An inexpensive method for GPS tracking turtles

as an alternative to conventional tracking via telemetry or relatively expensive wildlife specific GPS trackers.

A Blanding`s turtle (Emydoidea blandingii) affixed with a GPS tracking rig.

A Blanding`s turtle (Emydoidea blandingii) affixed with a GPS tracking rig. All work was conducted in association with SUNY Potsdam (1) and the New York State Department of Conservation (2).

The problem

The Blanding`s turtle (Emydoidea blandingii) is state listed as threatened across the majority of its range. In New York this turtle is distinctly distributed across the St. Lawrence River Valley (SLRV) and in parts of the Hudson River Valley, though the strongest populations persist in the former. Since 1999 efforts have been made in the SLRV to document the distribution of these turtles here and introduce mitigation strategies focused on threats which face this species.

The most significant threat facing this species is road mortality, specifically on mature females. The majority of road crossings occur during nesting season from mid to late June as gravid females leave their home wetlands in search of nesting habitat. Additionally it is likely that if a gravid female successfully manages to cross roadways during their nesting foray and reach a potentially suitable site, that site would be an agricultural field which pose an ecological trap to nesting turtles. Identifying potential nesting habitat or patterns of movement to potential nesting habitat is the first step in implementing mitigation strategies to limit the threat roadways or agricultural fields have on this species.

Conventional tracking methods using telemetry provide only a limited view, and require extensive man hours to acquire the data frequency required to accurately record these turtles forays. GPS trackers offer the means of accurately and persistently collecting the data needed to influence the implementation of mitigation strategies, though wildlife specific GPS trackers afford a significant monetary cost which limits the scope of the project in terms of how many turtles could be tracked.

The use of inexpensive commercially available GPS loggers for spatial ecology studies is not unheard of and these devices have seen wide application across a range of species including aquatic turtles. The challenge with properly implementing these devices is modifying them in a manner suitable for safe and accurate data collection. In regards to Blanding`s turtles these devices would have to be modified to be waterproof, lightweight, have a low profile, withstand being carried through wetlands and dense vegetation for an extended period of time, and be easily removed and replaced. Here we will detail the methods we used to modify commercially purchased IGOTU GT-120 GPS loggers (figure 1) for successful implementation on gravid female Blanding`s turtles.

During the 2017 and 2018 season the implementation and exploration of the efficacy of these devices to potentially identify nesting sites of gravid females was conducted. The limited success of these devices during these seasons was in large part due to the nature of the devices, which were not designed with this application in mind. Waterproofing, sizing, and mounting issues were of most concern. Prior to the 2019 season, ideas were explored for deriving solutions to the issues faced in the past seasons using these devices. Ultimately a solution was devised which addressed the main concerns surrounding the use of these devices. During the 2019 season these methods were tested, revised and ultimately settled upon as a safe and effective means of deploying these devices on a large number of individuals for and extended period of time.

Figure 1: IGOTU GT-120 GPS logger, approximately $50 retail.

The solution

The IGOTU GT-120 is small, (44.5 x 28.5 x 13 mm, 20g) but relatively large compared to a mature Blanding`s turtle when considering the additional equipment needed to facilitate the implementation of these devices (radio and epoxy 28g). Waterproofing the device out of the box only increases the size and weight of the device to the point where safely mounting of the device on the turtle is impractical. Mounting of the device is another issue. Due to the limited battery life of these devices under the data interval periods we wanted to track these turtles, the devices would need to be replaced approximately every 7 to 10 days. A method of quickly removing and replacing the devices was needed. During the 2017 and 2018 seasons the issues we faced with the successful implementation of these devices revolved mainly around these two issues. Data was often lost to water damage and replacement of devices involved having to remove the turtle from the field, remove the old device, epoxy a new device on and ferry the turtle back into the field as soon as possible, typically 6 hours later. This likely resulted in undue stress to the animals and was impractical when trying to track many individuals.

First a mounting solution had to be devised to address this issue. Our team had access to 3D printers which allowed us to explore the creation of devices to facilitate our needs. Several ideas were explored including cases in which the device could be placed in and epoxied to the turtle allowing for the removal of the device. The issues surrounding this method were of course size constraints as any case would need to be larger than the device, which was already too large to begin with. So, a more simple solution was explored. A simple mounting plate, slightly smaller than the device itself, with attachment points designed to be used with zip ties, seemed to be the most practical solution (Figure 2).

Figure 2: 3D model of GPS logger mounting plate used to affix IGOTU GT-120 GPS loggers to Blanding`s turtles.

This mounting plate would be epoxied to the turtle and the GPS devices could then be securely fastened to the plate using zip ties, allowing for the easy removal and replacement of the devices in the field, within minutes.

With the mounting issue resolved the focus was brought to waterproofing and sizing, which seemed to be trade offs. Waterproofing the device inevitably required increasing the size of the device because new material had to be added to facilitate this. So, the stock device would have to be made as small as possible prior to any waterproofing. The simple solution to this was removing the device from the manufacturers casing. Doing so significantly reduces the size of the device and the relative weight (Figure 3 & 4).

Figure 3: Backing of IGOTU GT-120 case removed, exposing device battery and GPS.

Figure 4: IGOTU GT-120 device removed from manufacturers casing.

With the device removed from the manufacturers casing, making the device as small as possible, it is vulnerable to damaging impacts and exposure to moisture and dust that could render the device inoperable. The device must be secured in a state which allows for the stable use of the device. In our case we decided to simply wrap the device in a semi-permanent casing of electrical tape, being sure to leave the LED indicators and the button operable. This not only secured the battery to the device but provided some protection from impacts and exposure to light moisture and dust (Figure 5).

Figure 5: IGOTU GT-120 device removed from manufacturers casing and wrapped in an electrical tape casing.

Devices were labeled to keep track of each device throughout the season.

During the 2018 season we had reasonable success waterproofing the stock devices with latex balloons. Latex balloons offered great waterproofing, and when coated in epoxy they offered good protection from punctures of this waterproofing barrier but proved to be more cumbersome than needed. So, an alternative, finger cots seemed to be the perfect size for the job. Three finger cots, applied in alternating directions and secured closed with super glue was placed over the prepared device in figure 5.

Figure 6: The first finger cot was placed over the modified IGOTU GT-120 device so the opening was not over the USB end of the device. The USB end of the device was protected with a small piece of electrical tape to prevent any super glue from getting on the port.

Figure 7: Super glue is placed in the mouth of the finger cot.

Figure 8: The finger cot mouth is pinched shut for approximately 10 seconds.

Figure 9: The lip on the opening of the finger cot is removed.

Figure 10: The excess is folded over and glued down. The process is repeated 3 times.

The first two finger cots are used for redundancy in waterproofing while the third is used to protect the first two finger cots from the Flex tape which will be added following this step and allows for the easy removal of the device from the tape.

The finger cots are rather delicate and vulnerable. They would not survive being released into the field in this state. Flex tape is used to protect the finger cots from damage which may be incurred in the field resulting in the puncture of the finger cots and the inevitable water damage of the device.

Figure 11: The waterproofed device from figure 10 is wrapped in Flex tape for protection while in the field. The exposed finger cot here is covered with another piece of flex tape before being ready for application.

The waterproofed device, seen in figure 11, is the same dimensions of the manufactures casing, if not slightly smaller, and the weight is similar at 22g. The profile of the device is now slightly smoother, cutting off the hard edges on the manufacturers casing.

The mounting plate should be affixed to the turtle prior to preparing any devices. After printing of the plate the plate must be sanded and drilled for proper adherence of the epoxy and to reduce the profile and size of the device.

Figure 12: A mounting plate, sanded to size prior to installation on a turtle (bottom). The top plate is not sanded.

Figure 13: A mounting plate sanded and drilled, ready to be mounted on a turtle. Drilling holes in the device increased the surface area insuring a stronger hold. Be sure the plate is cleaned of any residue prior to installation.

Prior to applying any epoxy on the mounting plate or the turtle be sure to find the proper position on the turtle for application. Typically the best position to avoid hindering turtle movements is on either the left or right posterior edge of the shell where the marginal scutes meet the costal scutes inside the widest point of the shell. You want to place the mounting plate in a position which reduces the impact the affixed device would have on increasing both the lateral and vertical extent the device has on the shell profile.

Figure 14: A bed of epoxy is placed on the shell where the mounting plate will be installed. This bed is left to dry until tacky before placing the mounting plate.

Prior to placing any epoxy on the mounting plate, zip ties are placed through the attachment point to keep them clear of any epoxy. Epoxy is then placed on the concave side of the mounting plate and placed into position on the turtle. The device is held in this position for approximately 5 minutes. Additional epoxy is placed around the mounting plate to fill any gaps and to smooth the profile of the leading edges. Be careful to not get any epoxy in the attachment points.

Figure 15: Mounting plate securely attached to the turtle. Zip ties have been removed.

Be sure to allow the epoxy to dry for at least 6 hours prior to installation any GPS device or release into the field.

With the mounting plate installed and the GPS device prepared and waterproofed it can now be installed. The device is secured to the plate using 4, 4-inch zip ties. The zip ties are positioned in such a manner so the heads of the ties are flush with the base of the mounting plate. The device is placed on the mounting plate as low as possible to reduce the gap between the zip ties and the device on the lower edge of the plate.

Zip ties are first place up through the lower attachment points (black).

After connecting the lower ties with the upper ties (white) at the upper tie heads, slide the upper tie heads flush with the base. The connection between the upper ties and the heads of the lower ties is used to secure the device to the base plate.

Figure 16: Water proofed IGOTU GT-120 affixed via mounting plate to a Blanding`s turtle.

Prior to release a strip of flex tape is placed over the top of the device and down the leading edge to cover the exposed zip ties here and to close any gaps formed along these edges, smoothing the profile of the mounted device (figure 16).

The device installed, including zip ties epoxy, radio and epoxy, weighs about 60g. This setup allows for tracking of the majority of mature Blanding`s turtles which typically weigh more than 1200g, the threshold in this case when following the widely accepted 5% rule of thumb regarding equipment on wildlife.

The results

All devices implemented in the 2019 season performed as expected. No devices were lost to water damage, or as a result of falling off. The mounting technique used did not seem to limit the ability of these turtles to continue activity as usual. In fact, one turtle was found in an abandoned beaver lodge with the device still securely attached, which demonstrates the device did not hinder the turtle’s ability to navigate tight spaces.

During the 2018 season a collection interval of 15 minutes was used. The goal of this data collection frequency was in pursuit of battery longevity. The devices in this case would last greater than 7 days though the exact longevity was never witnessed as a check period of 7 days was identified as being ideal when trying to keep track of gravid female turtles for a few reasons. One was gravid female Blanding`s turtles can travel significant distances in search of nesting habitat. The 7 day period from release to visual allowed us to keep track of all individuals activities without losing contact with an individual. Additionally the nesting period is relatively short so a collection period of 7 days allowed us to narrow down the day in which a turtle could of nested if, when captured, the turtle was noted as being non-gravid. The closer to the time of nesting we can pickup an individual turtle the more likely it could be that we could use the data from the GPS loggers to identify and optimistically locate the exact point of nesting. Though this was only possible on a single occasion in 2018, but the nest had been predated by the time we were able to identify the location.

It is possible the decision to use a 15 minute collection interval also interfered with the accuracy of the data collected. These devices do not get a GPS fix when submerged under water. Therefore a turtle must be out of the water for an extended period of time in order to collect accurate data using lengthy intervals such as this. The data was consequentially limited when the turtle was in a wetland. By decreasing the collection period these issues could be rectified. Prior to the 2019 season, longevity testing was conducted to find the shortest collection period that would allow for a battery life of at least 7 days. It was found that a 5 minute collection interval could last up to 10 days on a four hour charge with LED indicators disabled. This proved to be ideal as it gave enough of a buffer period from our desired check period of 7 days, just in case a turtle could not be located on the 7th day. As a result of this change, data from this seasons testing was much more consistent and accurate than in previous years.

A total of 19 turtles, across 3 main sites were tracked using GPS loggers and radio telemetry during the 2019 season. Just 20 IGOTU GT-120 GPS loggers were required to maintain the schedule of GPS changes among the 19 tracked turtles, at the peak, 17 turtles were being tracked simultaneously. Material costs are as follows;

  • 20 IGOTU GT-120 GPS Loggers ($55/ unit) - $1100
  • 19 Holohil Systems RI-2B radio transmitters ($150/ unit) - $2850
  • JB Weld Kwick Weld Epoxy (approximately 4, 10 oz tubes) - $50
  • Flex Tape ($12/ 4" x 5' roll x 3) - $36
  • Gorilla Gel Super Glue ($5 x 3 units) - $15
  • Finger cots ($5/ 35 count box x 5) - $35
  • Electrical tape ($10/ unit) - $10
  • 4 inch zip ties ($10/ x 2, 100 count bags) - $20
  • 3D prints - $0
  • Total cost ~ $4100

Alternative wildlife specific GPS trackers would only allow for at most, tracking of 2, maybe 3 individuals with a similar budget.

Of the turtles tracked 18 were gravid females. The first turtle identified as gravid was on 6/8. The last check of a turtle yielding gravidity was on 7/17. Logger data accounted for 264 days of track data (1 track day is one day of data for one turtle), and approximately 124 instances of telemetry to maintain continuous tracking and checks for gravidity. The average number of track days per turtle was 15, min = 7, max = 24. Approximate nest locations of 12 of the 18 gravid females was estimated. The remaining 5 turtles showed little to no sign of typical or expected nesting activity. An attempt was made to locate four of the nests though this proved to be a challenge and no nests could be found. Estimated nest locations were established by reviewing and attempting to interpret the track data. Waypoints inside potential nesting habitat were reviewed more closely until a cluster of points, presumed to fall within the likely nesting period based on the context of the individual turtles track data leading up to this point, were extracted. The mean coordinate of these waypoints was calculated and the average distance from each point was used to delineate an area around the mean waypoint for which the nest likely fell within. This method was validated on one occasion when one of our tracked turtles was observed nesting. Using the waypoints from the time period of nesting and the method of nest location estimation detailed above, the actual nest location fell within the 3.3m radius of the estimated nesting area derived from the track data. Due to the typical nesting habitat, open and clear of overstory, it is reasonable to assume if the nesting period could be determined, the nest location can be determined with a relatively high degree of accuracy.

Site 1

A total of 7 gravid females were tracked using telemetry and IGOTU GPS loggers for a total of 96 track days. Average number of track days was 13, min = 7, max = 24. First instance of gravidity within this group of turtles was noted on 6/8, and last instance of non-gravidity was noted on 7/17. Of the 7 turtles, 5 were relocated to artificial nesting sites in an attempt to facilitate nesting at these sites. Interpretation of Logger data did not seem to show nesting activity taking place after relocation. In fact, all of the turtles that were relocated returned, in varying degrees, to the wetland in which they were originally found or continued beyond the point in the direction they were heading at the time of capture prior to release at these nesting sites. Of the relocated turtles two made significant journeys after release. The approximate nest location of 4 of the 7 turtles tracked can accurately be determined. These four turtles nested on 6/22, 6/24, 6/26, and 6/30. The approximate location of these potential nesting sites fell into two categories; road/ driveway edges, and agricultural fields. All potential nest locations are within a significant distance from one another. For the remaining three turtles at this site the nesting location could not be determined. For two of the remaining three, typical nesting activity featured by forays into expected nesting habitat i.e. agricultural fields, road edges, or clearings, was not recorded. This begs the question as to the fate of these turtles’ clutches. The third turtle mentioned here may provide a possible explanation to this phenomenon. Logger data from this turtle shows activity conducive of nesting. Logger data augmented with telemetry captures and checks for gravidity show this turtle did not nest in any of the location’s logger data suggests. Checks for gravidity show a period of 3 weeks in which a gradual loss of eggs was noted until a single egg was noted as being retained and lost on the 3rd week. Logger data shows the turtle remained within a wetland during this time period. This potentially suggests absorption as a possibility to the disappearance of this turtles clutch over a 3-week period. One turtle`s GPS logger suffered damage conducive of a mammal attack. Puncture marks across the loggers’ protective covering potentially suggest this.

Figure 17: This logger suffered damage from a likely attack on the turtle it was attached. The device was not removed from the turtle. The turtle showed no injuries.

This damage is similar to damage witnessed on another turtle that was tracked but not at one of our three main sites. In this instance the device was found removed from the turtle, located on a muskrat cattail mound. The radio was also damaged and located just 5m away, the turtle was not located.

Figure 18: As mentioned above, this GPS logger suffered similar damage to one of the turtles loggers at Site 1. In this instance the device was removed from the turtle. Notice the similar puncture marks on this device.

Figure 19: As mentioned above this device was found removed from a tracked turtle. Note the distinct puncture mark on one of the USB connections in this image. A significant amount of force would be require to produce this type of damage to the device. The device was still functioning.

Figure 20: The radio recovered from this turtle also showed damage, but the damage was distinctly different than the damage seen on the logger. In this image you will notice the wide "scrapes" on the device. These marks look similar to incisors of a muskrat.

Site 2

A total of 6 turtles were tracked using telemetry and IGOTU GPS trackers at this site for a total of 86 track days, average of 17, min = 7, max = 21. Logger data from one turtle was not acquired due to a malfunction resulting from human error, 5 turtles were successfully tracked. There were 3 non-gravid female turtles that were not tracked at this site. The first instance of gravidity was noted on 6/11. Interpretation of logger data shows clear nesting activity for 4 of the successfully tracked turtles. The proximate location of these turtle’s potential nest falls within two categories; pasture and agriculture fields. Nesting for these 4 turtles took place on 6/22, 6/27, and 7/2.  Of the turtles tracked here 3 were tracked in 2018, one successfully. This turtle likely nested within 100 m of the potential nesting location last year. One turtle traveled nearly 2 kilometers from the wetland this turtle was captured. One turtle showed activity towards potential nesting habitat, but no nesting activity was recorded. Turtle showed interesting behavior during two separate nesting forays. This turtle was located on one instance in an abandoned beaver lodge, gravidity was noted on this instance, a behavior that was unexpected during nesting season.

Site 3

Just 3 gravid females were tracked using telemetry and IGOTU GPS trackers at this site, though 6 female turtles were tagged at this site, 3 turtles remained non-gravid throughout the nesting season. These turtles were GPS tracked for a total of 45 track days, average of 15, min = 9, max = 21. The first instance of gravidity was noted on 6/10. Interpretation of logger data shows clear nesting activity for all three of the turtles tracked here. The proximate location of these turtle’s potential nest location falls within three categories; agriculture field, road edge, and a private residence lawn. Nesting took place on 6/22 (2) or 6/23 (1). All turtles nested within a significant distance from one another. All turtles returned to the wetland in which they were originally captured post nesting.  

Our conclusions

While no accuracy assessment was conducted on the data we collected in this study, other studies have specifically looked at the accuracy of these devices in varying cover types. Moris and Conner, 2017, concluded that an accuracy of approximately 10m can be expected from these devices in areas with varying cover types. This is a reasonable and acceptable trade off considering the price point at which these devices can be applied considering the alternative.

Revision of all track data acquired during the course of this study yielded additional applications for these devices on fresh water turtles. Such applications might include temporal activity patterns. Reviewing the data showed patterns among individuals and across turtles which entice further speculation. The addition of temperature sensors with this rig could influence more detailed interpretation of activity patterns which might be influenced by temperature. Additionally more can be done with these devices regarding evaluating potential hot spots for road crossings, or analyzing the effectiveness of road crossing barriers. Several peculiar phenomenons were also recorded throughout the course of our study, including the possible absorption of a clutch by one individual. Logger data did not show expected nesting activity. Augmented with gravidity checks a gradual reduction in clutch was noted over a 2 week period. Similarly several individuals were also recorded as never leaving a wetland which begs the question is egg absorption a more common phenomenon than previously expected or is stress associated with actions conducted in this study a contributing factor to unsuccessful nesting. Some individuals were also relocated to artificial nesting sites, but no nesting activity was recorded at these sites. The application of these devices in this context could therefore be used to monitor the effectiveness of the relocation of gravid individuals to artificially created nesting sites. On the other hand they could also be used to better target areas most suited for the creation of artificial nesting sites by intercepting individuals along common nesting foray paths, if commonalities between years among individuals at a site exist. It is worth noting that all relocated individuals navigated with great accuracy back to their home wetlands or surprisingly back to the point of capture prior to relocation. This suggests an ability to navigate by some means. More work should also be done to identify patterns in track data associated with nesting activity to enhance interpretation techniques which in turn could yield more accurate estimations of probable nest locations. More accurate estimation could therefore provide field technicians with a greater ability to located the nests of tracked individuals where nest guards could then be implemented. In addition to this a search technique must also be developed to better located nests within a given area.

The efficacy of the methods and devices used in this study demonstrates the practical application of these devices as a valuable conservation tool that makes large scale and accurate tracking of target individuals possible. Conventional telemetry would have been impossible to implement on this scale. The data collected in this study has a wide range of applications focused around identifying a broad range of activity patterns which may exist within this species, which could potentially influence targeted conservation strategies to mitigate the threats affecting this species.

If you are interested in additional information regarding our methodology or if you are interested in obtaining the 3D model of the base plate, please contact Dr. Glenn Johnson, johnsong@potsdam.edu or Jase Briggs, briggs_jase@outlook.com.

A Blanding`s turtle (Emydoidea blandingii) affixed with a GPS tracking rig. All work was conducted in association with SUNY Potsdam (1) and the New York State Department of Conservation (2).

Figure 1: IGOTU GT-120 GPS logger, approximately $50 retail.

Figure 2: 3D model of GPS logger mounting plate used to affix IGOTU GT-120 GPS loggers to Blanding`s turtles.

Figure 3: Backing of IGOTU GT-120 case removed, exposing device battery and GPS.

Figure 4: IGOTU GT-120 device removed from manufacturers casing.

Figure 5: IGOTU GT-120 device removed from manufacturers casing and wrapped in an electrical tape casing.

Figure 6: The first finger cot was placed over the modified IGOTU GT-120 device so the opening was not over the USB end of the device. The USB end of the device was protected with a small piece of electrical tape to prevent any super glue from getting on the port.

Figure 7: Super glue is placed in the mouth of the finger cot.

Figure 8: The finger cot mouth is pinched shut for approximately 10 seconds.

Figure 9: The lip on the opening of the finger cot is removed.

Figure 10: The excess is folded over and glued down. The process is repeated 3 times.

Figure 11: The waterproofed device from figure 10 is wrapped in Flex tape for protection while in the field. The exposed finger cot here is covered with another piece of flex tape before being ready for application.

Figure 12: A mounting plate, sanded to size prior to installation on a turtle (bottom). The top plate is not sanded.

Figure 13: A mounting plate sanded and drilled, ready to be mounted on a turtle. Drilling holes in the device increased the surface area insuring a stronger hold. Be sure the plate is cleaned of any residue prior to installation.

Figure 14: A bed of epoxy is placed on the shell where the mounting plate will be installed. This bed is left to dry until tacky before placing the mounting plate.

Figure 15: Mounting plate securely attached to the turtle. Zip ties have been removed.

Zip ties are first place up through the lower attachment points (black).

After connecting the lower ties with the upper ties (white) at the upper tie heads, slide the upper tie heads flush with the base. The connection between the upper ties and the heads of the lower ties is used to secure the device to the base plate.

Figure 16: Water proofed IGOTU GT-120 affixed via mounting plate to a Blanding`s turtle.

Figure 17: This logger suffered damage from a likely attack on the turtle it was attached. The device was not removed from the turtle. The turtle showed no injuries.

Figure 18: As mentioned above, this GPS logger suffered similar damage to one of the turtles loggers at Site 1. In this instance the device was removed from the turtle. Notice the similar puncture marks on this device.

Figure 19: As mentioned above this device was found removed from a tracked turtle. Note the distinct puncture mark on one of the USB connections in this image. A significant amount of force would be require to produce this type of damage to the device. The device was still functioning.

Figure 20: The radio recovered from this turtle also showed damage, but the damage was distinctly different than the damage seen on the logger. In this image you will notice the wide "scrapes" on the device. These marks look similar to incisors of a muskrat.