Page 1
The European Breeding Bird Atlas
Combining environmental variables and distribution models for the
new European Breeding Bird Atlas
Pietro Milanesi PhD in Biodiversity and Evolution; Doctor Europaeus
Postdoctoral Researcher [email protected]
on behalf of
EBBA2 Steering Committee and EBBA2 coordination team and assistants
EBBA2 European Breeding Bird Atlas
Page 2
EBBA2 European Breeding Bird Atlas
The first EBCC atlas of European
Breeding Birds (EBBA1) was a
milestone in European ornithology,
and after 30 years fieldwork for
EBBA2 is underway;
EBBA1
Data collection 1980s Published 1997
Page 3
EBBA2 European Breeding Bird Atlas
Why a new Atlas? 1. Environmental changes (land use and climate) over the last 30 years
impacted on populations of birds across Europe;
HILDA v2.0 gross land changes European Environment Agency
Page 4
EBBA2 European Breeding Bird Atlas
Why a new Atlas? 1. Environmental changes (land use and climate) over the last 30 years
impacted on populations of birds across Europe; 2. Conservation actions require the most up-to-date information on
these impacts;
CEEweb for Biodiversity
Page 5
EBBA2 European Breeding Bird Atlas
Why a new Atlas? 1. Environmental changes (land use and climate) over the last 30 years
impacted on populations of birds across Europe; 2. Conservation actions require the most up-to-date information on
these impacts; 3. Bird occurrence and its change since the 1980s is fundamental for
decision making;
Northern Wheatear
Oenanthe oenanthe
Page 6
Grasshopper Warbler
Locustella naevia
1. Environmental changes (land use and climate) over the last 30 years impacted on populations of birds across Europe;
2. Conservation actions require the most up-to-date information on these impacts;
3. Bird occurrence and its change since the 1980s is fundamental for decision making;
4. To provide valuable data for science (e.g., future changes);
EBBA2 European Breeding Bird Atlas
Why a new Atlas?
Huntley et al., 2008
Page 7
Area of occurrence Network of protected areas
Area of occurrence Red List assessments
Niche-based models Ecological studies
EBBA2 European Breeding Bird Atlas
Some potential uses of this type of information
Gradients of occurrence Population studies
Etc.
Page 8
EBBA2 European Breeding Bird Atlas
What outputs are planned?
1. An atlas book, to be published in 2020;
2. Interactive maps and further information online;
3. Summaries to highlight the most crucial findings;
4. Bird database will be available for further scientific research.
EBBA1
Page 9
Peregrine falcon
Falco peregrinus
EBBA2 European Breeding Bird Atlas
EBBA1
Page 10
EBBA2 European Breeding Bird Atlas
Spatial resolution: 10x10 km.
The EBBA2 will cover
• 10,180,000 km2
• 50 Countries
• 5 years of fieldwork (2013-2017)
Mapping species distribution in EBBA2
Page 11
EBBA2 European Breeding Bird Atlas
Why 10x10 km maps?
50x50 km
(c. 5,000 sq)
10x10 km
(c. 120,000 sq)
Northern Wheatear
Oenanthe oenanthe
EBBA1 EBBA2
Page 12
EBBA2 European Breeding Bird Atlas
Why 10x10 km maps?
1. To determine finer-resolution species distribution patterns
2. To visualise relative prob. of occurrence – a proxy for relative abundance
3. To infer information in not surveyed areas
4. To standardise information across the whole Europe (if predictive accuracy is high)
Northern Wheatear
Oenanthe oenanthe
Page 13
How it works
EBBA2 European Breeding Bird Atlas
Species
distribution models
Predictor variables
Validations
Bird data
Sampling effort
Page 14
EBBA2 European Breeding Bird Atlas
Predictor variables
Predictor variables
Page 15
EBBA2 European Breeding Bird Atlas
Predictor variables (N=39)
GEOGRAPHY
TOPOGRAPHY
LAND-COVER
ANTHROPOGENIC FACTOR
SOIL TYPES
CLIMATE
Variable Units
Longitude (center of 10x10 cell) º
Latitude (center of 10x10 cell) º
Distance to the coastline m
Mean Annual Temperature ºC
Mean Temperature in the Breeding Period (April-July) ºC
Total Annual Precipitation mm
Total Precipitation in the Breeding Period (April-July) mm
Minimum temperature of the coldest month ºC
Maximum temperature of the warmest month ºC
Evapotranspiration in the Breeding Period (April-July) Ae / Pe
Mean elevation m
Mean slope %
Coniferous forests %
Bare areas %
Broadleaved forests %
Irrigated crops %
Rainfed tree crops %
Rainfed cropland %
Grassland %
Mixed Broadleaved and coniferous forests %
Mosaic cropland-natural vegetation %
Wetlands %
Permanent ice %
Shrubland %
Sparse vegetation %
Continental water bodies %
Urban areas %
Shannon habitat diversity Index H’ = −Σ ( pi × lnpi )
Accumulated NDVI in the Breeding Period (April-July) 0-500
Average forest canopy height meters
Wood biomass Mg / ha
Human population density number / km2
Young soils, weakly developed %
Well developed soils %
Well developed and acid soils %
Wet soils %
Soils rich in clay %
Saline soils %
Shannon soil diversity Index H’ = −Σ ( pi × lnpi )
HABITAT STRUCTURE
Page 16
EBBA2 European Breeding Bird Atlas
Example of predictor variables
Shannon habitat
diversity
Evapotranspiration
Human population
density
Acid soil
Altitude Coniferous
forests
Page 17
EBBA2 European Breeding Bird Atlas
Example of predictor variables
Average forest canopy height (LiDAR)
Wood biomass
Page 18
EBBA2 European Breeding Bird Atlas
Sources of predictor variables
- Worldclim dataset (http://www.worldclim.org);
- MODIS Evapotranspiration MOD16 global dataset (http://www.ntsg.umt.edu/project/mod16);
- Altitude and Slope from the ETOPO2 global dataset
(http://www.ngdc.noaa.gov/mgg/global/etopo2.html);
- Harmonized World Soil Database - HWSD (http://www.fao.org/soils-portal/soil-survey/soil-maps-
and-databases/harmonized-world-soil-database-v12/);
- land cover from GLOBCOVER dataset (http://maps.elie.ucl.ac.be/CCI/viewer/download.php);
- NDVI from MODIS
(https://lpdaac.usgs.gov/dataset_discovery/modis/modis_products_table/mod13c2);
- Forest canopy height from GLAS (http://lidarradar.jpl.nasa.gov);
- Wood biomass from Wageninge UR (http://www.wageningenur.nl/grsbiomass);
- Human population density from the Gridded Population of the World, GPW datasets
(http://beta.sedac.ciesin.columbia.edu/data/collection/gpw-v4).
Other predictor variables that can be potentially useful:
- Global tree density maps (http://dx.doi.org/10.6084/m9.figshare.3179986);
- Artificial night-light brightness from NOAA, NPP VIIRS – NASA
(http://visibleearth.nasa.gov/view.php?id=79765);
- Openstreetmap (https://www.openstreetmap.org/);
- Global roads (http://sedac.ciesin.columbia.edu/data/set/groads-global-roads-open-access-
v1/data-download).
Page 19
EBBA2 European Breeding Bird Atlas
Example of other predictor variables
Artificial night-light brightness
Global roads
Global tree density
Openstreetmap
Page 20
EBBA2 European Breeding Bird Atlas
Species
distribution models
Predictor variables
Validations
Bird data
Sampling effort
Bird data
Page 21
EBBA2 European Breeding Bird Atlas
Bird data
A (simplified) example of standardised data
36VUJ4_12 11.05.2014 60 Anas platyrhynchos
36VUJ4_12 11.05.2014 60 Tetrastes bonasia
36VUJ4_12 11.05.2014 60 Scolopax rusticola
36VUJ4_12 11.05.2014 60 Cuculus canorus
36VUJ4_12 11.05.2014 60 Dryocopus martius
36VUJ4_12 11.05.2014 60 Dendrocopos major
36VUJ4_12 11.05.2014 60 Dendrocopos leucotos
36VUJ4_12 11.05.2014 60 Anthus trivialis
36VUJ4_12 11.05.2014 60 Troglodytes troglodytes
36VUJ4_12 11.05.2014 60 Prunella modularis
36VUJ4_12 11.05.2014 60 Erithacus rubecula
36VUJ4_12 11.05.2014 60 Luscinia luscinia
36VUJ4_12 11.05.2014 60 Turdus merula
36VUJ4_12 11.05.2014 60 Turdus philomelos
36VUJ4_12 11.05.2014 60 Turdus iliacus
36VUJ4_12 11.05.2014 60 Sylvia borin
36VUJ4_12 11.05.2014 60 Sylvia atricapilla
36VUJ4_12 11.05.2014 60 Phylloscopus sibilatrix
36VUJ4_12 11.05.2014 60 Phylloscopus collybita collybita
36VUJ4_12 11.05.2014 60 Phylloscopus trochilus
36VUJ4_12 11.05.2014 60 Regulus regulus
36VUJ4_12 11.05.2014 60 Ficedula hypoleuca
36VUJ4_12 11.05.2014 60 Aegithalos caudatus
36VUJ4_12 11.05.2014 60 Parus palustris
36VUJ4_12 11.05.2014 60 Certhia familiaris
36VUJ4_12 11.05.2014 60 Fringilla coelebs
36VUJ4_12 11.05.2014 60 Spinus spinus
36VUJ4_12 11.05.2014 60 Coccothraustes coccothraustes
DateSurvey time
(minutes) Scientific name
10x10 km
square
≈ 100,000 timed surveys 12,057 10x10 km squares surveyed
Page 22
EBBA2 European Breeding Bird Atlas
Gaps identification (mMESS)
Combination of surveyed squares
and predictor variables
Modified version of the
Multivariate Environmental
Similarity Surface (mMESS).
Identification of poorly sampled
areas in Europe.
MESS is limited by its use of the
most dissimilar variable(s) as
indicator(s) of overall similarity;
mMESS weighs all variables in
the same way.
Page 23
EBBA2 European Breeding Bird Atlas
Species
distribution models
Predictor variables
Validations
Bird data
Sampling effort
Sampling effort
Page 24
EBBA2 European Breeding Bird Atlas
Sampling effort (averages over 100 simulated landscapes/communities)
Page 25
EBBA2 European Breeding Bird Atlas
Species
distribution models
Predictor variables
Validations
Bird data
Sampling effort
Species distribution models
Page 26
EBBA2 European Breeding Bird Atlas
SPECIES DISTRIBUTION MODEL (SDM) R PACKAGE
Artificial Neural Network (ANN) biomod2
Boosted Regression Trees (BRT) biomod2
Flexible Discriminant Analysis (FDA) biomod2
Multiple Adaptive Regression Splines (MARS) biomod2
Random Forest (RF) biomod2
Generalized Additive Model (GAM) biomod2
Generalized Linear Model (GLM) biomod2
Hierarchical Bayesian SDMs (hSDM) hSDM
Mapping Occurrence Probability (unmarked) unmarked
Bayesian SDM using Gaussian Processes (GP-MAP) GRaF
• Models for which autocorrelation of the residuals will be taken into account
through Kriging or Thin Plate Splines (TPS) functions;
• Models accounting for spatial autocorrelation.
Species distribution models (presence/absence data)
Page 27
EBBA2 European Breeding Bird Atlas
Species distribution models
Ensemble Modelling
(or model-averaging)
Tested to improve predictive performance
New algorithms
Evaluated and eventually included in the analyses
Page 28
EBBA2 European Breeding Bird Atlas
Species
distribution models
Predictor variables
Validations
Bird data
Sampling effort
Validations
Page 29
EBBA2 European Breeding Bird Atlas
Validations
VALIDATION STATISTICS
Area Under the Curve (AUC)
True Skill Statistic (TSS)
Cohen’s K (K)
Boyce Index (BY)
Correct Classification Rate (CCR)
Sensitivity - True Positive Rate (TPR)
Specificity - True Negative Rate (TNR)
10-folds cross-validations (90% vs. 10%)
WHOLE DATASET
Subsets to develop the models
# 1
# 2
# 3
Subsets to validate
the models
# … …
# 10
Page 30
EBBA2 European Breeding Bird Atlas
Luscinia luscinia Luscinia megarhynchos
Data collected
Page 31
EBBA2 European Breeding Bird Atlas
The invaluable role of the EBCC network of national coordinators, ornithologists and birdwatchers!
Bird Numbers 2016: Birds in a changing world Martin-Luther University in Halle (Saale), Germany
Page 32
EBBA2 European Breeding Bird Atlas
Luscinia luscinia Luscinia megarhynchos
A current exploratory modelling work (GLM)
Probability of occurrence
Low
High
Page 33
EBBA2 European Breeding Bird Atlas
Luscinia luscinia Luscinia megarhynchos
Spatial autocorrelation of the residuals (GLM)
Potential solution : Thin plate splines (Tps) regression of residuals
Page 34
EBBA2 European Breeding Bird Atlas
Luscinia luscinia Luscinia megarhynchos
Thin plate splines regression of the residuals (GLM)
Page 35
EBBA2 European Breeding Bird Atlas
Luscinia luscinia Luscinia megarhynchos
Thin plate splines regression of the residuals (GLM)
Probability of occurrence
Low
High
Page 36
EBBA2 European Breeding Bird Atlas
What will the new Atlas achieve?
1. Up-to-date distribution maps for birds across Europe;
2. Assess changes in species distribution since the 1980s.
3. Bird data collected will result in one of the most comprehensive biodiversity data sets in the world.
4. The Atlas will build capacity for conservation and monitoring in areas where this is most needed.
Page 37
EBBA2 Steering Committee: Verena Keller (Chair), Ruud Foppen, Mikhail Kalyakin, Lluis Brotons, Mark Eaton, Hans-Günther Bauer, Aleksi Lehikoinen & David Noble. EBBA2 coordination team and assistants: Sergi Herrando, Petr Vorisek, Marina Kipson, Marti Franch, Marc Anton, Dani Villero & Magda Pla. Spatial Modelling Group of the EBCC: Pietro Milanesi, Henk Sierdsema, Tomas Telensky, Alison Johnston, Jerome Guelat, Thomas Sattler, Marc Kery, Lluis Brotons, Nicolas Titeux, Frederic Jiguet, Christian Kampichler, Simon Gillings & Thomas Gottschalk. National coordinators and data providers for 2nd data provision: Taulant Bino (Albania), Clara Pladevall (Andorra), Karen Aghababyan & Hasmik Ter-Voskanyan (Armenia), Norbert Teufelbauer (Austria), Elchin Sultanov (Azerbaijan), Semion Levy & Marina Dmitrenok (Belarus), Dražen Kotrošan & Jovica Sjeničić (Bosnia and Herzegovina), Svetoslav Spasov & Stoycho Stoychev (Bulgaria), Vlatka Dumbovic & Tibor Mikuska (Croatia), Martin Hellicar (Cyprus), Karel Štastný & Vladimír Bejček (Czech Republic), Irina Levinsky & Thomas Vikstrom (Denmark), Jaanus Elts & Renno Nellis (Estonia), Janus Hansen (Faroe Islands), Aleksi Lehikoinen (Finland), Bernard Deceuninck, Nidal Issa, Jean-Philippe Siblet & Frederic Jiguet (France), Guille Mayor & Alexander Abuladze (Georgia), Christoph Sudfeldt & Christoph Grüneberg (Germany), Danae Portolou (Greece), Karoly Nagy, Zsolt Nagy, Tibor Szep (Hungary), Kristinn Skarphedinsson & Gudmundur Gudmundsson (Iceland), Brian Caffrey & Olivia Crowe (Ireland), Roberto Lardelli, Guido Tellini & Lorenzo Fornasari (Italy), Sergey Sklyarenko (Kazakhstan), Viesturs Kerus, Andris Dekants & Ainars Aunins (Latvia), Georg Willi (Liechtenstein), Liutauras Raudonikis (Lithuania), Patric Lorgé & Mikis Bastian (Luxembourg), Metodija Velevski, Ksenija Putilin & Danka Uzunova (Macedonia), Nicholas Barbara (Malta), Mihailo Jovičević & Darko Saveljić (Montenegro), Kjetil Mork, Ingar Jostein Øien, Magne Husby, John Atle Kålås & Roald Vang (Norway), Tomasz Chodkiewicz & Tomasz Wilk (Poland), Domingos Leitao, Joao Rabaca & Carlos Godinho (Portugal), Larisa Bogdea, Vitalie Ajder, Silvia Ursul & Emanuel Baltag (Republic of Moldova), Zoltan Szabo & Judit Veres-Szászka (Romania), Olga Voltzit & Mikhail Kalyakin (Russia), Dimitrije Radisic (Serbia), Jozef Ridzoň (Slovakia), Tomaž Mihelič (Slovenia), Juan Carlos del Moral & Blas Molina (Spain), Ake Lindstrom & Martin Green (Sweden), Hans Schmid & Peter Knaus (Switzerland), Chris van Turnhout & Dirk Zoetebier (The Netherlands), Dilek Sahin & Kerem Boyla (Turkey), Dawn Balmer, Simon Gillings & Justin Walker (UK), Igor Gorban & Tatiana Kuzmenko (Ukraine), as well as on-line portals, regional sources & foreign birdwatchers.
Thanks a lot to thousands of ornithologists and birdwatchers across Europe!!!
Page 38
EBBA2 European Breeding Bird Atlas
Thank you for
your attention