In jmlondon/kotzeb0912: kotzeb0912: An R package of data from a satellite telemetry study of bearded seals in Alaska

library(kotzeb0912)
library(dplyr)
library(ggmap)
library(knitr)

## Global options
options(max.print="75")
opts_chunk$set(echo=FALSE,
               prompt=FALSE,
               tidy=TRUE,
               comment=NA,
               message=FALSE,
               warning=FALSE)
opts_knit$set(width=75)

Josh M. London
Polar Ecosystems Program
Alaska Fisheries Science Center, NOAA
Seattle, Washington, USA
orcid: 0000-0002-3647-5046
[e-mail]

Michael F. Cameron
Polar Ecosystems Program
Alaska Fisheries Science Center, NOAA
Seattle, Washington, USA

Peter L. Boveng
Polar Ecosystems Program
Alaska Fisheries Science Center, NOAA
Seattle, Washington, USA

Last updated: r Sys.Date()

Executive Summary

data("kotzeb0912_locs")

locs <- kotzeb0912_locs %>% 
  dplyr::select(deployid,date_time,latitude,longitude) %>% 
  dplyr::filter(!is.na(latitude)) %>% 
  dplyr::arrange(deployid,date_time) %>% 
  data.frame()

med_lng <- median(locs$longitude)
med_lat <- median(locs$latitude) -3

map <- get_map(location=c(med_lng,med_lat),
                     zoom=4,source="google",maptype='terrain',color="bw")
gg <- ggmap(map) + geom_point(data=locs,mapping=aes(x=longitude,y=latitude),
                              alpha=0.25,size=0.75)
gg

Bearded seals (Erignathus barbatus) are one of the most important subsistence resources for the indigenous people of coastal northern and western Alaska, as well as key components of Arctic marine ecosystems, yet relatively little about their abundance, seasonal distribution, migrations, or foraging behaviors has been documented scientifically. Ice-associated seal populations may be negatively impacted by offshore oil and gas development as well as by climate change. Our ability to predict impacts, however, is limited by inadequate knowledge of seal population structure and foraging ecology. By working cooperatively with Alaska Native subsistence hunters we developed methods for live-capturing bearded seals in the Chukchi Sea using nets set in the shallow coastal waters where bearded seals were foraging. Capture efforts were based out of Kotzebue and various locations in the North Slope Borough from Wainwright to Barrow in June and July from 2009 to 2012. In all, 7 seals were caught (2 adults and 5 sub-adults; 4 males and three females; ranging in length and weight from 159 cm and 116 kg to 216 cm and 253 kg), all from Kotzebue Sound. Each seal was sampled for health and condition and released with three different types of satellite-linked bio-loggers: the SPOT5, attached to a rear flipper, provided information on the timing of hauling out and on the seal’s location for up to three years; the MK10, glued to the top of a seal’s head, provided Argos estimates of location and also provided data on the timing and depths of dives, for up to ten months; the Mk10-AF, also glued to the top of the head, provided GPS quality locations in addition to the Argos estimates of location and dive behavior data.

Bearded Seals

Seven bearded seals were captured between 2009 and 2012 and 14 tags were deployed. Each animal was released with a head-mounted satellite tag and a flipper-mounted satellite tag.

data('kotzeb0912_deployments')
d <- dplyr::arrange(kotzeb0912_deployments,speno,deployid) %>% 
  dplyr::select(-ptt)
hdrs <- c("Speno","Capture Time (UTC)","Age","Sex","DeployID","Tag Location")
knitr::kable(d,col.names = hdrs)

Telemetry Devices

All of the telemetry devices deployed in this study were manufactured by Wildlife Computers (Redmond, Washington, USA) [^2]. All of the tags relied on the Argos satellite network for location esimates and transfer of data. A few tags were equipped with Fastloc-GPS capabilities that provided a limited number of GPS quality locations [^3]. In addition to the Fastloc-GPS tags, flipper-mounted SPOT style tags were also deployed. These tags provided location and haul-out data for longer duration than the Mk10 tags (attached to the hair which molts each spring).

IMPORTANT NOTICE

Percent dry timeline data from the flipper-mounted SPOT style tags should NOT be used for any analysis. Several of the tags reported prolonged periods of 100% dry when we know this not to be true. The source of this is unknown but likely a result of sensor failure or compromise due to environmental conditions.

[^2]: mention of specific products or manufacturers does not constitute an endorsement by NMFS, NOAA, or the U.S. Department of Commerce

[^3]: fastloc-GPS was used sparingly because the technology was relatively new at the time of this study and there is a significant increase in battery consumption when using Fastloc-GPS compared to Argos.

Data Gathering and Processing

Tag Programming

There were three tag types/models deployed during this project

1. Mk10
2. Mk10-AF
3. SPOT5

Mk10 (-AF) Programming Summary

Histogram Data

Histogram Data sampling interval
  ~ 10 seconds
Dive Maximum Depth (m)
  ~ 14 bins:    10;30;50;70;90;100;150;200;250;300;400;500;600;>600
Dive Duration (min)
  ~ 14 bins:    1;2;3;4;6;8;10;12;16;20;30;40;50;>50
Time-at-Temperature (C) 
  ~ disabled
Time-at-Depth (m)
  ~ 14 bins:    4;10;30;50;70;90;100;150;200;250;300;400;500;>500
20-min time-line
  ~ disabled
Hourly % time-line (low resolution)
  ~ enabled
Hourly % time-line (high resolution)
  ~ disabled
Light-level locations
  ~ disabled

Histogram Collection

Hours of data summarized in each histogram
  ~ 6
Histograms start at
  ~ GMT 03:00

Dive & Timeline Definition

Depth reading to determine start and end of dive
  ~ Wet/Dry (Mk10)
  ~ 2m (Mk20-A)
Ignore dives shallower than 
  ~ 4m
Ignore dives shorter than
  ~ 1 min
Depth threshold for timelines
  ~ Wet/Dry
Haulout Definition
  ~ A minute is "dry" if Wet/Dry sensor is dry for any **30** seconds in a minute
  ~ Enter haulout state after **5** consecutive dry minutes
  ~ Exit haulout state if wet for any **50** seconds in a minute

Argos Transmissions

Transmission Hours
  ~ 0-23
Transmission Days
  ~ All Days
Transmission Monhts
  ~ All Months
Daily Transmission Cap
  ~ 250 Transmissions

Fast-GPS Settings

Fast-GPS sampling interval
  ~ 360 minutes
Deployment Latitude
  ~ 66.75 degrees
Deployment Longitude
  ~ -163 degrees
Deployment Altitude
  ~ 0 m
Transmit hours
  ~ 0 - 23
Fast-GPS Collection Days
  ~ January 4, 8, 12, 16, 20, 24, 28
  ~ February    1, 5, 9, 13, 17, 21, 25
  ~ March   1, 5, 9, 13, 17, 21, 25, 29
  ~ April   2, 6, 10, 14, 18, 22, 26, 30
  ~ May 4, 8, 12, 16, 20, 24, 28
  ~ June    1, 5, 9, 10, 14, 18, 22, 26, 30
  ~ July    4, 8, 12, 16, 20, 24, 28
  ~ August  1, 5, 9, 13, 17, 21, 25, 29
  ~ September   2, 6, 10, 14, 18, 22, 26, 30
  ~ October 4, 8, 12, 16, 20, 24, 28
  ~ November    1, 5, 9, 13, 17, 21, 25, 29
  ~ December    3, 7, 11, 15, 19, 23, 27, 31

Fast-GPS Control

Maximum successful Fast-GPS attempts
  ~ 1 per hour; 4 per day
Maximum failed Fast-GPS attempts
  ~ 3 per hour
Overall maximum Fast-GPS attempts
  ~ 12 per day
Supress Fast-GPS after good haulout location
  ~ enabled

SPOT5 Programming Summary

Argos Transmissions

Unused transmissions will be added to the next day's allowance.
Maximum transmissions per day
  ~ 150
Transmit on these hours
  ~ 1 - 4, 20 - 23
  ~ Tag will transmit during all of the hours before midnight on the first day of deployment
Transmit on these days, using an absolute calendar
  ~ Jan: 1, 15, 
  ~ Feb: 1, 7, 13, 19, 25, 
  ~ Mar: 1, 7, 13, 19, 25, 31
  ~ Apr: 6, 12, 18, 24, 30
  ~ May: 6, 12, 18, 24, 30, 
  ~ Jun: 6, 12, 18, 24, 30
  ~ Jul: 6, 12, 18, 24, 30, 
  ~ Aug: 1, 15, 
  ~ Sep: 1, 15, 
  ~ Oct: 1, 15, 
  ~ Nov: 1, 15, 
  ~ Dec: 1, 15,

Time at temperature histograms **are not collected**
Percent dry timelines **are being collected**

Argos Least-Squared vs. Kalman Processing

In 2011, Argos changed their location estimation algorithm to include a Kalman filter algorithm. This replaced their previous algorithm which relied on a least-squares process. The Kalman filter algorithm provides additional error data which is critical for many movement models. When this change was made, all data for tags in this project were reprocessed back to January of 2008. In most cases, for each location estimate in the least-squares dataset, there is a corresponding Kalman filter estimate for that satellite pass.

The reprocessed data and the originally delivered least-squares data were merged and reconciled by Wildlife Computers within their data portal. In this process, locations were matched by ptt, date-time, satellite and pass duration. date-time and pass duration were matched with fuzzy logic (i.e. they were allowed to not match exactly). In all cases, the reprocessed kalman locations were determined to be authoritative.

Wildlife Computers Data Portal

Data included in this package were downloaded from the Wildlife Computers Data Portal using the wcUtils package. After download, additional re-structuring and processing of the data was also done with the wcUtils package.

Those with collaborator permissions can access these deployments directly by searching for the kotzeb0912 projectid.

Data Components

The kotzeb0912 package is a data package for distribution of core data products resulting from this study. There are 7 data products distributed with this package:

1. kotzeb0912_locs
   • Argos location estimates for all deployments
2. kotzeb0912_gps
   • GPS location estimates from Fastloc solutions
3. kotzeb0912_status
   • tag status and performance messages for all deployments
4. kotzeb0912_depths
   • dive-depth histogram data; cleaned and tidy'd
5. kotzeb0912_durations
   • dive-duration histogram data; cleaned and tidy'd
6. kotzeb0912_tad
   • time-at-depth histogram data; cleaned and tidy'd
7. kotzeb0912_timelines
   • hourly percent dry (haul-out) data; cleaned and tidy'd
8. kotzeb0912_deployments
   • bearded seal morphometric and tag deployment data

kotzeb0912_locs

data("kotzeb0912_locs")
dplyr::glimpse(kotzeb0912_locs)

deployid

• An alphanumeric string that uniquely identifies the deployment.

  Since a deployment is a unqiue combination of animal and tag, the deployid is a concatenation of the animal id (speno) and the tag serial number (serialnum).

  By examining the deployid, one can discern several key details about the particular deployment. The first two characters in the deployid will specify the species of animal the tag was deployed on. These letters correspond to the genus and species (e.g. bearded seals (Erignathus barbatus) would be represented by EB). The next four characters represent the year the deployment started. The next section is separated by an underscore and is a unique number assigned to the animal. The first 11 characters in the deployid correspond to the speno for the deployment animal. The next section (also separated by an underscore) corresponds to the serial number for the tag.

ptt

• The Argos transmitter (PTT) identifier.

  It is important to understand that a PTT identifier can be assigned to different tags over the years. The PTT identifier should not be considered a unique identifier.

instr

• An alphanumeric string that identifies the make/model class of the tag

  r unique(kotzeb0912_locs$instr)

  Mk10 ~ Argos location tags with additional sensors for recording depth/dive behavior, temperature, conductivity. For this project, Mk10 tags were adhered to the hair of the seal on the head.

  SPOT ~ Argos location tags with conductivity. No pressure transducer for recording depth or dive behavior. For this project, SPOT tags were attached to the rear flipper.

date_time

• A POSIXct value representing the UTC time for the location estimate.

  r range(kotzeb0912_locs$date_time)

  Due to the nature of Argos data that is still not fully understood, there can be more than one record with the the same date_time value for the same deployment. See unique_posix for an adjusted value with no duplicate times.

type

• An character string that specifies the location estimate is based on the Argos process

  r unique(kotzeb0912_locs$type)

quality

• An alphanumeric value cooresponding to the Argos location quality.

  r unique(kotzeb0912_locs$quality)

  Argos location estimates were traditionally classified with a quality value that provides general guidelines regarding the error associated with the location estimates. From better to worse, the possible values are 3, 2, 1,0,A,B, and Z. Z values should be removed from any analysis and are only included here for completeness.

  In recent years, Argos has provided better estimates of error using their Kalman filter algorithm. See error_radius, error_semimajor_axis, error_semiminor_axis, and error_ellipse_orientation.

latitude, longitude

• Coordinates (decimal degrees, datum=WGS84) of the location estimate

error_radius, error_semimajor_axis, error_semiminor_axis,error_ellipse_orientation

• Error parameters provided as part of the Kalman filter algorithm

unique_posix

• The same as date_time except duplicate values have been eliminated by increasing the value of one duplicate by 1 second.

kotzeb0912_gps

data("kotzeb0912_gps")
dplyr::glimpse(kotzeb0912_gps)

deployid

• An alphanumeric string that uniquely identifies the deployment.

  Since a deployment is a unqiue combination of animal and tag, the deployid is a concatenation of the animal id (speno) and the tag serial number (serialnum).

  By examining the deployid, one can discern several key details about the particular deployment. The first two characters in the deployid will specify the species of animal the tag was deployed on. These letters correspond to the genus and species (e.g. bearded seals (Erignathus barbatus) would be represented by EB). The next four characters represent the year the deployment started. The next section is separated by an underscore and is a unique number assigned to the animal. The first 11 characters in the deployid correspond to the speno for the deployment animal. The next section (also separated by an underscore) corresponds to the serial number for the tag.

count

time_offset

locnumber

hauled_out

• a binary (1,0) indication whether the tag was in haul-out mode at the time the fastloc snapshot was captured

satellites

• total number of satellites captured in the snapshot. substract the number of bad_sats to get the total number of viable satellites

initlat, initlon

• Coordinates (decimal degrees, datum=WGS84) of the seed location used

inittime

• A POSIXct value representing the UTC time for the seed location used

inittype

• the source data for the seed coordinates: GPS, User or an letter corresponding to the Argos satellite

latitude, longitude

• Coordinates (decimal degrees, datum=WGS84) of the location estimate

bad_sats

• number of satellites with poor information in the snapshot and not usable

residual

time_error

date_time

• A POSIXct value representing the UTC time for the fastloc location estimate.

  r range(kotzeb0912_gps$date_time)

kotzeb0912_status

data("kotzeb0912_status")
dplyr::glimpse(kotzeb0912_status)

kotzeb0912_depths

data("kotzeb0912_depths")
dplyr::glimpse(kotzeb0912_depths)

deployid

• An alphanumeric string that uniquely identifies the deployment.

  Since a deployment is a unqiue combination of animal and tag, the deployid is a concatenation of the animal id (speno) and the tag serial number (serialnum).

  By examining the deployid, one can discern several key details about the particular deployment. The first two characters in the deployid will specify the species of animal the tag was deployed on. These letters correspond to the genus and species (e.g. bearded seals (Erignathus barbatus) would be represented by EB). The next four characters represent the year the deployment started. The next section is separated by an underscore and is a unique number assigned to the animal. The first 11 characters in the deployid correspond to the speno for the deployment animal. The next section (also separated by an underscore) corresponds to the serial number for the tag.

datadatetime

• A POSIXct value representing the UTC time for the start of the time bin.

  Dive behavior is summarized into user specified time bins. In this study, 6 hour bins were chosen.

num_dives

• The number of dives with a max depth that occured during the corresponding time period and within the corresponding depth bin. This value is capped at 255. Thus, in cases when the actual number of dives exceeds 255 some scaling occurs.

bin

• The bin label in the original data files provided by Wildlife Computers

limits

• The upper (numerically, not in the water column) limit of the depth bin in meters.

  The limits are determined from examining the data files provided from the Wildife Computers Data Portal. If the user has properly specified the programming schema for this deployment, those values are extracted and included in the data frame for easy reference.

kotzeb0912_durations

data("kotzeb0912_durations")
dplyr::glimpse(kotzeb0912_durations)

deployid

• An alphanumeric string that uniquely identifies the deployment.

  Since a deployment is a unqiue combination of animal and tag, the deployid is a concatenation of the animal id (speno) and the tag serial number (serialnum).

  By examining the deployid, one can discern several key details about the particular deployment. The first two characters in the deployid will specify the species of animal the tag was deployed on. These letters correspond to the genus and species (e.g. bearded seals (Erignathus barbatus) would be represented by EB). The next four characters represent the year the deployment started. The next section is separated by an underscore and is a unique number assigned to the animal. The first 11 characters in the deployid correspond to the speno for the deployment animal. The next section (also separated by an underscore) corresponds to the serial number for the tag.

datadatetime

• A POSIXct value representing the UTC time for the start of the time bin.

  Dive behavior is summarized into user specified time bins. In this study, 6 hour bins were chosen.

num_dives

• The number of dives with a duration (seconds) that occured during the corresponding time period and within the corresponding duration bin. This value is capped at 255. Thus, in cases when the actual number of dives exceeds 255 some scaling occurs.

bin

• The bin label in the original data files provided by Wildlife Computers

limits

• The upper limit of the duration bin in seconds.

  The limits are determined from examining the data files provided from the Wildife Computers Data Portal. If the user has properly specified the programming schema for this deployment, those values are extracted and included in the data frame for easy reference.

kotzeb0912_tad

data("kotzeb0912_tad")
dplyr::glimpse(kotzeb0912_tad)

deployid

• An alphanumeric string that uniquely identifies the deployment.

  Since a deployment is a unqiue combination of animal and tag, the deployid is a concatenation of the animal id (speno) and the tag serial number (serialnum).

  By examining the deployid, one can discern several key details about the particular deployment. The first two characters in the deployid will specify the species of animal the tag was deployed on. These letters correspond to the genus and species (e.g. bearded seals (Erignathus barbatus) would be represented by EB). The next four characters represent the year the deployment started. The next section is separated by an underscore and is a unique number assigned to the animal. The first 11 characters in the deployid correspond to the speno for the deployment animal. The next section (also separated by an underscore) corresponds to the serial number for the tag.

datadatetime

• A POSIXct value representing the UTC time for the start of the time bin.

  Dive behavior is summarized into user specified time bins. In this study, 6 hour bins were chosen.

pct_tad

• Percent time-at-depth represents the percentage of time the tag spent within the corresponding depth bin. Values across the depth bins for a given 6 hour time period should add to approximately 100.

bin

• The bin label in the original data files provided by Wildlife Computers

limits

• The upper (numerically, not in the water column) limit of the dive depth bin in meters

  The limits are determined from examining the data files provided from the Wildife Computers Data Portal. If the user has properly specified the programming schema for this deployment, those values are extracted and included in the data frame for easy reference.

kotzeb0912_timelines

data("kotzeb0912_timelines")
dplyr::glimpse(kotzeb0912_timelines)

deployid

• An alphanumeric string that uniquely identifies the deployment.

  Since a deployment is a unqiue combination of animal and tag, the deployid is a concatenation of the animal id (speno) and the tag serial number (serialnum).

  By examining the deployid, one can discern several key details about the particular deployment. The first two characters in the deployid will specify the species of animal the tag was deployed on. These letters correspond to the genus and species (e.g. bearded seals (Erignathus barbatus) would be represented by EB). The next four characters represent the year the deployment started. The next section is separated by an underscore and is a unique number assigned to the animal. The first 11 characters in the deployid correspond to the speno for the deployment animal. The next section (also separated by an underscore) corresponds to the serial number for the tag.

datadatetime

• A POSIXct value representing the UTC time for the start of the hour bin.

  Percent timeline data is summarized into hourly time bins.

percent_dry

• A percentage of the given hour the tag was dry (out of the water)

  Possible values include: 0,3,5,,10,20,30,40,50,60,70,80, 90,95,97, and 100

kotzeb0912_deployments

data("kotzeb0912_deployments")
dplyr::glimpse(kotzeb0912_deployments)

Publications and Data Availability

The final contract report for this research is available from the Bureau of Ocean and Energy Management

This R data package and associated vignette documents are archived with Zenodo. Please note each release of the R package generates a new, unique, and citable DOI. If you use this package and the data within, please cite the work as described in the DOI link below.

Acknowledgments and Funding

The research described here and the included data were obtained with significant financial contributions from the U.S. Department of Interior's Bureau of Ocean and Energy Management (BOEM) [^4] and the U.S. Department of Commerce's National Oceanic and Atmospheric Administration.

In addition to funding, significant leadership, participation and expertise was provided by the Kotzebue IRA and members of the Kotzebue community.

[^4]: funding administered under the Inter-agency Agreement M07RG13317

jmlondon/kotzeb0912 documentation built on May 19, 2019, 1:54 p.m.