Ecological Niche Modeling of Pteronotropis Hubbsi, the Bluehead Shiner
Total Page:16
File Type:pdf, Size:1020Kb
University of Texas at Tyler Scholar Works at UT Tyler Biology Theses Biology Fall 12-1-2015 Ecological Niche Modeling of Pteronotropis Hubbsi, the Bluehead Shiner: Evaluating the Effects of Spatial Filtering and Maxent Features Across Various Spacial Extents Justin Matthew eH rnandez Follow this and additional works at: https://scholarworks.uttyler.edu/biology_grad Part of the Biology Commons Recommended Citation Hernandez, Justin Matthew, "Ecological Niche Modeling of Pteronotropis Hubbsi, the Bluehead Shiner: Evaluating the Effects of Spatial Filtering and Maxent Features Across Various Spacial Extents" (2015). Biology Theses. Paper 30. http://hdl.handle.net/10950/309 This Thesis is brought to you for free and open access by the Biology at Scholar Works at UT Tyler. It has been accepted for inclusion in Biology Theses by an authorized administrator of Scholar Works at UT Tyler. For more information, please contact [email protected]. ECOLOGICAL NICHE MODELING OF PTERONOTROPIS HUBBSI, THE BLUEHEAD SHINER: EVALUATING THE EFFECTS OF SPATIAL FILTERING AND MAXENT FEATURES ACROSS VARIOUS SPATIAL EXTENTS by JUSTIN MATTHEW HERNANDEZ A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science of Biology Department of Biology Lance R. Williams, Ph.D., Committee Chair College of Arts and Sciences The University of Texas at Tyler December 2015 Acknowledgements To the people who have contributed their time and effort in helping me succeed at the University of Texas at Tyler, thank you. First, to Lance Williams, my committee chair, thank you for the opportunity to conduct research and your guidance in approaching this project. I am also grateful for providing me a networking opportunity with Cliff Boucher of Tyler Junior College. Because of your input I have enhanced my lecture and presentation skills. To Joshua Banta, thank you for your advice on Maxent-related questions and offering an opportunity to develop my skills in ecological niche modeling. To Harmony Hawley, thank you for your insight to hydrological concepts. Your engineering-based, interdisciplinary approach expanded my understanding of hydrology and its relationship to aquatic ecosystems. To Marsha Williams, thank you for your advice on ArcGIS-related questions and your feedback on my thesis. Of course, I would not have felt welcome without my fellow graduate students Larrimy Brown, Brianna Cierra, Kayla Key, and Melody Sain. Your time and support were invaluable to my experience at UT Tyler. Thank you, Sam Cline and Brenda Arras for being a part of our team in surveying for the bluehead shiner. I would also like to thank my former colleagues at the San Antonio River Authority: Ryan Burke, Larry Larralde, Karen Sablan, Jeanette Hernandez, Maru Gararyar, Katie Peché and Shannon Tollison. You all have made a great impact on my choice in continuing my education in aquatic biology. To John Rofolo, I also thank you for your time in explaining the NHDPlusV2 attribute and vector data. Lastly, to my wife, Amy Hernandez, thank you for being the catalyst in my applying to graduate school. Your support and motivation has pushed me beyond my comfort zone and into a realm of greater lifetime opportunities. Thank you, -Justin Hernandez Table of Contents List of Tables………………………………………………………………………………..……iii List of Figures……………………………………………………………………………...……..iv Abstract…………………………………………………………………………………..……….vi Chapter One: Introduction…………………………………………………………………..…….1 Chapter Two: Materials and Methods……………………………………………………...……..5 Study Area……………………………………………………………………….………..5 Occurrence Data…………………………….…………………………………….………7 Spatial Filtering…………………………………………………………….…….10 Environmental Variables………………………………………………………….……..11 Over-parameterization and Data Reduction………………………………..….…13 ArcGIS……………………………………………………………………………….…..14 Maxent……………………………………………………………………………..….…15 Constant and Adjusted Parameters………………………………………...…….16 Experimental Design……………………………………………………………….……17 Unfiltered vs. Filtered Extents…………………………………………………..17 Tuning Experiments……………………………………………………………..19 Quantitative Assessments……………………………………………………….20 Qualitative Assessments…………………………………………………….…..23 Jackknife Test of Variable Importance………………….…………….……23 Response Curves and Lambdas Files……………………………………….23 Chapter Three: Results..…………………………………………………………………….……25 Unfiltered vs. Filtered Datasets…………………………………………………..……..25 Tuning Experiments………………………………………………………..……………30 Quantitative Assessments………………………………………………………..30 Qualitative Assessments…………………………………………………………39 Jackknife Test of Variable Importance and Physical Habitat.……………...………..…43 Chapter Four: Discussion and Conclusion……….………………………………………………49 Discussion…..……………………………………………………………………………49 Conclusion.………………………………………………………………………………53 Literature Cited...…………………………………………………………………….…………..55 Appendix A: Species Occurrence Data……..……………………………………………………………..59 i Appendix B: Table of Environmental Variables Used Per Study Extent……..…………………………..62 Appendix C: Unfiltered vs. Filtered Occurrence Dataset Assessments, Their Respective Maxent Raw Output Maps, and Response Curves…...……………………………..…………………………….. 63 Appendix D: Threshold-dependent Assessments……………………..………………………………..…93 Appendix E: Threshold-independent Assessments……………………..…………………………………98 Appendix F: Jackknife Test of Variable Importance…………………………………………………….103 Appendix G: Best Performing Ecological Niche Models and Response Curves of the Seven Extents…111 Appendix H: Categorical Layer Attribute Tables…………...…………………………….……………..126 ii List of Tables Table 2.1: Original environmental layers incorporated in the full extent ENM analysis…. …………… .14 Table 2.2: Sample number for each unfiltered and filtered dataset……………………………………….19 Table 3.1: Summary of quantitative assessment results…………………………………………………..31 iii List of Figures Figure 2.1: River basin diagram of the seven extents…………………………………………….……...…6 Figure 2.2: Bluehead Shiner distribution by watershed…………….……………………………….……...9 Figure 2.3: Comparison of unfiltered and filtered occurrence records in geographic space……………...18 Figure 2.4: Diagram of the tuning experiments………………………………………………………...…20 Figure 3.1: Results of the seven extents’ unfiltered and filtered datasets regarding evaluation AUCs…...26 Figure 3.2: Results of the seven extents’ unfiltered and filtered datasets regarding calibration AUC minus evaluation AUC (AUCdiff)…..…...…………………………………………………………...…................26 Figure: 3.3: Results of the seven extents’ unfiltered and filtered datasets regarding omission rates…..…27 Figure 3.4: Logistic output of the unfiltered TX occurrence dataset (n = 18)………………………….…28 Figure 3.5: Logistic output of the filtered TX occurrence dataset (n = 10)…………………………….…29 Figure 3.6: Descriptive comparisons of linear and hinge features for the study’s full extent………….…32 Figure 3.7: Descriptive comparisons of linear and hinge features for the TX/OK half extent.…………...33 Figure 3.8: Descriptive comparisons of linear and hinge features for the AR/LA half extent……………34 Figure 3.9: Descriptive comparisons of linear and hinge features for the TX quarter extent………….…35 Figure 3.10: Descriptive comparisons of linear and hinge features for the OK quarter extent…………...36 Figure 3.11: Descriptive comparisons of linear and hinge features for the AR/N.LA quarter extent……37 Figure 3.12: Descriptive comparisons of linear and hinge models for the LA quarter extent……………38 Figure 3.13: Maxent models depicting relative habitat suitability of the bluehead shiner (Pteronotropis hubbsi) across the TX quarter extent…………………………………………………..………………….41 Figure 3.14: Linear and Hinge feature comparison of the Texas quarter extent at 2x regularization multiplier…………………………………………………………………………………………………..42 Figure 3.15: The Full Extent’s Average response curves (red) of suitable habitat predictions from the top four contributing environmental variables with standard deviation values (blue)………………………...43 Figure 3.16: The TX/OK half extent’s average response curves (red) of suitable habitat predictions from the top four contributing environmental variables with standard deviation values (blue)………………...44 iv List of Figures (continued) Figure 3.17: The AR/LA half extent’s average response curves (red) of suitable habitat predictions from the top four contributing environmental variables with standard deviation values (blue)………………...45 Figure 3.18: The TX quarter extent’s average response curves (red) of suitable habitat predictions from the top four contributing environmental variables. with standard deviation values (blue)………………..46 Figure 3.19: The OK quarter extent’s average response curves (red) of suitable habitat predictions from the top four contributing environmental variables. with standard deviation values (blue)…………….….47 Figure 3.20: The AR/N.LA quarter extent’s average response curve (red) of suitable habitat predictions from the only contributing environmental variable with standard deviation values (blue)……………….48 Figure 3.21: The LA quarter extent’s average response curves (red) of suitable habitat predictions from the top four contributing environmental variables with standard deviation values (blue)…………..….…48 v Abstract ECOLOGICAL NICHE MODELING OF PTERONOTROPIS HUBBSI, THE BLUEHEAD SHINER: EVALUATING THE EFFECTS OF SPATIAL FILTERING AND MAXENT FEATURES ACROSS VARIOUS SPATIAL EXTENTS Justin Matthew Hernandez Thesis Chair: Lance R. Williams, Ph.D. The University of Texas at Tyler December 2015 Ecological niche modeling (ENM) has been extensively applied as a reliable