Remotesensing correlates of biological diversity Catherine Graham Stony

  • Slides: 18
Download presentation
Remote-sensing correlates of biological diversity Catherine Graham Stony Brook University Graham lab: Jorge Velasquez

Remote-sensing correlates of biological diversity Catherine Graham Stony Brook University Graham lab: Jorge Velasquez Natalia Silva Pablo Menendez Other: Robert Hijmans Luis Coloma Santiago Ron NASA Funded: Tom Smith Sassan Saatchi Chris Schneider Robert Wayne

Remote-sensing correlates to biological diversity Effectively use RS data in species distribution modeling and

Remote-sensing correlates to biological diversity Effectively use RS data in species distribution modeling and decision support • Virtual species experiment • Modeling Andean bird species • Conservation planning Evaluate hypotheses explaining variability in species richness • Correlates of mammal richness across spatial scale • Effects of disease on amphibian richness pattern Simulate the impacts of and climate change on species distributions (in press, Global Change Biology) Train ecologists, evolutionary biologists and conservation biologists

Species distribution modeling 1) Extract environmental data for point localities; 2) Make statistical model

Species distribution modeling 1) Extract environmental data for point localities; 2) Make statistical model describing distribution in environmental space; 3) Project this model in geographic space to create a map. Elevation Annual Rainfall

Free download 16 modeling methods Presence-only training data Independent presence-absence test data 6 geographic

Free download 16 modeling methods Presence-only training data Independent presence-absence test data 6 geographic regions/series of taxonomic groups 250 species

Effectively use RS data in species distribution modeling • Problem: age and spatial accuracy

Effectively use RS data in species distribution modeling • Problem: age and spatial accuracy of point locality data in relation to RS data. • Solution: partition data in modeling – Use all point locality data with climate surfaces – Use only “accurate/recent” point locality data with remote-sensing layers

RS data in species distribution modeling: Virtual species experiment Points in Climate Original distribution

RS data in species distribution modeling: Virtual species experiment Points in Climate Original distribution (climate-only) Points in currentlyforested areas only Current distribution (climate & RS)

RS data in species distribution modeling: Virtual species experiment Points in RS-forest & climate

RS data in species distribution modeling: Virtual species experiment Points in RS-forest & climate only Climate+RS 0. 331 Sample size of 100 points 0. 632 Points partitioned by RS & climate 0. 462 *note correlations are between a binary and continuous prediction

RS data in species distribution modeling: Modeling Andean birds Treatments: Exp 1: climate only

RS data in species distribution modeling: Modeling Andean birds Treatments: Exp 1: climate only Exp 2 -4, climate and remote sensing layers without data splitting. - Exp 2: sampling from 1 km RS layers - Exp 3: sampling from 10 km RS layers - Exp 4: sampling from a neighborhood within a radius of 5 km Exp 5 -7, climate and remote sensing layers with data splitting. - Exp 5: sampling from 1 km RS layers - Exp 6: sampling from 10 km RS layers - Exp 7: sampling from a neighborhood within a radius of 5 km

Myadestes Ralloides (Andean Solitare) Exp 1 Exp 7 Exp 1: climate Exp 2 -4:

Myadestes Ralloides (Andean Solitare) Exp 1 Exp 7 Exp 1: climate Exp 2 -4: climate and RS without data splitting Exp 5 -7: climate and RS with data splitting

RS data in species distribution modeling: conservation planning In collaboration with CI & Pro.

RS data in species distribution modeling: conservation planning In collaboration with CI & Pro. Aves, we are redoing the analyses with all ~300 species and models built with both RS and climate data Preliminary conservation assessment with threatened parrots

Conservation planning Cerulean warbler listed as vulnerable by IUCN New protected area, 2005 Developing

Conservation planning Cerulean warbler listed as vulnerable by IUCN New protected area, 2005 Developing direct interactions with local conservation practitioners. - Courses - Data sharing - Decision support

Remote-sensing correlates to biological diversity Effectively use RS data in species distribution modeling and

Remote-sensing correlates to biological diversity Effectively use RS data in species distribution modeling and decision support • Virtual species experiment • Modeling Andean bird species • Conservation planning Evaluate hypotheses explaining variability in species richness • Correlates of mammal richness across spatial scale • Effects of disease on amphibian richness pattern Simulate the impacts of climate change on species distributions (in press, Global Change Biology) Train ecologists, evolutionary biologists and conservation biologists

Variability in species richness: Effects of disease on amphibian richness patterns Chytrid-thermal-optimum hypothesis Grey

Variability in species richness: Effects of disease on amphibian richness patterns Chytrid-thermal-optimum hypothesis Grey shading: estimated percentage of species lost from each altitudinal zone Optimum temperatures for chytrid: 17 C – 25 C Pounds et al. (2006)

Testing Chytrid-thermal-optimum hypothesis in Ecuador Temp 17 - 25 o C Temperature range does

Testing Chytrid-thermal-optimum hypothesis in Ecuador Temp 17 - 25 o C Temperature range does not correspond with declining frog distribution in Ecuador

Ecological Niche Hypothesis Chytrid distribution model Maxent Climatic & RS variables

Ecological Niche Hypothesis Chytrid distribution model Maxent Climatic & RS variables

PC II high 70% of variation explained low Primarily mean diurnal temperature range and

PC II high 70% of variation explained low Primarily mean diurnal temperature range and precipitation PCA of environmental space of chytrid and frogs labeled by IUCN categories low PC I Atelopus Colostethus Eleutherodactylus high Primarily temperatures during coldest and driest seasons

Forecasting Future Amphibian Declines • Tracking with RS data: rainfall of the driest quarter

Forecasting Future Amphibian Declines • Tracking with RS data: rainfall of the driest quarter is highly correlated with the mean leaf area index of the dry season • Forecasting: Use GCMs to investigate future changes in precipitation

Remote-sensing correlates to biological diversity: training CURSO–TALLER Métodos de modelamiento de distribución de especies

Remote-sensing correlates to biological diversity: training CURSO–TALLER Métodos de modelamiento de distribución de especies y sus aplicaciones Julio 10 al 15 de 2006