Skip to main content

Full text of "DTIC ADA571368: The Identification of Military Installations as Important Migratory Bird Stopover Sites and the Development of Bird Migration Forecast Models: A Radar Ornithology Approach"

See other formats


Environmental Laboratory erdc/elTR-12-22 



US Army Corps 
of Engineers© 

Engineer Research and 
Development Center 


The Identification of Military Installations as 
Important Migratory Bird Stopover Sites and 
the Development of Bird Migration Forecast 
Models: A Radar Ornithology Approach 

SERDP Project SI-1439 

Richard A. Fischer, Michael P. Guilfoyle, Jonathon Valente, August 2012 

Sidney A. Gauthreaux, Jr., Carroll G. Belser, 

Donald V. Blaricom, John W. Livingston, Emily Cohen, 
and Frank R. Moore 


Approved for public release; distribution is unlimited. 







ERDC/EL TR-12-22 
August 2012 


The Identification of Military Installations as 
Important Migratory Bird Stopover Sites and the 
Development of Bird Migration Forecast 
Models: A Radar Ornithology Approach 

SERDP Project SI-1439 

Dr. Richard A. Fischer, Dr. Michael P. Guilfoyle, 
and Jonathon Valente 

U.S. Army Engineer Research and Development Center 
Environmental Laboratory 
Vicksburg, MS 

Dr. Sidney A. Gauthreaux, Jr., Carroll G. Belser, Donald Van Blaricom, 
and John W. Livingston 

Clemson University Radar Ornithology Lab 
Clemson, SC 

Dr. Emily Cohen, and Dr. Frank R. Moore 

University of Southern Mississippi 
Department of Biological Sciences 
Hattiesburg, MS 


Final report 

Approved for public release; distribution is unlimited 



SERDP 

DOD ■ EPA ■ DOE 


Prepared for U.S. Army Corps of Engineers 
Washington, DC 20314-1000 


ERDC/EL TR-12-22 


ii 


Abstract 

Military lands and waters may be particularly valuable for migrating birds 
requiring stopover habitat to rest and refuel en route to very distant 
seasonal ranges. Recent developments in radar technology have provided 
powerful tools for investigating on a broad scale migrant use of military 
installations; thus providing an opportunity to improve both conservation 
and flight safety measures. In this study, spring and fall migrant use of 40 
military installations across the United States were qualitatively investi¬ 
gated. These times of year were selected since they are the periods when 
BASH is of most concern. Migratory patterns on three installations (Eglin 
Air Force Base, FL; Ft. Polk, LA; and Yuma Proving Ground, AZ) were then 
closely examined and migration forecast models for those locations were 
developed with the goal of providing a tool for reducing the probability of 
collisions between birds and military aircraft. A comparison was also made 
between radar estimates of migrant densities aloft during exodus events and 
more traditional ground-based surveys to evaluate the effectiveness of 
estimating migrant abundance in stopover habitat with radar data. At Fort 
Polk, movement ecology and migrant-habitat relations of the Red-eyed 
Vireo were investigated during migratory stopover. Lastly, migrant use of 
diverse riparian habitats was compared along water courses near the Yuma 
Proving Ground. Results indicated that approximately half of the installa¬ 
tions examined with radar data contained migrant stopover “hotspots,” 
reaffirming the fact that military installations are important to migrating 
birds. Interestingly, migrant abundances, and species turnover as estimated 
by ground-based surveys, were found to poorly reflect migrant densities 
estimated with radar data. Migrant abundance, species richness, and 
community composition were all also found to be influenced by riparian 
vegetation composition. This information collectively suggested that radar 
data can be used to identify migratory hotspots on military installations and 
improve flight safety on installations with an aviation mission. However, 
radar data may not be sufficient to distinguish fine-scale differences in 
habitat use by migrants within an installation’s boundaries. 


DISCLAIMER: The contents of this report are not to be used for advertising, publication, or promotional purposes. 
Citation of trade names does not constitute an official endorsement or approval of the use of such commercial products. 
All product names and trademarks cited are the property of their respective owners. The findings of this report are not to 
be construed as an official Department of the Army position unless so designated by other authorized documents. 

DESTROY THIS REPORT WHEN NO LONGER NEEDED. DO NOT RETURN IT TO THE ORIGINATOR. 




ERDC/EL TR-12-22 


iii 


Contents 

Abstract.ii 

Figures and Tables.v 

Preface.xiii 

Acronyms.xiii 

1 Objectives.1 

2 Background.3 

Objective 1: Migrant Use of Military Installations.4 

Objectives 2 and 3: Quantifying Migration and Developing Migration Forecast 

Models.5 

Objective 4: Comparison of Migrant Survey Techniques.8 

Objective 5: Avian Habitat Use in Southwestern Riparian Systems.9 

Objective 6: Movement Ecology and Habitat Use of a Neotropical Migrant during 
Spring Migratory Stopover.11 

3 Materials and Methods.12 

Objective 1: Migrant Use of Military Installations.12 

Objective 2: Quantifying Seasonal Migration.14 

Objective 3: Development of Migration Forecast Models.16 

Objective 4: Comparison of Migrant Survey Techniques.17 

Bird Surveys . 17 

Eglin Air Force Base, Florida . 23 

Fort Polk, Louisiana . 24 

Yuma Proving Ground , Arizona .26 

Radar Data .26 

Statistical Analyses .27 

Objective 5: Avian Habitat Use in Southwestern Riparian Systems.33 

Data Analyses . 33 

4 Results and Discussion.35 

Objective 1: Migrant Use of Military Installations.35 

Objective 2: Quantifying Seasonal Migration.39 

Fall Migration . 39 

Spring Migration . 42 

Discussion .44 

Objective 3: Development of Migration Forecast Models.45 

Discussion . 47 

Objective 4: Comparison of Migrant Survey Techniques.48 

Discussion .66 

Objective 5: Avian Habitat Use in Southwestern Riparian Systems.75 






































ERDC/EL TR-12-22 iv 


Discussion .75 

5 Conclusions and Implications for Future Research and Implementation.81 

References.83 

Appendix A: Movement Ecology and Migrant-Habitat Relations: Red-Eyed Vireos 

During Spring Stopover.89 

Appendix B: Composite Migration Maps Over U.S. Military Installations.122 

Report Documentation Page 








ERDC/EL TR-12-22 


v 


Figures and Tables 


Figures 

Figure 1. WSR-88D stations in the contiguous United States. Edges of the circles are at 

124 nautical mile range (230 km).6 

Figure 2. Diagram of neural net showing input variables, three nodes and output variable.17 

Figure 3. Three riparian transects established in Fall 2005 at Eglin Air Force Base, FL.18 

Figure 4. Transect endpoints along Tenmile Creek, Eglin AFB, FL.18 

Figure 5. Transect endpoints along Basin Creek, Eglin AFB, FL.18 

Figure 6. Transect endpoints along Alaqua Creek, Eglin AFB, FL.19 

Figure 7. Four riparian transects and one upland transect established in during 2005- 

2007 at Fort Polk, LA. The upland transect extends northwest from Drakes Creek.19 

Figure 8. Transect endpoints along Bundick Creek, Fort Polk, LA.20 

Figure 9. Transect endpoints along Drakes Creek, Fort Polk, LA.20 

Figure 10. Transect endpoints along Six Mile Creek, Fort Polk, LA.20 

Figure 11. Transect endpoints along the upland transect, Fort Polk, LA.21 

Figure 12. Transect endpoints along Whiskey Chitto Creek, Fort Polk, LA.21 

Figure 13. Three riparian transects established in spring 2006 near Yuma, AZ and 

Imperial National Wildlife Refuge, AZ.21 

Figure 14. Transect endpoints along All American Canal, Yuma, AZ.22 

Figure 15. Yuma transect endpoints along the Colorado River, Yuma, AZ.22 

Figure 16. Imperial transect endpoints along the Colorado River, Imperial National Wildlife 
Refuge, AZ.22 


Figure 17. Total number of birds recorded per morning transect at each Yuma site over 

time in the spring of 2006. The trends at the two sites tend to be mirror opposites of one 

another and the pattern reflects the fact that, for the most part, two observers alternated 

the days on which they sampled these sites (though a third observer was used on a few 

occasions). Thus, the trend indicates that one observer consistently counted more birds 

than the other regardless of which site they were both sampling.32 

Figure 18. Mean and standard error of the total number of birds per km recorded per 
morning transect by each observer at Yuma in 2006. Values were calculated by averaging 
the number of birds per km the observer counted on each morning he or she surveyed. 

Results indicate that observer 1 consistently counted more birds than observer 2 who 
consistently counted more birds than observer 3. Thus, since observers were rotated, it is 
impossible to determine how much of the change in migrant abundance from one day to 
the next at the site level was attributable to real turnover and how much to observer bias.32 

Figure 19. Map showing migration stopover areas based on WSR-88D detection of 
migrating birds during significant exodus events from Fort Polk, LA, during the fall 
migrations of 2000-2004. The data are quantified and displayed as standard deviations 
above mean.37 

Figure 20. Map showing migration stopover areas based on WSR-88D detection of 
migrating birds during significant exodus events from Eglin AFB, FL, during the fall 






















ERDC/EL TR-12-22 


vi 


migrations of 2000-2004. The data are quantified and displayed as standard deviations 
above mean.38 

Figure 21. Map showing migration stopover areas based on WSR-88D detection of 

migrating birds during significant exodus events from Fort Polk, LA, for the spring 

migrations of 2000-2003, and 2005. The colors represent standard deviations of above 

the mean density of birds per km 3 . Note that many of the stopover areas are associated 

with riparian habitat.38 

Figure 22. Map showing migration stopover areas based on WSR-88D detection of 
migrating birds during significant exodus events on and around Yuma Proving Ground, AZ 
for the spring migrations of 2000-2003 and 2005. The colors represent standard 
deviations above the mean density of birds per cubic km. The large red areas to the SW 
and SE of the radar site are not from migrant exodus events and are the results of ground 


clutter, breakthrough and radar blockage patterns in these areas. The DoD installation to 

the lower right is the Barry M. Goldwater Air Force Range, and the one above it is the 

Yuma Proving Ground.39 

Figure 23. Seasonal temporal pattern of nocturnal bird migration in fall over Eglin Air 

Force Base, FL for the years 2000-2005. The symbols represent the mean number of 

birds per km 3 and the bars indicate the standard error of the mean.40 

Figure 24. Seasonal temporal pattern of nocturnal bird migration in fall over Eglin Air 

Force Base, FL for the years 2000-2005. The symbols indicate the maximum value of 

birds per km 3 for each date of fall during the six year period.40 

Figure 25. Seasonal temporal pattern of nocturnal bird migration in fall over Fort Polk, LA 
for the years 2000-2005. Symbols represent the mean number of birds per km 3 and the 
bars indicate the standard error of the mean.41 

Figure 26. Seasonal temporal pattern of nocturnal bird migration in fall over Fort Polk, LA 
for the years 2000-2005. Symbols indicate the maximum value of birds per km 3 for each 
date of fall during the six year period.41 

Figure 27. Seasonal temporal pattern of nocturnal bird migration in spring over Fort Polk, 

LA for the years 2000-2003 and 2005-2006. Symbols represent the mean number of 

birds per km 3 and the bars indicate the standard error of the mean.42 

Figure 28. Seasonal temporal pattern of nocturnal bird migration in spring over Fort Polk, 

LA for the years 2000-2003 and 2005-2006. Symbols indicate the maximum value of 

birds per km 3 for each date of spring during the six year period.43 

Figure 29. Seasonal temporal pattern of nocturnal bird migration in spring over Yuma, AZ 
for the years 2000-2003 and 2005-2006. Symbols represent the mean number of birds 
per km 3 and the bars indicate the standard error of the mean.43 

Figure 30. Seasonal temporal pattern of nocturnal bird migration in spring over Yuma, AZ 
for the years 2000-2003 and 2005-2006. Symbols indicate the maximum value of birds 
per km 3 for each date of spring during the six year period.44 

Figure 31. Plot of actual birds per km 3 by predicted birds per km 3 for fall data from the 

years 2000-2005 at Eglin Air Force Base in Florida.46 

Figure 32. Plot of actual birds per km 3 by predicted birds per km 3 for fall data from 2005 

at Fort Polk Army Base in Louisiana.46 

Figure 33. Plot of actual birds per km 3 by predicted birds per km 3 for data from the spring 
seasons from 2001 through 2005 at Yuma, AZ.47 

Figure 34. Comparison of the daily change in nocturnal migrant survey abundance with 
daily exodus and peak migration densities (calculated from radar reflectivity) during 
spring migration of a) 2006 and b) 2007 near Yuma Proving Ground. The graphs depict 2 
y-axes; values on the left axis are in birds/km and represent A migrant abundance while 

















ERDC/EL TR-12-22 


vii 


the right axis is measured in mean birds birds per km 3 and represents exodus and peak 
migration densities. For A migrant abundance, values represent the number of migrants 
recorded on the morning of the plotted date minus the number of migrants recorded on 
the previous morning. Exodus and peak migration densities represent radar imagery 
captured in the early hours of the plotted date or the late hours of the previous evening, 
respectively. Note the differences in scale between figures a and b.58 

Figure 35. Comparison of the daily change in nocturnal migrant survey abundance with 
daily exodus and peak migration densities (calculated from radar reflectivity) during 
spring migration of a) 2006 and b) 2007 at Ft. Polk. The graphs depict 2 y-axes; values on 
the left axis are in birds/km and represent A migrant abundance while the right axis is 
measured in mean birds per km 3 and represents exodus and peak migration densities. 

For A migrant abundance, values represent the number of migrants recorded on the 

morning of the plotted date minus the number of migrants recorded on the previous 

morning. Exodus and peak migration densities represent radar imagery captured in the 

early hours of the plotted date or the late hours of the previous evening, respectively. Note 

the differences in scale between figures a and b.59 

Figure 36. Comparison of the daily change in nocturnal migrant survey abundance with 
daily exodus and peak migration densities (calculated from radar reflectivity) during fall 
migration of a) 2005, b) 2006 and c) 2007 at Eglin AFB. The graphs depict 2 y-axes; 
values on the left axis are in birds/km and represent A migrant abundance while the right 
axis is measured in mean birds per km 3 and represents exodus and peak migration 
densities. For A migrant abundance, values represent the number of migrants recorded 
on the morning of the plotted date minus the number of migrants recorded on the 
previous morning. Exodus and peak migration densities represent radar imagery 
captured in the early hours of the plotted date or the late hours of the previous evening. 

Note the differences in scale between figures a, b and c.60 

Figure 37. Comparison of the daily change in nocturnal migrant survey abundance with 
daily exodus and peak migration densities (calculated from radar reflectivity) during fall 
migration of a) 2005 and b) 2006 at Ft. Polk. The graphs depict 2 y-axes; values on the 
left axis are in birds/km and represent A migrant abundance while the right axis is 
measured in mean birds per km 3 and represents exodus and peak migration densities. 

For A migrant abundance, values represent the number of migrants recorded on the 
morning of the plotted date minus the number of migrants recorded on the previous 
morning. Exodus and peak migration densities represent radar imagery captured in the 
early hours of the plotted date or the late hours of the previous evening. Note the 


differences in scale between figures a and b.61 

Figure 38. Plots of linear regression models built to explain the daily change in migrant 

abundance recorded during bird surveys as a function of peak migration densities 

captured on radar at military installations during spring and fall migration.63 

Figure 39. Plots of linear regression models built to explain the daily change in migrant 

abundance recorded during bird surveys as a function of migrant exodus densities 

captured on radar at military installations during spring and fall migration.65 

Figure 40. Plots of linear regression models built to explain daily positive migrant turnover 
recorded during bird surveys as a function of peak migration densities captured on radar 
at military installations during spring and fall migration.67 

Figure 41. Plots of linear regression models built to explain daily negative migrant 

turnover recorded during bird surveys as a function of migrant exodus densities captured 

on radar at military installations during spring and fall migration.69 


Figure 42. Predicted change in migrant abundance values (and 95% confidence intervals) 
for days with different combinations of peak migration and exodus treatments. The first 
label for a bar indicates the peak migration treatment and the second indicates the 











ERDC/EL TR-12-22 


viii 


exodus treatment. Places where a bar is missing from the graph indicate that the 

treatment combination did not exist in the particular sampling season.70 

Figure 43. Mean (±SE) total migrant abundance per kilometer (a) and migrant species 
richness per transect section (b) recorded at 125 m transect sections of different habitat 


types near Yuma Proving Ground during spring migration in 2006 and 2007. Sections 

were classified as native tree (NT), native shrub (NS), native-dominated with non-natives 

(ND), non-native/invasive dominant with some natives (NND), or non-native/invasive 

shrub and tree community (NNI). Bars that do not have a letter in common indicate the 

response variable was significantly different between those habitat types.79 

Figure Al. Map of the state of Louisiana with translocation direction (arrow) from capture 
at Johnson’s Bayou to Kisatchie National Forest. Inset map of the study area within 
Kisatchie National Forest with release locations at Bundick (three • on left) and Drakes 
Creek (three • on right).94 

Figure A2. Number of mist-net captures of migrant species (corrected for daily netting 

effort) by date at Fort Polk, LA from 21 March to 5 May 2006 (excluding five days).101 


Figure A3. Number of Arthropods and Lepidoptera larvae detected along transects during 
spring of 2007 and 2008. The mean values are shown and the error bars represent 
standard deviation. The number of arthropods differed for all comparisons of pine, mixed 
and hardwood (p <0.001). The number of Lepidoptera larvae were different for pine and 
hardwood and pine and mixed habitat in (both comparisons p <0.001) but not for 


hardwood and mixed (p=0.136).103 

Figure A4. Mean number (bars represent standard error) of arthropods detected along 
transects (Random) and areas where migrants were located (Selected) in pine, mixed and 
hardwood habitat (pine p=0.01and mixed p=0.02 and hardwood p=0.05).104 

Figure A5. Mean number (bars represent standard error) of Lepidoptera larvae detected 
along transects (Random) versus areas where migrants were located (Selected) in each 
habitat type (pine p=0.03, mixed p=0.004 and hardwood p=0.10).104 

Figure A6. Mean linear displacement (m) by hour of the day and stopover day for the first 
three days of stopover. Mean values labeled and bars represent standard error.105 

Figure A7. Mean movement rate (m min 1 ) by hour of the day and stopover day for the first 
three days of stopover. Mean values labeled and bars represent standard error.106 

Figure A8. The number of successful attacks per time spent foraging in pine, mixed and 
hardwood habitat.107 

Figure A9. The correlation between fat score (Flelms and Drury 1960) and the condition 
index (R 2 = 0.56, P < 0.001). A condition index of zero corresponds to zero fat stores or 
lean body mass.110 

Figure A10. Duration of stay for red-eyed vireos radio tracked in Kistachie National Forest 

and the relationship between the condition of the bird (negative values are below lean 

body mass and positive are above) and the duration of stay in days.Ill 

Figure Bl. Composite maps indicating spring and fall migratory hotspots as recorded by 

NEXRAD station APX in northern Ml. The survey area encompasses Camp Grayling 

Military Reservation.123 

Figure B2. Composite maps indicating spring and fall migratory hotspots as recorded by 
NEXRAD station ARX in La Crosse, Wl. The survey area encompasses Fort McCoy.124 

Figure B3. Composite maps indicating spring and fall migratory hotspots as recorded by 

NEXRAD station CBX in Boise, ID. The survey area encompasses Saylor Creek Air Force 

Range.125 

Figure B4. Composite maps indicating spring and fall migratory hotspots as recorded by 
NEXRAD station CLX in Charleston, SC. The survey area encompasses Fort Stewart.126 



















ERDC/EL TR-12-22 ix 


Figure B5. Composite map indicating fall migratory hotspots as recorded by NEXRAD 

station EMX in Tucson, AZ. The survey area encompasses Fort Fluachuca.127 

Figure B6. Composite maps indicating spring and fall migratory hotspots as recorded by 

NEXRAD station EOX in southeastern AL. The survey area encompasses Fort Rucker 

Military Reservation.128 

Figure B7. Composite maps indicating spring and fall migratory hotspots as recorded by 

NEXRAD station EPZ in El Paso, TX. The survey area encompasses Fort Bliss and the Fort 

Bliss McGregor Range.129 

Figure B8. Composite maps indicating spring and fall migratory hotspots as recorded by 

NEXRAD station EVX in northwestern FL. The survey area encompasses Eglin Air Force 

Base.130 

Figure B9. Composite maps indicating spring and fall migratory hotspots as recorded by 
NEXRAD station EYX in southern CA. The survey area encompasses China Lake Naval 
Weapons Center, Edwards Air Force Base, and Fort Irwin.131 

Figure BIO. Composite maps indicating spring and fall migratory hotspots as recorded by 
NEXRAD station FDR in Frederick, OK. The survey area encompasses Fort Sill Military 
Reservation.132 

Figure Bll. Composite maps indicating spring and fall migratory hotspots as recorded by 

NEXRAD station GRK in central TX. The survey area encompasses Fort Flood and Camp 

Swift N. G. Facility.133 

Figure B12. Composite maps indicating spring and fall migratory hotspots as recorded by 

NEXRAD station FIDX in southern NM. The survey area encompasses Flolloman Air Force 

Base, White Sands Missile Range, Fort Bliss, and the Fort Bliss McGregor Range.134 

Figure B13. Composite maps indicating spring and fall migratory hotspots as recorded by 
NEXRAD station FIPX in southwestern KY. The survey area encompasses Fort Campbell.135 

Figure B14. Composite maps indicating spring and fall migratory hotspots as recorded by 
NEXRAD station LVX in central KY. The survey area encompasses Fort Knox.136 

Figure B15. Composite maps indicating spring and fall migratory hotspots as recorded by 
NEXRAD station LWX in Sterling, VA. The survey area encompasses Fort A.P. Hill Military 
Reservation and Quantico Marine Corps Base.137 

Figure B16. Composite maps indicating spring and fall migratory hotspots as recorded by 
NEXRAD station MHX in Morehead City, NC. The survey area encompasses Camp Lejeune 
Marine Corps Base.138 

Figure B17. Composite maps indicating spring and fall migratory hotspots as recorded by 

NEXRAD station MLB in Melbourne, FL. The survey area encompasses Avon Park Air 

Force Bombing Range.139 

Figure B18. Composite maps indicating spring and fall migratory hotspots as recorded by 

NEXRAD station MTX in Salt Lake City, UT. The survey area encompasses Hill Air Force 

Range and the Hill AFB Wendover Range.140 

Figure B19. Composite maps indicating spring and fall migratory hotspots as recorded by 
NEXRAD station MXX in eastern AL. The survey area encompasses Fort Benning.141 

Figure B20. Composite maps indicating spring and fall migratory hotspots as recorded by 

NEXRAD station NKX in San Diego, CA. The survey area encompasses Camp Pendleton 

Marine Corps Base.142 

Figure B21. Composite maps indicating spring and fall migratory hotspots as recorded by 

NEXRAD station PDT in Pendleton, OR. The survey area encompasses the Boardman 

Naval Bombing Range.143 




















ERDC/EL TR-12-22 


x 


Figure B22. Composite maps indicating spring and fall migratory hotspots as recorded by 
NEXRAD station POE in central, LA. The survey area encompasses Fort Polk.144 

Figure B23. Composite maps indicating spring and fall migratory hotspots as recorded by 
NEXRAD station PUX in Pueblo, CO. The survey area encompasses Fort Carson Military 
Reservation.145 

Figure B24. Composite maps indicating spring and fall migratory hotspots as recorded by 
NEXRAD station RAX in Raleigh-Durham, NC. The survey area encompasses Fort Bragg.146 

Figure B25. Composite maps indicating spring and fall migratory hotspots as recorded by 
NEXRAD station RGX in Reno, NV. The survey area encompasses the Sierra Army Depot.147 

Figure B26. Composite maps indicating spring and fall migratory hotspots as recorded by 
NEXRAD station TWX in Topeka, KS. The survey area encompasses Fort Riley Military 
Reservation.148 

Figure B27. Composite maps indicating spring and fall migratory hotspots as recorded by 
NEXRAD station TYX in northern NY. The survey area encompasses Fort Drum.149 

Figure B28. Composite maps indicating spring and fall migratory hotspots as recorded by 
NEXRAD station VBX in western CA. The survey area encompasses Vandenberg Air Force 
Base.150 

Figure B29. Composite maps indicating spring and fall migratory hotspots as recorded by 
NEXRAD station YUX in south-western AZ. The survey area encompasses Barry M. 

Goldwater Air Force Range and the Yuma Proving Ground.151 


Tables 

Table 1. Dates of fall migration bird surveys conducted at all sites around Eglin Air Force 
Base and Fort Polk from 2005-2007. Sites at Eglin AFB include Alaqua Creek (ALCR), 

Basin Creek (BACR) and Ten Mile Creek (TMCR); sites at Fort Polk include Bundick Creek 

(BUCR), Drakes Creek (DRCR), Six Mile Creek (SMCR), Upland Transect (UPTR), and 

Whiskey Chitto (WHICH). An “x” indicates that morning surveys were conducted on that 

date while an “o” indicates that only evening surveys were conducted.24 

Table 2. Dates of spring migration bird surveys conducted at all sites around Yuma and 
Fort Polk from 2006-2007. Sites at Yuma include All-American Canal (AAC), Imperial (IMP) 
and Yuma (YUMA); sites at Fort Polk include Bundick Creek (BUCR), Drakes Creek (DRCR), 


Six Mile Creek (SMCR), Upland Transect (UPTR), and Whiskey Chitto (WHCH). An “x” 

indicates that morning surveys were conducted on that date while an “o” indicates that 

only evening surveys were conducted.25 

Table 3. Migratory classification for all species recorded on ground transects during 

migration surveys at Eglin AFB, Fort Polk and Yuma (Poole 2005).28 

Table 4. Distribution and vegetation composition of 125 m transect sections from three 

sites 9 near Yuma, AZ that were surveyed for spring migrants in 2006 and 2007.34 

Table 5. DoD military installations greater than 200 km 2 and located within 120 km of 
NEXRAD stations. The columns labeled “Spring” and “Fall” indicate whether or not that 
installation served as a stopover hotspot for migrating birds. It was not possible to 


generate appropriate maps for some installations in some seasons due to complications 


with weather and beam blockage by proximal mountain ranges.36 

Table 6. Summary of the sampling effort and number of birds detected at each region by 

year and season. The N for surveys is the total number of 500m transects completed 

during each season; distance sampled is reported in km.49 

















ERDC/EL TR-12-22 


xi 


Table 7. Number of ground and flyover detections per species recorded during 1271 

morning and three evening transect surveys conducted during spring migration at Yuma 

in 2006-2007. Species highlighted in bold were included in all statistical analyses.49 

Table 8. Number of ground and flyover detections per species recorded during 981 

morning and 469 evening transect surveys conducted during spring migration at Ft. Polk 

in 2006-2007. Species highlighted in bold were included in all statistical analyses.52 

Table 9. Number of ground and flyover detections per species recorded during 426 
morning and 262 evening transect surveys conducted during fall migration at Ft. Polk in 
2005-2006. Species highlighted in bold were included in all statistical analyses.54 

Table 10. Number of ground and flyover detections per species recorded during 764 
morning and 448 evening transect surveys conducted during fall migration at Eglin AFB in 
2005-2007. Species highlighted in bold were included in all statistical analyses.55 

Table 11. Sample sizes for regression models built to explain A migrant abundance as a 

function of peak migration densities and migrant exodus densities captured on radar 

during spring and fall migration at three military installations.62 

Table 12. Parameter estimates 9 (± standard errors b ) and fit statistics for linear regression 

models built to explain the daily change in migrant abundance recorded during bird 

surveys as a function of peak migration densities captured on radar..62 

Table 13. Parameter estimates 9 (± standard errors b ) and fit statistics for linear regression 

models built to explain the daily change in migrant abundance recorded during bird 

surveys as a function of migrant exodus densities captured on radar..64 

Table 14. Parameter estimates 9 (± standard errors b ) and fit statistics for linear regression 
models built to explain daily positive migrant turnover recorded during bird surveys as a 
function of peak migration densities captured on radar..66 

Table 15. Parameter estimates 9 (± standard errors b ) and fit statistics for linear 

regression models built to explain daily negative migrant turnover recorded during bird 

surveys as a function of migrant exodus densities captured on radar.68 


Table 16. Mean abundance per kilometer for all migrant species recorded at 125 m 
transect sections of various habitat types during spring migration near Yuma Proving 
Ground in 2006 and 2007. The 20 most abundant species are indicated with an asterisk, 


and were tested for statistical differences among habitat types. Habitat types that were 

not statistically different from one another with regard to abundance of a species share a 

letter. It was not possible to construct appropriate habitat models for Orange-crowned 

Warbler or Bullock’s Oriole.76 

Table Al. Summary of banding effort and captures of migratory species in pine, mixed 

and hardwood habitat in Kisatchie National Forest from 27 March through 5 May 2006.101 

Table A2. Number of migrants released in pine, mixed or hardwood that left those habitat 

types during the first day of stopover and moved pine, mixed, hardwood habitat or 

another habitat type.107 


Table A3. Comparison of the relative influence of generalized linear models in predicting 

the movement rate and linear displacement of red-eyed vireos. The number of 

parameters (K), differences in AlCc values (AAlCc) and Akaike weights (w,) are shown for 

all top models (AAlCc < 4) as well as the null model. Models with AAlCc < 2 considered 

equally plausible. Results shown for all hours combined and for each two hour period of 

the day. Two outliers were removed from hour 2 (6:30 to 8:30).108 

Table A4. Relative importance and model-weighted averaged parameter estimates (when 
parameter was included in more than one supported model) of parameters included in 
top explanatory models (AAlCc < 2) for movement rate and linear displacement of red- 
















ERDC/EL TR-12-22 


xii 


eyed vireos. The conditional 95% confidence interval is calculated for parameters 

included in more than one top model.109 

Table A5. Relative influence of generalized linear models in predicting the time during 
the first five hours spent in the release habitat type for migrants released 1) at the 
same location in pine with and without playback of conspecific song, 2) in hardwood 
and pine in years without added playback (2007 & 2008) and 3) in hardwood and pine 
in years with added playback of conspecific song (2009). Number of parameters (K), 
differences in AlCc values (AAlCc), and Akaike weights (wi) are shown. All top models 
(MICc < 2) and the null model are shown. Parameter estimates and standard errors for 
variables influencing the time in release habitat during the first five hours after release.112 

Table A6. Comparison of generalized linear models comparing relative influence in 
predicting the time during the first five hours spent in pine for migrants released with and 
without playback added (Group). All top models (AAlCc < 2) and the null model are shown. 

Each set of models presented represents an addition of 3, 6 and 12 randomly selected 
pairs of individuals to each group. Number of parameters (K), differences in AlCc values 
(AAlCc), and Akaike weights (wi) are shown. All top models (AAlCc < 2) and the null model 
are shown.113 






ERDC/EL TR-12-22 


xiii 


Preface 

This report was prepared by Drs. Richard A. Fischer and Michael P. 
Guilfoyle, U.S. Army Engineer Research and Development Center- 
Environmental Laboratory (ERDC-EL); Jonathon J. Valente, Bowhead Inc., 
Vicksburg, MS; Dr. Sidney A. Gauthreaux, Jr., Carroll G. Belser, Donald Van 
Blaricom, and John W. Livingston, Clemson University Radar Ornithology 
Laboratory; and Drs. Emily Cohen and Frank R. Moore, University of 
Southern Mississippi (USM), Department of Biological Sciences. The 
authors wish to thank the Strategic Environmental Research and 
Development Program (SERDP) for providing the financial assistance for 
the project. Appreciation for technical assistance is extended to Mr. Bradley 
Smith and Dr. Jeffrey Marqusee, SERDP Executive Directors, former and 
present, and Drs. Robert Holst and John Hall, Sustainable Infrastructure 
Program Managers, former and present, and to the HydroGeoLogic, Inc., 
staff for their administrative assistance. The research would also not have 
been possible without the assistance of military base personnel granting 
access to the bases, helping to identify study sites, and providing site 
information, including Mr. Bruce Hagedorn and Ms. Kathleen Gault (Eglin 
Air Force Base, FL); Mr. Danny Hudson (Fort Polk, LA), and Mr. Randy 
English (Yuma Proving Ground, AZ). Thanks to Mr. Jim Johnson and Mr. 
Lynn Bennett for their logistical support and to Dr. Robb Diehl and Eben 
Paxton, USM, for help with analyses. The authors gratefully acknowledge 
field crew personnel who provided outstanding avian sampling during the 
course of the project: Archer Larned, Dr. Cliff Cordy, and Thomas Auer in 
Arizona; Beth Wright, Harley Winfrey, Sarah Winfrey, Kristina Baker, and 
Jordan Smith in Florida; and Phil Heavin, Emily Lain, Jaclyn Smolinsky, 
Kristin Comolli, Brian Wilson, Christine Roy, Christopher Nicholson, 
Shanna Everett, Lainie LaHaye, Dana Ripper, Christine Roy, Amy 
Scarpignato, Dave Haines, Mason Cline, Megan Hughes, Zoltan Nemeth, 
Clay Graham, Lisa Vormvold, Brian Bielfelt, and Pete Hosner in Louisiana. 

The report was reviewed by Drs. Eric R. Britzke and Nathan R. Beane, 
ERDC-EL. 

The study was conducted under the direct supervision of Tisa Webb, 
Branch Chief, Ecological Resources Branch, Dr. Edmond J. Russo, Chief, 



ERDC/EL TR-12-22 


xiv 


Environmental Evaluation and Engineering Division, EL; and Dr. Beth C. 
Fleming, Director, ERDC-EL. 

COL Gary E. Johnston was Commander and Executive Director of ERDC. 
Dr. Jeffery P. Holland was Director. 



ERDC/EL TR-12-22 


xv 


Acronyms 


AAC 

All-American Canal 

AFB 

Air Force Base 

AHAS 

Avian Hazard Advisory System 

AICc 

Akaike’s information criterion corrected for small 
sample size 

ALCR 

Alaqua Creek 

BACR 

Basin Creek 

BAM 

Bird Avoidance Model 

BASH 

Bird-Wildlife/Aircraft Strike Hazard 

BPCKM 

Birds per cubic kilometer 

BUCR 

Bundick Creek 

CUROL 

Clemson University Radar Ornithology Laboratory 

dBZ 

Decibels of reflectivity 

DoD 

Department of Defense 

DRCR 

Drakes Creek 

eBIRDRAD 

Enhanced Bird Radar 

EL 

Environmental Laboratory 

ERDC 

Engineer Research and Development Center 

GIS 

Geographic Information System 



ERDC/EL TR-12-22 


xvi 


GPS 

Global Positioning System 

hr 

hour 

IMP 

Imperial Valley 

INRMP 

Integrated Natural Resources Management Plan 

km 

kilometer 

m 

meter 

min 

minute 

NCDC 

National Climatic Data Center 

NEXRAD 

NEXt generation RADar (same as WSR-88D) 

ND 

Native-dominated with non-natives 

NND 

Non-native/invasive dominated with some natives 

NNI 

Non-native/invasive shrub and tree community 

NS 

Native shrubs 

NT 

Native trees 

PIF 

Partners in Flight 

SE 

Standard error 

SERDP 

Strategic Environmental Research and Development 
Program 

SMCR 

Six-mile Creek 

TMCR 

Ten-mile Creek 

UPTR 

Upland Transect 



ERDC/EL TR-12-22 


xvii 


USM 

University of Southern Mississippi 

WHCH 

Whiskey Chitto 

WSR-88D 

Weather Surveillance Radar-1988 Doppler 

YUMA 

Yuma transect site 

Z 

Absolute reflectivity 




ERDC/EL TR-12-22 


1 


1 Objectives 

This research project was performed with support from the Department of 
Defense (DoD) Strategic Environmental Research and Development 
Program (SERDP) to develop a radar-based monitoring strategy for 
migratory birds on military lands. The broad goal was to investigate migrant 
use of military installations as stopover habitat using ground-based surveys 
and Doppler weather surveillance radar (WSR-88D, also known as next 
generation radar [NEXRAD]) data. Specific objectives were to: 

1. use WSR-88D radar data to identify military installations that consistently 
support large migrant populations; 

2. examine the timing of migration on two eastern military installations and 
one western installation; 

3. develop migration forecast models for those three installations by 
combining migrant density estimates with weather variables; 

4. compare radar-based estimates of migrant density with more traditional 
ground-based bird surveys on the three installations; 

5. examine how vegetation influences migrant habitat use along riparian 
systems in the Southwest; and 

6. investigate movement ecology and habitat use by migrant Red-eyed Vireos 
in the spring within stopover habitat on Ft. Polk, Louisiana (Appendix A). 

Radar information showing bird migration on and near military 
installations is critically important for the protection of habitats used by 
migratory birds during stopover periods. This information is vital to DoD 
land managers who protect stopover areas on military lands. Similarly, 
radar data are particularly important to land managers of military air 
stations where bird-wildlife/aircraft collisions threaten lives and cost 
millions of dollars in damages to aircraft infrastructure every year. By 
identifying where, when, how long, and in what concentrations migratory 
birds inhabit temporary stopover sites or pass above military training 
airspace, affected installations will be able to improve both military 
readiness and species conservation. The work described in this document 
meets several of the objectives outlined in the DoD Partners in Flight (PIF) 
Strategic Plan (2002), including determining the current status of 
neotropical migratory bird populations on DoD lands, identifying and 
maintaining priority habitats on DOD lands for neotropical migratory bird 



ERDC/EL TR-12-22 


2 


populations, using radar technology as a BASH tool, improving 
communication with Air Operations personnel, improving hazard 
detection technology for pre-flight planning, and using information 
collected by partners to better support DoD mission requirements. 



ERDC/EL TR-12-22 


3 


2 Background 

Approximately half of all bird species that nest in the United States are 
classified as neotropical migratory birds. These species, which include about 
340 species of songbirds, shorebirds, waterfowl, and birds of prey, move 
annually between their breeding grounds in North America and wintering 
areas in Mexico, Central America, South America, and the Caribbean. 
Seasonal bird migration is a time-consuming, energetically expensive 
behavior that imposes numerous risks on the survival of individuals, with 
potential implications for long-term viability of populations (Alerstam 
2003, Moore et al. 1995). Migrants are often under pressure to complete the 
migratory passage quickly to procure suitable breeding opportunities or 
high quality over-winter territories (Greenberg 1980, Francis and Cooke 
1986). Although many landbird migrants are capable of making spectacular, 
nonstop flights over ecological barriers, few migrants actually engage in 
nonstop flights between points of origin and destination. Instead, migration 
is divided into alternating phases of flight and stopover, with each stopover 
lasting a few hours to a few days. In fact, the cumulative amount of time 
spent at stopover sites far exceeds time spent in flight and largely deter¬ 
mines the total duration of migration (Alerstam 2003). Thus, migrants are 
highly dependent on the availability of high quality stopover habitats with 
sufficient food and cover resources for refueling and avoiding predators 
(Morrison et al. 1992, Moore et al. 1995, 2005). 

Complicating — and potentially compromising — this already complex 
process, are decades of urban growth and agricultural expansion that are 
fragmenting and eliminating key migratory stopover habitats, such as 
riparian areas (e.g., Askins et al. 1990, Askins 1993, Rich 2002), for many 
species of birds. In particular, over 90% of western riparian areas have been 
lost (Dahl and Johnson 1991, Noss et al. 1995). Coastal areas in general, and 
the Northern Gulf Coast in particular (Moore et al. 1995), are critical for 
many trans-Gulf neotropical migrants. Yet human coastal populations were 
projected to increase over 60% from 1990 to 2010 (Cullitan et al. 1990). 

Loss and degradation of migratory stopover areas are frequently cited as key 
contributors to the observed long-term declines of migratory landbirds 
(Hutto 1985, Askins et al. 1990, Moore et al. 1995). Conservation organiza¬ 
tions (e.g., National Audubon Society, Nature Conservancy) rank protection 
of important migration stopover areas in the United States as a very high 



ERDC/EL TR-12-22 


4 


priority. Furthermore, Donovan et al. (2002) suggested that mapping 
migration stopover areas should be one of the highest research priorities for 
the conservation of migratory birds. 

Objective 1: Migrant Use of Military Installations 

DoD military installations, which include nearly 30 million acres of land 
and water in relatively large unfragmented tracts, provide a diversity of 
high-quality habitats (e.g., grasslands, wetlands, riparian areas, early 
successional habitats, and mature forests) for millions of en route migratory 
birds annually, including a large number of threatened, endangered, and 
otherwise sensitive species (Martin et al. 2000). Collectively and indivi¬ 
dually, military installations represent unique and important resources that 
likely play a critical role in the health and viability of migratory bird 
populations within and beyond the installation boundaries in all regions of 
the country. In recognition of this, the DoD has become an active member 
of the Partners in Flight (PIF) initiative, a voluntary, international coalition 
of government agencies, conservation groups, academic institutions, private 
businesses, and everyday citizens dedicated to “keeping common birds 
common.” The main goal of PIF is to direct resources toward the conserva¬ 
tion of birds and their habitats through cooperative efforts in North 
America and the neotropics. The DoD PIF program has been instrumental 
in assisting installations with conservation planning, facilitating avian 
inventories, and providing critical information from the various bird 
conservation initiatives to DoD land managers. In addition, the DoD PIF 
Strategic Plan (2002) clearly provides DoD natural resources managers 
with information about how to integrate bird conservation into Integrated 
Natural Resources Management Plans (INRMPs). 

Unfortunately, very little is known about specific stopover sites on military 
lands that provide critical in-transit habitats for migrating birds. Because 
of the potential conflicts between the conservation of migratory birds and 
mission readiness, natural resource managers need to know if their base 
occurs along a major bird migration flyway or serves as an important bird 
migration stopover area. Since the early 1940s, radar has been used to 
monitor bird migration. Most bird migration occurs under the cover of 
darkness, and radar is one of the few means of detecting and quantifying 
such movements (Gauthreaux 1970). Recent technological advances in 
radar systems have proven invaluable for improving knowledge of 
migratory bird movements. The newest weather surveillance radar, WSR- 
88D or NEXRAD (NEXt generation RADar), is ideal for studies of 



ERDC/EL TR-12-22 


5 


migratory bird movements. This sophisticated radar system can be used to 
map geographical areas of high bird activity (e.g., stopover, roosting and 
feeding, and colonial breeding areas). It also provides information on the 
quantity, general direction, and altitudinal distribution of birds aloft. 

From 1990 through 1997, nearly 150 new Doppler weather surveillance 
radars were installed in the United States (Figure 1), providing nearly 
complete NEXRAD coverage. Some of these units are on DoD installations, 
providing an opportunity to collect site-specific data on bird movements 
and use of DoD lands. Since 1993, the Clemson University Radar 
Ornithology Laboratory (CUROL) has examined the effectiveness of using 
these radars to detect, quantify, and monitor flying birds (Gauthreaux and 
Belser 1998,1999, 2003a and b). Research at the CUROL has shown that 
radar data can greatly enhance the ability to monitor broad-level migration 
patterns, assess annual trends of migratory bird passage, determine 
geographical areas of high stopover use, and gather information on the 
quantity, speed, and altitude of flying birds (Gauthreaux and Belser 1998, 
2003b; Diehl and Larkin 2004). By combining classified Landsat imagery 
and data from WSR-88D radar images, CUROL is mapping the geogra¬ 
phical distribution of critical migratory stopover areas and is characterizing 
the habitat of these areas (Gauthreaux and Belser 2003b). The first 
objective was to use geographic information systems (GIS) as a screening 
tool to identify all military installations within 120 km of NEXRAD stations, 
and then use NEXRAD data to identify installations that serve as migration 
stopover hotspots in the spring and fall. 

Objectives 2 and 3: Quantifying Migration and Developing Migration 
Forecast Models 

In addition to conservation obligations, the DoD also has safety and 
financial incentives to be concerned with migratory bird populations on 
military installations. Reduction of bird-wildlife/aircraft strike hazards 
(BASH) is a critically important issue on installations that have an aviation 
mission. Military aircraft are subject to potential bird strikes, especially 
during takeoff/landing and during low-level missions, and the problem is 
particularly acute during period of dense bird migration in spring and fall. 
Collisions between birds and aircraft pose a serious threat to the safety of 
passengers and flight crews on both civilian and military flights. Since 
1985, nearly 38,000 bird-military aircraft collisions have been reported. 
These collisions have killed 33 pilots, destroyed 30 aircraft, and resulted in 
over $500 million in damage (Lovell and Dolbeer 1999). Between 1988 



ERDC/EL TR-12-22 


6 



Figure 1. WSR-88D stations in the contiguous United States. Edges of 
the circles are at 124 nautical mile range (230 km). 


and 2004, approximately 194 people were killed from bird-aircraft strikes 
(Dolbeer 2006). In the United States, collisions between aircraft and 
wildlife cost the aviation industry over $600 million annually (Cleary et al. 
2007), while global costs are estimated at $1.2 billion (Allan and Orosz 
2001). On average, the U.S. Air Force (USAF) experiences costs of 
$35 million annually on bird-aircraft collisions, while over the past 
20 years, total costs of $98 million have been estimated for aircraft 
collisions with Turkey Vultures (Cathartes aura ) alone (Kelly and Wilkens 
2006). The number of wildlife-aircraft strikes has increased from 1,700 in 
1990 to over 7,000 in 2006; total reported collisions during this period 
exceed 70,000 (Cleary et al. 2007). The number of bird-aircraft collisions 
have increased recently due to increased air traffic and population 
increases of several large birds, particularly the Canada Goose (Branta 
canadensis ) (Dolbeer and Eschenfelder 2005). This situation was recently 
highlighted by the well-publicized story of a commercial passenger aircraft 
that landed in the Hudson River (U.S. Airways Flight 1549) in January 
2009 after colliding with several Canada Geese (Langer 2009). 

Bird migration is a dynamic process. Although predictable in terms of 
general timing, the specific times for determining peaks in activity often 
depend on current weather conditions. Migrants are known to respond to 
frontal systems (Alerstam 2003) during both the spring and the fall. During 













ERDC/EL TR-12-22 


7 


spring, radar data suggests that during peak migration, as many as 80,000 
birds per mile may pass northward over some sections of U.S. coastlines 
each day. After a stopover along the coastline to replenish energy reserves 
expended during long-distance flight, these migrants slowly filter northward 
to their breeding grounds, sometimes taking as long as six weeks to reach 
breeding sites. Typically, these migrants travel in large flocks that emerge 
from stopover sites shortly after dusk. From late-summer to fall, southward 
migration back to wintering grounds often includes large flocks of young 
birds and adults. 

Currently, two tools are available to DoD airfield personnel to assess bird 
strike risk during flight planning: the U.S. Bird Avoidance Model (BAM) 
and the Avian Hazard Advisory System (AHAS). The BAM, which is based 
on 30 years of historical bird activity data (primarily Audubon Society 
Christmas Bird Count, U.S. Biologic Survey Breeding Bird Survey, and bird 
refuge arrival and departure data for the conterminous U.S.), is used as a 
tool for analysis, predictability, and correlation of bird habitat, migration, 
and breeding characteristics, combined with key environmental and 
geographic data (USBAM 2004). The AHAS is an online, near real-time, 
tool based in a geographic information system (GIS) platform. This system 
also uses WSR-88D weather radars and models developed to predict bird 
movement, monitor bird activity, and forecast bird strike risk. When radar 
data indicate bird activity in a particular area, AHAS uses the BAM to assess 
risk for a particular time period. Yet, in addition to the BAM, AHAS also 
looks at migration and soaring bird data to determine if the risk should be 
higher than that indicated by the BAM. Although the BAM and AHAS are 
useful tools currently used by the military to reduce BASH incidences, there 
is need for an improved system that can more accurately predict timing and 
volume of migrating birds on specific installations. CUROL has also 
conducted preliminary work that suggests multivariate statistical analyses 
of forecast weather variables and bird migration density measures can be 
used to develop statistical models that accurately predict the density of 
migration events (http://www.birdsource.org/BirdCast/home.html) . Thus, the temporal 
pattern of migration was first examined on three installations, and then 
constructed models that can be implemented through a desktop computer 
spreadsheet to predict migrant densities over those installations on any 
given evening. 




ERDC/EL TR-12-22 


8 


Objective 4: Comparison of Migrant Survey Techniques 

The Engineer Research and Development Center (ERDC), Environmental 
Laboratory (EL), has conducted numerous intensive inventory and 
monitoring efforts to document the distribution, abundance, and diversity 
of avian populations on numerous military installations throughout the 
country. These surveys typically have been done on a seasonal basis (i.e., 
winter, spring migration, summer breeding, and fall migration). Typically, 
migration surveys are conducted by personnel on the ground either walking 
transects or sampling from fixed points. Ideally, sampling during migration 
seasons should be conducted over the course of many weeks, but this is 
often difficult to accomplish with limited personnel and funding. Investi¬ 
gators often use best professional judgment to determine when peaks in 
migration are likely to occur in an area, and when to conduct field surveys 
for migrants. For these ground-based surveys, however, timing maybe such 
that peak passage of birds is missed or the approach may overlook some 
important areas or habitats that are used heavily by stopover migrants. 
Radar data can be a significant tool to assist in improving the timing and 
location of ground-based migration surveys, and in identifying the impor¬ 
tance of military lands to migrant birds. While WSR-88D data are limited in 
resolution and complex in interpretation, the technology allows users to 
investigate migratory patterns on spatial and temporal scales that are not 
logistically or economically feasible with field surveys. Still, WSR-88D data 
have the potential to be a more efficient method than field survey data due 
to a number of biases associated with the former (Diehl and Larkin 2004). 

Few studies have actually compared indices of bird abundance on the 
ground with data from radar observations, and those that have done so have 
yielded mixed results. Buler and Diehl (2009) actually found a strong 
positive correlation between radar estimated bird densities and densities 
observed on the ground along transect surveys. A number of other investi¬ 
gations have shown a similar relationship between mist-net data and radar 
observations (Simons et al 2004, Peckford and Taylor 2008, Buler and 
Diehl 2009). However, mist-net capture data from multiple sites in 
California failed to detect significant relationships with radar data 
(DiGaudio et al. 2008). A lack of consistent correlation between on-the- 
ground field observations and radar may reflect a disparity in the resolution 
of the methods (see Diehl and Larkin 2004, Buler and Diehl 2009) and 
warrants further investigation. Thus, the investigation consisted of the 
relationship between abundance of transient migrants detected on the 
ground and measures of bird density in the atmosphere as determined 



ERDC/EL TR-12-22 


9 


through WRS-88D radar on or within the vicinity of three military 
installations. The results provide an opportunity compare and contrast the 
two methods of monitoring migrants to ultimately improve future sampling 
efforts on military installations. 

Objective 5: Avian Habitat Use in Southwestern Riparian Systems 

Lastly, in the arid southwestern United States, the DoD manages nearly a 
third of its total acreage (9 million ac). Much of this acreage is covered with 
a network of perennial, intermittent and ephemeral streams and drainages 
with relatively high-quality riparian habitat. Riparian areas typically 
comprise a small component of landscapes, especially in the southwestern 
United States where they are less than 1 percent of the total land area. 
However, they are used by more species of breeding birds than any other 
habitat in North America (Knopf et al. 1988, Stevens et al. 1977). The broad 
importance of riparian areas in providing stopover habitat, especially in the 
arid Southwest, is well documented (Yong and Finch 1997; Skagen et al. 
1998; Kelly et al. 2000). While width of riparian areas can strongly 
influence breeding bird community composition (though little information 
is available from the Southwest; Fischer 2000), research suggests that 
riparian habitats in this region are extremely important as stopover habitat, 
regardless of width or extent (Skagen et al. 1998). Some species may even 
adjust their migratory routes in order to maintain proximity to riparian 
habitats (Skagen et al. 2005). However, there is far less information 
available as to how migrants respond to finer scale differences in habitat 
variables within riparian areas such as vegetation structure and community 
composition (Carlisle et al. 2009). It is especially important to understand 
the influence of such variables on migrant communities in this region where 
human effects have, more often than not, heavily impacted riparian 
systems. 

Historically, riparian plant communities in the arid and semi-arid western 
United States were dominated by native willow (Salix spp.) and cottonwood 
(Populus spp.) trees. However, most riparian ecosystems in the West have 
been significantly degraded, with losses approaching 99 percent in some 
areas (Briggs 1996). Hydrologic modifications such as dams and water 
withdrawals for irrigation on most rivers and streams have heavily altered 
the natural functions and processes on these systems. Complicating the 
matter is the invasion of Saltcedar (Tamarix sp.), which possesses a number 
of weedy qualities that allow it to quickly out-compete native vegetation. In 
a study of riparian plant communities in the Gila and lower Colorado 



ERDC/EL TR-12-22 


10 


drainage basins of Arizona, Stromberg et al. (2007) found that reaches with 
a natural flood regime were dominated by willow and cottonwood, while 
those where natural spring flood pulses had been eliminated had some of 
the highest abundances of Saltcedar. Saltcedar is native to Eurasia and was 
introduced into the U.S. in the 1800s as a firewood source and for erosion 
control; it is now the second most dominant woody riparian tree in the 
western U.S. In the Lower Colorado River, Saltcedar has become the 
dominant riparian plant species, having replaced native vegetation in 
approximately 500,000 ha of riparian habitat (Zavaleta 2000; Friedman et 
al. 2005) and is expanding within western riparian areas at the rate of 
50,000 to 60,000 acres per year (Laccinole 2009). 

Research has thus far yielded little evidence that Saltcedar-dominated 
riparian corridors negatively impact breeding birds in comparison to those 
dominated by native vegetation (Fleishman et al. 2003; Durst 2004). Sogge 
et al. (2008) identified 49 different bird species that have been documented 
utilizing saltcedar during the breeding season in the U.S. and Sogge et al. 
(2005) actually showed a positive link between saltcedar coverage and the 
abundance and diversity of breeding bird communities. While bird 
abundance and density can be higher in riparian corridors dominated by 
native willow and cottonwood trees (Hunter et al. 1988), several researchers 
have suggested that these plants are no longer viable in the region due to 
widespread alterations in river hydrology, and thus riparian areas would 
actually have less value for breeding birds were it not for the presence of 
Saltcedar (Livingston and Schemnitz 1996; Sogge et al. 2008). 

Less is known about the value of various riparian plant communities as 
migratory bird stopover habitat. Skagen et al. (1998) deduced from their 
findings that all riparian areas dominated by native trees were important for 
migrants regardless of patch size or degree of isolation, yet they did not 
compare these areas to those dominated by other vegetation types. Hardy et 
al. (2004) showed that an overwhelming proportion (97%) of the migrants 
passing through a riparian system in the Sonoran Desert were utilizing 
xeroriparian vegetation over creosote bush (Larrea fndenfafa)-bursage 
(Ambrosia sp.), mixed cacti, or rock and cliff communities. However, 

Paxton et al. (2008) showed that Wilson’s Warblers (Wilsonia pusilla) 
actually base their selection of stopover sites on the flowering phenology of 
plants, rather than on the plant species themselves. Still other southwestern 
U.S. work has documented that migrant abundance is strongly associated 
with riparian species composition and that the highest abundances were 
recorded in areas dominated by saltcedar (Walker 2008). 



ERDC/EL TR-12-22 


11 


To meet this objective, this study also investigated how communities of 
spring migrants utilizing riparian stopover habitat near the U.S. Army 
Yuma Proving Ground in southwestern Arizona respond to differences in 
plant community composition. Specifically, differences in migrant 
abundance and richness among plant communities were identified, as 
were differences in migrant use of areas dominated by saltcedar versus 
native vegetation. The results will help natural resources managers 
develop management objectives and prioritize important migrant stopover 
habitats for conservation on southwestern military installations. 

Objective 6: Movement Ecology and Habitat Use of a Neotropical 
Migrant during Spring Migratory Stopover 

Due to the hazardous nature of migratory journeys, neotropical migrants 
require high-quality stopover habitat where they may rest and forage safely 
to refuel their energy stores. To date, little is known about how migrants 
identify, select, and move among habitats with varying resource availability. 
Understanding how migrants identify stopover areas and when or why they 
choose to move among habitat types is critical to improving researchers’ 
ability to provide quality stopover sites on DoD lands. To meet this 
objective, migrating Red-eyed Vireos were translocated (Vireo olivaceous ) 
to stopover sites on Ft. Polk during the spring of 2007 and 2008 and four 
hypotheses were investigated: 1) migrants stopping over in a heterogeneous 
landscape move to select high-quality habitat; 2) movement within a habitat 
type is related to the quality of that habitat type; 3) movement during 
stopover is related to the energetic condition of the bird; and 4) the duration 
of stay at a stopover site is related to the time of the season and the 
energetic condition of the bird. The effects of conspecific social cues on Red¬ 
eyed Vireo stopover site selection on Ft. Polk in the spring of 2009 were also 
investigated. Due to the uniqueness of this part of the project, these 
experiments are detailed separately in Appendix A. 



ERDC/EL TR-12-22 


12 


3 Materials and Methods 

Objective 1: Migrant Use of Military Installations 

The easiest way to identify migratory stopover hotspots with radar is to 
examine exodus data collected in the early evening (from 45 minutes to 
2 hours after sunset). When large, high density exodus events are recorded 
consistently in a location, that area is assumed to provide important 
stopover habitat. Because WSR-88D radars only reliably record exodus 
events within a 120 km radius, the first step was to identify all military 
installations within the effective survey area of NEXRAD stations. Using 
ArcView 3.2, shapefiles were overlaid outlining all military installations in 
the United States, and 120 km buffer regions around all NEXRAD stations. 
On 26 May 2005, Dr. Richard Fischer visited the CUROL and reviewed the 
output of this analysis. Smaller installations (< 200 sq. km) were eliminated 
from the potential base pool to concentrate on potential stopover habitats 
on larger installations. Drs. Fischer and Gauthreaux selected the 40 largest 
installations that met these criteria for generating preliminary stopover 
areas in relation to base boundaries (see results). For each of these 40 
installations, the CUROL used archived radar data for the spring migration 
periods of 2000 and 2001, and the fall migration periods of 2003 and 2004 
to characterize patterns of bird migration. All data were downloaded from 
the National Climatic Data Center (NCDC) in Ashville, North Carolina. 

To determine whether birds were responsible for the reflectivity in the 
display of the WSR-88D, the methods described in Gauthreaux and Belser 
(1998,1999, 2003a) and Gauthreaux et al. (2008) were followed. First, the 
base reflectivity and velocity files were examined for each night and those 
with precipitation in the sample area or with other issues such as excessive 
strobes and anomalous propagation of the radar beam were eliminated. In 
each case, it was noted whether migration could be observed outside the 
sample area where precipitation was not occurring. All information was 
recorded in a spreadsheet with a row for each evening. The series of saved 
base reflectivity and base velocity files without precipitation in the sample 
area was examined for each evening and the files showing peak density were 
selected and those with excessive insect contamination and other aerial 
reflectors were eliminated (e.g., smoke and dust particles). The last task was 
accomplished by comparing winds aloft with the mean speeds of targets in 
the base velocity images. The winds aloft data were obtained from the 



ERDC/EL TR-12-22 


13 


archive of the Department of Atmospheric Science at the University of 
Wyoming. This dataset is from the Universal RAwinsonde OBservation 
program (RAOB; Environmental Research Services, Matamoras, 
Pennsylvania) sounding plots and is available from 1973 to present for 
sounding plot locations. These atmospheric observations are measured by 
radiosonde near the beginning (00:00 Universal Time Coordinate [UTC]) 
and the end of the night (12:00 UTC). Text files were saved and converted 
into Excel spreadsheet files containing: date, time (UTC), altitude (m), wind 
speed (m per second [sec-i], and wind direction (from). 

If the maximum mean radial velocity of the targets was within 5 m sec 1 
(18 km per hour [hr 1 ]) of the velocity of following winds aloft, the 
associated base reflectivity file was eliminated from additional analysis 
because of the likelihood of insect contamination. Likewise, if winds aloft 
were calm and the base velocity file showed no mean radial velocities in 
keeping with velocities of songbirds (32.4 to 54 km hr 1 [17.5 knots to 
29.2 knots], Bruderer and Boldt 2001), the associated base reflectivity file 
was eliminated. If the maximum mean radial velocity of the targets was in 
keeping with birds and the direction of flight was not in the direction of the 
winds aloft, the base reflectivity file was saved for further analysis. Surviving 
mean base velocity information at least 5 m sec 1 above wind speed was used 
to determine the direction of flight and to determine the relative reflectivity 
values (decibels of reflectivity [dBZ]) in associated relative reflectivity pixels 
(i° x 1 km pulse volumes or resolution cells). Taking this conservative 
approach made it possible to measure the difference between the maximum 
relative reflectivity in a reflectivity file and the maximum relative reflectivity 
of the pixels that had velocities in keeping with bird flight speeds. The 
former is generally greater than the latter, but in some cases, the two 
measures are the same. A difference between the two measures indicates 
that insects and other particulates in the atmosphere are contributing to the 
relative reflectivity. Base reflectivity files with little or no reflectivity from 
insects or other particulates in the atmosphere were processed with an 
algorithm designed to measure the base reflectivity of each surviving pixel. 
The base reflectivity values were converted to Z values (absolute reflectivity) 
and then used to compute birds per cubic kilometer [knrs] (Z-value 
multiplied by 1.84, Gauthreaux et al. 2008). The conical beam of the WSR- 
88D is 1° in diameter and the center of the beam is tilted 0.5 0 above the 
horizontal for the lowest scan. Bats and migrating birds are very similar in 
size and shape. They are essentially indistinguishable from one another on 
reflectivity returns of the 88-D radar. However, there is little to no 



ERDC/EL TR-12-22 


14 


contamination of bird migration by bats in the reflectivity images used for 
analyses. Bats comprise <2% of the total number of individuals that pass 
through the disk of the moon during migration observation periods 
(Gauthreaux, unpublished data). Most of the bats that are observed during 
moon watching are actually foraging; thus, they have reflectivity but no 
radial velocity. Therefore, most bats are actually filtered out with algorithms 
and probably do not contaminate these estimates of migrating bird densities 
on radar. Second, the relative number of bats migrating on a given night is 
likely very small, and again, contributing very little to the reflectivity of 
migrants aloft. Future technology may allow better discrimination between 
the two and provide a more quantitative estimate of the relative proportion 
of bats aloft. 

The majority of analyses were completed using level-III data, but level-II 
data was used where available. Level-III data have a range resolution of 1 
km for both reflectivity and radial velocity, while for level-II data, these 
values are 250 m. With higher resolution Level-II data, it was possible to 
sample the mean radial velocity of each pulse volume or resolution cell in a 
volume of atmosphere and discriminate differences in the flight speeds of 
the different types of reflectors moving through the atmosphere, making it 
easier to separate birds from other particles. The resulting data from each 
evening in each location were combined into a rectangular raster for 
importing into GIS. Once in GIS, the data were mapped and displayed as 
deviation from the mean. Stopover areas with consistently large densities 
of migrants show high positive deviations above mean density while areas 
of lower density show low positive deviation above mean density. Spring 
and fall composite maps were generated for each of the 40 installations 
(Appendix B). Migrant exodus densities were qualitatively assessed over 
each installation in both seasons and identified those that contained 
stopover hotspots (defined as > 2 standard deviations above mean migrant 
density). It was not possible to generate appropriate composite maps for 
some site-by-season combinations due to lack of reliable data (see results). 

Objective 2: Quantifying Seasonal Migration 

After assessing radar data for each installation, two installations were 
selected for each detailed spring and fall migration analyses (Fort Polk, 
Louisiana; and Eglin AFB, Florida for fall migration; Yuma Proving Ground, 
Arizona; and Fort Polk for spring migration). These installations were 
selected because of the consistent patterns of exodus events during the 
respective migration seasons, and the relative ease of access to field sites 



ERDC/EL TR-12-22 


15 


suitable for establishment of transects within the surveillance area of the 
nearest radar station (see Objective 4). Originally, it was proposed to 
examine eastern and western study sites during both fall and spring 
migration seasons; however, the fall migration data on western military 
installations examined in this project appeared too dispersed and variable 
through the season to clearly identify fall migrant hot spots. Instead, a 
second eastern site (Eglin AFB) was selected for comparison during the fall 
migration season with data collected from Fort Polk. Past work on Eglin 
AFB (Tucker et al. 2003) suggested this site was not a significant stopover 
site for spring migrants; thus, it was only examined during fall. 

The first step for each of these individual installations was to investigate 
seasonal patterns of nocturnal bird migration so that the temporal scope 
for building forecast models could be limited and ground-based surveys 
could be conducted. However, rather than examine the same exodus data 
used to identify these stopover hotspots, peak nightly migration densities 
over each installation were quantified, as this would be a better indicator 
of regional migration activity. The data were processed and analyzed as 
described above for the exodus data. 

For the WSR-88D station (KEVX) at Eglin Air Force Base, Florida data for 
the fall migration period (15 August to 31 October) for the years 2000-2005 
were downloaded and used. The archive at NCDC has Level-Ill data for 
KEVX from 15 August 1994 to present and Level II data from 1 May 2001 to 
present. Level-Ill data were analyzed from the NCDC archive at a 0.5 0 
antenna elevation angle for the fall of 2000 and then analyzed Level-II data 
for the fall migrations of 2001-2005. Only rarefy were scans at 1.5 0 used 
when 0.5 0 scans were not suitable because of beam bending and strobe 
patterns on the display. 

For the WSR-88D station (KPOE) at Fort Polk Army Base, Louisiana, data 
for the spring and fall migrations periods (15 March-30 May, 15 August-15 
November) for the years 2000-2005 were downloaded and used. The 
archive at NCDC has level-III data for KPOE from 1 May 2001 to present 
and Level-II data from 28 June 1995 to 13 December 2001. Level-II data 
for the years 2000 and 2001 were analyzed and Level-III data for the years 
2002-2005 were used. Data with antenna elevations of 0.5 0 were used. 
When excessive beam bending downward or a strobe pattern in the display 
occurred, data with a higher antenna elevation were used. 



ERDC/EL TR-12-22 


16 


For the WSR-88D station (KYUX) at Yuma, Arizona, data for the spring 
migrations period (15 March-30 May) for the years 2000-2005 were 
downloaded and used. The archive at NCDC has level-III data for KYUX 
from 26 July 1996 to present and Level-II data from 29 July 1996 to 
present. Level-II data for the years 2000 through 2005 were analyzed and 
Level-III data for the same years were also used. Data with antenna 
elevations of 0.5 0 were used. When excessive beam bending downward or 
a strobe pattern in the display occurred, data with the next higher antenna 
elevation (e.g., i.45°-i.5°) were used. There was topographic blockage of 
the radar beam at this location that restricted the team’s analysis to areas 
without beam blockage. 

Objective 3: Development of Migration Forecast Models 

The team used the peak nightly migration data (described above) as the 
response variables for migration forecast model development, and 
explanatory variables were collected by downloading archived weather data 
from either the National Climatic Data Center (NCDC) in Asheville, North 
Carolina or from Weather Underground (http://www.wunderground.com) . The 
weather data were gathered for 10 PM local time and included the following 
weather variables: temperature, dew point temperature, humidity, 
barometric pressure, surface wind direction and speed, 925mb wind 
direction and speed, precipitation, and cloud cover. These weather variables 
are commonly forecasted by the National Weather Service and can be 
entered into a migration forecast model to predict bird migration densities. 
Because wind direction is a circular variable, the wind direction and speed 
variables were converted to head wind and cross wind variables using a 
wind component calculator (http://www.aeroplanner.com/calculators/avcalcdrift.cfm) . For 
fall, the course was entered as 180 (toward the south), and for spring the 
course was entered as 360 (toward the north). Weather data were 
downloaded for the period 15 August through 15 November 2001-2005 for 
Eglin Air Force Base and Ft. Polk and for the period 15 March through 31 
May 2001-2005 for Yuma and entered into an Excel spreadsheet. 

The statistical software JMP 8 (SAS Institute Inc., Cary, NC) was used to 
generate the predictions of migration density based on neural nets. Under 
the “Analyze” tab, Modeling > Neural Net was selected. Bird density (birds 
per km 3 ) was entered as the response variable and ordinal date, tempera¬ 
ture, dew point temperature, humidity, barometric pressure, surface head 
wind and cross wind, 925mb head wind and cross wind, amount of 
precipitation, and cloud cover were loaded as factors (input variables). 





ERDC/EL TR-12-22 


17 


Precipitation was not loaded in the neural net analysis for Yuma, because 
only one date had rain. Five-fold cross validation was selected. This 
procedure generates a cross-validation R 2 and the closer the Coefficient of 
Variation (CV) R 2 is to the neural net R 2 , the better the prediction ability of 
the neural net model. In the control panel the following were selected: 
hidden nodes 3 (Hi, H2, and H3 in Figure 2), overfit penalty 0.01 minimizes 
issues from overfitting and the effects of multicolinearity, number of tours 
20, maximum iterations 75, and converge criterion 0.00001. Additional 
details and explanations for the application of neural net algorithms in JMP 
8 can be found in the JMP White Paper by Marie Gaudard (2008). 


Ordinal date 
~[Temp C 

De w Pt C 
— Hum 

SLPress 
Shead 
Scross 
925 Head 
925 Cross 
Sky 

Figure 2. Diagram of neural net showing input variables, three nodes and 

output variable. 

After application of the neural net algorithm, the option “Save Formulas” 
was selected. This added four new columns to the data table: the formulas 
for the three hidden nodes and a formula for predicted birds per km 3 . When 
new weather variables are added to the data table, a new prediction of bird 
density will result when the cursor is in the empty cell in the predicted birds 
per km 3 column. With JMP software and the data tables, the forecast 
weather variable can be used to forecast bird migration density. 

Objective 4: Comparison of Migrant Survey Techniques 

Bird Surveys 

Three to five different riparian areas (sites) were identified on or adjacent to 
each installation (Yuma Proving Ground, Ft. Polk, and Eglin AFB) to be 
surveyed simultaneously using line-transects. Bird sampling was focused 
along riparian areas because numerous radar studies and field surveys have 
shown eastern (Gauthreaux and Belser 1998, 2005) and western (Skagen et 
al. 1998; Kelly and Hutto 2005) migrants to be highly dependent on 
riparian habitats during migration. Sites were selected based on radar 
information indicating stopover hotspots. At each site, 5-7 500m transects 
were established and each was comprised of a numbered start and end point 
(i.e., waypoints) (see Figures 3-16). 

















































ERDC/EL TR-12-22 


18 



Figure 3. Three riparian transects established in Fall 2005 at Eglin Air Force Base, FL. 



Figure 4. Transect endpoints along Tenmile Creek, Figure 5. Transect endpoints along Basin Creek, Eglin 
Eglin AFB, FL. AFB, FL. 







































ERDC/EL TR-12-22 


19 



Figure 6. Transect endpoints along Alaqua Creek, Eglin AFB, FL. 



Figure 7. Four riparian transects and one upland transect established in during 2005-2007 at 
Fort Polk, LA. The upland transect extends northwest from Drakes Creek. 
















ERDC/EL TR-12-22 


20 



Figure 8. Transect endpoints along Bundick 
Creek, Fort Polk, LA. 



Figure 9. Transect endpoints along Drakes 
Creek, Fort Polk, LA. 



Figure 10. Transect endpoints along Six Mile Creek, Fort Polk, LA. 


m 















ERDC/EL TR-12-22 


21 



Figure 11. Transect endpoints along the Figure 12. Transect endpoints along Whiskey Chitto Creek, Fort 
upland transect, Fort Polk, LA. Polk, LA. 



Figure 13. Three riparian transects established in spring 2006 
near Yuma, AZ and Imperial National Wildlife Refuge, AZ. 

































ERDC/EL TR-12-22 


22 



Figure 14. Transect endpoints along All American Canal, 
Yuma, AZ. 


Figure 15. Yuma transect endpoints along the Colorado River, 
Yuma, AZ. 



Figure 16. Imperial transect endpoints along the 
Colorado River, Imperial National Wildlife Refuge, AZ. 















































ERDC/EL TR-12-22 


23 


Experienced birders conducted simultaneous transect surveys at sites from 
fall 2005 through spring 2007; however, one technician left early in the 
2006 season at Yuma, and one site (Imperial Valley) was not surveyed that 
year. In addition, one technician left early during the spring 2007 season at 
Fort Polk; therefore, the surveys were conducted by two birders at Polk 
during that season. Each morning, field crews began monitoring at or 
shortly after sunrise, with each surveyor at a different site. Each surveyor 
used a handheld GPS unit preloaded with numbered waypoints that 
denoted the beginning and end of each 500 m transect. The team recorded 
the number and species of all birds detected. If a bird could not be identified 
to species, it was categorized to the lowest taxonomic level possible in order 
to separate and assess all potential migrants observed during the surveys. 
The team used transect data sheets that diagrammatically depicted the 
transect being surveyed, with beginning and end points indicated, plus a 
main line to transverse while walking the transect. The approximate loca¬ 
tion of any birds detected along each transect were noted (e.g., whether the 
bird was to the left or right of the surveyor, the approximate distance of the 
bird to the surveyor, and how far along the transect (0-500 m) the bird was 
detected). Each 500 m transect was completed in approximately 30 min¬ 
utes; therefore, each site survey could be completed in about three hours 
(surveys usually ended at approximately 10:00 am local time). Surveys were 
conducted >5 days a week (except in cases of inclement weather), and 
surveyors rotated among sites on a daily basis. A brief description of the 
areas surveyed on each installation, plus a short history of transects 
sampled are provided below. Sampling was conducted during a wide range 
of dates to capture peak bird migration at each site during all migration 
seasons (Tables 1 and 2). 

Eglin Air Force Base, Florida 

Three sites were selected along three different drainages on Eglin AFB, 
including Alaqua Creek (ALCR), Basin Creek (BACR), and Ten-mile Creek 
(TMCR). All transects at these sites were established in fall 2005 and 
included seven continuous waypoints each separated by 500m of habitat 
(see Figures 3-6). All transects on Elgin AFB were placed in drainages with 
well-defined, transitional habitats between upland and floodplain forests. 
Uplands were typically dominated by extensive longleaf pine (Pinus 
palustris ) sandhills characterized by an open, savanna-like structure with 
a moderate to tall canopy of longleaf pine, a sparse midstory of oaks and 
other hardwoods, and a diverse groundcover comprised mainly of grasses, 
forbs and low stature shrubs (Eglin Air Force Base, 2007). Riparian areas 



ERDC/EL TR-12-22 


24 


were comprised of a wide variety of hardwood tree species of Magnolia 
(Magnolia spp.), sweet gum (Liquidambar styraciflua), poplar (Populus 
spp.), hickory (Carya spp.), ash (Fraxinus spp.), and maple (Acer spp.). 
Although some minor changes were made in specific locations of some 
waypoints because of flooding along Basin and Alaqua Creeks, total 
distance covered and the number of transects remained consistent during 
the survey period. 


Table 1. Dates of fall migration bird surveys conducted at all sites around Eglin Air Force Base 
and Fort Polk from 2005-2007. Sites at Eglin AFB include Alaqua Creek (ALCR), Basin Creek 
(BACR) and Ten Mile Creek (TMCR); sites at Fort Polk include Bundick Creek (BUCR), Drakes 
Creek (DRCR), Six Mile Creek (SMCR), Upland Transect (UPTR), and Whiskey Chitto (WHCH). 
An "x” indicates that morning surveys were conducted on that date while an “o” indicates that 

only evening surveys were conducted. 


Region 

Site 

September 










October 











28 

29 

30 

1 

2 

3 

4 

5 

6 

7 

8 

9 

10 

11 

12 

13 

14 

15 

16 

17 

18 19 

20 

21 


ALCR 







X 

X 

X 

0 

X 

X 

X 

X 

X 

X 

X 







Eglin AFB 

BACR 







X 

X 

X 

0 

X 

X 

X 

X 

X 

X 

X 








TMCR 







X 

X 

X 

0 

X 

X 

X 

X 

X 

X 

X 








BUCR 

























DRCR 

X 

X 

X 





X 

X 




X 

X 

X 

X 

X 

X 






Fort Polk 

SMCR 

X 

X 

X 



X 

X 

X 

X 

X 




X 

X 

X 

X 

X 







UPTR 

























WHCH 

X 


X 




X 

X 

X 

X 



X 


X 

X 

X 

X 







ALCR 

X 

X 

X 

X 

X 

X 

X 

X 

X 

X 

X 

X 

X 

X 

X 

X 

X 

X 






Eglin AFB 

BACR 

X 

X 

X 

X 

X 

X 

X 

X 

X 

X 

X 

X 

X 

X 

X 

X 

X 

X 







TMCR 

X 

X 

X 

X 

X 

X 

X 

X 

X 

X 

X 

X 

X 

X 

X 

X 

X 

X 







BUCR 

X 

X 

X 

X 

X 

X 

X 

X 




X 

X 


X 

X 

X 

X 







DRCR 

X 

X 

X 

X 

X 

X 

X 

X 

X 


X 

X 

X 

X 

X 


X 

X 






Fort Polk 

SMCR 

X 

X 

X 

X 

X 

X 

X 

X 

X 


X 

X 

X 

X 


X 

X 

X 







UPTR 

























WHCH 

























ALCR 





X 

X 

X 

X 

X 

X 

X 

X 

X 

X 

X 

X 

X 

X 

0 

X 

0 

X 

X 

Eglin AFB 

BACR 





X 

X 

X 

X 

X 

X 

X 

X 

X 

X 

X 

X 

X 

X 

0 

X 

0 

X 

X 


TMCR 





X 

X 

X 

X 

X 

X 

X 

X 

X 

X 

X 

X 

X 

X 

0 

X 





BUCR 

























DRCR 
























Fort Polk 

SMCR 

























UPTR 

























WHCH 

























Year 


2005 


2006 


2007 


Fort Polk, Louisiana 

Throughout the study, the team established transects at five different sites 
at Fort Polk, Louisiana (see Figures 7-12). Transects at three sites were 
established in the fall 2005, including Whiskey Chitto (WHCH), Six-mile 
(SMCR), and Drakes Creeks (DRCR). Numerous logistic difficulties 
occurred along some of these transects that required modifications. After 
the fall 2005 season, the Whiskey Chitto Creek site could not be accessed 




ERDC/EL TR-12-22 


25 













ERDC/EL TR-12-22 


26 


because of military training restrictions; therefore, transects were 
established at Bundick Creek (BUCR) in the spring of 2006 and surveyed 
for the remaining period of the study. In addition, access along some 
transects at other sites was prohibited due to private landowner issues or 
logistic difficulties in accessing the sections (e.g., extensive flooding, 
downed woody debris, etc.). When this occurred, transects were moved up 
or down the drainage but always within the same drainage and within the 
radar-identified hotspot. Occasionally, migrants may use upland habitat 
disproportionally to riparian habitat (Rodewald and Matthews 2005). In 
order to assess the relative number of migrants detected outside of riparian 
transects, the team established an upland transect (UPTR) in the spring of 
2007. During this season, transect surveys were conducted on the Upland, 
Bundick Creek, and Drakes Creek sites. All transects, except those in upland 
habitat described above, were located in bottomland hardwood floodplains 
dominated by a variety of oak ( Quercus spp.), hickory ( Carya spp.), ash 
C Fraxinus spp.), and other hardwood species. 

Yuma Proving Ground, Arizona 

Because only xeroriparian (dry washes with very little riparian vegetation) 
habitat was present on Yuma Proving Ground, the team chose to locate 
study sites in Yuma outside the installation boundary along and near the 
Colorado River. In early 2006, three sites were established along two 
different drainages including the Colorado River and the All-American 
Canal (see Figures 13-16). The All-American Canal (AAC) site was located 
within an abandoned channel parallel with the All-American Canal. This 
site was characterized as having abundant shrub vegetation dominated by 
Palo Verde ( Parkinsonia spp.), Creosote Bush ( Larrea tridentata), and 
Saltcedar {Tamarix spp.). The Yuma transect (YUMA) was located 
immediately adjacent to the Colorado River and was within the city limits 
of Yuma, AZ. Vegetation varied in dominance among transects at this site, 
and most abundant species included Saltcedar, Willow ( Salix spp.), and 
Fremont Cottonwood ( Populus fremontii). The Imperial Valley site (IMP) 
was located approximately 18 km north of the other two sites and bisected 
the Imperial National Wildlife Refuge. Vegetation at this site was similar 
to other sites and was dominated by Mesquite ( Prosopis spp.), Palo Verde, 
Creosote Bush, and Saltcedar. 

Radar Data 

The team downloaded, analyzed, and processed archived migrant exodus 
data and nightly peak migration data for all days that ground surveys were 



ERDC/EL TR-12-22 


27 


conducted in all locations. The team used the same process described 
above (see Objective l methods). In some instances there were no usable 
radar data for a given evening due to the presence of precipitation or high 
densities of insects, and these were removed from analyses. Two spatially 
distinct exodus densities were estimated on each night on Eglin AFB and 
Ft. Polk due to the distance between ground sampling sites, and these two 
numbers were averaged to calculate a single exodus value. Only one peak 
migration value was recorded for each installation on each night. 

Statistical Analyses 

The team first classified all species recorded during transect surveys over 
the course of the study into migratory categories (Table 3), then eliminated 
all diurnal migrants and permanent residents from analyses; this ensured 
that only the relationship between migratory events captured on evening 
radar scans and changes in abundance of birds which had the potential to 
be captured by those radar scans was being modeled. For each morning a 
region was sampled, the team summed the number of spring or fall 
migrants recorded and divided it by the total distance walked (migrants/ 
km) that morning (this value was not necessarily constant on every morning 
because observers may have had to avoid or stop sampling transects for any 
number of logistical reasons). The team then calculated the change in 
migrant abundance between days by using the formula 

Amigrants / km d = migrants / km d - migrants / km d ^ 

where d = the survey date of interest. The team also calculated positive 
and negative species turnover from one day to the next; positive turnover 
is defined here as the number of species present on dayd which were not 
present on daya-i and negative turnover as the number of species not 
present on dayd which were present on dayd-i. 

The initial goal was to model changes in migrant abundance and species 
turnover as a function of migratory events captured on evening radar scans 
at the site level (i.e., individual riparian sites). However, due to a number of 
confounding issues, the team opted to combine all data recorded on each 
date at the regional (i.e., installation) level instead. First, due to the fact that 
observers at the site level were rotated within a season, it was impossible to 
determine how much variation in migrant data from one day to the next was 
attributable to real changes and how much was attributable to differences in 



ERDC/EL TR-12-22 


28 


Table 3. Migratory classification for all species recorded on ground transects during migration surveys at Eglin AFB, Fort Polk 

and Yuma (Poole 2005). 


Neotropical Migrants 

Nearctic Migrants 

Short-Distance Migrants 

Acadian Flycatcher 

Louisiana Waterthrush 

American Pipit 

American Crow 

Alder Flycatcher 

Lucy's Warbler 

Brewer's Blackbird 

American Goldfinch 

American Redstart 

MacGillivray's Warbler 

Brewer's Sparrow 

American Robin 

Ash-throated Flycatcher 

Magnolia Warbler 

Chipping Sparrow 

American Woodcock 

Baltimore Oriole 

Mourning Warbler 

Golden-crowned Kinglet 

Bachman's Sparrow 

Bank Swallow 

Nashville Warbler 

Green-tailed Towhee 

Belted Kingfisher 

Barn Swallow 

Northern Parula 

Hermit Thrush 

Bewick's Wren 

Bay-breasted Warbler 

Northern Rough-winged Swallow 

House Wren 

Black-chinned Sparrow 

Bell's Vireo 

Northern Waterthrush 

Lincoln's Sparrow 

Black-throated Sparrow 

Black-and-white Warbler 

Olive-sided Flycatcher 

Orange-crowned Warbler 

Bronzed Cowbird 

Black-billed Cuckoo 

Orchard Oriole 

Palm Warbler 

Brown Creeper 

Blackburnian Warbler 

Ovenbird 

Purple Finch 

Brown Thrasher 

Black-chinned Flummingbird 

Pacific-slope Flycatcher 

Ruby-crowned Kinglet 

Brown-crested Flycatcher 

Black-headed Grosbeak 

Painted Bunting 

Sage Thrasher 

Brown-headed Cowbird 

Blackpoll Warbler 

Philadelphia Vireo 

Say's Phoebe 

Cedar Waxwing 

Black-throated Blue Warbler 

Plumbeous Vireo 

Sedge Wren 

Common Grackle 

Black-throated Gray Warbler 

Prothonotary Warbler 

Swamp Sparrow 

Common Poorwill 

Black-throated Green Warbler 

Purple Martin 

Vesper Sparrow 

Costa's Hummingbird 

Blue Grosbeak 

Red-eyed Vireo 

White-crowned Sparrow 

Dark-eyed Junco 

Blue-gray Gnatcatcher 

Rose-breasted Grosbeak 

White-throated Sparrow 

Eastern Bluebird 

Blue-headed Vireo 

Ruby-throated Hummingbird 

Winter Wren 

Eastern Meadowlark 

Blue-winged Warbler 

Rufous Hummingbird 

Yellow-bellied Sapsucker 

Eastern Phoebe 

Bullock's Oriole 

Scarlet Tanager 

Yellow-headed Blackbird 

Eastern Towhee 

Calliope Hummingbird 

Summer Tanager 

Yellow-rumped Warbler 

European Starling 

Canada Warbler 

Swainson's Thrush 


Horned Lark 

Cassin's Vireo 

Swainson's Warbler 


Lesser Goldfinch 

Cerulean Warbler 

Tennessee Warbler 


Loggerhead Shrike 

Chestnut-sided Warbler 

Townsend's Warbler 


Marsh Wren 

Chimney Swift 

Tree Swallow 


Mourning Dove 

Chuck-will's-widow 

Vaux's Swift 


Phainopepla 

Clay-colored Sparrow 

Veery 


Pine Warbler 

Cliff Swallow 

Warbling Vireo 


Red-headed Woodpecker 

Common Yellowthroat 

Western Flycatcher 


Red-shafted Flicker 

Dickcissel 

Western Kingbird 


Red-winged Blackbird 

Eastern Kingbird 

Western Tanager 


Scott's Oriole 




ERDC/EL TR-12-22 


29 


Neotropical Migrants 

Nearctic Migrants 

Short-Distance Migrants 

Eastern Wood-Pewee 

Western Wood-Pewee 


Song Sparrow 

Golden-winged Warbler 

White-eyed Vireo 


Spotted Towhee 

Gray Catbird 

White-throated Swift 


Turkey Vulture 

Gray Flycatcher 

Willow Flycatcher 


Vermillion Flycatcher 

Gray-cheeked Thrush 

Wilson's Warbler 


Western Bluebird 

Great Crested Flycatcher 

Wood Thrush 


Western Meadowlark 

Hammond's Flycatcher 

Worm-eating Warbler 


White-winged Dove 

Hermit Warbler 

Yellow Warbler 


Yellow-shafted Flicker 

Hooded Oriole 

Yellow-bellied Flycatcher 



Hooded Warbler 

Yellow-billed Cuckoo 



Indigo Bunting 

Yellow-breasted Chat 



Kentucky Warbler 

Yellow-throated Vireo 



Lark Sparrow 

Yellow-throated Warbler 



Lazuli Bunting 

Unknown Empidonax 



Least Flycatcher 

Unknown Warbler 



Lesser Nighthawk 




Permanent Residents 

Waterbirds 

Raptors 

Unknown 

Abert's Towhee 

American Avocet 

Broad-winged Hawk 

Unknown Accipiter 

Anna's Hummingbird 

American Bittern 

Mississippi Kite 

Unknown Bird 

Barn Owl 

Black Tern 

Osprey 

Unknown Blackbird 

Barred Owl 

Blue-winged Teal 

Swainson's Hawk 

Unknown Buteo 

Black Phoebe 

Caspian Tern 

Merlin 

Unknown Cowbird 

Black Vulture 

Cinnamon Teal 

Northern Harrier 

Unknown Dove 

Black-tailed Gnatcatcher 

Eared Grebe 

Sharp-shinned Hawk 

Unknown Duck 

Blue Jay 

Franklin's Gull 

Red-shouldered Hawk 

Unknown Egret 

Boat-tailed Grackle 

Greater Yellowlegs 

Prairie Falcon 

Unknown Flycatcher 

Brown-headed Nuthatch 

Least Bittern 

American Kestrel 

Unknown Gnatcatcher 

Bushtit 

Least Sandpiper 

Cooper's Hawk 

Unknown Grackle 

Cactus Wren 

Lesser Yellowlegs 

Red-tailed Hawk 

Unknown Hawk 

Carolina Chickadee 

Little Blue Heron 


Unknown Heron 

Carolina Wren 

Long-billed Curlew 


Unknown Hummingbird 

Clapper Rail 

Long-billed Dowitcher 


Unknown Ibis 

Common Ground-Dove 

Snowy Egret 


Unknown Myiarchus 

Common Raven 

Solitary Sandpiper 


Unknown Nightjar 

Crissal Thrasher 

Sora 


Unknown Oriole 

Downy Woodpecker 

Spotted Sandpiper 


Unknown Owl 

Eastern Screech Owl 

Unknown Dowitcher 


Unknown Passerine 




ERDC/EL TR-12-22 


30 


Neotropical Migrants 

Nearctic Migrants 

Short-Distance Migrants 

Eastern Tufted Titmouse 

Virginia Rail 

Unknown Raptor 

Eurasian Collared Dove 

Western Sandpiper 

Unknown Sandpiper 

Fish Crow 

Willet 

Unknown Shorebird 

Gambel's Quail 

Ring-billed Gull 

Unknown Sparrow 

Gila Woodpecker 

Bufflehead 

Unknown Swallow 

Great Horned Owl 

California Gull 

Unknown Swift 

Greater Roadrunner 

Common Goldeneye 

Unknown Tanager 

Great-tailed Grackle 

Common Loon 

Unknown Thrasher 

Hairy Woodpecker 

Common Merganser 

Unknown Thrush 

House Finch 

Double-crested Cormorant 

Unknown Vireo 

Hutton's Vireo 

Forster's Tern 

Unknown Woodpecker 

Inca Dove 

Gadwall 

Unknown Wren 

Ladder-backed Woodpecker 

Northern Pintail 


Northern Bobwhite 

Northern Shoveler 


Northern Cardinal 

Ruddy Duck 


Northern Mockingbird 

Wilson's Snipe 


Pileated Woodpecker 

American Wigeon 


Red-bellied Woodpecker 

American Coot 


Red-cockaded Woodpecker 

Black-crowned Night-Heron 


Ring-necked Pheasant 

Black-necked Stilt 


Rock Pigeon 

Canada Goose 


Verdin 

Cattle Egret 


White-breasted Nuthatch 

Common Moorhen 


White-tailed Kite 

Great Blue Heron 


Wild Turkey 

Great Egret 



Green Heron 



Killdeer 



Mallard 



Pied-billed Grebe 



White-faced Ibis 



Wood Duck 



observers’ abilities. Moreover, a close examination of the data collected 
indicated that observer bias may have been a large and confounding 
problem during some sampling seasons (e.g., Figures 17 and 18). Pooling 
the data collected by all observers at the regional scale ensures that this 
source of error remains relatively constant from one day to the next, thereby 




ERDC/EL TR-12-22 


31 


making trends easier to detect. Second, it proved extremely difficult to 
calculate exodus densities for each site separately from the radar data. By 
the time migrants reached an altitude high enough to be captured in the 
radar beam they were displaced an unknown distance from their source 
habitat and the migrant images from the different sites began to blend 
together. Lastly, the diffuse nature of migration, once birds were in the 
atmosphere, did not lend itself to calculating site-specific peak migration 
values. 

High correlation (r > 0.8 in some cases) between peak migration and 
exodus densities prevented the team from including both explanatory 
variables in the same regression models. Additionally, because the same 
variables were recorded in the same place day after day, there was the 
possibility that turnover and abundance measurements were not inde¬ 
pendent over time (i.e., the change in migrant abundance on dayd was not 
independent of the change in migrant abundance on dayd-i). Thus, the team 
chose to use PROC AUTOREG (SAS Institute, Cary, NC 2009) for all 
regression analyses. PROC AUTOREG uses maximum likelihood theory to 
simultaneously estimate the regression coefficients of interest (p values) 
and the parameters for an autoregressive model (cp values) which explains 
the error term of a given day (Vt) as a function of the error value from the 
previous day (Vm) using the formulas 

Y t = (] 0 + &X t + V t 

V t = <pVt-i + e t 

Using these procedures, the team modeled the change in migrant 
abundance as a function of exodus and peak migration values separately for 
each region, year and season combination (this was necessary since PROC 
AUTOREG does not utilize categorical variables such as region or year). The 
team also modeled positive turnover as a function of peak migration radar 
values and negative turnover as a function of exodus radar values for each 
region, year and season combination. The significance level for parameter 
estimates was 0.05. 

Close inspection of the radar data revealed that both exodus and peak 
migration values tended to behave like continuous variables for low values 
and categorical variables for high values (see results). Because regression 
assumes continuous explanatory variables, the team also chose to divide 



ERDC/EL TR-12-22 


32 



Date 


Figure 17. Total number of birds recorded per morning transect at each Yuma site over 
time in the spring of 2006. The trends at the two sites tend to be mirror opposites of 
one another and the pattern reflects the fact that, for the most part, two observers 
alternated the days on which they sampled these sites (though a third observer was 
used on a few occasions). Thus, the trend indicates that one observer consistently 
counted more birds than the other regardless of which site they were both sampling. 



Figure 18. Mean and standard error of the total number of birds per km recorded 
per morning transect by each observer at Yuma in 2006. Values were calculated by 
averaging the number of birds per km the observer counted on each morning he or 
she surveyed. Results indicate that observer 1 consistently counted more birds 
than observer 2 who consistently counted more birds than observer 3. Thus, since 
observers were rotated, it is impossible to determine how much of the change in 
migrant abundance from one day to the next at the site level was attributable to 
real turnover and how much to observer bias. 










ERDC/EL TR-12-22 


33 


exodus and peak migration values into two categories representing high and 
low migratory events and investigate the impacts of these “treatments” on 
change in migrant abundance. Due to differences in the nature of migration 
across the country, the dividing point for these categories varied slightly for 
different regions. At Fort Polk and Eglin AFB, high migratory events 
included greater than too birds per cubic kilometer (birds per km 3 ), while 
low migratory events involved smaller densities. At Yuma high migratory 
events included greater than 70 birds per km 3 . The team used PROC GLM 
(SAS Institute, Cary, NC) to simultaneously model the change in migrant 
abundance as a function of these exodus and peak migration events for each 
year, region and season. A significance level of 0.05 was used for treatment 
effects as well. 

Objective 5: Avian Habitat Use in Southwestern Riparian Systems 

It was too difficult for observers to identify the exact location of each 
individual bird recorded during ground-based surveys near the Yuma 
Proving Ground. However, observers did divide each 500 m transect survey 
into four 125 m longitudinal sections and made note of which birds were 
recorded within each section. In 2007, a single observer then identified the 
dominant vegetation types at each of these 125 m sections at all three sites. 
The team then classified each section into one of five habitat categories: 
native trees (NT), native shrubs (NS), native-dominated with non-natives 
(ND), non-native / invasive shrub and tree community (NNI), or non-native 
/ invasive dominant with some natives (NND; Table 4). Due to the broad 
nature of these habitat categories and the fact that annual changes in 
vegetation cover and dominance are very minimal in this region, the team 
was able to use this information to compare abundance and richness data 
collected both in 2006 and 2007. Habitat within two surveyed sections was 
too heterogeneous to classify. 

Data Analyses 

Due to the fact that the team was primarily interested in terrestrial migrant 
habitat use, only species listed as neotropical or nearctic migrants in Table 3 
were retained. All swifts and swallows were also eliminated from the data 
because most were detected as flyovers, and it was not possible to determine 
whether these birds were actually using the habitat. Pacific-slope Fly¬ 
catchers (Empidonax difficilis ) and Cordilleran Flycatchers (Empidonax 
occidentalis) were considered Western Flycatchers because of the difficulty 
in distinguishing them in the field. Lastly, because the majority of the NT 



ERDC/EL TR-12-22 


34 


Table 4. Distribution and vegetation composition of 125 m transect sections from three sites 3 near Yuma, AZ 
that were surveyed for spring migrants in 2006 and 2007. 


Habitat b 

AAC 

YUMA 

IMP 

Total 

Dominant Woody Vegetation 

NS 

13 

0 

9 

22 

Palo Verde and Mesquite with or without Creosote Bush 

ND 

2 

0 

2 

4 

Palo Verde and Mesquite with or without Creosote Bush; 
some Saltcedar present but <50% dominance 

NND 

4 

3 

8 

15 

Saltcedar >50% in dominance with some Willow, Mesquite 
or Creosote Bush 

NNI 

9 

8 

1 

18 

Saltcedar-dominated community 

NT 

0 

11 

4 

15 

Cottonwood, Willow 


a Sites represent three geographically unique locations associated with the Colorado River called All-American Canal, 
Yuma and Imperial. 

b Sections were categorized according to dominant vegetation and were classified as either native shrub (NS), native- 
dominated with non-natives (ND), non-native/invasive dominant with some natives (NND), non-native/invasive shrub and 
tree community (NNI), or native tree (NT). 


sections (11 of 15) were located in Cottonwood restoration sites, the four 
sections located in mature Cottonwood stands were eliminated, as they were 
deemed too structurally different from the others. These mature Cotton¬ 
wood stands were not included in their own habitat category because they 
were only sampled in 2007. 

The team calculated total migrant abundance per kilometer, individual 
species abundance per kilometer, and total migrant species richness per 
section for each date a section was sampled. To account for the possibility of 
counting a single individual multiple times on different days, each metric 
was averaged across dates within years. Individuals not identified to species 
were removed from the data prior to calculating species richness. 

First, total abundance per kilometer and species richness per section were 
modeled as a function of habitat while including year as a fixed block effect. 
The 20 most abundant species (83% of detections; see results) were then 
selected and investigated as to whether or not habitat influences avian 
community composition. This was accomplished by including a species and 
species-by-habitat interaction term in the model. Results from this analysis 
indicated that habitat affected species differently, so abundance of each of 
those 20 species was also modeled individually as a function of habitat 
blocked on year. All modeling was conducted using PROC GLIMMIX (SAS 
Institute Inc., Cary, NC); abundance variables were best modeled by 
assuming a negative binomial distribution and all richness data were best 
modeled assuming a normal distribution. When there was evidence that 
habitat significantly affected any of the response variables (a = 0.05), all 
pair-wise comparisons of habitat types were investigated using a Tukey- 
Kramer adjustment of the P values. 




ERDC/EL TR-12-22 


35 


4 Results and Discussion 

Objective 1: Migrant Use of Military Installations 

Radar data indicated that 18 of the 40 installations investigated had 
significant exodus events occur during fall migration either on or directly 
adjacent to the installation (Table 5). Two such installations with very clear 
and repetitive exodus events were Fort Polk, Louisiana, and Eglin AFB, 
Florida. (Figures 19 and 20). Similarly, 17 installations showed consistently 
large exodus events during spring migration, including Fort Polk, Louisiana 
and the Yuma Proving Ground, Arizona (Figures 21 and 22). Spring and fall 
composite migration maps generated for each of the remaining 40 installa¬ 
tions can be found in Appendix B. 

These results reinforce the notion that military installations often play a 
critical role in supporting avian communities. Since installations tend to 
contain large expanses of natural areas, they often provide habitat oases in 
heavily developed landscapes. Such resources are particularly important to 
birds during their migratory journeys due to the precarious nature of the 
process and the high energy demands incurred. 

The team classified each installation into whether or not it contained 
stopover hotspots, but it should be noted that most installations showed 
significant spatial variability in migrant use. This is likely due to high 
variability in resources across a landscape and serves as further indication 
that migrants distinguish among habitat types. Natural resources 
managers on specific installations interested in providing and protecting 
stopover habitat should take a closer look at specific regions within the 
borders of their properties, using a combination of radar and ground 
surveys, in order to prioritize conservation areas. 

Lastly, while the team only investigated installations >200 km 2 in size, it 
should not be assumed that smaller installations do not provide valuable 
stopover habitat. This distinction was used as a means of narrowing down 
potential sites for further sampling, and not based on any biological signifi¬ 
cance. It is likely that many smaller installations also provide important 
stopover habitat, and this should not be discounted. 



ERDC/EL TR-12-22 


36 


Table 5. DoD military installations greater than 200 km 2 and located within 120 km of NEXRAD stations. The 
columns labeled “Spring” and “Fall” indicate whether or not that installation served as a stopover hotspot for 
migrating birds. It was not possible to generate appropriate maps for some installations in some seasons due to 
complications with weather and beam blockage by proximal mountain ranges. 


Installation 

State 

Region 

Area (km 2 ) 

NEXRAD Station 

NEXRAD Station Location 

Spring 

Fall 

Blair Lake Air Force Range 

AK 

W 

278 

APD 

Fairbanks, AK 

- 

- 

Fort Wainwright 

AK 

W 

3182 

APD 

Fairbanks, AK 

- 

- 

Fort Wainwright Maneuver Area 

AK 

w 

1203 

APD 

Fairbanks, AK 

- 

- 

Camp Grayling Military Reservation 

Ml 

MW 

538 

APX 

Gaylord, Ml 

No 

Yes 

Fort McCoy 

Wl 

MW 

239 

ARX 

La Crosse, Wl 

No 

No 

Saylor Creek Air Force Range 

ID 

W 

412 

CBX 

Boise, ID 

Yes 

No 

Fort Stewart 

GA 

E 

1130 

CLX 

Charleston, SC 

No 

No 

Fort Huachuca 

AZ 

W 

330 

EMX 

Tucson, AZ 

- 

Yes 

Fort Rucker Military Reservation 

AL 

E 

225 

EOX 

Fort Rucker, AL 

Yes 

Yes 

Eglin Air Force Base 

FL 

E 

1886 

EVX 

Eglin AFB, FL 

Yes 

Yes 

China Lake Naval Weapons Center 

CA 

W 

4035 

EYX 

Edwards AFB, CA 

No 

No 

Edwards Air Force Base 

CA 

W 

1244 

EYX 

Edwards AFB, CA 

No 

No 

Fort Irwin 

CA 

w 

2086 

EYX 

Edwards AFB, CA 

No 

No 

Fort Sill Military Reservation 

OK 

MW 

380 

FDR 

Frederick, OK 

Yes 

Yes 

Camp Swift N. G. Facility 

TX 

W 

210 

GRK 

Ft. Hood, TX 

No 

Yes 

Fort Hood 

TX 

W 

952 

GRK 

Ft. Hood, TX 

No 

Yes 

Holloman Air Force Base 

NM 

W 

213 

HDX 

Holloman AFB, NM 

Yes 

Yes 

White Sands Missile Range 

NM 

W 

8977 

HDX 

Holloman AFB, NM 

Yes 

Yes 

Fort Bliss 

TX 

W 

503 

HDtyEPZ 

Holloman AFB, NM/EI Paso, TX 

Yes 

Yes 

Fort Bliss McGregor Range 

TX 

W 

2727 

HDX/EPZ 

Holloman AFB, NM/EI Paso, TX 

Yes 

Yes 

Fort Campbell 

KY 

E 

406 

HPX 

Fort Campbell, KY 

Yes 

Yes 

Fort Knox 

KY 

E 

444 

LVX 

Louisville, KY 

Yes 

Yes 

Fort A. P. Hill Military Reservation 

VA 

E 

299 

LWX 

Sterling, VA 

No 

No 

Quantico Marine Corps Base 

VA 

E 

249 

LWX 

Sterling, VA 

Yes 

Yes 

Camp Lejeune Marine Corps Base 

NC 

E 

391 

MHX 

Morehead City, NC 

No 

Yes 

Avon Park AF Bombing Range 

FL 

E 

443 

MLB 

Melbourne, FL 

No 

No 

Hill Air Force Range 

UT 

W 

1444 

MTX 

Salt Lake City, UT 

Yes 

No 

Wendover Range (Hill AFB) 

UT 

W 

437 

MTX 

Salt Lake City, UT 

No 

No 

Fort Benning Military Reservation 

GA 

E 

736 

MXX 

Maxwell AFB, AL 

No 

No 

Camp Pendleton Marine Corps Base 

CA 

W 

491 

NKX 

San Diego, CA 

No 

Yes 

Boardman Naval Bombing Range 

WA 

W 

255 

PDT 

Pendleton, OR 

No 

No 




ERDC/EL TR-12-22 


37 


Installation 

State 

Region 

Area (km 2 ) 

N EXRAD Station 

N EXRAD Station Location 

Spring 

Fall 

Fort Polk Military Reservation 

LA 

E 

772 

POE 

Ft. Polk, LA 

Yes 

Yes 

Fort Carson Military Reservation 

CO 

W 

1131 

PUX 

Pueblo, CO 

No 

Yes 

Fort Bragg Military Reservation 

NC 

E 

571 

RAX 

Raleigh-Durham, NC 

No 

No 

Sierra Army Depot 

CA 

W 

381 

RGX 

Reno, NV 

No 

No 

Fort Riley Military Reservation 

KS 

MW 

414 

TWX 

Topeka, KS 

Yes 

Yes 

Fort Drum 

NY 

E 

447 

TYX 

Fort Drum, NY 

No 

No 

Vandenberg Air Force Base 

CA 

W 

449 

VBX 

Vandenberg AFB, CA 

Yes 

Yes 

Barry M. Goldwater AF Range 

AZ 

W 

4242 

YUX 

Yuma, AZ 

Yes 

Yes 

Yuma Proving Ground 

AZ 

w 

2927 

YUX 

Yuma, AZ 

Yes 

Yes 


POE Fall 2000 to 2004 Hotspots 



A / Rivers 
Poe_00_to_04 

□ 0.5- 1.0 Std. Dev. 

■ 1.0- 1.5 Std. Dev. 

I-1 1 5-2.0 Std. Dev. 

I-1 2.0- 2.5 Std. Dev. 

2.5- 3.0 Std. Dev. 
H : 3 Std. Dev. 

• Stations 

| | 60 Nautical Mile Limit 

I | Military4_lam 

I | States_outline 

■ Water 

I | Military4_lam 

■ States 

I | Canada 

□ Mexico 


10 0 10 20 30 40 Kilometers 



Copyright 2005 CUROL 


Figure 19. Map showing migration stopover areas based on WSR-88D detection of 
migrating birds during significant exodus events from Fort Polk, LA, during the fall 
migrations of 2000-2004. The data are quantified and displayed as standard deviations 

above mean. 






























ERDC/EL TR-12-22 


38 


EVX Fall 2000 to 2004 Hotspots 



A/ Rivers 
Evx_fallsto04 

□ 0.5-1.0 Std. Dev. 

■11.0-1.5 Std. Dev. 

■ 1.5-2.0 Std. Dev. 

□ 2.0-2.5 Std. Dev. 
2.5- 3.0 Std. Dev. 

> 3 Std. Dev. 

• Stations 

| | 60 Nautical Mile Limit 

I | Military4_lam 

| | States_outline 

■ Water 
Military4_lam 

■ States 

■■ Canada 
I | Mexico 


10 0 10 20 30 40 Kilometers 


% IS. 


Copyright 2005 


CUROL 


Figure 20. Map showing migration stopover areas based on WSR-88D 
detection of migrating birds during significant exodus events from Eglin AFB, 
FL, during the fall migrations of 2000-2004. The data are quantified and 
displayed as standard deviations above mean. 


POE Spring 2000-2003 & 2005 Hotspots 



A / Rivers 
Poe_00-03_05 
□ 0.5- 1.0 Std. Dev. 

■ 1.0-1.5 Std. Dev. 

■ 1.5-2.0 Std. Dev. 
_2.0- 2.5 Std. Dev. 

2.5- 3.0 Std. Dev. 
H > 3 Std. Dev. 

• Stations 

I | 60 Nautical Mile Limit 

I | Military4_lam 

| | States_outline 

■ Water 
Military4_lam 

■ States 
Canada 

I I Mexico 


10 0 10 20 30 40 Kilometers 


0 ysU,\y, 

■ > w %. 


Figure 21. Map showing migration stopover areas based on WSR-88D 
detection of migrating birds during significant exodus events from Fort Polk, 
LA, for the spring migrations of 2000-2003, and 2005. The colors 
represent standard deviations of above the mean density of birds per km 3 . 
Note that many of the stopover areas are associated with riparian habitat. 





















ERDC/EL TR-12-22 


39 


YUX Spring 2000-2003 & 2005 Hotspots 



& 

□ 


Rivers 
00_03_05 
0.5- 1.0 Std.Dev. 
1.0-1.5 Std.Dev. 
1.5- 2.0 Std. Dev. 


_ 2.0- 2.5 Std. Dev. 

2.5- 3.0 Std.Dev. 
■I > 3 Std. Dev. 

• Stations 

| | 60 Nautical Mile Limit 

I | Military4_lam 

| | States_outline 

■ Water 

Military4_lam 
HI States 
■■ Canada 
I I Mexico 


10 0 10 20 30 40 Kilometers 


U 


% 





Copyright 2006 CUROL 


Figure 22. Map showing migration stopover areas based on WSR-88D 
detection of migrating birds during significant exodus events on and 
around Yuma Proving Ground, AZ for the spring migrations of 2000-2003 
and 2005. The colors represent standard deviations above the mean 
density of birds per cubic km. The large red areas to the SW and SE of the 
radar site are not from migrant exodus events and are the results of 
ground clutter, breakthrough and radar blockage patterns in these areas. 
The DoD installation to the lower right is the Barry M. Goldwater Air Force 
Range, and the one above it is the Yuma Proving Ground. 


Objective 2: Quantifying Seasonal Migration 

Fall Migration 


The seasonal temporal pattern of nocturnal bird migration in fall over Eglin 
Air Force Base in Florida for the years 2000-2005 can be found in 
Figures 23 and 24. Migration in the month of August was minimal as were 
the values of the standard error (SE) of the mean. The quantity of migration 
increased in mid-September as did the values of the standard error of the 
mean. Peak of fall migration occurred between the end of September and 
the middle of October when the values of the standard error of the mean 
reached were greatest. Values of the standard error of the mean are greatest 
during this period because large movements occurred when conditions were 
favorable for migration, but migration was absent when conditions were 
poor (adverse winds, rain). After the middle of October migration density 
declined, and the values of the standard error of the mean also declined. 
During the six year period, a maximum migration density of 583 birds per 
km 3 was recorded at Eglin Air Force Base on two dates: 28 September 2003 
and 13 October 2004 (Figure 24). 










ERDC/EL TR-12-22 


40 



^(N^ooH^hO(N^ooHTthOfn^a\(N^ooH^hOfn^(^H^ 
OO 00 OO 00 00 OO 00 O'nO'nC^C^C^O'nO'n^h^h^hOOOOOOO^h^h 

Date ^ 


Figure 23. Seasonal temporal pattern of nocturnal bird migration in fall over Eglin Air 
Force Base, FL for the years 2000-2005. The symbols represent the mean number of 
birds per km 3 and the bars indicate the standard error of the mean. 



0^(N^00H^hO(N^00H^hOf0V0^(N^00H^hOf0^(^H^ 

00000000000000 0 \ 0 \ 0 \ 0 \ 0 \ 0 \ 0 \^^^< 0 >< 0 >< 0 >< 0 >< 0 >< 0 >< 0 >^^ 

Date 


Figure 24. Seasonal temporal pattern of nocturnal bird migration in fall over Eglin Air 
Force Base, FL for the years 2000-2005. The symbols indicate the maximum value of 
birds per km 3 for each date of fall during the six year period. 

The seasonal temporal pattern of fall, nocturnal bird migration over Fort 
Polk in Louisiana for the years 2000-2005, can be found in Figures 25 
and 26. Migration in the month of August was minimal with low standard 
error of the mean values. In September, migration and standard error of 
the mean values increased. The peak of migration occurred from mid- 
September through the middle of October, and the values of the standard 













































ERDC/EL TR-12-22 


41 


error of the mean reached a maximum during this period. From mid- 
October through the end of October, the amount of migration declined, 
and the values of the standard error of the mean also declined. During the 
six-year period, the maximum density of a migratory flight was 184 birds 
per km 3 and this value was reached on 13 dates between 9 September and 
24 October (Figure 26). 



Figure 25. Seasonal temporal pattern of nocturnal bird migration in fall over Fort 
Polk, LA for the years 2000-2005. Symbols represent the mean number of birds per 
km 3 and the bars indicate the standard error of the mean. 



Figure 26. Seasonal temporal pattern of nocturnal bird migration in fall over Fort Polk, 
LA for the years 2000-2005. Symbols indicate the maximum value of birds per km 3 for 
each date of fall during the six year period. 






































ERDC/EL TR-12-22 


42 


Spring Migration 

The seasonal temporal pattern of nocturnal bird migration in spring over 
Fort Polk, Louisiana for the years 2000-2003 and 2005-2006 can be found 
in Figures 27 and 28. The mean density (number of birds per km 3 ) was low 
until the end of March when it began to increase. By the middle of April, the 
mean density of migration was near peak values, and the period of peak 
density extended from mid-April until approximately 10 May. After this 
date, the density of migration declined steadily until the end of May. Values 
of the standard error of the mean followed the behavior of the mean. Near 
the beginning and end of the spring migration season, the SE values were 
low, because the differences between no migration and weak migration were 
small. During the period of peak migration, however, the SE values were 
high, because the differences between no migration and high density 
migration were great. The maximum density (birds per km 3 ) for each date 
of spring during the six year period can be found in Figure 28. Maximum 
densities of 583 occurred on 17, 28 April and 1,3, and 6 May. 

The seasonal temporal pattern of nocturnal bird migration in spring over 
Yuma, Arizona for the years 2000-2003 and 2005-2006 can be found in 
Figures 29 and 30. The mean density (number of birds per km 3 ) was low 
mid-March, increased steadily until mid-April, and then declined a bit 
until the beginning of May. The highest mean density of migration was in 



i-H 0 o m os ■ 

m m m m cn m m 


^j-t^om'sD^(NUDoo^H^|-t^om^o^(NUDoo^Hm'sO 

^ ^ r" r" r" r" n n ^ ^ ^ ^ ^ n n M n ^ ^ 


^ 

Date 


imniruoioioioio 


Figure 27. Seasonal temporal pattern of nocturnal bird migration in spring over Fort 
Polk, LA for the years 2000-2003 and 2005-2006. Symbols represent the mean 
number of birds per km 3 and the bars indicate the standard error of the mean. 







































ERDC/EL TR-12-22 


43 


S 

# c-> 

2 

sm 

OJ 


es 

3 

a 

*h 



^j- o m ^ 
m m m m cn m m 


^j-t^om'sO^(N‘^oc 
" " (N (N (N 


'sf Tt ■ 


^l" ^l" ^1" 


Date 


i-f'OM'ooiN'floortmio 
<r, 'r, 'r, ~ ' ' ' rl rl C'' X \C 

'r, <r. ir, ir, ir, ir, 'r, ir, 


Figure 28. Seasonal temporal pattern of nocturnal bird migration in spring over Fort 
Polk, LA for the years 2000-2003 and 2005-2006. Symbols indicate the maximum 
value of birds per km 3 for each date of spring during the six year period. 



Figure 29. Seasonal temporal pattern of nocturnal bird migration in spring over Yuma, 
AZ for the years 2000-2003 and 2005-2006. Symbols represent the mean number of 
birds per km 3 and the bars indicate the standard error of the mean. 







































ERDC/EL TR-12-22 


44 



< (N (N (N (N ; 


" " H <N <N <N ; 


^J-OOm'sO^(N‘^ 
" " H Cn| (N 


m m m m m m m ^ ^ ^ ^ ^ ^ ^ ir> in m m m ir> 


oonrn^ 

vo 

i n in 


Date 


Figure 30. Seasonal temporal pattern of nocturnal bird migration in spring over 
Yuma, AZ for the years 2000-2003 and 2005-2006. Symbols indicate the maximum 
value of birds per km 3 for each date of spring during the six year period. 


early May and then the mean density declined steadily until the end of 
May (Figure 29). The maximum density (birds per km 3 ) for each date of 
spring during the six year period can be found in Figure 30. The highest 
density recorded during the six spring seasons was 463 birds per km 3 on 
3 May 2005. The next highest density was 184 birds per km 3 ; this density 
was recorded on 12 dates ranging from 11 April through 21 May. 


Discussion 

The amount of migration over military installations for a particular date in 
spring and fall varies from year-to-year, and variation is greatest for dates 
during the period of peak seasonal migration. In this study, the periods of 
peak migration were clearly defined based on six years of data, but the 
amount of migration one observes on a particular date is strongly depen¬ 
dent on the number of grounded migrants in the area ready to migrate and 
the weather conditions present at the end of the day when nocturnal 
migration begins. When weather conditions are ideal during the period of 
peak movements very high density migrations can occur, but when weather 
conditions are poor (rain, or adverse winds, or both), no migration is 
recorded. At the beginning and end of seasonal migration, movements are 
generally of low density even when weather conditions are ideal for a flight, 
because the maximum number of birds ready to migrate is small. 


Migrations in the fall with the highest mean densities were recorded at 
Eglin Air Force Base. This is not surprising because winds prevailing from 








ERDC/EL TR-12-22 


45 


the west over the United States drift birds migrating southward towards 
the east and southeast, and as a result, fall migration densities in the 
Southeast are greater than in other areas of the South. Migrations in the 
spring with the highest mean densities were recorded at Fort Polk. This 
too is not surprising as the greatest densities of migrants entering the 
United States in spring arrive from over the Gulf of Mexico on the upper 
Texas coast and on the southwestern Louisiana coast. These flights 
originate south of the Gulf and the birds depart at the beginning of the 
night. Because of the distance across the Gulf, the flights arrive on the 
northern coast in the afternoon and early evening and continue inland 
passing over Fort Polk. The mean densities of migration in the spring are 
twice as great at Fort Polk as those passing over Yuma, and the magnitude 
of trans-Gulf migration is largely responsible for this difference. 

Objective 3: Development of Migration Forecast Models 

The neural net model for Eglin Air Force Base was based on the peak 
density of migration on 183 nights during the fall seasons of 2000 through 
2005. The R 2 is 0.8217 and the CV R 2 is 0.4228. Of the 20 tours, two 
converged at best, 15 converged less than best, and three reached the maxi¬ 
mum number of iterations. The plot of actual bird density by predicted bird 
density can be found in Figure 31. The most influential predictor variables 
were temperature, humidity, surface cross winds, and 925mb head winds. 
As temperature and humidity declined, the density of migration increased, 
and as surface winds shifted to blowing from the west, the density of 
migration increased. As winds at 3000 ft increased in velocity from the 
north, the amount of migration increased. 

The neural net model for Fort Polk was based on the peak density of 
migration on 40 nights during the fall of 2005. The R 2 is 0.9906 and no 
cross validation was used because only one season of data was used. Of the 
16 tours, two converged at best, 13 converged worse than best, and one 
reached the maximum number of iterations. The plot of actual bird density 
by predicted bird density can be found in Figure 32. The most influential 
predictor variables were surface head and cross winds, 925mb head winds. 
As surface winds from the north increased in velocity, the density of migra¬ 
tion increased, and as surface cross winds from the west increased (as 
happens just before a cold front passage), the density of migratory 
movements increased. The density of migratory movements increased as 
winds at 3000 ft increased in velocity from the north. 



ERDC/EL TR-12-22 


46 



Figure 31. Plot of actual birds per km 3 by predicted birds per 
km 3 for fall data from the years 2000-2005 at Eglin Air Force 
Base in Florida. 



Figure 32. Plot of actual birds per km 3 by predicted birds per 
km 3 for fall data from 2005 at Fort Polk Army Base in 
Louisiana. 


The neural net model for Yuma, Arizona was based on the peak density of 
migration on 186 nights during the spring seasons from 2001 through 
2005. The overfit penalty for this analysis was 0.3. The R 2 is 0.4774 for the 
neural net model, and the CV R 2 is 0.1266 for the 5-fold cross validation. 
The fit results indicate that eight of the 20 tours converged at best, nine 
converged worse than best, and three reached the maximum number of 
iterations. None of the predictor variables showed a great influence on the 
prediction of migration density. As ordinal date increased the density of 
bird migration increased, and as surface winds from the south increased 




















ERDC/EL TR-12-22 


47 


the density of migration tended to increase. When 925mb winds increased 
from the east the density of migratory flights increased. The plot of actual 
bird density by predicted bird density can be found in Figure 33. 


Actual by Predicted Plot 

110 

100 

90 

80 

? 70 

m 



0 10 20 30 40 50 60 70 80 90 100 110 

Predicted birds km-3 


Figure 33. Plot of actual birds per km 3 by predicted birds per 
km 3 for data from the spring seasons from 2001 through 
2005 at Yuma, AZ. 


Discussion 

All of the forecast models performed poorly when the densities of migration 
were low. The neural net models for Eglin Air Force Base and Fort Polk do a 
fairly good job at predicting dense migratory movements, and these are the 
movements that should be of the greatest concern for aviation safety. Poor 
performance for low density movements is of lesser concern. The neural net 
model for Yuma is not a good predictor of migration density, and this is 
likely the result of the number of low density movements that were used for 
building the model. Because winds are rarely favorable over Yuma, 
migratory flights frequently occur when unfavorable winds are present, and 
the flights are of low density. The density of migration actually increased 
when temperature decreased in spring. 

In both spring and fall, temperature, wind, and rain are the most influential 
variables (Gauthreaux 1978). In an investigation of correlations between 
radar measures of bird migration density and 20 weather variables in 
Lithuania from 1974 to 1977, Zhalakyavichus (1985) found that the most 
important variables were air temperature and cloud type in spring and 
cloud type and wind direction in fall. The correlation was better for spring 
than in fall and for migrations in the continental part of the region. Van 














ERDC/EL TR-12-22 


48 


Belle et al. (2007) also found that significant input variables for regression 
models that predicted migration intensity included: seasonal migration 
trend, wind profit, 24-hour trend in barometric pressure, and rain, but they 
cautioned that the mismatch between measurements and predictions was 
large for existing models, and that existing models are only valid locally and 
cannot be extrapolated to new locations. 

Most of the multivariate analyses of the influence of weather variables on 
migration have R 2 values ranging from 0.40 to 0.62 in spring and 0.26 to 
0.61 in fall (Gauthreaux 1978). Several factors are responsible when 
weather variables fail to explain much of the night-to-night variance in the 
density of bird migration: 

• There are no birds physiologically ready to migrate when weather 
conditions are ideal. 

• In the fall most birds that are ready to migrate leave the first night after 
the passage of a cold front. If the weather is still very favorable for 
migration on subsequent night(s), few birds are present to migrate. 

• If unfavorable weather conditions persist too long, migrants will depart 
in unfavorable weather conditions. 

• Weather conditions well away from the study site may influence peak 
migration density more than local conditions. 

Objective 4: Comparison of Migrant Survey Techniques 

Between the fall of 2005 and the spring of 2007, the team conducted 
1900 fall migrant surveys and 2724 spring migrant surveys along 2,312 km 
of transects across the three study regions (Table 6). The team made 
149,231 bird detections of 274 different species but 25,258 of those detec¬ 
tions were birds flying over the site and were not included in the data 
analyses. Species richness tended to be greater during spring surveys and 
was greater at Yuma (Table 7) than at either Ft. Polk (Tables 8 and 9) or 
Eglin AFB (Table 10). Overall, the team had 202 paired sampling dates from 
which A migrant abundance could be calculated (Figures 34-37). However, 
only peak migration densities could be estimated for 192 of those paired 
dates and migrant exodus densities for 179 (Table 11). 

Peak migration density was not a significant predictor of A migrant 
abundance observed during ground surveys for any region during either 
season in any year (Table 12). In fact, though we expected to see a positive 
relationship between peak migration densities and A abundance, the data 



ERDC/EL TR-12-22 


49 


Table 6. Summary of the sampling effort and number of birds detected at each region by year 
and season. The N for surveys is the total number of 500m transects completed during each 
season; distance sampled is reported in km. 


Season Region 

Year 

Surveys 

AM PM Total 

Distance Sampled 

AM PM Total 

Ground Detections 

AM PM Total 

Flyove 

AM PM 

r 

Total 

Spring 

Yuma 

2006 

439 

3 

442 

219.5 

1.5 

221 

50404 

161 

50565 

3838 

0 

3838 



2007 

832 

0 

832 

416 

0 

416 

31133 

0 31133 

5937 

0 

5937 


Ft. Polk 

2006 

633 

272 

905 

316.5 

136 

452.5 

13147 2250 

15397 

577 

192 

769 



2007 

348 

197 

545 

174 

98.5 

272.5 

7083 2116 

9199 

146 

79 

225 

Fall 

EglinAFB 2005 

166 

71 

237 

83 

35.5 

118.5 

2268 

673 

2941 

42 

6 

48 



2006 

314 

172 

486 

157 

86 

243 

8214 2715 

10929 

99 

18 

117 



2007 

284 

205 

489 

142 

102.5 

244.5 

6989 3313 

10302 

78 

31 

109 


Ft. Polk 

2005 

194 

115 

309 

97 

57.5 

154.5 

2921 

718 

3639 

96 

133 

229 



2006 

232 

147 

379 

116 

73.5 

189.5 

2895 

873 

3768 

64 

22 

86 


Table 7. Number of ground and flyover detections per species recorded during 1271 morning and three 
evening transect surveys conducted during spring migration at Yuma in 2006-2007. Species highlighted in 

bold were included in all statistical analyses. 


Species 

Ground 

Flyovers 


Species 

Ground 

Flyovers 

Mourning Dove 

7968 

532 


Green Heron 

81 

17 

White-winged Dove 

5859 

379 


Northern Mockingbird 

81 

8 

Brown-headed Cowbird 

4182 

574 


Summer Tanager 

78 

0 

Red-winged Blackbird 

4118 

2184 


Black Phoebe 

75 

0 

Verdin 

3372 

0 


Unknown Sparrow 

68 

14 

Common Yellowthroat 

3161 

1 


Bell's Vireo 

67 

0 

Wilson's Warbler 

3048 

4 


Spotted Sandpiper 

62 

1 

Brewer's Sparrow 

2463 

204 


Tree Swallow 

61 

8215 

House Finch 

2166 

232 


Rock Pigeon 

59 

52 

Gambel's Quail 

2022 

13 


Ruby-crowned Kinglet 

58 

0 

Abert's Towhee 

1798 

3 


Blue-gray Gnatcatcher 

53 

0 

Ash-throated Flycatcher 

1791 

8 


Black-crowned Night-Heron 

51 

6 

Unknown Warbler 

1334 

110 


Least Bittern 

50 

0 

Pacific-slope Flycatcher 

1301 

0 


Sora 

50 

0 

Warbling Vireo 

1209 

1 


Mallard 

48 

31 

Unknown Passerine 

1182 

164 


N. Rough-winged Swallow 

46 

975 

Orange-crowned Warbler 

1130 

i 


Western Flycatcher 

45 

0 

Black-tailed Gnatcatcher 

1118 

2 


Vaux's Swift 

43 

464 

Cliff Swallow 

1056 

4884 


Hermit Warbler 

42 

0 

Gila Woodpecker 

1008 

8 


Cassin's Vireo 

41 

0 

Ladder-backed Woodpecker 

939 

14 


Olive-sided Flycatcher 

41 

2 

American Coot 

935 

0 


House Wren 

39 

0 

Yellow-headed Blackbird 

882 

146 


Green-tailed Towhee 

37 

0 



























































ERDC/EL TR-12-22 


50 


Species 

Ground 

Flyovers 


Species 

Ground 

Flyovers 

Song Sparrow 

815 

0 


Cinnamon Teal 

37 

0 

Great-tailed Grackle 

815 

494 


Unknown Flycatcher 

35 

11 

Nashville Warbler 

634 

0 


Great Blue Heron 

34 

32 

Yellow-rumped Warbler 

566 

10 


European Starling 

34 

11 

Unknown Empidonax 

536 

2 


Unknown Vireo 

34 

0 

Western Kingbird 

526 

121 


Snowy Egret 

34 

42 

Anna's Hummingbird 

469 

4 


Least Sandpiper 

29 

3 

Cactus Wren 

444 

0 


Lesser Goldfinch 

28 

17 

Western Tanager 

409 

19 


Western Meadowlark 

28 

0 

Black-throated Gray Warbler 

408 

2 


Cedar Waxwing 

27 

0 

Yellow Warbler 

400 

1 


Dark-eyed Junco 

25 

0 

Lucy's Warbler 

378 

0 


American Kestrel 

22 

6 

MacGillivray's Warbler 

351 

0 


Red-tailed Hawk 

22 

3 

Bullock's Oriole 

304 

10 


Great Egret 

22 

97 

Unknown Hummingbird 

301 

31 


Cooper's Hawk 

21 

5 

White-faced Ibis 

295 

570 


Unknown Thrasher 

21 

0 

Blue Grosbeak 

292 

0 


Inca Dove 

19 

0 

Townsend's Warbler 

277 

0 


Ruddy Duck 

18 

0 

Phainopepla 

277 

25 


Brown-crested Flycatcher 

17 

0 

Chipping Sparrow 

225 

0 


Say's Phoebe 

16 

0 

Western Wood-Pewee 

224 

1 


Long-billed Dowitcher 

16 

0 

Black-chinned Hummingbird 

219 

3 


Western Sandpiper 

16 

0 

Yellow-breasted Chat 

212 

0 


Unknown Ibis 

15 

0 

Marsh Wren 

208 

0 


Osprey 

15 

16 

Black-headed Grosbeak 

201 

11 


Loggerhead Shrike 

15 

1 

Lesser Nighthawk 

177 

71 


Bufflehead 

15 

1 

White-crowned Sparrow 

177 

0 


Common Raven 

15 

5 

Crissal Thrasher 

169 

0 


Common Ground-Dove 

14 

8 

Common Moorhen 

159 

0 


Great Horned Owl 

14 

0 

Greater Roadrunner 

142 

0 


Northern Harrier 

12 

6 

Killdeer 

126 

26 


Unknown Cowbird 

12 

0 

Barn Swallow 

125 

113 


Swainson's Thrush 

12 

0 

Lazuli Bunting 

125 

6 


Unknown Myiarchus 

12 

12 

Turkey Vulture 

115 

114 


Red-shafted Flicker 

12 

0 

Black-necked Stilt 

in 

68 


Cattle Egret 

11 

4 

Pied-billed Grebe 

102 

0 


Hooded Oriole 

ii 

0 




ERDC/EL TR-12-22 


51 


Species 

Ground 

Flyovers 


Species 

Ground 

Flyovers 

Unknown Swallow 

86 

2022 


Gray Flycatcher 

10 

0 

Unknown Shorebird 

86 

20 


Unknown Thrush 

10 

0 

Unknown Sandpiper 

10 

11 


Eurasian Collared Dove 

2 

0 

Belted Kingfisher 

10 

4 


White-tailed Kite 

2 

1 

Double-crested Cormorant 

9 

95 


American Pipit 

1 

0 

Hermit Thrush 

9 

0 


Broad-winged Hawk 

1 

0 

Common Merganser 

9 

0 


Swainson's Hawk 

1 

0 

Greater Yellowlegs 

9 

4 


Unknown Hawk 

1 

1 

White-throated Swift 

7 

33 


American Crow 

1 

1 

Sharp-shinned Hawk 

7 

6 


American Robin 

1 

0 

American Goldfinch 

7 

2 


Horned Lark 

1 

0 

Unknown Oriole 

7 

4 


Unknown Heron 

1 

1 

Ring-necked Pheasant 

7 

0 


Alder Flycatcher 

i 

0 

Willow Flycatcher 

6 

0 


Baltimore Oriole 

i 

0 

Bewick's Wren 

6 

0 


Plumbeous Vireo 

i 

0 

Vermillion Flycatcher 

6 

0 


Unknown Tanager 

i 

2 

Unknown Gnatcatcher 

6 

0 


Common Poorwill 

1 

0 

Unknown Woodpecker 

6 

4 


Scott's Oriole 

i 

0 

Franklin's Gull 

5 

1 


Spotted Towhee 

i 

0 

Black-throated Sparrow 

5 

0 


Lesser Yellowlegs 

i 

9 

Unknown Raptor 

4 

0 


Solitary Sandpiper 

i 

0 

Northern Parula 

4 

0 


Unknown Dowitcher 

i 

0 

Gad wall 

4 

0 


Virginia Rail 

1 

0 

Long-billed Curlew 

4 

14 


Willet 

i 

13 

Wilson's Snipe 

4 

1 


Barn Owl 

1 

0 

Unknown Bird 

4 

0 


Bushtit 

1 

0 

Unknown Owl 

4 

0 


Clapper Rail 

1 

0 

Northern Shoveler 

3 

0 


Hutton's Vireo 

1 

0 

Prairie Falcon 

3 

1 


Unknown Wren 

1 

0 

Unknown Accipiter 

3 

0 


Brewer's Blackbird 

0 

3 

Unknown Buteo 

3 

0 


Bank Swallow 

0 

11 

American Wigeon 

3 

0 


Unknown Swift 

0 

24 

Black Tern 

3 

1 


Unknown Blackbird 

0 

25 

Sage Thrasher 

3 

0 


California Gull 

0 

22 

Black-and-white Warbler 

3 

0 


Caspian Tern 

0 

2 

Clay-colored Sparrow 

2 

0 


Common Loon 

0 

5 




ERDC/EL TR-12-22 


52 


Species 

Ground 

Flyovers 


Species 

Ground 

Flyovers 

Calliope Hummingbird 

2 

0 


Forster's Tern 

0 

1 

Rufous Hummingbird 

2 

0 


Ring-billed Gull 

0 

2 

Bronzed Cowbird 

2 

0 


Unknown Egret 

0 

15 

Costa's Hummingbird 

2 

0 


Lark Sparrow 

0 

2 

Western Bluebird 

2 

0 


American Avocet 

0 

72 

Lincoln's Sparrow 

2 

0 


American Bittern 

0 

1 

Hammond's Flycatcher 

2 

0 


Common Goldeneye 

0 

i 

Black-chinned Sparrow 

2 

0 


Northern Pintail 

0 

i 

Blue-winged Teal 

2 

0 


Unknown Duck 

0 

6 

Eared Grebe 

2 

0 






Table 8. Number of ground and flyover detections per species recorded during 981 morning and 
469 evening transect surveys conducted during spring migration at Ft. Polk in 2006-2007. Species 
highlighted in bold were included in all statistical analyses. 


Species 

Ground 

Flyovers 


Species 

Ground 

Flyovers 

Eastern Tufted Titmouse 

2723 

0 


Broad-winged Hawk 

26 

51 

Red-eyed Vireo 

2402 

0 


American Goldfinch 

26 

40 

Northern Cardinal 

2289 

0 


Magnolia Warbler 

26 

0 

Hooded Warbler 

1493 

0 


Common Yellowthroat 

24 

0 

Carolina Wren 

1378 

0 


Yellow-bellied Sapsucker 

23 

0 

Yellow-rumped Warbler 

1263 

6 


Veery 

22 

0 

Carolina Chickadee 

1213 

0 


Red-headed Woodpecker 

20 

1 

Pine Warbler 

1071 

0 


Yellow-breasted Chat 

20 

0 

Red-bellied Woodpecker 

918 

3 


Eastern Phoebe 

18 

0 

Ruby-crowned Kinglet 

781 

0 


Yellow-bellied Flycatcher 

18 

0 

Blue-gray Gnatcatcher 

741 

0 


American Redstart 

17 

0 

Blue Jay 

710 

17 


Eastern Bluebird 

16 

4 

American Crow 

617 

220 


Brown Thrasher 

16 

0 

White-throated Sparrow 

590 

0 


Northern Waterthrush 

15 

0 

Pileated Woodpecker 

446 

9 


Rose-breasted Grosbeak 

15 

0 

Unknown Warbler 

432 

6 


Tennessee Warbler 

15 

0 

Summer Tanager 

344 

0 


Unknown Empidonax 

15 

0 

Acadian Flycatcher 

343 

0 


Nashville Warbler 

14 

0 

Great Crested Flycatcher 

331 

0 


Scarlet Tanager 

14 

0 

Yellow-throated Vireo 

287 

0 


Brown-headed Nuthatch 

13 

0 

Downy Woodpecker 

255 

0 


Brown-headed Cowbird 

12 

19 

Louisiana Waterthrush 

243 

1 


Warbling Vireo 

ii 

0 





ERDC/EL TR-12-22 


53 


Species 

Ground 

Flyovers 


Species 

Ground 

Flyovers 

Wood Thrush 

189 

0 


Swainson's Warbler 

10 

0 

Barn Swallow 

181 

122 


Oven bird 

9 

0 

Blue-headed Vireo 

177 

0 


Turkey Vulture 

8 

52 

White-eyed Vireo 

177 

0 


Yellow Warbler 

8 

0 

Barred Owl 

145 

0 


Canada Warbler 

7 

0 

Indigo Bunting 

144 

9 


Eastern Towhee 

7 

0 

Hermit Thrush 

143 

0 


Unknown Swallow 

6 

15 

Black-throated Green Warbler 

121 

0 


Unknown Hawk 

6 

1 

Ruby-throated Hummingbird 

117 

1 


Gray-cheeked Thrush 

6 

0 

Red-shouldered Hawk 

114 

64 


Prothonotary Warbler 

6 

0 

Northern Parula 

114 

0 


Blackpoll Warbler 

5 

0 

Hairy Woodpecker 

103 

0 


Philadelphia Vireo 

5 

0 

Cedar Waxwing 

98 

89 


Bachman's Sparrow 

5 

0 

Yellow-billed Cuckoo 

96 

0 


Chimney Swift 

4 

102 

Unknown Passerine 

88 

0 


Tree Swallow 

4 

3 

Mourning Dove 

86 

3 


Little Blue Heron 

4 

0 

Unknown Woodpecker 

76 

0 


Blue-winged Warbler 

4 

0 

Wood Duck 

73 

1 


Wilson's Warbler 

4 

0 

Swainson's Thrush 

69 

0 


Northern Bobwhite 

4 

0 

Kentucky Warbler 

68 

0 


Belted Kingfisher 

4 

0 

Unknown Bird 

67 

17 


Eastern Kingbird 

3 

0 

Black-and-white Warbler 

53 

0 


Common Grackle 

3 

0 

Worm-eating Warbler 

51 

0 


Black-billed Cuckoo 

3 

0 

Yellow-shafted Flicker 

46 

0 


Cerulean Warbler 

3 

0 

Orange-crowned Warbler 

44 

0 


Golden-winged Warbler 

3 

0 

Baltimore Oriole 

44 

0 


Orchard Oriole 

3 

0 

Yellow-throated Warbler 

44 

0 


Unknown Sparrow 

3 

0 

Brown Creeper 

42 

0 


Unknown Duck 

3 

0 

Wild Turkey 

42 

0 


Black Vulture 

3 

14 

Eastern Wood-Pewee 

41 

0 


Cooper's Hawk 

2 

0 

Great Blue Heron 

39 

29 


Red-tailed Hawk 

2 

1 

Blackburnian Warbler 

38 

0 


Unknown Raptor 

2 

1 

Unknown Thrush 

38 

0 


American Robin 

2 

0 

Golden-crowned Kinglet 

37 

0 


Swamp Sparrow 

2 

0 

Unknown Vireo 

37 

0 


Alder Flycatcher 

2 

0 

Chestnut-sided Warbler 

31 

0 


Blue Grosbeak 

2 

0 




ERDC/EL TR-12-22 


54 


Species 

Ground 

Flyovers 


Species 

Ground 

Flyovers 

Chipping Sparrow 

28 

0 


Chuck-will's-widow 

2 

0 

Winter Wren 

27 

0 


Red-cockaded Woodpecker 

2 

2 

Gray Catbird 

27 

0 


Unknown Owl 

2 

0 

Green Heron 

1 

0 


Killdeer 

0 

3 

Mississippi Kite 

1 

17 


Bank Swallow 

0 

1 

Sharp-shinned Hawk 

1 

3 


Cliff Swallow 

0 

3 

Unknown Buteo 

1 

1 


Purple Martin 

0 

38 

Unknown Heron 

1 

0 


Unknown Accipiter 

0 

1 

Palm Warbler 

1 

0 


Red-winged Blackbird 

0 

5 

Bay-breasted Warbler 

1 

0 


Black-crowned Night-Heron 

0 

2 

Black-throated Blue Warbler 

1 

0 


Cattle Egret 

0 

7 

Mourning Warbler 

1 

0 


Double-crested Cormorant 

0 

6 

Painted Bunting 

1 

0 


Dickcissel 

0 

3 

Common Ground-Dove 

1 

0 


Fish Crow 

0 

1 

Unknown Wren 

1 

0 






Table 9. Number of ground and flyover detections per species recorded during 426 morning and 262 evening 
transect surveys conducted during fall migration at Ft. Polk in 2005-2006. Species highlighted in bold were 

included in all statistical analyses. 


Species 

Ground 

Flyovers 


Species 

Ground 

Flyovers 

Blue Jay 

1033 

33 


Red-eyed Vireo 

4 

0 

Eastern Tufted Titmouse 

946 

0 


Chipping Sparrow 

3 

0 

Carolina Chickadee 

692 

0 


American Kestrel 

3 

0 

American Crow 

685 

110 


Unknown Hawk 

3 

2 

Carolina Wren 

633 

3 


American Robin 

3 

0 

Red-bellied Woodpecker 

618 

2 


Hermit Thrush 

3 

0 

Northern Cardinal 

433 

0 


Blue-headed Vireo 

3 

0 

Yellow-shafted Flicker 

349 

1 


Gray-cheeked Thrush 

3 

0 

Pileated Woodpecker 

304 

1 


Unknown Buteo 

2 

0 

Downy Woodpecker 

174 

0 


Red-headed Woodpecker 

2 

0 

Pine Warbler 

167 

0 


House Wren 

2 

0 

Wood Thrush 

134 

0 


Swainson's Thrush 

2 

0 

Wood Duck 

133 

7 


Yellow-throated Vireo 

2 

0 

Brown Thrasher 

116 

0 


Ruby-throated Hummingbird 

1 

0 

Unknown Warbler 

109 

1 


Unknown Swift 

1 

2 

Ruby-crowned Kinglet 

97 

0 


Broad-winged Hawk 

1 

0 

Red-shouldered Hawk 

73 

25 


Sharp-shinned Hawk 

1 

0 

Unknown Woodpecker 

64 

0 


Brown-headed Cowbird 

1 

0 

Hooded Warbler 

51 

0 


Eastern Meadowlark 

1 

0 





ERDC/EL TR-12-22 


55 


Species 

Ground 

Flyovers 


Species 

Ground 

Flyovers 

Gray Catbird 

36 

0 


Unknown Egret 

1 

0 

Barred Owl 

36 

0 


Orange-crowned Warbler 

1 

0 

Yellow-bellied Sapsucker 

34 

0 


Blackburnian Warbler 

1 

0 

Eastern Phoebe 

31 

0 


Chuck-will's-widow 

1 

0 

Unknown Thrush 

29 

0 


Kentucky Warbler 

1 

0 

White-eyed Vireo 

29 

0 


Least Flycatcher 

1 

0 

Hairy Woodpecker 

24 

0 


Louisiana Waterthrush 

1 

0 

Brown-headed Nuthatch 

23 

0 


Magnolia Warbler 

1 

0 

Summer Tanager 

22 

0 


Nashville Warbler 

1 

0 

Black-throated Green Warbler 

21 

0 


Oven bird 

1 

0 

Great Crested Flycatcher 

20 

0 


Painted Bunting 

1 

0 

Blue-gray Gnatcatcher 

19 

0 


Prothonotary Warbler 

1 

0 

Eastern Wood-Pewee 

19 

0 


Rose-breasted Grosbeak 

1 

0 

Golden-crowned Kinglet 

17 

0 


Yellow-throated Warbler 

1 

0 

Veery 

16 

0 


Eastern Towhee 

1 

0 

Unknown Bird 

16 

99 


Snowy Egret 

1 

0 

Wild Turkey 

14 

0 


Common Ground-Dove 

1 

0 

Eastern Bluebird 

13 

5 


Eastern Screech Owl 

1 

0 

Mourning Dove 

13 

0 


Fish Crow 

1 

0 

American Redstart 

13 

0 


White-breasted Nuthatch 

1 

0 

Black-and-white Warbler 

13 

0 


Unknown Owl 

1 

0 

Common Yellowthroat 

9 

0 


Unknown Wren 

1 

0 

Yellow-billed Cuckoo 

9 

0 


Canada Goose 

0 

1 

Belted Kingfisher 

9 

0 


Killdeer 

0 

1 

Northern Mockingbird 

8 

0 


Chimney Swift 

0 

1 

Winter Wren 

7 

0 


Tree Swallow 

0 

6 

Great Blue Heron 

6 

0 


Cooper's Hawk 

0 

1 

American Goldfinch 

6 

1 


Merlin 

0 

1 

Unknown Empidonax 

6 

0 


Red-tailed Hawk 

0 

2 

Northern Bobwhite 

6 

0 


Turkey Vulture 

0 

8 

Indigo Bunting 

5 

1 


Black Vulture 

0 

1 

Northern Parula 

4 

0 






Table 10. Number of ground and flyover detections per species recorded during 764 morning and 448 evening 
transect surveys conducted during fall migration at Eglin AFB in 2005-2007. Species highlighted in bold were 

included in all statistical analyses. 


Species 

Ground 

Flyovers 


Species 

Ground 

Flyovers 

Gray Catbird 

5884 

5 


Scarlet Tanager 

9 

0 

Northern Cardinal 

2609 

3 


American Robin 

8 

0 

Blue Jay 

2041 

25 


Yellow-bellied Flycatcher 

8 

0 





ERDC/EL TR-12-22 


56 


Species 

Ground 

Flyovers 


Species 

Ground 

Flyovers 

Carolina Wren 

2016 

0 


Eastern Screech Owl 

8 

0 

Red-bellied Woodpecker 

1633 

1 


Swamp Sparrow 

7 

0 

Wood Thrush 

1311 

0 


Orange-crowned Warbler 

6 

0 

Carolina Chickadee 

961 

3 


Yellow-throated Vireo 

5 

0 

Eastern Tufted Titmouse 

834 

3 


Great Blue Heron 

4 

0 

Pileated Woodpecker 

637 

6 


American Kestrel 

4 

2 

White-eyed Vireo 

595 

2 


Sharp-shinned Hawk 

4 

0 

Yellow-shafted Flicker 

509 

9 


Cedar Waxwing 

3 

0 

Pine Warbler 

507 

4 


Unknown Raptor 

3 

4 

Unknown Passerine 

416 

66 


Brown-headed Cowbird 

3 

0 

Downy Woodpecker 

380 

1 


Turkey Vulture 

3 

45 

Common Yellowthroat 

359 

0 


Blue-headed Vireo 

3 

0 

Blue-gray Gnatcatcher 

310 

0 


Worm-eating Warbler 

3 

0 

Eastern Towhee 

284 

1 


Black Vulture 

3 

4 

American Crow 

229 

17 


Chipping Sparrow 

2 

0 

Unknown Warbler 

217 

0 


Brown Creeper 

2 

0 

Brown Thrasher 

199 

1 


Broad-winged Hawk 

2 

1 

American Redstart 

138 

0 


Red-headed Woodpecker 

2 

0 

Unknown Thrush 

135 

12 


Great Egret 

2 

1 

Red-shouldered Hawk 

123 

5 


Golden-crowned Kinglet 

2 

0 

Eastern Phoebe 

118 

0 


Blackburnian Warbler 

2 

0 

Unknown Bird 

97 

15 


Canada Warbler 

2 

0 

Yellow-bellied Sapsucker 

95 

0 


Great Crested Flycatcher 

2 

1 

Swainson's Thrush 

91 

0 


Kentucky Warbler 

2 

0 

Common Grackle 

87 

6 


Nashville Warbler 

2 

0 

Hooded Warbler 

86 

1 


Yellow-breasted Chat 

2 

0 

Brown-headed Nuthatch 

83 

0 


Bewick's Wren 

2 

0 

Hairy Woodpecker 

78 

0 


Unknown Sparrow 

2 

0 

Barred Owl 

75 

0 


Mallard 

2 

0 

House Wren 

72 

0 


House Finch 

2 

0 

Unknown Woodpecker 

57 

1 


White-breasted Nuthatch 

2 

0 

Eastern Wood-Pewee 

52 

0 


Purple Finch 

1 

0 

Ruby-crowned Kinglet 

46 

0 


Cooper's Hawk 

1 

0 

Eastern Bluebird 

44 

6 


Mississippi Kite 

1 

1 

Summer Tanager 

44 

0 


Red-winged Blackbird 

1 

0 

Magnolia Warbler 

43 

0 


Common Loon 

1 

0 

Red-eyed Vireo 

42 

0 


Unknown Heron 

1 

0 

Wild Turkey 

42 

0 


Sedge Wren 

i 

0 

Blue Grosbeak 

40 

2 


Unknown Nightjar 

i 

0 

Acadian Flycatcher 

31 

0 


Vesper Sparrow 

i 

0 




ERDC/EL TR-12-22 


57 


Species 

Ground 

Flyovers 


Species 

Ground 

Flyovers 

Black-and-white Warbler 

28 

0 


Winter Wren 

1 

0 

Unknown Empidonax 

27 

0 


Baltimore Oriole 

1 

0 

Yellow-billed Cuckoo 

26 

0 


Black-throated Blue Warbler 

1 

0 

Ruby-throated Hummingbird 

23 

0 


Indigo Bunting 

1 

0 

Rose-breasted Grosbeak 

23 

0 


MacGillivray's Warbler 

1 

0 

Veery 

23 

0 


Northern Parula 

1 

0 

Great Horned Owl 

23 

0 


Olive-sided Flycatcher 

1 

0 

Gray-cheeked Thrush 

21 

0 


Prothonotary Warbler 

1 

0 

Unknown Wren 

21 

0 


Unknown Vireo 

1 

0 

Belted Kingfisher 

21 

1 


Yellow Warbler 

1 

0 

Mourning Dove 

20 

1 


American Woodcock 

1 

0 

Northern Mockingbird 

19 

0 


Wood Duck 

1 

1 

Oven bird 

17 

0 


Fish Crow 

1 

0 

Hermit Thrush 

16 

0 


Red-cockaded Woodpecker 

1 

0 

Black-throated Green Warbler 

14 

0 


Unknown Dove 

1 

0 

Tennessee Warbler 

14 

0 


Osprey 

0 

1 

Unknown Grackle 

13 

0 


Unknown Accipiter 

0 

1 

Chestnut-sided Warbler 

12 

0 


Unknown Buteo 

0 

1 

American Goldfinch 

10 

4 


Unknown Hawk 

0 

3 

Chimney Swift 

9 

5 


Boat-tailed Grackle 

0 

2 


actually exhibited a negative trend for six of the nine sampling seasons 
modeled (Figure 38) and seven of these models explained less than 20% of 
the variation in the data. The model for Eglin AFB fall 2006 explained the 
largest proportion of the variance of any of these models (51%), but exami¬ 
nation of the data indicates that the trend line may be heavily influenced by 
two data points representing abnormally large migratory events (Figure 
38b). Nonetheless, the autoregressive error model parameter was the only 
significant model parameter in any sampling season. 

Migrant exodus densities were also not a significant predictor of A migrant 
abundance observed during ground surveys for any region during either 
season in any year (Table 13). In fact, none of the linear models created for 
each of the nine sampling seasons explained > 16% of the variation in the 
data. Moreover, five of the regression lines had positive slopes (Figure 39), 
which was the opposite of the expected trend. The autoregressive error 
model parameter was significantly greater than o in 5 of the models 
indicating auto-correlated error terms. 




ERDC/EL TR-12-22 


58 


a.2006 


■A Migrant Abundance 
Exodus Density 
Peak Migration Density 



b. 2007 



90 95 100 105 110 115 120 125 130 135 140 

Julian Date 


Figure 34. Comparison of the daily change in nocturnal migrant survey abundance with daily exodus and 
peak migration densities (calculated from radar reflectivity) during spring migration of a) 2006 and b) 2007 
near Yuma Proving Ground. The graphs depict 2 y-axes; values on the left axis are in birds/km and 
represent A migrant abundance while the right axis is measured in mean birds birds per km 3 and represents 
exodus and peak migration densities. For A migrant abundance, values represent the number of migrants 
recorded on the morning of the plotted date minus the number of migrants recorded on the previous 
morning. Exodus and peak migration densities represent radar imagery captured in the early hours of the 
plotted date or the late hours of the previous evening, respectively. Note the differences in scale between 

figures a and b. 






















ERDC/EL TR-12-22 


59 


a.2006 


•A Migrant Abundance 
Exodus Density 
Peak Migration Density 


15 700 



80 90 100 110 120 130 140 

Julian Date 


b. 2007 



2000 

1800 

1600 

1400 

1200 

1000 

800 

600 

400 

200 

0 


2 

u 

o_ 

00 

2 


Figure 35. Comparison of the daily change in nocturnal migrant survey abundance with daily exodus and 
peak migration densities (calculated from radar reflectivity) during spring migration of a) 2006 and b) 2007 
at Ft. Polk. The graphs depict 2 y-axes; values on the left axis are in birds/km and represent A migrant 
abundance while the right axis is measured in mean birds per km 3 and represents exodus and peak 
migration densities. For A migrant abundance, values represent the number of migrants recorded on the 
morning of the plotted date minus the number of migrants recorded on the previous morning. Exodus and 
peak migration densities represent radar imagery captured in the early hours of the plotted date or the late 
hours of the previous evening, respectively. Note the differences in scale between figures a and b. 




















ERDC/EL TR-12-22 


60 


a.2005 


A Migrant 
Abundance 



b. 2006 


20 1500 



270 275 280 285 290 295 

Julian Date 


c. 2007 



Figure 36. Comparison of the daily change in nocturnal migrant survey 
abundance with daily exodus and peak migration densities (calculated 
from radar reflectivity) during fall migration of a) 2005, b) 2006 and 
c) 2007 at Eglin AFB. The graphs depict 2 y-axes; values on the left axis 
are in birds/km and represent A migrant abundance while the right axis 
is measured in mean birds per km 3 and represents exodus and peak 
migration densities. For A migrant abundance, values represent the 
number of migrants recorded on the morning of the plotted date minus 
the number of migrants recorded on the previous morning. Exodus and 
peak migration densities represent radar imagery captured in the early 
hours of the plotted date or the late hours of the previous evening. Note 
the differences in scale between figures a, b and c. 



















ERDC/EL TR-12-22 


61 


a.2005 


■A Migrant Abundance 
Exodus Density 
Peak Migration Density 



5 

u 

Q. 

GO 

2 


b. 2006 



s 

u 

Q. 

CQ 

2 


Figure 37. Comparison of the daily change in nocturnal migrant survey abundance with daily exodus and 
peak migration densities (calculated from radar reflectivity) during fall migration of a) 2005 and b) 2006 at 
Ft. Polk. The graphs depict 2 y-axes; values on the left axis are in birds/km and represent A migrant 
abundance while the right axis is measured in mean birds per km 3 and represents exodus and peak 
migration densities. For A migrant abundance, values represent the number of migrants recorded on the 
morning of the plotted date minus the number of migrants recorded on the previous morning. Exodus and 
peak migration densities represent radar imagery captured in the early hours of the plotted date or the late 
hours of the previous evening. Note the differences in scale between figures a and b. 
















ERDC/EL TR-12-22 


62 


Table 11. Sample sizes for regression models built to explain A migrant abundance as 
a function of peak migration densities and migrant exodus densities captured on radar 
during spring and fall migration at three military installations. 


Season 

Region 

Year 

Peak 

Exodus 

Spring 

Yuma 

2006 

36 

32 

2007 

39 

40 

Ft. Polk 

2006 

26 

24 

2007 

31 

20 

Fall 

Eglin AFB 

2005 

8 

8 

2006 

17 

17 

Ft. Polk 

2007 

14 

13 

2005 

9 

11 

2006 

12 

14 


Table 12. Parameter estimates 3 (± standard errors b ) and fit statistics for linear regression models built to explain 
the daily change in migrant abundance recorded during bird surveys as a function of peak migration densities 

captured on radar. 


Season 

Region 

Year 

Intercept 

Peak Migration 

0 

9- 

R 2 

Spring 

Yuma 

2006 

4.50 

(4.12) 

-0.09 

(0.07) 

0.33 

(0.17) 

0.05 

2007 

-1.10 

(1.32) 

0.02 

(0.02) 

0.52 

(0.15) 

0.01 

Ft. Polk 

2006 

-0.23 

(1.25) 

< 0.01 

(< 0.01) 

0.26 

(0.23) 

0.29 

2007 

0.18 

(1.35) 

< 0.01 

(< 0.01) 

0.43 

(0.17) 

0.18 

Fall 

Eglin 

2005 

1.87 

(0.95) 

< 0.01 

(< 0.01) 

-0.21 

(0.73) 

0.16 

2006 

-0.06 

(1.36) 

0.01 

(< 0.01) 

0.61 

(0.26) 

0.51 

2007 

-1.12 

(1.62) 

0.01 

(< 0.01) 

0.50 

(0.25) 

0.12 

Ft. Polk 

2005 

2.15 

(1.40) 

-0.03 

(0.01) 

-0.36 

(0.44) 

0.03 

2006 

0.39 

(0.62) 

< 0.01 

(< 0.01) 

0.42 

(0.40) 

0.03 


a Parameter estimates which were significantly different from 0 are indicated in bold. 


standard error estimates are indicated in parentheses next to the parameter estimate. 

cParameter estimate for the autoregressive error model indicating the magnitude of the effect of the error term for dayt-i on dayt. 


For eight of the nine sampling seasons, peak migration densities were not a 
significant predictor of positive migrant turnover, and linear models did not 
explain > 19% of the variance in their respective data sets (Table 14). For Ft. 
Polk, fall 2006 peak migration density actually showed a significant 
negative effect on positive species turnover (Figure 40d). This trend is 
opposite of what was expected as it indicates that larger densities of birds 
migrating over the region result in more migrant species leaving the region. 
While this model explains 37% of the variance in the data for that season, 
the trend line is likely heavily influenced by a large migratory event on one 





a. Yuma (spring) 


60 

40 

8 20 

s 

■a 0 

5 -20 

< 

<1 -40 
-60 


-80 



R 2 = 0.01 


O 


R 2 = 0.05 


o 


b. Eglin AFB (fall) 


02005 



MBPCKM 


c. Ft. Polk (spring) 


d. Ft. Polk (fall) 



Figure 38. Plots of linear regression models built to explain the daily change in migrant abundance recorded during bird surveys as a function of 
peak migration densities captured on radar at military installations during spring and fall migration. 


ERDC/EL TR-12-22 



























ERDC/EL TR-12-22 


64 


Table 13. Parameter estimates 8 (± standard errors b ) and fit statistics for linear regression models built to explain 
the daily change in migrant abundance recorded during bird surveys as a function of migrant exodus densities 

captured on radar. 


Season 

Region 

Year 

Intercept 

Exodus 

0 

0- 

R 2 

Spring 

Yuma 

2006 

7.60 

(4.39) 

-0.05 

(0.03) 

0.27 

(0.19) 

0.09 

2007 

-0.22 

(0.97) 

< 0.01 

(0.01) 

0.52 

(0.14) 

< 0.01 

Ft. Polk 

2006 

-1.10 

(1.04) 

< 0.01 

(0.01) 

0.55 

(0.21) 

< 0.01 

2007 

0.29 

(1.53) 

-0.02 

(0.01) 

0.58 

(0.21) 

0.16 

Fall 

Eglin 

2005 

0.92 

(2.04) 

< 0.01 

(< 0.01) 

-0.23 

(0.84) 

< 0.01 

2006 

-0.21 

(1.83) 

< 0.01 

(< 0.01) 

0.56 

(0.26) 

0.08 

2007 

- 2.35 

(1.86) 

< 0.01 

(< 0.01) 

0.50 

(0.28) 

0.12 

Ft. Polk 

2005 

-0.06 

(0.94) 

0.01 

(0.01) 

0.53 

(0.39) 

0.03 

2006 

0.20 

(0.46) 

< 0.01 

(< 0.01) 

0.57 

(0.25) 

0.08 


a Parameter estimates which were significantly different from 0 are indicated in bold. 


standard error estimates are indicated in parentheses next to the parameter estimate. 

c Parameter estimate for the autoregressive error model indicating the magnitude of the effect of the error term for dayt-i on dayt. 


evening that was approximately twice as large as those recorded on all other 
evenings. For all nine models the intercept was significantly greater than 
zero, indicating that a certain number of species are present or detected on 
day x which were not on day x -i regardless of the migration intensity between 
days. 

As with peak migration density, there was also no significant effect of 
migrant exodus density during eight of nine sampling seasons on negative 
migrant turnover, and five of those models explained < 5% of the variance 
in the respective data (Table 15). Migrant exodus density did have a 
significant negative effect on A migrant abundance at Eglin AFB in the fall 
of 2007 (Figure 41), but again this was the opposite of what would have 
been expected, as it indicates that smaller exodus events result in more 
species leaving the region. This model explains 37% of the variance in the 
data and the model for Eglin AFB fall 2007 explains 30% of the variance in 
the data, yet these negative trend lines are counterintuitive. The intercept 
for all nine models was significantly greater than o, indicating that a certain 
number of species are not present or detected on day x , which were on day x -i, 
regardless of the migrant exodus intensity between days. 

Finally, treating peak migration and exodus densities as categorical 
variables did not yield different results. There was no significant effect of 
high vs. low peak migration densities or high vs. low migrant exodus 




a. Yuma (spring) 



MBPCKM 


b. Eglin AFB (fall) 


02005 

A2006 



MBPCKM 


c. Ft. Polk (spring) 


d. 


Ft. Polk (fall) 



Figure 39. Plots of linear regression models built to explain the daily change in migrant abundance recorded during bird surveys as a function of 
migrant exodus densities captured on radar at military installations during spring and fall migration. 


o 

tfl 


ERDC/EL TR-12-22 




















ERDC/EL TR-12-22 


66 


Table 14. Parameter estimates 8 (± standard errors 13 ) and fit statistics for linear regression models built to explain daily 
positive migrant turnover recorded during bird surveys as a function of peak migration densities captured on radar. 


Season 

Region 

Year 

Intercept 

Peak Migration 

0 

9- 

R 2 

Spring 

Yuma 

2006 

8.83 

(0.82) 

-0.01 

(0.01) 

0.35 

(0.17) 

0.01 

2007 

7.37 

(0.63) 

0.02 

(0.01) 

0.32 

(0.16) 

0.10 

Ft. Polk 

2006 

4.29 

(0.30) 

< 0.01 

(< 0.01) 

0.67 

(0.16) 

< 0.01 

2007 

6.15 

(0.79) 

< 0.01 

(< 0.01) 

0.02 

(0.20) 

< 0.01 

Fall 

Eglin 

2005 

3.97 

(1.05) 

< 0.01 

(< 0.01) 

0.77 

(0.31) 

0.06 

2006 

4.89 

(0.58) 

< 0.01 

(< 0.01) 

0.18 

(0.26) 

0.19 

2007 

5.37 

(0.57) 

< 0.01 

(< 0.01) 

0.41 

(0.29) 

0.01 

Ft. Polk 

2005 

6.70 

(1.21) 

-0.01 

(0.02) 

0.54 

(0.36) 

0.08 

2006 

3.78 

(0.31) 

< 0.01 

(< 0.01) 

0.84 

(0.14) 

0.37 


a Parameter estimates which were significantly different from 0 are indicated in bold, 
standard error estimates are indicated in parentheses next to the parameter estimate. 

c Parameter estimate for the autoregressive error model indicating the magnitude of the effect of the error term for dayt-i on dayt. 


densities on A migrant abundance for any region during either season in 
any year (Figure 42). In fact, not a single predicted A migrant abundance 
value was significantly different from o for any combination of treatments 
in any sampling season. There was apparently an enormous amount of 
variation in A migrant abundance values which simply could not be 
explained by peak migration and migrant exodus values alone. 

Discussion 

The team’s results did not show any significant relationships between 
ground-based avian transect data and two radar measures of bird density 
over the three study regions in either the eastern or western United States. 
As far as is known, this is the first study to investigate the relationship 
between daily changes in migrant communities recorded using field surveys 
and nightly radar data. Interestingly, Buler and Diehl (2009) had very 
different results when they analyzed the correspondence between radar and 
ground-estimated bird densities at 24 survey sites in Mississippi and 
Louisiana. Their study involved averaging ground and radar data over 
seasons and using a large number of sites as replicates, whereas the present 
study used days as replicates within a small number of sites. Unfortunately, 
the present team was unable to take the same analytical approach as Buler 
and Diehl for a direct comparison because, in the context of the present 
team’s experimental design, there were only three study sites. However, 
these results seem to indicate that daily variation associated with migrant 
movements or detection probabilities may be too high to allow for an 
accurate daily estimate of input or exodus due to migration. 




a. Yuma (spring) 


25 



0 50 100 

MBPCKM 


c. Ft. Polk (spring) 


O 


R 2 = 0.10 ^ 


_O 

R 2 = 0.01 

▲ 

-r -e-1 

150 200 


b. Eglin AFB (fall) 02005 

A2006 



d. Ft. Polk (fall) 




Figure 40. Plots of linear regression models built to explain daily positive migrant turnover recorded during bird surveys as a function of peak 
migration densities captured on radar at military installations during spring and fall migration. 


O) 


ERDC/EL TR-12-22 






























ERDC/EL TR-12-22 


68 


Table 15. Parameter estimates 8 (± standard errors 13 ) and fit statistics for linear regression models built to explain daily 
negative migrant turnover recorded during bird surveys as a function of migrant exodus densities captured on radar. 


Season 

Region 

Year 

Intercept 

Exodus 

-G 

0 

R 2 

Spring 

Yuma 

2006 

9.31 

(0.56) 

< 0.01 

(< 0.01) 

0.66 

(0.14) 

0.04 

2007 

7.95 

(0.33) 

< 0.01 

(< 0.01) 

0.41 

(0.16) 

0.03 

Ft. Polk 

2006 

7.27 

(1.24) 

< 0.01 

(0.01) 

-0.60 

(0.18) 

0.02 

2007 

6.89 

(0.80) 

< 0.01 

(< 0.01) 

0.28 

(0.24) 

0.05 

Fall 

Eglin 

2005 

6.76 

(1.38) 

< 0.01 

(< 0.01) 

0.86 

(0.19) 

0.30 

2006 

5.42 

(0.59) 

< 0.01 

(< 0.01) 

-0.08 

(0.28) 

0.01 

2007 

6.34 

(0.35) 

<0.01 

(< 0.01) 

0.67 

(0.22) 

0.37 

Ft. Polk 

2005 

4.58 

(1.83) 

0.02 

(0.02) 

-0.18 

(0.47) 

0.11 

2006 

3.42 

(0.36) 

< 0.01 

(< 0.01) 

0.72 

(0.20) 

0.09 


a Parameter estimates which were significantly different from 0 are indicated in bold, 
standard error estimates are indicated in parentheses next to the parameter estimate. 

c Parameter estimate for the autoregressive error model indicating the magnitude of the effect of the error term for dayt-i on dayt. 


The hypothesis that change in daily migrant abundance can be explained by 
nightly migrant exodus or input assumes that virtually all change in migrant 
abundance stems from birds either arriving or leaving as part of their 
migratory journey. It does not take into account that birds may remain 
resident in stopover habitat for varying amounts of time, or move short 
distances within a small geographical area in search of resources. Residence 
time by transient migrants in stopover habitat varies and is a function of 
many factors, including weather (e.g., prevailing winds, weather fronts), 
habitat quality, species, and physiological condition of individual birds 
(reviewed in Moore et al. 2005). Also, birds utilizing a region as stopover 
habitat do not have territories established as they would on their breeding 
grounds and thus may not be found in the same place on consecutive days, 
despite being present. Migrants often move in flocks, especially during the 
fall, and changes in abundance recorded during transect surveys could be 
heavily impacted by whether or not one of these flocks were present near 
the transect on a given day independently of when those birds actually 
arrive or depart. Similarly, the hypothesis does not take into account that 
individuals, despite being primarily nocturnal migrants, may actually arrive 
or depart during the day, avoiding capture on nightly radar images (Lowery 
1955). Moreover, daily changes in the detectability of species due to 
weather-related factors could also confound the results. Large fluctuations 
in daily abundance of permanent residents during some of our study 
seasons lend credence to the theory that not all changes in the number of 
migrant birds encountered could be attributed to individuals leaving or 
arriving as part of their migratory journey. 




a. Yuma (spring) 



c. Ft. Polk (spring) 



b. Eglin AFB (fall) 02005 


A2006 



d. Ft. Polk (fall) 



Figure 41. Plots of linear regression models built to explain daily negative migrant turnover recorded during bird surveys as a function of migrant 

exodus densities captured on radar at military installations during spring and fall migration. 


ERDC/EL TR-12-22 





















30 


0J 
C J 

s 

d 

d 

s 

5 

pQ 


20 

10 

0 


-10 

-20 


<1 


-30 


a. Yuma (spring) 

2006: p = 0.52 
2007: p = 0.21 



d 

d 


pfi 


d 

in 

DX 


25 

20 

15 

10 

5 

0 

-5 


<1 


-10 


b. Eglin AFB (fall) 

2005: p = 0.84 

2006: p = 0.81 

2007: p = 0.65 



■ High Low 

■ Low Low 

■ High High 

■ Low High 



-40 


QJ 

(J 

c 

d 

d 

S3 

d 

-o 

< 

ss 

d 

u 

OX 


8 

6 

4 

2 

0 

-2 

-4 

-6 


-10 

-12 

-14 


2006 2007 


c. Ft. Polk (spring) 

2006: p = 0.54 
2007: p = 0.43 



2006 2007 


-15 


2005 


2006 


2007 


d 

d 


d. Ft. Polk (fall) 

2005: p = 0.15 

10 2006: p = 0.61 

8 

6 - 
4 - 
2 - 


pQ 

d 0 
d 

OX -2 - 

-6 - 
-8 - 




2005 


2006 


Figure 42. Predicted change in migrant abundance values (and 95% confidence intervals) for days with different combinations of peak migration 
and exodus treatments. The first label for a bar indicates the peak migration treatment and the second indicates the exodus treatment. Places 
where a bar is missing from the graph indicate that the treatment combination did not exist in the particular sampling season. 


o 


ERDC/EL TR-12-22 






























































ERDC/EL TR-12-22 


71 


Another factor to consider is that the change in migrant abundance, which 
is attributable to birds actually arriving or departing the region, is in fact 
some combination of the two values. It was not possible to include both 
measures in the same regression models here due to high correlation 
between the values, and it was not possible to subtract one from another, 
due to the fact that the measures were not quite on par with one another. 
While it can be reasonably assumed that all individuals recorded via radar 
during an exodus event are leaving from the study region, an unknown 
proportion of birds detected by radar during peak nightly migration are 
settling down into the habitat. This is currently a difficult parameter to 
measure with any monitoring technique. Moreover, that proportion may 
change throughout the course of the night, depending on when the radar 
reflectivity values were calculated. Peckford and Taylor (2008), for instance, 
found that the correlation between their ground censuses and radar data 
varied throughout the course of a night and peaked just before sunrise on 
nights with unfavorable headwinds, and just after sunset on nights with 
favorable tailwinds. The fact that all radar reflectivity data for “input” were 
collected closer to sunset each night (and during the peak of exodus as 
shown by radar) may suggest that the team was not collecting “input” value 
at the right time of night. Radar data collected at a smaller spatial scale and 
over a larger temporal scale may be required to accurately estimate how 
many individuals are actually arriving on any given night. 

The ability to assign birds — captured on radar during an exodus event to 
specific stopover habitat — as aloft is a challenge that is being addressed by 
various researchers (e.g., Buler and Diehl 2009). The displacement, or 
distance, between actual stopover habitat and the location where birds enter 
the radar beam, is an issue and is related to several factors, including the 
distance of the birds from the radar, how quickly birds are climbing into the 
night sky, and atmospheric conditions that may affect bending of the radar 
beam (Diehl and Larkin 2004). The WSR-88D provides relatively coarse 1° 
x l-km (pulse volume) resolution cells. At that level of resolution, it can be 
difficult to be extremely specific about the origin of departing migrants. 
Bonter et al. (2009) successfully assigned birds detected aloft by WSR-88D 
radar to habitat on the ground. They concluded that radar is a powerful tool 
for identifying stopover habitats of migratory birds, especially when those 
habitats are discrete and easily identified via land cover maps. Their work 
linked areas of high migrant activity as evidenced by radar-indicated exodus 
events to various land-cover types in near-shore terrestrial habitats in the 
Great Lakes basin. Stopover concentrations of departing migrants were 



ERDC/EL TR-12-22 


72 


readily identified via radar as “exodus images.” Important habitats included 
forest fragments that were dispersed in a sea of agriculture and develop¬ 
ment. The team is confident that within the effective range of the radar 
beam from point of origin (within approximately 120 km; Gauthreaux and 
Belser 1998), it was possible to successfully identify stopover habitats at 
approximately the same scale as the radar. A couple of examples illustrate 
this point. First, work by Gauthreaux and Belser (1998, 2005) in southern 
Louisiana and coastal South Carolina strongly suggested that the geometry 
of reflectivity images from departing migrants were strongly associated with 
the landscape geometry of forested wetlands and riparian areas along major 
drainages. This was the main reason riparian habitats the team selected 
focused ground sampling. Second, at the Yuma, Arizona sites along the 
Colorado River, the only suitable stopover habitats were discrete riparian 
areas adjacent to the river (most other vegetated areas were either Sonoran 
desert or agriculture). Radar data showed that the geometry of the radar- 
indicated exodus is of similar shape to the available habitat on the ground. 

Since radar properly identified broad-scale hotspots at the three sites, 
another explanation for the lack of correspondence may be that migrants 
were keying in on specific microhabitats that are not identifiable at the 
resolution used with the WSR-88D imagery. In other words, exodus 
imagery may allow the proper labeling of a particular riparian system as a 
hotspot, but birds may be localized in their habitat use within these systems. 
Thus, it maybe possible that the ground surveyors missed large groups of 
transient migrants using these microhabitats. This could be especially true 
in the eastern sites where riparian areas are embedded in a much larger 
matrix of upland forest. In heavily forested landscapes with few identifiable 
discrete habitat patches (i.e., discriminated via aerial imagery), such as what 
is typically found at the study sites on Eglin AFB and Fort Polk, the 
identification of stopover habitats via radar imagery can be more difficult. 
The team detected very few transient migrants each day at the eastern sites 
during both spring and fall and were unable to ascertain whether migrants 
that were stopping over were using very specific microhabitats in radar- 
indicated hotspots, or whether they were dispersed over larger areas within 
these extensive drainages and across adjacent ecotone and upland habitats. 
If the former is true, and the team didn’t sample in the exact microhabitats 
where migrants were concentrated, correlations would be expected to be 
low. In the latter case, large numbers of transient migrants dispersed widely 
throughout the extensive forested habitat would be difficult to detect in 
sufficient numbers to show strong relationships with radar data. However, 



ERDC/EL TR-12-22 


73 


transect sampling covered >2.5 km of stopover habitat as delimited by the 1 
km radar resolution images, suggesting that a significant amount of habitat 
within radar-indicated hotspots were covered and should have traversed at 
least some small microhabitats where large flocks would have been detec¬ 
ted. This is especially true at Yuma sites, where migrants were concentrated 
in very discrete riparian stopover habitats and where the majority of hotspot 
habitat (as indicated by radar) was sampled. This lends further evidence 
that abundance estimates derived solely from transect surveys of birds in 
stopover habitat simply are not a good metric for comparison with radar¬ 
generated density estimates of birds aloft during exodus. 

Additionally, observer bias issues could have contributed significantly to 
the lack of correspondence between data types. The team attempted to 
control variation in observer abilities by using skilled birders and having 
individuals rotate among sites on a daily basis whenever possible. A 
thorough analysis of ground data, however, showed that observer abilities 
to detect migrants did vary significantly. To reduce this problem, the team 
pooled data across sites and developed a single metric of abundance for 
each region; yet there was likely still some variation introduced by the 
rotating observers. In retrospect, it may have been prudent to keep each 
observer at a single site within seasons. 

Several recent studies have attempted to relate radar data to avian 
information collected on the ground. Each of these studies met varying 
degrees of success and all used different approaches to seeking a relation¬ 
ship between the two. Three of these studies have shown that mist-netting is 
perhaps the ground-sampling method which yields results most reflective of 
migratory events captured on radar (Zehnder et al. 2001, Simons et al. 

2004, Peckford and Taylor 2008). Simons et al. (2004) found that the 
number of migrants captured per net-hour in southern Louisiana was 
significantly correlated with WSR-88D-indicated numbers of migrants aloft. 
Peckford and Taylor (2008) used three indices of birds on the ground, 
including mist-netting, and found significant correlations with migrants 
detected along the Maine coast with a small mobile marine radar. Unlike 
mist-netting, transect censuses cannot tell researchers whether the 
migrants they are detecting have been resident in stopover habitat for 
several days or are the result of migration input. Similarly, on days with low 
numbers of detections, it cannot be discriminated whether birds departed 
on migration or whether the birds may have dispersed through the 
landscape beyond the transects and did not migrate. Daily numbers of 



ERDC/EL TR-12-22 


74 


transient migrants varied greatly at each of the study regions. Future 
research should focus on gaining a better understanding of this variation, as 
it has significant implications for interpreting the results obtained. 

Lastly, finer-scale radar data, which is not currently possible with the WSR- 
88-D, would have been useful for generating metrics for individual sites. 

The resolution of the WSR-88D is too coarse to detect individual birds as 
they depart from migration stopover areas. High resolution mobile radar 
units like eBIRDRAD, can be moved and placed in strategic locations where 
individual birds can be tracked as they depart from migration stopover 
areas. Such an effort would allow quantification of echo size, flight direction, 
flight speed, and quantity of migrant birds leaving from different habitat 
types. For example, the bird survey data collected along riparian transects 
on the All-American Canal just North of Yuma, Arizona suggested that 
migrant bird densities varied significantly among native versus non-native 
reaches of the transect, with migrants being highly concentrated in native 
shrub/tree habitats. Although the WSR-88D Doppler weather radar was 
used to locate important migration stopover areas near Yuma, the 
resolution of the WSR-88D is too coarse (i° x l km for base reflectivity data 
and 1° x 0.25 km for base radial velocity data) to detect individual birds as 
they depart from migration stopover areas. The resolution cells contained 
many different targets and the data reported for the resolution cells are 
average values. Because eBIRDRAD is mobile it can be moved to several 
stopover areas to gather high-resolution data on habitat preferences of 
migrants. By strategically locating eBIRDRAD between habitat types, the 
team could attempt to quantify the number of migrant birds leaving from 
different habitat types. 

Near the end of the study, the team initiated some very promising 
collaborative work with Dr. Andrew Farnsworth (Cornell Laboratory of 
Ornithology) at the Yuma sites, using automated acoustic recording 
devices. Six of these devices were placed along the YUMA and AAC sites 
(three at each site) and set up to record nocturnal flight calls from dusk to 
dawn during each night of the 2007 sampling season. The team’s intention 
is to use these tools as a third measure of bird abundance and species 
richness. The combination of ground-based survey and acoustics data, 
along with radar data, will add an additional and interesting analysis of 
bird migration from this discrete stopover habitat. 



ERDC/EL TR-12-22 


75 


Objective 5: Avian Habitat Use in Southwestern Riparian Systems 

Data from 70 different sections was analyzed, and over the two years, the 
team recorded 20,665 migrant detections on the ground from 49 unique 
species (Table 16). In 2006,1,496 bird surveys were conducted 
(26-33 surveys per section) and in 2007, 2,896 bird surveys were 
conducted (37 and 42 per section). Yearly estimates of mean daily total 
migrant abundance for individual sections ranged from 5.9 birds/km to 
178.4 birds/km and yearly estimates of mean daily migrant species 
richness ranged from 0.5 species/section to 5.7 species/section. 

There was a significant overall habitat effect on both total migrant 
abundance (F = 15.89, P < 0.01) and migrant species richness (F = 15.64, P 
< 0.01), and the results from the analyses of these metrics were remarkably 
similar (Figure 43). The team encountered the greatest abundance and 
richness values in NS habitats, followed respectively by ND, NND, NT, and 
NNI habitat sections. In both cases, NNI habitats had significantly lower 
values than all but ND habitats, and NT habitats had significantly lower 
values than NS sections. 

There was strong evidence that individual species differed in their 
response to habitat type (F = 5.31, P < 0.01). Of the 20 species investigated 
individually, the abundance of 12 differed significantly between habitats 
while six did not (Table 16). Models for two species were not able to be 
constructed in PROC GLIMMIX. Nineteen of these species were found in 
all habitat types, while Yellow-headed Blackbirds were found in all but ND 
habitats. Peak densities were found of nearly 75% of species analyzed in 
NS communities, while no species peaked in NNI communities. In fact, 
abundances were lowest in NNI habitats for 40% of bird species. 

Discussion 

The team found that spring migrant abundance, species richness, and 
community composition are all affected by riparian vegetation composition. 
Results indicate that riparian habitats completely dominated by invasive 
saltcedar support fewer migrants and migrant species in the spring than 
other riparian habitat types in the study area. The presence of native 
vegetation, even in small concentrations, appears to greatly improve habitat 
value for birds at this time of year. Many species tended to prefer habitats 
comprised entirely or partially of native shrubs (which here include some 
shrubs more typical of upland systems), rather than riparian trees. 



ERDC/EL TR-12-22 


76 


Table 16. Mean abundance per kilometer for all migrant species recorded at 125 m transect sections of various 
habitat types during spring migration near Yuma Proving Ground in 2006 and 2007. The 20 most abundant 
species are indicated with an asterisk, and were tested for statistical differences among habitat types. Habitat 
types that were not statistically different from one another with regard to abundance of a species share a letter. It 
was not possible to construct appropriate habitat models for Orange-crowned Warbler or Bullock’s Oriole. 


Species 

Scientific Name 

NT 

NS 

ND 

NND 

NNI 

Total 

Counted 

^Wilson’s 

Warbler 

Wilsonia pusilla 

4.22 ± 
(0.65) 

AC 

8.73 ± 
(1.14) 

B 

4.32 ± 
(0.96) 

ABC 

5.36 ± 
(1.07) 

AB 

2.68 ± 
(0.40) 

C 

2817 

^Common 

Yellowthroat 

Geothlypis 

trichas 

9.16 ± 
(1.43) 

A 

2.13 ± 
(0.55) 

B 

1.97 ± 
(1.51) 

AB 

5.61 ± 
(1.75) 

A 

5.1 ± 
(1.52) 

AB 

2396 

* Brewer’s 
Sparrow 

Spizella breweri 

0.03 ± 
(0.03) 

AB 

10.29 ± 
(4.25) 

B 

10.63 ± 
(10.44) 

AB 

0.94 ± 
(0.32) 

A 

0.59 ± 
(0.35) 

A 

2069 

*Ash-throated 

Flycatcher 

Myiarchus 

cinerascens 

0.77 ± 
(0.14) 

A 

4.73 ± 
(0.47) 

B 

4.22 ± 
(1.35) 

BC 

3.16 ± 
(0.78) 

BC 

2.39 ± 
(0.33) 

C 

1617 

Unknown 

Warbler 

— 

2.6 ± 
(0.49) 


2.61 ± 
(0.34) 


1.86 ± 
(0.44) 


3.31 ± 
(0.81) 


1.67 ± 
(0.20) 


1255 

^Western 

Flycatcher 

— 

1.02 ± 
(0.16) 

A 

3.85 ± 
(0.33) 

B 

3.81 ± 
(0.62) 

BC 

1.96 ± 
(0.36) 

AC 

1.08 ± 
(0.18) 

A 

1230 

*Warbling 

Vireo 

Vireo gilvus 

1.28 ± 
(0.30) 

A 

3.21 ± 
(0.44) 

B 

2.44 ± 
(0.79) 

AB 

2.66 ± 
(0.57) 

B 

1.05 ± 
(0.16) 

A 

1128 

Grange- 

crowned 

Warbler 

Vermivora 

celata 

2.29 ± 
(0.40) 


2.84 ± 
(0.30) 


1.49 ± 
(0.45) 


1.84 ± 
(0.42) 


1.45 ± 
(0.24) 


1066 

*Yel low¬ 
headed 

Blackbird 

Xanthocephalus 

xanthocephalus 

6.48 ± 
(2.31) 

A 

0.3 ± 
(0.12) 

B 

0±(0) 

AB 

0.62 ± 
(0.36) 

B 

0.49 ± 
(0.22) 

B 

781 

*Nashville 

Warbler 

Vermivora 

ruficapilla 

0.5 ± 
(0.10) 

AC 

1.9 ± 
(0.27) 

B 

1.59 ± 
(0.59) 

ABC 

1.14 ± 
(0.26) 

AB 

0.43 ± 
(0.11) 

C 

600 

Unknown 

Empidonax 

— 

0.27 ± 
(0.06) 


1.47 ± 
(0.20) 


1.17 ± 
(0.27) 


1.15 ± 
(0.32) 


0.41 ± 
(0.08) 


460 

*Yel low- 
rum ped 

Warbler 

Dendroica 

coronata 

0.46 ± 
(0.14) 

A 

1.07 ± 
(0.15) 

A 

0.31 ± 
(0.15) 

A 

0.74 + 
(0.17) 

A 

0.74 ± 
(0.16) 

A 

406 

^Western 

Kingbird 

Tyrannus 

verticalis 

0.72 ± 
(0.18) 

A 

1.23 ± 
(0.35) 

A 

0.6 ± 
(0.37) 

A 

0.53 ± 
(0.14) 

A 

0.49 ± 
(0.13) 

A 

390 

^Western 

Tanager 

Piranga 

ludoviciana 

0.34 ± 
(0.10) 

A 

1.2 ± 
(0.24) 

B 

0.53 ± 
(0.23) 

AB 

0.86 ± 
(0.23) 

AB 

0.36 ± 
(0.06) 

A 

385 

* Black- 
throated Gray 
Warbler 

Dendroica 

nigrescens 

0.53 ± 
(0.11) 

AB 

1.08 ± 
(0.21) 

A 

1.19 ± 
(0.48) 

AB 

0.75 ± 
(0.18) 

AB 

0.19 ± 
(0.04) 

B 

380 

* Lucy’s 

Warbler 

Vermivora luciae 

0.01 ± 
(0.01) 

A 

0.81 ± 
(0.21) 

A 

0.92 ± 
(0.49) 

A 

1.36 ± 
(0.51) 

A 

0.33 ± 
(0.13) 

A 

365 

*MacGillivray’s 

Warbler 

Oporornis 

tolmiei 

0.41 ± 
(0.08) 

A 

0.86 ± 
(0.12) 

A 

1.05 ± 
(0.42) 

A 

0.67 ± 
(0.17) 

A 

0.18 ± 
(0.07) 

A 

328 

* Ye Now 

Warbler 

Dendroica 

petechia 

0.5 ± 
(0.12) 

A 

0.85 ± 
(0.13) 

A 

0.37 ± 
(0.15) 

A 

0.44 ± 
(0.10) 

A 

0.38 ± 
(0.09) 

A 

320 

Glue 

Grosbeak 

Passerina 

caerulea 

0.94 ± 
(0.16) 

A 

0.15 ± 
(0.05) 

B 

0.51 ± 
(0.15) 

AB 

0.63 ± 
(0.20) 

A 

0.64 ± 
(0.16) 

AB 

265 




ERDC/EL TR-12-22 


77 


Species 

Scientific Name 

NT 

NS 

ND 

NND 

NNI 

Total 

Counted 

^Townsend’s 

Warbler 

Dendroica 

townsendi 

0.49 ± 
(0.12) 

AB 

0.71 ± 
(0.12) 

A 

0.38 ± 
(0.10) 

AB 

0.59 ± 
(0.14) 

AB 

0.16 ± 
(0.04) 

B 

260 

*Bullock’s 

Oriole 

Icterus bullockii 

0.38 ± 
(0.12) 

_ 

0.93 ± 
(0.17) 

_ 

0.1 ± 
(0.07) 

- 

0.28 ± 
(0.06) 

_ 

0.24 ± 
(0.08) 

_ 

239 

Unknown 

Hummingbird 

— 

0.31 ± 
(0.12) 


0.58 ± 
(0.12) 


0.38 ± 
(0.20) 


0.44 ± 
(0.09) 


0.51 ± 
(0.14) 


237 

*Black- 

chinned 

Hummingbird 

Archilochus 

alexandri 

0.14 ± 
(0.05) 

A 

0.66 ± 
(0.17) 

A 

0.1 ± 
(0.04) 

A 

0.42 ± 
(0.14) 

A 

0.2 ± 
(0.07) 

A 

204 

Black-headed 

Grosbeak 

Pheucticus 

melanocephalus 

0.27 ± 
(0.09) 


0.39 ± 
(0.07) 


0.45 ± 
(0.13) 


0.38 ± 
(0.07) 


0.26 ± 
(0.07) 


179 

Western Wood- 

Pewee 

Contopus 

sordidulus 

0.09 ± 
(0.03) 


0.52 ± 
(0.06) 


0.35 ± 
(0.06) 


0.46 ± 
(0.10) 


0.1 ± 
(0.03) 


175 

White-crowned 

Sparrow 

Zonotrichia 

leucophrys 

0.08 ± 
(0.06) 


0.51 ± 
(0.12) 


0.66 ± 
(0.32) 


0.12 ± 
(0.04) 


0.22 ± 
(0.07) 


160 

Yellow¬ 
breasted Chat 

Icteria virens 

0.36 ± 
(0.10) 


0.19 ± 
(0.09) 


0.41 ± 
(0.26) 


0.76 ± 
(0.22) 


0.04 ± 
(0.02) 


160 

Chipping 

Sparrow 

Spizella 

passerina 

0±(0) 


0.37 ± 
(0.22) 


0.24 ± 
(0.16) 


0.31 ± 
(0.20) 


0.41 ± 
(0.25) 


152 

Lazuli Bunting 

Passerina 

amoena 

0.18 ± 
(0.07) 


0.3 ± 
(0.07) 


0.35 ± 
(0.25) 


0.18 ± 
(0.07) 


0.09 ± 
(0.03) 


111 

Bell’s Vireo 

Vireo bellii 

0±(0) 


0.2 ± 
(0.13) 


0±(0) 


0.26 ± 
(0.18) 


0±(0) 


67 

Blue-gray 

Gnatcatcher 

Polioptila 

caerulea 

0.05 ± 
(0.02) 


0.18 ± 
(0.05) 


0.03 ± 
(0.03) 


0.05 ± 
(0.03) 


0.08 ± 
(0.03) 


50 

Ruby-crowned 

Kinglet 

Regulus 

calendula 

0.1 ± 
(0.03) 


0.1 ± 
(0.03) 


0.22 ± 
(0.11) 


0.05 ± 
(0.03) 


0.07 ± 
(0.02) 


45 

Olive-sided 

Flycatcher 

Contopus 

cooperi 

0.03 ± 
(0.02) 


0.07 ± 
(0.02) 


0.1 ± 
(0.07) 


0.11 ± 
(0.04) 


0.06 ± 
(0.02) 


41 

Hermit 

Warbler 

Dendroica 

occidentalis 

0.07 ± 
(0.03) 


0.05 ± 
(0.02) 


0.03 ± 
(0.03) 


0.13 ± 
(0.05) 


0.1 ± 
(0.04) 


40 

Cassin’s Vireo 

Vireo cassinii 

0.03 ± 
(0.02) 


0.11 ± 
(0.03) 


0.06 ± 
(0.06) 


0.11 ± 
(0.04) 


0.01 ± 
(0.01) 


38 

Green-tailed 

Towhee 

Pipilo chlorurus 

0.03 ± 
(0.02) 


0.12 ± 
(0.03) 


0.06 ± 
(0.06) 


0.07 ± 
(0.03) 


0.02 ± 
(0.01) 


35 

Unknown 

Flycatcher 

— 

0.04 ± 
(0.02) 


0.07 ± 
(0.02) 


0.13 ± 
(0.09) 


0.03 ± 
(0.02) 


0.04 ± 
(0.01) 


33 

House Wren 

Troglodytes 

aedon 

0.02 ± 
(0.02) 


0.12 ± 
(0.04) 


0.1 ± 
(0.07) 


0.04 ± 
(0.03) 


0.02 ± 
(0.01) 


32 

Unknown Vireo 

— 

0.04 ± 
(0.02) 


0.1 ± 
(0.03) 


0±(0) 


0.05 ± 
(0.02) 


0.03 ± 
(0.01) 


32 

Summer 

Tanager 

Piranga rubra 

0±(0) 


0.03 ± 
(0.01) 


0±(0) 


0.18 ± 
(0.10) 


0.01 ± 
(0.01) 


26 

Say’s Phoebe 

Sayornis saya 

0.01 ± 
(0.01) 


0.06 ± 
(0.03) 


0±(0) 


0.02 ± 
(0.02) 


0.01 ± 
(0.01) 


13 

Unknown 

Myiarchus 

— 

0±(0) 


0.02 ± 
(0.01) 


0±(0) 


0.06 ± 
(0.02) 


0.01 ± 
(0.01) 


12 




ERDC/EL TR-12-22 


78 


Species 

Scientific Name 

NT 

NS 

ND 

NND 

NNI 

Total 

Counted 

Swainson’s 

Thrush 

Catharus 

ustulatus 

0.02 ± 
(0.02) 


0.03 ± 
(0.01) 


0±(0) 


0.02 ± 
(0.01) 


0.01 ± 
(0.01) 


10 

Unknown 

Thrush 

— 

0.02 ± 
(0.02) 


0.02 ± 
(0.01) 


0.03 ± 
(0.03) 


0.02 ± 
(0.01) 


0.01 ± 
(0.01) 


10 

Hermit Thrush 

Catharus 

guttatus 

0.02 ± 
(0.02) 


0.02 ± 
(0.02) 


0±(0) 


0.01 ± 
(0.01) 


0.01 ± 
(0.01) 


8 

Hooded Oriole 

Icterus 

cucullatus 

0±(0) 


0.05 ± 
(0.03) 


0±(0) 


0±(0) 


0±(0) 


8 

Unknown 

Oriole 

— 

0±(0) 


0.03 ± 
(0.02) 


0±(0) 


0±(0) 


0±(0) 


6 

Northern 

Parula 

Parula 

americana 

0±(0) 


0.02 ± 
(0.02) 


0±(0) 


0±(0) 


0±(0) 


4 

Willow 

Flycatcher 

Empidonax 

traillii 

0±(0) 


0.02 ± 
(0.01) 


0±(0) 


0±(0) 


0.01 ± 
(0.01) 


4 

Sage Thrasher 

Oreoscoptes 
montan us 

0.03 ± 
(0.02) 


0±(0) 


0±(0) 


0±(0) 


0.01 ± 
(0.01) 


3 

Calliope 

Hummingbird 

Stellula calliope 

0.01 ± 
(0.01) 


0.01 ± 
(0.01) 


0±(0) 


0±(0) 


0±(0) 


2 

Clay-colored 

Sparrow 

Spizella pallida 

0±(0) 


0.01 ± 
(0.01) 


0±(0) 


0.01 ± 
(0.01) 


0±(0) 


2 

Gray 

Flycatcher 

Empidonax 

wrightii 

0±(0) 


0.01 ± 
(0.01) 


0.04 ± 
(0.04) 


0±(0) 


0±(0) 


2 

Hammond’s 

Flycatcher 

Empidonax 

hammondii 

0±(0) 


0.01 ± 
(0.01) 


0±(0) 


0.01 ± 
(0.01) 


0±(0) 


2 

Lincoln’s 

Sparrow 

Melospiza 

lincolnii 

0±(0) 


0±(0) 


0.03 ± 
(0.03) 


0.01 ± 
(0.01) 


0±(0) 


2 

Plumbeous 

Vireo 

Vireo plumbeus 

0±(0) 


0±(0) 


0±(0) 


0.01 ± 
(0.01) 


0±(0) 


1 

Rufous 

Hummingbird 

Selasphorus 

rufus 

0±(0) 


0.01 ± 
(0.01) 


0±(0) 


0±(0) 


0±(0) 


1 

Unknown 

Tanager 

— 

0±(0) 


0±(0) 


0.03 ± 
(0.03) 


0±(0) 


0±(0) 


1 


NNI habitat sections supported significantly fewer migrants and species 
than almost all other habitat types. Previous investigations into migrant use 
of Saltcedar have produced mixed results. In the middle Rio Grande Valley 
of New Mexico, Kelly et al. (2000) recorded greater spring and fall migrant 
capture rates in riparian areas dominated by willows than other plants. 
Walker (2008), however, found that migrant abundance and energy con¬ 
sumption were actually highest in Saltcedar habitats during the fall. Avian 
use of Saltcedar may be influenced by climatic variables (Hunter et al. 
1988), but in the team’s study region, high concentrations of the invasive 
plant appear to reduce habitat value for migrants. While the investigation 
did not examine the cause for the more depauperate bird community in 
Saltcedar-dominated habitats, it maybe that such areas offer reduced 
structural complexity, cover, or food resources. 




ERDC/EL TR-12-22 


79 


a. 


70 

60 


S 40 

a 


pg 30 - 

u 

S 20 
10 


0 



NT NS ND NND 

Habitat 


I 

NNI 


b. 


ABC 



NT NS ND NND NNI 

Habitat 


Figure 43. Mean (±SE) total migrant abundance per kilometer (a) and migrant 
species richness per transect section (b) recorded at 125 m transect sections of 
different habitat types near Yuma Proving Ground during spring migration in 2006 
and 2007. Sections were classified as native tree (NT), native shrub (NS), native- 
dominated with non-natives (ND), non-native/invasive dominant with some natives 
(NND), or non-native/invasive shrub and tree community (NNI). Bars that do not have 
a letter in common indicate the response variable was significantly different between 

those habitat types. 


Van Riper et al. (2008) surveyed bird communities throughout the year 
just north of the present study area and found that abundance of many 
bird species was highest at intermediate concentrations of Saltcedar. 
These researchers suggested that there may be a threshold for Saltcedar 
composition, above and below which the avian habitat value is reduced. 
The results of the present study somewhat contradict this hypothesis; 
while NND sections, which included low levels of native vegetation, did 
attract more migrants than NNI sections, it was also found that habitats 




















ERDC/EL TR-12-22 


80 


completely dominated by native shrubs supported the greatest abundance 
and richness of migrant birds. Van Riper et al. (2008) did not investigate 
avian use of habitats dominated by shrub vegetation, but focused on plants 
that have historically been more typical of riparian communities. 

However, it is important to consider avian use of these habitats as well, 
given that it may not be realistic to expect regeneration of Cottonwood and 
Willows in systems that have been impacted by intense hydrologic 
alterations (Livingston and Schemnitz 1996; Sogge et al. 2008). 

While both NS and ND sections supported greater migrant richness, total 
migrant abundance, and abundance of several individual species than did 
NT sections, it is important to remember that all of the did NT sections 
included in the present analysis were located at Cottonwood restoration 
sites and were comprised of very young trees. When the team considered 
avian communities from the four mature Cottonwood sections surveyed in 
2007, it was found they also supported lower numbers of migrants (46.33 ± 
10.61 birds per kilometer) than NS sections. However, they had 
substantially greater species richness (3.69 ± 0.80 species per section) than 
any other habitat type; however, only a small number of mature 
Cottonwood sections were sampled in one of the sample years. Indeed, 

Szaro and Jakle (1985) found bird densities and species richness values in 
areas dominated by riparian trees were greater than or similar to those 
found in adjacent shrub communities. Thus, further research is necessary to 
compare migrant habitat use between mature native riparian trees and 
those dominated by other native shrubs. 

The United States Department of Agriculture recently decided to shut 
down its Saltcedar biological control program over concerns that it was 
destroying critical Willow Flycatcher nesting habitat. In light of this, it is 
important to note that nearly half of the migrant species the team 
detected, including Willow Flycatchers, were found in greatest abundance 
in shrub habitats. By contrast, nearly 35% of species were found in lowest 
abundance in NNI habitats. Consequently, allowing the uninhibited 
expansion of Saltcedar may have detrimental effects on many western 
migrant species by causing reductions in the availability of preferred 
stopover habitat. However, results of the present study also suggest that 
complete restoration of hydrologic processes to support native riparian 
tree communities may not be necessary to provide decent-quality habitat 
for en route migrants. Full consideration should be given to the annual 
life-cycle requirements of riparian species prior to making decisions about 
Saltcedar contol. 



ERDC/EL TR-12-22 


81 


5 Conclusions and Implications for Future 
Research and Implementation 

The national WSR-88D radar system proved to be an effective coarse tool 
for investigating migrant use of U.S. military installations. The team 
successfully downloaded and processed archived data to identify installa¬ 
tions with high density estimates of migrant use in the spring and the fall, to 
identify specific locations on installations with varying densities of migrants 
departing stopover habitat, to summarize temporal peaks in migration, and 
to develop relatively accurate migration forecast models. Such information 
can be used by natural resources personnel to identify priority habitats or 
areas within installations (or, in some cases, contiguous areas outside base 
boundaries) for conservation and management, and to help reduce the 
likelihood of BASH incidents by minimizing the overlap between training 
missions and large migratory events. While the migration forecast models 
that were developed had high explanatory power, they should be validated 
with radar and weather data collected in the future to fully understand their 
predictive capabilities. The authors of this study also recommend testing 
these models at other military installations in proximity to those investi¬ 
gated in this study. Such information will help determine the usefulness of 
these models and whether similar models should be developed for other 
regions or installations. 

WSR-88D radar data do have some limitations, however, that cannot be 
overcome with current technology. First, this system cannot be used to 
monitor migration on military installations that are located too far away 
from a NEXRAD radar. Second, WSR-88D radar data can, at best, only 
quantify migrant densities in 250 m 3 volumes, meaning that it cannot be 
used to identify smaller-scale migration hotspots or fine-scale movements 
on and around airfields. Third, WSR-88D radar data cannot provide 
detailed species composition information. Lastly, the team’s results 
indicated that changes in migrant density calculated from radar surveys 
and ground-based surveys expressed very different results, which may be 
related to differences in the scale of observations. The team suggests that 
future research comparing ground surveys in discrete patches of stopover 
habitat be compared with radar data collected with high-resolution, 
mobile eBIRDRAD units to help further understand the relationship 
between traditional and progressive survey methods. 



ERDC/EL TR-12-22 


82 


While it was only possible to closely examine stopover habitat on three 
installations, this broad-scale survey suggested that many other installa¬ 
tions support stopover habitats having high densities of migrants in both 
spring and fall migration seasons (Appendix B). It is the team’s recom¬ 
mendation that these installations use available resources to implement or 
continue migration monitoring using a combination of radar and ground- 
based surveys to inventory species and identify priority stopover habitats for 
conservation and management. These results also suggested that migrant 
use of such habitats is highly variable both spatially and temporally, so such 
information should be taken into account when planning inventory and 
monitoring programs. In particular, special care should be taken to monitor 
a variety of plant communities because the composition of vegetation seems 
to have a significant impact on avian use. 

Finally, the team’s collaborative work with Dr. Andrew Farnsworth and 
acoustics (Cornell Laboratory of Ornithology) at the Yuma sites may provide 
additional information on sound-based indices of migrants on the ground 
versus the transect survey data. The team is currently analyzing those data 
as a comparison between acoustic data and results of the ground-based 
surveys. The combination of ground-based survey and acoustics data, along 
with radar data, will add an additional and interesting analysis of bird 
migration from this discrete stopover habitat. 



ERDC/EL TR-12-22 


83 


References 


Alerstam, T. 2003. Bird Migration. Cambridge University Press, United Kingdom. 

Allan, J. R., and A. P. Orosz. 2001. The costs of birdstrikes to commercial aviation. Bird 
Strike Committee Proceedings. 2001 Bird Strike Committee-USA/Canada, third 
joint annual meeting, Calgary, AB. 

Askins, R. A. 1993. Population trends in grassland, shrubland, and forest birds in eastern 
North America. Current Ornithology 11:1-34. 

Askins, R. A., J. F. Lynch, and R. Greenberg. 1990. Population declines in migratory birds 
in eastern North America. Current Ornithology 7:1-57. 

Bonter, D. N., S. A. Gauthreaux, and T. M. Donovan. 2009. Characteristics of important 
stopover locations for migrating birds: remote sensing with radar in the Great 
Lakes Basin. Conservation Biology 23:440-448. 

Briggs, M. K. 1996. Riparian ecosystem restoration in arid lands: Strategies and 
references. Tucson, AZ: The University of Arizona Press. 

Bruderer, B, and A. Boldt. 2001. Flight characteristics of birds: 1. radar measurements of 
speeds. Ibis 143:178-204. 

Buler, J. J., and R. H. Diehl. 2009. Quantifying bird density during migratory stopover 
using weather surveillance radar. IEEE Transactions on Geoscience and Remote 
Sensing 47:2741-2751. 

Carlisle, J. D., S. K. Skagen, B. E. Kus, C. Van Riper, III, K. L. Paxton, and J. F. Kelly. 
2009. Landbird migration in the American West: recent progress and future 
directions. Condor 111:211-225. 

Cleary, E. C., R. A. Dolbeer, and S. E. Wright. 2007. Wildlife strikes to civil aircraft in the 
United States 1990-2006. U.S. Department of Transportation, Federal Aviation 
Administration. 

Cullitan, T. J., M. A. Warren, T. R. Goodspeed, D. G. Remer, C. M. Blackwell, and J. J. 

McDonough III. 1990. Fifty years of population change along the nation’s coasts 
1960-2010. Silver Spring, MD: National Ocean Service, National Oceanic and 
Atmospheric Administration. 

Dahl, T. E., C. E. Johnson. 1991. Status and trends of wetlands in the conterminous 

United States, mid-1970s to mid-1980s. Washington, DC: U.S. Department of the 
Interior, Fish and Wildlife Service. 

Diehl, R. H., and R. P. Larkin. 2004. Introduction to the WSR-88D (NEXRAD) for 

ornithological research. USDA Forest Service General Technical Report PSW- 
GTR-191. 



ERDC/EL TR-12-22 


84 


DiGaudio, R., D. Humple, and G. Geupel. 2008. Analysis of California migration 

patterns using NEXRAD and on-the-ground data. Report to the Department of 
Defense Legacy Resource Management Program, Project 06-329. 

Department of Defense Partners in Flight. 2002. DoD Partners in Flight Strategic Plan. 
The Conservation and Management of Migratory and Resident Landbirds and 
their Habitats on Department of Defense Lands, www.dodpif.org . 

Dolbeer, R.A. 2006. Height distribution of birds recorded by collisions with civil aircraft. 
The Journal of Wildlife Management 70:1345-1350. 

Dolbeer, R.A., and P. Escheenfelder. 2002. Population increases of large birds, 

airworthiness standards, and high-speed flight: A precarious combination. In 
Proceedings of the 55th International Air Safety Seminar, Dublin, Ireland. 

Donovan, T. M., C. J. Beardmore, D. N. Bonter, J. D. Brawn, R. J. Cooper, J. A. 

Fitzgerald, R. Ford, S. A. Gauthreaux, T. L. George, W. C. Hunter, T. E. Martin, J. 
Prince, K. V. Rosenberg, P. D. Vickery, and T. B. Wigley. 2002. Priority research 
needs for the conservation of neotropical migrant landbirds. Journal of Field 
Ornithology V 3 - 3 ^ 9 ~ 339 - 

Durst, S. L. 2004. Southwestern Willow Flycatcher potential prey base and diet in native 
and exotic habitats. M.S. thesis, Flagstaff, AZ: Northern Arizona University. 

Eglin Air Force Base. 2007. Integrated Natural Resources Management Plan. Prepared by 
Science Applications International Corporation (SAIC), Shalimar, FL. 

Fischer, R. A. 2000. Width of riparian zones for birds. TN-EMRRP-SI-09. EMRRP 

Technical Notes Collection. Vicksburg, MS: U.S. Army Engineer Research and 
Development Center. 

Fleishman, E., N. McDonal, R. MacNally, D. D. Murphy, J. Walters, and T. Floyd. 2003. 
Effects of floristics, physiognomy and non-native vegetation on riparian bird 
communities in a Mojave Desert watershed. Journal of Animal Ecology 72:484- 
490. 

Francis, C. M., and F. Cooke. 1986. Differential timing of spring migration in wood 
warblers ( Parulinae).Auk 103:548-556. 

Friedman, J. M., G. T. Auble, P. B. Shafroth, M. L. Scott, M. F. Merigliano, M. D. 

Freehling, and E. R. Griffin. 2005. Dominance of non-native riparian trees in 
western USA. Biological Invasions 7:747-751. 

Gaudard, M. 2008. Classification of breast cancer cells using JMP. JMP White Paper, 
Section 7: 20-29. Cary, NC: SAS Institute Inc. 

Gauthreaux, S. A., Jr. 1970. Weather radar quantification of bird migration. BioScience 
20:17-20. 

Gauthreaux, S. A., Jr. 1978. Migratory behavior and flight patterns. In Proceedings 
Impact of Transmission Lines on Migratory Birds, January 31-February 2, 

1978. Oak Ridge, TN: Oak Ridge Associated Universities. 




ERDC/EL TR-12-22 


85 


Gauthreaux, S.A., Jr., and C.G. Belser. 1998. Displays of bird movements on the WSR- 
88D: Patterns and quantification. Weather and Forecasting 13:453-464. 

Gauthreaux, S.A., Jr., and C.G. Belser. 1999. “Reply” to displays of bird movements on the 
WSR-88D: Patterns and quantification. Weather and Forecasting 14:1041-1042. 

Gauthreaux, S.A., Jr., and C.G. Belser. 2003a. Bird movements on Doppler weather 
surveillance radar. Birding 35:616-628. 

Gauthreaux, S. A., Jr., and C. G. Belser. 2003b. Radar ornithology and biological 
conservation. An A: 120:266-277. 

Gauthreaux, S. A., Jr., and C. G. Belser. 2005. Radar ornithology and the conservation of 
migratory birds. USDA Forest Service General Technical Report PSW-GTR-191. 

Gauthreaux, S. A., Jr., John W. Livingston and Carroll G. Belser. 2008. Detection and 
discrimination of fauna in the aerosphere using Doppler weather surveillance 
radar. Integrative and Comparative Biology 48 (1): 12-23. 

Greenberg, R. 1980. Demographic aspects of long-distance migration. In ed. A.Keast and 
E. S. Morton, editors. Migrant Birds in the Neotropics. 493-504. Washignton, 

DC: Smithsonian Institution Press. 

Hardy, P. C., D. J. Griffin, A. J. Kuenzi, and M. L. Morrison. 2004. Occurrence and 
habitat use of passage neotropical migrants in the sonorant desert. Western 
North American Naturalist 64:59-71. 

Hunter, W. C., R. D. Ohmart, and B. W. Anderson. 1988. Use of exotic Saltcedar 

(Tamarix chinensis) by birds in arid riparian systems. Condor 90:113-123. 

Hutto, R. L. 1985. Habitat selection by nonbreeding, migratory land birds. Pp. 455-476 in 
Habitat selection in birds (M. L. Cody, ed.). Academic Press, New York. 

Kelly, J. F., D. M. Finch, and W. Yong. 2000. Vegetative associations of wood warblers 
migrating along the middle Rio Grande Valley, New Mexico. The Southwestern 
Naturalist 45:159-168. 

Kelly, J. F., and R. L. Hutto. 2005. An east-west comparison of migration in North 
American wood warblers. Condor 107:197-211. 

Kelly, T.A., and T. Wilkens. 2006. Bird diction radar development: from Dare County to 
the Cape! United States Air Force Flying Safety Magazine. October 2006. 

Knopf, F. L., R. R. Johnson, T. Rich, F. B. Samson, and R. C. Szaro. 1988. Conservation of 
Riparian Ecosystems in the United States. Wilson Bulletin 100:272-284. 

Laccinole, J. A. 2009 . Tamarisk Invasion, http://www.westwindtechnology.com/cnwr/ 
problem.html 

Langer, G. 2009. Bird strikes: the feathers fly. ABC News Blogs, http://biogs.abcnews.com/ 
thenumbers/2009/04/bird-strikes-th.html . July 18, 2009. 







ERDC/EL TR-12-22 


86 


Livingston, M. F., and S. D. Schemnitz. 1996. Summer bird/vegetation associations in 
Tamarisk and native habitat along the Pecos River, southeastern New Mexico. 
Pages 171-180 in Desired future conditions for southwestern riparian ecosystems: 
Bringing interests and concerns together (D. W. Shaw and D. M. Finch, technical 
coordinators). RM-GTR-272. United States Department of Agriculture Forest 
Service, Rocky Mountain Research Station, General Technical Report. 

Lovell, C. D., and R. A. Dolbeer. 1999. Validation of the United States Air Force bird 
avoidance model. Wildlife Society Bulletin 27(i):i67-i7i. 

Lowery, G. H., Jr. R. J. Newman. 1955. Direct studies of nocturnal bird migration. In: 

Recent studies in avian migration, ed. A.Wolfson, pp. 238-263. Champaign, IL: 
University of Illinois Press. 

Martin, C. O., M. G. Harper, A. M. T. Shapiro, and R. A. Fischer. 2000. A Community- 
based regional plan for managing threatened and endangered species on 
military lands in the southeastern United States. SERDP Technical Notes 
Collection, ERDC/EL TN-SERDP-01-01. Vicksburg, MS: U.S. Army Engineer 
Research and Development Center. 

Moore, F. R., R. J. Smith, and R. Sandberg. 2005. Stopover ecology of intercontinental 
migrants: en route problems and consequences for reproductive performance. 
Pages 251-261 in Birds of two worlds: The ecology and evolution of migration, 
ed. R. Greenberg and P. P. Marra. Baltimore, MD: The Johns Hopkins University 
Press. 

Moore, F. R., S. A. Gauthreaux, P. Kerlinger, and T. R. Simons. 1995. Habitat 

requirements during migration: important link to conservation. Pages 121-144 In 
Ecology and management of neotropical migratory birds, ed. T. E. Martin and 
D. M. Finch. New York: Oxford University Press, Inc. 

Morrison, R. I. G., R. K. Ross, and M. Torres. 1992. Aerial surveys ofNearctic 
shorebirds in Mexico: Some preliminary results. Can. Wildl. Serv. Progr. Notes 
No. 201, Ottawa, Canada. 

Noss, R. F., E. T. LaRoe, III, and J. M. Scott. 1995. Endangered ecosystems of the United 
States: A preliminary assessment of loss and degradation. Washington, DC: 

U.S. Department of the Interior, National Biological Service. 

Paxton, K. L., C. Van Riper, III, and C. O’Brien. 2008. Movement patterns and stopover 
ecology of Wilson’s Warblers during spring migration on the lower Colorado 
River in southwestern Arizona. Condor 110:672-681. 

Peckford, M. L., and P. D. Taylor. 2008. Within night correlations between radar and 

ground counts of migrating songbirds. Journal of Field Ornithology 79:207-214. 

Poole, A. F. (editor). 2005. The birds of North America online: 

http://bna.birds.cornell.edu/BNA/ . Ithaca, NY: Cornell Laboratory of Ornithology. 

Rich, T. D. 2002. Using Breeding Land Birds in the Assessment of Western Riparian 
Systems. Wildlife Society Bulletin 30:1128-1139. 

Rodewald, P. G., and S. N. Matthews. 2005. Landbird use of riparian and upland forest 
stopover habitats in an urban landscape. Condor 107:259-268. 




ERDC/EL TR-12-22 


87 


Simons, T. R., F. R. Moore, and S. A. Gauthreaux. 2004. Mist netting trans-gulf migrants 
at coastal stopover sites: the influence of spatial and temporal variability on 
capture data. Studies in Avian Biology 29:135-143. 

Skagen, S. K., C. P. Melcher, W. H. Howe, and F. L. Knopf. 1998. Use of riparian corridors 
and oases by migrating birds in southeast Arizona. Conservation Biology 12:896- 

909. 

Skagen, S. K., J. F. Kelly, C. Van Riper, III, R. L. Hutto, D. M. Finch, D. J. Krueper, and C. 
P. Melcher. 2005. Geography of spring landbird migration through riparian 
habitats in southwestern North America. Condor 107:212-227. 

Sogge, M. K., D. L. Felley, and M. Wotawa. 2005. A quantitative model of avian 

community and habitat relationships along the Colorado River in the Grand 
Canyon. In The Colorado Plateau II: biophysical, socioeconomic, and cultural 
research, ed. C. van Riper III and D. Mattson. Tucson, AZ: University of Arizona 
Press. 

Sogge, M. K., S. J. Sferra, and E. H. Paxton. 2008. Tamarix as habitat for birds: 
Implications for riparian restoration in the southwestern United States. 
Restoration Ecology 16:146-154. 

Stevens, L. E., B. T. Brown, J. M. Simpson, and R. R. Johnson. 1977. The importance of 
riparian habitat to migrating birds. In Importance, preservation and 
management of riparian habitat: A symposium July 9, 1977, ed. R. R. Johnson 
and D. A. Jones, Jr. U.S. Department of Agriculture Forest Service, General 
Technical Report RM-43. 

Stromberg, J. C., S. J. Lite, R. Marler, C. Paradzick, P. B. Shafroth, D. Shorrock, J. M. 

White, and M. S. White. 2007. Altered stream-flow regimes and invasive plant 
species: the Tamarix case. Global Ecology and Biogeography 16:381-393. 

Szaro, R. C., and M. D. Jakle. 1985. Avian use of a desert riparian island and its adjacent 
scrub habitat. Condor 87:511-519. 

Tucker, J. W., G. E. Hill, and N. R. Holler. 2003. Longleaf pine restoration: implications 
for landscape-level effects on bird communities in the lower Gulf Coastal Plain. 
Southern Journal of Applied Forestry 27(2): 107-121. 

Lovell, C. D., and R. A. Dolbeer. 1999. Validation of the United States Bird Avoidance 
Model. Journal of Wildlife Management 27:167-171. 

Van Belle, J., J. Shamoun-Baranes, E. van Loon, W. Bouten. 2007. An operational model 
predicting autumn bird migration intensities for flight safety. Journal of Applied 
Ecology 44:864-874. 

Van Riper, III, C., K. L. Paxton, C. O’Brien, P. B. Shafroth, and L. J. McGrath. 2008. 

Rethinking avian response to Tamarix on the lower Colorado River: a threshold 
hypothesis. Restoration Ecology 16:155-167. 

Walker, H. A. 2008. Floristics and physiognomy determine migrant landbird response to 

Tamarisk ( Tamarix ramosissima ) invasion in riparian areas. The Auk 125:520-531. 



ERDC/EL TR-12-22 


88 


Yong, W. and D. M. Finch. 1997. Migration of the Willow Flycatcher along the middle Rio 
Grande. Wilson Bulletin 109:253-268. 

Zavaleta, E. 2000. The economic value of controlling an invasive shrub. Ambio 29:462- 

467. 

Zehnder, S., S. Akesson, F. Liechti, B. Bruderer. 2001. Nocturnal autumn bird migration 
at Falsterbo, South Sweden. Journal of Avian Biology 32:239-248. 

Zhalakyavichus, M. M. 1985. Radar investigations of correlations between bird migration 
and weather in Lithuania. Zoological Journal 64: 72-77. USSR Academy of 
Sciences [in Russian]. 



ERDC/EL TR-12-22 


89 


Appendix A: Movement Ecology and Migrant- 
Habitat Relations: Red-Eyed Vireos During 
Spring Stopover 

Introduction 

Movement ecology, or the study of why and how an organism moves from 
one spatial location to another, is a component of nearly all aspects of 
animal behavior. An animal’s decision to move is likely influenced by a 
complex combination of internal state and external stimuli (Nathan et al. 
2008). An understanding of movement in relation to changing internal state 
or in a heterogeneous environment is important for understanding 
ecological processes ranging from life history strategy (Farmer and Weins 
1999) to meta-population analyses (With and King 2001). Temporal and 
spatial decisions about movement are likely to have costs and benefits 
(Nathan et al. 2008) and costs may be intensified by current environmental 
changes such as habitat loss or fragmentation (Belisle et al. 2001, Turcotte 
and Desrochers 2003). Given its importance, there have been relatively few 
hypothesis driven studies of individual movement decisions within a 
landscape context (Lima and Zollner 1996, Graham 2001, but see Belisle et 
al. 2001, Turcotte and Desrochers 2003). 

Songbird migrants spend the majority of the migratory period refueling en 
route (Alerstam 2003) and stopover periods are uniquely well suited to 
studying movement decisions and their consequences. First, migrants are 
unlikely to be biased by previous information or experience at stopover sites 
because extrinsic factors such as weather influence the migratory route 
(Gauthreaux 1971, Moore and Kerlinger 1991, Nemeth and Moore 2007). 
Second, decisions are likely to be based on immediate costs and benefits 
because stopover is temporally restricted by the benefits of arriving early to 
the breeding sites (Moore et al. 2003, Smith and Moore 2005) and migrants 
landing after long nocturnal flights are likely to be energetically constrained 
and needing to refuel to continue their energetically demanding journey 
(i.e., Bairlein 1985, Moore and Kerlinger 1987, and Wang and Moore 1997). 
Pressures to arrive early to breeding locations should act to minimize time 
spent migrating and these pressures are expected to increase as the season 
progresses (Weber et al. 1998). However, migration is a time of exceptional 
energetic demands (Blem 1980) and migrants with fewer fuel reserves may 



ERDC/EL TR-12-22 


90 


require more time for refueling at stopover sites (Kuenzi et al. 1991, Wang 
and Moore 1993). Therefore, movement decisions during stopover reflect 
strategies for minimizing assessment time and maximizing refueling rate 
while avoiding predators and the consequences of decisions are likely to be 
expressed in a short time period. 

Migratory birds may exhibit two types of movement during stopover: 1) 
within habitat patch for direct resource acquisition and avoidance of stress 
and 2) between habitat patch for selection of quality. When migratory birds 
stopover to refuel, habitat quality may vary in terms of prey abundance (i.e., 
Delingat and Dierschke 2000, Parrish 2000, Farrington 2003, Smith et al. 
2004), predation pressure (Lindstrom et al. 1990, Cimprich and Moore 
1999), and competitor abundance (Hutto 1985a). While there is likely some 
innate preference for specific habitat types (see Berthold 1990), temporal 
variability in resources between habitat types has led to shifts in use by 
migrants (i.e., Bairlein 1983, Hutto 1885b, Moore et al. 1990), indicative of 
selection for habitat quality. To optimize refueling, migrants in unfamiliar 
landscapes are expected to select the highest quality habitat, containing the 
most abundant food resources, while expending the least searching effort 
(e.g., Lindstrom et al. 1990, Moore 1994, Cimprich et al. 2005). Migrants do 
preferentially select habitat during stopover because they are not randomly 
distributed across habitats types (Petit 2000). However, due to the 
difficulties in following passerines along the migratory route, when that 
selection occurs is not well understood. Songbird night vision probably is 
not good enough to make any more than a rough distinction of terrain at 
night (Martin 1990) and distributions of migrants captured in multiple 
habitat types differ from the morning to the afternoon (Bairlein 1983, 
Winker et al. 1995), suggesting that most selection likely occurs the day after 
landing during an initial exploratory phase (Aborn and Moore 1997). There 
have been few studies of individual movements in relation to habitat of 
birds newly arrived at stopover sites but Sedge warblers (Acrocephalus 
schoenobaenus ) that landed the previous night moved out of poor habitat 
during the first hours after dawn (Spina and Bezzi 1990) and Ovenbirds 
translocated into low and high quality habitat moved out of the low quality 
habitat but not out of the high quality habitat (Buler 2006). 

Migrants have to balance the need to move to access food resources while 
having little time to assess the predation risk from avian predators attracted 
to movement (Cimprich et al. 2005). Therefore, a migrant is expected to 
offset the risk of predation by foraging within as restricted an area as 



ERDC/EL TR-12-22 


91 


contains the necessary food resources (Cimprich et al. 2005). In habitat 
characterized by greater density of food resources, migrants do not need to 
move as far to secure those resources (Delingat and Dierschke 2000). A 
migrant landing in a high quality habitat can maximize refueling rate and 
reduce predation risk by exhibiting area-restricted movements and is 
unlikely to move out of that habitat type. Energetic condition upon arrival at 
a stopover site is also expected to influence migratory refueling strategy 
because energetically constrained individuals are under more pressure to 
replenish fuel stores (Wang and Moore 2005). Constrained migrants 
increase their predation risk in favor of refueling (Cimprich and Moore 
2006) by foraging both more rapidly and over a greater area to access prey 
resources (Loria and Moore 1990, Moore and Aborn 2000, Wang and 
Moore 2005, Buler 2006). An energetically constrained migrant may 
remain within a lower quality habitat, moving over a larger area to acquire 
fewer resources until it gains some threshold level of fat reserves, after 
which it may make the decision to search for a better habitat type. Wang 
and Moore (2005) found that foraging strategies during stopover varied 
with energy reserves for four species of thrushes. The greater energetic cost 
of searching for resources and greater risk of predation within a lower 
quality habitat may mean the constrained migrant is less likely to survive to 
continue migration. Habitat associations then result in costs and benefits to 
the individual that affect future movement decisions and, ultimately, 
survival to the next stage of migration. 

Further, after landing at a stopover site, migrants likely use multiple 
ecological cues to make movement and habitat selection decisions, yet 
these cues remain poorly understood (Moore and Aborn 2000). Direct 
sampling of food resources is the most accurate measure of habitat quality, 
but time and energetic constraints may require migrants to rely on 
structural or vegetative cues (Moore and Aborn 2000, Buler et al. 2007, 
McGrath et al 2009). In addition to personally acquired information, 
social information may be especially useful during migratory stopover 
when there is environmental uncertainty and little time for assessment of 
unfamiliar surroundings (Chernetsov 2006, Nemeth and Moore 2007). 
Acoustic cues or soundscape orientation may be an important component 
of the available social information (Slabbekoorn and Bouton 2008). 

The team conducted an experiment to test the following central 
assumptions about migrant movement ecology and habitat relations 
during stopover: 



ERDC/EL TR-12-22 


92 


• Hypothesis A. Migrants stopping over in a heterogeneous landscape 
move to select high quality habitat. 

Prediction. Migrants move l) across habitat boundaries to change 
habitat types and 2) the frequency of movement out of low quality 
habitat types is higher than out of high quality habitat types. 

• Hypothesis B. Movement within a habitat type is related to the quality 
of that habitat type. 

Prediction. A migrant in higher quality habitat, with greater densities 
of food resources, will exhibit more area-restricted movement than in a 
lower quality habitat. 

• Hypothesis C. Movement during stopover is related to the energetic 
condition of the bird. 

Prediction. During stopover energetically constrained individuals will 
move faster and further than will individuals with more fuel reserves. 

• Hypothesis D. The duration of stay at a stopover site is related to the 
time of the season and to the energetic condition of the bird. 

Prediction. Migrants stopping over 1) in better energetic condition or 
2) later in the season will spend less time at stopover sites than those 
arriving earlier or in worse condition. 

• Hypothesis E. Migratory songbirds use conspecific song as a cue to 
assess the quality of a habitat at stopover sites. 

Prediction. Migrants will remain longer in poor quality habitat with 
added conspecific song and will behave more similarly to migrants in a 
higher quality habitat type. 


Methods 

Focal Species 

The red-eyed vireo ( Vireo olivaceus) was chosen as focal species for this 
study for several reasons: First, it is one of the most common migrants 
captured at the netting location (see below) and the most common migrant 



ERDC/EL TR-12-22 


93 


detected during spring surveys at the release site (see below and this 
report). Although red-eyed vireos use a variety of substrates for foraging, 
they concentrate activity in the canopy and subcanopy (Cimprich et al. 
2000), so an analysis of their stopover movement ecology and habitat 
relations is applicable to other canopy gleaners. Furthermore, while red¬ 
eyed vireos are primarily foliage-gleaners of Lepidoptera larvae during 
breeding (Cimprich et al. 2000), they have been found to vary their 
foraging behaviors according to energetic condition during spring stopover 
(Loria and Moore 1990). This result suggests that red-eyed vireos change 
their stopover strategy with energetic condition. The result also provides 
evidence for making hypotheses about how they will move in relation to 
habitat quality. The term “migrant” refers to red-eyed vireos throughout 
the remainder of this chapter. 

Capture site 

The team captured migrants in chenier habitat in coastal southwestern 
Louisiana near Johnson’s Bayou (29 0 45’ N 93 0 30’ W, Figure Al). 

Cheniers are narrow strips of coastal woodlands dominated by hackberry 
(Celtis laevigata ) along the northern coast of the Gulf of Mexico. They are 
the first available habitat to rest and replenish fat stores following the 
trans-Gulf flight (Moore 1999). Extensive mist-netting for spring migrants 
has been conducted at the site since 1993, with the exception of one year 
(1997). Capture rates vary annually and with netting effort but up to 223 
red-eyed vireos have been captured in one season. Because the netting 
effort covers the chenier and most migrants arrive over the Gulf after 1000 
h (see Gauthreaux 1971,1972), it is relatively certain that a bird captured 
in the afternoon is a migrant that arrived that day. 

Release site 

The Vernon Unit of the Calcasieu Ranger District in Kisatchie National 
Forest (30° 57’ N 93 0 08’ W) is in the western portion of Louisiana and is 
adjacent to Fort Polk, the United States Army’s Joint Training and 
Readiness Center (Figure 1). This site was initially chosen because it was 
found to have a high density of spring migrants using radar ornithology 
(this report). In addition, forest cover types in the Kisatchie National 
Forest are characteristic of those found throughout the Gulf Coast region 
(Keddy 2009). They include longleaf pine, pine-hardwood, hardwood, pine 
regeneration, and harvested or open areas (Evans 1994). There is active 
management for upland longleaf pine (Pinus palustris) stands, and pine- 



ERDC/EL TR-12-22 


94 


hardwood stands are largely composed of a longleaf pine canopy with a 
hardwood understory. Hardwood dominated forests are primarily along 
creeks and slightly lower in elevation than the longleaf pine savannas 
(Figure Ai inset). 



Figure Al. Map of the state of Louisiana with translocation direction (arrow) from capture at Johnson’s 
Bayou to Kisatchie National Forest. Inset map of the study area within Kisatchie National Forest with 
release locations at Bundick (three • on left) and Drakes Creek (three • on right). 


Experimental releases 

To simulate arrival at a stopover site, the team translocated red-eye vireos 
with varying amounts of fuel reserves to an unfamiliar heterogeneous 
landscape and released them before dawn into habitat types varying in 
quality. Migrants captured in mist nets at Johnson’s Bayou were trans¬ 
ported the afternoon or evening of the day of capture approximately 17 0 and 
143 km north to Kisatchie National Forest (Figure Ai). Birds were held in 
individual cages for up to 12 hours and provided with food (meal worms and 
monkey biscuits) and water ad libitum. The migrants were fitted with radio¬ 
transmitters weighing less than 3% of their body mass (models LB-2 and 

















ERDC/EL TR-12-22 


95 


LB-2N, Holohil Systems). Feathers were removed from the synsacrum and 
the transmitter was glued to the exposed area using nontoxic glue. Each 
migrant’s mass, fat score (Helms and Drury i960), wing length, and tarsus 
length was recorded and a Fish and Wildlife Service band and a unique 
combination of colored leg bands were placed on the birds. 

Migrants were released in the two landscapes at six predetermined locations 
in three habitat types representative of the region: upland pine savanna 
(pine), deciduous forests along creeks (hardwood), and an intermediate 
between the two (mixed) (Figure Ai inset). The hardwood release sites were 
placed adjacent to Drakes Creek and Bundick Creek along transects where 
concurrent migrant surveys for radar ground-truthing were being con¬ 
ducted. One pine and one mixed release site were associated with each creek 
(Drakes and Bundick). The pine and mixed sites were placed in the closest 
accessible locations that were surrounded by predominately pine or mixed 
habitat, respectively (Figure Ai). The mass and fat score at the time of 
release was recorded; then the birds were released simultaneously before 
first light. In the case of rain, birds were released as soon as possible later in 
the morning. To determine energetic condition, size-specific fat-free masses 
for red-eyed vireos were calculated. The team ran a regression of wing chord 
length on mass for all fat score zero (Helms and Drury i960) red-eyed 
vireos captured from 1998 to 2006 at Johnson’s Bayou (n= 1775). The 
energetic condition of each individual was determined by subtracting the 
fat-free mass specific to their wing chord length from their release mass 
(Owen and Moore 2006). Migrants in positive energetic condition had fuel 
reserves while migrants in negative energetic condition were below lean 
body mass and therefore significantly energetically constrained. 

During the spring of 2009, the team released migrants at only two of the 
predetermined locations in upland pine savanna (pine) and deciduous 
forests (hardwood) associated with Bundick Creek. The team conducted 
only paired daily releases of one migrant in pine and one in hardwood; each 
simultaneously released at 6:30 am. From 24 April through 6 May 2009, 
the team conducted six paired releases of migrants in pine and hardwood 
with playback (see “Audio recording and playback” below) added to pine. 

Movement observations 

Individuals were continuously radio-tracked from dawn to dusk for the first 
three days after release with locations taken every 15 minutes. To minimize 
the impact of the observer, the birds were located to within 30 m and then 



ERDC/EL TR-12-22 


96 


the locations were circled to verify the accuracy before approaching and 
attempting to make a visual observation. Whenever possible, the team 
recorded the substrate, height, the success or failure of foraging maneuvers, 
prey items and other behaviors such as preening or resting (Remsen and 
Robinson 1990). In 2008, when the team visually observed a bird more 
detailed recordings were made of all behaviors according to definitions (i.e., 
foraging, vocalizing, perched, preening, or flying) with the number of 
seconds that each behavior was continuously observed. 

In 2007, migrants were tracked continuously for the first five hours and 
the last hour of each day of stopover. In 2008, migrants were tracked 
continuously throughout the entire day for the first three days. In 2008, 
migrants were located once or twice daily after the first three days to 
determine the duration of stay in the landscape. Searching for a bird 
tracked the previous day began at first light at the last known location. 
Once an individual was no longer detected, a set of locations surrounding 
the entire study site were visited daily to check signals and verify that 
migrants no longer detected had continued migration. In addition, once 
per season personnel from Fort Polk verified from a helicopter that signals 
not detected on the ground were no longer in the landscape. 

Audio recording and playback 

On 20 April 2009, the team used a Sennheiser© ME-65 omindirectional 
microphone, a Telinga© Pro parabola, and an Olympus© WS-110 WMA 
digital voice recorder to record singing red-eyed vireos near the hardwood 
release site. The team was reasonably certain that different individuals were 
recorded based on locations and counter-singing. The recordings were 
minimally edited to reduce background noise and then chose the one 
minute with the least background noise and most song clarity for each of 
three individuals. Three iPods (Apple©) with folding speakers (©Radio- 
Shack) were placed in the same distances and directions surrounding the 
pine release site as the detected red-eyed vireos surrounding the hardwood 
release site (we refer to this artificially added song as “playback”). The 
playback setup was situated as high in the vegetation as logistically possible 
to replicate the canopy singing height of red-eyed vireos singing in the 
hardwood. Each playback setup looped a one minute recording of a single 
individual and played continuously as long as the migrant tracked remained 
within 300 m of the release site. 



ERDC/EL TR-12-22 


97 


Designation of habitat quality 

While categorization of habitat quality in a changing environment can be 
difficult, the combination of sampling for prey, predators, and competitors 
should provide a strong line of evidence for habitat quality within the 
spatiotemporal frame of the study. To measure prey abundance, the team 
used arthropod sampling and to measure avian predators and density of 
songbirds (potential competitors), banding and transect surveys were used. 

Migrant distribution. During the spring of 2006, the team sampled the 
relative use of adjacent pine, mixed and hardwood habitats by migratory 
species using mist-nets to capture and band birds. Twenty standard 12 by 
2.5 m, 30 mm mesh mist-nets were used and placed in each habitat type 
near the ground, mid-canopy, and upper-canopy. Nets were kept open 
throughout the day except in the case of rain, flooding or high winds. 

Daily surveys were conducted along transects associated with each of the 
release sites in each habitat type in 2008. Transects were 500 m long 
passing through each release site and flagged every 25 m. The same 
surveyor conducted the surveys the entire season. Each day from 10 min 
prior to sunrise until 10 am, he walked at a constant pace (1 km/ hr) along 
all three of the transects associated with one of the two creeks (pine, mixed, 
and hardwood). The daily order of the habitat types surveyed was 
systematically rotated and the creek sampled (Bundick or Drakes) was 
alternated each day. Surveys were not conducted in rain or high winds. The 
surveyor recorded the first detection location, species, detection method, 
and age or sex (whenever possible) for every bird observed within 50 m of 
the survey transect. 

Arthropod abundance. To quantify food resources, the team used canopy 
branch clipping (Cooper and Whitmore 1990), a method that has been 
shown to be effective in measuring arthropod prey density on and near 
vegetation used by foliage-gleaning birds (Johnson 2000). Each year, 
twenty-four samples were collected at the same location points — every 
100 m along transects associated with the release sites — in each habitat 
type during a week in the early, middle and late spring. The team con¬ 
sidered the samples collected systematically in the habitat types to be 
random in terms of bird use and samples were also collected to compare to 
food resources in locations where migrants were moving. Some arthropod 
samples were also taken in locations where migrants were located in 2007 
and in 2008. For the majority of birds tracked, the team returned within ten 



ERDC/EL TR-12-22 


98 


days to take an arthropod sample at the first location point of every second 
hour along a migrants’ movement track for the entire first day of movement. 
A hoop net was used on an extending survey pole to encompass and then 
clip a branch 4 to 6 m above the ground (Johnson 2000). The team alter¬ 
nated the times of day that each habitat type was sampled, but did not 
sample in the very early morning because insects were not as active when it 
was cooler. Branches approximately 0.25 m long were chosen from the pine 
(Pinus spp.), oak ( Quercus spp.), or sweetgum ( Liquidambar styraciflua ) 
canopy tree (> 7 m tall), with a branch 4 to 6 m above the ground closest to 
the sampling point. If there were no trees that fit these criteria within 40 m 
of the point, then the team sampled the highest branches of the closest of 
the specified species that was taller than 3 m. When there was more than 
one designated species present within the sampling area, the team alter¬ 
nated between them at successive sampling locations. Clipped branches 
were weighed and collected arthropods were identified to order and 
measured to length. 

Statistical Analyses 

Designation of habitat quality. Testing was conducted to check for 
differences in the number of arthropods and the number of Lepidoptera 
larvae per sample among habitat types and time of season (categorized as 
early, middle and late in the spring). Testing was also conducted to check 
for differences in arthropods and Lepidoptera larvae between years and 
creeks. To determine if migrants were selecting areas with higher food 
resources in each habitat type, the team tested for differences in the number 
of arthropods and Lepidoptera larvae per sample between locations selected 
by migrants and locations along transects. The transect survey data was 
used to test for differences in the number of red-eyed vireos and avian 
predators in each habitat type or between drainages. For pair-wise 
comparisons the non-parametric Mann-Whitney U-test was used because 
the normality assumptions for parametric tests were rarely met. One-tailed 
tests were used when a priori predictions were directional. Pair-wise 
comparisons were conducted in SPSS 15.0 (2006). 

Movement patterns. Because it was expected that multiple factors to 
influence migrant behavior, a multi-model inference and an information - 
theoretic approach were used to analyze the relative explanatory power of 
seven factors on movement patterns throughout the day — in two hour 
increments — for each individual during the first three days of stopover. 

Two parameters were used to quantify movement patterns: Linear 



ERDC/EL TR-12-22 


99 


displacement (the linear distance between the first and last location of the 
time period; m) and rate (the cumulative distance between all locations 
divided by the time in a time period; m hr-i). Due to the fact that the team 
quantified movement throughout the day for the same individuals, linear 
mixed-effects models were fitted with restricted maximum likelihood 
parameter estimation (REML function in library nlme for R, Pinheiro et al. 
2010) with the individual bird as the random component. This allowed the 
team to control for correlations between observations from the same 
individual. REML estimation was used because it is less sensitive to small 
sample sizes relative to the number of fixed-effects than traditional 
maximum likelihood estimation (Zuur et al. 2009). First, models were 
compared with the same full set of fixed-effects and different random 
components: 1) no random term, 2) random intercept only and 3) random 
intercept and slope (Zuur et al. 2009). Akaike’s Information Criterion and 
Bayesian Information Criterion were used to compare models and found the 
models, including a random intercept — but not a random slope — to be the 
most parsimonious (Zuur et al. 2009). 

To measure the effects of factors on movement throughout the day, 43 
biologically plausible models were created and compared. The models 
were composed of six fixed effects: hour of day (in two hour increments: 
6:30 to 8:30, 8:31 to 10:30,10:31 to 12:30,12:31 to 14:30,14:31 to 16:30 
and 16:31 to 18:30 CST; hour 2, 4, 6, 8,10, and 12, respectively); arrival 
energetic condition (condition); habitat type of the release site (release 
habitat); creek of the release site (creek); day of season (season); and day 
of stopover (day) as well as one interaction term, arrival energetic 
condition by release habitat type. There were no differences between years 
in exploratory analyses, so the team did not include year in candidate 
models. A subset of the same set of models without hour (28 candidate 
models) was used to assess the differential influence of five fixed effects 
(condition, release habitat, creek, season, and day) and one interaction 
term (condition by release habitat) on movement during each time period 
throughout the day. Data transformation (log [x +1]) was used to achieve 
homogeneity of variance for movement rate and linear displacement. 

Akaike’s Information Criterion for small sample sizes (AICc) was used to 
rank, compare, and evaluate all candidate model sets (Burnham and 
Anderson 2002). All models were presented with a AAICc < 2 as plausible 
competing models (considered the subset of best supported models, 
Burnham and Anderson 2002). The team also presented the null (intercept 



ERDC/EL TR-12-22 


100 


only) model for assessment of the relative explanatory power of the 
plausible models. For variables in more than one top model (AAICc < 2), 
parameter estimates were averaged across models containing each 
explanatory variable and standard errors were calculated from conditional 
variances due to model selection uncertainty (Burnham and Anderson 
2002). The relative influence of each variable (j) was also estimated by 
calculating w+ (j), the sum of wi (Akaike weights) across all models in the 
dataset in which variable j occurred (Burnham and Anderson 2002). Model 
building was conducted using library nlme (Pinheiro et al. 2010) for R (R 
version 2.1.11, R Core team 2010) and comparison of competing models was 
conducted using library bbmle for R (Bolker 2010). Relative importance of 
variables and model-averaged parameter estimates were calculated in Excel 
according to Burnham and Anderson 2002. 

Stopover duration. To analyze the relative influence of condition, release 
habitat, creek, and season on the duration of stay at the study site, the 
team built and compared 15 biologically plausible generalized linear 
models. Akaike’s Information Criterion for small sample sizes (AICc) was 
used to rank, compare, and evaluate all candidate model sets (Burnham 
and Anderson 2002) as described above. 

Conspecific song. To determine if migrants used conspecific song as a cue to 
habitat quality, the team quantified the amount of time migrants spent 
within their release habitat for all birds tracked from Bundick pine (with 
and without playback added) and hardwood during 2007 to 2009. The team 
constructed and compared the same set of four generalized linear models 
including all combinations of two main effects: energetic condition (condi¬ 
tion) and group (pine, pine with added playback, and hardwood) and one 
interaction (condition* group) on the amount of time spent in the release 
habitat during the first five hours after release. The team tested the expecta¬ 
tion that added song would be a top supported variable for comparisons of 
birds released at the same site in pine (with and without song) but not for 
birds released in pine with song as compared to birds released in hardwood. 
Therefore, models containing the variable group (pine versus hardwood) 
would be more supported for comparisons of birds released in pine and 
hardwood without the addition of playback in the pine (during 2007 and 
2008) than for comparisons of birds released in the same two habitat types 
but with the addition of playback in pine (during 2009). Akaike’s Informa¬ 
tion Criterion for small sample sizes (AICc) was used to rank, compare, and 
evaluate all candidate model sets (Burnham and Anderson 2002) as 
described above. 



ERDC/EL TR-12-22 


101 


Results 

Designation of habitat quality 

Migrant distribution. Nets were run for 35 days from 27 March to 5 May 
2006 (closed five days due to weather) and 116 individuals of 28 migratory 
species were caught (Table Ai). The majority of the migratory species 
captured breed in the landscape so most of the individuals captured could 
have been either breeding or migratory. Some individuals were recaptured 
on successive days (Table Ai). Migrants were found using more than one 
habitat type, with the majority of the captures in the mixed habitat (86%), 
few in the hardwood (5%), and a moderate number in the pine (13%). 
Capture numbers varied during the season but were highest the last week 
of April and first week of May (Figure A2). 


Table AI. Summary of banding effort and captures of migratory species in pine, mixed 
and hardwood habitat in Kisatchie National Forest from 27 March through 5 May 2006. 



Pine 

Mixed 

Hardwood 

Total 


Number of nets 

6 

6 

8 

20 

Net hours 

1254.8 

1286.7 

1418.3 

3959.7 

Number captured 

13 

98 

5 

116 

Number recaptures 

4 

35 

0 

39 

Number caught per net hour 

0.010 

0.076 

0.004 

0.030 

Number caught per net hour 
including recaptures 

0.014 

0.103 


0.040 




c£> 
















A<s> 


a<£ 




A<s> 


a<£ 


^ ////// # 


■xo 


Figure A2. Number of mist-net captures of migrant species (corrected for daily netting 
effort) by date at Fort Polk, LA from 21 March to 5 May 2006 (excluding five days). 


























ERDC/EL TR-12-22 


102 


The team recorded 2337 birds on 99 transect surveys during 33 sampling 
days from 7 April to 18 May 2008. There were no red-eyed vireos detected 
in pine; singing red-eyed vireos were detected three times in mixed habitat 
(at Bundick transect on 15 April, 22 April and 16 May). Two to seven red¬ 
eyed vireos were detected daily (x = 4.36 ± 1.03) on hardwood transects, 
with more at Bundick than at Drakes Creek (Bundick x = 4.94 ± 0.84, n= 
17, Drakes x = 3.75 ± 0.86, n= 16, p < 0.001). Throughout the season only, 
3 broad-winged hawks, 17 red-shouldered hawks, 1 sharp-shinned hawk, 
and 4 hawks of unknown species were detected. The sharp-shinned was 
detected in the pine habitat, the broad-winged hawks were detected as fly¬ 
overs in all three habitat types, and the red-shouldered were detected 
primarily in pine habitat (n=i2) but also in mixed (n=2) and pine (n=3). 
There were more hawks detected near Bundick (n=i7) than Drakes Creek 
(n=8) but they were detected in all three habitats at both sets of transects. 

Arthropod abundance. During the early, middle and late spring transect 
sampling periods, the team collected a total of 180 arthropod samples in 
2007 and 216 in 2008. At locations selected by migrants, 36 samples were 
collected in 2007 and 368 in 2008. There was no difference between the 
creeks in number of arthropods (n=i98 Drakes, n=i98 Bundick, p=o.n) 
or the number of Lepidoptera larvae ^=198,198, p=o.is). There was also 
no annual difference in number of arthropods (n=i8o, 216, p=o.59) or the 
number of Lepidoptera larvae (n=i8o, 216, p=o.i8). The team combined 
years and creeks for comparisons of food resources during times of spring 
and habitat types. 

There were no seasonal differences in the number of arthropods (n=io8 
early, n= 144 middle, n= 144 late, p=o.47) but there were significantly more 
Lepidoptera larva early versus late in the spring (early n=io8, x=i.42 ± 
7.46; late n=i44, x=o.i7 ± 0.61, p=o.02), with no difference between those 
timeframes and the middle of the spring. Significant differences were found 
in the number of arthropods for all comparisons of pine, mixed and hard¬ 
wood (p <0.000, Figure A3) and differences between pine and hardwood 
and pine and mixed habitat in the number of Lepidoptera larvae (both 
comparisons p <0.000, Figure A3), but no difference between hardwood 
and mixed (p=o.i36, Figure A3). In pine and mixed habitats there were 
more arthropods in areas selected by migrants (pine p=o.oi and mixed 
p=0.02) but the converse was true for hardwood habitat (p=0.05, 

Figure A4). The number of Lepidoptera larvae was greater in areas where 



ERDC/EL TR-12-22 


103 


migrants selected in pine (p=0.03) but were lower in mixed (p=0.004) and 
not different in hardwood (p=o.io, Figure A5). 

Experimental releases and movement observations 

Fifty red-eyed vireos were successfully translocated and tracked during 
April and May of 2007 (11=17) and 2008 (n=33). In 2007, the team recorded 
1,093 individual locations in over 47 days of tracking and in 2008,3,305 
individual locations were recorded on over 66 days of tracking. A total of 
24 migrants were released and tracked at the Drakes Creek release sites and 
26 migrants at the Bundick Creek release sites for a total of 17 released in 
hardwood, 16 in mixed and 17 in pine. Migrants moved 21 to 2347 m linear 
distances from release locations during the first day of stopover (x = 618 ± 
75 m). The furthest movements occurred during the first two hours of the 
first day and distances gradually declined throughout the day and with 
successive days (Figure A6). However, hours four and six (8:31 to 12:30) 
and ten and twelve (14:31 to 18:30) were similar in displacement distances 
(Figure A6). Movement rate was also fastest during the first two hours of 
the first day of stopover (Figure A7). There was not a clear pattern to rate on 
the second day of stopover but on the first and third day rate gradually 
declined throughout the day until the last hour when there was an increase 
(Figure A7). 



Figure A3. Number of Arthropods and Lepidoptera larvae detected along transects 
during spring of 2007 and 2008. The mean values are shown and the error bars 
represent standard deviation. The number of arthropods differed for all 
comparisons of pine, mixed and hardwood (p <0.001). The number of Lepidoptera 
larvae were different for pine and hardwood and pine and mixed habitat in (both 
comparisons p <0.001) but not for hardwood and mixed (p=0.136). 


























ERDC/EL TR-12-22 


104 


9 

8 

7 

6 

5 

4 

3 

2 

1 

0 


2.75 


1.44 

JL 


4.82 


4.97 

JL 


7.10 

L 


6.02 

L 


Random Selected 


Pine 


Random Selected 


Mixed 


Random 


Selected 


Hardwood 


Figure A4. Mean number (bars represent standard error) of arthropods detected along 
transects (Random) and areas where migrants were located (Selected) in pine, mixed 
and hardwood habitat (pine p=0.01 and mixed p=0.02 and hardwood p=0.05). 


1.4 

1.2 

1.0 

0.8 

0.6 

0.4 

0.2 

0.0 


0.60 


0.86 


0.40 


0.11 

T 


I 


0.31 

X 


Random Selected Random Selected 
Pine Mixed 


0.50 

L. 


Random Selected 
Hardwood 


Figure A5. Mean number (bars represent standard error) of Lepidoptera larvae detected 
along transects (Random) versus areas where migrants were located (Selected) in each 
habitat type (pine p=0.03, mixed p=0.004 and hardwood p=0.10). 



























































ERDC/EL TR-12-22 


105 



Figure A6. Mean linear displacement (m) by hour of the day and stopover day for the first three days of stopover. 

Mean values labeled and bars represent standard error. 


The team visually observed 38 radio-tagged individuals on 306 occasions. 
On five occasions interactions were observed between the migrant that 
was being tracked and a verified unbanded conspecific individual. Two of 
these migrants were observed having aggressive interactions with another 
red-eyed vireo and one had a total of three different interactions in 
different times and locations, all in hardwood habitat. In all cases the 
migrant tracked subsequently moved away from the area of the 
interaction. A third tagged migrant was observed perched within three 
inches of an unhanded red-eyed vireo which was fluttering its wings while 
making twittering calls and later chased the migrant. 

The team did not observe any vocalizations from migrants. Foraging was the 
most commonly observed behavior (49% of observations). Other commonly 
observed behaviors were preening (9%), flying (11%), and perching (25%). 
Migrants were observed foraging in all three habitat types and all except 
three of the 39 prey items identified were Lepidoptera larvae. Migrants were 
observed foraging once in Sassafras albidum, Pinus palustris and Ostrya 



















































ERDC/EL TR-12-22 


106 



Figure A7. Mean movement rate (m min 1 ) by hour of the day and stopover day for the first three days of 
stopover. Mean values labeled and bars represent standard error. 


virginiana infrequently in Cornus florida (2% of observations), Acer 
rubrum (4%), Ilex spp. (5%), Fagus grandifolia (6%), and Magnolia spp. 
(10%) and frequently in Quercus spp. (51%) and Liquidambar styraciflua 
(20%). There was at least one observation of a successful prey attack in all of 
the tree species observed frequently or infrequently except Ilex spp. The 
team observed migrants from 1 to 35 m above the ground with a mean of 
14.3 ± 8.6 m (n = 192). The number of successful attacks increased from 
pine to mixed to hardwood habitat (Figure A8). 

Hypothesis A 

Twenty nine of the 47 birds that we were able to follow continuously 
throughout the first day changed habitat types after release. The time from 
release until changing habitat types ranged from 33 min to seven hours, 
with a mean of 2.65 ± 0.269 hours. The majority of the birds released in the 
pine and mixed left those habitat types (pine 88% left out of n=i6 released; 
mixed 71%, n=i4 released) while the majority of the birds released in hard¬ 
wood remained in hardwood (29% left, n=i7 released, Table A2). Most 




























































ERDC/EL TR-12-22 


107 


migrants that left pine or mixed habitat types moved into higher quality 
habitat types (pine to mixed, pine to hardwood or mixed to hardwood, 75%, 
11=24, Table A2). Many of the migrants that initially selected one habitat 
type later moved again to select another habitat type. 


o 

CD 

(/) 

~c/5 

o 

ro 

-4—* 

CO 


c/5 

c/5 

CD 

O 

O 

C/D 



Figure A8. The number of successful attacks per time spent foraging in pine, mixed and 

hardwood habitat. 


Table A2. Number of migrants released in pine, mixed or hardwood that left those habitat 
types during the first day of stopover and moved pine, mixed, hardwood habitat or 

another habitat type. 


Selected Habitat 


Release habitat 

Pine 

Mixed 

Hardwood 

Other 

Total 

Pine 

— 

8 

4 

2 

14 

Mixed 

1 

— 

6 

3 

10 

Hardwood 

0 

3 

— 

2 

5 

Total 

1 

11 

10 

7 

29 


Hypotheses B and C 

When all hours and stopover days were combined, movement rate and 
linear displacement were influenced by migrant arrival condition and 
landscape (creek) but not by the habitat type of the release patch or the day 
of the season (Table A3). Rate and displacement also decreased with the 
hour of the day and the day of stopover (Table A4). Habitat type of the 
release site was only influential in the linear displacement during the first 
two hours after release. During the first two hours of the first day, migrants 
released in pine moved further and faster (pine x = 462 ± 80 m) than those 


















ERDC/EL TR-12-22 


108 


Table A3. Comparison of the relative influence of generalized linear models in predicting 
the movement rate and linear displacement of red-eyed vireos. The number of parameters 
(K), differences in AlCc values (MICc) and Akaike weights (wi) are shown for all top models 
(AAlCc < 4) as well as the null model. Models with AAlCc < 2 considered equally plausible. 
Results shown for all hours combined and for each two hour period of the day. Two 
outliers were removed from hour 2 (6:30 to 8:30). 


Time Period 

Model description 

AlCc 

K 

Wi 

Individuals Observation 

All Hours 

Linear Displacement 




50 

382 


Hour of day, Arrival condition, Creek, Day of stopover 

0.0 

7 

0.924 




Null 

33.9 

3 

<0.001 




Movement Rate 







Hour of day, Arrival condition, Creek, Day of stopover 

0.0 

7 

0.990 




Null 

43.3 

3 

<0.001 



6:30- 8:3(5 

Linear Displacement 




48 

74 


Habitat, Arrival condition 

0.0 

5 

0.475 




Arrival condition, Creek 

2.3 

5 

0.151 




Habitat, Arrival condition, Creek, Condition*Habitat 

2.8 

7 

0.117 




Arrival condition, Creek, Day of stopover 

2.9 

6 

0.109 




Null 

8.2 

3 

0.008 




Movement Rate 







Habitat 

0.0 

4 

0.266 




Null 

0.0 

3 

0.261 




Arrival condition, Creek 

1.4 

5 

0.132 




Creek 

1.9 

4 

0.105 




Habitat, Arrival condition 

2.5 

5 

0.078 




Arrival condition 

3.4 

4 

0.048 




Day of stopover 

3.6 

4 

0.043 



8:31 - 10:30 

Linear Displacement 




49 

94 


Arrival condition, Creek 

0.0 

5 

0.460 




Arrival condition 

0.4 

4 

0.368 




Null 

10.4 

3 

0.003 




Movement Rate 







Arrival condition, Creek 

0.0 

5 

0.952 




Null 

13.0 

3 

0.001 



10:31 - 12:30 

Linear Displacement 




39 

71 


Arrival condition, Creek 

0.0 

5 

0.656 




Arrival condition, Creek, Day of season 

3.9 

6 

0.092 




Null 

7.1 

3 

0.019 




Movement Rate 







Arrival condition, Creek 

0.0 

5 

0.680 




Arrival condition, Creek, Day of season 

3.7 

6 

0.105 




Null 

6.3 

3 

0.029 



12:31 - 14:30 Linear Displacement 




33 

59 


Day of season 

0.0 

4 

0.226 




Null 

1.9 

3 

0.158 




Arrival condition, Day of season 

2.7 

5 

0.101 




Arrival condition 

3.9 

4 

0.057 




Movement Rate 







Arrival condition, Creek 

0.0 

5 

0.406 




Day of Season 

1.0 

4 

0.236 




Arrival condition, Day of season 

2.7 

5 

0.107 




Null 

2.8 

3 

0.093 



14:31 - 16:30 Linear Displacement 




32 

56 


Null 

0.0 

3 

0.573 




Arrival condition 

3.0 

4 

0.131 




Creek 

3.5 

4 

0.099 




Movement Rate 







Null 

0.0 

3 

0.819 



16:31 - 18:30 Linear Displacement 




14 

26 


Null 

0.0 

3 

0.713 




Movement Rate 







Null 

0.0 

3 

0.288 




Arrival condition 

0.7 

4 

0.206 




Arrival condition, Day of stopover 

0.7 

5 

0.204 




Day of stopover 

1.3 

4 

0.148 

















ERDC/EL TR-12-22 


109 


Table A4. Relative importance and model-weighted averaged parameter estimates (when 
parameter was included in more than one supported model) of parameters included in top 
explanatory models (AAlCc < 2) for movement rate and linear displacement of red-eyed 
vireos. The conditional 95% confidence interval is calculated for parameters included in more 

than one top model. 



Time Period 

Model description 

Parameter estimate a 

Standard error 

Degrees of freedom 

All Hours 

Linear Displacement 
Hour of Day 

-0.1205 

0.0226 

330 


Arrival Condition 

-0.3189 

0.0784 

47 


Creek b 

1.5359 

0.3089 

47 


Day of Stopover 
Movement Rate 

-0.2437 

0.0989 

330 


Arrival Condition 

-0.1351 

0.0272 

47 


Hour of Day 

-0.0495 

0.0074 

330 


Creek 

0.5845 

0.1003 

47 


Day of Stopover 

-0.0861 

0.0320 

330 

6:30 to 8:30 

Linear Displacement 
Arrival Condition 

-0.2306 

0.0662 

45 


Habitat c 

-0.4249 

0.1604 

45 


Movement Rate 

Null a top model 



8:31 to 10:30 

Linear Displacement 11 





Arrival Condition 

-0.3595 

0.0994 

46 


Creek 

Movement Rate 

1.1419 

0.4058 

46 


Arrival Condition 

-0.1606 

0.0322 

46 


Creek 

0.5894 

0.1205 

46 

10:31 to 12:30 

Linear Displacement 
Arrival Condition 

-0.3576 

0.1145 

36 


Creek 

Movement Rate 

2.4922 

0.4755 

36 


Arrival Condition 

-0.1312 

0.0340 

36 


Creek 

0.6436 

0.1274 

36 

12:31 to 14:30 

Linear Displacement 
Movement Rate 

Null a top model 




Arrival Condition 

-0.1503 

0.0391 

30 


Creek 

0.6562 

0.1499 

30 


Day of season 

0.0297 

0.0070 

25 

14:31 to 16:30 

Linear Displacement 

Null a top model 




Movement Rate 

Null a top model 



16:31 to 18:30 

Linear Displacement 

Null a top model 




Movement Rate 

Null a top model 




a Values back-transformed from LoglO + 1 
b Values for creek are Drake = 1 and Bundick = 2 
0 Values for habitat are pine = 1, mixed = 2 and hardwood = 3 
d Arrival condition parameter estimate and error averaged from top two models 


released in mixed (x = 238 ± 75 m) or hardwood habitat (x =185 ± 52 m). 
The team released birds with energetic conditions ranging from well below 
lean body mass (-2.28) to far above (6.3) and the measure of energetic 
condition was well correlated with visual estimation of fat scores for the 
migrants released (R 2 = 0.562, p < 0.001, Figure A9). Condition was a 







ERDC/EL TR-12-22 


110 


highly influential variable throughout the day for movement rate and linear 
displacement (Table A3). Throughout the day, as energetic condition 
increased, movement rate and linear displacement decreased (Table A4). 
The null (intercept only) model was the top model for most periods in the 
afternoon, indicating that none of the variables included in candidate 
models influenced linear displacement after 12:30 or movement rate after 
14:30, migrants also moved less during the afternoon hours (Table A3). The 
only other period when variables did not explain movement pattern was for 
rate during the first two hours of the day. 



Figure A9. The correlation between fat score (Helms and Drury 1960) and the condition 
index (R 2 = 0.56, P < 0.001). A condition index of zero corresponds to zero fat stores or 

lean body mass. 


The release landscape (creek) also consistently explained variability in 
movement after the first two hours of the day (Table A3). Migrants moved 
both faster and further at Bundick Creek than they did at Drakes Creek 
(Bundick rate x = 2.444 ± 0.153 m min 1 and displacement x = 157.046 ± 
13.660 m for two hour periods; Drakes rate x = 1.494 ± 0.132 m min 1 , 
displacement x = 115.880 ± 12.535 m). When all hours were combined, 
the day of stopover was an important explanatory model, with movement 
decreasing as migrants stayed successive days (Tables A3 and A4). Time of 
the season was not highly influential, except from 12:31 to 14:30 when the 
movement rate increased slightly as the season progressed (Table A4). 








ERDC/EL TR-12-22 


111 


Hypothesis D 

The team was able to determine the stopover duration of 43 migrants (15 in 
2007, 28 in 2008). One migrant stopped over for thirteen days and this bird 
was excluded from stopover duration analyses because the duration may 
have been extended due to an unseasonably cold week. For the remaining 
cases, weather or the end of the season prevented the team from deter¬ 
mining whether migrants remained in the landscape. Migrants stopped over 
from one to eight days (2.857 ±0.3 days). One third of migrants left the 
night of the release day (33 percent), 21 percent stayed an additional day 
and an equal 12 percent stayed for three, four and five days. Finally, four 
birds stayed longer; one stayed six days, two stayed seven days, and one 
stayed eight days. 

There were two supported models for stopover duration, which explained 
61.1 percent of the variation in the data (percent from sum of Akaike 
weights, Wj), compared to the null model, which accounted for 0.034 per¬ 
cent of the total variation. Both top models included energetic condition and 
one included day of season. The model-averaged parameter estimate and SE 
for energetic condition (-0.403 ± 0.156) reflects decreasing duration of stay 
at the stopover site with increasing arrival energetic condition (Figure A10). 
In addition, as the spring progressed, migrants spent less time at the 
stopover site (-0.048 ± 0.027) but most of this relationship was due to the 
four birds that stayed longer than five days, all of which occurred during the 
first three weeks of April. When these four individuals are excluded, day of 
season was no longer a supported variable. 



Figure A10. Duration of stay for red-eyed vireos radio tracked in Kistachie National 
Forest and the relationship between the condition of the bird (negative values are 
below lean body mass and positive are above) and the duration of stay in days. 








ERDC/EL TR-12-22 


112 


Hypothesis E 

The team did not find support for differences in behavior of migrants 
released in pine with and without the addition of conspecific song 
(Table A5). Energetic condition alone was the only supported model for 
both the amount of time spent in the release habitat type (Table A5). 
Without added playback in pine, the habitat type was the only influential 
variable on the time spent in release habitat. Habitat type, together with 
condition, continued to influence the time spent in the release habitat with 
added song in pine (Table A5). Without added playback in pine, migrants 
released in hardwood spent 2.913 more hours in hardwood than migrants 
released in pine spent in that habitat type. With added playback, migrants 
released in hardwood spent 3.381 more hours in their release habitat type 
than did birds released in pine (Table A5). 

Table A5. Relative influence of generalized linear models in predicting the time during the first 
five hours spent in the release habitat type for migrants released 1) at the same location in pine 
with and without playback of conspecific song, 2) in hardwood and pine in years without added 
playback (2007 & 2008) and 3) in hardwood and pine in years with added playback of 
conspecific song (2009). Number of parameters (K), differences in AlCc values (MICc), and 
Akaike weights (w,) are shown. All top models (MICc < 2) and the null model are shown. 
Parameter estimates and standard errors for variables influencing the time in release habitat 

during the first five hours after release. 


Time in release habitat 

K 

AAlCc 

Wj 

Variable 

Parameter estimate 

se 

n 

Pine with and without added song (2007 to 2009) 







12 

Condition 

3 

0.00 

0.789 

Condition 

0.244 

0.058 


Null 

2 

8.60 

0.011 





Hardwood and pine without added song (2007 & 2008) 







17 

Habitat 

3 

0.00 

0.756 

Habitat 

2.913 

0.685 


Null 

2 

10.50 

0.004 





Hardwood and pine with added song (2009) 







10 

Habitat and Condition 

4 

0.00 

0.878 

Habitat 

3.381 

0.522 






Condition 

0.427 

0.104 


Null 

2 

9.40 

0.008 






Since sample sizes were small, the team conducted a post hoc analysis to 
determine whether increasing the sample size (assuming the current sample 
is representative of the population) would increase the team’s ability to 
detect the predicted differences. The team also wanted to determine 
whether the predicted results would be achieved with the addition of a 
sample size that could be obtained in one logistically possible additional 
field season. The current sample was randomly sampled to increase each 
group by factors of three. The sample was built and compared to the 
resulting models for comparisons of migrants released in pine with and 
without playback. Six individuals per release type were added, with added 
playback influencing the amount of time spent in the pine habitat and 
remaining important with the addition of 12 individuals per release type 
(Table A6). 





ERDC/EL TR-12-22 


113 


Table A6. Comparison of generalized linear models comparing relative influence in predicting 
the time during the first five hours spent in pine for migrants released with and without 
playback added (Group). All top models (MICc < 2) and the null model are shown. Each set of 
models presented represents an addition of 3, 6 and 12 randomly selected pairs of 
individuals to each group. Number of parameters (K), differences in AlCc values (MICc), and 
Akaike weights (w ( ) are shown. All top models (MICc < 2) and the null model are shown. 


Model description 

K 

AAlCc 

W; 

Time in release habitat 

Pine with and without song with n increased by three 

Cond 

3 

0.00 

0.554 

Null 

2 

16.10 

<0.001 

Pine with and without song with n increased by six 

Cond 

3 

0.00 

0.39 

Cond, Group 

4 

0.50 

0.31 

Null 

2 

19.20 

<0.001 

Pine with and without song with n increased by twelve 

Cond, Group 

4 

0.00 

0.37 

Cond, Group, Group*cond 

5 

0.50 

0.29 

Cond, Group, Day, Group*Cond 

6 

1.90 

0.14 

Null 

2 

29.80 

<0.001 


Discussion 

The first necessary component in understanding migrant-habitat relations 
is characterizing habitat quality in terms of availability of food resources 
and threats or sources of stress. The team found support in terms of food 
resources for the expectation of variability in habitat quality with hardwood 
the highest quality, pine lowest and mixed intermediate. There were more 
arthropods in hardwood than mixed or pine and more in mixed than pine 
and there were more Lepidoptera larvae in hardwood than pine and in 
mixed than pine but no difference between mixed and pine. Variability in 
food resources was also reflected in foraging observations, with migrants 
having the most successful attacks in hardwood followed by mixed and the 
least success in pine. However, except for three occasions in mixed habitat, 
all red-eyed vireos detected on daily surveys were in hardwood habitat and 
behavioral observations of interactions with breeding red-eyed vireos 
suggest that conspecific competition may be a source of stress for migrants 
in hardwood. It’s possible that migrant distribution wasn’t detectable 
migrants because the migrants rarely vocalized. Radio-tagged migrants 
were never observed vocalizing during extensive visual observations. In 
addition, densities of detected red-eyed vireos along transects remained 
fairly consistent throughout the survey period indicating a majority of them 
were breeding birds. Using mist-nets the team detected the greatest density 
of migratory species in mixed habitat. This could be reflective of greater 






ERDC/EL TR-12-22 


114 


usage of mixed habitat by migratory — versus breeding — individuals which 
are less likely to be detected on surveys. However, the difference in canopy 
height between the hardwood (> 20 m) and mixed (5 to 10 m) habitat in the 
area sampled makes it difficult to compare capture rates. Even if migrants 
eventually moved out of the mixed habitat they may have been more likely 
to be captured in nets while there than while in hardwood. The team’s 
combined measures indicate that hardwood and mixed habitat also 
contained greater densities of migratory species as potential competitors for 
the greater abundance of food resources. However, behavioral results 
suggest that food resources may be the most influential measure of quality 
for migrants during stopover in this landscape. 

Habitat selection may occur at multiple scales from the region to the 
landscape to the habitat patch (Moore et al. 2005, Chernetsov 2006, Buler 
et al. 2007 and Packett and Dunning 2009). While some level of selection 
likely occurs at the regional or landscape scale (Chernetsov 2006, Buler et 
al. 2007, but see Packett and Dunning 2009), limitations of vision in 
nocturnal migrants (Martin 1990) mean that selection likely also occurs 
the morning after landing in a novel landscape. The true frequency of 
active habitat selection has been difficult to document during stopover 
because trapping and tracking most often occur after an unknown period 
of time has already been spent at a stopover site (but see Buler 2006 and 
Cochran and Wikelski 2005). Present study results indicate that the 
majority of migrants that found themselves in low quality habitat moved 
further and faster to select better quality habitat than migrants in higher 
quality habitat. Selection of locations within habitat types was most 
strongly related to distribution and abundance of food resources in the 
poorest quality habitat type but not related to food resources in the highest 
quality habitat type. The highest quality habitat type was characterized by 
both greater food resources and foraging success rates. 

The present study found support for multiple ecological factors affecting 
red-eyed vireo movement from one spatial location to another. Movement 
patterns were also reflective of behavioral variability in relation to habitat 
quality. Migrants released in the poorest quality moved the furthest and 
fastest, suggesting searching, and then most changed habitat types, whereas 
migrants released in high quality habitat initially did not move as far or as 
fast and moved in a more restricted area. Therefore, migrants that land in 
high quality habitat may not expose themselves to threats or lose time 
searching and begin foraging earlier in the morning. Multiple extrinsic 



ERDC/EL TR-12-22 


115 


factors contributed to the decision to change spatial locations. Rate and 
displacement distances of movement were also influenced by the release 
landscape. The release sites were chosen for homogeneity of surrounding 
habitat in as large an area as possible but sites were embedded within a 
heterogeneous landscape. Differences between release sites in the same 
habitat type indicate that landscape context, or the pattern and distribution 
of habitat types, is also important in determining migratory movement. For 
example, many of the migrants released in pine moved into small patches of 
mixed habitat composed of young pine and hardwood along seasonal creeks 
and surrounded by pine savanna. The distances and directions to these 
patches differed between the two pines release sites. Also, Bundick Creek, 
where birds moved both further and faster, has a larger and wider hard¬ 
wood drainage than Drakes Creek. These results suggest that landscape 
context may be an important factor influencing migrant behavior. 

Migrants made movement decisions based on their internal state in terms of 
the amount of fat stored and the temporal proximity to the breeding season. 
Energetic condition and time of season also influenced the duration of time 
spent in the landscape. Energetic condition did not determine whether 
migrants moved because even birds with large fat reserves moved out of low 
quality habitat types to forage in higher quality habitat types, but condition 
was influential in determining how migrants moved. Fat migrants in high 
quality habitat moved faster and further than lean migrants in the same 
habitat type while fat migrants in poorer quality habitat types moved slower 
and stayed closer than lean migrants also in poorer quality habitat types. 
This suggests that migrants with stored energy took advantage of foraging 
opportunities in high quality habitat but were less likely to expose them¬ 
selves to stress or predation risk by moving to foraging in lower quality 
habitat. There was minimal support for time of season, which may be due to 
the fact that it was not possible to accurately measure individual temporal 
and spatial distance to breeding areas. Inclusion of individual breeding 
locations as well as age and sex could strengthen the influence of this 
variable on movement. 

Nocturnal migrants arriving in an unfamiliar landscape likely use a 
combination of cues to assess their surroundings while balancing the need 
to refuel and avoid predators (Chernetsov 2006, Moore and Aborn 2000). 
Behavior during stopover may also be influenced by internal factors such as 
energetic condition and an individuals’ time program (Wang and Moore 
2005, Moore et al. 2003). Therefore, the team did not expect conspecific 



ERDC/EL TR-12-22 


116 


song to be the only factor involved or cue used to make movement deci¬ 
sions. Little is known about the role of breeding conspecific individuals in 
areas where migratory birds stopover during migration. Resident, or 
breeding, individuals may have had more time to assess a landscape or 
habitat patch and therefore have more information about the quality. 
Conversely, a high density of breeding birds may also represent more 
competition for food resources (Moore and Wang 1991). So it is also 
possible that conspecific song could have repellant qualities as an indication 
of density. The team did observe aggressive interactions between migrants 
and resident singing vireos on several occasions (see above, this report). 

This suggests that conspecific song may be an attractant at the landscape 
level (i.e., where the high quality habitat types are) but it may also be a 
deterrent at the patch level (i.e., where the defended territories are). The 
majority of the migrants released at the site in the pine habitat (with or 
without song added) eventually moved into small mixed hardwood patches 
characterized by seasonal creeks that were similar in plant composition to 
hardwood but were much smaller in area. There were no breeding red-eyed 
vireos detected in these patches. This suggests that these habitat types may 
provide resources valuable for the short-term of stopover but not sufficient 
for a breeding red-eyed vireo. 

These results are suggestive of the value of conspecific song but also 
indicate that a larger sample size is needed to make any definitive 
conclusions about the value of conspecific song during stopover. Condition 
was also found to be important, a factor that could not be logistically 
controlled for in a field setting. This finding illustrates the value of testing 
the relative influence of representative models in a field experimental 
setting. To date, there is little information about cues used to make 
decisions during stopover. The team would like to continue this 
experiment with another season of sampling and conduct another 
experiment in the fall because hatch year birds would be expected to rely 
more heavily on conspecific song since they are less experienced with the 
variability of vegetative cues (Stamps 1988). 

Migrating songbirds are challenging subjects with brief stopover periods 
along unpredictable routes and there is currently no technological capability 
to follow individuals for the duration of migration. However, migration may 
be when most mortality occurs (Sillett and Holmes 2002) and most of the 
migratory period is spent refueling at stopover sites (Alerstam 2003). 
Therefore, understanding the causes of mortality during this period is 



ERDC/EL TR-12-22 


117 


essential for the conservation of the many migratory species currently in 
decline (Kirby et al. 2008). In the present study, an experimental 
hypothesis-driven approach was used to better understand migrant-habitat 
relations and movement ecology at a stopover site characteristic of the 
region. The results show migrants made movement decisions in relation to 
both intrinsic and extrinsic factors and that hardwood habitat — as well as 
mixed-woody habitat — may be important for migratory refueling at Fort 
Polk and in Kisatchie National Forest. 

Literature Cited 

Aborn, D. and F. R. Moore. 1997. Pattern of movement by Summer Tanagers ( Piranga 
rubra ) during migratory stopover: a telemetry study. Behaviour 134:1077-1100. 

Alerstam, T. 2003. Bird migration speed. Pp. 253-267 In Avian Migration. Berthold, P., 
Gwinner, E. & Sonnenschein, E., eds. Berlin: Springer-Verlag. 

Bairlein, F. 1983. Habitat selection and associations of species in European passerine 

birds during southward, post-breeding migrations. Ornis Scandinavica. 14:239- 
245 . 

Bairlein, F. 1985. Efficiency of food utilisation during fat deposition in the long-distance 
migratory garden warbler (< Sylvia borin ). Oecologia 68:118-125. 

Belisle, M., Desrochers, A. and Fortin, M. J. 2001. Influence of forest cover on the 
movements of forest birds: A homing experiment. Ecology 82:1893-1904. 

Berthold, P. 1990. Genetics of migration. Pp. 269-280 In Bird Migration: the physiology 
and ecophysiology E. Gwinner, ed. Berlin: Springer-Verlag. 

Blem, C. R. 1980. Avian energy storage. Pp. 59-113 In: Current ornithology, vol 7. (D. M. 
Power, ed). New York, NY: Plenum Press. 

Bolker, B. 2010. Package ‘bbmle’ Tools for general maximum likelihood estimation 
Version 0.9.0. Maintainer Ben Bolker < bolker@zoo.ufl.edu > 

Buler, J. J. 2006. Understanding habitat use by landbirds during migration along the 
Gulf Coast using a scale-dependent approach [Dissertation].Hattiesburg, MS: 
University of Southern Mississippi. 

Buler, J. J., F. R. Moore, and S. Woltmann. 2007. A multi-scale examination of stopover 
habitat use by birds. Ecology 88:1789-1802. 

Burnham, K. P., and D. R. Anderson. 2002. Model selection and multimodel inference: A 
practical information-theoretic approach. 2nd Edition. New York, NY: Springer- 
Verlag. 

Chernetsov, N. 2006. Habitat selection by nocturnal passerine migrants en route: 
mechanisms and results. Journal of Ornithology 147:185-191. 




ERDC/EL TR-12-22 


118 


Cimprich, D. and F. R. Moore. 1999. Energetic constraints and predation pressure during 
stopover. Acta XXII Congressus Internationalis Ornithologici. 

Cimprich, D. and F. R. Moore. 2006. Fat affects predator-avoidance behavior in gray 
catbirds (Dumetella carolinensis) during migratory stopover. Auk 123:1069- 
1076. 

Cimprich, D. A., F. R. Moore, and M. P. Guilfoyle. 2000. Red-eyed Vireo. In The Birds of 
North America, No. 527 A. Poole and F. Gill, eds.. Philadelphia, PA: The Birds of 
North America, Inc. 

Cimprich, D., M. Woodrey, and F. R. Moore. 2005. Passerine migrants respond to 
variation in predation risk during stopover. Animal Behavior 69:1173-1179 

Cochran, W. W. and M. Wikelski. 2005. Individual migratory tactics of New World 

Catharsus thrushes, p. 274- 289. In Birds of two worlds R. Greenberg and P.P. 
Marra, eds. Baltimore, MD: Johns Hopkins University Press. 

Cooper, R. J., and R. C. Whitmore. 1990. Arthropod sampling methods in ornithology. Pp 
29-37 In . Avian foraging: theory, methodology and applications M. L. 

Morrison, C. J. Ralph, J. Verner, and J. J. R. Jehl, eds. Lawrence, KS: Cooper 
Ornithological Society. 

Delingat, J., and V. Dierschke. 2000. Habitat utilization by Northern Wheatears 

(Oenanthe oenanthe) stopping over on an offshore island during migration. Die 
Vogelwarte 40:271-278. 

Evans, D. L. 1994. Forest cover from landsat thematic mapper data for use in Catahoula 
ranger district geographic information system. General Technical Report SO- 
99. New Orleans, LA: U.S. Dept of Agriculture, Forest Service, Southern Forest 
Experiment Station. 

Farmer, A. H., and J. A. Wiens. 1999. Models and reality: time-energy trade-offs in 

Pectoral Sandpiper ( Calidris melanotos ) migration. Ecology 80(8): 2566-2580. 

Farrington, J. W. 2003. Habitat use by migratory birds during fall migration at a stopover 
site in Alabama. M.S. Hattiesburg, MS: The University of Southern Mississippi. 

Gauthreaux, S. A., Jr. 1971. A radar and direct visual study of passerine spring migration 
in southern Louisiana. Auk 88:343-365. 

Gauthreaux, S. A., Jr. 1972. Behavioral responses to migrating birds to daylight and 
darkness: A radar and direct visual study. Wilson Bulletin 84:136-148. 

Graham, C. A. 2001. Factors influencing movement patterns of Keel-billed Toucans in a 
fragmented tropical landscape in southern Mexico. Conservation Biology 15: 
1789-1798. 

Helms, C. W. and W. H. Dury. i960. Winter and migratory weight and fat: field studies 
on some North American buntings. Bird-Banding 31:1-40. 

Hutto, R. L. 1985a. Seasonal changes in the habitat distribution of transient insectivorous 
birds in southeastern Arizona: competition mediated? Auk 102:120-132. 



ERDC/EL TR-12-22 


119 


Hutto, R. L. 1985b. Habitat selection by nonbreeding, migratory land birds. Pp. 455-476 
In Habitat selection in birds, ed. M. L. Cody. New York: Academic Press. 

Johnson, M.D. 2000. Evaluation of an arthropod sampling technique useful in measuring 
food availability for forest insectivores. Journal of Field Ornithology 71:88-109. 

Keddy, P. 2009. Thinking Big: A Conservation Vision for the Southeastern Coastal Plain 
of North America. Southeastern Naturalist 8:213-226. 

Kirby, J. W., A. J. Stattersfield, S. H. M. Butchart, M. I. Evans, R. F. A. Grimmett, V. R. 

Jones, J. O’Sullivan, G. M. Tucker, and I Newton. 2008. Key conservation issues 
for migratory land- and waterbird species on the world’s major flyways. Bird 
Conservation International 18: Suppl. 1: S49-S73. 

Kuenzi, A. J., F. R. Moore, and Simons, T. R. 1991. Stopover of Neotropical landbird 
migrants following trans-Gulf migration. Condor 93:869-883. 

Lima, S. L., and P. A. Zollner. 1996. Anti-predatory vigilance and the limits to collective 
detection: visual and spatial separation between foragers. Behavior Ecology and 
Sociobiology 38:355-363* 

Lindstrom, A., Hasselquist, D., Bensch, S., and Grahn, M. 1990. Asymmetric contests 
over resources for survival and migration: a field experiment with Bluethroats. 
Animal Behavior 40:453-461. 

Loria, D. E., and F. R. Moore 1990. Energy demands of migration on Red-eyed Vireos 
(Vireo olivaceus). Behavioral Ecology 1:24-35. 

Martin, G. R. 1990. The visual problems of nocturnal migration. Pp. 185-197 In Bird 
migration. E. Gwinner, ed. Berlin: Springer-Verlag. 

McGrath, L. J., Charles van Riper III and J. J. Fontaine. 2009. Flower power: tree 

flowering phenology as a settlement cue for migrating birds. Journal of Animal 
Ecology 78:22-30. 

Moore, F. R. 1994. Resumption of feeding under risk of predation: effect of migratory 
condition. Animal Behaviour 48:975-977. 

Moore, F. R. 1999. Neotropical migrants and the Gulf of Mexico: The Cheniers of 

Louisiana and stopover ecology. Pp. 51- 62 In Gatherings of Angels migrating 
birds and their ecology. K.P. Able, ed. Ithaca: Comstock Books. 

Moore, F. R., and D. A. Aborn. 2000. Mechanisms of en route habitat selection: How do 
migrants make habitat decisions during stopover? Studies in Avian Biology 
20:34-42. 

Moore, F. R., and P. Kerlinger. 1987. Stopover and fat deposition by North American 
wood-warblers (Parulinae) following spring migration over the Gulf of Mexico. 
Oecologia 74: 47-54. 

Moore, F. R., P. Kerlinger, and T. R. Simons. 1990. Stopover on a Gulf coast barrier island 
by spring trans-Gulf migrants. Wilson Bulletin 102: 487-500. 



ERDC/EL TR-12-22 


120 


Moore, F. R., S. Mabey, and M. S. Woodrey. 2003. Priority access to food in migratory 
birds: Age, sex and motivational asymmetries. Pages 281-292 In Avian 
Migration, eds. P. Berthold, E. Gwinner, and E. Sonnenschein. Berlin: Springer- 
Verlag Berlin Heidelberg. 

Moore, F. R., and Y. Wang. 1991. Evidence of food-based competition among passerine 
migrants during stopover. Behavioral Ecology & Sociobiology 28: 85-90. 

Moore, F. R., R. J. Smith, and R. Sandberg. 2005. Stopover ecology of intercontinental 

migrants: En route problems and consequences for reproductive performance. In 
Birds of two worlds: The ecology and evolution of migration. 251-261. 
Baltimore, MD: Johns Hopkins University Press. 

Nathan, R., W. M. Getz, E. Revilla, M. Holyoak, R. Kadmon, D. Saltz, and P. E. Smous. 
2008. A movement ecology paradigm for unifying organismal movement 
research. PNAS 105 (49) 19052-19059. 

Nemeth, Z. and F. R. Moore. 2007. Unfamiliar stopover sites and the value of social 

information during migration. Journal of Ornithology 148 (Suppl 2): S369-S376. 

Owen, J. C. and F. R. Moore 2006. Seasonal differences in immunological condition of 
three species of thrushes. Condor 108: 389- 398. 

Packett, D. L. and J. B. Dunning, Jr. 2009. Stopover habitat selection by migrant 
landbirds in a fragmented forest-agricultural landscape. Auk 126: 579-589. 

Parrish, J. D. 2000. Behavioral, energetic, and conservation implications of foraging 
plasticity during migration. Studies in Avian Biology 20:53-70. 

Petit, D. R. 2000. Habitat use by landbirds along Nearctic-Neotropical migration routes: 
Implications for conservation of stopover habitats. Studies in Avian Biology 
20:15-33. 

Pinheiro, J., D. Bates, S. DebRoy, D. Sarkar and the R Development Core Team (2010). 
nlme: Linear and Nonlinear Mixed Effects Models. R package version 3.1-97. 

Remsen, J. V., Jr., and S. K. Robinson. 1990. A classification scheme for foraging 

behavior of birds in terrestrial habitats. Studies in Avian Biology 13:144-160. 

Sillett, S. T., and R. T. Holmes. 2002. Variation in survivorship of a migratory songbird 
throughout its annual cycle. Journal of Animal Ecology 71:296-308. 

Slabbekoorn, H. and N. Bouton. 2008. Soundscape orientation: a new field in need of 
sound investigation. Animal Behavior 76: es~ e8. 

Smith, R. J., M. J. Hamas, D. N. Ewert, and M. E. Dallman. 2004. Spatial foraging 

differences in American redstarts along the shoreline of northern Lake Huron 
during spring migration. Wilson Bulletin 116:48-55. 

Smith, R. and F. R. Moore. 2005. Arrival timing and seasonal reproductive performance 
in a long-distance migratory landbird. Behavioral Ecology and Sociobiology 

57:231-239. 



ERDC/EL TR-12-22 


121 


Spina, F. and E. M. Bezzi. 1990. Autumn migration and orientation of the Sedge Warbler 
(Acrocephalus schoenobaunus ) in northern Italy. Journal filr Ornithologie 131: 
429-438. 

Stamps, J. A. 1988. Conspecific Attraction and Aggregation in Territorial Species. 
American Naturalist 131: 329-347. 

Turcotte, Y. and Desrochers, A. 2003. Landscape-dependent response to predation risk 
by forest birds. Oikos 100: 614-618. 

Wang, Y. and F. R. Moore. 1993. Relation between migratory activity and energetic 
condition among thrushes (Turdinae) following passage across the Gulf of 
Mexico. Condor 95:934-943. 

Wang, Y., and F. R. Moore. 1997. Spring stopover of intercontinental migratory thrushes 
along the northern coast of the Gulf of Mexico. Auk 114:263-278. 

Wang, Y., and F. R. Moore. 2005. Long-distance bird migrants adjust their foraging 
behavior in relation to energy stores. Acta Zoologica Sinica 51:12-23. 

Weber, T. P., B. J. Ens, and A. I. Houston. 1998 Optimal avian migration: a dynamic 
model of nutrient stores and site use. Evolutionary Ecology 12: 377-401. 

Winker, K., J. H. Rappole, and M. A. Ramos. 1995. The use of movement data as an assay 
of habitat quality. Oecologia 101:211-216. 

With, K. A., and A. W. King. 2001. Analysis of landscape sources and sinks: the effect of 
spatial pattern on avian demography. Biological Conservation 100: 75-88. 

Zuur A.F., E.N. Leno, N.J. Walker, A.A. Saveliev and G.M. Smith. 2009. Mixed effects 
models and extensions in ecology with R. Statistics for Biology and Health. 
Springer-Science + business Media LLC. New York, New York. 



ERDC/EL TR-12-22 


122 


Appendix B: Composite Migration Maps Over 
U.S. Military Installations 



ERDC/EL TR-12-22 


123 


APX Spring 2000 & 2001 Hotspots 



/\y Rivers 
Apx_QQ_01 


I 10-5- 
■ 1 . 0 - 

■ 1 - 5 - 

■ 2 . 0 - 
H 2.5- 


- 1.0 Std.Dev. 

■ 1.5 Std.Dev. 

- 2.0 Std. Dev. 

- 2.5 Std. Dev. 

- 3.0 Std. Dev. 

> 3 Std. Dev. 

• Stations 

| | 60 Nautical Mile Limit 

I | Military4_lam 

| | States_outline 

■ Water 

Military4_lam 
States 
Canada 
I | Mexico 


10 0 10 20 30 40 Kilometers 



Copyright 2006 CUROL 


APX Fall 2003 & 2004 Hotspots 



/\J Rivers 
Apx 

□ 0.5- 1.0 Std.Dev. 

■ 1.0- 1.5 Std. Dev. 

■ 1.5-2.0 Std.Dev. 

|-1 2.0- 2.5 Std. Dev. 

H 2.5- 3.0 Std.Dev. 

■ > 3 Std. Dev. 

• Stations 

| | 60 Nautical Mile Limit 

I | Military4_lam 

| 1 States_outline 

■ Water 
Military4_lam 

■ States 

■ Canada 
I | Mexico 


10 0 10 20 30 40 Kilometers 


wl I Ur. 





Copyright 2005 CUROL 


Figure Bl. Composite maps indicating spring and fall migratory hotspots as recorded by 
NEXRAD station APX in northern Ml. The survey area encompasses Camp Grayling Military 

Reservation. 




















ERDC/EL TR-12-22 


124 


ARX Spring 2000 & 2001 Hotspots 



A / Rivers 
Arx 00 01 


□ 0.5- 
■ 1 . 0 - 

■ 1 - 5 - 

■ 2 . 0 - 
H 2.5- 


- 1.0 Std.Dev. 

■ 1.5 Std.Dev. 

- 2.0 Std. Dev. 

- 2.5 Std. Dev. 

- 3.0 Std. Dev. 

> 3 Std. Dev. 

• Stations 

| | 60 Nautical Mile Limit 

I | Military4_lam 

| | States_outline 

■ Water 

Military4_lam 
States 
Canada 
I | Mexico 


10 0 10 20 30 40 Kilometers 



Copyright 2006 CUROL 


ARX FALL 2003 & 2004 Hotspots 



/\y Rivers 

□ 0.5- 1.0 Std.Dev. 
H 1.0- 1.5 Std.Dev. 

■ 1.5-2.0 Std.Dev. 

□ 2.0-2.5 Std.Dev. 
H 2.5- 3.0 Std.Dev. 

■ > 3 Std. Dev. 

• Stations 

| | 60 Nautical Mile Limit 

I I Military4_lam 

| | States_outline 

■ Water 
Military4_lam 

■ States 

I I Canada 

I I Mexico 


10 0 10 20 30 40 Kilometers 



Copyright 2005 CUROL 


Figure B2. Composite maps indicating spring and fall migratory hotspots as recorded by 
NEXRAD station ARX in La Crosse, Wl. The survey area encompasses Fort McCoy. 

























ERDC/EL TR-12-22 


125 


CBX Spring 2000 & 2001 Hotspots 



A / Rivers 
Cbx 00 01 


□ 0.5- 
■ 1 . 0 - 

■ 1-5- 

■ 2 . 0 - 
H 2.5- 


- 1.0 Std.Dev. 

■ 1.5 Std.Dev. 

- 2.0 Std. Dev. 

- 2.5 Std. Dev. 

- 3.0 Std. Dev. 

> 3 Std. Dev. 

• Stations 

| | 60 Nautical Mile Limit 

I | Military4_lam 

| | States_outline 

■ Water 

Military4_lam 

States 

^B Canada 
I | Mexico 


10 0 10 20 30 40 Kilometers 




Copyright 2006 CUROL 


CBX FALL 2003 & 2004 Hotspots 



/\J Rivers 
Cbx 

| | 0.5- 1.0 Std. Dev. 

Hi 10- 1.5 Std. Dev. 

■ 1.5- 2.0 Std. Dev. 
□□2.0-2.5 Std.Dev. 
H 25-3.0 Std.Dev. 

> 3 Std. Dev. 

• Stations 

| | 60 Nautical Mile Limit 

I | Military4_lam 

| | States_outline 

■ Water 
^B Military4_lam 

□ States 
^B Canada 

□ Mexico 


10 0 10 20 30 40 Kilometers 



Copyright 2005 CUROL 


Figure B3. Composite maps indicating spring and fall migratory hotspots as recorded by 
NEXRAD station CBX in Boise, ID. The survey area encompasses Saylor Creek Air Force 

Range. 


















ERDC/EL TR-12-22 


126 


CLX Spring 2000 & 2001 Hotspots 



A/ Rivers 
Clx_00_01 

| | 0.5- 1.0 Std. Dev. 

^B 1.0- 1.5 Std. Dev. 

|-1 1.5- 2.0 Std. Dev. 

|-1 2.0- 2.5 Std. Dev. 

2.5 - 3.0 Std. Dev. 

■ > 3 Std. Dev. 

• Stations 

| | 60 Nautical Mile Limit 

I I Military4_lam 

| | States_outline 

■ Water 

I I Military4_lam 

■ States 
■■ Canada 
□ Mexico 


10 0 10 20 30 40 Kilometers 



Copyright 2005 CUROL 


CLX Fall 03 & 04 Hotspots 



I | Military4_lam 
/\y Rivers 
Clx 

|-1 0.5- 1.0 Std. Dev. 

■ 1.0- 1.5 Std. Dev. 
|1.5 - 2.0 Std. Dev. 

I | 2 0-2.5 Std. Dev. 

2.5 - 3.0 Std. Dev. 
^B > 3 Std. Dev. 

• Stations 

I | 60 Nautical Mile Limit 

| | States_outline 

■ Water 
Military4_lam 

□ States 
BB Canada 
I I Mexico 


10 0 10 20 30 40 Kilometers 







Copyright 2005 CUROL 


Figure B4. Composite maps indicating spring and fall migratory hotspots as recorded by 
NEXRAD station CLX in Charleston, SC. The survey area encompasses Fort Stewart. 























ERDC/EL TR-12-22 


127 


EMX Fall 2003 & 2004 Hotspots 



A/ Rivers 
Emx 

| | 0.5- 1.0 Std. Dev. 

1.0- 1.5 Std. Dev. 
H 15 - 2.0 Std. Dev. 

|-1 2.0- 2.5 Std. Dev. 

^2.5- 3.0 Std. Dev. 

■ > 3 Std. Dev. 

• Stations 

| | 60 Nautical Mile Limit 

I I Military4_lam 

| | States_outline 

■ Water 

I I Military4_lam 

■ States 
Canada 

□ Mexico 


10 0 10 20 30 40 Kilometers 



Copyright 2005 CUROL 


Figure B5. Composite map indicating fall migratory hotspots as recorded by NEXRAD 
station EMX in Tucson, AZ. The survey area encompasses Fort Huachuca. 












ERDC/EL TR-12-22 


128 


EOX Spring 2000 & 2001 Hotspots 



A/ Rivers 
Eox 00_01 


I-1 0 5 - 

^B i o- 

■ 1.5- 

■ 2 . 0 - 
M 2.5- 


- 1.0 Std. Dev. 

- 1.5 Std. Dev. 

- 2.0 Std. Dev. 

- 2.5 Std. Dev. 

- 3.0 Std. Dev. 

> 3 Std. Dev. 

| 60 Nautical Mile Limit 
I I Military4_lam 

| | States_outline 

• Stations 
^B Water 
I I Military4_lam 
■ States 
□ Canada 
I I Mexico 


10 0 10 20 30 40 Kilometers 



Copyright 2005 CUROL 


EOX Fall 2003 & 2004 Hotspots 



A/ Rivers 
Eox 

| | 0.5- 1.0 Std. Dev. 

^B 1.0- 1.5 Std. Dev. 

|-1 1 5 - 2.0 Std. Dev. 

|-1 2.0- 2.5 Std. Dev. 

H2.5- 3.0 Std. Dev. 

■ > 3 Std. Dev. 

• Stations 

| | 60 Nautical Mile Limit 

I Military4_lam 
| | States_outline 

■ Water 

I I Military4_lam 

■ States 
BB Canada 
□ Mexico 


10 0 10 20 30 40 Kilometers 



Copyright 2005 CUROL 


Figure B6. Composite maps indicating spring and fall migratory hotspots as recorded by 
NEXRAD station EOX in southeastern AL. The survey area encompasses Fort Rucker 

Military Reservation. 

























ERDC/EL TR-12-22 


129 


EPZ Spring 2000 & 2001 Hotspots 



& 


Rivers 
pz_00_01 


I-1 0 5 - 

K i o- 

■ 1.5- 

■ 2 . 0 - 
M 2.5- 


- 1.0 Std. Dev. 

- 1.5 Std. Dev. 

- 2.0 Std. Dev. 

- 2.5 Std. Dev. 

- 3.0 Std. Dev. 

> 3 Std. Dev. 

• Stations 

| | 60 Nautical Mile Limit 

I I Military4_lam 

| | States_outline 

■ Water 

I I Military4_lam 

■ States 
■■ Canada 
□ Mexico 


10 0 10 20 30 40 Kilometers 



Copyright 2005 CUROL 


EPZ Fall 2003 & 2004 Hotspots 





Rivers 
:pz 
I-10.5- 

H i o- 
■ 1-5- 
□ 2 . 0 - 


- 1.0 Std. Dev. 

- 1.5 Std. Dev. 

- 2.0 Std. Dev. 

- 2.5 Std. Dev. 

- 3.0 Std. Dev. 

> 3 Std. Dev. 

• Stations 

| | 60 Nautical Mile Limit 

| | Military4_lam 

| | States_outline 

■ Water 

I | Military4_lam 

■ States 

I I Canada 

| | Mexico 


10 0 10 20 30 40 Kilometers 


U;w 





Copyiight 2005 CUROL 


Figure B7. Composite maps indicating spring and fall migratory hotspots as recorded by 
NEXRAD station EPZ in El Paso, TX. The survey area encompasses Fort Bliss and the Fort 

Bliss McGregor Range. 






















ERDC/EL TR-12-22 


130 


EVX Spring 2000 & 2001 Hotspots 



A/ Rivers 
Evx_00_01 

| | 0.5- 1.0 Std. Dev. 

1.0- 1.5 Std. Dev. 
H 15 - 2.0 Std. Dev. 

|-1 2.0- 2.5 Std. Dev. 

^2.5- 3.0 Std. Dev. 

■ > 3 Std. Dev. 

• Stations 

| | 60 Nautical Mile Limit 

I I Military4_lam 

| | States_outline 

■ Water 

I I Military4_lam 

■ States 
■■ Canada 
□ Mexico 


10 0 10 20 30 40 Kilometers 



Copyright 2005 CUROL 


EVX Fall 2000 to 2004 Hotspots 



A/ Rivers 
Evx_fallsto04 
□ 0.5- 1.0 Std. Dev. 

■ 1.0- 1.5 Std. Dev. 

■ 1.5 - 2.0 Std. Dev. 
_2.0- 2.5 Std. Dev. 
H 2.5- 3.0 Std. Dev. 

■ > 3 Std. Dev. 

• Stations 

| | 60 Nautical Mile Limit 

I | Military4_lam 

| | States_outline 

■ Water 
Military4Jam 

I 1 States 
■■ Canada 
I I Mexico 


10 0 10 20 30 40 Kilometers 





Copyright 2005 CUROL 


Figure B8. Composite maps indicating spring and fall migratory hotspots as recorded by 
N EX RAD station EVX in northwestern FL. The survey area encompasses Eglin Air Force 

Base. 























ERDC/EL TR-12-22 


131 


EYX Spring 2000 & 2001 Hotspots 



A/ Rivers 
Eyx_00_01 

r I 0.5- 1.0 Std. Dev. 
^B 1.0- 1.5 Std. Dev. 
H 15 - 2.0 Std. Dev. 

|-1 2.0- 2.5 Std. Dev. 

^2.5-3 0 Std. Dev. 

■ > 3 Std. Dev. 

• Stations 

| | 60 Nautical Mile Limit 

I I Military4_lam 

| | States_outline 

■ Water 

I I Military4_lam 

■ States 
■■ Canada 
□ Mexico 


10 0 10 20 30 40 Kilometers 





Copyright 2005 CUROL 


EYX Fall 2003 & 2004 Hotspots 



A/ Rivers 
Eyx 

r I 0.5- 1.0 Std. Dev. 

■ 1.0- 1.5 Std. Dev. 

■ 1.5- 2.0 Std. Dev. 

|-1 2.0- 2.5 Std. Dev. 

^2.5- 3.0 Std. Dev. 
^B > 3 Std. Dev. 

• Stations 

| | 60 Nautical Mile Limit 

I I Military4_lam 

| | States_outline 

Bi Water 
^B Military4_lam 

■ States 

^B Canada 
I I Mexico 


10 0 10 20 30 40 Kilometers 





Copyright 2005 CUROL 


Figure B9. Composite maps indicating spring and fall migratory hotspots as recorded by 
NEXRAD station EYX in southern CA. The survey area encompasses China Lake Naval 
Weapons Center, Edwards Air Force Base, and Fort Irwin. 





















ERDC/EL TR-12-22 


132 


FDR Spring 2000 & 2001 Hotspots 



A/ Rivers 
Fdr_00_01 

| | 0.5- 1.0 Std. Dev. 

1.0- 1.5 Std. Dev. 
H 15 - 2.0 Std. Dev. 

|-1 2.0- 2.5 Std. Dev. 

^2.5-3 0 Std. Dev. 

■ > 3 Std. Dev. 

H Water 
• Stations 

| | 60 Nautical Mile Limit 

| | States_outline 

I | Military4_lam 

| Military4_lam 
H States 
■■ Canada 
□ Mexico 


10 0 10 20 30 40 Kilometers 


O 

% 


U.V/j 


'a, 






Copyright 2005 CUROL 



Figure BIO. Composite maps indicating spring and fall migratory hotspots as recorded by 
NEXRAD station FDR in Frederick, OK. The survey area encompasses Fort Sill Military 

Reservation. 























ERDC/EL TR-12-22 


133 


GRK Spring 2000 & 2001 Hotspots 



A/ Rivers 
Grk_00_01 

| | 0.5- 1.0 Std. Dev. 

1.0 - 1.5 Std. Dev. 
H 15 - 2.0 Std. Dev. 

|-1 2.0- 2.5 Std. Dev. 

2.5 - 3.0 Std. Dev. 

■ > 3 Std. Dev. 

■ Water 

| Military4_lam 
| | Military4_lam 

. Stations 

I | 60 Nautical Mile Limit 

I I States_outline 

■ States 
■■ Canada 
□ Mexico 


10 0 10 20 30 40 Kilometers 



Copyright 2005 CUROL 


GRK Fall 2003 & 2004 Hotspots 



f\J Rivers 
Grk 

| | 0.5- 1.0 Std. Dev. 

Hi 10 - 1.5 Std. Dev. 

■ 1.5- 2.0 Std. Dev. 
□□2.0- 2.5 Std. Dev. 
H 2 5-3.0 Std. Dev. 

> 3 Std. Dev. 

• Stations 

| | 60 Nautical Mile Limit 

I | Military4_lam 

| | States_outline 

■ Water 
Military4_lam 

□ States 
Canada 

□ Mexico 


10 0 10 20 30 40 Kilometers 





Copyright 2005 CUROL 


Figure Bll. Composite maps indicating spring and fall migratory hotspots as recorded by 
NEXRAD station GRK in central TX. The survey area encompasses Fort Hood and Camp 

Swift N. G. Facility. 




























ERDC/EL TR-12-22 


134 


HDX Spring 2000 & 2001 Hotspots 





Rivers 
Idx 00 01 


| 0.5- 1.0 Std. Dev. 
^B 1.0- 1.5 Std. Dev. 
H 15 - 2.0 Std. Dev. 

|-1 2.0- 2.5 Std. Dev. 

2.5 - 3.0 Std. Dev. 

■ > 3 Std. Dev. 

• Stations 

| | 60 Nautical Mile Limit 

I I Military4_lam 

| | States_outline 

■ Water 

I I Military4_lam 

■ States 
■■ Canada 
□ Mexico 


10 0 10 20 30 40 Kilometers 



Copyright 2005 CUROL 


HDX Fall 2004 Hotspots 



A/F 

Hdx_0 


' Rivers 
Hdx_04 

I-10.5- 

^B 1.0- 

H 1.5- 


- 1.0 Std. Dev. 

- 1.5 Std. Dev. 

- 2.0 Std. Dev. 

|-1 2.0- 2.5 Std. Dev. 

Bl 2.5 - 3.0 Std. Dev. 
H : 3 Std. Dev. 

• Stations 

| | 60 Nautical Mile Limit 

| | Military4_lam 

| | States_outline 

■ Water 

I | f.1ilitary4_lam 

■ States 

I I Canada 

| | Mexico 


10 0 10 20 30 40 Kilometers 


-^U/w 

V ^ %. 



OjlWTYK-A 


Copyright 2005 CUROL 


Figure B12. Composite maps indicating spring and fall migratory hotspots as recorded by 
NEXRAD station HDX in southern NM. The survey area encompasses Holloman Air Force 
Base, White Sands Missile Range, Fort Bliss, and the Fort Bliss McGregor Range. 























ERDC/EL TR-12-22 


135 


HPX Spring 2001 Hotspots 





Rivers 
lpx_00_01 


I-1 0 5 - 

K i o- 

■ 1.5- 

■ 2 . 0 - 
M 2.5- 


- 1.0 Std. Dev. 

- 1.5 Std. Dev. 

- 2.0 Std. Dev. 

- 2.5 Std. Dev. 

- 3.0 Std. Dev. 

> 3 Std. Dev. 

• Stations 

| | 60 Nautical Mile Limit 

I I Military4_lam 

| | States_outline 

■ Water 

I I Military4_lam 

■ States 
■■ Canada 
□ Mexico 


10 0 10 20 30 40 Kilometers 



HPX Fall 2003 & 2004 Hotspots 



f^ F 


7 Rivers 
Hpx 
□ 05 - 

K i o- 

H 1.5- 


- 1.0 Std. Dev. 

- 1.5 Std. Dev. 

- 2.0 Std. Dev. 

|-1 2.0- 2.5 Std. Dev. 

H2.5- 3.0 Std. Dev. 

■ > 3 Std. Dev. 

• Stations 

| | 60 Nautical Mile Limit 

I Military4_lam 
| | States_outline 

■ Water 

I I Military4_lam 

■ States 
■■ Canada 
□ Mexico 


10 0 10 20 30 40 Kilometers 



Copyright 2005 CUROL 


Figure B13. Composite maps indicating spring and fall migratory hotspots as recorded by 
NEXRAD station HPX in southwestern KY. The survey area encompasses Fort Campbell. 
























ERDC/EL TR-12-22 


136 


LVX Spring 2000 & 2001 Hotspots 



A/ F 

LvxJX 

I-1 0 5 - 

| | 1 . 0 - 

■ 1.5- 

■ 2 . 0 - 
M 2.5- 


/ Rivers 
Lvx v _00_01 

- 1.0 Std. Dev. 

- 1.5 Std. Dev. 

- 2.0 Std. Dev. 

- 2.5 Std. Dev. 

- 3.0 Std. Dev. 

> 3 Std. Dev. 

• Stations 

| | 60 Nautical Mile Limit 

I I Military4_lam 

| | States_outline 

■ Water 

I I Military4_lam 

■ States 
■■ Canada 
□ Mexico 


10 0 10 20 30 40 Kilometers 



Copyright 2006 CUROL 


LVX FALL 2003 & 2004 Hotspots 



c/ 


Rivers 


I-1 0 5 - 

K 1-0- 

■ 1.5- 

■ 2 . 0 - 
H 2.5- 


- 1.0 Std. Dev. 

- 1.5 Std. Dev. 

- 2.0 Std. Dev. 

- 2.5 Std. Dev. 

- 3.0 Std. Dev. 

> 3 Std. Dev. 

• Stations 

| | 60 Nautical Mile Limit 

I Military4_lam 
| | States_outline 

■ Water 

I I Military4_lam 

■ States 
■■ Canada 
□ Mexico 


10 0 10 20 30 40 Kilometers 



Copyright 2005 CUROL 


Figure B14. Composite maps indicating spring and fall migratory hotspots as recorded by 
NEXRAD station LVX in central KY. The survey area encompasses Fort Knox. 


























ERDC/EL TR-12-22 


137 




LWX Spring 2000 & 2001 Hotspots 


/V/ Rivers 
Lwx_00_01 

| | 0.5- 1.0 Std. Dev. 

1.0- 1.5 Std. Dev. 
H 15 - 2.0 Std. Dev. 

|-1 2.0- 2.5 Std. Dev. 

H 2.5-3.0 Std. Dev. 

■ > 3 Std. Dev. 

• Stations 

| | 60 Nautical Mile Limit 

I Military4_lam 
| | States_outline 

■ Water 

I I Military4_lam 

■ States 
■■ Canada 
□ Mexico 




v 


V 




LWX FALL 2003 & 2004 Hotspots 


A/F 

Lwx 


i Rivers 
Lwx 

| | 0.5- 1.0 Std. Dev. 

Hi 10 - 1.5 Std. Dev. 

■ 1.5- 2.0 Std. Dev. 
□□2.0- 2.5 Std. Dev. 
H 2 5-3.0 Std. Dev. 
H > 3 Std. Dev. 

• Stations 

| | 60 Nautical Mile Limit 

I | Military4_lam 

| | States_outline 

■ Water 
Military4_lam 

□ States 
^B Canada 

□ Mexico 




Figure B15. Composite maps indicating spring and fall migratory hotspots as recorded by 
NEXRAD station LWX in Sterling, VA. The survey area encompasses Fort A.P. Hill Military 
Reservation and Quantico Marine Corps Base. 
































ERDC/EL TR-12-22 


138 


MHX Spring 2000 & 2001 Hotspots 



/V/ Rivers 
Mhx 00 01 


I-1 0 5 - 

K 1-0- 

■ 1.5- 

■ 2 . 0 - 
M 2.5- 


- 1.0 Std. Dev. 

- 1.5 Std. Dev. 

- 2.0 Std. Dev. 

- 2.5 Std. Dev. 

- 3.0 Std. Dev. 

> 3 Std. Dev. 

• Stations 

| | 60 Nautical Mile Limit 

I Military4_lam 
| | States_outline 

■ Water 

I I Military4_lam 

■ States 
■■ Canada 
□ Mexico 


10 0 10 20 30 40 Kilometers 



Copyright 2005 CUROL 



Figure B16. Composite maps indicating spring and fall migratory hotspots as recorded by 
NEXRAD station MHX in Morehead City, NC. The survey area encompasses Camp Lejeune 

Marine Corps Base. 
























ERDC/EL TR-12-22 


139 


MLB Spring 2000 & 2001 Hotspots 



/V F 

Mlb_0( 

I-1 0 5 - 

B i o- 

■ 1.5- 

■ 2 . 0 - 
B 2.5- 


7 Rivers 
v1lb v _00_01 

- 1.0 Std. Dev. 

- 1.5 Std. Dev. 

- 2.0 Std. Dev. 

- 2.5 Std. Dev. 

- 3.0 Std. Dev. 

> 3 Std. Dev. 

• Stations 

| | 60 Nautical Mile Limit 

I I Military4_lam 

| | States_outline 

■ Water 

I I Military4_lam 

■ States 
■■ Canada 
□ Mexico 


10 0 10 20 30 40 Kilometers 


A, 


5 

U 

S’ 


Qt! 


V'a 


.vrrvvJ^ 


Copyright 2006 CUROL 


MLB Fall 2003 & 2004 Hotspots 

iviy 



Rivers 
llb v 

I-10.5- 

B 1 . 0 - 


- 1.0 Std. Dev. 

- 1.5 Std. Dev. 

| | 1.5- 2.0 Std. Dev. 

|B 2.0- 2.5 Std. Dev. 
B 2.5-3.0 Std. Dev. 
B| > 3 Std Dev. 

■ Water 

• Stations 

I | 60 Nautical Mile Limit 
I | Military4_lam 

I I States_outline 

I ] Military4_lam 

■ States 

I I Canada 

| | Mexico 


10 0 10 20 30 40 Kilometers 


U 

& 


r'vX 

>S 

\ Sr 


Copyright 2006 CUROL 


Figure B17. Composite maps indicating spring and fall migratory hotspots as recorded by 
NEXRAD station MLB in Melbourne, FL. The survey area encompasses Avon Park Air Force 

Bombing Range. 























ERDC/EL TR-12-22 


140 


MTX Spring 2000 & 2001 Hotspots 



A/ F 

MtX_0( 

I-1 0 5 - 

K 1-0- 

■ 1.5- 

■ 2 . 0 - 
M 2.5- 


7 Rivers 
Vltx'_00_01 

- 1.0 Std. Dev. 

- 1.5 Std. Dev. 

- 2.0 Std. Dev. 

- 2.5 Std. Dev. 

- 3.0 Std. Dev. 

> 3 Std. Dev. 

• Stations 

| | 60 Nautical Mile Limit 

I I Military4_lam 

| | States_outline 

■ Water 

I I Military4_lam 

■ States 
■■ Canada 
□ Mexico 


10 0 10 20 30 40 Kilometers 



Copyright 2005 CUROL 


MTX Fall 2003 & 2004 Hotspots 





/ Rivers 
VI tx 
□ 0.5- 

K i o- 

H 1.5- 


- 1.0 Std. Dev. 

- 1.5 Std. Dev. 

- 2.0 Std. Dev. 

|-1 2.0- 2.5 Std. Dev. 

H2.5- 3.0 Std. Dev. 

■ > 3 Std. Dev. 

• Stations 

| | 60 Nautical Mile Limit 

I I Military4_lam 

| | States_outline 

■ Water 

I I Military4_lam 

■ States 
■■ Canada 
□ Mexico 


10 0 10 20 30 40 Kilometers 



Copyright 2005 CUROL 


Figure B18. Composite maps indicating spring and fall migratory hotspots as recorded by 
NEXRAD station MTX in Salt Lake City, UT. The survey area encompasses Hill Air Force 
Range and the Hill AFB Wendover Range. 


























ERDC/EL TR-12-22 


141 


MXX Spring 2000 & 2001 Hotspots 



A/ Rivers 
Mxx_00_01 

| | 0.5- 1.0 Std. Dev. 

^B 1.0- 1.5 Std. Dev. 
^B 1.5- 2.0 Std. Dev. 

|-1 2.0- 2.5 Std. Dev. 

^2.5-3 0 Std. Dev. 

B1 > 3 Std. Dev. 

• Stations 

| | 60 Nautical Mile Limit 

I I Military4_lam 

| | States_outline 

■ Water 

I I Military4_lam 

■ States 
BH Canada 
□ Mexico 


10 0 10 20 30 40 Kilometers 



Copyright 2005 CUROL 


MXX Fall 2003 & 2004 Hotspots 



A/ Rivers 
Mxx 03 04hs 


I- 105- 

H 1.0- 
■ 1.5- 
12 . 0 - 
M 2.5- 


- 1.0 Std. Dev. 

- 1.5 Std. Dev. 

- 2.0 Std. Dev. 

- 2.5 Std. Dev. 

- 3.0 Std. Dev. 
^B > 3 Std. Dev. 

• Stations 

| | 60 Nautical Mile Limit 

I I Military4_lam 

| | States_outline 

Bi Water 
^B Military4_lam 
■ States 
^B Canada 
I I Mexico 


10 0 10 20 30 40 Kilometers 





Copyright 2005 CUROL 


Figure B19. Composite maps indicating spring and fall migratory hotspots as recorded by 
N EX RAD station MXX in eastern AL. The survey area encompasses Fort Benning. 























ERDC/EL TR-12-22 


142 


NKX Spring 2000 & 2001 Hotspots 





Rivers 
Ikx 00 01 


| 0.5- 1.0 Std. Dev. 
^B 1.0- 1.5 Std. Dev. 
H 15 - 2.0 Std. Dev. 

|-1 2.0- 2.5 Std. Dev. 

H 2-5-3.0 Std. Dev. 

■ > 3 Std. Dev. 

• Stations 

| | 60 Nautical Mile Limit 

I I Military4_lam 

| | States_outline 

■ Water 

I I Military4_lam 

■ States 
■■ Canada 
□ Mexico 


10 0 10 20 30 40 Kilometers 



NKX Fall 2003 & 2004 Hotspots 





/ Rivers 
Klkx 

I-1 0 5 - 

H 1.0- 
H 1.5- 


- 1.0 Std. Dev. 

- 1.5 Std. Dev. 

- 2.0 Std. Dev. 

|-1 2.0- 2.5 Std. Dev. 

H?.5- 3.0 Std. Dev. 
H > 3 Std. Dev. 

• Stations 

| | 60 Nautical Mile Limit 

I I Military4_lam 

| | States_outline 

■ Water 
Military4_lam 

■ States 
Canada 

I I Mexico 


10 0 10 20 30 40 Kilometers 





Copyright 2005 CUROL 


Figure B20. Composite maps indicating spring and fall migratory hotspots as recorded by 
NEXRAD station NKX in San Diego, CA. The survey area encompasses Camp Pendleton 

Marine Corps Base. 




























ERDC/EL TR-12-22 


143 


PDT Spring 2000 & 2001 Hotspots 



A/ Rivers 
Pdt_00_01 

| | 0.5- 1.0 Std. Dev. 

^B 1.0- 1.5 Std. Dev. 
H 15 - 2.0 Std. Dev. 

|-1 2.0- 2.5 Std. Dev. 

^2.5-3 0 Std. Dev. 

■ > 3 Std. Dev. 

• Stations 

| | 60 Nautical Mile Limit 

I I Military4_lam 

| | States_outline 

■ Water 

I I Military4_lam 

■ States 
■■ Canada 
□ Mexico 


10 0 10 20 30 40 Kilometers 




5 

u 


>5 






Copyright 2006 CUROL 


PDT Fall 2003 & 2004 Hotspots 



\ ( A -Y~" v \ • Y . f&i 

( \\V )J -V 

a ) c\ JUa..- 

\ \ \ 



A/ Rivers 
Pdt 

|-1 0.5- 1.0 Std. Dev. 

^B 1.0- 1.5 Std. Dev. 
^B 15 - 2.0 Std. Dev. 

|-1 2.0- 2.5 Std. Dev. 

^2.5- 3.0 Std. Dev. 

■ > 3 Std. Dev. 

• Stations 

| | 60 Nautical Mile Limit 

I Military4_lam 
| | States_outline 

■ Water 

I I Military4_lam 

■ States 
BH Canada 
□ Mexico 


10 0 10 20 30 40 Kilometers 


o 

so 

\LSr ^ 


Copyright 2005 CUROL 


Figure B21. Composite maps indicating spring and fall migratory hotspots as recorded by 
NEXRAD station PDT in Pendleton, OR. The survey area encompasses the Boardman Naval 

Bombing Range. 






















ERDC/EL TR-12-22 


144 


POE Spring 2000 & 2001 Hotspots 



A/ Rivers 
Poe 00 01 


I-1 0 5 - 

K i o- 

■ 1.5- 

■ 2 . 0 - 
M 2.5- 


- 1.0 Std. Dev. 

- 1.5 Std. Dev. 

- 2.0 Std. Dev. 

- 2.5 Std. Dev. 

- 3.0 Std. Dev. 

> 3 Std. Dev. 

• Stations 

| | 60 Nautical Mile Limit 

I I Military4_lam 

| | States_outline 

■ Water 

I I Military4_lam 

■ States 
■■ Canada 
□ Mexico 


10 0 10 20 30 40 Kilometers 



Copyright 2005 CUROL 


POE Fall 2000 to 2004 Hotspots 



A/ Rivers 
Poe_00_to_04 


□ 0.5- 
H i.o- 

■ 1-5- 

■ 2.0 - 
■■ 2.5- 


- 1.0 Std. Dev. 

- 1.5 Std. Dev. 

- 2.0 Std. Dev. 

- 2.5 Std. Dev. 

- 3.0 Std. Dev. 

> 3 Std. Dev. 

• Stations 

| | 60 Nautical Mile Limit 

I I Military4_lam 

| | States_outline 

■ Water 
Military4_lam 

■ States 

I I Canada 

I I Mexico 


10 0 10 20 30 40 Kilometers 



Copyright 2005 CUROL 


Figure B22. Composite maps indicating spring and fall migratory hotspots as recorded by 
NEXRAD station POE in central, LA. The survey area encompasses Fort Polk. 



























ERDC/EL TR-12-22 


145 


PUX Spring 2000 & 2001 Hotspots 



A/ Rivers 
Pux 00 01 


I-1 0 5 - 

H 1.0- 


- 1.0 Std. Dev. 

- 1.5 Std. Dev. 
H 15 - 2.0 Std. Dev. 

|-1 2.0- 2.5 Std. Dev. 

H 2.5-3.0 Std. Dev. 

■ > 3 Std. Dev. 

• Stations 

| | 60 Nautical Mile Limit 

I I Military4_lam 

| | States_outline 

■ Water 

I I Military4_lam 

■ States 
■■ Canada 
□ Mexico 


10 0 10 20 30 40 Kilometers 







Copyright 2005 CUROL 


PUX Fall 2003 & 2004 Hotspots 



A / Rivers 
Pux 

| | 0.5- 1.0 Std. Dev. 

Hi 10 - 1.5 Std. Dev. 

■ 1.5- 2.0 Std. Dev. 

|- 1 2 0-2.5 Std. Dev. 

H 2 5-3.0 Std. Dev. 
H > 3 Std. Dev. 

• Stations 

| | 60 Nautical Mile Limit 

I | Military4_lam 

| 1 States_outline 

■ Water 
Military4_lam 

□ States 
^B Canada 

□ Mexico 


10 0 10 20 30 40 Kilometers 



Copyright 2005 CUROL 


Figure B23. Composite maps indicating spring and fall migratory hotspots as recorded by 
NEXRAD station PUX in Pueblo, CO. The survey area encompasses Fort Carson Military 

Reservation. 

























ERDC/EL TR-12-22 


146 


RAX Spring 2000 & 2001 Hotspots 



A / Rivers 
Rax_ QQ_Q1 

| | 0.5- 1.0 Std. Dev. 

1.0- 1.5 Std. Dev. 
H 15 - 2.0 Std. Dev. 

|-1 2.0- 2.5 Std. Dev. 

H 2.5-3.0 Std. Dev. 

■ > 3 Std. Dev. 

• Stations 

| | 60 Nautical Mile Limit 

I I Military4_lam 

| | States_outline 

■ Water 

I I Military4_lam 

■ States 
■■ Canada 
□ Mexico 


10 0 10 20 30 40 Kilometers 



!Jvtrm^ T 


Copyright 2005 CUROL 


RAX Fall 2003 & 2004 Hotspots 



A / Rivers 
Rax 

| | 0.5- 1.0 Std. Dev. 

Hi 10 - 1.5 Std. Dev. 

■ 1.5- 2.0 Std. Dev. 
□□2.0- 2.5 Std. Dev. 
H 2 5-3.0 Std. Dev. 

> 3 Std. Dev. 

• Stations 

| | 60 Nautical Mile Limit 

I | Military4_lam 

| | States_outline 

■ Water 
Military4_lam 

□ States 
Canada 

□ Mexico 


10 0 10 20 30 40 Kilometers 





Copyright 2005 CUROL 


Figure B24. Composite maps indicating spring and fall migratory hotspots as recorded by 
NEXRAD station RAX in Raleigh-Durham, NC. The survey area encompasses Fort Bragg. 


























ERDC/EL TR-12-22 


147 


RGX Spring 2000 & 2001 Hotspots 



A/ Rivers 
Rgx_00_01 

| | 0.5- 1.0 Std. Dev. 

^B 1.0- 1.5 Std. Dev. 
H 15 - 2.0 Std. Dev. 

|-1 2.0- 2.5 Std. Dev. 

^2.5-3 0 Std. Dev. 

■ > 3 Std. Dev. 

• Stations 

| | 60 Nautical Mile Limit 

I I Military4_lam 

| | States_outline 

■ Water 

I I Military4_lam 

■ States 
■■ Canada 
□ Mexico 


10 0 10 20 30 40 Kilometers 



RGX Fall 2003 & 2004 Hotspots 



A/ Rivers 
Rgx 

| | 0.5- 1.0 Std. Dev. 

■ 1.0- 1.5 Std. Dev. 

■ 1.5- 2.0 Std. Dev. 

|-1 2.0- 2.5 Std. Dev. 

^2.5- 3.0 Std. Dev. 
^B > 3 Std. Dev. 

• Stations 

| | 60 Nautical Mile Limit 

I I Military4_lam 

| | States_outline 

Bi Water 
^B Military4_lam 

■ States 

■H Canada 
I I Mexico 


10 0 10 20 30 40 Kilometers 


V: 




Copyright 2005 CUROL 


Figure B25. Composite maps indicating spring and fall migratory hotspots as recorded by 
NEXRAD station RGX in Reno, NV. The survey area encompasses the Sierra Army Depot. 





















ERDC/EL TR-12-22 


148 


TWX Spring 2000 & 2001 Hotspots 



A/F 

Twx 0 


' Rivers 
Twx_00 
I-1 0 5 - 

K i o- 

■ 1.5- 

■ 2 . 0 - 
M 2.5- 


- 1.0 Std. Dev. 

- 1.5 Std. Dev. 

- 2.0 Std. Dev. 

- 2.5 Std. Dev. 

- 3.0 Std. Dev. 

> 3 Std. Dev. 

• Stations 

| | 60 Nautical Mile Limit 

I I Military4_lam 

| | States_outline 

■ Water 

I I Military4_lam 

■ States 
■■ Canada 
□ Mexico 


10 0 10 20 30 40 Kilometers 




Q,,, 


;\TTV\0^ 


Copyright 2006 CUROL 


TWX Fall 2003 & 2004 Hotspots 



A/F 

Twx 


f Rivers 

TWX 

| | 0.5- 1.0 Std. Dev. 

1.0- 1.5 Std. Dev. 

■ 1.5- 2.0 Std. Dev. 
□□2.0- 2.5 Std. Dev. 
H 2.5 - 3.0 Std. Dev. 
H > 3 Std. Dev. 

• Stations 

| | 60 Nautical Mile Limit 

I | Military4_lam 

| | States_outline 

■ Water 
Military4_lam 

□ States 
^B Canada 

□ Mexico 


10 0 10 20 30 40 Kilometers 


U P\\, ~ 3 

Ol.VjTVlO^' 


Copyright 2005 CUROL 


Figure B26. Composite maps indicating spring and fall migratory hotspots as recorded by 
NEXRAD station TWX in Topeka, KS. The survey area encompasses Fort Riley Military 

Reservation. 


























ERDC/EL TR-12-22 


149 


TYX Spring 2000 & 2001 Hotspots 



A/ F 

Tyx 0( 


f Rivers 
Yyx v _ 00_01 

r I 0.5- 1.0 Std. Dev. 

1.0 - 1.5 Std. Dev. 
H 15 - 2.0 Std. Dev. 

|-1 2.0- 2.5 Std. Dev. 

H 2.5-3.0 Std. Dev. 

■ > 3 Std. Dev. 

• Stations 

| | 60 Nautical Mile Limit 

I I Military4_lam 

| | States_outline 

■ Water 

I I Military4_lam 

■ States 
■■ Canada 
□ Mexico 


10 0 10 20 30 40 Kilometers 


u * 


Copyright 2006 CUROL 


TYX Fall 2003 & 2004 Hotspots 





Rivers 


| 0.5- 1.0 Std. Dev. 
Hi 10- 1.5 Std. Dev. 

■ 1.5- 2.0 Std. Dev. 
□□2.0- 2.5 Std. Dev. 
H 2.5- 3.0 Std. Dev. 

> 3 Std. Dev. 

• Stations 

| | 60 Nautical Mile Limit 

I | Military4_lam 

| | States_outline 

■ Water 
^B Military4_lam 

□ States 
^B Canada 

□ Mexico 


10 0 10 20 30 40 Kilometers 


u ^ 


Copyright 2005 CUROL 


Figure B27. Composite maps indicating spring and fall migratory hotspots as recorded by 
NEXRAD station TYX in northern NY. The survey area encompasses Fort Drum. 























ERDC/EL TR-12-22 


150 


VBX Spring 2000 & 2001 Hotspots 



A/ Rivers 
Vbx 00_01 


□ 0.5 
H 1.0 


1.0 Std. Dev. 

1.5 Std. Dev. 
H 1-5- 2.0 Std. Dev. 

|-1 2.0- 2.5 Std. Dev. 

^B 2.5- 3.0 Std. Dev. 

■ > 3 Std. Dev. 

• Stations 

| | 60 Nautical Mile Limit 

I I Military4_lam 

| | States_outline 

■ Water 

I I Military4_lam 

■ States 
BB Canada 
□ Mexico 


10 0 10 20 30 40 Kilometers 



Copyright 2006 CUROL 


VBX Fall 2003 & 2004 Hotspots 



A/ Rivers 
Vbx 

| | 0.5- 1.0 Std. Dev. 

■ 1.0- 1.5 Std. Dev. 

■ 1.5- 2.0 Std. Dev. 

|-1 2.0- 2.5 Std. Dev. 

^2.5- 3.0 Std. Dev. 
^B > 3 Std. Dev. 

• Stations 

| | 60 Nautical Mile Limit 

I I Military4_lam 

| | States_outline 

■ Water 
^B Military4_lam 

■ States 

^B Canada 
I I Mexico 


10 0 10 20 30 40 Kilometers 





Copyright 2005 CUROL 


Figure B28. Composite maps indicating spring and fall migratory hotspots as recorded by 
NEXRAD station VBX in western CA. The survey area encompasses Vandenberg Air Force 

Base. 

























ERDC/EL TR-12-22 


151 




YUX Spring 2000 & 2001 Hotspots 




Rivers 
ux 00 01 


| 0.5- 1.0 Std. Dev. 
1.0- 1.5 Std. Dev. 
H 15 - 2.0 Std. Dev. 

|-1 2.0- 2.5 Std. Dev. 

H 2-5-3.0 Std. Dev. 

■ > 3 Std. Dev. 

• Stations 

| | 60 Nautical Mile Limit 

I I Military4_lam 

| | States_outline 

■ Water 

I I Military4_lam 

■ States 
■■ Canada 
□ Mexico 




v 


* 






YUX Fall 2004 Hotspots 


Rivers 

Water 


] 0.5- 1.0 Std. Dev. 
j 1.0- 1.5 Std. Dev. 

| 1.5- 2.0 Std. Dev. 
|2.0-2.5 Std. Dev. 
|2.5-3.0 Std. Dev. 

_| > 3 Std. Dev. 

• Stations 

| | 60 Nautical Mile Limit 

I | Military4_lam 

I I States_outline 

I ] Military4_lam 

■ States 
■I Canada 
I I Mexico 


* UAr/p, 


V 




Figure B29. Composite maps indicating spring and fall migratory hotspots as recorded by 
NEXRAD station YUX in south-western AZ. The survey area encompasses Barry M. 
Goldwater Air Force Range and the Yuma Proving Ground. 




























REPORT DOCUMENTATION PAGE 


Form Approved 
OMB No. 0704-0188 


Public reporting burden for this collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources, gathering and maintaining 
the data needed, and completing and reviewing this collection of information. Send comments regarding this burden estimate or any other aspect of this collection of information, including suggestions for 
reducing this burden to Department of Defense, Washington Headquarters Services, Directorate for Information Operations and Reports (0704-0188), 1215 Jefferson Davis Highway, Suite 1204, Arlington, 
VA 22202-4302. Respondents should be aware that notwithstanding any other provision of law, no person shall be subject to any penalty for failing to comply with a collection of information if it does not 
display a currently valid OMB control number. PLEASE DO NOT RETURN YOUR FORM TO THE ABOVE ADDRESS._ 


1. REPORT DATE (DD-MM-YYYY) 2. REPORT TYPE 3. DATES COVERED (From - To) 

August 2012 F inal 


4. TITLE AND SUBTITLE 5a. CONTRACT NUMBER 


The Identification of Military Installations as Important Migratory Bird Stopover Sites and 
the Development of Bird Migration Forecast Models: A Radar Ornithology Approach: 
SERDP Project SI-1439 


5b. GRANT NUMBER 


5c. PROGRAM ELEMENT NUMBER 


6. AUTHOR(S) 

Richard A. Fischer, Michael P. Guilfoyle, Jonathon Valente, Sidney A. Gauthreaux, Jr., 
Carroll G. Belser, Donald V. Blaricom, John W. Livingston, Emily Cohen, 
and Frank R. Moore 


5d. PROJECT NUMBER 


5e. TASK NUMBER 


5f. WORK UNIT NUMBER 


7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES) 

U.S. Army Engineer Research and Development Center 
Environmental Laboratory 

3909 Halls Ferry Road, Vicksburg, MS 39180-6199; 


8. PERFORMING ORGANIZATION REPORT 
NUMBER 

ERDC/EL TR-12-22 


9. SPONSORING / MONITORING AGENCY NAME(S) AND ADDRESS(ES) 

U.S. Army Corps of Engineers 
Washington, DC 20314-1000 


10. SPONSOR/MONITOR’S ACRONYM(S) 


11. SPONSOR/MONITOR’S REPORT 
NUMBER(S) 


12. DISTRIBUTION / AVAILABILITY STATEMENT 

Approved for public release; distribution is unlimited. 

13. SUPPLEMENTARY NOTES 


14. ABSTRACT 

Military lands and waters may be particularly valuable for migrating birds requiring stopover habitat to rest and refuel en route to very 
distant seasonal ranges. Recent developments in radar technology have provided powerful tools for investigating on a broad scale migrant 
use of military installations; thus providing an opportunity to improve both conservation and flight safety measures. In this study, spring 
and fall migrant use of 40 military installations across the United States were qualitatively investigated. These times of year were selected 
since they are the periods when BASH is of most concern. Migratory patterns on three installations (Eglin Air Force Base, FL; Ft. Polk, 

LA; and Yuma Proving Ground, AZ) were then closely examined and migration forecast models for those locations were developed with 
the goal of providing a tool for reducing the probability of collisions between birds and military aircraft. A comparison was also made 
between radar estimates of migrant densities aloft during exodus events and more traditional ground-based surveys to evaluate the 
effectiveness of estimating migrant abundance in stopover habitat with radar data. At Fort Polk, movement ecology and migrant-habitat 
relations of the Red-eyed Vireo were investigated during migratory stopover. Lastly, migrant use of diverse riparian habitats was compared 
along water courses near the Yuma Proving Ground. Results indicated that approximately half of the installations examined with radar data 
contained migrant stopover “hotspots,” reaffirming the fact that military installations are important to migrating birds. Interestingly, migrant 
abundances, and species turnover as estimated by ground-based surveys, were found to be poorly reflected migrant densities estimated with 
radar data. Migrant abundance, species richness, and community composition were all also found to be influenced by riparian vegetation 
composition. This information collectively suggested that radar data can be used to identify migratory hotspots on military installations and 
improve flight safety on installations with an aviation mission. However, radar data may not be sufficient to distinguish fine-scale 
differences in habitat use by migrants within an installation’s boundaries. 


15. SUBJECT TERMS 

BASH 

Migrating birds 


Stopover hotspots 

Radar estimates of ground densities 
Radar ornithology 

Bird migration forecast models 

16. SECURITY CLASSIFICATION OF: 

17. LIMITATION 
OF ABSTRACT 

18. NUMBER 
OF PAGES 

19a. NAME OF RESPONSIBLE 

PERSON 

a. REPORT 

UNCLASSIFIED 

b. ABSTRACT 

UNCLASSIFIED 

c. THIS PAGE 

UNCLASSIFIED 


170 

19b. TELEPHONE NUMBER (include 
area code) 


Standard Form 298 (Rev. 8-98) 

Prescribed by ANSI Std. 239.18