README: Wind energy development can lead to guild-specific habitat loss in boreal forest bats

https://doi.org/10.5061/dryad.x0k6djhrp

This data was collected in between July and October 2020 at Marker Wind Park in Ørje, Norway.

Bat acoustic monitoring took place from 1 July to 29 September 2020 at seven wind turbines. Wildlife Acoustics Song-Meter4-BATFS detectors combined with either SMM-U2 or SMM-U1 microphones were deployed at ground-level (microphone approximately 2 m high).

Images were taken with a a digital camera (Ricoh WG-6 Waterproof 20m/65.6ft; Model R02050 2019) oriented skyward attached to an external battery supply which made it possible for the camera to take images every 10 minutes continuously.

A more detailed description of the methods can be found in the research article associated with DOI: 10.1002/wlb3.01168.

NOTE: The weather data used in this study is the property of Marker Wind Park and therefor cannot be shared openly. The datasets in this workflow which include weather data are not available to share but public weather data can be used instead.

Description of the data and file structure

The following datasets are described in the order that they are used in the workflow documented in the GitHub repository: GitHub repository: https://github.com/airmckay/Bats-Insects-BorealForest-WindPark.

**Datasets **

__________________________________________________________________________________________________________

1. cleaed_manual.output_27.11.2023.csv

• General description

All bat acoustic files that were manually identified as well as those that the automatic classification software identified as noise files (which were not manually checked). Each row is a unique, approximately 5-15 second WAV file acoustic recording collected from a Wildlife Acoustics SM4-BATFS ultrasonic acoustic detector.

• R Markdown(s) where the dataset appears within the GitHub repository

1. Cleaning Bat Acoustic Data.Rmd
https://github.com/airmckay/Bats-Insects-BorealForest-WindPark/blob/main/1.%20Cleaning%20Bat%20Acoustic%20Data.Rmd

• Description of columns/data contents:

The columns correspond to the id.csv format which the Wildlife Acoustics Kaleidoscope Pro software produces when automatically classifying bat acoustic recordings. The following column descriptions are from the Kaleidoscope Pro version 5 user manual on page 92 (http://condor.wildlifeacoustics.com/Kaleidoscope.pdf) with some minor edits made by the author of this dataset for further clarification:

The id.csv file has a row for each detected signal, and the following columns:
INDIR Absolute path to input directory.
OUTDIR Absolute path to output directory.
IN FILE Input file name.
DURATION Duration in seconds of the output file.
OUT FILE FS Name of the output file (full-spectrum).
OUT FILE ZC Name of output file (zero-crossing).
DATE Date in form YYYY-MM-DD of the recording.
TIME Time in form of hh:mm:ss of the recording.
HOUR Hour of the recording (0-23) for convenient pivot tables by hour.
DATE-12 Date 12 hours prior to date of recording (e.g. for night vs. day) in the form YYYY-MM- DD.
TIME-12 Time 12 hours prior to time of recording (e.g. for night vs. day) in the form hh:mm:ss.
HOUR-12 Hour 12 hours prior to time of recording (e.g. for night vs. day).
AUTO ID Automatic classification result.
If NA - then relevant data is instead stored in "AUTO.ID" or was not classified.
AUTO.ID The same as 'AUTO ID' but software formatted the column name differently for some portions of the data. .
If NA - then relevant data is instead stored in "AUTO ID" or was not classified.
PULSES Number of pulses detected in the file that were identified to species.
If NA - then no pulses were detected - a noise file or faint recording
MATCHING Number of pulses matching the auto classification result.
If NA - then no pulses were detected to match - a noise file or faint recording.
MATCH RATIO The ratio of MATCHING over PULSES.
If NA - then no pulses were detected to match - a noise file or faint recording.
MARGIN Classification margin - this is an uncalibrated confidence score and should not be subject to much interpretation other than that within a given species, higher values are more confident than lower values.
If NA - then no pulses were detected to match/assign confidence - a noise file or faint recording.
ALTERNATE 1 First alternate.
If NA - then no alternate automatic classification ID was detected.
ALTERNATE 2 Second alternate. In addition to the species identification, these fields list zero or more alternate species identifications separated by semicolons based on other pulse-level classifications detected in the file ranked from highest probability to lowest probability. This might suggest an alternate identification or multiple bats present.
If NA - then no alternate automatic classification ID was detected.
N Total number of pulses detected. This is used to derive average values for the following 12 parameters

(If NA - then no pulses were detected - a noise file or faint recording):

Fc Average characteristic frequency (kHz) - the body of the call is the portion of the call consisting of the flattest slope where the characteristic frequency is typically the frequency at the latest part of the call body.
Sc Average characteristic slope (Octaves per Second) - this is the slope of the body of the call. Positive values correspond to decreasing frequency while negative values correspond to increasing frequency.
Dur Average duration (ms) - this is the duration of the call. o Fmax Average maximum frequency (kHz) - the maximum frequency detected in the call.
Fmin Average minimum frequency (kHz) - the minimum frequency detected in the call.
Fmean Average mean frequency (kHz) - the time-weighted mean frequency of the call. Kaleidoscope Pro 5 User Guide Wildlife Acoustics, Inc. 94
TBC Average time between calls (ms) - if N above is greater than one, this is the average period of the calls from the start of one call to the start of the next.
Fk Average frequency of the knee (kHz) - the frequency at the beginning of the call body.
Tk Average time to the knee (ms) - the time from the beginning of the call to the beginning of the call body.
S1 Average initial slope (octaves per second) - the initial slope of the call.
Tc Average time to the characteristic (ms) - the time from the beginning of the call to the end of the call body.
Qual Average call quality (%) - a measure of the smoothness of the call where smaller values indicate a smoother call.

FILES The number 1, indicating one file, as a convenience for pivot tables by file count.
MANUAL ID Manual identification (this field populated during review with Viewer).
If NA - then relevant data is instead stored in "MANUAL.ID" or was not classified.
MANUAL.ID The same as 'MANUAL ID' but software formatted the column name differently for some portions of the data.

If NA - then relevant data is instead stored in "MANUAL.ID" or was not classified.

---

2. Marker_night_inventory_0704.2022_edit.csv

• General description

The nights in which a bat acoustic detector was active at a given site.

• R Markdown(s) where the dataset appears within the GitHub repository

2. Aggregate bat data to night with weather.Rmd
https://github.com/airmckay/Bats-Insects-BorealForest-WindPark/blob/main/2.%20Aggregate%20bat%20data%20to%20night%20with%20weather.Rmd

3. Aggregate bat data to hour with weather.Rmd
https://github.com/airmckay/Bats-Insects-BorealForest-WindPark/blob/main/3.%20Aggregate%20bat%20data%20to%20hour%20with%20weather.Rmd

4. Insect - Bat models and figures .Rmd
https://github.com/airmckay/Bats-Insects-BorealForest-WindPark/blob/main/4.%20Insect%20-%20Bat%20models%20and%20figures%20.Rmd

• Description of columns/data contents:

Site: The first letter corresponds to rather or not the detector location was a ground level natural site ("C"), a ground level turbine pad location ("P") or a meteorological tower location (MetA, MetB). For ground level detector locations, the number following the first letter corresponds to the number ID of the closest turbine.

night: DD.MM.YYYY

active.night: Indicates that the detector was actively recording at a given site that night. All rows are "TRUE" - nights when the detector were not active are omitted.

__________________________________________________________________________________________________________
3. raw insect data tidied_27112023.csv

• General description

Each row corresponds to an image from a camera trap that was manually analyzed to count the number of flying insects in the photo.

• R Markdown(s) where the dataset appears within the GitHub repository

3. Aggregate bat data to hour with weather.Rmd
https://github.com/airmckay/Bats-Insects-BorealForest-WindPark/blob/main/3.%20Aggregate%20bat%20data%20to%20hour%20with%20weather.Rmd

• Description of columns/data contents:

Date: DD.MM.YYYY

Time: HH:MM:SS

Hour: Integers between 0-23 where 0 is midnight and 23 is 23:00.

Count: Integer describing the number of insects manually counted in an image

InsectPresAbs: Binary (1 or 0) where 1 is an image where insects were counted and 0 is an image where insects were not found.

Site: The first letter corresponds to rather or not the camera trap site was a natural site ("C") or a turbine pad location ("P"). The number following the first letter corresponds to the number ID of the closest turbine (08 or 11).

Viable: Binary (Yes or No) describing the quality of the image and therefore the ability of the analyst to count the number of insects in the image. Yes = image was clear enough the detection was plausible, No = poor quality image that was difficult to analyze.

_________________________________________________________________________________________________________
**4. Turbines(3).kml**

• General description

Spatial data of where the turbines and the meteorological data were located in the wind farm.

• R Markdown(s) where the dataset appears within the GitHub repository

6. Site map.Rmd
https://github.com/airmckay/Bats-Insects-BorealForest-WindPark/blob/main/6.%20Site%20map.Rmd

• Description of columns/data contents:

A KML file spatial data attached describing the locations of the 15 turbines that were on the wind farm in the autumn and summer in 2020; easy to read into GoogleEarth, QGIS and other spatial software.

Sharing/Access information

Acoustic files and other files associated with these datasets are stored on Norwegian University of Life Sciences research network drives and plan to be archived. We welcome requests to use the data; please contact Reed April McKay (she/her) by emailing reed.april.mckay@nmbu.no for arranging access.

Code/Software

All code for preparing these datasets as well as analyzing them to create the results and figures described in the research article associated with the DOI: 10.1002/wlb3.01168 are available at the GitHub repository: https://github.com/airmckay/Bats-Insects-BorealForest-WindPark.