Chris Thaxter 11/03/2026
MoveRakeR contains a collection of functions to help streamline data
processing and common data wrangling tasks that often occurs with GPS
tracking data. It operates at the initial stages after data acquisition.
Although a number of R packages are now available for analysis of animal
movement data (Joo et al. 2020), MoveRakeR tackles some fundamental
data processing steps that are often encountered to help streamline data
preparation ahead of further analytical steps. It is intended to support
the other existing workflows in other R packages, such as move
(Kranstauber et al. 2018) and move2 (Kranstauber et al. 2024) and
provides some further tools to address common manipulations of data
routinely encountered that often require a degree of subjectivity in
decision making. Should the user be interested in exploring location
error in GPS devices in more detail, we would recommend the approach of
Flemming et al. (2020) and the ctmm R package (Fleming & Calabrese
2023).
MoveRakeR needs the following packages installed:
adehabitatLT, dplyr, tibble, lubridate, tidyr, purrr, data.table, move, RODBC, suncalc, geosphere, DT, leaflet, shiny, shinyWidgets, suncalc, htmltools, sf, sfheaders, sp, ggplot2, R.rsp, readr, flexdashboard, plotly, shinydashboard, shinyWidgets, shinybusy, DT, rmapshaper, units, magrittr and viridisLite.
MoveRakeR is not yet available on CRAN therefore please install it by
using:
devtools::install_github('BritishTrustForOrnithology/MoveRakeR', build_vignettes = TRUE)There are currently two vignettes:
-
clean_track This that gives a more detailed overview of the initial error-checking and data annotation/filtering processes involved.
-
ShinyRakeR This details the
ShinyRakeRapp in more detail for the interactive means of carrying out annotation of your data through a number of processes available withinMoveRakeRas detailed in the clean_track vignette.
These can be accessed through:
vignette(package="MoveRakeR")
vignette("clean_track", package = "MoveRakeR")
vignette("ShinyRakeR", package = "MoveRakeR")A further vignette will be added for defining trips and computing trip metrics.
An example dataset will be added to the package for a small number of example Lesser Black-backed Gulls Larus fuscus tracked from a breeding colony in the UK.
MoveRakeR was borne out of seabird tracking work. Much of the focus of
the package is on movements of tracked animals from a central place,
used within numerous papers to date (e.g. Thaxter et al. 2015, Langley
et al. 2021, Clewley et al. 2023, Thaxter et al. 2025, O’Hanlon et
al. 2025). The MoveRakeR pipeline requires a dataset with the
following columns: TagID, DateTime, latitude and longitude. The
processing is built around a Track class object. This is not a strict
requirement but aids in visualisation of printing, plotting and
summarising.
data <- read.csv(data.csv,sep = ",", header=TRUE)
data <- Track(data)
summary(data)
plot(data) # slower for large datasets
tabulate_history(data)The workflow within MoveRakeR and the further accompanying
visualisation package RakeRvis is depicted below.
The MoveRakeR workflow uses a number of functions. As there are a many
of these, the below schematic outlines how these fit into the process.
Data sourcing functions allow data to be read into R from databases,
from MoveBank through the read_track_MB() function, and the University
of Amsterdam Bird-tracking System database (UvA-BiTS) (Bouten et
al. 2013), using the read_track_UvA() function. For the UvA-BiTS
connection, you will need to first follow the steps outlined under
?read_track_UvA to set up an ODBC connection using R package RODBC.
The MoveBank login details are also required for use and create this
calling:
library(RODBC)
db <- odbcConnect("GPS") # if you called your connection "GPS"
library(move)
login <- MoveBankLogin("username","password")Then data can be read in using read_track_UvA() and read_track_MB()
that allows concatenation of animal IDs and start/end times of sections
of data required:
# UvA-BiTS
TagID <- c('1', '2', '3') # example birds
start <- c("2016-06-01 13:53:50", "2016-06-15 12:22:13", "2016-06-05 08:07:23") # example start times
end <- c("2016-07-15 09:17:14", "2016-07-20 01:08:58", "2016-07-18 14:22:45") # example end times
dataUvA <- read_track_UvA(TagID, start=start, end=end, pressure = FALSE) # reading in data
# MoveBank
TagID <- c('4', '5', '6')
start <- c("2016-06-01 13:53:50", "2016-06-15 12:22:13", "2016-06-05 08:07:23")
end <- c("2016-07-15 09:17:14", "2016-07-20 01:08:58", "2016-07-18 14:22:45")
repo = "your_MoveBank_repo"
dataMB <- read_track_MB(TagID, start=start, end=end, repo = repo)The MoveBank read in depends on the move package, which of course can
be done directly using that package; the read_track_MB() function
seeks to align different datasets to a common form for further
combination. The raw MoveBank form of the data can be preserved if need
be. Note also, that the read in for MoveBank data automatically filters
out duplicate data - see also move2 (Kranstauber et al. 2024) for more
details of such filtering.
It is necessary early on in the workflow to check for duplicate data.
This is to avoid pitfalls of carrying out initial sorting by timestamps
within animals that could mask issues, such as sudden data jumps in
time. The duplicate_track() function helps identify duplicate time
streams of valid data and other duplication errors such as identical
timestamps for the same xy position. However, care is needed to make
sure the behaviour of duplication handling is as expected, and that
location duplicates are not carrying other needed data alongside, such
acceleration information. A means to resolve these duplicates are also
provided, such as random selection of a duplicate or based on highest
quality data from other columns in your data.
MoveRakeR provides some built-in mapping options shaped around two
leaflet Shiny apps. Firstly, plot_leaflet() can allow
visualisation of tracked animals, via:
plot_leaflet(data)
A second function is called plot_leaflet_trips() which visualises
individual trips of animals in more detail with further tabs for data
summaries.
The user may also find useful a further R package for use alongside
MoveRakeR called RakeRvis. This is a further Shiny app with
detailed plotting functionality for neater visualisation using
leaflet, leafgl and mapdeck, with further option to either load
data or download data from MoveBank or UvA-BiTS.
library(RakeRvis)
RakeRvis(data)
In plot_leaflet(), as also in RakeRvis, there are further options
for plotting shapes, additional point shape layers (using the sf
package), and options to plotby a different variable, e.g. a year,
cohort, behaviour etc that you may have in your data; further coloration
and point/line options are available.
A generic S3 plot function for is also available for Track objects,
which uses base R graphics and can be overlain on top static maps and
shapefiles, with arrows indicating direction also included. For example:
ukmap <- sf::st_transform(ukmap, 4326) #n example base map of hour choice in geographic WGS84
# using the generic plot function for the S3 method for class 'Track' (which LBBGWalB201416 is)
Cpal <- grDevices::colorRampPalette(c("red", "green", "blue"))
par(mar=c(0,0,0,0))
plot(ukmap$geometry, xlim=c(-4.7,-2), ylim = c(53.2,54.5), col = "wheat", border= "wheat")
plot(data, anims = "uniq",
Lines = TRUE,
gap = FALSE,
ADD=TRUE,
Legend = TRUE,
cex_p = 0.2,
col = Cpal,
p4s = 4326)Should the user be interested in voltages, then these can also be
plotted through an S3 generic plot function, that plots objects of class
Vo. This is directly compatible with voltage data that can be accessed
from the UvA-BiTS database, through the read_voltage_UvA() function.
For use with wider Track data, if you have voltage and charging data
in the same dataset, this can be coerced to a Vo object using the
Track2Vo function. This visualisation may be useful to see how tags
are performing, after deployment to check sustainability of sampling
protocols, and can be plotted alongside GPS data. Further options also
exist to source acceleration and pressure data from UvA-BiTS directly
into R.
# This needs conversion to a Vo object first, where data is an object with TagID and DateTime columns:
Vo_obj = Track2Vo(data, vbat = "tag-voltage", vsll = "battery-charging-current")
Vo_obj_subset = Vo_obj[Vo_obj$TagID == "animal_1",] # subsetting one animal if wanted from Track data
track_data = data[data$TagID == "animal_1",]
# plot with Track data alongside for additional day/night bars, with a subset of time period
plot_st = plot(Vo_obj_subset, dataTrack = track_data, start = "2022-04-05 00:00:00", end = "2022-04-10 00:00:00")An example plot below is shown from several animals of a separate dataset:
Thereafter, the task may involve understanding other potential sampling
biases in your data, such as routine summaries of data amounts per
animal through a summary() function, spatial/temporal spans of the
data that you may have or expect a priori through the data_spans()
function, and assessing density of fixes of animals through time and
potential outlying individuals using tag_timeline(). You may be
interested to understand likely tag sampling rates, that can be assigned
through assign_rates(). Moreover, the spacing between sampling rates
may vary considerably if the tag has stopped recording, for example due
to low battery, necessitating identification of ‘gaps’ in your data,
available through the gap_section() function.
Movement validity can be assessed through checks of trajectory speed
between two fixes, through the speed_filt() function. MoveRakeR
offers flexibility and understanding around the sorts of filtering you
may want to do, such as based on raw trajectory speed, e.g. assessing
influencing of tag sampling rate using tspeed_filt(), and other
averaging/smooths such as root mean square within speed_filt() -
similar processes available in other R packages such as trip (see also
Sumner 2011).
Alternatively, the user may consider fixes directly unsuitable due to
poor quality measurement, for example through too few satellites or DOP
values - this is easily done outside of MoveRakeR but a function
filt_err() is also available to help with this. Further measurement
errors may also be apparent, linked to the devices themselves. It should
be said MoveRakeR can help with some of these issues but it may be
worth investigating outliers in the data using other R packages. For
example, the ctmm package (Fleming & Calabrese 2023) through
ctmm::outlie(), uses available information on location error such as
DOP values or number of satellites, as well as trajectory speed
estimates (Flemming et al. 2020) in assessing potential error
generation. A further function within MoveRakeR called rake_outlie()
can also be used to look at potential measurement errors from variables
in the data, exploring potential outlying data using boxplot, quantile
or Hampel approaches to outlier detection.
The above steps span a process of initial error/outlier assessment and
potential effects of removal of GPS data. This we term as “raking”, a
subset of what we may call data “cleaning”. A further general function
called rake() is also included to scan the data for potential issues,
using most of the above processes, although deeper dives into your data
are always advised. In addition, a further shiny app has also been
included as a more interactive tool to examine and test some of these
biases visually, called ShinyRakeR. This can be called as:
ShinyRakeR(data)This app displays two main tabs, a threshold explorer and a temporal inspector, which can be synchronised. The threshold explorer is a means of annotating the data for potentially outlying fixes, the combined effects of which can be explored geographically using the leaflet map as well as individually, with visual and tabular breakdowns also evaluated for proportion of fixes affected by the choices at the animal level. Options for annotating include minimum numbers of GPS fixes per animal, duplicate data, minimum numbers of satellites, dilution of precision values, trajectory speeds, turning angles and spatial xy extents. Gaps in the data can be investigated where isolated fixes may be left.
The temporal inspector tab allows for visualisation of variables associated with the data through time as line plots, alongside a map. The histogram distribution of the variable chosen can be explored and a plot for the density of tracking data across animals through time is also displayed.
Among other packages, the ExMove toolkit (Langley et al. 2024)
provides a Shiny App to investigate the effect of filter effects of
speed and net displacement based on prior knowledge of the study system.
Note, as of July 2025, there was an update to shiny and subsequent
shinydashboard that affects the way menu items are coded. For
functions such as ShinyRakeR there is a chance, depending on the
versions of those packages you have, that apps may stall and complain
about deprecation of menuItems. This has yet to be explored, but for
now I suggest using a stable combination of those packages to run the
shiny apps:
remotes::install_version("shiny", version = "1.8.1")
remotes::install_version("shinydashboard", version = "0.7.2")
# then
ShinyRakeR(data) # should work as intended - get in touch if not!
Annotation or filtering of the data following data inspection can be
made using the clean_track() function. This is a wider process similar
to the threshold explorer tab of the ShinyRakeR() function, to flag a
combined errant data label per row. clean_track() can also filter
automatically for these decisions. For example:
clean_track(data = data, drop_sats = 3, sp_thres = 40, GAP = 28800, drop_single_gap = FALSE)In previous MoveRakeR releases prior to version 1.1.4,
clean_track() was called clean_GPS() although the legacy code as the
former function name is still included for backwards compatibility. This
cleaning function process is a rather blunt instrument and we strongly
advise that clean_track() is not used out-of-the box using the
defaults but that a thorough investigation of your data is carried out
using the above “raking” steps. The clean_track() function should be
seen as a final all-encompassing shortcut to annotate or filter your
data based on decisions that you are arrived at.
A common task is to sub-sample GPS data to a common rate; this may be
needed for further analytical steps or to align data in some way. As
with all these tasks, modelling solutions may exist to not require this
step. However,MoveRakeR provides a sub-sampling algorithm sub_samp()
that builds upon existing tools available in some other packages such as
AMT (Signer et al. 2019) and move (Kranstauber et al. 2018), in
particular including a bootstrap method to sub-sample data and maximise
data availability, useful for animals traversing space for only a
limited time such as on migration. Other approaches using interpolation
and simulation are not provided in MoveRakeR for which other R
packages are available to provide those operations, such as
adehabitatLT, move and continuous-time movement modelling approaches
such as crawl (Johnson et al. 2008) ctmm (Fleming & Calabrese 2023)
and aniMotum (Jonsen et al. 2023).
Further annotation of the data is often required for central place
foraging consideration. Metrics for an animal or population may be
needed to assess how far typically animals move from their central
place, how long trips are, and other metrics such as tortuosity,
bearing, total distance and probably many others. The trips are first
identified provided using function define_trips() using user-specified
definitions of the central place such as a radius around a centre point,
a rectangle or polygon to define the central place, or alternatively use
of specific nest locations for individual animals; on the latter, nest
locations may be known directly or perhaps where not, estimates from
other approaches such as nestR (Picardi et al. 2020) or recurse
(Bracis et al. 2019) may be informative; see also the trip package
(Sumner et al. 2009, Sumner 2011). These locations are needed by the
function to estimate where to measure the trip start/end point in space
and time. The function trip_stats() can then be used to generate
movement metric summaries of trips for each animal and the group of
animals. Further tabulation functions are provided such as
tabulate_trips() which takes the output from trip_stats() and
summarise the information neatly. The trip_stats() function will
return a new object of class Trip. The default of the trip_stats()
function is to use the first and last fix of the animal within the
central place boundary to compute trip duration and distances.
newdata <- data %>%
clean_GPS(GAP = 3600*8, speedfilt = TRUE, drop = FALSE) %>%
sub_samp(dt=300, tol = 0.2, method = "sequencer") %>% # sub-sample data
define_trips(method = "rect", lls = LLS, plot = TRUE, verbose = TRUE) # (e.g. LLS = c(-3.2, 54.0553,-3.1689, 54.0437))
plot_leaflet(newdata)
trips <- trip_stats(newdata) # a new dataset of individual trips using first and last known positions in CP
tabulate_trips(trips) # neater tabulation
# visualising individual trips
plot_leaflet_trips(newdata, trips = trips)Note that complexities can arise for trips that are determined as
arriving and then departing again from the central place with a single
shared GPS fix for starts and ends of trips. These are handled in
MoveRakeR through duplicate row records, stored as an attribute from
the define_trips() function. However should substantial data
manipulation, such as base R processing, be carried out between these
two functions being run, it is best to add these duplicate data directly
into the data before such manipulation and ahead of trip_stats() being
run; beware that this data will likely need removing again ahead of
downstream analyses.
LLS <- c(-3.2, 54.0553)
newdata <- define_trips(data = data, method="circ", lls = LLS) %>%
add_cag_trips() # this will add duplicate rows directly to the Track using column "extra_row" 1 or 0.
### other potential data manipulation e.g.
newdata <- data.frame(newdata) # removes any attributes and Track data class
# trip_stats() can still be run given pseudo duplicate data directly included demarcate start/ends for all trips
trips <- trip_stats(newdata)
# but then best put the extra rows back as an attribute, removing them from the main Track data
newdata <- Track(newdata) %>% reset_cag_trips()The MoveRakeR seeks to employ ‘tidy’ principles at heart, although
further data.table and base R operations are also well-utilised. The
use of tibble and dplyr ensured that operations were carried out at
very least at the grouping of the animal level. However, in case further
listed formats of data are wanted, the user can convert the Track data
to a list of animals TrackStack or a further multi-list
TrackMultiStack e.g. for separate stacks of years. These are just
convenience functions to convert to base R operations.
The MoveRakeR package is mainly focused on xy position, but further
functionality may be included for vertical assessment. For example
assessing fix-to-fix changes in GPS altitude and whether any outlying
points are worthy of considering as erroneous. Such consideration may be
useful for considering the flight altitude of birds, but this is another
topic that deserves separate consideration, and often includes modelling
of such sources of error (Ross-Smith et al. 2016, Johnston et al. 2023).
These are all routine GIS operations, that MoveRakeR doesn’t seek to
replicate. However, should you wish to add a distinction for offshore
and onshore (likely very specific circumstances) then a function
offshore() is available to annotate as a simple binary vector. Should
you be interested in distance to coast as a metric for foraging trips,
then dist2coast() is also provided, then can be fed through the
trip_stats() routine. These functions depend on you definitions of the
coastline for which an sf shape is provided to each function, with a
buffer extent available to crop the overall shape for increased
efficiency.
newdata <- data %>%
offshore(shape = shape_sf, crop = TRUE, buffer_extent = 10000, p4s = 3035) %>%
dist2coast(shape = shape_sf, crop = TRUE, buffer_extent = 50000, p4s = 3035)The MoveRakeR R package source code is released under the MIT License.
This permits free use, modification, and redistribution of the code,
including for commercial purposes.
Any datasets included within this package are licensed separately under the Creative Commons Attribution–NonCommercial 4.0 International (CC BY-NC 4.0) license.
This means:
- You are free to use, share, and adapt the data, but
- Commercial use of the data is not permitted, and
- Attribution is required when the data are used or redistributed.
The non-commercial data license applies only to the included datasets and does not restrict use of the package code itself.
For full license texts, see: LICENSE (MIT license for code), and inst/extdata/data-license.txt (CC BY-NC 4.0 license for data).
Bouten, W., Baaij, E.W., Shamoun-Baranes, J. & Camphuysen, K.C.J. (2013) A flexible GPS tracking system for studying bird behaviour at multiple scales. Journal of Ornithology, 54, 571–580. https://doi.org/10.1007/s10336-012-0908-1
Bracis, C., Bildstein, K.L. & Mueller, T. (2018) Revisitation analysis uncovers spatio-temporal patterns in animal movement data. Ecography. https://doi.10.1111/ecog.03618
Calenge, C. (2006) The package adehabitat for the R software: tool for the analysis of space and habitat use by animals. Ecological Modelling, 197, 516-519. https://doi.org/10.1016/j.ecolmodel.2006.03.017
Clewley, G.D., Thaxter, C., Scragg, E.S., Masden, E.A., Barber, L.J., Conway, G., Clark, N.A. & Burton, N.H.K. (2023) Daily, seasonal and annual variation in area use of Lesser Black-backed Gulls (Larus fuscus) related to offshore renewable developments. Bird Study, 70. https://doi:10.1080/00063657.2023.2190080
Cullen, J. (2020) R package ‘bayesmove’ v 0.1.0. https://github.com/joshcullen/bayesmove
Fleming, C.H., Drescher-Lehman, J., Noonan, M.J., Akre, T.S.B., Brown, D.J., Cochrane, M.M., Dejid, N., DeNicola, V., DePerno, C.S., Dunlop, J.N., Gould, N.P., Harrison, A.-L., Hollins, J., Ishii, H., Kaneko, Y., Kays, R., Killen, S.S., Koeck, B., Lambertucci, S.A., LaPoint, S.D., Medici, E.P., Meyburg, B.-U., Miller, T.A., Moen, R.A., Mueller, T., Pfeiffer, T., Pike, K.N., Roulin, A., Safi, K., Séchaud, R., Scharf, A.K., Shephard, J.M., Stabach, J.A., Stein, K., Tonra, C.M., Yamazaki, K., Fagan, W.F. & Calabrese, J.M. (2020) A comprehensive framework for handling location error in animal tracking data. bioRxiv 2020.06.12.130195. https://doi.org/10.1101/2020.06.12.130195
Fleming, C.H. & Calabrese, J.M. (2023). ctmm: Continuous-Time Movement Modeling. R package version 1.2.0. https://CRAN.R-project.org/package=ctmm
Johnson, D.S., London, J.M., Lea, M.-A. & Durban, J.W. (2008) Continuous-time correlated random walk model for animal telemetry data. Ecology, 89, 1208-1215. https://doi:10.1890/07-1032.1
Jonsen, I.D., Grecian, W.J., Phillips, L., Carroll, G., McMahon, C., Harcourt, R.G., Hindell, M.A. & Patterson, T.A. (2023) aniMotum, an R package for animal movement data: Rapid quality control, behavioural estimation and simulation. Methods in Ecology & Evolution, 14(3), 806-816. https://doi.org/10.1111/2041-210X.14060
Joo, R., Boone, M.E., Clay, T.A., Patrick, S.C., Clusella-Trulas, S. & Basille, M. (2020) Navigating through the R packages for movement. BES, 89, 248-267. https://doi.org/10.1111/1365-2656.13116
Kranstauber, B., Smolla, M., & Scharf, A. K. (2018). move: Visualizing and analyzing animal track data. R package version 3.1.0. https://CRAN.R-project.org/package=move
Kranstauber, B., Safi, S. & Scharf, A.K. (2024) move2: R package for processing movement data. Methods in Ecology and Evolution, 15(9), 1561-1567. https://doi.org/10.1111/2041-210X.14383
Langley, L.P., Bearhop, S., Burton, N.H. Banks, A.N., Frayling, T., Thaxter, C.B., Clewley, G.D., Scragg, E. & Votier, S.C. (2021) GPS tracking reveals landfill closures induce higher foraging effort and habitat switching in gulls. Movement Ecology, 9, 56 https://doi.org/10.1186/s40462-021-00278-2
O’Hanlon, N.J., Clewley, G.D., Johnston, D.T., Thaxter, C.B., Langlois Lopez S., Quinn, L.R., Boersch-Supan, P.H., Masden, E.A., Daunt, F., Wilson, J., Burton, N.H.K., Humphreys, E.M. Partial Niche Partitioning in Three Sympatric Gull Species Through Foraging Areas and Habitat Selection. Ecology & Evolution, 15(7), e71577 https://doi.org/10.1002/ece3.71577
Signer, J. Fieberg, J. & Avgar, T. (2019) Animal movement tools (AMT): R package for managing tracking data and conducting habitat selection analysis. Ecology and Evolution, 9, 880-890. https://doi.org/10.1002/ece3.4823
Sumner, M.D. (2011) The Tag Location Problem. Ph.D. thesis, University of Tasmania. https://eprints.utas.edu.au/12273/3/sumner.pdf.
Sumner, M.D., Wotherspoon, S.J. & Hindell, M.A. (2009) Bayesian estimation of animal movement from archival and satellite tags. PLoS ONE, 4(10). http://dx.plos.org/10.1371/journal.pone.0007324.
Thaxter, C.B., Ross-Smith, V.H., Bouten, W., Clark, N.A., Conway, G.J., Rehfisch, M.M. & Burton, N.H.K. (2015) Seabird–wind farm interactions during the breeding season vary within and between years: A case study of lesser black-backed gull Larus fuscus in the UK. Biological Conservation, 186 https://doi.10.1016/j.biocon.2015.03.027
Thaxter, C.B., Quinn, L., Atkinson, P., Booth Jones, K.A., Clark, N.A., Clewley, G.D., Green, R.M.W., O’Hanlon, N.J., Johnston, D.T., Masden, E.A., Ross-Smith, V.H., Sage, E., Scragg, E., Taylor, R., Burton, N.K.H. & Humpreys, E.H. 2025. Breeding and foraging habitat are important in determining foraging ranges of sympatric generalist species. Ibis https://doi:10.1111/ibi.13446