align_move does not preserve data (colour) column

[nmasto]

Hi Jakob,

I searched but couldn't find an answer. I'm trying to color each track by a behavioral class. This worked fine with a smaller dataset (one month), two colours (sex = male/female), and 20 individuals. However, I'm trying on a larger dataset of 39 individuals, each with ~2,000 locations. At first I thought it was because the timezone was "America/Chicago" which would switch from 'CDT' to 'CST' in these migratory tracks. However, I switched to UTC and encounter the same issue. I then found that when using df2move there were issues with a single individual...so I removed that individual. Everything seemed fine however, despite trying different temporal alignments, I can't seem to get align_move to preserve my data. See below my code. Any assistance is appreciated.

# Data Management
dat <- read.csv("data/full_dat_env_cov_classify.csv") %>% 
  mutate(time = ymd_hms(time, tz = "America/Chicago"),
         time_utc = ymd_hms(time_utc, tz = "UTC"))

# Time is handled; make study_season a factor
dat$study_season <- as.factor(dat$study_season)

# Make a `colour` column for behavioral states
dat <- dat %>% group_by(ID) %>%
  distinct(time_utc, .keep_all = TRUE) %>%   # keep distinct records
  mutate(colour = case_when(class == "postbreeding" ~ "red",
                            class == "migrating" ~ "green",
                            class == "stopover" ~ "purple",
                            class == "winter" ~ "blue",
                            TRUE ~ "red")) %>%                      # Give post-breeding color to those NAs that will be before segmentation began
  filter(ID != "LLR-b1142021-F_fall_winter_4") %>%  # Filter ID that was previously giving me trouble (390 NAs when `df2move`
  ungroup()

# Separate yrs. Wish I knew how to do all years but can't
yr1 <- dat %>% filter(study_season == "fall_winter_2")
yr2 <- dat %>% filter(study_season == "fall_winter_3")
yr3 <- dat %>% filter(study_season == "fall_winter_4")

# Order
yr1 <- yr1[order(yr1$ID, yr1$time_utc),]
yr2 <- yr2[order(yr2$ID, yr2$time_utc),]
yr3 <- yr3[order(yr3$ID, yr3$time_utc),]

# Year 3----

check_locs <- yr3 %>% group_by(BirdsID_season) %>% count() # at least 400 locs

yr3 <- as.data.frame(yr3) # not a tibble

yr3 <- df2move(yr3, x = "x", y = "y",
                    time = "time_utc", track_id = "BirdsID_season",
                    proj = "+proj=longlat +datum=WGS84 +no_defs", data = yr3) # append data; importantly "colour"

sum(table(yr3$colour))  # 46273
sum(is.na(yr3$colour))  # 0 NAs

lag = unlist(timeLag(yr3, unit = "mins")) # sampling varies
median(lag)  # 60 min
mean(lag)     # 77 min
sd(lag)          # 328 min -- maybe the lag? Try different temporal alignments

# Maybe something to do with the standard deviation/timelag?
mdat <- align_move(yr3, res = 240, digit = 0, unit = "mins") # Every 4 hrs
mdat2 <- align_move(yr3, res = 360, digit = 0, unit = "mins") # Every 6 hrs

# `align_move` does not preserve colour column
 sum(is.na(mdat$colour))
[1] 14866
sum(is.na(mdat2$colour)) 
[1] 9904

What am I missing? Align move worked fine and did not convert colour class to NAs when using a much smaller dataframe of only a single month and only two "colours". Any help appreciated. Best, Nick

[J4SJA]

I've had the same. Because I keep the resolution the same as it was, I 'just' take the colour out and then put it back after i've used align_move

[kaijagahm]

@nmasto I'm curious if you ever found a solution to this? I am not keeping the resolution the same unfortunately so I can't use @J4SJA's solution. Manually annotating the behavioral segments is really onerous and won't work in my case.