proper software was obtained, the units were relatively easy to program.

Several issues were encountered during predeployment testing of the GPS units. A significant problem was difficulty in deactivating the units after testing. Seven of the units experienced software errors that apparently prevented data download. For one of these units, actual time intervals for collecting locations were inconsistent with those programmed; and for another unit, the time on the internal clock was not stable and shifted causing locations to be recorded at incorrect intervals. The VHF transmitter on one of these units also exhibited a rapid pulse rate that would have significantly reduced battery life. The VHF transmitter on another unit was not operating when the unit was delivered. All malfunctioning units were returned to the manufacturer for repair, but 2 units were not sent back in time for deployment on foxes. Finally, during testing of the remote download function, only one attempt to download data was successful. Apparently, the whip antenna supplied with the base station had an extremely short range (approximately 10 m). However, switching to a Yagi antenna (3 or 6 elements) resolved this issue, and subsequent download attempts were more successful.

During annual trapping efforts on Rosa conducted by the NPS, 14 GPS units were deployed on foxes during September-November 2009. In an effort to provide a larger safety margin in the event that the weight of the units proved to be a burden for foxes, all of the units were placed on adult males because of their larger body size. Fox weights ranged from 1.75 to 2.9 kg, resulting in unit-to-fox mass ratios of 2.2%–3.7% (Table 2). All the units were deployed on foxes on the east side of the island to facilitate monitoring. Of the 14 units deployed, 8 were SOB and 6 were RD.

After the units were deployed on foxes, the most prevalent and serious problem encountered during monitoring was the performance and reliability of the VHF transmitters, all of which failed prematurely. Among the 14 units deployed, the average number of days that VHF transmitters operated was just 27. Five transmitters failed after only 1 day of operation, and the longest any functioned was just 92 days. Signal strength was relatively weak on all of the units, and consequently, the signal detection distance was quite short. Typically, signals could be heard only from a distance of ≤100 m and often only when in the lineof-sight distance (i.e., no topographic or other features between the transmitter and observer). Thus, even when the transmitters were operating, the weak signal strength and short distances made tracking the units challenging. Another problem associated with the VHF transmitters was malfunctioning mortality sensors. Three of the units emitted false mortality signals; in all 3 cases, live foxes were observed or recaptured, thereby confirming the false signals.

While the units were deployed on foxes, attempts were made to remotely download data from the 6 RD units. Such efforts were rendered significantly more difficult by the VHF transmitter failures described above. Downloads could be performed only at a maximum distance of approximately 300 m. Thus, getting within this distance was challenging without the aid of the VHF transmitters. Fortunately, most foxes remained in the general vicinity of the location where they were trapped and collared.

In December 2009, downloads were attempted from the ground on 3 units, of which only one had a functioning VHF transmitter. The downloads were successful, and the data from one unit indicated that the fox was likely dead because all of the most recent locations were from a single location. The coordinates for this location were used to conduct a ground search, and the carcass of the fox was located and collected. Cause of death could not be determined due to the advanced state of carcass decomposition, but there was no evidence to suggest that the GPS unit was responsible for the death. In February 2010, downloads were attempted on the remaining 3 RD units from the air by using an antenna attached to a fixed-wing aircraft. None of these remaining units had functioning VHF transmitters, but because the foxes all were in the general area where they had been trapped, the download attempts were successful.

Live trapping was conducted from June 2010 to January 2011 in an effort to recapture foxes and recover the GPS units, and 8 foxes were recaptured. No injuries associated with collars were observed among the recaptured foxes that wore GPS units. Excluding the fox recovered dead, the other foxes recovered wore the units for 274–432 days (Table 2). Of the 8 foxes recaptured, 4 had gained weight, 3 had lost weight, and 1 was the same weight (Table 2). The weight losses were not considered excessive, particularly given the relatively long period that the foxes wore the units. The condition of the units upon recovery varied. Most exhibited excessive wear on the ends of the epoxy housing that resulted in exposed wires. The antenna on one unit was broken off where the antenna exited the housing.

GPS units were recovered from the 8 foxes recaptured and from the one fox found dead. Of the 5 foxes not recaptured, 3 had RD units and data were successfully downloaded from these resulting in at least partial GPS data sets for 12 of the 14 foxes that received GPS units. All 12 data sets included usable locations. Excluding the partial data sets from the 3 unrecovered units and the fox found dead, the average operational time for the remaining 8 units was 17.5 weeks (range 10–28 weeks). Of these 8, two malfunctioned and ceased operating due to damages (e.g., broken antenna or transmitter housing). The expected operational life of the GPS battery varied from 23 to 30 weeks, depending on the frequency of location attempts programmed into each unit. Only one unit met or exceeded the expected operating time. On average, the units successfully collected a location in 81.7% (range 73.0%–92.8%) of attempts. The units collected an average of 357 (range 223–552) locations while they were deployed (Table 2).

In the second GPS collar study on Catalina, 5 Telemetry Solution GPS units were deployed on foxes (3 males and 2 females) in November 2010. Fox weights ranged from 2.05 to 3.00 kg resulting in unit-to-fox mass ratios of 2.33%–3.41% (Table 2). The study for which the units were deployed was relatively short, and foxes were retrapped in January 2011 to remove collars. Four of the 5 foxes were recaptured, and these 4 had worn the collars for 47–58 days (Table 2). All 4 had hair loss and abrasions on their necks from the collars. Also, all of the foxes had lost weight (range 0.15–0.35 kg). One unit malfunctioned (both the VHF and GPS functions ceased working) just before the fox wearing that unit was recaptured. Data were successfully downloaded from all units during the study. The number of locations collected ranged from 178 to 517. The proportion of attempts in which the units successfully obtained a location ranged from 19.7% to 56.0%. The range on the VHF transmitter was relatively short and was estimated at 150–300 m. However, the transmitters were readily heard during aerial searches.

Proximity Logger Units

For annual trapping efforts conducted by the NPS on Miguel, 17 proximity logger units were deployed on foxes during December 2009–January 2010. All were deployed within an approximately 3.5-km² area to facilitate monitoring and to increase the potential for recording contacts between individuals. Units were placed on both males and females and on foxes ranging in age from <1 year to >7 years. Fox weights ranged from 2.05 to 2.75 kg, resulting in unit-to-fox mass ratios of 2.2%–2.9% (Table 2).

The performance of the VHF transmitters on the units met expectations. None of the transmitters failed while deployed on foxes. Also, there were no observed deviations in signal strength, frequency, or pulse rate during the period of deployment.

One of the foxes with a proximity logger was found dead on 27 August 2010. The carcass was too decomposed to determine the cause of death; but the fox had worn the unit for several months and also had worn multiple conventional radio-collars in the past, and NPS staff felt that the logger unit likely did not contribute to the death. Live trapping was conducted during summer and fall 2010 in an effort to recapture foxes and recover the proximity logger units. Of the 16 foxes still wearing units, 14 were recaptured during July–August 2010, one was recaptured in November 2010, and one eluded recapture.

Excluding the fox recovered dead, the recaptured foxes wore the proximity logger units for an average of 226 (range 205–323) days (Table 2). No injuries associated with collars were observed among the foxes. Of the 15 foxes recaptured, 4 had gained weight, 5 had lost weight, and 5 were the same weight (Table 2). The weight losses were not considered excessive, particularly given the relatively long period that the foxes wore the units. Generally, the units appeared to be in good condition upon recovery. They exhibited some wear, but none of the wear was considered excessive for collars deployed on wild foxes for multiple months. The housing and antenna were intact on all units.

The proximity logger units all recorded contacts between foxes wearing the units. Of the 16 recovered units, 12 recorded at least 1000 contacts, and 3 units recorded >4000 contacts (Table 3). Difficulties were detected for only one unit: it malfunctioned after 177 days on the fox and afterward recorded 16,516 unusable records. Otherwise, all units performed as expected. Most of the foxes were recaptured prior to the estimated termination of battery life (276 days), but the last fox recaptured wore the unit for 323 days and the unit was still operating. After recovery, the units could not be deactivated by simply passing a magnet near the external activation site and magnets had to be taped to units in order to deactivate them, but this was a minor issue. Also, the base station failed within a few weeks of deployment, but a replacement sent by Sirtrack worked fine during the remainder of the project.

The number of contacts and the total duration of contacts between units within a dyad generally were very similar (Table 3). For example, M265 recorded 63 contacts with F352, totaling 2134 s; whereas F352 recorded 58 contacts with M265, totaling 2257 s. This general concordance was common among almost all the dyads, indicating a relatively high level of accuracy among the data recorded by the units.

GPS Units

The performance of the GPS units was mixed. Some aspects of these units were very successful, whereas others fell well short of expectations or advertised performance. The units were generally worn successfully by island foxes. Some minor injuries were noted on the 4 foxes recaptured on Catalina that wore Telemetry Solutions units. These injuries were likely due to the positioning and placement of the VHF and satellite antennae. Instead of running parallel along the animal's back, the antennae were mounted perpendicular to the ground for better satellite reception. The antenna extended approximately 10 cm above the fox's head and therefore struck vegetation as the fox navigated through dense brush. These impacts most likely caused the collar to rock front to back on the animal's neck, thus causing the noted abrasions. However, no injuries were noted on the 8 foxes recaptured on Rosa that also wore Telemetry Solutions units.

Table 3.

Contacts between island foxes wearing proximity loggers, San Miguel Island, 2009–2010. Values are the number of times that a fox in a row recorded contact with a fox in a column.

One fox with a Televilt unit on Catalina was found dead, but the fox was too decomposed to determine whether the unit contributed to the death. One fox died on Santa Rosa while wearing a Telemetry Solutions unit, but there was no evidence that the unit contributed to this death. Weight loss was commonly observed among foxes wearing GPS units, but again it is unclear whether the units contributed to this loss. On Catalina, the Televilt units were deployed in late fall when island foxes may be heavier, and they were removed in early summer when the foxes are commonly lighter. Winter to summer weight loss of similar magnitude also was observed among foxes that did not wear GPS units (J. King unpublished data). Thus, at least some of this weight loss may have been attributable to natural physiological patterns among the foxes. Kit foxes (Vulpes macrotis) exhibit similar seasonal variations in weight (Warrick and Cypher 1999).

The GPS units were bulkier than standard VHF units, which probably resulted from bundling multiple functions (e.g., GPS receiver, VHF transmitter, mortality sensor, associated batteries, antenna) into a single package. However, the additional bulk did not appear to produce any deleterious ef fects. The units also were a bit heavier than standard VHF units (typically <45 g). A general recommendation when using telemetry equipment to conduct research on animals is to limit the weight of the equipment to ≤5% of body weight. Federal and state permits for handling island foxes included a 3% or 4% limit, and the collarto-body mass ratios for all foxes receiving GPS units was <4% and in most case <3%. At this limit, the Televilt units could have been placed on foxes weighing ≥1.8 kg and the Telemetry Solutions units could have been placed on foxes weighing ≥1.625 kg. In an analysis of radio-collar effects on 542 endangered San Joaquin kit foxes (V. m. mutica), possible detrimental effects were detected only when equipment exceeded 6% of body weight, and these effects were primarily detected among juveniles (Cypher 1997).

The Televilt units were factory-programmed thereby reducing the chance of user error. However, due to a miscommunication, the units were programmed improperly with regards to the UHF transmitter time. This error made tracking the units significantly more difficult. For the Telemetry Systems units, once all of the proper software and instructions were in hand, the units generally were easy to program. Also, the variety of programming options allowed the user to more effectively address study objectives and also to maximize battery life.

GPS receivers need to communicate with orbiting satellites in order to calculate and record locations. Obstacles such as dense vegetation or topographic features and behaviors such as den use can impede communications between the units and satellites and thus result in failed attempts to obtain locations (e.g., Johnson et al. 2002). Island foxes occasionally use dens (Moore and Collins 1995) and commonly use areas of rugged terrain, such as canyons. The numbers of locations collected by the Televilt units were considerably lower than expected, although all of the units that did not malfunction did produce data. For the Telemetry Solutions units, the proportions of attempts for which successful locations were secured were relatively high, at least on Rosa, demonstrating that the units were quite effective in collecting the desired data. It is unknown whether a particular factor or factors (e.g., topography, vegetation, den use) were consistently associated with failed location attempts. During pretesting, the units successfully collected locations in grassland, chaparral, and mixed woodland habitats, although the rates of successful locations were not determined.

The most significant issue with the GPS function on the Telemetry Solutions units deployed on Rosa was that the duration of operation fell well short of expectations for all but one unit. Almost all of the units operated for several weeks, and in some cases several months, less than expected. Consequently, the number of locations collected also fell short of expectations. The expected battery life and programmed parameters should have yielded approximately 670 locations. However, even accounting for failed location attempts, only one unit achieved the expected number of locations (which was the same unit that also exceeded expected battery life).

Six of the GPS units on Rosa included a remote-download function, which was very successful when used with the proper antenna. The short range on the whip antenna supplied with the base station rendered it essentially incapable of remotely downloading data from units, particularly under field conditions. However, switching to a Yagi antenna resolved this issue. Data were successfully downloaded from all 6 units. Downloading from the ground did necessitate maneuvering to within about 300 m of the foxes, which positioning could be challenging depending on factors such as road access, topographic ruggedness, and vegetation density. However, downloading also was effective from the air. Aerial downloading may seem more expensive due to the costs of aircraft charter, but ground downloading could consume significantly more staff time and thus reduce or even negate any differences in cost efficiency between the 2 methods. Regardless of method, the immense value of the remotedownload function was highlighted by the inability to recapture 5 of the Rosa foxes. Data were obtained from 3 of these foxes that had collars with remote-download functions, whereas no data were collected from the 2 without this function.

The accuracy of the locations obtained by all models of GPS units was not precisely quantified. However, qualitative evidence suggests that the locations were reasonably accurate. The locations were effective in leading field biologists to the fox that had died on Rosa. Because the foxes are small and the carcass was decomposed and not obvious, the locations had to lead to a relatively small area for the carcass to be found. After the fox died, the GPS unit collected 90 locations. On average, these locations were <10 m (range 0.4–81 m) from the coordinates for the car cass provided by a hand-held GPS unit. Such precision is sufficient for conducting detailed spatial analyses, such as examining use of habitats and landscape features by foxes. Also, the locations from the collar GPS units were consistent with field biologists' knowledge of the space-use patterns of the foxes that wore the units.

The performance of the VHF transmitters associated with the Telemetry Solutions units deployed on Rosa was unacceptable. None of the transmitters came close to operating for the advertised life expectancy of 200 days. The longest any operated was 92 days, and 6 of the 14 deployed units failed after <10 days. The failure of VHF transmitters precluded tracking and locating animals for status checks, targeted trapping, or remote download of data, and also precluded the detection of dead foxes, as the mortality sensor was an altered pulse rate of the VHF signal. Fortunately, data were obtained from 5 units with malfunctioning VHF transmitters by activating the download function and searching over broad areas from the air.

Proximity Logger Units

The performance of the proximity logger units was excellent. The units functioned as expected. The small number of problems that did surface, almost all of which were relatively minor, were to be expected for a field study, particularly one in which relatively novel equipment and techniques are being tested.

Of greatest importance, the units were worn successfully by island foxes without causing any detectable injuries to foxes, and no adverse effects on survival or condition were detected. One fox died while wearing a unit, but there was no evidence that the unit contributed to this death. At 60 g, proximity loggers could be placed on foxes ≥1.5 kg.

The units were easy to program. Part of this programming ease was due to the limited parameters to program, and each parameter had just a few options. Probably the most important parameters are the “UHF coefficient,” which determines the distance at which a unit will detect another unit and record a contact, and the time with no contact recorded required to terminate a contact. The first parameter can be adjusted to alter the detection distance based on study objectives, whereas setting the second for too short a time is likely to result in an extended interaction between 2 foxes being recorded as a series of short contacts.

The units held up well under field conditions and did not exhibit any signs of damage or excessive wear. None of the units experienced premature battery failures. The VHF transmitters all functioned per expectations, and this greatly facilitated targeted trapping efforts to recapture foxes and recover the units. This ability was extremely important, as the proximity logger units do not have a remote-download function or automated drop-off system, thus necessitating recapture of the animals in order to recover the stored data. One fox was not recaptured, and therefore no data were obtained from this animal.

Abundant data were collected using the proximity logger units, and the quality of those data appeared to be quite high, based on the examination of concordance within unit dyads. In most cases, the number and total duration of contacts did not match exactly between 2 units in a dyad. However, this difference could easily be attributable to several factors. First, the sensitivity of each unit in detecting another unit likely was not identical across units due to inherent variations in the electronics of each unit. Second, the ability of a unit to detect another also is influenced by the orientation of each unit with respect to the other unit (e.g., height, position of fox, obstacles, etc.). Consequently, one unit may have detected a second unit whereas the second unit may not have detected the first, particularly when foxes were near the limits of unit detection abilities. This would lead to the observed discrepancies within unit dyads. In particular, an extended contact may be recorded as a series of shorter contacts and the number and length of these contacts may differ between the 2 collars (Prange et al. 2006). However, these discrepancies are relatively small, and trends and patterns were easily detected in the data. Scientists using proximity loggers on other species have found similar occasional, small discrepancies between the data recorded by 2 units in a dyad and have developed ways to deal with them during data analysis (Prange et al. 2006, Hamede et al. 2009, Hauver et al. 2010).