Movement and space use
Cohort outlines, one tortoise's fixes, or sighting effort, on one map.
Viewing entire cohort · records through 17 Aug 2026
Movement map
Recorded positions never imply a traveled path, so no line joins two fixes. An area outline is a kernel contour or a dashed occurrence footprint, never an averaged cohort range. Use-intensity cells are sighting effort, not abundance.Preparing the recorded positions. The rest of the page is ready.
Generated records on a real landscape. The marks on this map are deterministic sample records placed on the real Santa Rita Experimental Range. They describe no real survey, animal, plant, or treatment; the basemap and context layers are the only real things drawn.
How precision is drawn
Marks draw at their generated precision; nothing is rounded or moved for display. Context layers are committed public snapshots of OpenStreetMap (ODbL), credited in the map's attribution line; choosing a road, wash or trail shows its layer and any name the source map gives it. Layers that start switched off are still listed in the layers button at the top right; tick one to draw it.
Surface
The window scopes positions and outlines. Sighting effort always reads every year.
Sighting effort counts every recorded position on this load, in every year and for every animal. It shows where effort went, not where tortoises are.
Season
From record: Fixed 15 June to 30 September window, the definition in force since 01 Jan 2008.
Map context
Committed public context, credited in the map's attribution line. The layers button on the map also switches the Topo, Light and Imagery basemaps.
Playback
- Play steps through the chosen tortoise's recorded positions in date order, one fix per step: earlier fixes stay as faint dots and the current fix is the filled disc.
- No line joins two fixes. Relocations are days apart, and a line between them would draw a route nobody observed.
- The map keeps its zoom while playing and moves only when the current fix would leave the view; after a drag it waits 4 seconds before it follows again.
- With a playback control focused, the Left and Right arrow keys step one fix, and Home and End jump to the first and last.
- A step longer than 90 days says that no position is recorded across the gap.
Position ledger ·
One row for every mark the map draws, whichever surface is showing.
Typical-move ledger · 24 animals · 1,948 moves
What the columns mean
- Period
- The years a row pools, set by Combine years.
- Season
- With Year and season, whether a move ends inside the Monsoon window (15 Jun to 30 Sep) or outside it.
- Sex
- The animal's recorded sex.
- Typical move (95% interval)
- The median straight-line distance between two fixes of one animal 3 to 90 days apart, with a seeded bootstrap interval over the row's moves (600 resamples). It is not scaled for the time between the two fixes.
- Most common band
- Moves vote on a fixed ladder with breaks at 50, 150 and 400 m; the band holding the most moves is shown, and a tie prints no dominant band.
- Median gap
- The median days between the two fixes of a move. Compare it before comparing distances across rows.
- Animals / moves
- Animals with a move in the row, then the moves. An animal with moves in several rows counts in each.
- Withheld
- A row with fewer than 5 animals prints Suppressed and withholds its values.
Behind this number
3 functions compute it. Each is shown exactly as this build runs it.
.tor_move_steps
apps/hub/R/tor_movement_revamp.R
.tor_move_steps <- function (loaded)
{
d <- .tor_geo_frame(loaded)
roster <- .tor_context_roster(loaded)
none <- data.frame(animal_id = character(0), date = as.Date(character(0)),
year = integer(0), gap = numeric(0), disp = numeric(0), sex = character(0),
stringsAsFactors = FALSE)
d <- d[d$animal_id %in% roster$animal_id, , drop = FALSE]
if (!nrow(d))
return(none)
steps <- do.call(rbind, lapply(split(d, d$animal_id), function(one) {
if (nrow(one) < 2L)
return(NULL)
one <- one[order(one$event_date), , drop = FALSE]
gap <- as.numeric(diff(one$event_date))
disp <- .tor_dist_m(utils::head(one$lon, -1L), utils::head(one$lat,
-1L), utils::tail(one$lon, -1L), utils::tail(one$lat, -1L))
keep <- gap >= TOR_MOVE_GAP_DAYS[1] & gap <= TOR_MOVE_GAP_DAYS[2]
if (!any(keep))
return(NULL)
later <- one$event_date[-1L]
data.frame(animal_id = one$animal_id[1], date = later[keep], year = as.integer(format(later,
"%Y"))[keep], gap = gap[keep], disp = disp[keep], stringsAsFactors = FALSE)
}))
if (is.null(steps) || !nrow(steps))
return(none)
sex <- stats::setNames(.tor_chr(roster$sex), roster$animal_id)
steps$sex <- unname(sex[steps$animal_id])
rownames(steps) <- NULL
steps
}
.tor_move_ledger_data
apps/hub/R/tor_movement_revamp.R
.tor_move_ledger_data <- function (loaded, combine = 3L, group = "year")
{
steps <- .tor_move_steps(loaded)
if (!nrow(steps))
return(NULL)
by_season <- identical(group, "season")
win <- if (by_season)
tor_season_window(loaded)
else NULL
if (by_season && !isTRUE(win$ok))
by_season <- FALSE
steps$sex <- as.vector(ddl_decode(loaded, "tortoise", "TortoiseSexDomain",
toupper(steps$sex)))
yr_min <- min(steps$year)
steps$period <- if (identical(as.character(combine), "all"))
"All years"
else {
k <- as.integer(combine)
lo <- yr_min + ((steps$year - yr_min)%/%k) * k
hi <- pmin(lo + k - 1L, max(steps$year))
ifelse(lo == hi, as.character(lo), sprintf("%s–%s", lo, hi))
}
steps$season <- if (by_season)
ifelse(tor_in_season(steps$date, win), sprintf("%s window", win$name),
"Outside the window")
else ""
keys <- if (by_season)
list(steps$period, steps$season, steps$sex)
else list(steps$period, steps$sex)
band_label <- sprintf("%s m", c(sprintf("0–%s", TOR_MOVE_BANDS_M[1]),
sprintf("%s–%s", utils::head(TOR_MOVE_BANDS_M, -1L), utils::tail(TOR_MOVE_BANDS_M,
-1L)), sprintf("%s+", TOR_MOVE_BANDS_M[length(TOR_MOVE_BANDS_M)])))
rows <- do.call(rbind, lapply(split(steps, keys, drop = TRUE), function(g) {
n_animals <- length(unique(g$animal_id))
lead <- data.frame(Period = g$period[1], Season = g$season[1], Sex = g$sex[1],
stringsAsFactors = FALSE)
if (n_animals < TOR_MOVE_FLOOR_ANIMALS)
return(cbind(lead, data.frame(Typical = sprintf("Suppressed (%s)",
.tor_move_one(n_animals, "animal", "animals")), Band = "Withheld",
`Median gap` = "Withheld", `Animals / moves` = sprintf("%s / %s",
n_animals, nrow(g)), .median_m = NA, .lo_m = NA, .hi_m = NA,
.gap_days = NA, .animals = n_animals, .moves = nrow(g), .suppressed = TRUE,
check.names = FALSE, stringsAsFactors = FALSE)))
med <- stats::median(g$disp)
boot <- local({
set.seed(20260817L + nrow(g))
vapply(seq_len(TOR_MOVE_BOOT_N), function(b) stats::median(sample(g$disp,
nrow(g), replace = TRUE)), numeric(1))
})
ci <- stats::quantile(boot, c(0.025, 0.975), names = FALSE)
counts <- table(factor(band_label[findInterval(g$disp, TOR_MOVE_BANDS_M) +
1L], levels = band_label))
top <- which(counts == max(counts))
gap_med <- stats::median(g$gap)
cbind(lead, data.frame(Typical = sprintf("%s m [%s–%s]", fmt_n(round(med)),
fmt_n(round(ci[1])), fmt_n(round(ci[2]))), Band = if (length(top) >
1L)
"no dominant band"
else names(counts)[top], `Median gap` = sprintf("%s days", fmt_n(round(gap_med))),
`Animals / moves` = sprintf("%s / %s", n_animals, nrow(g)), .median_m = round(med),
.lo_m = round(ci[1]), .hi_m = round(ci[2]), .gap_days = round(gap_med),
.animals = n_animals, .moves = nrow(g), .suppressed = FALSE,
check.names = FALSE, stringsAsFactors = FALSE))
}))
rows <- rows[order(rows$Period, rows$Season, rows$Sex), , drop = FALSE]
if (!by_season)
rows$Season <- NULL
rownames(rows) <- NULL
attr(rows, "grouping") <- list(combine = as.character(combine), season = by_season)
rows
}
.tor_dist_m
apps/hub/R/map_tortoise.R
.tor_dist_m <- function (lon1, lat1, lon2, lat2)
{
r <- 6371008.8
p <- pi/180
a <- sin((lat2 - lat1) * p/2)^2 + cos(lat1 * p) * cos(lat2 * p) * sin((lon2 -
lon1) * p/2)^2
2 * r * asin(pmin(1, sqrt(a)))
}Read-only excerpts of this build's calculation code (hash ef2625aa4521). No record appears in them, and this page displays them without running them.