birdrec: Bird Species Detections

In psolymos/opticut: Likelihood Based Optimal Partitioning and Indicator Species Analysis

Bird Species Detections

Description

Data set listing 156 species (mostly birds, few amphibians and mammals) detected at 127 sites (367 point locations) in Alberta, Canada in 2015, using autonomous recording technology (ARU; Wildlife Acoustic Song Meter) for sound recordings.

Usage

data("birdrec")

Format

A list with 3 elements with matching ordering: xtab is a sample x species matrix with number of detections, samp is a data frame with sample level attributes. taxa is a data frame with species level attributes.

Multiple random recordings at each location were selected according to a stratified random design (based on combination of TOY and TOD). These recordings were listened to by trained analysts and species were identified based on auditory cues.

This data set lists detections from the first 1-minute segment of each recording. Dates for the 3967 1-minute segments range between 2015-03-31 and 2015-07-29. Variables in birdrec$samp are the following:

PKEY: primary key for location/time combinations.

POINT: unique spatial location IDs, each point had its own ARU unit.

SITE: site ID (1-4 ARU units deployed per site).

YEAR: year, 2015.

MONTH: month from 3 (March) to 7 (July).

MDAY: day of month, 1-31.

HOUR: 24-hour of day, values between 0-12.

MINUTE: minute, 0-59.

YDAY: ordinal day of the year, 89-209.

RAIN, WIND, INDUSTRY, NOISE: level of rain, wind, industrial noise, and background noise. 0 = no; 1 = light; 2 = moderate; 3 = heavy.

MICROPHONE: Every recording contains a certain level of background static due to the pre-amplifiers; however, problems, such as, electrostatic discharge on the microphones, faulty wiring, poorly installed microphones and/or missing microphones can occur causing excess static or dead channels. 0 = no microphone related issues; 1 = left microphone cuts out intermittently; 2 = right microphone cuts out intermittently; 3 = both microphones cut out intermittently; 4 = left channel failed; 5 = right channel failed; 6 = both channels failed (no cases in the data set); 7 = left side extra static; 8 = right side extra static; 9 = both sides extra static; 10 = other issues; 11 = unbalanced channels.

TOY: time of year intervals used for stratified random selection of dates. 8 intervals divided into 3 major units (early, mid, and late breeding season; YDAY 140 and 180 were used as threshold between the major units).

TOD: time of day, midnight (HOUR = 0) or morning (HOUR > 0).

Variables in birdrec$taxa are the following: Species, CommonName, ScientificName, Family, Order, Class, and MigratoryBehaviour.

Methodology and metadata is described in ABMI (2016), and Lankau et al. (2015).

Source

Alberta Biodiversity Monitoring Institute (ABMI, www.abmi.ca)

References

Alberta Biodiversity Monitoring Institute (ABMI), 2016. Terrestrial field data collection protocols (abridged version) 2016-05-18. Alberta Biodiversity Monitoring Institute; Edmonton, Alberta, Canada.

Lankau, H.E., MacPhail, A., Knaggs, M. & Bayne, E., 2015. Acoustic recording analysis protocol. Bioacoustic Unit, University of Alberta, and Alberta Biodiversity Monitoring Institute; Edmonton, Alberta, Canada.

Examples

data(birdrec)
str(birdrec)

aggregate(rowSums(birdrec$xtab),
    list(TOY=birdrec$samp$TOY, TOD=birdrec$samp$TOD), mean)
boxplot(rowSums(birdrec$xtab) ~ TOD + TOY, birdrec$samp,
    col=c("gold", "tomato"), ylab="# detections")

## Not run: 
y <- ifelse(birdrec$xtab > 0, 1, 0)
g <- paste0(gsub("[[:digit:]]", "", as.character(birdrec$samp$TOY)),
    substr(as.character(birdrec$samp$TOD), 1, 4))
g <- factor(g, levels=c("EarlyMorn", "MidMorn", "LateMorn",
    "EarlyMidn", "MidMidn", "LateMidn"))
## binary response model
oc <- opticut(y ~ 1, strata=g, dist="binomial")
## multi-level response model
mc <- multicut(y ~ 1, strata=g, dist="binomial")

## testing equality of labels
splito <- as.character(summary(oc)$summary$split)
splitm <- as.character(summary(mc)$summary$split)
table(splito == splitm)
## seeing how much those differ
bpo <- summary(oc)$bestpart
bpm <- summary(mc)$bestpart
rs <- rowSums(abs(bpo-bpm))
table(rs)
10 * bpo[rs > 0,] + bpm[rs > 0,]

## End(Not run)

psolymos/opticut documentation built on Nov. 27, 2022, 11:29 a.m.