Delineation of Winter and Pre-breeding Habitats of Rocky Mountain Population
Trumpeter Swans
Core tri-state wintering area of RMP trumpeter swans
History and Status of Trumpeter Swans
•Restoration efforts underway since the 1930s
•Translocations have resulted in more trumpeter swans in North America
•The Canada breeding segment of the RMP has increased
•The number of breeding swans in the core tri-state area of Montana, Idaho, and Wyoming remains low
Entire Tri-state breeding population in 2006:312 adults and only 33 successful nests
•Rocky Mountain Population is between the black dashed lines
•Notice the large Canada breeding area and the small tri-state core area
Tri-state swans increase to several thousand in winter
•The Canada breeding portion of the Rocky Mountain Population has increased to >4000
•Canadian birds migrate to winter in the tri-state–<1,000 wintered in the tri-state before 1980–>5,400 in February 2006 winter survey
•Competition with swans from Canada in winter may reduce productivity of the tri-state breeders
Wintering trumpeters seek submergent plants and tubers in low-velocity streams and rivers
Example of slow-moving river habitat in Yellowstone National Park
On the National Elk Refuge, wintering swans removed 42% of foliage and 28% of tubers of
Stuckenia pectinatus (Squires 1991)
Other important foods include Chara, Elodea, and Myriophyllum exalbescens
The 2002 Trumpeter Swan Implementation Plan (TSIP) calls for reducing swans wintering
the core tri-state area to 1,500
•Redistribution efforts have included
–Trapping and translocating hundreds of cygnets to areas outside the core area
–Hazing of concentrations of swans from lakes and rivers
Harriman State Park, Idaho, is a favorite wintering site where swans were trapped and hazed
TSIP also called for:
•Inventory of suitable habitat throughout the range of RMP•The development of a comprehensive database and model of potential habitat availability
Our goal:
•Use image analysis and GIS to delineate areas of both used and potential wintering habitat in the tri-state core and beyond
Method – Supervised Classification
•To use image analysis software to–determine the reflectance characteristics of wetlands known to be used by wintering swans–find other locations that have those same characteristics (potential habitat)
•Used the Leica Image Analysis Extension for ArcGIS
Landsat TM dataThe positives:•Provides 8 spectral bands
–visible, infrared, and thermal
•Each scene is big–All of Yellowstone NP in 1
Landsat TM dataThe positives:•Provides 8 spectral bands
–visible, infrared, and thermal
•Each scene is big–All of Yellowstone NP in 1
•Computer can easily handle the 9-scene mosaic required
–Our tri-state mosaic•It’s available and free
Landsat TM dataThe positives:•Provides 8 spectral bands
–visible, infrared, and thermal
•Each scene is big–All of Yellowstone NP in 1
•Computer can easily handle the 9-scene mosaic required
–Our tri-state mosaic•It’s available and free
The negatives:•15-m to 60-m pixel size
–Might miss smaller wetlands
Swan winter location data
•Neckband resighting records collected over the last 17 years (since 1990)
–1990-2002 have been tabulated with lat –long locations
•Nov-Feb (winter) resightings within tri-state region represent >16,000 records of swans in >100 separate locations
•Yellow dots each represent from 1 to several hundred neck band sightings during winter (Nov-Feb)
•Yellow dots each represent from 1 to several hundred neck band sightings during winter (Nov-Feb)
Second set of location data
•Winter aerial surveys flown annually for the last 5 years (2002-2006)
–Separate counts for >200 named areas in the Montana, Idaho, and Wyoming
•Lat-long locations were not usually precise enough for our purposes here
•Only Wyoming data for 2006 were considered precise enough
•Blue dots each represent an individual aerial count of a swan or single group of swans
Selection of pixel for classification as swan habitat:
•One pixel per survey point
•Only wetland (dark) pixels used
•Within 2 pixels of survey point
Selection of pixel for classification as swan habitat:
•One pixel per survey point
•Only wetland (dark) pixels used
•Within 2 pixels of survey point
Supervised classification:
Software found all pixels on the image within the spectral range of those selected
•Analysis used all 8 Landsat TM spectral bands
Supervised classification:
Software found all pixels on the image within the spectral range of those selected
•Analysis used all 8 Landsat TM spectral bands
Interim Conclusions(work is on-going)
•The technique seems promising•More, geographically precise swan location data are needed over the entire area•We hope to use higher resolution imagery only within areas identified as wetlands by NWI maps •Would like to incorporate a classification of wetland sites known to NOT be suitable to swans to improve the classification•Ultimately need to validate the final product against independent swan locations
If successful
•The map of potential wintering sites would be useful in carrying out the Flyway’s Trumpeter Swan Implementation Plan
–Identify possible translocation sites–Model probabilities swans will find these sites
•On their own, based on known swan food-searching and dispersal characteristics•Upon hazing from traditional wintering areas
Acknowledgements
Eastern Kentucky University
USGS
Northern Rocky Mountain Science
Center
Montana State University
US Fish and Wildlife
Migratory Birds
Pacific Flyway