A Basic Introduction to Wildlife Mapping & Modeling ~~~~~~~~~~ Rev. Ronald J. Wasowski, C.S.C. Associate Professor of Environmental Science University of Portland Portland, Oregon 8 December 2015 Introduction to Wildlife Modeling Purpose of wildlife conservation Maintain maximum plant & animal diversity Genetic traits Ecological functions Bio-geo-chemical cycles Strategies for wildlife conservation Success in establishing parks & reserves Balanced communities of plants & animals Unique natural landscapes Failure in protecting complete ecological units Limiting factors Knowledge of species abundance Knowledge of species distribution Knowledge of causal factors Conservation & Reserve Management Conservation strategies Acquisition & protection of critical wildlife habitat Parks & reserves ~ 7.74. 10 6 km 2 (~ 5.2% of land surface) Many set aside as aesthetic or geologic wonders May or may not also be wildlife ecosystems Conservation problems Protected status no guarantee of conservation Other threats may exist Protecting resources no guarantee of conservation Inadequate understanding of impacting phenomena Fragmentation of wildlife habitat outside reserves Resting areas along flyways may work Migration corridors may not work A Small Migrating Herd Mapping Wildlife Distribution 1 Distribution data Traditionally derived from field observations Enhanced by modern technologies Aerial observation useful for colonial species Radio & GPS observation useful for migratory species Enhanced by satellite remote sensing imagery Limitations of spatial & temporal resolution Limitations of basic scientific understanding Limitations of classification algorithms & accuracy Mapping Wildlife Distribution 2 Some major successes of remote sensing Landsat TM/aerial photography coral reef studies Need to account for light attenuation with water depth Thermal InfraRed (TIR) studies of animal numbers Successes with bison, moose, deer & elk Problems with solar heated objects & non-target animals Radar useable if a species changes surface texture Radar sees both macro- & micro-geometry [texture]
Live Elephants Dead Elephants An Elephant Herd Wildlife Resource Requirements 1 Necessary resources Food, water, shelter, nesting Vegetation maps as species habitat maps Information present Species composition, physiognomy, density Information absent Vegetation quality, useable biomass Wildlife Resource Requirements 2 Strategies for preparing vegetation maps NDVI Single-date Rare for environmental satellites: Clouds Multi-date Weekly / monthly / seasonal Annual Integrated / crop calendar Decadal Long-term looking to climate, not weather Problems in high-relief areas Sun-lit & sun-shaded slope effects Wildlife Habitats & Habitat Maps Focus on the wildlife resource base Surrogate for the species of interest Vexing language issues Two traditional definitions of habitat Habitare Actual place where a species does live Type of environment where a species may or does live The territory occupied by a particular species Elk, grizzly bear, steelhead habitat Yes! Woodland, riverine habitat No! Ambiguity is one inevitable result Suitable habitat Redundant definition Unsuitable habitat Contradiction in terms
A Small Elk Herd An Even Smaller Elk Herd Pronghorn Antelope Mapping Suitability for Wildlife 1 Wildlife suitability map Land or water suitability for a particular species Examples of satellite-derived maps Suitability for Lesser prairie chickens Snow-free south-facing slopes for Alpine Ibex Suitability for Wood stork foraging Mapping Suitability for Wildlife 2 Fundamental problem Vegetation as the only explanatory variable There are almost always multiple explanatory variables GIS can handle multiple variables in multiple layers Weakness of conceptual models Both explanatory variables & their interrelationships Significant advancement for Black bears Ground cover Elevation, slope & aspect Roads Streams, forests & grasslands Accuracy of Suitability Maps 1 Validation strategies Poor from a theoretical perspective Seldom even attempted Factors determining accuracy Identification of all relevant variables Understanding of cause & effect relationships Unbiased estimation of model parameters
Accuracy of Suitability Maps Error types Type 1 Errors of omission Animals are not found in suitable areas Type 2 Errors of commission Animals are found in unsuitable areas Issue to remember Models predict suitability, not presence or absence Factors Influencing Distribution Causal environmental considerations Resource base factors Physico-chemical factors Cultural factors Spatial scale considerations First-order Global range of a species Global climate change, El Niño Second-order Home range of individuals Productivity & climatic patterns Third-order Resource utilization in home range Soil micro-fertility, moisture Additional considerations Competition between species Predation Random events Floods, fires A Stampede Species-Environment Relationships Two approaches to modeling Theoretical / deductive General Specific Very common approach since the 1980 s Solid theoretical understanding is essential Empirical / inductive Specific General Increasingly common approach since the 1990 s Analysis of data reveals patterns One significant problem Datasets tend to be highly correlated Use of PCA to reduce dimensionality of spectral bands Use of other strategies with field datasets Static & Dynamic Models Some basic issues Static models: Conditions at one point in time Multiple static models Actual present, possible future Present from existing data Future from some model Dynamic models: Conditions changing in time One continuous model Past Present Future Unbroken sequence of conditions Transferability of Species Site-specificity of models Almost all models are developed for specific sites May be local, regional, continental or global Conditions vary enough to complicate transferability Concerns degree of knowledge of explanatory variables Time-specificity of models Almost all models are developed for specific times Assumption that species are minimally adaptable Time of the model may not be representative Hotter & drier, cooler & wetter
Innovative Wildlife Env l Mapping Scale dependency Broad strokes modeled at continental scales Slope & aspect become texture Fine details modeled at plot scales Slope & aspect become critical Two examples Fires Displace & kill wildlife Use of TIR images to monitor fires through smoke Floods Damage sustaining vegetation Use of radar images to monitor floods through clouds One problem Critical gap between field & satellite observations Which dataset best represents the explanatory variables Some Conclusions Great potential for RS & GIS Increasing quantity & quality of data GPS-aided field data RS-aided regional data Increasing sophistication of models Theoretical & algorithmic Some components are critically lacking Financial resources Expertise resources Possible research directions More realistic dynamic models Include changes in both time & space Random events Catastrophic processes