Advancing maps of ignorance on the distribution of biodiversity
description
Transcript of Advancing maps of ignorance on the distribution of biodiversity
Advancing maps of ignorance on the distribution of biodiversity
Visualising the future of our planet – Can we do better than heat maps?
Museo Nacional de Ciencias Naturales (CSIC), Spain
http://jhortal.com/; [email protected] / [email protected]
Joaquín Hortal
Microsoft Research Ltd.Cambridge, 16-17/2012
Summary
Biodiversity information
Quality (and quantity) of data: Wallacean
Shortfall
Mapping unknown species distributions
Mapping ignorance
Imagine a magnificent and omniscient GIS for all the Earth’s living species, with the capacity to display any level of the Linnaean hierarchy on any spatial scale, for any season of the year.
biodiversity and biogeography
Colwell & Coddington Phil Trans Roy Soc B 1994
digitize available distributional information: Natural History collections
▪ Institutional (Museums, Herbaria)▪ Private collections
gathering biodiversity information
digitize available distributional information: Natural History collections
▪ Institutional (Museum, Herbaria)▪ Private collections
Literature
gathering biodiversity information
digitize available distributional information: Natural History collections
▪ Institutional (Museums, Herbaria)▪ Private collections
Literature ad hoc surveys
gathering biodiversity information
integrate all information on the distribution of biodiversity
biodiversity databases
Map of Life
http://www.gbif.org/ ; http://www.mappinglife.org/ ; http://splink.cria.org.br/
Adequate distribution data is lacking for many of the known species and higher taxa (Lomolino 2004)
Whittaker et al. Div Distr 2005; Hortal et al. Conserv Biol 2007
Wallacean shortfall
Adequate distribution data is lacking for many of the known species and higher taxa (Lomolino 2004)
1,131 species1,084,971 records
960 records/species128 records/grid cell
Tenerife seed plants
Whittaker et al. Div Distr 2005; Hortal et al. Conserv Biol 2007
Wallacean shortfall
Tenerife seed plantsRecords Observed
Richness
Whittaker et al. Div Distr 2005; Hortal et al. Conserv Biol 2007
Wallacean shortfall
Adequate distribution data is lacking for many of the known species and higher taxa (Lomolino 2004)
taxonomic error
Lozier et al J Biogeogr 2009
taxonomic error
Lozier et al J Biogeogr 2009
taxonomic bias
Baselga et al Biodiv Conserv 2007
recorder’s home rangehotspots
spatial bias
Dennis & Thomas J Insect Conserv 2000
accessibility: ‘roadside bias’
spatial bias
Kadmon et al Ecol Appl 2003; Hurlbert & Jetz PNAS 2007
butterflies scarab dung beetles
bias differs between groups
Hortal et al Biod Conserv 2001; Hortal et al Ecography 2004
spatial bias
Onthophagus fracticornis
Lobo et al. Div Distr 2007
temporal bias
Historical survey process has been incomplete and biased:
Taxonomic bias Temporal bias Spatial bias
quality of distributional data
Pineda & Lobo J Anim Ecol 2009
Historical survey process has been incomplete and biased:
Taxonomic bias Temporal bias Spatial bias
quality of distributional data
Pineda & Lobo J Anim Ecol 2009
Current biodiversity picture depends on the survey process
Historical survey process has been incomplete and biased:
Taxonomic bias Temporal bias Spatial bias
quality of distributional data
Pineda & Lobo J Anim Ecol 2009
Current biodiversity picture depends on the survey process
Current knowledge on species distribution patterns may depend on survey unevenness rather than on their actual distributions
fill in the gapsexpert opinion
predictive models
mapping species distributions
Carabus granulatus Copris hispanus
Hortal J Biogeogr 2008; Penev et al The genus Carabus in Europe 2007; Chefaoui et al Biol Conserv 2005
neither the species are present everywhere within their range maps, nor all their known occurrences are within these range maps
inconsistencies with atlas data
Hurlbert & White Ecol Lett 2005
these mismatches are scale dependent
inconsistencies with atlas data
Hurlbert & Jetz PNAS 2007
prob
abili
ty o
f pre
senc
eenvironmental gradient
land classes
limited knowledge on the predictors
the actual responses of the species to the environment are unknown
data incompleteness
0
10
20
30
40
50
60
70
80
90
100
1900 1935 1970 1998
Num
ber o
f spe
cies
Niche coverage100%
90%
75%50%
1900 1935 1970 19980
20
40
60
80
100
1900 1935 1970 19980
20
40
60
80
100
Tota
l
All species 23 First recorded species
the descriptions of the environmental responses of most species are incomplete and biased
Hortal et al Oikos 2008
Chefaoui et al Anim Biodiv Conserv 2011
expert-drawnobserved plots
predictive modelshybrid approach
fine
coarse
uncertainty in predictions
different techniques predict different distribution patterns
whorl snailVertigo mouninsiana
southern damselflyCoenagrion mercuriale
GLM GAM NNET
Araújo & Rahbek Science 2006; Lawler et al Global Change Biol 2006
uncertainty in future projections
other determinants of the distribution
historical effects
e.g., Lobo et al. Div Distr 2006Chefaoui & Lobo J Wildl Man 20º7
Spanish moon mothGraellsia isabelae
ensemble forecasting
Araújo & New Tree 2006
dealing with uncertainty?
maps of ignorance
Boggs Proc Am Phil Soc 1949
a region is an “area of ignorance” if the total library resources of the outside world do not cover it (Boggs 1949)
maps of ignorance
Boggs Proc Am Phil Soc 1949
a region is an “area of ignorance” if the total library resources of the outside world do not cover it (Boggs 1949)
accuracy of knowledge
Hortal, Ladle et al in prep.
Kp = f ( [K0·C] , Lt , Ls )
Kp = accuracy of the knowledge about a given taxon or community at area pK0 = knowledge about such taxon or community at each area in the moment of the surveyC = degree of completeness of the surveyLt = loss of knowledge across time
Ls = loss of knowledge across space
0 200 400 600 800 1000 1200
em
0
20
40
60
80
100
Sobs
– Taxonomic accuracy– Detectability (crypsis, phenology)– Adequacy of sampling method and dates– Interactions– Size of focal unit– Habitat heterogeneity– Sampling effort and success
quality of initial knowledge
Hortal, Ladle et al in prep.
Temporal decay of similarity:- Changes in taxonomy- Turnover of species (mobility, phenotypic traits)- Area of unit (small higher turnover)- Range shifts (climate change)- Local extinctions (land use changes, biological invasions)
temporal loss of knowledge
Hortal, Ladle et al in prep.
Magersfontein battlefield, South Africa
Magersfontein battlefield, South Africa(from Moustakas et al Front Biogeogr 2010)
1899
2005
Distance decay of similarity:- Taxon specific- Biogeographical changes- Environmental gradients- Metacommunity structure- Habitat specificity (niche width)
spatial loss of knowledge
Hortal, Ladle et al in prep.
Magersfontein battlefield, South Africa
(from Green et al Nature 2004)
•Species: metacommunity structure / habitat specificity (niche width) / changes in climatic scenopoetic conditions
1. develop tools to map ignorance- how to measure taxonomic uncertainty- how to assess uncertainty in observations- how to map the degree of reliability of species distribution models in each point of space- how to determine when distribution is being extrapolated- how…
2. attach maps of ignorance as metadata for any distributional map
suggestions are welcome!
looking forward
1. develop tools to map ignorance- how to measure taxonomic uncertainty- how to assess uncertainty in observations- how to map the degree of reliability of species distribution models in each point of space- how to determine when distribution is being extrapolated- how…
2. attach maps of ignorance as metadata for any distributional map
suggestions are welcome!
looking forward
gracias thank youMuseo Nacional de Ciencias Naturales (CSIC), Spain
http://jhortal.com/ ; [email protected] , [email protected]
Joaquín Hortal
Richard J. Ladle
Geiziane Tessarolo
Jorge M. Lobo
Duccio Rocchini
and many others...