How often should dead-reckoned animal movement paths be corrected for drift?

Abstract

Background: Understanding what animals do in time and space is important for a range of ecological questions, however accurate estimates of how animals use space is challenging. Within the use of animal-attached tags, radio telemetry (including the Global Positioning System, 'GPS') is typically used to verify an animal's location periodically. Straight lines are typically drawn between these 'Verified Positions' ('VPs') so the interpolation of space-use is limited by the temporal and spatial resolution of the system's measurement. As such, parameters such as route-taken and distance travelled can be poorly represented when using VP systems alone. Dead-reckoning has been suggested as a technique to improve the accuracy and resolution of reconstructed movement paths, whilst maximising battery life of VP systems. This typically involves deriving travel vectors from motion sensor systems and periodically correcting path dimensions for drift with simultaneously deployed VP systems. How often paths should be corrected for drift, however, has remained unclear.

Methods and results: Here, we review the utility of dead-reckoning across four contrasting model species using different forms of locomotion (the African lion Panthera leo, the red-tailed tropicbird Phaethon rubricauda, the Magellanic penguin Spheniscus magellanicus, and the imperial cormorant Leucocarbo atriceps). Simulations were performed to examine the extent of dead-reckoning error, relative to VPs, as a function of Verified Position correction (VP correction) rate and the effect of this on estimates of distance moved. Dead-reckoning error was greatest for animals travelling within air and water. We demonstrate how sources of measurement error can arise within VP-corrected dead-reckoned tracks and propose advancements to this procedure to maximise dead-reckoning accuracy.

Conclusions: We review the utility of VP-corrected dead-reckoning according to movement type and consider a range of ecological questions that would benefit from dead-reckoning, primarily concerning animal-barrier interactions and foraging strategies.

Keywords: Animal movement; Animal tracking; Biologging; Dead-reckoning; Drift; GPS correction; Global Positioning System (GPS); Tilt-compensated compass.

Fig. 1

Boxplots summarising the magnitude of net error according to the VP correction rate per species. Mean values were aggregated per individual and VP correction rate. Boxes encompass the 25–75% interquartile range and horizontal bars denote the median value with ‘loess’ smooth line (grey shading shows the standard error and Whiskers extend to 1.5 * Interquartile range). Note net error drops to zero when the VP correction rate equates with GPS recording frequency (1 Hz for the lions, penguins and cormorants, and 1 fix/min for the tropicbirds). The inserts zoom in on the net error between VP correction rates of one fix per hour and one fix per second

Fig. 2

Boxplots demonstrating the total distance moved (km) during the tag deployment period according to VP correction rate for the study species. Solid lines show the distance moved calculated using successive dead-reckoned positions (distance moved (see methods) was 3-D and computed for penguins and cormorants operating at varying depths and tropicbirds at varying altitudes, and 2-D computed for lions and penguins walking on land). Dashed lines reflect the distance moved calculated from successive GPS positions according to the level of VP under-sampling stated (only 2-D distances were computed). Mean values were aggregated per individual and per VP correction rate. Boxes encompass the 25–75% interquartile range and horizontal bars denote the median value with ‘loess’ smooth line (grey shading shows the standard error and Whiskers extend to 1.5 * interquartile range). Note that the high spread of each species boxplot is due to the high intra-specific variability of distances moved—e.g., with the tropicbirds, differences in foraging/distance roamed may be due to breeding vs. non-breeding status

Fig. 3

A lion’s dead-reckoned movement path (approx. 12 days) in relation to (all) GPS positions (black), plotted both as a function of GPS correction rate (a no correction, b GPS corrected every 12 h, c GPS corrected once every hour) and initial subset of data used to create the path (red = all data (no VeDBA threshold for speed), blue = only data that surpassed VeDBA threshold (> 0.1 g) used for speed, green = only data during periods depicted as proper movement (using the MVF protocol [23])). Note the difference in y-scales across the net error graphs

Fig. 4

VP-corrected dead-reckoned movements of lions in the Kgalagadi Transfrontier Park. The top left plots show a pride of 5 lions (2 males—blue and 3 females—red). Both male and female movements abutted the Botswana fence boundary (dashed green line), although only the females crossed (illustrated in the dotted inserts, with yellow, cyan and purple tracks denoting individual females) The bottom right insert shows one female pacing along the fence line in an attempt to re-join the other two that crossed hours earlier. Note the extent of (unfiltered) GPS error that occurs (particularly during resting behaviours) (top right)

Fig. 5

A 45-min section of a tropicbird’s foraging flight at sea, encompassing periods of thermal soaring. The top plot characterises stylised trends in the raw values and select derivatives from the motion sensor and GPS unit outputs (2D waveforms vs time), including differentiating flapping flight from thermal soaring (marked events—primarily based on magnetism data). The sine waves appearing in two of the magnetometer channels simultaneously reflect circling. The bottom plot graphs the dead-reckoned track (coloured according to VeDBA) in 3-D, relative to all available GPS fixes obtained (black) (including an insert of circling behaviour). Note periods of thermal soaring are not apparent with GPS at the recording frequency of 1 fix/1 min as used here. Note that climb rate increases as a function of the inverse of the rate of change of pressure

Fig. 6

A 15-min duration of a Magellanic penguin’s foraging trip at sea. The top plot characterises stylised trends in the raw values and select derivatives from motion sensor and GPS unit output (2D waveforms vs time), including differentiating between dives and surface periods (marked events—primarily based on depth data). Note that pressure is inverted to reflect depth. The bottom left plot maps the entire (17 h) VP-corrected dead-reckoned foraging trip. The bottom right plot graphs the resultant VP-corrected dead-reckoned track (coloured according to VeDBA) in 3-D, relative to all available GPS fixes obtained (black)). Note the latency delays in GPS recordings (as seen in the top plot), with a temporal offset of fixes (red dot projections) occurring at green-marked events (at depth)). Fixes that occurred at depth were removed from the analysis

Fig. 7

Seven minutes of tropicbird flight with dead-reckoned tracks advanced according to 3 different allocations of speed, plotted alongside GPS (1 fix/min) both pre- and post-VP correction. This demonstrates the main error that can arise during the VP correction procedure (using heading and distance correction factors (see “Discussion” section)), when there is a large disparity in distance between consecutive VPs and consecutive dead-reckoned positions, primarily due to inaccurate speed allocation and/or VP error. Note how a segment of thermalling behaviour was disproportionately expanded during the VP correction process when using GPS-derived speeds and DBA-based estimates, because there was no differentiation between thermal soaring and flapping flight (cf. Fig. 5). Using a much lower speed value during thermal soaring value (a quarter of the magnitude allocated for flapping flight) greatly improved track estimates because the magnitude of linear drift correction works as a function of the underlying speed allocation

Fig. 8

a One penguin’s dead-reckoned track calculated with- (green) and without- (blue) current integration and 3 variant VP correction rates (left panel). b Differences in net error when dead-reckoned tracks were iteratively integrated with space- and time-correction (net error estimates obtained from 5 penguin datasets). The boxes denote the median and 25–75% interquartile range and whiskers extend to 1.5*IQR. c An uncorrected dead-reckoned tropicbird flight path, relative to GPS, both with (green) and without (blue) current integration. Note the clustering of fixes (e.g., due to animal not moving much for extended periods of time) that can occur when using temporal sub-sampling routines (a more refined method could include using a VP correction rate of 1 fix every ‘x’ m moved (e.g., as estimated between VPs)).

Fig. 9

Schematic diagram to illustrate the various elements that modulate VP and VP-corrected dead-reckoning accuracy. Black dots illustrate the element’s graphical position