Data from: Digging deeper: habitat selection within the home ranges of a threatened marsupial
Cite this dataset
Cornelsen, Kate; Elphinstone, Andrew; Jordan, Neil (2024). Data from: Digging deeper: habitat selection within the home ranges of a threatened marsupial [Dataset]. Dryad. https://doi.org/10.5061/dryad.s1rn8pk9w
Abstract
While resource selection varies according to the scale and context of study, gathering data representative of multiple scales and contexts can be challenging especially when a species is small, elusive, and threatened. We explore resource selection in a small, nocturnal, threatened species—the greater bilby (Macrotis lagotis)—to test (a) which resources best predict bilby occupancy, and (b) whether responses are sex-specific and/or vary over time. We tracked a total of 20 bilbies and examined within home range resource selection over multiple seasons in a large (110ha) fenced sanctuary in temperate Australia. We tested a set of plausible models for bilby resource selection, showing that food biomass (terrestrial and subterranean invertebrates, and subterranean plants) and soil textures (% sand, clay and silt) best predicted bilby resource selection for all sampling periods. Selection was also sex-specific; female resource use, relative to males, was more closely linked to the location of high-quality resources (sandier soils, and terrestrial invertebrate biomass). Bilby selection for roads was independent of season but varied over time with females selecting for areas closer to roads when plants increased in density off roads. Our findings demonstrate the importance of considering resource selection over multiple contexts and highlight a method to collect such data on a difficult to study, threatened species. Collecting such data is critical to understanding the habitat required by species.
README: Data from: Digging deeper: habitat selection within the home ranges of a threatened marsupial
https://doi.org/10.5061/dryad.s1rn8pk9w
This README file lists the supplementary datasets, shapefiles, rasters, and R code used in the associated paper. It also provides a brief description of how each file was used in generating the results, and definitions for any abbreviated variables within datasets. Note: dates are all in the format ‘date-month-year’.
Description of the data and file structure
Shapefiles.zip (and auxiliary files):
Sanctuary fenceline.shp
- Sanctuary fenceline.qpj
- Sanctuary fenceline.prj
- Sanctuary fenceline.dbf
- Sanctuary fenceline.cpg
- Sanctuary fenceline.shx
water sources.shp
- water sources.shx
- water sources.prj
- water sources.dbf
- water sources.cpg
Roads.shp
- Roads.shx
- Roads.prj
- Roads.dbf
- Roads.cpg
Individual Minimum Convex Polygons (MCPs):
Minimum convex polygons (MCPs) were generated for individual bilbies tracked in each season using movement data collected once during the day while resting (to a burrow), and several locations, while active at night. There are two ‘MCP’ files per bilby, per season. One shapefile is the raw polygon (Bilby Name.shp) and the other is the polygon with unavailable areas removed (i.e. actual area available) such as buildings and enclosed areas (Bilby Name_MCP_Available.shp). Parentheses indicate the number of bilby MCP shapefiles in each folder (excluding auxiliary files). The shapefiles are used to calculate MCP area in the associated paper and can be used in conjunction with the ‘third-order_RSF.R’ script. Metadata for bilbies can be found in the bilby_metadata.csv file.
Individual MCPs.zip
- MCPs_autumn2020 (30)
- MCPs_autumn2021 (16)
- MCPs_spring (18)
- MCPs_summer2020 (28)
- MCPs_summer2021 (21)
- MCPs_winter (20)
Rasters:
These include processed (i.e. clipped) soil texture data originally obtained from the SEED portal (State of New South Wales, Office of Environment and Heritage, 2018). Raw soil texture data has been clipped to the sanctuary extent by creating a bounding box around the sanctuary in QGIS. Seasonal food biomass rasters generated from the ‘Interpolation.R’ scripts are also provided. Rasters need to be in your working directory to run the ‘Third-order_RSF.R’ script.
Note: all shapefiles and rasters are projected to GDA94 / MGA zone 55 (EPSG:28355; https://epsg.io/28355) and units are in meters.
Rasters.zip
Soil type
- soil_silt_0_30_clipped.tif
- soil_sand_0_30_clipped.tif
- soil_clay_0_30_clipped.tif
Terrestrial invertebrate biomass (for each season)
- raster_Sum21.tif
- raster_Sum21.tfw
- raster_Sum20.tif
- raster_Sum20.tfw
- raster_Spr20.tif
- raster_Spr20.tfw
- raster_Win20.tif
- raster_Win20.tfw
Subterranean invertebrate biomass (for each season)
- raster_Sum21_soil.tif
- raster_Sum21_soil.tfw
- raster_Sum20_soil.tif
- raster_Sum20_soil.tfw
- raster_Spr20_soil.tif
- raster_Spr20_soil.tfw
- raster_Win20_soil.tif
- raster_Win20_soil.tfw
Subterranean plant biomass (for each season)
- raster_Sum21_plants.tif
- raster_Sum21_plants.tfw
- raster_Sum20_plants.tif
- raster_Sum20_plants.tfw
- raster_Spr20_plants.tif
- raster_Spr20_plants.tfw
- raster_Win20_plants.tif
- raster_Win20_plants.tfw
Data.zip
bilby_metadata.csv
Seasonal home range area (in m2 and hectares) determined from the 100% minimum convex polygon (MCP) around all locations collected for each individual bilby (indicated under ‘bilby_ID’) in summer, spring, winter and autumn. Number of observations included in the calculation of 100% MCPs is indicated under the ‘days_obs’ column. Whether the bilby was born within the sanctuary (W) or was a founder released to the sanctuary (C) and whether they are male (M) or female (F) is also provided.
biomass.csv
Food biomass (g/m2) data obtained from the sanctuary over four seasons and 16 sampling sites. Biomass was calculated by dividing the total mass (g) of food collected within a site and season by the sample area (m2). For assessing terrestrial invertebrate biomass (Type = “TER”), invertebrates were captured in five pitfall traps, within a 5 m2 sample area, over seven consecutive nights each season. For assessing subterranean invertebrates and plant material (seed, roots, and bulbs) biomass (Type = “SOIL”), soil cores were removed from a 1 m2 quadrat in five randomly selected locations for each site and season. More detail on the methods used for sampling can be found in the supplementary information and below (Table 1; Figure 2 in the main text of the associated paper). This data can be used to generate the interpolation rasters, and to replicate biomass comparisons described in the supplementary information to the associated paper.
Table 1: A description of the variables used and their units of measurement (if applicable).
| Variable (as stated in spreadsheet) | Units of measurement | Description |
|---|---|---|
| Date Collected | DD/MM/YYYY | Date that the sample was obtained in the field. |
| Site | Unique site ID for each of the 16 sample sites. Each site corresponds to a unique Lat/Long. | |
| Lat | Latitude of the Site. | |
| Long | Longitude of the Site. | |
| Sample | Number indicates the order that each sample was taken from the sites (first to last; 1 to 5). | |
| Type | How sample was obtained (SOIL = biomass obtained from a soil core, TER = biomass obtained from terrestrial pitfall trapping). | |
| Season | Season and year each sample was taken (SUM21 = summer 2021, SUM20 = summer 2020, WIN = winter 2020, SPR = spring 2020). Note: no biomass sample was taken in autumn. | |
| Sample area | m2 | Area that the corresponding sample represents, either 1m2 for the soil core or 5m2 for the terrestrial pitfall trapping. |
| Plant mass | g | Mass of plant material (i.e. roots and woody debris) in the sample. |
| Invert mass | g | Mass of invertebrate material (entire or parts of) in the sample. |
| Seed mass | g | Mass of seeds in the sample. |
| Bulb mass | g | Mass of plant bulbs in the sample. |
| Total mass | g | Mass of all edible material (plants, seeds, bulbs and inverts) in the sample. |
| Invert bio | g/m2 | Biomass of invertebrate material adjusted for the sample area. As the soil core samples represent 1m2, ‘invert mass’ is equal to ‘invert bio’ for soil core samples. However, as the terrestrial pitfall samples represent 5m2, total mass is greater than ‘invert bio’. |
| Hab Type | Generalised habitat type classification for each of the sample sites. Either G = grassland, C = Callitris glaucophylla woodland, ECO = Ecotone (interface between two habitat types), and A = Acacia deanei shrubland (see Figure 1). |
weather.csv
Daily rainfall (mm) and maximum temperatures (°C) over the two years of this study (2020–2021) and drought year prior (2019). Data can be used to generate Figure A1 presented in the supplementary information.
Coefficient-Est-Results.csv
Results from the generalised linear mixed model for each sampling period including a unique number for each season/sampling period (seasonID) in ascending order from the oldest to most recent period, the predictors specified in the model (see Table 2 for abbreviations used), the coefficient estimate (estimate) for each predictor, the standard error (se) of the estimate, the lower 95% confidence interval (lcl), and upper 95% confidence level (ucl) of the estimate, the reference level for the model (sex), the p-value, and whether there was a significant result (signif) as either a yes (Y) or no (N). Where sex = F the reference level was female for the corresponding model. Where sex = M the reference level was male for the corresponding model.
Table 2: Predictor abbreviations and their meaning.
| Variable (as stated in spreadsheet) | Units of measurement | Description |
|---|---|---|
| Intercept | Intercept of the model | |
| TIB | g/m2 | Terrestrial invertebrate biomass |
| SIB | g/m2 | Subterranean invertebrate biomass |
| SPB | g/m2 | Subterranean plant biomass |
| %Sand | Percent of sand in the soil | |
| %Clay | Percent of clay in the soil | |
| %Silt | Percent of silt in the soil | |
| DTR | m | Shortest Euclidean distance to roads |
| DTW | m | Shortest Euclidean distance to a water source |
| Sex(male) | M = male, F = female | Sex of the bilby (either male or female). |
Note: a ‘*’ between two predictors indicates an interaction term.
Sharing/Access information
Soil texture data was derived from the following source:
State of New South Wales and Office of Environment and Heritage. 2018. Digital soil mapping of key soil properties over New South Wales. Version 1.2. https://datasets.seed.nsw.gov.au/dataset/digital-soil-maps-for-key-soil-properties-over-nsw. Office of Environment and Heritage, Sydney, Australia.
Weather data was derived from the following source:
Bureau of Meteorology (Australia). 2022. Climate data online. http://www.bom.gov.au/climate/averages/tables/cw_065070.shtml. Accessed 19 May 2022.
Code/Software
R code.zip
Biomass_comp.R
Annotated R script is provided to generate Figure A5 presented in the supplementary information. The code also contains methods for food biomass comparisons described in Appendix of the main paper. For the code to run, you will need to have the ‘biomass.csv’ file in your working directory.
Interpolation.R
Annotated R script is provided to replicate the food biomass interpolation rasters. You will need to have the ‘biomass.csv’ file in your working directory for the code to run.
Weather.R
Annotated R script is provided to generate Figure A1 presented in the supplementary information. For the code to run, you will need to have the ‘weather.csv’ file in your working directory.
Coefficient-Est-Plots.R
Annotated R script is provided to generate Figure 4 presented in the main text. For the code to run, you will need to have the ‘Coefficient-Est-Results.csv’ file in your working directory.
Third-order_RSF.R*
Annotated R script is provided for third-order (i.e. within home range) resource selection analysis presented in the associated paper. For the code to run, you will need the shapefiles and animal movement data in your working directory.
Full model.R (file for each sampling period: Sum21, Sum 20, Aut20, Aut21, Spring, Win)*
Annotated R script to replicate the model selection steps described in the associated paper. For the code to run, you will need the output generated from the Third-order_RSF.R script in your working directory for the corresponding season.
*Scripts require animal movement data to run. GPS/VHF radio-tracking data from the associated paper can be requested from the corresponding author (Kate Cornelsen: cornelsen.kate@gmail.com) upon reasonable request.
Methods
Data includes terrestrial invertebrate, and subterranean invertebrate and plant biomass data collected over four seasons (summer 2020, winter 2020, spring 2020, and summer 2021). We have also provided the R code used for generating the interpolation rasters and the necessary shapefiles (generated in QGIS) and rasters for this interpolation. The R code used for construction of resource selection functions and for generating 'Figure 4' in the main paper is also provided. The GPS data associated with this analysis can be requested from the corresponding author upon reasonable request.
Funding
Wildlife Preservation Society of Australia
Taronga Conservation Society Australia
Centre for Ecosystem Science, University of New South Wales