Multi-scale habitat selection and nest survival of whinchats and stonechats: Implications for grassland conservation management
Abstract
Effective management for endangered grassland birds requires a sound understanding of habitat requirements and factors shaping reproductive success. We studied multi-level habitat selection of endangered whinchats (Saxicola rubetra) and locally increasing stonechats (Saxicola rubicola) in a nature reserve using Sentinel-2 images, LiDAR data, and habitat variables measured onsite. We monitored nest fates, and estimated how the causes of nest mortality varied with nest initiation date. Further, we analyzed how nest-site selection, precipitation, nest initiation date, and nest age influenced daily nest survival.
At landscape- and site-level, both species primarily selected for territories in lightly farmed meadows. However, stonechats appeared to be less specialized and used most available habitat types. Within home-ranges, both species selected for nest-sites in fallow patches, but in contrast to stonechats, whinchats also selected for large viewsheds.
In both species, the main causes of nest failure were predation by red foxes (Vulpes vulpes), weather-related, and nest desertion. In whinchats, the probability of nest desertion increased and predation risk decreased with nest initiation date, possibly due to better nest concealment in the later season. Daily nest survival decreased with nest age and precipitation in both species, and increased with nest initiation date in whinchats. None of the habitat variables examined were positively linked to daily nest survival, and viewshed was negatively related to whinchat daily nest survival.
Our study suggests that extending core habitat may help to promote settlement and potentially reproductive success of whinchats. However, as habitat selection was not positively associated with daily nest survival, land management measures alone might not be sufficient to halt rapid population declines. To effectively conserve whinchats and other grassland bird populations, the causes of nest failures must be identified and addressed. Within the examined reserve and similar habitats, targeted flood and predator control may yield the biggest short-term conservation gains.
README: Multi-scale habitat selection and nest survival of whinchats and stonechats: implications for grassland conservation management
This dataset contains the data used for our analysis of multi-scale habitat selection and nest survival in whinchats and stonechats. In this paper, we analyzed territory selection of whinchats and stonechats at landscape- and site-level, and nest-site selection at home-range-level. We monitored nest fates, and analyzed how the causes of nest mortality varied with nest initiation date. Further, we estimated how environmental and temporal factors influenced daily nest survival.
The dataset is organized in five folders following the structure of the paper: Territory selection at reserve-level, territory selection at site-level, nest-site selection at home-range level, causes of nest mortality, and daily nest survival. Each folder contains multiple files with the data used for the respective analysis.
The folder “habitat_selection_reserve_level”
contains four files with the data used to analyze territory selection of whinchats and stonechats at reserve-level. Each table contains percentages of 11 remotely sensed land use/cover classes (representing either use or availability of these classes). The column names are:
o.bog: open bog, not overgrown by pines
w.bog: wooded bog, overgrown by pines
t.bog: transitional bog, in the transition stage from fen to raised bog
built-up: e.g. roads, buildings, etc.,
light: lightly farmed meadows, i.e. mown only once a year not before 1 August or 1 September
moderate: moderately farmed meadows, i.e. mown once or twice a year not before 1 June, 15 June or 1 July
fallow: fallow meadows, not mown for a year or longer
forest: broadleaf and coniferous forests
intensive: intensively farmed meadows, mown up to 6 times/year without any time restrictions
reed: reed beds, dense stands of Phragmites sp.
water: waterbodies such as rivers and lakes
avail_whin_reserve.csv
This file contains the percentages of 11 remotely sensed land use/cover classes (as defined above) within the nature reserve, representing available areas. To run a compositional analysis via ‘compana’ from the adehabitat package, these percentages are repeated for the number of whinchat territories.
used_whin_reserve.csv
This file contains the percentages of 11 land use/cover classes (as described above) in each of 66 whinchat territories.
avail_stone_reserve.csv
This file contains the percentages of 11 remotely sensed land use/cover classes (as defined above) within the nature reserve, representing available areas. To run a compositional analysis via ‘compana’ from the adehabitat package, these percentages are repeated for the number of stonechat territories.
used_stone_reserve.csv
This file contains the percentages of 11 land use/cover classes (as described above) used in each of 238 stonechat territories.
The folder “habitat_selection_site_level”
contains eight files with the data used to analyze territory selection of whinchats and stonechats at site-level. Each table contains percentages of up to seven land use/cover classes assessed onsite and from aerial pictures. The percentages represent either use or availability. The column names are:
light: lightly farmed meadows, mown once a year not before 1 August or 1 September, unfertilized
moderate: mown either once or twice a year not before 15 June or 1 July, infrequently fertilized with horse dung
intensive: mown 4-6 times a year, frequently fertilized with liquid manure
pasture: cattle pastures
reed: dense stands of Phragmites sp.
shrub: lower shrubs e.g. Pinus mugo or Salix sp.
forest: broadleaf and coniferous forests, groups of large trees
Note that marginally used and unused classes were removed from the respective sample to get an unbiased ranking of selected land use/cover classes. The percentages are repeated for the number of whinchat territories.
avail_whin_A.csv
Percentages of 7 land use/cover types within site A (defined as available).
used_whin_A.csv
Percentages of 7 land use/cover classes used in each of 48 whinchat territories at site A.
avail_stone_A.csv
Percentages of 7 land use/cover types within site A (defined as available).
used_stone_A.csv
Percentages of 7 land use/cover classes used in each of 17 stonechat territories at site A.
avail_whin_B.csv
Percentages of 5 land use/cover types within site B (defined as available).
used_whin_B.csv
Percentages of 5 land use/cover classes used in each of 43 whinchat territories at site B.
avail_stone_B.csv
Percentages of 7 land use/cover types within site B (defined as available).
used_stone_B.csv
Percentages of 7 land use/cover classes used in each of 48 stonechat territories at site B.
The folder “habitat_selection_nest_site”
contains four files used to analyze nest-site selection of whinchats and stonechats at home-range-level. The files contain measurements of habitat variables at nest-sites (presence) and each ten control points distributed randomly in 240 m radius home-range (absence) for whinchats at site A. The column names are:
viewshed: the area visible from a nest location (in m², see the paper for a detailed description of the settings)
viewshed.z: scaled viewshed (as described in the paper)
dist_int: distance to intensively farmed meadows (in meter).
dist_int.z: scaled dist_int (as described in the paper)
fallow: 0/1 denotes whether a nest was found inside or outside a fallow (unmown area)
no: unique nest ID
PA: Presence/Absence (1 means nest, 0 means random point)
Note that in the columns dist_int and dist_int_z are only present in the files concerning site A, and the column fallow is only present in the files concerning site B.
In addition, the folder contains a file with the orientation of whinchat nests and stonechat nests.
data_A_whin.csv
This file contains measurements of habitat variables at nest-sites (presence) and each ten control points distributed randomly in 240 m radius home-range (absence) for whinchats at site A.
data_B_whin.csv
This file contains measurements of habitat variables at nest-sites (presence) and each ten control points distributed randomly in 240 m radius home-range (absence) for whinchats at site B.
data_A_stone.csv
This file contains measurements of habitat variables at nest-sites (presence) and each ten control points distributed randomly in 240 m radius home-range (absence) for stonechats at site A.
data_B_stone.csv
This file contains measurements of habitat variables at nest-sites (presence) and each ten control points distributed randomly in 240 m radius home-range (absence) for stonechats at site B.
orientation.csv
This file contains the orientation of 81 whinchat nests and 40 stonechat nests with a recognizable lateral entrance. The column names are:
no: unique nest ID
degree: orientation of the lateral nest opening in degree
Species: species, whinchat or stonechat
The folder “mortality”
contains one file with the data used to analyze how the causes of mortality for whinchat nests varied with nest initiation date.
nest_fates.csv
This file contains the fate (success, or failure due to predation, desertion, weather) of 114 whinchat nests and the day of the year (yday) of nest initiation. The column names are:
fate: the fate of the nest (success, or failure due to predation, desertion, weather)
yday: day of the year when the nest was initiated
The folder “survival”
contains six files with the data used to analyze daily nest survival of whinchats and stonechats with the overall models and with the site-specific models. The .RDS-file can be read in R and contains a list of 11 and 12 elements for the overall and the site-specific models, respectively. The element names are:
y: survival matrix with individual nests in rows and days in columns (1 = nest is alive, 0 = nest is dead)
NNests: number of nests
last: last observation (alive or dead)
first: first observation (alive or dead)
cover: viewshed (in m², a covariate of the nest)
distance: distance (in meter) to the nearest intensively farmed meadow. As intensively farmed meadows were only present at site A, this column is only present in the site-specific datasets for site A.
fallow: indicating whether a nest was inside or outside a fallow. As fallows were created only at site B, this column is only present in the site-specific datasets for site B.
init: nest initiation day (day of the year when the nest was initiated, a covariate of the nest)
age: nest age (in days, a covariate of the date)
maxage: maximum of age
rain: a matrix containing the amount of rain (in mm) for each day (in columns) of each nest (in rows).
parentID: a vector with mother ID (as integer)
Nparents: number of unique mother IDs
Numeric covariates were scaled and centered.
datax_whin
This file contains all data used for the overall model of daily nest survival in whinchats. The .RDS-file can be read in R and contains a list of 11 elements (element names are described above).
datax_stone
This file contains all data used for the overall model of daily nest survival in stonechats. The .RDS-file can be read in R and contains a list of 11 elements (element names are described above).
datax_whin_A
This file contains all data used for the site-specific model of daily nest survival in whinchats at site A. The .RDS-file can be read in R and contains a list with 12 elements (element names are described above).
datax_stone_A
This file contains all data used for the site-specific model of daily nest survival in stonechats at site A. The .RDS-file can be read in R and contains a list with 12 elements (element names are described above).
datax_whin_B
This file contains all data used for the site-specific model of daily nest survival in whinchats at site B. The .RDS-file can be read in R and contains a list with 12 elements (element names are described above).
datax_stone_B
This file contains all data used for the site-specific model of daily nest survival in stonechats at site B. The .RDS-file can be read in R and contains a list with 12 elements (element names are described above).
Sharing/access information
Satellite images were downloaded from the Copernicus Open Access Hub (https://scihub.copernicus.eu/), flood data and the digital surface model from Landesamt für Digitalisierung, Breitband und Vermessung (https://www.ldbv.bayern.de/), and weather data from meteorological stations run by Deutscher Wetterdienst (https://www.wetter-by.de/).
Code/software
Spatial analyses were conducted in QGIS, and all models were run in R and Stan. Code is available upon request. RDS files can be read in R using the readRDS().
Methods
This dataset contains the data for our analysis of multi-scale habitat selection and nest survival in whinchats and stonechats. In this paper, we analyzed territory selection of whinchats and stonechats at landscape- and site-level using compositional analysis, and nest-site selection at home-range-level using conditional logistic regression. We monitored the nest fates and analyzed how the causes of nest mortality varied with nest initiation date using a multinomial model. Further, we used known fate models to estimate how environmental and temporal factors influenced daily nest survival.