LINKING WATERFOWL DISTRIBUTION AND ABUNDANCE TO SPATIAL AND TEMPORAL DISTRIBUTION AND ABUNDANCE OF WETLAND HABITAT _______________________________________ A Thesis Presented to the Faculty of the Graduate School at the University of Missouri-Columbia _______________________________________________________ In Partial Fulfillment of the Requirements for the Degree Master of Science _____________________________________________________ by BRIAN HIDDEN Dr. Lisa Webb, Thesis Supervisor December 2016 The undersigned, appointed by the dean of the Graduate School, have examined the thesis entitled LINKING WATERFOWL DISTRIBUTION AND ABUNDANCE TO SPATIAL AND TEMPORAL DISTRIBUTION AND ABUNDANCE OF WETLAND HABITAT presented by Brian Hidden, a candidate for the degree of Master of Science, and hereby certify that, in their opinion, it is worthy of acceptance. Elisabeth B. Webb, Ph.D. University of Missouri Thesis Advisor Andy Raedeke, Ph.D. University of Missouri Committee Member Joanna Whittier, Ph.D. University of Missouri Committee Member Robert Jacobson, Ph.D. U.S. Geological Survey Committee Member ACKNOWLEDGEMENTS I wish to thank Dr. Lisa Webb for this fantastic opportunity to advance my professional and academic career. Her mentorship and guidance were central to the success of this project, and has enabled me to embark on a professional career in which I am fully prepared. I also thank the other members of my committee, Dr. Raedeke, Dr. Whittier, and Dr. Jacobson. Their vast knowledge in the fields related to my project has contributed tremendously to the outcome of my graduate research. Further, I would like to thank the other faculty and staff at the University of Missouri who have assisted with my project through technical and/or professional support. I would like to thank the Missouri Department of Conservation (MDC) for funding my graduate research project. Not only did the MDC provide the funding necessary for research of this magnitude, but they provided technical support from many of their employees. I would like to thank Phillip Marley and Craig Scroggins for outstanding GIS support. Xiaoming Gaos’ statistical expertise was of immense help with designing the aerial survey portion of my research. I am indebted to Alicia Burke for sacrificing many hours to counting waterfowl with me whether she had motion sickness or not. Finally, I would like to thank the MDC pilots for being exceptional pilots and preventing the aircraft from plummeting back to earth. I am indebted to the private landowners and public wetland managers that allowed access to their wetland units. This project would not have succeeded without their permission and enthusiasm for the progression of wetland and waterfowl science. Specifically, Chris Freeman, Vic Bogosian, Steve Whitson, Wedge Watkins, Clayton Light, Roland Lohmar, Jay McGarraugh, and the Cunningham family not only provided access to their wetlands but also shared their time and knowledge of wetland systems with me. ii Technical aspects of this project were physically demanding and time consuming. Because of this, I could not have completed this project without the assistance of several field and laboratory technicians. I am grateful for the dedication of Natalie Martin, Samuel Newman, Samuel Daugherty, Chelsy Simpson, and Abigail Kaemmerer for working countless hours in the field and lab. I am forever grateful for the support of my loving family. From a young age I observed my mom’s work ethic and integrity and have strived to mirror her character in every way. My step father has been a friend and mentor to me and his friendship has been critical to my success. All my uncles imparted in me a burning affection for the outdoors and instilled in me a lifelong desire to work in the natural resource field. I am extremely thankful to have two of the most perfect children. Their beaming splendor was often my motivation to continue pushing with this project. Finally, I must thank my beautiful wife Natalia. She has remained with me during the worst of times and was always there for support. Without her, I would not be the man I am today. iii TABLE OF CONTENTS ACKNOWLEGEMENTS………………………………………………………………………...ii LIST OF TABLES……………...………………………………………………...……………...vii LIST OF FIGURES……………...……………………………………………………………..…x CHAPTER I: MODELING FACTORS INFLUENCING SPATIAL AND TEMPORAL OCCURRENCE OF WETLAND INUNDATION IN AUTUMN IN THE GRAND AND MISSOURI RIVER ECOREGION, MISSOURI………………………………………………....1 Study Area………………………………………………………………………………...5 Methods…………………………………………………………………………………...7 Sampling Design and Data Collection……………………………………………7 Statistical Analysis……………………………………………………………….10 Results……………………………………………………………………………………12 Discussion..………………………………………………………………………………15 Citations …………………………………………………………………………………20 CHAPTER II: DESIGN AND EVALUATION OF AN AERIAL STRIP-TRANSECT SURVEY TO ASSESS WATERFOWL ABUNDANCE AND DISTRIBUTION IN THE GRAND AND MISSOURI RIVER ECOREGION, MISSOURI……….………………………38 Study Area……………………………………………………………………………….43 Methods………………………………………………………………………………….44 Survey Design……………………………………………………………………44 iv Sampling Methods………………………………………………………………..45 Estimation and Analysis………………………………………………………….47 Results……………………………………………………………………………………50 Discussion………………………………………………………………………………..53 Management Implications………………………………………………………………..56 Citations …………………………………………………………………………………58 CHAPTER III: EFFECTS OF WETLAND HABITAT AVAILABILITY ON AUTUMN DABBLING DUCK ABUNDANCE ON PUBLIC WETLAND AREAS IN THE GRAND AND MISSOURI RIVER ECOREGION, MISSOURI…...…………………………………………...69 Study Area……………………………………………………………………………….72 Methods…………………………………………………………………………………..73 Seed Sampling……………………………………………………………………73 Influence of habitat availability on public wetland area duck abundance………76 Results……………………………………………………………………………………79 Discussion………………………………………………………………………………..81 Management Implications………………………………………………………………..86 Citations………………………………………………………………………………….87 APPENDIX A. Landsat 5 Thematic Mapper satellite images from November 2004-2006 and 2007- 2008 were used to identify inundated wetland conditions. Missouri Department of Conservation (MDC) and U.S. Fish and Wildlife Service (USFWS) public wetland areas were used as training sites to perform a supervised classification in Earth v Resources Data Analysis System Imagine. I identified known land cover pixels (i.e., inundated wetland, cropland, forest) and created specific land classes that could be converted to a raster dataset to determine National Wetland Inventory wetland basin inundation..........................................................……………………………………104 B. Corrected waterfowl observations from an aerial strip-transect waterfowl survey flown in the Grand and Missouri River Ecoregion in north Missouri autumn and winter 2014 and 2015-2016. Observation were corrected for observer bias using a two sample capture-recapture Petersen estimator………………………………………108 vi LIST OF TABLES Table 1.1. Descriptions of variables used to model wetland inundation in the Grand and Missouri River ecoregion during autumn 2004-2006, and 2008-2010…………………26 Table 1.2. Pearson correlation test identifying highly correlated predictor variables. Data were pooled among strata to assess overall correlation of variables for the entire Grand and Missouri River Ecosystem (GMRE)………………………………27 Table 1.3. Annual percent National Wetland Inventory (NWI) and Wetland Reserve Program (WRP) wetlands inundated in the Grand and Missouri River ecoregion, Missouri in November 2004-2006 and 2008-2010. Inundation status was determined using Landsat 5 Thematic Mapper (TM) imagery…………………………………………28 Table 1.4. Top three ranked models explaining National Wetland Inventory (NWI) wetland inundation for three strata in the Grand and Missouri River Ecoregion. Models were ranked using Akaike Information Criterion for small sample size (AIC for the effects of c) variables on inundation of NWI wetlands (model terms: Annual = annual precipitation, October = October precipitation, Ha = hectares, Cult = proportion land cover cultivated in associated watershed, Hyd = hydric soils)………………………………….29 Table 1.5. Mean and range of inundation probability, optimal predictive threshold and misclassification frequency for National Wetland Inventory wetland inundation predictions in November for the Grand and Missouri River Ecoregion, Missouri 2004- 2006 and 2008-2010. Inundation predictions were developed using generalized linear mixed models and ranked using AIC . A receiver operator characteristics (ROC) curve c was used to identify an optimal cutoff value for wetland inundation predictions……….30 Table 1.6. Top three ranked Landsat wetland inundation models for each strata. Models were ranked using Akaike Information Criterion for small sample size (AIC for the effects of c) variables on inundation of NWI wetlands (model terms: Cult = proportion land cover cultivated in associated watershed, Hyd = hydric soils)…………………………………31 Table 1.7. Mean and range of inundation probability, optimal predictive threshold and misclassification frequency for wetland inundation predictions for wetland basins developed using Landsat 5 Thematic Mapper scenes in November for the Grand and Missouri River Ecoregion, Missouri 2004-2006 and 2008-2010. Inundation predictions were developed using generalized linear mixed models and ranked using AIC . A c receiver operator characteristics (ROC) curve was used to identify an optimal cutoff value for wetland inundation predictions………………………………………………..32 Table 1.8. Percent area of National Wetland Inventory (NWI) wetland, Wetland Reserve Program (WRP) wetland and NWI, WRP area combined in the Grand and Missouri River Ecoregion, Missouri identified as inundated based on Landsat 5 TM aerial imagery. Landsat 5 TM aerial imagery was collection for years 2004-2010 though years 2005 and 2007 were not included because spatial coverage of the study area was obscured by cloud cover……………………………………………………………………………………33 Table 2.1. Indices, standard error, and mean coefficient of variation (CV) of abundance estimates after bootstrapping estimates for an aerial waterfowl survey autumn/winter 2014-2016 in vii the Grand and Missouri River Ecoregion, Missouri. I analyzed abundance estimates by Canada Goose (CAGO), all waterfowl species combined (Waterfowl Combined), and all duck species combined (Duck)………………………………………………….61 Table 2.2. Mean and standard error of waterfowl density (ducks/ha) estimates for an aerial strip transect survey performed autumn 2014 and autumn and winter 2015-2016 in the Grand and Missouri River Ecoregion, Missouri…………………………………………..62 Table 2.3. Comparison of mean waterfowl densities (birds/ha) over three sample periods in 2014 and six sample periods in 2015-2016 for an aerial waterfowl survey in the Grand and Missouri River Ecoregion, Missouri. Aerial survey sampling effort was proportionally allocated to strata for all waterfowl surveys…………………………………………..63 Table 2.4. Comparison of public wetland area waterfowl ground estimates to aerial waterfowl survey estimates in the Grand and Missouri River Ecoregion, Missouri. Public wetland area estimates and aerial survey estimates were conducted on the same day or within two days of each other……………………………………………………………………….64 Table 3.1. Variables used to develop models explaining waterfowl abundance on public managed wetland areas in the Grand and Missouri River Ecoregion, Missouri during autumn and winter 2004-2006, and 2008-2010…………………………………………………..92 Table 3.2. Pearson correlation test identifying highly correlated predictor variables used to develop models explaining waterfowl abundance on public managed wetland areas in the Grand and Missouri River Ecoregion, Missouri during autumn and winter 2004-2006, and 2008-2010………………………………………………………………………93 Table 3.3. Frequency (percent samples a species was detected) of plant species collected from seed head clippings and soil core samples after plants had gone dormant in October. Samples were collected in the Grand and Missouri River Ecoregion, Missouri to evaluate potential waterfowl food abundance……………………………………………….94 Table 3.4. Number of sites sampled, number of samples collected, and mean (± SE) total seed biomass (kg/ha) collected from public wetland area and Wetland Reserve Program (WRP) wetlands in the Grand and Missouri River Ecoregion, Missouri. Combined seed head clippings and soil core samples were collected in October 2014 and 2015 to estimate total seed biomass…………………………………………………………………..95 Table 3.5. Mean and standard error (SE) peak public wetland area dabbling duck abundance estimates in the Grand and Missouri River Ecoregion, Missouri during hunting season (HS) and post-hunting season (PS). Weekly ground count estimates were collected by public wetland area managers from the onset of autumn migration in October to January 31st 2004-2016 and 2008-2010……………………………………………………….96 Table 3.6. Mean and Standard error estimates of number of National Wetland Inventory (NWI), NWI area, number of Wetland Reserve Program (WRP), and WRP area within 30 km of a public wetland area in the Grand and Missouri River Ecoregion, Missouri. All NWI and WRP data was collected from the Grand and Missouri River floodplains because regional viii aerial survey estimates indicate >90% of dabbling ducks occur within the floodplains in autumn and winter. NWI estimates were based on probability of inundation from modeling in chapter 1 and WRP estimates were collected from the Natural Resource Conservation Service……………………………………………………………….97 Table 3.7. Top three ranked models explaining peak dabbling duck abundance estimates on public wetland areas during hunting season (DS) and post-hunting season (PS) in the Grand and Missouri River Ecoregion, Missouri. Models were ranked using Akaike Information Criterion for small sample size (AIC for the effects of variables on c) probability of peak duck abundance (model terms: N_NWI = estimated number of available National Wetland Inventory (NWI) wetlands, A_NWI = estimated area of NWI wetlands available, N_WRP = estimated number of Wetland Reserve Program wetlands available, and O_Date = ordinal date of peak waterfowl abundance on public wetland areas…………………………………………………………………………………..98 ix