DU Mobile Apps
World Leader in Wetlands Conservation

Methods - Ducks Unlimited Gadwall Study

SIGN IN    SAVE TO MY DU    PRINT    AAA

Survival and Habitat Use of Female Gadwalls Along the Gulf Coast

 

Methods

Researchers captured gadwalls in Louisiana’s coastal marsh zone within the Gulf Coast Chenier Plain during winters of 2007-08 and 2008-09. They implanted 38g satellite transmitters (PTT-100 Microwave Telemetry, Columbia, MD) into 60 female gadwalls (2007-08 n=14; 2008-09 n=46) ranging from 750-1015g. PTTs were programmed to run on a continuous duty cycle of 6 hours on and 32 hours off with an expected transmitter life of seven months.

Detailed methods as described by researcher Jacob Gray.



Trapping:

I captured gadwalls with rocket nets fired by remote detonators from portable platforms. These methods have been effective in capturing other waterfowl in coastal marsh habitats. I selected capture sites on areas where I had permission to trap, based on the distribution of gadwall after their arrival to the wintering area.

I used unbaited and baited rocket net sites to capture gadwall. I baited some trap sites with a mixture of barley, corn, milo, rice, sweet potatoes, wheat, and oyster shells. I selected unbaited trap sites in smaller, more secluded areas where concentrations of gadwall were located. I utilized existing marsh vegetation at unbaited sites to hide rocket nets and then waited for gadwall to get into the throw of the net without the use of any attractant or hazing.

During the two winter study, I established trap sites in southwest Louisiana on Rockefeller Refuge, White Lake Conservation Area, and Cameron Prairie National Wildlife Refuge. Capture dates were 12 December 2007 through 9 January 2008 during the first winter and 25 October 2008 through 19 December 2008 during the second winter.

Marking:

I implanted 38g PTTs (Microwave Telemetry; Columbia, MD) with external whip antennas in 60 female gadwalls as described in Korschgen et al. (1996). I marked 14 female gadwalls at Rockefeller Refuge during winter 2007-08; 6 females were marked at Rockefeller Refuge and 40 at Cameron Prairie National Wildlife Refuge during winter of 2008-09.

I marked female gadwalls weighing over 750g to reduce potential transmitter effects. I recorded body mass, natural wing cord, total tarsus length, culmen length, and head length on all captured gadwalls.

Tracking

PTTs were programmed to run on a continuous duty cycle of 6 hours on and 32 hours off. This duty cycle allowed for 4-6 transmissions per week and produced a theoretical transmitter life of approximately 7 months (Microwave Telemetry, Inc., Columbia, MD, USA). Transmission periods varied by time of day to allow categorization of nocturnal and diurnal locations. Initial data processing was preformed by Argos and data were sent to me in daily e-mails in both DIAG and DS format. I checked locational accuracy of all locations with the Douglas Argos-Filter Algorithm V7.03. This program compares each location to other locations taken during the same time period, helping to ensure locational accuracy and remove erroneous locations.

I also had data downloaded at seaturtle.org and managed using STAT, a satellite tracking and analysis tool. The STAT program archives and backs-up the data retrieved from Argos. I retrieve the data from seaturtle.org, after being converted into Microsoft® Excel© comma separated values (.csv) files by STAT. I stored all data on my computer hard drive and backed-up the data daily on an external hard drive.

Habitat Use

I used the Douglas-Argos Algorithm to select the highest quality location per transmission for each bird, but did not include any locations that were < location class 2. Locations chosen by the filter were assigned one of five habitat categories; (1) SALT, (2) BRACKISH, (3) INTERMEDIATE, (4) FRESH using the Sasser et al. (2007) marsh classification map and non-marsh locations were combined and categorized as (5) OTHER. I used the sigma-1 error term provided by Argos to calculate an error ellipse with 95% confidence around each PTT location. I determined the 95% confidence ellipses to be 411m and 817m for class 3 and 2 locations respectively. I generated 1000 random points following a bivariate normal distribution within the 95% error ellipse for each estimated PTT location to account for habitat misclassification as result of potential triangulation error.

I obtained few locations from females during the [PRE] and [FIRST] hunt periods. Therefore, I divided each winter into hunted [HUNT] or non-hunted time periods [NOHUNT]. I constructed 4 log-ratios by dividing the proportional use of each habitat by proportional use of OTHER, and used napierian logarithms of these ratios as response variables in our full and reduced models. I replaced zero values with 0.0001 (an order of magnitude smaller than the lowest nonzero habitat recorded for any bird in either hunt period). I then used the transformed proportional habitat use data in split-plot MANOVAs to test for differences in use of habitat types among individual female, female age, winter, hunt period within winter, and all possible interactions. I used variation due to individual female as the error term to test for effects of female age, winter, and their interaction, and residual error to test for effects of individual female, hunt period, and all other interactions. I began with a full model and used backward, step-wise procedures to eliminate non-significant (P > 0.05) terms, beginning with highest order interactions. Once the final model was determined, I compared use of habitats relative to OTHER by testing whether least-square means of log-ratios differed (P < 0.05) from zero (Aebischer et al. 1993a). The habitat use results in this report are for diurnal and nocturnal locations combined. I will provide separate diurnal and nocturnal analyses in the next report.

Refuge Use

I classified all estimated PTT locations within the Chenier Plain as being on (1) areas where hunting was not prohibited for the entire hunting season [HUNTED] or (2) areas where hunting was never permitted [UNHUNTED]. I obtained few locations from females during the [PRE] and [FIRST] hunt periods. Therefore, I divided each winter into hunted [HUNT] or non-hunted time periods [NOHUNT]. I constructed 2 log-ratios by dividing the proportional use of each habitat by proportional use of [UNHUNTED], and used napierian logarithms of these ratios as response variables in our full and reduced models. I replaced zero values with 0.001961 (an order of magnitude smaller than the lowest nonzero area recorded for any bird in either hunt period). I used a mixed model ANOVA to test for differences in use of habitat types among individual female, female age, winter, hunt period within winter, time of day (diurnal or nocturnal) and all possible interactions. I began with a full model and used backward, step-wise procedures to eliminate non-significant (P > 0.05) terms, beginning with highest order interactions. Once the final model was determined, I compared use of areas among levels of significant effects by using least squares means.

Research HomeDucks Unlimited Research TeamDucks Unlimited Research ProjectsDU Research Publications

SIGN IN    SAVE TO MY DU    PRINT    AAA

Free DU Decal

Receive a free DU decal when you signup for our free monthly newsletter.

  DU is a Better Business Bureau Accredited Charity DU Holds a 4-Star Rating with Charity Navigator