Mapping global bird distributions
NCEAS working group meeting 16-20 July 2001
Walter JetzDept Zoology
Oxford
Benefits
• Large-scale conservation priority setting (refining the hotspot approach with species distributions)
• Rapid assessment of diversity in regions under threat
• Coarse-resolution basis for deductive modelling of species’ fine-scale distributions
• Scrutiny of hypothesis in large-scale ecology
Why a free, public global vertebrate distribution database would be valuable
Mapping Global Vertebrate DistributionsWalter Jetz, University of Oxford
Conservation NGOs• prioritise conservation efforts• taxonomically: range size as
measure of extinction risk• geographically: refining
hotspots using species data
Museums• identify holes in distribution and
gaps of specimen records• prioritise areas for fieldwork• link morphological data and
biogeographic perspective
Academia• identify determinants of patterns in
species richness• detect mechanisms and environmental
correlates of speciation• understand environmental determinants
of biological patterns
General Public• custom species lists for home
region or eco-tourism destinations
Private Sector Land-use• information tool for land
development projects, impact assessments
General GOs and NGOs• tool for resource and land
management
Beneficiaries
Mapping Global Vertebrate DistributionsWalter Jetz, University of Oxford
Conservation NGOs• ground network of expertise• source and gap identification• facilitation of digitisation and
gap filling
Museums• taxonomic expertise• source identification and selection• specimen records
Academia• methodological expertise• source selection and prioritisation• GIS tools
Joint Effort
Working Groupmeetings at (and funded by)
National Centre for Ecological Analysis and Synthesis, University of Santa Barbara
Mapping Global Vertebrate DistributionsWalter Jetz, University of Oxford
Diversity of Terrestrial Vertebrates• Birds (Sibley)
– Passerines (Passeriformes): 5879– Nonpasserines (Non-Passeriformes): 4075
• Mammals (Wilson & Reeder 1993)– Platypus, Echidnas (Monotremata): 3– Oppossums, Kangaroos etc. (Marsupalia): 273– Placental Mammals (Eutheria): 4353-78 (whales)
• Amphibians (Duellman & Trueb 1986) – Frogs and Toads (Salientia): 3438– Salamanders and Newts (Caudata): 352– Caecilians (Gymnophiona): 162
• Reptiles (Uetz) – Lizards (Sauria): 4582– Snakes (Serpentes): 2910– Turtles (Testudines): 296– Crocodiles (Crocodylia): 23– Amphisbaenians (Amphisbaenia): 158– Tuataras (Rhynchocephalia): 2
9954
4275
3952
7971
--------26152
Mapping Global Vertebrate DistributionsWalter Jetz, University of Oxford
State of country- and continent-wide mapping efforts for bird distributions. Dark green: advanced, light green: weak
Birds regional databases
9954 speciesin 176 families
Mapping Global Vertebrate DistributionsWalter Jetz, University of Oxford
Sources
Identify key sources
Regional atlas projects
Meta-data collections
Museum specimen
Experts’ opinion
Species accounts
Regional species listsCongo Peacock
Square-tailed Kite
Mapping Global Vertebrate DistributionsWalter Jetz, University of Oxford
America
AustraliaPasserines not covered
Nonpasserines not covered
Europe & Africa
Gridded databases
Not covered
Handbook of the Birds of theWorld
Other Mono-graphs
Birds the knowledge base
I. Major regional atlases(proportion of 9954 bird species)
II. Major monographs(proportion of 176 bird families)
Mapping Global Vertebrate DistributionsWalter Jetz, University of Oxford
Taxonomically organised sources Distributions
Gaps
Geographically organised sources
source overlap
Inter-relationship of Source Types
• Monographs• Specimen Collections• etc. ...
• Atlases• Regional databases• Regional monographs• Regional specimen collections• etc. ...
• IUCN Red List data • Conservation NGO data and
distribution maps
Sources organised by extinction risk
Taxonomically organised sources
Example: Bird Distributions in Asia
• Threatened Birds of the World• Regional threatened species databases
Geographically organised sources
• HBW - Nonpasserines• Thrushes of the World• Finches and Sparrows of the World• Old World Warblers• etc. ...
• The Birds of China• Birds of the Indian Subcontinent• Birds of Japan• etc. ...
Sources organised by extinction risk
Distributions
Gaps
source overlap
Taxonomically organised sources
Full Distributions
Sources organised by extinction risk
Geographically organised sources
Birds: 9,954 species
I.Handbook of the Birds of the Worldfull ranges for 3,666 species
I.Birdlife: Threatened Birds of the Worldfull ranges for 1,189 species
I.• ABI-CABS Birds of the Americas
Databasepartial ranges for ca. 3,680 species
• Atlas of Birds of Australiapartial ranges for ca. 1,030 species
• Atlas of Birds of Europepartial ranges for ca. 430 species
• Birds of Oceanic islands, from WWF eco-regions and other sourcespartial ranges for ca. 450 species
* listed are potential sources pending agreement with authors/publishers
I. 8,200 species
II.• Birds of the Western Palearctic
full ranges for ca. 520 species• Birds of China• Keith et al: Birds of Africa, Atlases
from Southern Africa, Tanzania, Kenia, Somalia, Liberia, etc. …partial ranges for ca. 2000 species
II. 9,450 species III. Various family monographs
III.• Various regional sources and species
lists.
III. 9,954 species
III.Birdlife: Endemic Bird Areasfor Orientalis, Wallacea
Concatenated,original resolution
How to map a species’ range from a variety of sources?Overlaying disparate sources
Source 1 HBW
Source 2Regional Monograph
Source 3Regional Atlas
Hierarchical Decision Rule
S4 > S3 > S2 > S1
Gridded,fixed resolution
Source 4 Point Data
• Climate and vegetation layers, remotely sensed
• Species habitat preference information
confirmed fine scale presence and absence,extent of occurrence maps for biogeographic validation
Modelled (inductive and deductive) species distribution
General MethodologyRange of potential sources
Identify available sources
Evaluate sources for quality, accessibility and complementarity
Selected sources
Identify most efficient method of digitisation
Digitise
Multitude of regional and taxonomic databases of different resolution and quality
Devise hierarchical algorithm for query where sources overlap
Queried database
Mapping Global Vertebrate DistributionsWalter Jetz, University of Oxford
Taxonomies
Data Reality• Taxonomies will always differ somehow by expert and region• Various initiatives: ISIS, Species2000…. BCIS
Master Taxonomies ?• Herps: Master Taxonomies?• Mammals: Wilson & Reeder• Birds: Sibley & Ahlquist
Solution …?• Create database of all potential taxonomies (or ask data provider to provide)• Link all taxonomies to master taxonomy
Source Types - Problems
Extent of occurrence maps • poor temporal and spatial resolution• wide coverage• Frame/size of maps printed in books does not scale with extent of range
• potentially high spatial error• size of error should be directly related to map scale, can perhaps be
incorporated in modelling • inter- and extrapolated in unstandardised way, false presences• great resource for range modelling together with point data
Atlas Data• good temporal resolution• geographically limited• differences in observer effort, holes in distribution, false absences
Point Data: specimen, community studies, observations• perfect temporal and spatial resolution• coverage scattered, patchy, biased• great basis for ranges modelling using remotely sensed data and extent of
occurrence maps for biogeographic component
Towards a standardised source database
Fields to include:• Usual reference information (author, year, title, journal/publisher)• Extent: temporal, taxonomic, geographic (description), spatial object• Procedural information: processes undertaken, dates, people behind• Evaluation:
– spatial resolution– quality: correct species identification– quality: spatial error data– quality: spatial error digitisation
• Notes: Similar sources
Source types:• Published or expert-based extent of occurrence maps, atlas data,
gridded databases, regional or local community studies, point localities (observations and specimen)
Square-tailed Kite White-collared Kite
Time Efficient Data Entry
Streamlining the
digitisation process
Mapping Global Vertebrate DistributionsWalter Jetz, University of Oxford
Ranges of year-round residents, min=1 to max=70 speciesData from Handbook of the Birds of the World, resampled to 200km grid
Global Patterns of Diversity in Diurnal Raptors
Mapping Global Vertebrate DistributionsWalter Jetz, University of Oxford
Species Richnessnatural breaks, min=1, max=332
Geom. Mean of Range Sizesnatural breaks, min=12990km2,
max=13642403km2
New World Passerines
Collaboration with Lisa Manne and Stuart Pimm. Mapping Global Vertebrate DistributionsWalter Jetz, University of Oxford
Determinants of species richness- Hypotheses -
• Energy availability• Habitat Heterogeneity• Evolutionary Time• Biome area
• Geometric constraints
Plethora of hypotheses
• one single factor ???• many studies to date:
– limited to one specific hypothesis/variable
– limited to small region, taxonomic sub-sample
– limited to one dimension (e.g. latitude)
– excluding the tropics
Lessons so far
Natural breaks classification, 2-615 species per quadrat
Species richness pattern All species (n=1902), continental Africa
Collaboration with Carsten Rahbek.
NDVI mean of all ten-day images (1982-99)
Productivity & Habitat Heterogeneity
• remotely sensed from AVHHR satellites at 7.6km resolution
• NDVI (normalised difference vegetation index) is measure of greenness of vegetation, often used for vegetation classification
• NDVI is synthesis of climatic condition that regulate productivity
Productivity & Habitat Heterogeneityspatial pattern, observed vs. predicted
Natural breaks classification, left 3-558 species per quadrat, right 28-371
observed predicted(NPP, NPP2, HabHet)
Productivity & Habitat Heterogeneityspatial pattern of residuals
Residual from model NPP+NPP2+HabHetStandard deviation classification, <-3 to >+3s.d.; left -8.962 to 8.612 ; right -214 to 262
cyan: -white: 0red: +
A signature of history?
• Past climate events and their potential regional significance difficult to reconcile
• Species data as proxy• Assumption: Regions with restricted range species
(Centers of Endemism) have distinct evolutionary history• Prediction: species richness in such defined regions with
distinct evolutionary history is– likely to be higher than in surrounding regions– much less well predicted from contemporary environmental
variables
Standard deviation classification<-3 to >+3s.d., -214 to 262 species
Natural breaks classification 3-558 species per quadrat
Observed species richness Residual from model NPP+NPP2+HabHet
Centers of endemism: quadrats with species that have <= 10 quadrats range size
The signature of historyobserved and predicted species richness
in and outside Centers of Endemism (CoE)
The signature of historyobserved and predicted species richness
in and outside Centers of Endemism (CoE)
Mean
inside CoE outside CoE
Spe
cies
Ric
hnes
s
0
50
100
150
200
250
300
350
predictedobserved
species withrange size <=10
n = 359 quadrats n = 1379 quadrats
Environmental change and rates of evolution: the phylogeographic pattern within the hartebeest complex as related to climatic variationFlagstad et al. Proc. R. Soc Lond. B (2001) 268, 667-677
Phylogeography
Mapping Global Vertebrate DistributionsWalter Jetz, University of Oxford
The ‘End’ Product?
One-off database, downloadable from the internet
Continuously updated, peer supervised internet based database embedded in a multi-level access, graphical web-portal with facilities for down- and uploading data etc. ...
Source, species, lat, lon17,1245,45,4217, 1245,45,4317, 1245,45,4117, 1245,44,4317, 1245,43,4317, 1245,43,4217, 1246,02,2217, 1246,02,2217, 1246,03,2017, 1246,03,19………………
Mapping Global Vertebrate DistributionsWalter Jetz, University of Oxford
Mapping Global Vertebrate DistributionsWalter Jetz, University of Oxford