Exploring dog movement
Serge Morand
2024-09-15
1 Load libaries
library(ggplot2)
library(move)
library(move2)
library(trackdf)
library(trajr)
library(circular)
library(wildlifeDI)
library(move2)
library(adehabitatLT)
library(igraph)
2 Managing collected data
Three data types are monitored in the field:
- 1. data from
sensors: a list of sensors, here, GPS collars
- 2. data from
animals: a list of individual animals equipped with GPS collars (tags)
- 3. data of events: a list of logs (per unit of time) with GPS
positions for each tagged animal (deployment)
figure
from movebank
2.1 Get (load) RData file
A full dataset of logs of individual dogs for all sessions, acquired by the ANR project SEAdogSEA.
2.1.1 Display the structure of the dataframe using ‘str’
## tibble [224,665 × 9] (S3: tbl_df/tbl/data.frame)
## $ session : num [1:224665] 1 1 1 1 1 1 1 1 1 1 ...
## $ dog_id : chr [1:224665] "D206" "D206" "D206" "D206" ...
## $ dogName_en : chr [1:224665] "Mhee" "Mhee" "Mhee" "Mhee" ...
## $ village_name: chr [1:224665] "Huak (4)" "Huak (4)" "Huak (4)" "Huak (4)" ...
## $ logtime : POSIXct[1:224665], format: "2019-11-14 00:16:23" "2019-11-14 01:03:36" ...
## $ habitat : chr [1:224665] "teak" "urban" "urban" "urban" ...
## $ distance : num [1:224665] 51.1 19.3 39.9 19.6 19.1 ...
## $ long : num [1:224665] 101 101 101 101 101 ...
## $ lat : num [1:224665] 19.1 19.1 19.1 19.1 19.1 ...
2.1.2 Distribution of distances between each log and the house of dog owner (all sessions) for all dogs.
ggplot(data = all_sessions,aes(x=distance, colour = dog_id))+
geom_density() +
labs(title="Distribution of distances between logs and dog's owner house") +
theme_classic() + theme(legend.position="none")
3 Visualization of movement
A first visualization using the package “move”
Here, we select
all sessions for the dog “D211”
mydog <- all_sessions |>
dplyr::select(long,lat,logtime,dog_id) |>
dplyr::arrange(logtime)|>
dplyr::filter(dog_id == "D211")
Prepare a move object
mydog_session <- move(
x=mydog$long, y=mydog$lat,
time=as.POSIXct(mydog$logtime, format="%Y-%m-%d %H:%M:%OS", tz="Asia/Bangkok"),
proj=CRS("+proj=longlat +ellps=WGS84 +datum=WGS84"),
data=mydog,
animal=mydog$dog_id)
Print the movement of “D211”
plot(mydog_session, xlab="Longitude", ylab="Latitude",main="Movement of D211",
type="l", pch=16, lwd=0.5)
points(mydog_session, pch=20, cex=0.5)
Prepare a move2 object
mydog_session2 <- move2::mt_as_move2(mydog_session,
coords = c("long", "lat"), time_column = "logtime",
track_id_column = "dog_ID") |>
sf::st_set_crs(4326L)
Plot using mapview
Prepare a track object of D211 and plot using ggplot
track_mydog <- track(x = mydog$long, y = mydog$lat, t = mydog$logtime,
id = mydog$dog_id, proj = "+proj=longlat", tz = "Asia/Bangkok")
ggplot(data = track_mydog) +
aes(x = x, y = y, color = id) +
geom_path() +
coord_map()
Create a trajectory from the log coordinates of D211 using ‘trajr’ and plot
trj <- TrajFromCoords(mydog|>
dplyr::select(long,lat),
spatialUnits = "pixels")
plot(trj, xlab="Longitude", ylab="Latitude")
Using ‘circular’ to identify potential corridor
mydog_sessionCorr <- corridor(mydog_session)
plot(mydog_sessionCorr, type="l", xlab="Longitude", ylab="Latitude", col=c("black", "grey"), lwd=c(1,2))
legend("bottomleft", c("Corridor", "Non-corridor"), col=c("black", "grey"), lty=c(1,1), bty="n")
If center=TRUE: the center of the coordinate system is the center of the track. Units are in meters
mydog_session.prj <- spTransform(mydog_session, center=TRUE)
dBB.mydog_session <- brownian.bridge.dyn(mydog_session.prj, ext=.85, raster=100, location.error=20)
UDmydog_session <- getVolumeUD(dBB.mydog_session)
Plot the Utization Distribution (UD) of D211 The UD represents the minimum area in which an animal has some specified probability of being located
plot(UDmydog_session, main="UD and contour lines for D211")
contour(UDmydog_session, levels=c(0.5, 0.95), add=TRUE, lwd=c(0.5, 0.5), lty=c(2,1))
plot UD85 and UD50
par(mfrow=c(1,2))
## mantaining the lower probabilities
ud95 <- UDmydog_session
ud95[ud95>.95] <- NA
plot(ud95, main="UD95")
## UD 50
ud50 <- UDmydog_session<=.5
plot(ud50, main="UD50")
3.0.1 Analysing speed
Compute speed and accelation using ‘TrajDerivatives’, plot isong
# Smooth before calculating derivatives
smoothed <- TrajSmoothSG(trj, 2, 15)
# Calculate speed and acceleration
derivs <- TrajDerivatives(smoothed)
# Plot change-in-speed and speed
plot(derivs$acceleration ~ derivs$accelerationTimes, type = 'l', col = 'red',
yaxt = 'n',
xlab = 'Time (s)',
ylab = expression(paste('Change in speed (', m/s^2, ')')))
axis(side = 2, col = "red")
lines(derivs$speed ~ derivs$speedTimes, col = 'blue')
axis(side = 4, col = "blue")
mtext('Speed (m/s)', side = 4, line = 3)
abline(h = 0, col = 'lightGrey')
Plot speed over time.
4 Contacts
The package ‘wildlifeDI’ allows the investigation of contacts between individuals. - 1. create as sf object of the full data (all_sessions) - 2. transform into a move2 object
# sf object
alldog_sf <- sf::st_as_sf(all_sessions, coords = c("long", "lat"),
crs= "+proj=longlat +datum=WGS84 +ellps=WGS84 +towgs84=0,0,0") |>
dplyr::select(logtime,dog_id,session,village_name,habitat)
# move2 object
alldog_obj <- mt_as_move2(alldog_sf ,
time_column = "logtime", track_id_column = "dog_id", session = "session")
Plot contact distances, selecting one session (“3) and villages 6 and 7 (one reason is to decrease the computation time)
dcPlot(alldog_obj |>
dplyr::filter(session =="3",
village_name %in% c("Huay Muang (6)", "Sun Ti Suk (7)")),
tc=15*60,dmax=1000)
## [1] 156.5153 439.2527 570.5236 833.0654
Compute contact distances, selecting one session (“3) and villages 6 and 7 (one reason is to decrease the computation time)
doecons <- conProcess(alldog_obj |>
dplyr::filter(session =="3",
village_name %in% c("Huay Muang (6)", "Sun Ti Suk (7)")),
dc=50,tc=15*60)
doephas <- conPhase(doecons, pc=60*60)
consum <- doephas |>
sf::st_drop_geometry() |>
dplyr::filter(!is.na(contact_pha)) |>
dplyr::group_by(contact_pha) |>
dplyr::summarise(
nfix = dplyr::n(),
t1 = min(logtime),
t2 = max(logtime),
duration = max(logtime)-min(logtime),
avg_d = mean(contact_d,na.rm=T),
min_d = min(contact_d,na.rm=T),
max_d = max(contact_d,na.rm=T)
)
consum
Map contacts of dog “D219” (contact 1) with other dogs (other contacts)
pha_lin <- doephas |>
dplyr::filter(!is.na(contact_pha)) |>
dplyr::group_by(contact_pha) |>
dplyr::summarise(n = dplyr::n(),do_union=FALSE) |>
dplyr::filter(n > 1) |>
sf::st_cast("LINESTRING")
ggplot() +
geom_sf(data=alldog_obj |>
dplyr::filter(session =="3",
village_name %in% c("Huay Muang (6)", "Sun Ti Suk (7)")),
aes(color=mt_track_id(alldog_obj |>
dplyr::filter(session =="3",
village_name %in% c("Huay Muang (6)", "Sun Ti Suk (7)"))))) +
geom_sf(data=pha_lin)
Prepare object for network analysis
cons <- conProcess(alldog_obj |>
dplyr::filter(session =="3",
village_name %in% c("Huay Muang (6)", "Sun Ti Suk (7)")),
dc=50,tc=15*60,
return='contacts')
cons
Prepare igraph object
tab_cnt <- cons |>
dplyr::count(id1,id2)
gr <- graph_from_data_frame(tab_cnt,directed=FALSE)
E(gr)$weight <- tab_cnt$n
plot(gr)
Detect modules and plot unipartite network of contacts between individual dogs
## IGRAPH clustering fast greedy, groups: 3, mod: 0.46
## + groups:
## $`1`
## [1] "D219" "D220" "D236" "D237" "D238"
##
## $`2`
## [1] "D245" "D246" "D247" "D248" "D249"
##
## $`3`
## [1] "D242" "D243" "D244"
## 
Map overlap of home range between dogs, villages 6 and 7, session 3
idcol <- mt_track_id_column(alldog_obj |>
dplyr::filter(session =="3",
village_name %in% c("Huay Muang (6)", "Sun Ti Suk (7)")))
mcphr <- alldog_obj |>
dplyr::filter(session =="3",
village_name %in% c("Huay Muang (6)", "Sun Ti Suk (7)")) |>
dplyr::group_by_at(idcol) |>
dplyr::summarise() |>
sf::st_convex_hull()
mapview::mapview(mcphr)