UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Developing conservation action for data-poor species using seahorses as a case study Aylesworth, Lindsay 2016

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2016_september_aylesworth_lindsay.pdf [ 30.61MB ]
Metadata
JSON: 24-1.0308076.json
JSON-LD: 24-1.0308076-ld.json
RDF/XML (Pretty): 24-1.0308076-rdf.xml
RDF/JSON: 24-1.0308076-rdf.json
Turtle: 24-1.0308076-turtle.txt
N-Triples: 24-1.0308076-rdf-ntriples.txt
Original Record: 24-1.0308076-source.json
Full Text
24-1.0308076-fulltext.txt
Citation
24-1.0308076.ris

Full Text

DEVELOPING CONSERVATION ACTION FOR DATA-POOR SPECIES USING SEAHORSES AS A CASE STUDY  by Lindsay Aylesworth  B.S.F.S., Georgetown University, 2006 M.E.M., Duke University, 2009  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Zoology)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)   August 2016  © Lindsay Aylesworth, 2016 ii  Abstract In this thesis I explore how to develop conservation action strategically for data-poor marine fishes.  The dearth of information about populations, habitats and threats for many marine fishes makes it difficult to know how or where to initiate conservation strategies. My PhD research explores what type of information is essential for conservation management, and how it can be generated and applied for data-poor marine fishes. I use the case study of seahorses (Hippocampus spp), because they are notoriously understudied and yet their trade is regulated under the Convention on International Trade in Endangered Species (CITES).  I further focus on Thailand, the largest exporter of seahorses, which has come under considerable international scrutiny.   In my first two chapters I generated the spatial data that are vital to support conservation and management efforts. My results showed that using local knowledge to inform a presence / absence study, one that incorporated detection probabilities, was the most expedient way to produce the necessary spatial data.   In my next two chapters I explored two approaches to understanding incidental capture of data-poor species in non-selective fishing gear. I found that vulnerability analysis yielded greater return on fewer data than data-poor fisheries stock assessment. However, data-poor fishery stock assessment made it possible to estimate stock status and revise management measures.   For my fifth chapter, I applied findings from my previous chapters to meet CITES obligations, by assuming the role of a Thai government agent confronted with the external technical advice iii  that I had generated.  I found that implementation was most successful if I addressed three main questions: (1) What are the pressures on species?; (2) Is management in place to mitigate those pressures?; and (3) Are the species responding as hoped to management?    My thesis highlights ways that management can move forward with limited data to address conservation issues for marine species. Some of these ways include valuing the use of local knowledge and using new advances in data-poor assessment methods in fisheries. Whenever fisheries are involved, conservationists need to respect the challenges that managers face in simultaneously seeking to protect wild species and meet human needs. iv  Preface This thesis represents my own work, some of which has been published elsewhere. One chapter in this thesis has been accepted for publication in a peer-reviewed journal (Chapter 3) and another is in review (Chapter 2). Three chapters are being prepared for submission (Chapters 4, 5, 6) at peer-reviewed journals. I am (or will be) the lead author on all five papers. I was primarily responsible for conceptualization, experimental design, collecting information (with help from research assistants), data management, data analysis, and writing in each of the manuscripts. Dr. Vincent had a central role in facilitating research in Thailand, and conceptualizing the thesis and ideas for each chapter. Both Dr. Vincent and Dr. Sarah Foster (also of Project Seahorse) played critical roles in securing funding for my research. Dr. Foster generously contributed feedback on chapter ideas and provided logistic support for working in Thailand. My co-authors have made significant contributions and improved the manuscripts substantially. I list my co-authors and outline their contributions to each specific chapter below.   A version of Chapter 2 has been submitted for peer review as “Aylesworth L, Loh T-L, Rongrongmuang W, Vincent ACJ Approaches to locating cryptic and data-poor marine fishes in a conservation context.” Dr. Loh and Wansiri Rongrongmuang assisted with experimental design, implementation of surveys and provided edits to the manuscript. Dr. Vincent gave feedback on experimental design and provided edits throughout the analysis and writing of the manuscript.  A version of Chapter 3 has been accepted for publication at Biodiversity & Conservation as “Aylesworth L, Phoonsawat R, Suvanachai P, Vincent ACJ. Generating spatial data for marine v  conservation and management” BIOC-S-15-01194.  Ms. Phoonsawat collated the research trawl dataset from the Thailand Department of Fisheries and provided edits to the manuscript. Dr. Suvanachai shared Thai geospatial data that assisted with editing fisher spatial data and provided edits to the manuscript. Dr. Vincent gave feedback on the design of scientific surveys and fisher interviews and provided edits throughout the analysis and writing.  A version of Chapter 4 is in preparation for submission as “Aylesworth L, Phoonsawat R, Vincent ACJ. Effects of indiscriminate fisheries (commercial and small-scale) on small data-poor species”.  Ms. Phoonsawat provided suggestions on fisher interview locations and dominant fishing gear types in Thailand, shared relevant Thai fishing literature, and provided edits to the manuscript. Dr. Vincent gave feedback on the design of fisher interviews and provided edits throughout the analysis and writing.  A version of Chapter 5 is in preparation for submission as “Aylesworth L, Phoonsawat R, Walters CJ, Vincent ACJ. Developing a data-poor species stock assessment using seahorses as a case study”.  Ms. Phoonsawat shared relevant Thai fishing literature, supported fisher interview research with recommended fishing port locations, recommended appropriate management scenarios and provided edits to the manuscript. Dr. Walters created the general model structure for the stock assessment, supported model scenario analyses, and created the model variations for the two hypotheses relating habitat and recruitment as causes for reported seahorse declines. Dr. Vincent gave feedback on the design of fisher interviews and provided manuscript edits throughout the writing process.  vi  A version of Chapter 6 is in preparation for submission as “Aylesworth L, Foster SJ, Vincent ACJ. Taking a dose of our own medicine: implementing conservation policy for marine fishes”. Dr. Foster and Dr. Vincent provided significant input into this chapter with the creation of the CITES NDF framework for seahorses and with brainstorming the storyline of this research. Dr. Foster provided critical review of the risk evaluation of seahorses. Both Dr. Foster and Dr. Vincent contributed key insights into how the research in this chapter contributed to the larger issue implementation of CITES for marine fishes based on their experience. Dr. Foster edited the chapter for content throughout the analyses and writing. Dr. Vincent provided final writing edits for this chapter.   All fieldwork in this dissertation was approved by UBC’s animal care committee (A12-0288) and UBC’s human ethics committee (H12-02731).   vii  Table of Contents  Abstract .......................................................................................................................................... ii!Preface ........................................................................................................................................... iv!Table of Contents ........................................................................................................................ vii!List of Tables .............................................................................................................................. xiv!List of Figures ............................................................................................................................. xvi!List of Symbols ........................................................................................................................... xix!List of Abbreviations ................................................................................................................. xxi!Acknowledgements ................................................................................................................... xxii!Dedication ................................................................................................................................. xxiv!Chapter 1: Introduction ............................................................................................................... 1!1.1! Rationale ............................................................................................................................ 1!1.2! Background ........................................................................................................................ 2!1.3! Case Study ....................................................................................................................... 11!1.4   Context And Collaborators ............................................................................................... 18!1.5! Research Questions .......................................................................................................... 19!1.6! Thesis Outline .................................................................................................................. 19!Chapter 2: Approaches to Locating Cryptic And Data-poor Marine Fishes For Conservation ................................................................................................................................ 23!2.1! Summary .......................................................................................................................... 23!2.2! Introduction ...................................................................................................................... 23!2.3! Materials And Methods .................................................................................................... 27!viii  2.3.1! Data Collection ......................................................................................................... 27!2.3.1.1! Determining Relative Abundance Assuming Detection Probability = 1 ........... 27!2.3.1.2! Comparing Methods Assuming Detection Probability < 1 ................................ 28!2.3.2! Data Analysis ............................................................................................................ 29!2.3.2.1! Determining Relative Abundance Assuming Detection Probability = 1 ........... 29!2.3.2.2! Comparing Methods Assuming Detection Probability < 1 ................................ 30!2.3.2.3! Re-evaluating With Occupancy Models When Detection Probability < 1 ........ 30!2.3.2.4! Sensitivity Analyses And Future Study Design Assuming Detection ............... 31!Probability < 1 ................................................................................................................... 31!2.4! Results .............................................................................................................................. 32!2.4.1! Determining Relative Abundance Assuming Detection Probability = 1 .................. 32!2.4.2! Comparing Methods Assuming Detection Probability < 1 ....................................... 33!2.4.3! Re-evaluating With Occupancy Models Assuming Detection Probability < 1 ........ 34!2.4.3.1! Relative Abundance Dataset (2.3.1.1) ............................................................... 34!2.4.3.2! Comparing Methods Dataset – Belt Transects And Timed Swims (2.3.1.2) ..... 34!2.4.4! Sensitivity Analyses and Future Study Design For When Detection  Probability < 1 ....................................................................................................................... 35!2.5! Discussion ........................................................................................................................ 36!Chapter 3: Generating Spatial Data For Marine Conservation And Management ............. 48!3.1! Summary .......................................................................................................................... 48!3.2! Introduction ...................................................................................................................... 48!3.3! Methods ............................................................................................................................ 52!3.3.1! Data Collection ......................................................................................................... 53!ix  3.3.1.1! Dataset 1: Local Knowledge – Fisher Interviews .............................................. 53!3.3.1.2! Dataset 2: Department of Fisheries Research Trawls ........................................ 54!3.3.1.3! Dataset 3: Scientific Diving Surveys ................................................................. 55!3.3.1.4! Dataset 4: Citizen Science Diver Contributions ................................................ 55!3.3.2! Data Analysis ............................................................................................................ 56!3.3.2.1! Comparison 1: Commercial And Small-scale Fishers ....................................... 57!3.3.2.2! Comparison 2: Trawl Captain Interviews And Department of Fisheries  Research Trawls ................................................................................................................ 57!3.3.2.3! Comparison 3: Small-scale Fisher And Diver (Scientific And Citizen Science) Generated Data .................................................................................................................. 58!3.4! Results .............................................................................................................................. 58!3.4.1! Comparison 1: Commercial And Small-scale Fishers .............................................. 58!3.4.2! Comparison 2: Trawl Captain Interviews And Department of Fisheries (DoF) Research Trawls .................................................................................................................... 59!3.4.3! Comparison 3: Small-scale Fisher And Diver (Scientific And Citizen Science) Generated Data ...................................................................................................................... 60!3.5! Discussion ........................................................................................................................ 60!Chapter 4: Effects Of Indiscriminate Fisheries On Small Data-poor Species ...................... 76!4.1! Summary .......................................................................................................................... 76!4.2! Introduction ...................................................................................................................... 77!4.3! Material And Methods ..................................................................................................... 82!4.3.1! Data Collection ......................................................................................................... 83!4.3.2! Data Analysis ............................................................................................................ 85!x  4.4! Results .............................................................................................................................. 88!4.5! Discussion ........................................................................................................................ 90!Chapter 5: Developing A Data-poor Species Stock Assessment Using Seahorses As A Case Study ........................................................................................................................................... 106!5.1! Summary ........................................................................................................................ 106!5.2! Introduction .................................................................................................................... 107!5.3! Methods .......................................................................................................................... 112!5.3.1! Building A Seahorse Age-structured Population Model ........................................ 112!5.3.2! Reconstructing Fishing Effort And Seahorse Catches ............................................ 114!5.3.3! Finding A Model To Fit The Data .......................................................................... 117!5.3.4! Evaluating Management Options ............................................................................ 118!5.4! Results ............................................................................................................................ 119!5.4.1! Building A Seahorse Age-structured Population Model ........................................ 119!5.4.2! Reconstructing Fishing Effort And Seahorse Catches ............................................ 119!5.4.3! Finding A Model To Fit The Data .......................................................................... 120!5.4.4! Evaluating Management Options ............................................................................ 120!5.5! Discussion ...................................................................................................................... 121!Chapter 6: Taking A Dose Of Our Own Medicine: Implementing Conservation Policy For Marine Fishes ............................................................................................................................ 138!6.1! Summary ........................................................................................................................ 138!6.2! Introduction .................................................................................................................... 139!6.3! Methods .......................................................................................................................... 143!6.4! Results ............................................................................................................................ 146!xi  6.5! Discussion ...................................................................................................................... 151!Chapter 7: Conclusion .............................................................................................................. 194!7.1! Overview ........................................................................................................................ 194!7.2! Research Findings .......................................................................................................... 194!7.2.1! Research Question 1: What Is The Most Efficient Way To Search For Cryptic, Rare Or Data-poor Marine Species In The Field? (Chapter 2) ................................................... 196!7.2.2! Research Question 2: What Is The Most Efficient Way To Generate Spatial Data For A Data-poor Species? (Chapter 3) ...................................................................................... 197!7.2.3! Research Question 3: How Can We Best Discern The Scale Of The Fisheries Problem? (Chapter 4) .......................................................................................................... 199!7.2.4! Research Question 4: How Can We Best Determine What Fisheries Management Responses Might Work Best? (Chapter 5) .......................................................................... 200!7.2.5! Research Question 5: How Can We Get Practical Movement Implementing CITES At A National Level? (Chapter 6) ....................................................................................... 202!7.3! Practical Applications Of My Research ......................................................................... 202!7.4! Broader Implications / Future Directions ...................................................................... 206!Bibliography .............................................................................................................................. 209!Appendices ................................................................................................................................. 250!Appendix A Supporting Material For Chapter 2 .................................................................... 250!A.1! Total Number Of Sites Surveyed For Seahorses By Search Strategy, Thai Coast And Search Method. Due To Logistical Constraints Not All Methods Were Conducted At  Each Site. ............................................................................................................................ 250!xii  A.2! Rapid Assessment Sites On The Andaman And Gulf Coast Of Thailand. The Number Of Timed Swims And Transects Varied By Location But Total Area Covered For The Majority Of Sites Was Between 1000-2000 m2. Searches Were Conducted In Coral Reef, Seagrass, Mangrove And Sandy Bottom Habitat. .............................................................. 251!A.3! Detection Rate (+/- 95% Confidence Intervals) Of Sites With Low And High Abundance Of Seahorses At Sandy Soft Bottom Research sites In Thailand .................... 259!A.4! Estimated Number Of Survey Replicates Per Site Needed To Obtain A 90% Confidence That Zeros Represent True Absence Of Seahorses (Hippocampus spp.) At  The Site. .............................................................................................................................. 260!Appendix B Supporting Material For Chapter 3 ..................................................................... 261!B.1! Defining Characteristics Of Commercial And Small-scale Fishers In Thailand (Lymer et al., 2010) ......................................................................................................................... 261!B.2! Small-scale And Commercial Fisher Interview Locations ....................................... 261!B.3! Sampling Effort By Province Of Fisher Villages And Commercial Ports Along The Andaman Coast ................................................................................................................... 262!B.4! Department of Fisheries Survey Grid For Research Trawls ..................................... 263!B.5! Items Included In The Cost Of Generation For The Four Spatial Seahorse Datasets264!B.6! Editing And Verification Procedures Of Fishers’ Maps ........................................... 264!B.7! Evaluating Effort Discrepancies Between Trawl Fishers And DoF Research  Trawls. ................................................................................................................................ 267!Appendix C Supporting Material For Chapter 4 ..................................................................... 270!C.1! Percent Catch Under Length At Maturity And Sex Ratio By Seahorse Species By Fishing Gear Type. .............................................................................................................. 270!xiii  Appendix D Supporting Material for Chapter 5 ..................................................................... 272!D.1! Life History Indicators For Five Thai Seahorse Species.  All Indicators Taken From Froese and Pauly 2016 Except Length At Reproductive Maturity.  (Lawson et al 2015*). ......................................................................................................... 272!D.2! Simulation: Sensitivity Analysis With Fishing Mortality And Life History  Variables ............................................................................................................................. 272!D.3! Fisher Interviews ....................................................................................................... 273!D.4! Model Variations ....................................................................................................... 276!Appendix E Supporting Material For Chapter 6 ..................................................................... 278!E.1! Risk Assessment For Pressures Facing Seahorse Species. ........................................ 278!E.2! Potential Management Responses And Their Appropriateness For Mitigating Pressures On Seahorse Populations From Fisheries And Habitat Pressures. ..................... 286!E.3! Identifying Spatial Overlap Of Management Measures And Hippocampus Spp. Observations ....................................................................................................................... 292!Appendix F Fisher Interview Questionnaire ........................................................................... 295!Appendix G Acknowledgements Continued .......................................................................... 300! xiv  List of Tables Table 2.1 Mean number of surveys, minutes and area covered before observing a seahorse (Hippocampus spp.) at a research site for belt transect and timed swim search methods. ........... 41!Table 2.2 Results of single season occupancy models exploring the effects of method, time of day and visibility on detection probabilities compared to the simplest model with a constant detection co-variate. ...................................................................................................................... 44!Table 2.3 Costs and savings of three survey designs scenarios (Mackenzie and Royle 2005) to search for seahorses (Hippocampus spp.) based on per diem costs of 2014 field season. The standard design (scenario 1) employs the same number of replicates at all sites while the removal survey stops surveys at a site once a seahorse has been found or until a pre-determined maximum number of replicates has been reached (scenario 2 / scenario 3). ................................................. 45!Table 2.4 The usefulness of search strategy for informing future research questions. ................. 47!Table 3.1 Seahorse species knowledge gathered from 1) fisher interviews, 2) DoF research trawls, 3) scientific diving surveys, and 4) citizen science contributions. .................................... 68!Table 3.2 Sampling effort expended to generate the four spatial datasets on the Andaman coast, Thailand. ....................................................................................................................................... 70!Table 4.1 Life history indicators for five Thai seahorse species (Hippocampus spp).  All indicators taken from Froese and Pauly 2015 (fishbase.org) except for those heights at reproductive maturity estimates indicated with *, which came from Lawson et al 2015. ............ 97!Table 4.2 Productivity and susceptibility attributes and rankings (modified from Patrick et al 2009). ............................................................................................................................................ 98!Table 4.3 Data summary of port sampled seahorses from commercial and small-scale fishing gears in coastal Thailand. ............................................................................................................ 100!xv  Table 4.4 Port sampling results (number of individuals) by species by gear type. .................... 102!Table 4.5 Percent catch under length at maturity and sex ratio by species. ............................... 102!Table 5.1 Seahorse age-structured model variations, variables, and equations representing life history relationships. ................................................................................................................... 128!Table 5.2 Fisher reported changes in catches per haul collectively and by gear type. ............... 132!Table 5.3 Model comparisons of habitat and recruitment for observed historical catches ......... 133!Table 5.4 Evaluation of three management strategies compared to the ‘do nothing’ alternative. Strategies compare 2015 model results to predicted results in 2029. ......................................... 134!Table 5.5 Current Thai management measures and their simulated outcome in changing catch trends and decline rates by 2029. ................................................................................................ 135!Table 6.1 Datasets available to evaluate pressures and risk on H. kuda and H. trimaculatus. ... 168!Table 6.2 Datasets available to evaluate management measures related to seahorse species generally. ..................................................................................................................................... 172!Table 6.3 Evaluation of data availability, risk, management response, enforcement and effectiveness of management for H. trimaculatus in Thailand. .................................................. 175!Table 6.4 Evaluation of data availability, risk, management response, enforcement and effectiveness of management for H. kuda in Thailand. .............................................................. 183!Table 6.5 Summary of spatial overlap of marine management measures and sightings of H. trimaculatus and H. kuda. ........................................................................................................... 191!Table 6.6 Detailed information on fishing gears & management measures specifically as they relate to H. trimaculatus and H. kuda. ........................................................................................ 192! xvi  List of Figures Figure 2.1 Survey efforts on Thailand’s Gulf and Andaman coasts to determine relative abundance or presence absence and detection. Sites visited opportunistically for presence absence and detection are also included. ...................................................................................... 40!Figure 2.2 Estimates of detection probabilities (+/- 95% Confidence Intervals) of search methods (belt transects versus timed swims) based on a) relative abundance (2013) and b) methods comparison (2014) datasets. .......................................................................................................... 42!Figure 2.3 Site detection probability based on the total number of seahorses found per site (n=10) by search method based on 2014 research data. Seahorses had higher detection probabilities when they were more abundant. ................................................................................................... 43!Figure 2.4 Simulation results highlight that the optimal number of sites to visit increases as with increased need for confidence in determining future occupancy of seahorses at sites. ................ 46!Figure 3.1 I made three comparisons among four datasets to evaluate if fisher knowledge was the most cost-efficient to generate seahorse spatial data. ................................................................... 72!Figure 3.2 Commercial (a) and small-scale (b) fisher identified seahorse locations, along the Andaman Coast. A total of 86% (63of 73) of commercial and 96% (116 of 120) of small-scale fishers had spatial knowledge of seahorses. Darker areas are where more fishers identified locations of seahorses. .................................................................................................................. 73!Figure 3.3 Locations where trawl captains and DoF research trawls captured seahorses. Trawl captains covered a larger area with their fishing effort, whereas DoF research trawls covered a smaller area but provided information on distribution of seahorses at the individual species level. ....................................................................................................................................................... 74!xvii  Figure 3.4 Small-scale fisher locations and diver sightings (scientific and citizen science contributions) of seahorses across the Andaman Coast. Diver sightings are reported by species with a) sightings of Hippocampus comes and b) all other seahorse species, ................................ 75!Figure 4.1 Locations of fisher interviews with commercial and small-scale fishers along Thailand’s coast. ......................................................................................................................... 103!Figure 4.2 Vulnerability analysis comparing pressure from fishing and the population’s ability to recover amongst seahorse species observed in port sampling. ................................................... 104!Figure 4.3 Susceptibility scores by fishing gear for the three seahorse species most vulnerable to fisheries in Thailand. ................................................................................................................... 105!Figure 5.1 Commercial (a) and small-scale (b) fishing vessel numbers from 1970-2015 taken from the Department of Fisheries and published literature. ....................................................... 136!Figure 5.2 Observed (black) vs. predicted catches from model variations representing (i) historic & current fishing vessel numbers (grey); (ii) IUU fishing vessel estimates (yellow); (iii) fast habitat recovery & weak recruitment compensation (blue); (iv) slow habitat recovery and strong recruitment compensation (green). ............................................................................................. 137!Figure 6.1 Flow chart describing the Non-Detriment Finding Framework for Seahorses. Section numbers refer to sections of the NDF framework. ..................................................................... 158!Figure 6.2 Locations of H. kuda and H. trimaculatus observations from research trawls, scientific surveys and citizen science contributions in relation to national parks on the Andaman Coast. 159!Figure 6.3 Locations of H. kuda and H. trimaculatus observations from research trawls, scientific surveys and citizen science contributions in relation to national parks on the Southern Gulf Coast. .......................................................................................................................................... 160!xviii  Figure 6.4 Locations of H. kuda and H. trimaculatus observations from research trawls, scientific surveys and citizen science contributions in relation to national parks on the Upper Gulf Coast. ..................................................................................................................................................... 161!Figure 6.5 Locations of H. kuda and H. trimaculatus observations from research trawls, scientific surveys and citizen science contributions in relation to no trawl zones on the Andaman Coast.162!Figure 6.6 Locations of H. kuda and H. trimaculatus observations from research trawls, scientific surveys and citizen science contributions in relation to no trawl zones on the Southern Gulf of Thailand. ..................................................................................................................................... 163!Figure 6.7 Locations of H. kuda and H. trimaculatus observations from research trawls, scientific surveys and citizen science contributions in relation to no trawl zones on the Upper Gulf of Thailand. ..................................................................................................................................... 164! Figure 6.8 Locations of H. kuda and H. trimaculatus observations from research trawls, scientific surveys and citizen science contributions in relation to seasonal closures on the Andaman coast. ........................................................................................................................... 165!Figure 6.9 Locations of H. kuda and H. trimaculatus observations from research trawls, scientific surveys and citizen science contributions in relation to seasonal closures on the Southern Gulf coast. ........................................................................................................................................... 166!Figure 6.10 Locations of H. kuda and H. trimaculatus observations from research trawls, scientific surveys and citizen science contributions in relation to seasonal closures on the Upper Gulf coast. ................................................................................................................................... 167! xix  List of Symbols a- Multiplicative value in the fish length-weight relationship  b- How fish length relates to fish weight B0 – Initial fish biomass at time 0 Bt – Fish biomass at time t Ct,a - The number of individuals of age a, caught during time t E – Fishing effort Eg,t– Fishing effort per gear type, at time t ea – Age-specific fecundity Et – Total egg production Et,a – The number of eggs produced by each cohort Ft – Fishing mortality rate for fully vulnerable individuals (exploitation rate at time T) Ht – Total unfished habitat h – habitat remaining it – Ratio of unreported to reported catch per decade It,g – Additional number of fishing vessels added per year to account for IUU effort j- Habitat specific value at which juvenile survival rate (pre-recruitment) drops to ½ of its ‘normal’ value La – Length at age Linf – Length of a fish if it were to grow indefinitely Lmat- Length of a fish at reproductive maturity Lvul- Length of a fish when it recruits to the fishery M – Natural mortality rate xx  Nt,a – Number of seahorses alive at age a, and time t Nt+1,a+1 - Number of seahorses alive at age a + 1 month, and time t+ 1 month p- How habitat recovery rate depends on the current habitat state PmatL – Proportion of fish at length at maturity qt- Catchability coefficient  qh – Relative catchability coefficient of the habitat itself; relative probability of loss of a sponge / coral when swept over by a single trawl R – Annual recruitment rh – Habitat regrowth / recovery rate S - Survival Va – An age-specific vulnerability to fishing mortality vbK - The rate (1/year) at which Linf is approached Wa – Weight-age relationship α – Recruitment compensation / eggs per recruit at time 0 β- (Recruitment compensation – 1) / (recruitment at time 0 * eggs per recruit at time 0)   xxi  List of Abbreviations CBD – Convention on Biological Diversity CITES – Convention on International Trade in Endangered Species of Wild Fauna and Flora CMS – Convention on Migratory Species CPUE – Catch per unit effort DMCR – Thailand Department of Marine and Coastal Resources DoF – Thailand Department of Fisheries  EU – European Union IUCN – International Union for the Conservation of Nature IUU – Illegal, unregulated, and unreported fishing LEK – Local ecological knowledge MEY – Maximum economic yield MPA – Marine protected area MSY – Maximum sustainable yield NDF – Non detriment finding PSA- Productivity-susceptibility analysis RFMO – Regional fisheries management organization RST – Review of significant trade SSC- Species Survival Commission UBC – University of British Columbia xxii  Acknowledgements My most sincere and heartfelt thanks goes to my PhD supervisor, Dr. Amanda Vincent. This was one of the most challenging and rewarding experiences I have embarked on. Thank you for allowing me to take this journey under your guidance and for helping me to be a better communicator, scientist and conservationist. Thank you to my amazing committee members - Dr. Scott Hinch, Dr. Mary O’Connor and Dr. Anne Salomon - for providing support and encouragement both during and outside of committee meetings helping me focus my thesis and grow as a scholar.   I received financial and in-kind support from many donors including: Ocean Park Conservation Foundation of Hong Kong, Riverbanks Zoo and Garden Conservation Fund, The Explorer’s Club Exploration Fund, the numerous contributors to my SciFund Challenge, Bottom Billion Fieldwork Fund, FBR Capital Investments, John G. Shedd Aquarium, Guylian Chocolates and an anonymous donor. I would like to thank the National Research Council of Thailand (permit no. 0002/1306), and the Thailand Department of Fisheries for their in-kind support.  My research would not have been possible for the wonderful support of numerous colleagues and collaborators in Thailand. I have listed some here, but there is also an Appendix dedicated to my on the ground collaborators because I am only allowed two pages for acknowledgements. A heartfelt thank you to Praulai Nootmorn and Ratanawaree Phoonsawat, whose patience and persistence with an enthusiastic farang led to a wonderful working relationship.   xxiii  I could not have survived in Thailand without the guidance and support of my Thai family members. Top, thank you for your willingness to adventure around Thailand in search for seahorses with me and for your sense of humor. Kaew and Oh, thank you for opening your home, hearts, and kitchen to support your sister. To Isaac, thank you for being my adventure buddy, confidant and for providing much needed laughter to put life in perspective. Uncle John and Marcella, thanks for being my home away from home  Dr. Nathan Bennett, Dr. Sarah Foster and Dr. Tse-Lynn Loh - you have been wonderful mentors throughout my dissertation, thank you for sharing your PhD wisdom with me. Thank you to the students and post-docs who have helped me in innumerable ways: Kat Anderson, Joey Bernhardt, Dr. Iain Caldwell, Mariana Diaz-Gomez, Kyle Gillespie, Sarah Klain, Danika Kleiber, Ting-Chun Kuo, Julia Lawson, Roberto Licandeo, Ravi Maharaj, Dr. Phil Molloy, Shannon Obradovitch, Jennifer Selgrath, Allison Stocks, Dr. Stephanie Tomscha, Tanvi Vaidyanathan, Dr. Beth Volpov and Brianna Wright. Thank you to the Project Seahorse Team, past and present, including Gina Bestbier, Tarah Brachman, Scott Finestone, Riley Pollom and Tyler Steim. Thank you to Dr. Ierece Rosa – who helped make my transition to a PhD with Project Seahorse possible. Thank you to Dr. Daniel Pauly and Dr. Carl Walters who supported me to take crap data and turn it into ‘fertilizer’.  Acknowledgements would not be complete without giving thanks to my wonderful family. Mom and Dad, and the Thompson family, thanks for sharing in the adventure. Heidi and Angela – thanks for listening and reminding me that life should be enjoyed. Danielle (Paris) and Danielle (Colorado) thank you for sharing the laughter, the tears, and for your unwavering support.  xxiv  Dedication  With love and gratitude this thesis is dedicated to my parents Skip and Maryanne. Thank you for encouraging me to follow my passion and explore the underwater world. Your unwavering support gave me the strength to make this thesis possible, even with countries between us. Thank you for being there when I needed you.   1 Chapter 1: Introduction  1.1 Rationale As conservation challenges in the ocean grow (McCauley et al., 2015), we face a practical need to develop conservation action for data-poor species, those with inadequate information on life history, population dynamics or threats. Determining conservation action is challenging enough for species where such information is available (Myers et al., 2007), but natural resource managers find it very difficult when the information is lacking. Most conservation studies inevitably call for ‘more data’ (Hamann et al., 2010; Young and Van Aarde, 2011), often using a lack of data to justify management inaction (Johannes, 1998).  Maintaining the status quo may be appealing when issues are complicated or controversial (Joseph, 1994), where there is high uncertainty about management effectiveness (Boersma and Parrish, 1999) or priorities are placed on species with existing management (Bentley and Stokes, 2009a).  Nonetheless, increasing human pressure on the marine environment (Halpern et al., 2008) has created an urgent need for pragmatic solutions to advance species conservation in the ocean.   This dissertation seeks to develop pragmatic approaches for how to address the challenges related to a data-poor marine fish genus requiring national conservation action. I explore a case study of Thailand’s implementation of the Convention on International Trade of Endangered Species of Flora and Fauna (CITES) for seahorses (Hippocampus spp.). Seahorses are examples of data-poor fishes because half of the species (20 of 41 species) (Lourie et al., in press)  are listed as Data-Deficient on the IUCN Red List of Endangered Species, meaning there are not enough data on distribution, population trends, habitats and threats to evaluate their risk of   2 extinction  (IUCN, 2015), including in Thailand (UNEP-WCMC, 2012). Seahorses are also a pioneer for CITES and marine fishes because they were among the very first marine fishes to be placed on CITES Appendix II (in 2002), and the first to be subject to a review of how the 182 member countries were implementing the Convention (Vincent et al., 2013). Thailand is the world’s largest exporter of seahorses, and was identified by CITES as exporting seahorses without assessments of sustainability. Thailand was asked to provide scientific evidence that their export levels were not harmful to wild populations (called a non-detriment finding – NDF), but with a lack of sufficient data on distribution, life history and threats, it struggled to do so (CITES, 2012a; UNEP-WCMC, 2012). The challenges faced by Thailand in this respect are not unique; implementing an international treaty at the national level for a data-poor group of species is not a quick or simple task. Indeed there is a gap in general about the understanding of how to implement CITES for marine fishes when the ideal data are lacking. This thesis attempts to bridge that gap, highlighting lessons learned from the process of assisting Thailand to understand its seahorse populations – in terms of where they live (Chapter 2), how we generate that knowledge (Chapter 3), the threats they face (Chapter 4 and Chapter 5), and determining how to move forward with CITES implementation (Chapter 6).    1.2 Background In the field of conservation, advice is plenty but action is limited. Many species continue to decline both on land (Bland et al., 2014), and in the ocean (McCauley et al., 2015). There is no prescription to ensure success of species conservation, but scientists offer a great deal of advice to those responsible for taking action (e.g. policy makers, government agencies) (Carwardine et al., 2009; Gullison et al., 2000; Nicholson et al., 2013; O’Donnell et al., 2010a).  Some of this   3 advice consists of long lists of the types of data that are the most important for various conservation goals (Hamann et al., 2010; Mackenzie et al., 2002; Pitcher et al., 2013; Rosser, 2008) while other advice covers frameworks to evaluate conservation effectiveness and facilitate decision making (CITES, 2013a; Garmestani and Benson, 2013; Martin et al., 2009; Salafsky et al., 2002). All such guidance is necessarily generic, as it must apply to many species in numerous countries, each with different situations, limitations and opportunities (CITES, 2013a; Rosser, 2008).  The actual value of such guidance is often unclear in practical terms, not least because budgetary constraints may make it impractical or unfeasible to maximize conservation gains (Bottrill et al., 2008; Carwardine et al., 2009; T. G. Martin et al., 2012). Seldom do the people who give advice have to ‘face the music’ and take action with implementation of species conservation management.    In relation to natural resources, conservation refers to halting, stemming or preventing the loss of biological diversity (Soule 1985; Brussard 1991). Conservation biologists are concerned about population depletions for two main reasons: (a) small populations may lead to species extinction (Brussard, 1991) and (b) declining populations may lead to species extinction (Robinson, 2006).  This heightened concern about extinction risk led to an initial focus on maintaining minimum viable population sizes1 or commercially viable2 populations (Tear et al., 2005). With advances in ecosystem and landscape ecology in the 1980’s, the focus of conservation biology expanded to include maintaining the complex nature of ecosystems, as a way to prevent small or declining                                                 1 Minimum viable population size is defined as the smallest isolated population that would prevent the species from becoming extinct for the next 1000 years despite the foreseeable effects of demographic, environmental and geographic variability and natural catastrophes (Shaffer, 1981) 2 Commercially viable population sizes are those that support commercial exploitation activities such as hunting and fishing (Tear et al., 2005).    4 populations (Meine et al., 2006). Based on this new understanding of the importance of ecosystems, a transition occurred in the 1990s, towards conservation of flagship, umbrella and keystone species, with the idea that supporting these animals or plants would benefit other species in the ecosystem (Simberloff, 1998). At present there is a dual focus in how to advance the goals of conservation biology with both species and ecosystem-based management (Cowling et al., 2004; Meine et al., 2006). Conservation targets include concepts such as recovery goals for threatened and endangered species, national biodiversity protection goals, or maintaining minimum viable populations (Tear et al., 2005). Many of these are now enshrined in national and international law, which was historically most concerned about species but has become increasingly directed at more holistic approaches to conservation.  This thesis focuses on the advancement of species conservation in the context of declining populations, but retains awareness that conservation actions may benefit the ecosystem more broadly.   Increasing global commitments to marine conservation (e.g. Aichi Biodiversity Targets, representative networks of Marine Protected Areas see Wabnitz, Andréfouët, & Muller-Karger, 2010) have created an urgent need for practical advice on how to implement international conservation goals at the national level (Smith et al., 2011). Indeed concern has already surfaced about implementation and subsequent achievement of the Aichi Biodiversity targets (Harrop and Pritchard, 2011; Mace et al., 2010; Stuart and Collen, 2013). The reality of practical implementation is challenging because of the inherent nature of international law (Bragdon, 1992; Victor et al., 1998). Successful achievement of obligations needed for conservation action may be hindered by a country’s lack of financial resources, personnel capacities, political will, or legislative infrastructure (Victor et al., 1998). Bridging the divide in governance gaps and   5 implementation between international, national, and local levels (Braunisch et al., 2012; Hind et al., 2010; Pajaro et al., 2010) is a challenge that leaves in-country managers desperate for the best strategy to marshal all tools available for expediency to meet conservation targets (Carpenter et al., 2005; Yates et al., 2012). There is an urgent need to strategically evaluate the ways by which countries can effectively meet obligations. Consequently, countries are in need of guidance on how to collect ‘useful’ science3 (McNie, 2007) to meet conservation targets.  Conservation for data-poor species is particularly challenging because of the dearth of information on species populations, habitats and threats.  Taking action for species conservation involves making decisions about what habitats to protect and how to regulate extractive activities and pressures (Rodrigues et al., 2006). Without data on where the species lives, the current population trend, and pressures on species and their habitats, creating a management and monitoring plan is a stab in the dark at best (Tear et al., 2005). With limited conservation resources (time, money, personnel), managers need to effectively make decisions that prioritize species conservation action to maximize conservation benefits and meet identified targets (Bottrill et al., 2008; Tear et al., 2005). For data-poor species, there may be high levels of uncertainty or controversy surrounding when to prioritize funding on research and on what aspect (distribution, population trend, threats), compared to evaluating effectiveness of current management measures, or implementing new management altogether (Bottrill et al., 2009). In the face of uncertainty, a lack of data and finite resources, managers may decide to maintain the status quo until more information becomes available, or implement precautionary measures, with                                                 3 McNie 2007 defines ‘useful’ as research that expands alternatives, clarifies choices and enables policy makers to achieve desired outcomes.   6 the hopes of refining it later with more data (adaptive management) (Johannes, 1998; Walters and Holling, 1990).    Addressing the increasing number of threats on data-poor species requires a delicate balance of data-gathering, smart decisions, and taking action. There are different ways to achieve knowledge development. Science, judged to be a professional domain, acquires knowledge through lengthy studies, with large datasets, and conclusions based on statistics to minimize uncertainty (Holmes and Clark, 2008; McNie, 2007). Local knowledge, often rated as trade knowledge, is acquired through human-environment interactions in specific locations by local inhabitants (Byg et al., 2012). Some conservation tools (e.g. IUCN Red List of Threatened Species) allow for the use of both to advance the conservation agenda (Rodrigues et al., 2006), but the perception that local knowledge involves greater bias results in much uncertainty and debate around its use in the decision making process . All forms of information carry uncertainty, but uncertainty in a natural resource management context tends to default into the ‘do-nothing’ option (Johannes, 1998).  Decision analysis tools are available to incorporate uncertainty into the decision-making process with mostly scientific knowledge (Wade, 2000). However it is evident that science is not the only factor in the conservation decision-making process (McClanahan et al., 2008; Rands et al., 2010).  On land much progress has been made to obtain vital information quickly and creatively. To overcome difficulties of acquiring knowledge about data-poor terrestrial species, scientists have used camera traps, scat sampling, and genetic analysis to make population estimates for species in need of conservation (Foran et al., 1997; Kelly et al., 2013; Scotts and Craig, 1988). Studies   7 aimed at estimating species occupancy (presence and absence), given the challenges of finding the species of interest (detection), have increased considerably in the terrestrial ecology and conservation literature over the past 12 years (Hurme et al., 2005; Mackenzie et al., 2002; Mazerolle et al., 2007; Petracca et al., 2013; Robley et al., 2014).  Such studies estimate the likelihood that the species was present at a site, given the data, which typically consists of counts of species that include zeros (Mackenzie et al., 2002). Predictive models have informed us that most data-poor species on land, are under threat (Bland et al., 2014). Taking action through adaptive management – learning by implementing management decisions as experiments to determine what actions do and do not work in the conservation context (Walters and Holling, 1990) – is a potential solution to situations with a lack of data. Indeed, literature identifying key components for initiation and maintenance of adaptive management is growing to support terrestrial conservation objectives (Greig et al., 2013; Westgate et al., 2013).  Insights into taking action for ocean conservation in the face of a lack of data and uncertainty are emerging rather slowly compared with terrestrial conservation. The use of some terrestrial methods is becoming more common in the ocean. Examples include underwater videos (Goetze and Fullwood, 2012), scat sampling (Bowles and Trites, 2013) and genetic analysis (Mobley et al., 2011). Applying adaptive management in the fisheries context has yet to be successful, according to a review of 100 case studies (Walters 2007), with institutional failures and a lack of resources for gathering strategic science identified as the main causes for the lack of success (Walters 2007). With continuing marine species decline (McCauley et al., 2015), managers urgently need new ways to handle data scarcity and facilitate effective decision-making for the ocean.   8  Common to both the land and the ocean are several analytical techniques to handle challenging species datasets. When data are less than ideal, simulation techniques and hypothetical datasets allow explorations of relationships in the data that approximate real-world results but require less time, effort and/or money than other approaches or gathering additional data (Huntsman and Schaaf, 1994; Molloy et al., 2013; Regan et al., 2013). Bayesian analyses or Markov chain Monte-Carlo simulations can be used to incorporate uncertainty in species models such as those identifying habitat relationships or evaluating population dynamics in the face of exploitation (Collier et al., 2012; Punt and Hilborn, 1997). Bootstrapping, a computer intensive re-sampling method, is another common technique, used to gain confidence in the accuracy of sample or parameter estimates (e.g. variance, confidence intervals, prediction error) (Nakagawa and Cuthill, 2007). These analytical tools support conservation biologists to draw conclusions about the natural world despite imperfect datasets and the complicated relationships that exist in nature.  In the marine realm, a major challenge remains with reconciling sustainable fisheries management with conservation (Davies and Baum, 2012; Salomon et al., 2011). Sustainable fisheries management seeks to use technical knowledge to reach specific conservation targets or goals (e.g. maximum sustainable yield (MSY), maximum economic yield (MEY)) and stems from broader goals of trying to achieve maximum productivity without extinguishing populations (Robinson, 2006). Conservation in the context of fisheries is concerned with declining fish populations but recognizes that exploitation activities are also linked to livelihood and food security (Salomon et al., 2011). These two fields share similar goals such as preventing overfishing, recovering depleted stocks or maintaining jobs, and indeed recent research   9 comparing fisheries and conservation metrics about the status of marine fish align well (Davies and Baum, 2012). However, the path forward to achieve sustainable fisheries goals remains contentious and unclear (Davies and Baum, 2012; Salomon et al., 2011), because it involves addressing societal values about sustainable use and whether the priority is placed on ‘use’ (i.e. ensuring that fishing can continue to provide job and food security) or on ‘sustainability’ (i.e. preventing fish declines and biodiversity loss) (Bowman, 2013; Salomon et al., 2011; Tisdell et al., 2007).    The largest challenge in reconciling fisheries and conservation may well be for species that are incidentally caught to a level that threatens wild populations without us knowing much about them.  Typical fisheries management at the very least requires information on catches and fishing effort (Walters and Martell, 2002), which can be hard to come by for data-poor species. A lack of data-collection for non-target species may occur for three main reasons (1) they are not a management priority; (2) they are economically unimportant to the fishery or (3) non-target fish are not retained (Bentley and Stokes, 2009b; Bentley, 2015). A lack of data on non-target species has most likely led to management inaction because recent research indicates that 64% of unassessed fisheries have stock biomasses below fisheries management targets (e.g. MSY) (Costello et al., 2012). Such evidence leads to the conclusion that non-target fish are not commonly valued in terms of sustainability or use. Although non-target fishes and the fisheries that catch them are not a priority, with decreasing abundances of many marine species (both large and small) (McCauley et al., 2015), there is an urgent need to identify fisheries management tools that support the persistence of data-poor species (Johannes, 1998; Walters and Holling, 1990).    10  Management measures at national levels are complemented by regional fisheries management organizations (RFMOs), which are commonly even less effective than national governments at reducing pressure on non-target species.  Many RFMO mandates now include goals of eco-system based management, some of which address fisheries impacts on non-target species (Cullis-Suzuki and Pauly, 2010; Gilman et al., 2014). However, the overall progress of many global RFMO’s to implement best practice in bycatch governance is slow despite some RFMO’s moving faster (e.g. Commission for the Conservation of Antarctic Marine Living Resources) than others (e.g. Regional Commission for Fisheries) (Gilman et al., 2014). For example, three RFMOs do not have any binding bycatch conservation and management measures, four RFMO’s do not have regional observer programmes and five RFMOs do not have data protocols in place for non-target fish (Gilman et al., 2014). Other research highlights discrepancies between how well RFMO’s meet conservation best practice standards on paper compared to in practice, suggesting that RFMOs have prioritized the exploitation of fish stocks over a commitment to conservation (Cullis-Suzuki and Pauly, 2010). With numerous high seas fish stocks in decline (FAO, 2012) there is an urgent need to prioritize issues of sustainability over use.   With questionable effectiveness of RFMOs, several international agreements have shifted their focus to include marine environmental issues (Fox et al., 2012; Vincent et al., 2013), the implementation of which has placed a responsibility on national government agencies to achieve new conservation targets (Davies and Baum, 2012; Salomon et al., 2011).  For example, countries that have signed onto the Convention on Biological Diversity (CBD) have committed to the goal of protecting and effectively managing 10% of the sea with marine protected areas by   11 2020 (CBD, 2010). Of a similar nature, five shark species and all manta rays joined the few marine species regulated by the international wildlife trade treaty, The Convention on International Trade in Endangered Species (CITES) (CITES, 2013b). Implementation of international agreements such as the CBD and CITES lies with national government agencies responsible for ocean management (CBD, 1992; CITES, 1973). Taking action for such treaties involves not only identifying where to place new management areas such as no-take marine protected areas (Boersma and Parrish, 1999; Katsanevakis et al., 2011; Toonen et al., 2013), but also how to ensure such management is effective (Carwardine et al., 2009; DeSanto, 2013).  Such response undoubtedly involves taking action in the face of a lack of data and uncertainty to meet targets and deadlines of implementation, yet government response to new mandates and responsibilities is resoundingly slow (Martin et al., 2012). The result is an urgent need for practical approaches to implement international treaties and meet national conservation commitments  1.3 Case Study In this thesis, I used the case study of Thailand’s implementation of CITES for seahorses (Hippocampus spp.) as a means to examine how to develop pragmatic approaches for conservation of data-poor species.  Seahorses are a marine fish genus that exemplifies the challenges involved with data-poor fishes. With 41 species in the genus Hippocampus (Lourie et al, in press), surprisingly little is known about where seahorses live and what drives distribution patterns across the different species as there are both habitat generalists (e.g. Hippocampus kuda) (Lourie et al., 2004) and specialists   12 (Hippocampus bargibanti) (Reijnen et al., 2011). Seahorses live in both temperate and tropical areas, with recorded presence in virtually every habitat type (e.g. coral reefs, seagrass beds, mangroves, sandy bottoms, rocky reefs etc.) (Lourie et al., 2004). Despite recorded occurrence in many habitats, seahorses are difficult to find in the wild because they camouflage well, have a patchy distribution, and are found in low numbers (Foster and Vincent, 2004). The trade of seahorses is global and complex, involving 25 known species, and more than 80 countries (Foster et al., 2016). However, half (n = 20 of 41) of species in trade are listed as ‘Data-Deficient’ by the IUCN, meaning there is not enough information about their population status, distributions, or threats to evaluate extinction risk at the global level (IUCN, 2015). This is in part because the majority of seahorses are captured as non-target catch in fisheries and fishers catch a low number of seahorses per night, there are few records of catches, effort, or landings, at national levels (but see Baum and Vincent, 2005; Lawson et al., in press; Perry et al., 2010). Records are available for international seahorse trade volumes because seahorses are listed on CITES Appendix II, but such records are full of discrepancies creating uncertainty in their use (Foster et al., 2016). Because of the limited global seahorse knowledge, determining the severity of threats at national levels remain challenging (CITES, 2013a).   The issue of practical approaches to implementation at the national level is particularly relevant for those countries party to the CITES treaty, one of the largest multilateral environmental agreements (Vincent et al., 2013). Created in the 1970s, this Convention seeks to prevent international trade from contributing to the extinction of species (CITES, 1973). Over the last forty years CITES has achieved some success (e.g. improvements in conservation status of taxa such crocodiles and cockatoos (Cahill et al., 2006)), and faced continued difficulties (e.g.   13 emergence of black markets caused by trade bans (Cooney and Jepson, 2012; Lemieux and Clarke, 2009)). For two of the more well known species listed on CITES, rhinos and elephants, the mixed results exemplify the aforementioned successes and difficulties with CITES (Amin et al., 2006; Bennett, 2015; Biggs et al., 2013; Wittemyer et al., 2014).  The effectiveness of CITES has attracted debate (Bowman, 2013), the heart of which comes down to whether conservation or trade is the main priority for the treaty (Bowman, 2013; Vincent et al., 2013). All countries struggle with enforcement and implementation (Heng, 1999; Liljeblad, 2008; Nijman, 2010; Shepherd and Nijman, 2008) in part because of a lack of political will and resources (financial and personnel). Such constraints support the need for a practical approach to achieve implementation at national levels, one that focuses on maintaining momentum with conservation action instead of defaulting to a do-nothing attitude.   CITES is a new tool to secure sustainable exports in marine fishes and, by proxy sustainable fisheries (Vincent et al., 2013). Seahorses are one of the few marine fish taxa regulated by CITES (Vincent et al., 2013). Many seahorse species (n=20 of 41; Lourie et al, in press) are considered Data-Deficient by the IUCN Red List of Threatened Species, meaning there is not enough information on their distribution, population trends, habitats and threats to evaluate their risk of extinction (IUCN, 2015). Because of this, information on seahorse species at national levels, also suffers from a lack of data, and therefore seahorses can be considered data-poor4. Additionally seahorses are an example of small non-target fishes, meaning that commercial and small-scale fishers incidentally capture them, and they are not typically prioritized for data                                                 4 Where there is little information on species distribution, life history or threats, such species are considered data-poor in this thesis.    14 collection by fishers or fisheries departments (Reuter et al., 2010). Thailand is the world’s largest exporter of seahorses (Foster et al., 2016), and as a country that has signed the CITES treaty was asked to provide scientific evidence that export levels of seahorses were sustainable. This thesis supports Thailand with this task.  The strength of CITES, to support sustainable wildlife trade for terrestrial and marine species, lies in its enforcement process. CITES currently regulates the international trade of only very few marine fish species, by listing them in its Appendices (Vincent et al., 2013). Countries that trade in Appendix II species5 must prove that exports do not harm wild populations, called making a non-detriment finding (NDF) (CITES, 1973). Parties must overcome two main challenges associated with this process: (1) uncertainties about trade levels, population status and management options and (2) institutional problems associated with stakeholder involvement, financial support, and lack of capacity (Vincent et al., 2013). Failure to declare the sustainability of exports of a species appropriately can lead CITES to propose an evaluation of a Party’s trade (called a Review of Significant Trade), and eventually to suspend the Party’s right to trade the species in question (e.g. queen conch [Strombus gigas] (CITES, 1999) (UNEP-WCMC, 2012).  The first policy advice for regulating exports for any marine fish came in the form of an NDF framework for seahorses (Hippocampus spp.) (CITES, 2013). Indeed seahorses are a pioneer for CITES and marine fishes in general. They were among the very first marine fishes to be placed                                                 5 Species can be listed on one of three appendices under CITES. Appendix I includes species threatened with extinction that are or may be affected by trade. Appendix II lists species that may become threatened with extinction if trade is not regulated. Appendix III species are taxon of national concern where individual countries ask support from other CITES parties because exploitation is prohibited by national law (CITES, 1973).   15 on CITES Appendix II (in 2002), and the first to be subject to a review of how Parties were implementing the convention, dubbed a Review of Significant Trade (RST) (Vincent et al 2013). As part of this RST process, Thailand and Vietnam were identified as having exports warranting concern in 2012 (CITES, 2012b).  Seahorses thus also became the first marine fish species to have experienced a trade suspension, when CITES closed exports of H. kuda for Vietnam because the latter failed to meet its obligations for this species (CITES, 2013c).    As the world’s largest exporter of seahorses, Thailand has been subject to CITES recommendations (effectively instructions) for four species, with the goal of ensuring that their exports are not harmful to wild populations. Based on analysis of CITES export records, Thailand is responsible for 75% of global trade volumes, exporting approximately 3.0 to 6.5 million seahorses per year (Foster et al., 2016). There are seven species of seahorses that inhabit Thai waters (Lourie et al., 2004), and five of which have been recorded in trade (Laksanawimol et al., 2013; Perry et al., 2010). Four of these five (Hippocampus kelloggi, Hippocampus kuda, Hippocampus spinosissimus, and Hippocampus trimaculatus) dominate global trade volumes (Foster et al., 2016), and Thailand’s large export volumes in these species led to their designation as ‘urgent concern’ by the CITES Animal Committee (CITES, 2014, 2012a).  Thailand was unable to scientifically determine if their export volumes were harming local populations, therefore, the CITES Animal Committee made ten recommendations to Thailand to assist with this process, which the country must implement in order to continue with their export trade in a sustainable manner (CITES, 2012b). Of these ten recommendations, four are particularly relevant to this thesis and include the following: (i) undertake studies to provide evidence on variation of spatial and temporal abundance of seahorses; (ii) implement additional measures   16 such as spatial or temporal restrictions on fishing activities;  (iii) model population response to exploitation pressures to review and revise management measures; and (iv) establish an adaptive management program.   Thailand is one of the world’s largest fishing nations, and with numerous commercial and small-scale fisheries, catching target and non-target species alike, is challenged with the goal of sustainable fisheries management (DoF, 2015). Commercial and small-scale fisheries are defined based on the size of the boat, the type of site where fishers land fish, and the distance typically fished from shore (Lymer et al., 2010), with small-scale fisheries tending towards smaller boats that fish closer to shore. Thai fishers use many different types of fishing gear but common gears for commercial fishers include otter and pair trawls, purse seines, and pushnets, whereas small-scale fishers typically use gillnets and or cages (DoF, 2015). Both commercial and small-scale fishing gears in Thailand have expanded beyond the effort required to achieve maximum sustainable yield for numerous target species (Chuenpagdee and Pauly, 2003; Lymer et al., 2008; Pomeroy, 2012). There have been large declines of catch per unit effort especially in the trawl fisheries, and the bulk of catches consist of non-target species (Chuenpagdee and Pauly, 2003; DoF, 2015)  Despite knowledge of declining resources and over-capacity since the 1980’s, Thailand still struggles with overcapacity challenges today (Christensen, 1998; Chuenpagdee and Pauly, 2003; DoF, 2015; Pauly, 1988, 1986).  Declining resources and overcapacity have created many conflicts between commercial and small-scale fishers over remaining resources (Chuenpagdee and Pauly, 2003; DoF, 2015). Of a similar note are problems for fisher compliance with management measures (e.g. spatial fisheries closures or no-take marine protected areas) along with a lack of confidence in government enforcement capabilities (DoF,   17 2015; Lunn and Dearden, 2006a; Panjarat and Bennett, 2012; Siriprasertchok, 2015).  Thailand’s fisheries management has recently undergone reform based on a yellow card issued to Thailand by the European Commission for not taking sufficient measures against illegal fishing (IUU). Thailand was given an ultimatum, to reform their fisheries management or face a potential ban on import of their fisheries products to EU countries (European Commission, 2015).   The small number of published articles about seahorses in Thailand (Laksanawimol et al., 2013, 2006; Panithanarak et al., 2010a; Perry et al., 2010) provided far too few data to support Thailand to implement CITES effectively. The published articles covered a diverse array of topics related to seahorses including, reproductive physiology (Laksanawimol et al., 2006), genetics (Panithanarak et al., 2010b), trade and fisheries (Laksanawimol et al., 2013; Perry et al., 2010). Additionally the Thailand Department of Fisheries had collected one year of trawl research survey data where the presence and species of seahorses were recorded. While previously published information on trade and fisheries provided evidence of why there should be concern for seahorses in Thailand (e.g. multiple species, multiple gears, local knowledge of declines) (Laksanawimol et al., 2013; Perry et al., 2010), it did not provide scientifically sound evidence to evaluate if export levels were harming wild populations. With a lack of data on seahorse distribution, threats and management, Thailand was not supported to implement CITES effectively.  Thailand’s management of the marine environment is split amongst several government agencies. Thailand’s marine fisheries are managed by the Department of Fisheries (DoF), who has overall responsibility for fisheries monitoring, control and surveillance. The Thai Marine   18 Department is responsible for new vessel registration, vessel permit renewal, change of vessel lists, and captain certificates. Management of the marine environment (e.g. endangered species) and monitoring of coral reefs, mangroves and seagrass beds, falls to the Department of Marine and Coastal Resources (DMCR). The Royal Thai Navy is responsible for maritime transport and security issues, and coordinates enforcement and protection of marine resources. There is inter-agency integration in five areas – national observer programs, port sampling, compliance and enforcement activities, data collection, and management (DoF, 2015). Despite integration efforts, the differing mandates of each government agency make it challenging to ensure consistent management of the Thai marine environment.  1.4   Context And Collaborators At the CITES Secretariat’s request, Thailand agreed to work with Project Seahorse, the IUCN designated global expert group for seahorses and their relatives, to provide guidance on implementation. As a PhD candidate with Project Seahorse and also a member of this IUCN Seahorse, Pipefish and Stickleback Specialist Group for Seahorses, I was in a unique position to join this collaboration to assist Thailand achieve sustainable trade. The Thai CITES Scientific and Management Authorities for marine fishes lie with the DoF. Project Seahorse and the DoF signed a memorandum of understanding to work together to support Thailand to implement the recommendations of the CITES Animals Committee. The DoF supported Project Seahorse to receive a research permit from the National Research Council of Thailand (permit no. 0002/1306). This collaboration provided the context for me to obtain most of the data presented in my thesis.     19 Although my main collaborator was the DoF, I was also supported by other local collaborators in Thailand. Such collaborators include local universities, government agencies, non-governmental organizations, dive shops, boat captains, traders, conservation and community groups. I introduce my collaborators in each thesis chapter, and note their intended contributions to my research.  1.5 Research Questions In this thesis I focus on four main research questions as they relate to data-poor marine fishes. The research questions are as follows:  1) What is the most efficient way to search for rare marine species in the field? (Chapter 2) 2) What is the most efficient way to generate spatial data for a data-poor species? (Chapter 3) 3) How can we best discern the scale of fisheries problems? (Chapter 4) 4) How can we best determine what fisheries management responses might work    best? (Chapter 5) 5) How can we get practical movement implementing CITES at the national level? (Chapter 6)  1.6 Thesis Outline This thesis has five research chapters, followed by a general discussion of the implications for conservation and management of data-poor species.  In Chapter 2, I take the perspective of a marine ecologist to explore the best search strategy and method for a data-poor, rare species underwater. Cryptic data-poor species may not be well served by classic ecological techniques for determining relative abundance because data   20 typically consists of species counts that include many zeros (Fairweather, 1991). One way terrestrial researchers deal with cryptic and data-poor species is to first deploy a presence / absence search strategy among sites that includes the probability of detecting the species if present (Mackenzie et al., 2002). Such a strategy is uncommon in the ocean (Issaris et al., 2012; Katsanevakis et al., 2012), but may represent a solution to determine how much effort to allocate to research when there are substantial and pressing threats for data-poor species. I use my experience searching for seahorses in Thailand to determine if there is a difference between two search strategies - (1) searching to determine relative abundance and (2) searching for presence / absence with detection probabilities - for each of two methods (transects and random swims).  I then evaluated the hypothesis that terrestrial solutions to searching for data-poor or rare species are applicable in the ocean.  In Chapter 3, I compare and contrast four methods used to generate spatial data in the ocean for data-poor species and hypothesize that fishers know best when it comes to identifying areas with rare and depleted fish species. The global conservation crisis demands that managers marshal all available datasets to inform conservation management plans for depleted species, yet the level of trust placed in local knowledge remains uncertain. To evaluate if fishers know best, I compare four methods for inferring species distribution: (i) fisher interviews; (ii) government research trawls, (iii) scientific diving surveys, and (iv) citizen science contributions. With the data generated by these methods I make three comparisons (i) commercial versus small-scale fishers; (ii) commercial trawl fishers versus government research trawls; and (iii) small-scale fishers versus scientific diving surveys and citizen science contributions. I evaluated these comparisons   21 at both the genus and individual species levels to determine conclusions about seahorse spatial occurrence, diversity of species present and the cost effectiveness of sampling effort.  In Chapter 4, I investigate the use of a data-poor fishery assessment method to understand the relative pressures of commercial and small-scale fisheries on seahorses. First, I gathered data by interviewing commercial and small-scale fishers and through port sampling of landed catch. I estimate annual catches of seahorses from all gears in Thailand based on interview data and determine the relative impacts of fisheries on seahorses by species and gear type from port sampling data. Finally, I explore the use of a data-poor fishery assessment method to prioritize management efforts by evaluating the risk posed to each species by gear type.    In Chapter 5, I develop the first seahorse stock assessment for any Hippocampus species and explore its usefulness to revise Thai fisheries management, past and present. First I build an age-structured model incorporating previously published life history information into the model. I explore this simple model to evaluate the sensitivity of life history parameter estimates and determine if fishing mortality has been sustainable over the course of fishing industry growth in Thailand. I then reconstruct fishing effort and seahorse catches from 1970 to 2013 with a combination of data from published literature, DoF, and fisher interviews. Next, I explore four model variations to test the following hypotheses: (1) seahorse catch is best predicted by historic fishing & current fishing effort; (2) seahorse catch is best predicted by incorporating estimates of illegal, unreported and unregulated fishing (IUU); (3) seahorse catch is best predicted by habitat decline with fast recovery and weak recruitment compensation; (4) seahorse catch is best predicted by habitat decline with slow recovery and strong recruitment compensation. I   22 determined these four hypotheses based on a combination of available data (e.g. historic and current fishing vessel numbers), and expert knowledge about Thai fisheries (challenges with IUU) and seahorses (links with habitat). I then determine which model best fits the reconstructed data. Finally, I use the model of best fit to review and revise management strategies by comparing four future management scenarios, and evaluating five current Thai fisheries management efforts.  In Chapter 6, I change perspective and force myself, the provider of conservation advice, to take policy and management action based on that advice by assuming the role of the government of Thailand faced with implementing CITES for seahorses. I run through a CITES-prescribed framework to attempt to determine whether export levels of two seahorse species, H. kuda and H. trimaculatus are harming wild populations. I do so by evaluating the risk to these species from various pressures and determine whether appropriate management is enforced and effective. I then assess whether trade can be allowed, not allowed or allowed with conditions, and what remedial actions should be prioritized for each species in support of sustainable trade.   In Chapter 7, I end with a general discussion of the findings presented in this thesis and how those findings might provide useful input on how to take conservation action for data-poor species.     23 Chapter 2: Approaches to Locating Cryptic And Data-poor Marine Fishes For Conservation  2.1 Summary When seeking to conserve data-poor species, we need to decide how to allocate research effort, especially when threats are substantial and pressing.  My study provides guidance for sampling marine fishes that are particularly difficult to find – those species that are cryptic or rare and or where little information exists on local distribution (data-poor). I used my experience searching for seahorses (Hippocampus spp.) in Thailand to evaluate two search strategies for marine conservation: (1) determining relative abundance and (2) searching for presence / absence with detection probabilities. My fieldwork indicated that using the presence / absence framework was more likely to lead to inferences that seahorses could be found in the site than when using the relative abundance framework. This realization would support a common-sense approach, where presence/absence with detection probabilities is centrally important to marine conservation planning for cryptic and or data-poor marine species.   2.2 Introduction Cryptic data-poor species are not well served by classic ecological techniques for determining relative abundance.  In the absence of baseline information, pilot studies are often conducted (Fairweather, 1991) to determine relative abundance to inform research or to create management plans. But what happens when pilot studies fail to find the species of interest, which may commonly happen for cryptic or depleted species (Durso et al., 2011; Engler et al., 2004)?  Such   24 failures to locate cryptic species may result from limited data, linked to funding or logistic restraints resulting in a type II error (Underwood 1996) or because the species was not detected or is truly not there (Mackenzie et al., 2002). We need to deploy rigorous yet realistic methods to assess conservation status and enact management for data-poor species.  On land, researchers are finding ways to cope with data collection for relative abundance of cryptic, data-poor species. To overcome difficulties of terrestrial sampling, scientists have used camera traps, scat sampling, and genetic analysis to make population estimates for species in need of conservation (Foran et al., 1997; Kelly et al., 2013; Scotts and Craig, 1988). Studies aimed at estimating species occupancy (presence and absence), given the challenges of finding the species of interest (detection), have increased considerably in the terrestrial ecology and conservation literature over the past 12 years (Hurme et al., 2005; Mackenzie et al., 2002; Mazerolle et al., 2007; Petracca et al., 2013; Robley et al., 2014).  Such studies estimate the likelihood that the species was present at a site, given the data, which typically consists of counts of species that include zeros (Mackenzie et al., 2002).  Studies using presence / absence with detection probabilities have generated sound science for management decisions on a range of terrestrial species.  Guidance on the implementation of management measures such as habitat corridors (Petracca et al., 2013), predator control (Robley et al., 2014),  and invasive species remediation (Britton et al., 2011) stem from the results of detection research. Management decisions regarding the evaluation of monitoring programs or procedures (de Solla et al., 2005; van Strien et al., 2010), conservation planning (Hurme et al., 2005; Weller, 2008), and effectiveness of conservation measures (Zipkin et al., 2010) have been   25 supported by studies using presence / absence with detection probabilities. The increasing use of studies using presence / absence with detection probabilities to inform terrestrial management activities may represent a paradigm shift or simply a response to the challenges associated with data-poor species (Mackenzie et al., 2005; McNie, 2007; Underwood and Chapman, 2003).   In the ocean, methods for assessing data-poor species are still in their infancy compared to those established for terrestrial counterparts (Katsanevakis et al., 2012; Mobley et al., 2011). The aquatic environment is more complex because researchers cannot easily see what lies beneath the surface (Monk, 2013), and the technology for sampling and surveying is less efficient and more costly than in terrestrial systems (Katsanevakis et al., 2012). This adds important concerns about estimates of population density, abundance and simple presence / absence data (Issaris et al., 2012; Monk, 2013). When animals are rare, cryptic or depleted, revealing their presence requires large amounts of effort (Guillera-Arroita and Lahoz-Monfort, 2012; Maxwell and Jennings, 2005). The use of some terrestrial methods is becoming more common in the ocean.  Examples include underwater videos at baited sites (Goetze and Fullwood, 2012), scat sampling (Bowles and Trites, 2013) and genetic analysis (Mobley et al., 2011).  However, these methods are not widely employed for cryptic data-poor marine species (Jerde et al., 2011). Only one study has compared presence /absence with detection probabilities in relation to two diver-based methods – stationary census, where a diver remains stationary and records species within a given distance, and belt transects, where a diver swims along a transect at a constant speed recording species within a fixed distance of the transect (Green et al., 2013). The results indicated that both diver-based methods had poor detection abilities (Green et al., 2013). With continuing marine species   26 decline (McCauley et al., 2015), managers urgently need new ways to handle data scarcity and facilitate effective decision-making for the ocean.   Seahorses (Hippocampus spp.) are such a challenging fish genus to locate that they are good representative candidates for a detection study. They are difficult to find in the wild because they camouflage well, have a patchy distribution, and are found in low numbers (Foster and Vincent 2004). Seahorses have small home ranges (Perante et al., 2002), and live in shallow, coastal habitats including coral reefs, seagrass beds, mangroves, and sandy soft bottoms (Lourie et al., 2004). Worldwide there are 41 species of seahorses (Lourie et al. in press). The IUCN Red List of Threatened Species (www.redlist.org) lists 12 seahorse species as threatened with 20 classified as ‘Data Deficient’.   I here conduct the first study to focus on using presence / absence with detection probabilities for cryptic, data-poor marine fishes. Other species-specific detection research has focused on comparatively large marine animals including marine mammals and sharks, or lionfish (Green et al., 2013; Williams and Thomas, 2009). I simultaneously explored efforts to locate seahorses and contribute to the literature on search methods and detection for cryptic and data-poor marine fishes. First I evaluated the success of searches aimed at determining relative abundance, where I assume detection probabilities are = 1, to locate my species of interest.  Next, I compared two methods - belt transects, where the search area is fixed along a transect, and timed swims, where the search area is unrestricted - for their efficiency to locate cryptic, data-poor marine fishes. I then re-evaluated my datasets comparing belt transects and timed swims, with presence/absence models with detection co-variates that enable me to determine if my original estimates of   27 occupancy were biased. Finally, I explored how studies using presence / absence with detection probabilities can be incorporated into future marine research for data-poor species with simulations, sensitivity analyses, and a cost-benefit analysis.  2.3 Materials And Methods 2.3.1 Data Collection 2.3.1.1 Determining Relative Abundance Assuming Detection Probability = 1 My search to determine relative abundance of seahorses began on the west (Andaman Sea) coast of Thailand using common underwater survey protocols based on Reef Check (Andrew and Mapstone, 1987; Hodgson, 2004). My surveys took place from February to May 2013, and covered 19 sites, across all six Andaman provinces, in mangrove, seagrass, and coral reef habitats (Fig 2.1). I could not select sampling sites in the national parks because of permit restrictions. Therefore sites were chosen based on the best available habitat outside national parks, according to local expert opinion (from marine biologists, fishers, and dive operators).  I conducted searches with belt transects at 10 sites and timed swims at nine sites (Appendix A.1).  My multiple searches (mean 8, range 1-20) using snorkel and scuba at each site covered a mean area of approximately 1100 m2 but ranged between 500 – 2000 m2 (Appendix A.2).   The challenges encountered on the Andaman coast (see results), led me to alter the methods for surveying the east (Gulf of Thailand) coast. I began by making a concerted effort to extract seahorse sightings from all available information including, but not limited to, internet resources, popular media, survey reports, published articles and direct contact with scuba divers and dive shops. I also encouraged people to report their observations to the global database of seahorse   28 sightings, iSeahorse.org. I used the responses from this outreach to select 35 sites in the Gulf of Thailand (Fig 2.1). My search efforts on the eastern coast took place in September and October 2013 still with the aim of determining relative abundance of seahorses at sites using Reef Check protocols (Hodgson 2004). I conducted only timed swims for each site. The number of repeat searches per site varied from one to six (mean of two) based on logistic constraints and the need to visit as many sites as possible for wider geographic coverage (Appendix A.1).   2.3.1.2 Comparing Methods Assuming Detection Probability < 1 Based on the results of my attempts to determine relative abundance, I focused my second field season on comparing the relative efficiency of belt transects (searching in a fixed area) and timed swims (searching in an unrestricted area) - controlling for search time in both - as underwater search methods to locate seahorses (Andrew and Mapstone, 1987). I chose research sites based on the knowledge derived from the aforementioned outreach efforts (Fig 2.1). My challenges in locating my species of interest with Reef Check search protocols based on relative abundance in the first season highlighted the importance of defining the size of my search site. I therefore defined each site as an area roughly 75 000 – 80 000 m2 (approximately 400 m x 200 m, or 8 ha). Fieldwork occurred from March to July 2014 at 10 sandy, soft bottom sites – seven on the Andaman Coast and three on the Gulf Coast. I chose sandy, soft bottom sites (known habitat for five of the seven Thai seahorse species (Lourie et al., 2004)) to maximize my chances of locating a seahorse; seahorses are easier to find in this type of habitat because it has low relief and rugosity, and few places to hide. I conducted five replicates of each method per site. An additional ten sites (five on each coast) were visited opportunistically, but not surveyed fully because of logistic constraints (Appendix A.1).    29  I surveyed sites over a three to four-day period to minimize the probability of changes in seahorse occupancy (presence or absence) over the sampling period. For transects and timed swims, I generated a random start point, using a random number table, within the site for each of the five replicate searches. All replicates per site were conducted for the same amount of total search time (ranging from 50 to 75 minutes depending on site depth and allowable dive time), with divers swimming in the same general direction. For example, if a site had a depth that allowed a dive time of 60 minutes, regardless of where the starting point was located at site X, all timed swims headed in a northern direction and lasted for 60 minutes. I measured the distance covered by timed swims with calibrated fin kick cycles (to determine length), and assumed a search width of two meters. Transect surveys consisted of as many 2 x 50 m transects as dive time would allow (ranging from two to five). I laid out transects perpendicular to the water current direction, in alternating directions, with 50 meters in between each transect. For each method, upon finding a seahorse, the species, time, and distance covered were recorded.  2.3.2 Data Analysis 2.3.2.1 Determining Relative Abundance Assuming Detection Probability = 1 First, I determined the proportion of sites where seahorses were located, for both the Andaman and Gulf Coasts. This calculation assumed that any zeros represented a true absence of seahorses. Next, I identified the number of sites on each coast where local fishers or divers reported seahorses as common (based on fieldwork notes) but where I observed no seahorses in my search efforts. These sites would suggest that seahorses were indeed present but my search efforts were not robust or sensitive enough to detect them.    30  2.3.2.2 Comparing Methods Assuming Detection Probability < 1 I analyzed my data based on search times, search distance, and survey effort with homoscedastic t-tests, to determine which method was more effective at finding the first and second seahorses at a site (Zar, 1999). For this analysis I used the data from my second field season, but only from sites where seahorses were observed. I also conducted a post-hoc power-test to determine if I had enough power to determine if there was a difference in the proportion of surveys with seahorses per site between timed swims and transects for an effect size of 0.2 at the α = 0.2 level (Zar, 1999).   2.3.2.3 Re-evaluating With Occupancy Models When Detection Probability < 1 I re-evaluated my data from relative abundance surveys (2.3.1.1) and from comparing belt transects and timed swims (2.3.1.2), with occupancy models to explore how my results might differ if detection probabilities were accounted for in my analysis. For my relative abundance dataset (initial assumed detection probability = 1) I used single season occupancy models to generate estimates of detection probabilities < 1 (Mackenzie et al. 2002). Next I compared my estimates of seahorse occupancy across the seascape (proportions of sites where I located seahorses) to identify how site-specific occupancy varies with and without accounting for detection probabilities. Occupancy models were run with the software Presence (Hines, 2006).   I also explored the possibility that survey method, visibility or time of day influenced my detection probability of seahorses by creating single season occupancy models using each of these factors as detectability covariates. I hypothesized that these three variables would be most   31 likely to affect the efficiency and reliability of finding a small cryptic fish, given my decision to search in sandy soft bottom habitats. I also created multistate, single season models to explore the relationship between detection probabilities and relative abundance (high vs. low) of seahorses per site (Royle and Nichols, 2003). I defined a site as high abundance if I observed four or more seahorses during any given search effort (timed swim or transect). This is because seahorses are commonly found in pairs and rarely found in groups of three or more (Perante et al., 2002).  A site was defined as a low abundance site when three or fewer seahorses were observed during any given search effort. I conducted all analyses with two datasets from 2014 (2.3.1.2) - research sites only (n=10) and research sites plus opportunistically visited sites (n=20) (Appendix A.1). The presence / absence of seahorses at each site was determined by the pooled data from all surveys with both protocols.  2.3.2.4 Sensitivity Analyses And Future Study Design Assuming Detection  Probability < 1 I explored how an occupancy and detection framework could be incorporated into planning for future marine research. I ran a cost-benefit analysis with several sampling design scenarios to determine the most cost-efficient study design for a data-poor species. There are two main study designs for occupancy and detection models (Mackenzie and Royle, 2005). A ‘standard design’ is one where all sites are surveyed the same number of times (scenario 1), while a ‘removal design’ is one where surveying stops once the species of interest is found or until a pre-determined maximum number of replicates has been reached (scenario 2 and scenario 3 respectively).  For my simulations I calculated the number of recommended replicate searches for a 90% probability that any zeros would represent true absence at a site (Mackenzie and Royle   32 2005). This procedure required me to input initial estimates of occupancy and detection, and I used estimates from one of my single season models. This calculation resulted in five replicate searches for the standard design and a maximum of seven searches for the removal designs. Scenario 1 is a typical standard survey design where all sites are searched five times. Scenario 2 is a removal design simulation where a small, cryptic data-poor marine species is found within the first day of searching at each site – before my maximum replicate search number is reached. Scenario 3 is a removal design simulation when the focal species is not observed in early searches. It takes 2 or more days to either observe the species or reach the maximum number of pre-determined replicates for all sites. For this scenario I estimated the number of days it would take (2+) to reach this number (7 replicates) based on observations from my second research season.   To build on the results of simulations and survey design, I wanted to identify the most important factor driving the optimal number of sites to survey. This required me to conduct a sensitivity analysis. I evaluated the change in number of sites to survey with changes in initial occupancy estimates, detection probabilities, and desired confidence level in future estimates of seahorse presence or absence (occupancy) per site (Mackenzie and Royle 2005).  2.4 Results  2.4.1 Determining Relative Abundance Assuming Detection Probability = 1 Searching to determine relative abundance among sites resulted in low numbers of seahorses during my first field season. A total of ten individual seahorses was found on the Andaman Coast at 36% (n=7 of 19) of sites surveyed (Appendix A.2). These results did not allow me to   33 determine relative abundance or identify sites for future research. In comparison, on the Gulf Coast, a total of 39 individuals was found across 38% of the sites visited (n = 10 of 35) (Appendix A.2). I gained some ground in that these results identified two sites for future research.   When I compared my search results to reports of where seahorses were common (Appendix A.2), I found discrepancies at 20% (n=11 of 54) of the sites I surveyed. At eleven sites (five on the Andaman and six on the Gulf) where seahorses were not encountered, local fishers or divers reported seahorses as common based on capture in fishing nets or tourism experience (Appendix A.2). These results highlighted that at these sites (and potentially others), my search efforts were not sufficient to detect seahorse occupancy.  It remained unclear, however, what level of search effort was actually needed.   2.4.2 Comparing Methods Assuming Detection Probability < 1 On average, timed swims were more effective than belt transects at finding seahorses based on analysis with effort metrics (number of surveys, time searched and area covered) (Table 2.1). Timed swims found seahorses in fewer surveys and a shorter amount of time than transects (Table 2.1). However, transects found seahorses after searching a smaller area than timed swims (Table 2.1). The amount of time and distance to the second seahorse observed was shorter on timed swims than on transects (Table 2.1). Homoscedastic t-tests found no significant difference in the total number (p=0.59) or proportion of searches that found seahorses (p=0.43) between transects and timed swims. A post-hoc power test indicated that I had 24% power to detect a difference in the proportion of surveys with seahorses per site for each search method.    34  2.4.3 Re-evaluating With Occupancy Models Assuming Detection Probability < 1 2.4.3.1 Relative Abundance Dataset (2.3.1.1) My re-evaluation of relative abundance data revealed that estimates of detection probability  were < 1 (Fig 2.2). I found overlapping confidence intervals in detection probabilities - for timed swims (0.23 ± 0.07) and transects (0.12 ± 0.10) on the Andaman coast, and timed swims (0.41 ± 0.16) on the Gulf coast (Fig 2.2). Once I factored in a detection probability of < 1, my estimates of sites occupied by seahorses were higher than my initial estimates (Andaman coast occupancy 58.5% ± 29%, Gulf Coast 73.9% ± 26%), but still maintained high standard errors.   2.4.3.2 Comparing Methods Dataset – Belt Transects And Timed Swims (2.3.1.2) Drawing in prior information (garnered through outreach activity) increased my probability of finding seahorses at research sites. More seahorses were found during the 2014 research season; a total of 69 seahorses were observed, at 70% (seven of the ten) of sites surveyed. By including the sites I visited opportunistically, a total of 98 seahorses at 11 of 20 (55%) surveyed sites were found. When I accounted for detection probabilities < 1, my estimates of site occupancy increased with both my dataset with 10 sites (70%) and with 20 sites (65%).   The only factor significantly influencing detection probabilities was abundance. I found overlapping confidence intervals between timed swims and transects for both my datasets, although timed swims had consistently higher detection probabilities (Fig. 2.2).  Sites with higher total numbers of seahorses had higher detection probabilities for both methods (Fig. 2.3). The multi-state single season model for the dataset with 10 research sites had overlapping 95%   35 confidence intervals in detection probabilities for sites with high and low abundance of seahorses. However when I used the dataset with 20 sites (refers to research plus opportunistic sites), there was a significant difference in detection probabilities between sites with high and low abundance of seahorses (Appendix A.3). Detectability did not improve with improved visibility, nor was detectability of seahorses better in the morning or afternoon (Table 2.2).   2.4.4 Sensitivity Analyses and Future Study Design For When Detection Probability < 1 A removal sampling design is the most cost efficient way to sample for small, cryptic, data-poor marine fish (Table 2.3).  The average cost per day for research in the 2014 field season was US$177.55 (Table 2.3). The total savings for a removal design with minimal sampling would have provided the opportunity to visit an additional 29 sites. The total savings for a standard sampling design or for a removal design with maximum sampling were close to equal, as were the additional number of sites to visit (Table 2.3).   Simulations indicated that the factor most influencing the optimal number of sites to visit was the acceptable level of confidence of future seahorse presence / absence (occupancy) per site (Fig 2.4). As the acceptable level of confidence increased, the number of optimal sites to visit increased (Fig 2.4). In terms of identifying the optimal number of replicate surveys needed to obtain confidence that zeros represent absence at a site, my simulation at 90% confidence indicated that as detection probability increased, the number of required replicate surveys decreased (Appendix A.4).    36 The presence / absence detection framework was the most informative for effective initial management (Table 2.4). Executing timed swims only identified two sites for future research and or management, but could not be used to determine most effective search method or overcome issues of detection (Table 2.4). My search based on relative abundance on the Andaman Coast where I conducted both haphazard transects and timed swims did not yield enough information about seahorses for initial management measures. When I evaluated my data by comparing methods in a presence / absence detection framework, I identified three sites for future research, provided supporting evidence that both methods were equally effective at finding seahorses at sites with high abundance, and provided information to deal with issues of non-detection (Table 2.4).  2.5 Discussion My research highlights that using a framework based on presence absence with detection probabilities (Mackenzie et al., 2002) is more useful for cryptic data-poor marine fishes than one using relative abundance, despite the latter being more widely used in marine conservation research (Andrew and Mapstone, 1987). The presence absence/ with detection probability framework enabled me to determine the probability that sites with zeros represented the true absence of species given my sampling efforts, as commonly done in terrestrial detection studies (Kéry and Schmidt, 2008; MacKenzie et al., 2006).  This prevents the Type II error of assuming absence of an organism when it is present, a serious matter when one is considering remedial management action for a species of conservation concern.    37 Gathering local knowledge advanced my understanding of where my cryptic data-poor species could be found, supporting similar findings about the value of local knowledge for research on data-poor species (Thornton and Scheer, 2012). Despite the growing use of local knowledge to inform marine species distributions, its use in a management context remains controversial because of potential data bias (Thornton and Scheer, 2012; Usher, 2000).  Indeed, I also initially placed more value on systematic technical approaches than on gathering local knowledge (Turvey et al., 2015), and thus hampered my first attempts to find sites for seahorses. Selecting sites based upon prior knowledge is entirely reasonable if one is trying to figure out where the animals can be found, in order to mobilize protection as soon as possible.  Further research can work out presence/absence more broadly and even address relative abundance eventually.  My research shows that when surveying data-poor species, factoring in detection probabilities changes the conclusions drawn whether in the ocean (Monk, 2013) or on land (Bailey et al., 2007). Initially, assuming detection probability = 1 caused me to underestimate the number of sites with seahorses, as has been the case with all other seahorse studies relating to density, abundance, distribution and habitat preferences. For example, Yasué et al., (2012) would have done well to incorporate detection probabilities before declaring that marine protected areas had no significant effect on densities of a seahorse species (H. comes) in the Philippines. This is particularly true as I would expect detection probabilities to decrease in more complex environments, such as seagrass beds and coral reefs and marine protected areas may often be more complex than heavily fished sites (Green et al., 2013).     38 Similar to land based conservation efforts, I found when planning a detection study for cryptic marine fishes, a removal sampling design is less expensive than the standard design (Bailey et al., 2007; Mackenzie and Royle, 2005). The standard survey design, where all sites are repeatedly sampled the same number of times, is a common choice for monitoring programs because a balanced survey design supports the use of relative abundance analyses (Andrew and Mapstone, 1987). However when searching for cryptic species, a removal sampling design – one where surveying stops once the species is found or until a maximum number of surveys has been reached – allows for the maximization of resources to visit as many sites as possible (Mackenzie and Royle, 2005). When faced with a need to gather baseline data on the presence and absence of a data-scarce species expediently, where detection of individuals is constant for all surveys, marine managers should opt for a removal survey design (Mackenzie and Royle, 2005).   Techniques to assess data-poor species are rapidly evolving for both terrestrial and aquatic systems (MacKenzie et al. 2006; Durso, Willson, and Winne 2011). On land, occupancy and detection work has provided scientific support for decisions about habitat corridors (Zeller et al., 2011), predator control (Robley et al., 2014), invasive species remediation (Britton, Pegg, and Gozlan 2011) and conservation planning (Weller, 2008). Despite such advances on land, I have only begun to incorporate these ideas into ocean conservation and research (Green et al., 2013; Issaris et al., 2012). Studies using presence / absence with detection probabilities are flexible and practical for science-based decision making (MacKenzie et al. 2006) as compared to the data challenges of comparing densities and relative abundance, which require empirical rigor that is seldom available (Zar, 1999). Species presence and absence across the land or seascape forms the basis for identifying sites for protection and creating a monitoring protocol, both critical steps   39 during the initial phase of management. Finding a solution to the conservation crisis that is both pragmatic and expedient will involve increased communication among conservation fields to share beneficial lessons such as those to be learned from occupancy and detection.    40  Figure 2.1 Survey efforts on Thailand’s Gulf and Andaman coasts to determine relative abundance or presence absence and detection. Sites visited opportunistically for presence absence and detection are also included.   41 Table 2.1 Mean number of surveys, minutes and area covered before observing a seahorse (Hippocampus spp.) at a research site for belt transect and timed swim search methods. Method Mean # of surveys to 1st seahorse sighting Mean Time to 1st sighting  (minutes) Mean Distance Covered to 1st sighting Time of 2nd sighting (minutes) Mean Distance Covered to  2nd sighting Transect 2.3 122.5 891 m2 49.8 343 m2 Timed Swims 1.5 72.0 1 095 m2 17.8 184 m2    42  Figure 2.2 Estimates of detection probabilities (+/- 95% Confidence Intervals) of search methods (belt transects versus timed swims) based on a) relative abundance (2013) and b) methods comparison (2014) datasets.     43  Figure 2.3 Site detection probability based on the total number of seahorses found per site (n=10) by search method based on 2014 research data. Seahorses had higher detection probabilities when they were more abundant.      44 Table 2.2 Results of single season occupancy models exploring the effects of method, time of day and visibility on detection probabilities compared to the simplest model with a constant detection co-variate.  2014 research sites Δ AIC Support for model (ω) Number of variables (K) Occupancy (+/-SE) Detection rate  (+/- SE) Simplest (null) model 0.00 0.4449 2 0.7057 (0.14) 0.38 (0.05) Detection varies by method 1.46 0.2144 3 0.7055 (0.14) 0.34 (0.08) Transect  0.42(0.08) – Timed swim Detection varies by time of day  1.88 0.1738 3 0.7054 (0.14) 0.40 (0.08) – AM 0.36 (0.08) - PM Detection varies by visibility  1.96 0.167 3 0.7053 (0.14) 0.39 (0.07) –  < 1 m 0.39 (0.08)-  > 1 m      45 Table 2.3 Costs and savings of three survey designs scenarios (Mackenzie and Royle 2005) to search for seahorses (Hippocampus spp.) based on per diem costs of 2014 field season. The standard design (scenario 1) employs the same number of replicates at all sites while the removal survey stops surveys at a site once a seahorse has been found or until a pre-determined maximum number of replicates has been reached (scenario 2 / scenario 3).    Scenario 1 Standard Design Scenario 2 Removal Design  (species found in 1st day) Scenario 3 Removal Design  (species found in 2+ days or max. searches reached) Estimated cost per day from 2014 field season $178 $178 $178 Simulated total number of days needed to survey 20 sites 34 20 36 Simulated total cost to survey 20 sites  $6 036 $3 551  $6 391 Total savings compared to actual 2014 research costs to survey 20 sites $2 663 $5 149  $2 308 Simulated additional sites to visit using total savings 15 29 13    46  Figure 2.4 Simulation results highlight that the optimal number of sites to visit increases as with increased need for confidence in determining future occupancy of seahorses at sites.       47 Table 2.4 The usefulness of search strategy for informing future research questions.  Future Research Questions Effective search method Overcomes issues of non-detection Identification of improvements for future research design Identified Sites for Future Research / Management Relative abundance   - mixed methods No No No No Relative abundance – 1 method No No Yes Yes Presence absence detection framework Yes Yes Yes Yes    48 Chapter 3: Generating Spatial Data For Marine Conservation And Management 3.1 Summary Do fishers know best when it comes to identifying areas with rare and depleted fish species? The global conservation crisis demands that managers marshal all available datasets to inform conservation management plans for depleted species, yet the level of trust placed in local knowledge remains uncertain. This study compares four methods for inferring species distributions of an internationally traded, rare and depleted genus of marine fishes (Hippocampus spp.): the use of (i) fisher interviews; (ii) government research trawls, (iii) scientific diving surveys, and (iv) citizen science contributions. I analyzed these four datasets at the genus and individual species levels to evaluate my conclusions about seahorse spatial occurrence, diversity of species present and the cost effectiveness of sampling effort. I found that fisher knowledge provided more information on my data-poor fish genus at larger spatial scales, with less effort, and for a cheaper price than all other datasets. One drawback was that fishers were unable to provide data down to the species level. People embarking on conservation endeavors for data-poor species may wish to begin with fisher interviews and use these to inform the application of government research, scientific diving, or citizen science programs.   3.2 Introduction Spatial data lies at the heart of current conservation and management efforts. Numerous management plans for species of conservation concern call for monitoring of animals or for protecting their critical habitats (NOAA, 2014). Identification of important populations   49 (Thornton and Scheer, 2012) and locations with extractive activities (Reid, 2007) are common research and management objectives both on land and in the ocean. The data requirements underlying these objectives are spatial in nature and are critical to achieving conservation outcomes.   Challenges remain about the best ways to generate spatial data to achieve conservation outcomes, especially for species where baseline information on local distribution is non-existent or very limited (data-poor). Common methods to generate spatial data for such species include field surveys to detect presence or absence (Mackenzie et al., 2002), camera trapping (Bender et al. 2014), acoustic sampling (Koslow, 2009), and animal tagging (Lewis et al., 2009). Growing in use are citizen science programs, which encourage the public to report their observations of wildlife (Braschler, 2009). Biases and benefits are associated with each of these methods (Dickinson et al. 2010; Katsanevakis et al. 2012), and guidance is still needed on where natural resource managers should prioritize scarce resources.  Considerable information about data-poor species often exists in local knowledge (Gilchrist et al. 2005), but its application for management remains uncertain (Mallory et al., 2003). Local ecological knowledge (LEK) refers to ‘knowledge generated and reproduced through human-environment interactions in specific locations by local inhabitants’ (Byg et al., 2012). LEK has supplied sightings data for rare species (Turvey et al., 2015) both on land and in the ocean, in a cost effective manner (Thornton and Scheer, 2012; Turvey et al., 2014). Providing a source for historical baseline information (McClanahan et al. 2012), LEK can complement scientific data to expand spatial or temporal scales (Tobias, 2010). Despite a growing movement to verify LEK   50 (Beaudreau and Levin, 2014; Shepperson et al., 2014; Turner et al., 2015), this practice remains uncommon, lending uncertainty to its use (Usher 2000). There are many biases associated with LEK, such as observation error, perception bias, shifting baselines, and telling researchers what they want to hear (Stake et al. 2005; Byg et al. 2012). This knowledge is also limited to locals’ experience (what they need to know)(Mallory et al., 2003) and tends to incorporate more than one time period (Shepperson et al., 2014). These biases influence the validity of extracting and translating LEK into scientific formats (Berkes, 2009), but all forms of information carry uncertainty.   Despite the challenges of generating spatial data for marine species, current ocean management strategies are commonly spatial. Many national and international policies are geared towards creating marine protected areas (Chape et al., 2005), and many countries are now prioritizing marine spatial planning (Douvere, 2008; Stelzenm et al., 2012). Common fisheries management strategies include gear restrictions in certain areas or seasonal closures (Cadrin et al., 1995; Panjarat and Bennett, 2012). In order to evaluate the effectiveness of conservation and management strategies, spatial data are required on species (Moller and Berkes, 2004), habitats (Chape et al., 2005), and human activities (Reid, 2007).   Many international environmental treaties, such as those on global fisheries or air pollution, use spatial data as part of the monitoring and enforcement process (Lorenz, 1995; Lorenzen et al., 2010), and one new application for spatial data in the ocean is with the Convention on International Trade of Endangered Species (CITES) (Rosser and Haywood, 2002). The CITES treaty regulates international wildlife trade to ensure it does not contribute to the extinction of   51 species in the trade (Vincent et al., 2013). Internationally traded wildlife species may be listed on one of three treaty Appendices, the second of which (Appendix II) requires countries to regulate trade to ensure exports are not harmful to wild populations (UNEP-WCMC, 2012). As part of the enforcement process for Appendix II species, any country may be asked to provide evidence that exports are not harmful to wild populations (non-detriment finding) and specimens are legally sourced. Meeting these conditions requires spatial data in several contexts including species locations, habitats, extractive activities and management areas (Rosser and Haywood 2002, CITES 2013). Recently guidelines for making non-detriment findings were published for seahorses and sharks (CITES, 2013a; Mundy-Taylor et al., 2014), but these were not available at the time when several countries were asked to undergo the first enforcement process for marine fish in 2012 (UNEP-WCMC 2012).   The need for guidelines became apparent when Thailand (and Vietnam) were asked to make changes in their process to evaluate the sustainability of trade for marine fishes (UNEP-WCMC, 2012). Thailand is the world’s largest exporter of seahorses, (Hippocampus spp.), a cryptic marine fish genus listed on the CITES Appendix II (UNEP-WCMC, 2012). Seahorses were the first marine fishes to be listed on CITES Appendix II since its inception, and the first marine fishes on the CITES treaty to have countries undergo the Appendix II enforcement process (Vincent et al., 2013). Insufficient data on seahorse distribution and extractive activities were available when Thailand was asked to make changes to its process, which necessitated the exploration of appropriate data sources (CITES, 2013a).     52 This paper compares information gained and costs associated with different types of information that can be used to infer species distributions. I determine the most cost-effective method to identify where seahorses live and evaluate these methods at i) the genus and ii) individual species levels. I compare knowledge of these marine fishes generated from fisher interviews to data gathered from governmental research trawls, scientific diving surveys and citizen scientists contributions, and evaluate the conclusions drawn from the four methods.  Two years of collaborative research efforts yielded four datasets with spatial information about seahorses (Hippocampus spp.): i.fisher interviews, ii. Thailand Department of Fishery (DoF) research trawls, iii. scientific diver surveys, and iv. citizen science diver contributions) with spatial information about seahorses (Hippocampus spp.). I determine which dataset (or combination of datasets) was the most useful in a conservation context and compared the knowledge from the various datasets in terms of in situ seahorse occurrence, species composition, and cost of generation.  3.3 Methods I predicted that (i) fisher generated knowledge would provide information at the largest spatial scale, for the most number of species, and be the most cost efficient method. I therefore made three comparisons among four datasets (Fig. 3.1). To explore the available information in different subsets of (i) fisher LEK, I first compared the information generated by commercial fishers with that of small-scale fishers. Since commercial and small-scale fisheries have defining characteristics in Thailand (Lymer et al. 2010), (Appendix B.1) I expected their knowledge of seahorses to differ. Second, I compared a subset of the data from the (i) commercial fishers, trawl captains, to the (ii) DoF research trawls. Third, I compared (i) small-scale fisher data to   53 diver-generated ([iii.] scientific diving surveys and (iv.) citizen science diver contributions) data because I expected both small-scale fisher and diver reports of seahorses to be localized at similar spatial scales. As part of this analysis I also compared the similarities and differences of data generated by scientific divers to that of citizen science divers.   3.3.1 Data Collection I selected data from Thailand’s western coast, an area covering 865 km in length, with six provinces (Lymer et al. 2010). I chose this coast of Thailand because it had the most comprehensive coverage across all four datasets. I now describe the methods used to produce these datasets.   3.3.1.1 Dataset 1: Local Knowledge – Fisher Interviews To select respondents, I targeted two sets of landing locations along the Andaman Coast: fishing ports for commercial fishers and coastal villages for small-scale fishers (Appendix B.2). I visited 26 locations including 87% of commercial ports (n=7 of 8) and 3% of small-scale fisher villages (n=19 of 621) based on the recommendations of provincial DoF staff. These ports and villages were representative of the fishing gears used in each province. For the percentage of ports and villages visited by province, see Appendix B.3.  To gather seahorse spatial data, I interviewed 73 commercial captains and 120 small-scale fishers at the ports and villages in six provinces along the Andaman Coast with semi-structured interviews. I determined the number of fishers to interview in each location as either 10% of the estimated total number of fishing boats landing catch at that location or the saturation method   54 (Tobias, 2010). I defined saturation as the point where if the 6th, 7th and 8th interviews yielded no new information on gear type, fishing grounds, or seahorse locations, then saturation had been reached (Tobias, 2010). I additionally confirmed saturation through my observations of gear type at each landing location. To participate in my study, fishers had to be a captain and participated in fishing activity within the last year because I assumed that captains would have the most spatial knowledge of seascape based on navigation experience. Interviews occurred in Thai through the use of a local translator. All interviews followed UBC Human Ethics Protocols (H12-02731).   To identify seahorse locations across the Andaman Sea, I conducted spatial mapping with fishers using a computer tablet with iGIS software (Geometry Pty Ltd, 2008). I oriented fishers on how to use the tablet before mapping started. If fishers stated they did not feel comfortable using the tablet after the orientation, I guided mapping efforts by drawing seahorse polygons in iGIS based on the fisher’s description. Fishers were asked to identify a) the spatial extent of their fishing grounds b) locations where they knew seahorses lived, and c) places where they had captured seahorses with fishing gear. They were asked to describe the depths of all the aforementioned locations. Fishers were also asked to describe the seahorse species they found based on a seahorse species identification guide. If fishers could not identify the species, they were invited to classify the type of seahorse as either smooth or spiny.   3.3.1.2 Dataset 2: Department of Fisheries Research Trawls Trawl surveys were executed by the Thailand Department of Fisheries in 2010, 2012 and 2013 at 22 pre-determined sampling locations throughout the Andaman Sea, originally intended for the   55 purpose of sampling commercially exploited fish species. Each location represented a grid equivalent to 15 x 15 nautical miles (Appendix B.4). The grids were sampled four times per year using a 23.5m otter-board trawl research vessel, with trawl speed set at 2.5 nautical miles / hour. Trawling took place within each gridded area for one hour.  GPS locations were marked at the beginning and ending of each trawl, and the depth of the trawl recorded. All fish from each trawl were sorted and identified by Department of Fisheries staff. Presence or absence of seahorses was noted for each trawl, and seahorses were identified to species level.   3.3.1.3 Dataset 3: Scientific Diving Surveys I used data from Aylesworth et al. 2015 (Chapter 2) to identify locations of seahorses in the Andaman Sea. These data were collected through underwater scientific diving surveys over two three-month field seasons. A total of 26 sites were visited over the two field seasons in coral, seagrass, mangrove and sandy soft bottom habitats (Aylesworth et al. 2015)(Chapter 2). These data included GPS coordinates for each site, and when seahorses were observed, the species, depth, and habitat were recorded (Aylesworth et al. 2015) (Chapter 2). This research was conducted in accordance with UBC Animal Ethics protocols (A12-0288).   3.3.1.4 Dataset 4: Citizen Science Diver Contributions I evaluated citizen science contributions of seahorse sightings from the global seahorse database, iSeahorse.org. All data available through April 23, 2015 were included for analysis, excluding any data contributed by the authors. The iSeahorse program encourages divers to report sightings of seahorses to the database with the option to include species names, habitat, depth, and any additional information, e.g. behavior. A concerted effort was made during my project to promote   56 the iSeahorse program within Thailand’s diving community through presentations, training workshops, and appeals to divers (English and Thai) on social media. I had no control over who contributed sightings and from which locations, but I used all sightings spatially located in Thai waters.  3.3.2 Data Analysis I conducted three comparisons (Fig. 3.1) to assess if the conclusions I drew about the locations of seahorses at the genus and individual species levels changed based on the data I used. For each analysis I compared the information on where seahorses lived, how many species were identified and which types, and the effort needed to generate the data. I included all available data from all datasets, after an initial review determined all spatial data was located inside Thailand’s exclusive economic zone. To compare the knowledge about where seahorses lived, I analyzed the total number of seahorses, depth ranges, and the number of seahorses found inshore (within 3 km of shore) compared to offshore (more than 3 km from shore). I defined inshore areas to be within 3 km of the shore, because this coastal area had been set-aside for small-scale fishers for many years (Lymer et al. 2010). Pushnets and trawlers are not allowed within 3 km of shore based on Ministerial Notification from 20 July 1972. To compare knowledge about seahorse species, I analyzed the number of species found (or reportedly found), the composition of those species, and determined the dominant species present. To compare the effort expended to generate the datasets I compared the number of sites visited, total number of days, number of interviews conducted, number of staff deployed, and cost per day in US dollars to generate each dataset. I determined cost based on the amount of money expended in Thailand to generate each dataset (Appendix B.5). For example, with the citizen science dataset, I did not include creation   57 or maintenance costs associated with the iSeahorse.org database of seahorse sightings. I only included outreach costs conducted in Thailand from January 2013 until September 2014.  3.3.2.1 Comparison 1: Commercial And Small-scale Fishers To compare the identified spatial extent of seahorses across the Andaman coast from commercial and small-scale fishers, I created seahorse presence maps by overlaying individual fisher’s maps of seahorse locations. Prior to analysis, I standardized the maps’ spatial accuracy because fishers’ knowledge varied in how they described the locations of seahorses and their fishing grounds (Appendix B.6). All fisher shapefiles required some standardization to ensure spatial accuracy.  By reviewing the data for spatial accuracy I also ensured there were no outliers in these data (Appendix B.6). I counted the number of shapefiles that overlapped in each area. Then I calculated the proportion of fishers interviewed who reported catching seahorses in any given area. I binned the output count shapefiles to represent equal thirds in the overlap values (Commercial: 0-9% 10-19%, 20-33%; Small-scale: 0-5%, 5-10%, 10-15%). I verified these maps with two methods, internal consistency and fisher interviews prior to use in analysis (Appendix B.6). A table reporting the errors identified during the verification process can be found in Appendix B.6. All spatial analyses were conducted in ArcGIS 10.1 (ESRI 2011).  3.3.2.2 Comparison 2: Trawl Captain Interviews And Department of Fisheries Research Trawls  I compared commercial trawl captains’ knowledge about seahorse locations with the DoF research trawl data. From the commercial fisher interviews, I selected those shapefiles from trawl captains and executed the same analysis as described above to create a seahorse presence   58 map. With the DoF trawl survey data, I created point shapefiles. Next I executed a simple overlay analysis with the trawl captain seahorse maps and the DoF trawl survey data to identify the similarities and differences between the two datasets. I ran an additional analysis of trawling effort to highlight discrepancies in the datasets that were related to effort (Appendix B.7).   3.3.2.3 Comparison 3: Small-scale Fisher And Diver (Scientific And Citizen Science) Generated Data  I compared the knowledge generated from small-scale fishers to that of scientific diving surveys and citizen science diver contributions of seahorses throughout the Andaman Sea. I executed a simple overlay analysis with these datasets to identify spatial similarities and differences among the datasets.  3.4 Results  3.4.1 Comparison 1: Commercial And Small-scale Fishers The knowledge from both commercial and small-scale fishers was sufficient to provide guidance on where seahorses live at the genus level along the Andaman Coast of Thailand (Fig. 3.2; Table 3.1). Commercial fishers identified areas farther offshore, with deeper ranges, and for a greater extent of the coast for seahorses than small-scale fishers. More small-scale fishers (80%) provided knowledge about the depths of seahorses than commercial fishers (64%). Fishers could not identify seahorses to species level (Table 3.1). Many fishers (78% commercial, 99% small-scale) were able to comment about whether the seahorses were smooth or spiny, but could not identify species based on the seahorse identification guide.    59 In terms of sampling effort, my coverage of landing locations and number of respondents varied greatly between small-scale and commercial fishers (Table 3.2). I visited a larger proportion of commercial landing locations than small-scale fisher landing locations, given similar metrics of cost per day, total number of days and staff needed for conducting fisher interviews (Table 3.2). However, I interviewed many more small-scale fishers than commercial fishers over a relatively similar time period (Table 3.2).   3.4.2 Comparison 2: Trawl Captain Interviews And Department of Fisheries (DoF) Research Trawls The trawling dataset comparison shows differences in spatial scale and inference about where seahorses live (Fig. 3.3) at both the genus and individual species level. At a broad scale, and the genus level, seahorses were found throughout the areas where trawling occurred, across all depths and distances from shore (Table 3.1). However, the trawl captain responses provide more detail at the genus level, across a larger area, because they fished more often, more frequently, and for longer periods of time.   At the individual species level, only the DoF research trawls provided information on the number of species observed, and species composition (Table 3.1). Hippocampus trimaculatus Leach 1814 represented 97% of the individuals captured, and approximately 50% of all seahorses were captured just south of Koh Lanta (Fig. 3.3). Hippocampus trimaculatus was found mostly in areas south of Phuket (Fig. 3.3), whereas three species were found from Phuket north to Ranong.     60 Interviewing trawl captains was the least expensive way to gather the most information about seahorse locations across the Andaman Sea (Table 3.2). In terms of effort, the trawl captain responses provided a larger sample size of trawling vessels and had more fishing effort per vessel (Appendix B.7). Trawl captains reported a greater range of seahorses captured per haul (Appendix B.7).   3.4.3 Comparison 3: Small-scale Fisher And Diver (Scientific And Citizen Science) Generated Data  The knowledge from small-scale fishers and diver sightings generated distribution maps (Fig. 3.4) at the genus level, but only diver sightings provided information for individual species. Small-scale fishers and citizen science contributions informed me that seahorses were found farther offshore and in deeper waters than the seahorse locations documented by scientific diving surveys (Table 3.1). My scientific diving surveys and citizen science contributions provided similar reports of species diversity (Table 3.1) and the dominant species observed in both datasets was H. comes Cantor 1849 (Table 3.1). In terms of sampling effort, the small-scale fisher data were less expensive and took fewer days to gather given equal staffing requirements (Table 3.2).  3.5 Discussion My results emphasize the most expeditious way to generate spatial data for data-poor marine fishes is through interviews with marine stakeholders (Johannes 2000; Gilchrist et al. 2005).  As with my fisher interviews for seahorses on the Andaman coast of Thailand, spatial data on rare or depleted marine fishes often exists in local knowledge (Gilchrist et al. 2005; Thornton & Scheer   61 2012). When urgency arises to ensure the sustainability of data-poor species, as in my study, local knowledge is cheaper to generate and provides a starting point to make inferences about species distribution at larger spatial scales (Moller and Berkes, 2004; Thornton and Scheer, 2012). Similar to other studies, I found supporting evidence that local knowledge was complementary to scientific data (Castellanos-Galindo et al., 2011; Hamilton et al., 2012), but challenging to verify at the individual species level (Golden et al., 2014; Turvey et al., 2015) .  However, my study is unique because I compared local knowledge to three different external datasets and found that in all cases, fisher interviews were the most cost-effective method to generate spatial data.   Despite its great capacity to support conservation needs as in my study and others, (Thornton and Scheer, 2012), one limitation of local knowledge is the lack of species-specific information, especially for hard-to-distinguish species (Turvey et al., 2014). Natural science methods (such as fishery research, scientific diving) or citizen science programs tend to be focused on species-level data, but take considerable amount of time and money to generate (Dickinson et al. 2010), as seen with my results. Local stakeholders may not initially be focused on gathering species level data (i.e. unaware of species differences or concentrating on other tasks) or may possess a folk taxonomy that distinguishes only amongst species that are commercially important (Beaudreau et al., 2011). However training stakeholders such as fishers to provide species level data forms the basis for many fishery dependent data collection methods (Morgan and Burgess, 2005; Stanley and Wilson, 1990). Indeed both commercial and small-scale fishers in Thailand record information on exploited species for Department of Fisheries research activities (R.Phoonsawat, pers.comm), and it may be feasible to train these fishers to identify and record   62 seahorse species. While training activities are underway, genus level data provided by local knowledge can still provide a basis for modeling species distributions that can be further refined once additional data become available (Van Strien et al. 2013).   My research comparing commercial and small-scale fisher knowledge differs from the usual practice of drawing mainly on small-scale fisher knowledge (Golden et al., 2014; Lunn and Dearden, 2006b).  In many countries data from commercial fisheries are generated from logbooks, landings, or vessel monitoring systems, and rarely come from commercial fishers themselves but see (Hall et al. 2009; Shepperson et al. 2014; Turner et al. 2015). Indeed local knowledge in a fishery context generally discusses knowledge from small-scale fishers (Anuchiracheeva et al., 2003; Lunn and Dearden, 2006b).  In terms of comparison to external datasets, my commercial fishers had sufficient knowledge to compare with fishery independent research as seen in other work (Hall et al., 2009; Shepperson et al., 2014). By contrast, I found small-scale fishers had more localized knowledge such as in (Anuchiracheeva et al., 2003; Lunn and Dearden, 2006b) and covered similar spatial scales to diver generated datasets.   In contrast to most research, my results show that comparing local knowledge to governmental research trawls decreased the uncertainty associated with both datasets (Hamilton et al., 2012; Lewis et al., 2009). Commercial trawl captains provided a more complete spatial picture of my data-poor fishes whereas the DoF research trawls were more clearly quantitative, providing individual species information. Despite the increased length of time to generate the DoF dataset, and its high expense, one area with extremely high catch despite a smaller sampling effort was identified.  Discrepancies in spatial or temporal data collection between fisher knowledge and   63 fishery research can increase the uncertainty in both datasets if the conclusions drawn are not comparable (Lewis et al., 2009; Usher, 2000). Such uncertainties have led to controversies surrounding the use of both local knowledge and scientific data to inform management (Bohensky and Maru, 2011; Usher, 2000).  These controversies highlight the importance of knowledge integration and assessment criteria to ensure conservation can proceed in the face of uncertainty (Bohensky and Maru, 2011; Hamilton et al., 2012).   My comparison of diver generated data confirms that citizen science volunteers can provide data similar to scientists about species-level occurrence (Crall et al., 2011). If citizen science programs are well established, then informing the creation of a management plan with such data may be acceptable because of a large sample sizes or identification of appropriate analytical techniques (Crall et al. 2011; Van Strien et al. 2013). The iSeahorse.org program, a global citizen science initiative and database was launched in October 2013. Therefore when Thailand underwent the initial CITES enforcement process, the option to inform management was unavailable (UNEP-WCMC, 2012). Citizen science programs are expensive (Dickinson et al. 2010). I coupled my iSeahorse citizen science activities with my scientific diving surveys to offset outreach costs, but it was still expensive. Two years of outreach efforts yielded a substantial amount of sightings (n=95) considering my scientific dataset (n=56) for the same area, over the same time period, failed to produce an equal number of sightings. However, compared to more established citizen science programs (e.g. ebird, 3.1 million bird sightings in North America for March 2012) it was a high cost with minimal return.     64 Five inherent biases may have influenced the complementarity of my datasets. First, by selecting to interview captains, I may have missed respondents with species level knowledge of seahorses because in some Thai fisheries (e.g. trawl fisheries), crew sort the fish and commonly set aside seahorses to sell in port. Second, there were large amounts of spatial and temporal variation in seahorse occurrence from the government research trawls, lending uncertainty to extrapolating seahorse locations at broad spatial scales. Third, challenges with detection of rare and cryptic species most likely biased scientific diving surveys, as documented in Aylesworth et al 2015. Fourth, both the scientific diving surveys and citizen science contributions are biased towards shallow depths because of recreational safe diving limits. Lastly, there may be quality control issues with species identification in the citizen science contributions, as documented in other research (Dickinson et al., 2010). However the iSeahorse.org program allows participants the opportunity to request assistance with identification or report seahorse species as ‘unknown,’ which may have helped to minimize this uncertainty.   From my research I can deduce that various types of data gathering are beneficial for different management and conservation objectives, as discussed in other studies (Beaudreau and Levin, 2014; Castellanos-Galindo et al., 2011). If you need to design and implement a species-specific management program (NOAA, 2014), my research suggests that local fisher knowledge or citizen scientists can provide an informed starting point for determining species distribution. Such information can be refined to a species level at a later time through more advanced methods (e.g. scientific or fisheries independent surveys). If your objective is to identify key habitat areas as part of large scale conservation planning (NOAA, 2014), then my research highlights that fisher knowledge is the most expeditious method to do so.  If the management objective is to   65 identify species threats, than my research, similar to others (Thornton and Scheer, 2012), supports that stakeholder (fisher) interviews are the most cost effective data collection method. In my case, information about threats at the genus level was relevant to all species, and highlighted where current management and conservation efforts should be focused (e.g. specific gear types). While I found that fisher interviews enabled me to identify a path forward with management action, I recognize that they may not be an appropriate starting point in all circumstances because of perception bias, shifting baselines or telling researchers what they want to hear (Panjarat and Bennett, 2012; Stake et al., 2005)  Taking the specific example of implementing CITES for seahorses in Thailand, my work exemplifies how spatial data can be used in the various stages of treaty implementation (CITES, 2012a; UNEP-WCMC, 2012). CITES focus is to ensure that export levels do not harm local populations. To implement CITES successfully, Thailand needed to ensure the sustainability of wild populations that initially it could not locate. In this case, had local knowledge been available, Thailand could have prioritized gathering species-specific data in these locations. Since such information was unavailable, Thailand underwent the CITES enforcement process, where recommendations were given regarding actions to take to ensure sustainability (CITES, 2012a). One of these recommendations included undertaking studies to provide evidence on seahorse spatial abundance and use this information to consider spatial area restrictions for fisheries (CITES, 2012a). My fisher knowledge provided spatial data at the genus level for where seahorses could be found, and also where fishing effort occurred. I am currently working to identify where management overlaps to evaluate if current management measures are appropriate to address the threats to local seahorse populations. Evident from my experience, is the   66 importance of spatial data on distribution, threats and management to supporting CITES implementation for parties trading in Appendix II species.   Meeting international conservation actions are typically implemented by national governments, but to ensure effectiveness requires both top down and bottom up approaches (Moller and Berkes, 2004; UNEP-WCMC, 2012).  With the example of seahorses in Thailand, the duty of CITES implementation for marine fishes lies within the Department of Fisheries. Currently Thailand is working to create a monitoring and adaptive management plan for seahorses based largely on the spatial data related to distribution, threats and current management. If new management measures are enacted, previous studies in Thailand suggest their success may be linked to consultation with commercial and small-scale fishers (Lunn & Dearden 2006; Panjarat & Bennett 2012). In these instances fishers disagreed with the timing of closed seasons, doubted government enforcement capabilities, or were unaware of management measures, leading to lower compliance and support for restricted areas (Panjarat and Bennett 2012). A key component for management effectiveness is promoting bottom up participation in natural resource governance (Berkes, 2009; Moller and Berkes, 2004), and encouraging expeditious data collection from resource users provides an opportunity for local input in to the management process.   My study supports the use of local knowledge to enable managers to act in the spirit of adaptive management for data-poor species (Johannes, 1998; Walters and Holling, 1990). Such knowledge can later be refined once more advanced data become available.  With continued species declines both on land and in the ocean (McCauley et al., 2015), natural resource   67 managers cannot afford to wait for the perfect dataset because the costs of inaction are high (Johannes 1998; Johannes 2000). The use of spatial data to evaluate conservation outcomes is growing in application for both international and national environmental agreements (Lorenz, 1995; Lorenzen et al., 2010). Indeed spatial data are critical to achieve conservation and management objectives such as monitoring populations, protecting habitats or regulating extractive activities (NOAA, 2014; Reid, 2007). I encourage others involved at the interface of science, management and policy to consider gathering local knowledge prior to establishing initial management plans, prioritizing areas for conservation, and evaluating management effectiveness.       68 Table 3.1 Seahorse species knowledge gathered from 1) fisher interviews, 2) DoF research trawls, 3) scientific diving surveys, and 4) citizen science contributions.   Depth in meters (mean) Majority reports inshore,  offshore or approximately equal Seahorse species (number) Comparison 1 Commercial fishers 5-120 (48)  equal ------------ Small-scale fishers 0-60 (14)  inshore ------------ Comparison 2 Trawl captains 0-110 (47) equal ------------ DoF research trawls 21-90 (46) offshore H.trimaculatus (524) H. spinosissimus (13) H. kelloggi (1) Comparison 3 Small-scale fishers 0-60 (14) inshore ------------ Scientific diving surveys 1.2-17.5 (5.4) inshore H. comes (31) H. kuda (17) H. spinosissimus (4) H.trimaculatus (2) H. kelloggi (1) H. mohnikei (1)   69  Depth in meters (mean) Majority reports inshore,  offshore or approximately equal Seahorse species (number) Citizen science contributions 3-32 (15.5) inshore H. comes (61) H. kuda (9) H. spinosissimus (8) H. histrix (6) H. kelloggi (4) H.trimaculatus (3) H .mohnikei (2) Hippocampus spp. (2)   70 Table 3.2 Sampling effort expended to generate the four spatial datasets on the Andaman coast, Thailand.  # sites visited Total landing or sampling sites  # days Interviews Total # of seahorses found # of staff deployed Amount spent per day (USD) Total cost for data generation (USD) Comparison 1  Commercial fishers 7 8 19 73 _________ 2 $141.98 $2,697 Small-scale fishers 19 621 21 120 _________ 2 $141.98 $2,981 Comparison 2  Trawl captains 6 7 15 45 _________ 2 $141.98 $2,129 DoF research trawls 22 44 180 _________ 538 10-15 $1000.00 $180,000   71 Table 3.2 (Con’t) Sampling effort expended to generate the four spatial datasets on the Andaman coast of Thailand Comparison 3   # sites visited Total landing or sampling sites  # days Interviews Total # of seahorses found # of staff deployed Amount spent per day (USD) Total cost for data generation (USD) Small-scale fishers 19 621 21 120 _________ 2 $141.98 $2,981 Scientific surveys 29 _________ 86 _________ 56 2 $177.75 $15,286 Citizen science contributions 26 _________ 158 _________ 95 4 $228.36 $36,080     72  Figure 3.1 I made three comparisons among four datasets to evaluate if fisher knowledge was the most cost-efficient to generate seahorse spatial data.      73  Figure 3.2 Commercial (a) and small-scale (b) fisher identified seahorse locations, along the Andaman Coast. A total of 86% (63of 73) of commercial and 96% (116 of 120) of small-scale fishers had spatial knowledge of seahorses. Darker areas are where more fishers identified locations of seahorses.   74  Figure 3.3 Locations where trawl captains and DoF research trawls captured seahorses. Trawl captains covered a larger area with their fishing effort, whereas DoF research trawls covered a smaller area but provided information on distribution of seahorses at the individual species level.  75  Figure 3.4 Small-scale fisher locations and diver sightings (scientific and citizen science contributions) of seahorses across the Andaman Coast. Diver sightings are reported by species with a) sightings of Hippocampus comes and b) all other seahorse species,  76 Chapter 4: Effects Of Indiscriminate Fisheries On Small Data-poor Species  4.1 Summary As catches of economically valuable fishes decline, and demand continues to increase, commercial and small-scale fishers retain and sell more non-target marine fishes. Some of these catches are destined for international markets and subject to international trade regulations. Many of these species are considered “data-poor” in that there are limited data on their biology, ecology and exploitation, which poses a serious management challenge for sustainable fisheries and trade. My research explores the relative pressure exerted by such indiscriminate fisheries on a data-poor marine fish genus – seahorses (Hippocampus spp.) – whose considerable international trade is regulated globally. My focus is Thailand, a dominant fishing nation and the world’s largest exporter of seahorses (Hippocampus spp.), where I gathered data by interviewing commercial and small-scale fishers and through port sampling of landed catch. I estimate that annual catches were more than three times larger than previously documented, approximating 29 million individuals from all gears. Three fishing gears - two commercial (otter and pair trawl) and one small-scale (gillnet) - caught the most individuals. Results from port sampling and my vulnerability analysis confirmed that H. kelloggi, H. kuda, and H. trimaculatus were the three species (of seven found in Thai waters) most vulnerable to fishing pressure. Small-scale gillnets captured the majority of specimens under length at maturity, largely due to catches of juvenile H. kuda and H.trimaculatus. This research indicates a role for vulnerability analysis in initiating precautionary management plans while more extensive studies can be conducted. My intention is to identify paths towards sustainable fisheries management when minimal data are available which, in the case of seahorses in Thailand, should focus on commercial trawling gears and small-scale gillnets.   77 4.2 Introduction The pressure on non-target species from fisheries remains a serious problem (Davies et al., 2009; Kelleher, 2005). Many non-target species - such as juvenile fish, shellfish and invertebrates - are incidentally captured in fishing gears around the world (Kelleher, 2005). In many fisheries non-target species, once considered a nuisance by fishers, are becoming incorporated as part of target catches (FAO, 2012). The availability of new markets are driving this practice because non-target species are being used to make products such as animal feeds or fertilizers (FAO, 2012). These markets provide additional revenue for fishers and little incentive to reduce amounts of non-target catch, turning these fisheries into indiscriminate fisheries (FAO, 2012). As a result, both commercial and small-scale fishers are taking and utilizing an increasing number of species, continuing the pressure on non-target species. With the additional threats to the marine environment such as coastal run-off, sedimentation, and changes in water quality (Halpern et al., 2008), maintaining healthy populations of target and non-target species alike remains an elusive task (Costello et al., 2012).   Because non-target fish enter markets from small-scale and commercial fishers, some of which fall under international regulation (UNEP-WCMC, 2012), countries must determine how to manage such non-target fisheries sustainably (CITES, 2013a). The sustainable management of national fisheries must include the social, economic and biological nature of fisheries (Gerrodette et al., 2002; Salomon et al., 2011).  The intersection of national fisheries management, species conservation, and international trade becomes even more complicated because of issues linked to national sovereignty, conflicting mandates, and multi-agency and organization involvement in these issues (Cullet and Kameri-Mbote, 1996; Joseph, 1994; Vincent et al., 2013). Every country   78 has the right to manage its own natural resources but the vast majority of countries have also agreed to international treaties related to sustainability, such as the Convention on International Trade in Endangered Species (CITES), (CITES, 2013a). By agreeing to international treaties countries have accepted the obligation to implement these treaties at the national level (Joseph 1994).   For countries where national fisheries catch marine species (target or non-target) destined for international markets, balancing social, economic, and international obligations is mandatory, yet difficult. For example under the CITES convention, signatory countries must demonstrate that export of any Appendix II listed species is not harmful to the local populations (CITES, 2013a). If a country is unable to produce such information, it can lead to a ban on exports of that species from that country (CITES, 2013a). Bans on the exports of wildlife products typically do not solve issues related to sustainability, and can even have negative consequences such as forcing the trade underground (Cooney and Lang, 2007; Rivalan et al., 2007). However, implementing the necessary changes within a country to address sustainability issues with any wildlife trade is a slow and expensive process (Nijman, 2010). For the few marine species listed on the CITES convention validating the process and necessary data requirements needed to determine harm to local populations is still underway (UNEP-WCMC, 2012; Vincent et al., 2013).   Determining fisheries management solutions to address increasing pressures on non-target species is challenging, primarily due to lack of data for rigorous analysis (Reuter et al., 2010). Many fisheries remain un-assessed and unmanaged because natural resource managers lack the   79 data to perform both simple or complex stock assessments (Costello et al., 2012; Honey et al., 2010). Typical fisheries management at the very least requires information on catches and fishing effort (Walters and Martell, 2002), which can be hard to come by for data-poor species. Stock assessments typically guide the fisheries decision-making process by providing an understanding of fish population dynamics in relation to fishery input (e.g. number of fishing vessels) or output controls (e.g. total allowable catch, quotas) (Sparre and Venema, 1998). However, even fish stocks with the necessary data to perform various stock assessments can be mismanaged or suffer from poor data quality (Myers et al., 1997; Walters and Maguire, 1996). With decreasing abundances of many marine species (both large and small) (McCauley et al., 2015), waiting for data to become available may be detrimental to sustainable management (Johannes, 1998; Walters and Holling, 1990).  How then should management decisions be made for data limited species?   Several new methods for dealing with data-poor stocks (Honey et al., 2010; Patrick et al., 2009; Reuter et al., 2010) have been proposed recently but have yet to be evaluated in the CITES context. CITES data requirements at the very least require information on species and fisheries (CITES, 2013a; UNEP-WCMC, 2012). Specifically to identify pressure from fisheries CITES recommends looking at the (i) diversity of fishing methods / gears, (ii) fishing mortality, (iii) fishing selectivity, (iv) discarding practices and (v) indicators of fishing impacts. However, there is little guidance on the spatial or temporal nature of such data or appropriate analytical techniques required to evaluate the pressure on species from fisheries (CITES, 2013a).  A recent review of data-poor fishery assessment methods outlines various methods based on data richness (Honey et al., 2010), and ranges from stock assessment methods to analyze trends and set catch   80 limits (Carruthers et al., 2014; Rago and Sosebee, 2010), to vulnerability analyses (Brown et al., 2013; Patrick et al., 2009) and extrapolation methods (Smith et al., 2009). The two least data intensive options are extrapolation and vulnerability analysis (Honey et al., 2010). Extrapolation methods require no scientific data, and instead incorporate fisher knowledge and scientific research from “sister” species to inform a starting point for fishery managers (Honey et al., 2010; Smith et al., 2009). Vulnerability analyses such as the Product and Susceptibility Analysis (PSA) assess a stock’s vulnerability to fishing based on life history characteristics and what is known about the fishery, and are commonly used to evaluate multiple bycatch stocks (Brown et al., 2013; Patrick et al., 2009; Stobutzki et al., 2001). This method has the potential to be a good compromise in the CITES context between the more data intensive assessment methods such as stock assessment and the least data intensive methods like extrapolation (CITES, 2013a; Honey et al., 2010).  The future of seahorses (Hippocampus spp.), a marine fish genus captured as a non-target animal in many fisheries around the world, lies at the nexus of conservation, fisheries and international trade. Traded internationally for use in traditional medicine and aquariums (Vincent et al., 2011), seahorses were the first marine fishes to be listed on CITES Appendix II since its inception, and the first marine fishes on the Convention to have countries undergo the enforcement process (UNEP-WCMC, 2012; Vincent et al., 2013). As part of the enforcement process, any country exporting an Appendix II species must understand if trade is harmful to wild populations (UNEP-WCMC, 2012). CITES trade records indicate seahorse trade volumes range between 3.3 and 7.6 million individuals annually, but most likely under-estimate actual trade volumes due to reporting challenges (Foster et al., 2016). From a fisheries perspective, annual estimates of   81 bycatch suggest that more than 37 million seahorses are captured by global fisheries each year (Lawson, Foster & Vincent, in press). The trade of seahorses is global and complex, involving 25 of 41 known species, and more than 80 countries (Foster et al., 2016). However, half of species in trade (n=20 of 41) (Lourie et al., in press) are listed as ‘Data-Deficient’ by the IUCN, meaning there is not enough information about their population status, distributions, or threats to evaluate extinction risk (IUCN, 2015). Because of the limited global seahorse knowledge, determining at the national level if seahorse exports are harmful to local populations remains challenging (CITES, 2013a).   As the world’s largest exporter of seahorses, Thailand has undergone the CITES enforcement process twice, to determine if their seahorse exports were harmful to wild populations. Thailand has faced challenges related to insufficient data regarding the seahorse species and fisheries involved in the trade, making an evaluation of impact to local populations difficult (CITES, 2013a; UNEP-WCMC, 2012). CITES export records, despite their short comings, indicate that Thailand accounts for 75% of global seahorse exports with annual estimates between 3 million and 6 million seahorses from 2004-2012 (Foster et al., 2016). Seven out of the fourteen seahorse species found in Southeast Asia live in Thai waters, with five of them reported in trade (Perry et al., 2010). All five trade species have IUCN Red List global and national assessments documenting a ‘vulnerable’ risk of extinction based on high species trade volumes coupled with habitat degradation and threat of overfishing for seahorses in general (IUCN, 2015; Vidthayanon, 2005). Importantly, little is known about how seahorses in Thailand are influenced by fisheries. Thailand is home to many small-scale and commercial fisheries representing a range of fishing gears (FAO, 2012; Lymer et al., 2010) and catch of seahorses for dry trade have been   82 documented in both types of fisheries (Laksanawimol et al., 2013). Previous research in the late 1990s estimated Thailand’s trawl fleet caught approximately 2 million seahorses per year, but this study did not investigate catch rates for any other commercial or small-scale gears (Perry et al., 2010). Identifying appropriate methods to evaluate Thai fisheries with minimal data in relation to local seahorse populations will support Thailand to identify which species and gear to focus management measures.  The goal of my research was to determine current annual estimates of a non-target fish genus and determine which fishing gear exerts the most pressure on these data-limited fishes. Additionally I explore the use of vulnerability analysis as a data-limited fishery assessment method for closely related bycatch species and evaluate its usefulness against results from fisher interviews and port sampling. Finally, I evaluate my results in the CITES context to identify management measures in support of a country evaluating the sustainability of its marine fish exports.  4.3 Material And Methods To better understand how indiscriminate fisheries affect a data-limited fish genus, I interviewed fishers and executed port sampling of seahorses throughout coastal Thailand. I conducted my research with the support of the Thailand Department of Fisheries (DoF), seeking input on where to collect data based on locations representative of local fishing activities. I first describe the methods used to collect the data, and then describe the analyses used to identify initial management action for fisheries catching seahorse in Thailand.     83 4.3.1 Data Collection  To identify what gears and seahorse species were involved in seahorse bycatch, I conducted semi-structured fisher interviews in 11 provinces on both the Andaman (n=6) and Gulf (n=5) coasts of Thailand. I interviewed a total of 306 (137 commercial, 169 small-scale) fishers from both coasts. Since commercial and small-scale fisheries have defining characteristics in Thailand (Lymer et al., 2010), I expected their annual bycatch rates of seahorses to differ. I focused my sampling efforts on four types of commercial gears (otter trawl, pair trawl, purse seine, pushnet) and two types of small-scale gears (gillnet, cage) based on recommendations of dominant gear types from the DoF.   To select respondents, I targeted two sets of landing locations: fishing ports for commercial fishers and coastal villages for small-scale fishers. Commercial fishing port locations were identified as ports located near government port facilities, whereas landing locations in coastal villages were typically on local beaches or in an estuary arm near fishers’ houses. I visited a total of 13 ports (Andaman = 7, Gulf = 6) and 28 coastal villages (Andaman = 21, Gulf = 7) (Figure 4.1). I determined the number of fishers to interview in each location as either 10% of the estimated total number of fishing boats landing catch at that location or the saturation method (Tobias, 2010) given my need to maximize the number of landing locations visited. I defined saturation as the point where if the 6th, 7th, and 8th, interviews introduced no more new information on gear type, fishing grounds, or seahorse locations, then saturation had been achieved. I additionally confirmed saturation through my observations of gear type at each landing location. All interviews followed UBC Human Ethics Protocols (H12-02731).    84 During semi-structured interviews (Stake et al., 2005) I asked fishers about the frequency and duration of their fishing activities. Fishers were asked to describe gears, depths fished, number of hauls (or sets) per day, and typical fishing duration per day, month and year. I asked fishers to identify the gears they use that capture seahorses, along with numbers and frequency of catch. I offered fishers the opportunity to report seahorse catch by haul, day, fishing trip, week, month or year. Fishers could report catch for more than one effort estimate, for example some fishers report mean seahorse catch per day and also per month. This type of reporting would enable me to scale catches up to annual estimates based on additional information gathered throughout the interview on number of hauls per day, total days fished per month, and total months fished per year.  To identify what gears and seahorse species were involved in seahorse bycatch, I conducted port sampling opportunistically while at landing locations conducting fisher interviews. When I encountered seahorses in a fisher’s catch, the species, size, and sex were determined from all available individuals in the catch. Seahorses were then returned to the fisher to retain, discard or sell.   To identify the total number of registered fishing vessels by gear type, I accessed the most recent fisher census data from the Thailand Department of Fisheries (www.platalay.com/boatsurvey2554/index.php). Within the census data, small-scale and commercial fishing gears are categorized by gear type and reported by province. I summarized the available numbers for my six focal gears – (i) otter and (ii) pair trawls, (iii) purse seines, (iv)   85 pushnets, (v) gillnets (including surrounding nets) for fish, crab and shrimp, and (vi) cages (for fish, crab and squid)- at the provincial level to create national totals for each gear type.   4.3.2 Data Analysis First I calculated annual catch rates for each fisher by scaling up his/her reported mean catch rate with his/her reported frequency of fishing (e.g. daily catch rate x total days fished per month x total months fished per year). For those fishers who reported more than one catch rate (e.g. fishers reporting catch rates per day and per month) (n=42), I selected the catch rate for the longer period of time (e.g. per month) because many fishers upon reflection (n=40/42), adjusted their estimated catch rate for the longer period of time to account for zero-catches.   Calculating annual catch by gear type was a two-step process. Because there was high variation in fisher reported catch rates, I first determined the percentage of fishers reporting 0 yearly catch and made the assumption that the fleet (for each gear) had the same proportion of zero catch.  For the positive catches per year (where fishers reported catch rates > 0), I used the mean catch rate per vessel, after exploring the potential for recall bias based on reporting time slice (Aylesworth and Kuo, in prep) and validation with external data (Aylesworth and Kuo, in prep).  Second, to calculate yearly catch I multiplied the mean catch rate per vessel by the proportion of the fleet size with positive catches. Additionally I summed each annual catch estimate for all gear types to determine a national estimate of seahorse catch.    I made four key assumptions in the development of my Thai national estimate of seahorse catch. First, I assumed that fishers had a consistent catch for all seasons. Second, I assumed that   86 individual fisher catch rates did not differ spatially with fishing ground location.  Third, I assumed that fishers were reporting current effort and catches. Fourth, I assumed that the number of fishers reporting zero catch per gear type was proportional to the number of vessels across the entire fleet with zero catch.  To identify which gear catches the most seahorses and which species was caught the most, I summarized the port-sampled data by gear type and species. I then compared the results from port sampling to annual seahorse catch rates per vessel from fisher interviews. Additionally I looked for indicators of overfishing by gear type and for species by calculating the majority of sampled catch under length at maturity, and the presence of sex bias in the catch. Identifying the gear with the largest percentage of catch under height at maturity can be an indicator of potential overfishing (Froese, 2004), as can sex ratios significantly different from unity (Rowe and Hutchings, 2003). I sourced heights at maturity from Froese and Pauly 2016 for H. histrix, H. kelloggi, and H. kuda and from Lawson et al 2015 for H. spinosissimus and H. trimaculatus. I used a chi-squared test to identify if sex ratios by gear type and species were significantly different from unity (p < 0.05) (Zar, 1999).   To further understand the pressure exerted on my non-target fish genus by indiscriminate fisheries, I executed a vulnerability analysis, one of the recommended data-limited fishery assessment methods (Honey et al., 2010). I modified a productivity-susceptibility analysis (PSA) (Patrick et al., 2009; Stobutzki et al., 2001), based on the data available for seahorses in Thailand, which enabled me to incorporate life history of each species with the data available from fisher interviews and port sampling. A typical productivity-susceptibility analysis ranks   87 multiple species with respect to susceptibility to fishing pressure and the stock’s capacity to recover after depletion (typically life history characteristics) (Stobutzki et al., 2001). The rank represents the stock’s relative capacity to sustain fishing pressure and can be used to prioritize fisheries research and management (Stobutzki et al., 2001). This method of analysis is growing in its use to assess data-poor bycatch species (Brown et al., 2013; Patrick et al., 2009). I piloted this method for its relevance to small con-generic fishes, since typically PSA’s compare vulnerability for species with a wide breadth of life history characteristics (e.g. gobies to sharks) (Patrick et al., 2009; Stobutzki et al., 2001). Additionally I wanted to test this method to compare what I could deduce about prioritizing conservation action from quantitative analysis of fisher interviews and port sampling of landed catch.   I identified life history attributes of the five seahorse species represented in port sampling from Froese and Pauly 2016 or new literature (Lawson et al 2015) for the following characteristics: population growth rate, maximum length (Lmax), maximum asymptotic length (Linf), von Bertalanffy growth coefficient (k), natural mortality (M), life span, length at maturity, age at maturity, and mean trophic level (Table 4.1). I then created fishery pressure attributes based on the data from my analysis of catch rates, annual fleet catch, catch under length at maturity and sex ratios (Table 4.2). As commonly done in productivity and susceptibility analysis (Patrick et al., 2009; Stobutzki et al., 2001), I divided the range in attributes of my five species for both life history and fishing pressure into three categories for each variable with scores representing low (1), moderate (2), and high (3) risk (Table 2). I anchored scores around median values, which were assigned to the moderate (2) risk category (Patrick et al., 2009; Stobutzki et al., 2001), and   88 weighted all attributes equally. Then I summarized the values and divided by the number of total attributes, to determine an average score from 1 to 3 for life history (1 = low risk, 3 = high risk) and fishing pressure (1 = low pressure from fishing, 3 = high pressure from fishing) (Table 2) (Patrick et al., 2009; Stobutzki et al., 2001).  4.4 Results Based on fisher interviews, mean catch rates per vessel per year ranged from 106 to 6,534 seahorses (Table 4.3).  Otter and pair trawl gears had the highest catch rates per vessel per year - in the thousands - (Table 4.3) out of all fishing gears. Commercial pushnets, purse seines, and small-scale cages had the lowest catch rates per year (less than 200) (Table 4.3). Out of small-scale fishing gears, gillnets had catch rates more than two times larger than cages (Table 4.3). All gears had high variation in reported in catch rates per vessel per year (Table 4.3).  I estimated that the annual number of seahorses captured by Thai fishing gears was 29.4 million seahorses (95% confidence interval: 5.0 – 55.9 million) (Table 3).  Annual estimates of seahorses by fishing gear type ranged from 78,440 (pushnets) to 13,614,696 (otter trawls) (Table 4.3). Pair trawls and gillnets both had large annual estimates (in the millions), where as purse seines, pushnets and cages all had annual estimates less than 500,000 (Table 4.3).   Port sampling results confirmed the results of fisher interviews that otter trawls, pair trawls, and gillnets caught the largest amount of seahorses (Table 4.3). A total of 498 (n=251 Andaman, n=247 Gulf) seahorses were port sampled from 55 (n=25 Andaman coast, n=20 Gulf coast)   89 fishers. Purse seines, pushnets, and cages had the lowest sample sizes of port sampling – with each gear having five or fewer individuals.   In terms of indicators of overfishing, the port sampling results by gear type found that the percent of sampled catch under length at maturity ranged from 0 to 50% (Table 4.3). Small-scale fishing gears had the largest percentage of catch under length at maturity (Table 4.3). For commercial gears, purse seines (20%) and pair trawls (15%) caught the highest amounts of catch under length at maturity, whereas pushnets had the lowest (0%) (Table 4.3). No gears had sex ratios significantly different than unity (all p > 0.05) (Table 4.3).   I observed five seahorse species during port sampling, H. histrix, H. kelloggi, H. kuda, H. spinosissimus, and H. trimaculatus (Table 4.4). Port sampling results by species indicate that most seahorses in sampled catches were H. trimaculatus (Table 4.4). I observed H. histrix the least frequently (Table 4.4). H. spinosissimus and H. trimaculatus were observed in the most number of fishing gears (Table 4.4), where as H. histrix was observed in the least number of fishing gears. H. histrix, H. kelloggi, H. spinosissimus, and H. trimaculatus were observed most in trawl gears, whereas H. kuda was most observed in gillnet catches (Table 4.4). The percentage of sampled catch under length at maturity by species ranged from 0 (H. histrix) to 66% (H. kuda) (Table 4.5). Gillnets caught the most immature H. kuda and H. trimaculatus whereas immature H. kelloggi and H. spinosissimus were observed most in trawls (Table 5). Only H. trimaculatus had a sex ratio statistically different than 1, where more females were captured than males (X2 = 5.8384, d.f. = 1, p-value = 0.01568, Table 4.5). For additional analyses of catch under length at maturity and sex ratio by species by gear type, see Appendix C.1.   90  My modified productivity and susceptibility analysis revealed that H. kelloggi, H. kuda and H. trimaculatus are the three species most vulnerable to Thai fisheries (Fig 4.2) after taking into account life history traits. By comparing only the fishery pressure attributes of these species by gear type I identified that H. kelloggi is under the largest pressure from otter trawls where as H. kuda is under the largest pressure from gillnets (Fig 4.3). H. trimaculatus is the species under the most pressure from two types of gears – otter trawls and gillnets (Fig 4.3).  4.5 Discussion My research shows with a Thai example that both commercial and small-scale indiscriminate fisheries have large impacts on non-target species. I estimated the mean annual catch of Hippocampus spp. in Thailand at 29.4 million individuals, more than three times higher than previous estimates for Thailand or indeed for some estimates of global catch (Foster et al., 2016; Lawson et al., in press). My research confirms the growing literature that small-scale fisheries exert fishing pressure greater than or equivalent to commercial gears (e.g. Peckham et al. 2007; Shester & Micheli 2011). My results from port sampling and vulnerability analyses both identified three species particularly susceptible to fisheries, whereas other studies have sometimes found results from these methods contradictory (FAO, 2010a). The data-limited fishery assessment I employed enabled me to gain confidence in results from my port sampling analysis, and provided guidance to initiate a precautionary management plan, as found in other studies (Brown et al., 2013; Tuck et al., 2011). However, one critical drawback, as commonly cited with this method, was that I was unable to determine species specific targets or reference points to ensure sustainable management (FAO, 2010a; Tuck et al., 2011).    91  My research confirms that commercial indiscriminate fisheries in Thailand catch large annual numbers of seahorses, which are just one of the many Thai non-target fish genera (Lymer et al., 2010). My annual estimates from the Thai otter-trawl fleet (13.6 million seahorses / year) are much larger than previous otter trawl estimates in Thailand (~2 million / year) (Perry et al., 2010), most likely because my study had larger sample sizes. Indeed the annual estimates from the commercial trawl fleet are also larger than estimated seahorse trade volumes from Thailand or any other country (Foster et al., 2016). My large annual catch estimates from Thailand call into question the latest global estimate of seahorse bycatch (37 million individuals) because my commercial estimates are larger than those for all of Thailand (~9 million individuals) cited in that study (Lawson, Foster & Vincent, in press).  In my study not all commercial gears exerted strong pressures on seahorses. Despite documentation of the destructive nature of pushnets operating in coastal waters (Changsang and Poovachiranon, 1994; Tokrisna et al., 1997), I found that the annual capture of seahorses from pushnets was comparatively low to other gears both in terms of catch rate and fleet size. I found the Thai purse-seine fleet had moderate catch rates and fleet size, despite the fact that bycatch from this gear type is not well recorded in the seahorse literature (Lawson, Foster & Vincent, in press).   Small-scale fishing gears are typically considered as benign compared to destructive fishing gears such as trawlers (Kaiser et al., 2002), but my study supports recent research indicating that small-scale fisheries can have large impacts (Shester and Micheli, 2011). My study expands on research in Malaysia (Lawson et al., 2015) documenting that seahorses are captured incidentally from small-scale gears, with annual estimates on par with commercial gears such as pair   92 trawlers. Small-scale fishers in Thailand have four times more registered fishing boats than the commercial fleet (DoF 2015), an attribute which led to annual seahorse catch estimates to be equivalent or greater than some commercial gears. Such results are similar to the pressures exerted on sea turtles by small-scale fishers in Mexico, where catches per trip were low at the individual level, but the vast number of participants in small-scale fisheries lead to large annual catches (Peckham et al., 2007). Official catch statistics in Thailand, as in many countries, often under-report small-scale fisheries catches and fail to include small-scale gears in management plans (Teh et al., 2015).  The large number of small-scale catches in my study, along with previously documented perceptions and motivations of small-scale fishers in Thailand, highlight that the inclusion of small-scale fishers is critical for successful fishery management (Bennett and Dearden, 2014; Lunn and Dearden, 2006a, 2006b; Panjarat and Bennett, 2012).   I found overfishing indicators (Froese, 2004; Rowe and Hutchings, 2003) a useful tool to assess pressure from fishing gears. My study found large proportions of juvenile seahorses in small-scale fishing gears but not commercial ones, suggesting that small-scale gears are less selective than their commercial counter parts for juvenile and immature fishes (Foster and Vincent, 2010; Watson et al., 2006). One explanation for this divergence may be that small-scale fishing gears in Thailand tend to fish in coastal waters that act as refuges for young fishes where as commercial gears fish farther from shore (Lymer et al., 2010). Although not well-documented to inhabit coastal areas (Lourie et al., 2004) H.trimaculatus and its coastal congener H. kuda were the species most susceptible to capture under length at maturity in small-scale gears. My second overfishing indicator, sex bias, did not identify any particular gear type at risk for overfishing but was useful at the level of species, as seen in Malaysia (Lawson et al., 2015). Similar to my study,   93 sex ratios of H. trimaculatus catches in the south-west region of Peninsular Malaysia were female biased (94% of samples were female) (Lawson et al., 2015). However, other studies of exploited seahorses found male biased ratios (Baum et al., 2003), and it was hypothesized that pregnant males may be more susceptible to capture because they are less likely to expend energy avoiding fishing gears (Baum et al., 2003). For seahorses a bias in sex ratio be it for males or females is equally undesirable because seahorses form pair bonds and are monogamous through the breeding season (Foster and Vincent, 2004). Therefore losing a mate, be it male or female, to fishing pressure may alter the reproductive success of these fishes (Foster and Vincent, 2004).  My research confirms there is value to modifying vulnerability analyses to match available data (Brown et al., 2013; Tuck et al., 2011). My results modified the classic first paper on productivity and susceptibility analysis by drawing on the attributes available for my fishes related to productivity, and then creating my own susceptibility attributes based on available fishery data (Stobutzki et al., 2001). Species-specific modifications made to vulnerability analyses, such as this one with seahorses, have been deemed successful for other groups of species such as seabirds and marine mammals (Brown et al., 2013; Tuck et al., 2011). The procedure I used to generate data for the vulnerability analysis – using life history characteristics from Froese and Pauly 2016 and data from several months of interviews and port sampling – could easily be replicated for other groups of fishes where more sophisticated data is lacking. This suggests vulnerability analysis may be an acceptable compromise between data-rich fishery stock assessments and extrapolation from sister groups of species (Honey et al., 2010)    94 The vulnerability and port sampling analyses enabled me to gain a better understanding of the diversity of fishing gears exerting pressure on seahorses and evaluate potential indicators of fishing impacts (CITES, 2013a). These are two important components for evaluating the pressures of fisheries and trade on local populations of marine fishes listed under CITES (CITES, 2013a). In the spirit of adaptive management (Walters and Holling, 1990), my results indicate the value of focusing on commercial trawl and small-scale gillnet gears since they captured the largest numbers and exerted pressure on the most vulnerable species (H. kelloggi, H. kuda, H. trimaculatus). Incidentally, H. histrix and H. spinosissimus would benefit from management measures aimed at trawling gears, even though these species scored the lowest in terms of pressure from fishing gears. Given that spatial management is one of the best options for trawl fisheries management, Thailand is potentially well situated because it already has spatial management measures (e.g. marine national parks, no trawl zones) in place (CITES, 2013a). Evaluating management effectiveness is the next step in the CITES enforcement process (CITES, 2013a) and should be addressed in future papers given the complex nature of fisheries in Thailand (Janekitkosol et al., 2003; Tokrisna, 2006). However, for other countries this process may be much simpler if few fishing gears are involved in the capture and subsequent trade of CITES-listed species.      One critical drawback of my vulnerability analysis was that it did not provide information on how fish populations had changed over time in response to fishing pressure, one of the key recommendations given to Thailand by CITES (CITES, 2012b). Identifying such a change over time requires information about the current stock status (Honey et al., 2010), and with my limited data of seahorses and fisheries in Thailand I could not achieve this result. Even if I had this type   95 of information, I most likely would have used a different fishery analysis tool (Honey et al., 2010). In addition to a lack of understanding about current stock status, another drawback of vulnerability analysis is that previous research in Thailand has called into question the validity of this method because it failed to highlight the risk of an overfished stock (FAO, 2010a). In order to gain confidence in the results of the vulnerability analysis from a Thai management perspective and meet one of the key CITES recommendations, fisheries modeling and simulations will need to be done, requiring additional skill and expertise. However, Thailand can begin to evaluate the effectiveness of current management measures with the knowledge provided by the vulnerability analysis and draft a precautionary management plan – the final recommendation from CITES (CITES, 2012b; Walters and Holling, 1990). An adaptive management plan with frequent monitoring would enable further evaluation of management effectiveness while additional data are collected (Walters and Holling, 1990).   As found in my study, both small-scale and large scale fisheries can have significant impacts on fish populations (Kelleher, 2005; Shester and Micheli, 2011). Resource managers need flexible tools to initiate sustainable management plans for both small and large scale fisheries as well as for target and non-target species (Honey et al., 2010). Creative solutions that use available or easy to generate data should be considered in all contexts, because with continued marine species declines, I cannot afford to wait for better data (Johannes, 1998; McCauley et al., 2015). Such solutions are not easy to identify or implement given the challenges of data-limited fisheries, non-target fisheries, and new markets increasing demand for all species captured by fishing gears (FAO, 2012; Honey et al., 2010). The path forward to sustainable fisheries may lie in simple   96 assessments like vulnerability analyses that support the development of precautionary management plans while longer term data are collected.    97 Table 4.1 Life history indicators for five Thai seahorse species (Hippocampus spp).  All indicators taken from Froese and Pauly 2015 (fishbase.org) except for those heights at reproductive maturity estimates indicated with *, which came from Lawson et al 2015.   Hippocampus sp. N (from port sampling) Population growth rate (r) Max height (cm) Linf (cm) von Bertalanffy growth coefficient (k) Nat. Mort. (M)   Maximum age (years) Height at reproductive maturity (cm) Age at first maturity (years) Mean trophic level H. histrix 2 5.72 17.0 18.0 0.81 1.61 3.5 11.2 1.0 3.5 H. kelloggi  67 4.58 . 28.0 29.4  1.04 1.95 2.7 15.0 0.7 3.4 H. kuda 71 12.3   17 21.9 2.47 3.93 1.1 14.0 0.3 3.6 H. spinosissimus  97  5.19  22.7  23.9   0.97 1.86 2.9  12.3* 0.7 3.4 H. trimaculatus  261 5.97  22.0 23.2 1.01 1.83 2.9  12.1* 0.8 3.8    98 Table 4.2 Productivity and susceptibility attributes and rankings (modified from Patrick et al 2009).  Attribute  Low Risk (Score 1) Moderate risk (Score 2) High risk (Score 3) Life history  “productivity” Population growth rate (r) (1/year) > 8 5-8 <5  Maximum height (cm)  < 20  20-25 > 25  von Bertalanffy growth coefficient (1/year) >2 1-2 <1  Estimated natural mortality (1/year) <2  2-3  >3   Maximum age (years) < 2 2-3 > 3  Age at maturity (years) < 0.4  0.4-0.8  > 0.8   Mean trophic level < 3.5 3.5 >3.5      99 Table 4.2 (Con’t) Productivity and susceptibility attributes and rankings (modified from Patrick et al 2009). Fishing pressure “susceptibility”  Low (1) Moderate (2) High (3)  % of port sampled catch < 25% of port sampled catch 25% < port sampled catch < 50% > 50% port sampled catch  % catch under length at maturity < 33% catch under length at maturity 33% < catch under length at maturity < 67% Catch under length at maturity > 67%  Sex bias in catch  No --- Yes  Fleet size < 1000 1000-10,000 >10,000  Mean catch rate (per year for 1 vessel) < 200 200-1000 >1000     100 Table 4.3 Data summary of port sampled seahorses from commercial and small-scale fishing gears in coastal Thailand.  Gear type # fishers interviewed (n) Fisher reported mean catch per vessel / year (95% CI) Fleet size Estimated yearly mean total catch (95% CI) Number of port sampled seahorses  % sampled catch under length at maturity  Sex ratio (proportion male) Commercial Otter trawl 38 5472  (0-11,446) 2553 13,614,696 (0 - 28,478,688) 195 10.2 0.43 Pair trawl 48 6534  (0-14,055) 912 5,821,929 (0 – 12,523,612) 148 15.5 0.43 Purse seine 38 188  (0-400) 1774 207,750 (0-442,014) 5 20.0 0.5 Pushnet 10 106  (0-237) 1233 78,440 (0 – 175,489) 2  0 0.5    101 Table 4.3 (con’t) Data summary of port sampled seahorses from commercial and small-scale fishing gears in coastal Thailand.  Gear type # fishers interviewed (n) Fisher reported mean catch per vessel / year (95% CI) Fleet size Estimated yearly mean total catch (95% CI) Number of port sampled seahorses  % sampled catch under length at maturity  Sex ratio  (proportion male)   Gillnet 260 411  (231-590) 30,952 8,911,286  (5,025,057 -12,797,515) 146 47.2 0.51 Cage 51  191 (17-366) 6,410 448,700  (30,043 – 858,764) 2 50.0 0.5 Total  306 --- 43,834 29,418,249 (5,095,790 -55,914,882) 498 22.8 0.45       102 Table 4.4 Port sampling results (number of individuals) by species by gear type.  Gear type Commercial Small-scale  Species Otter trawl Pair trawl Purse seine Pushnet Gillnet  Cage Total H. histrix 0 2 0 0 0 0 2 H. kelloggi 38 27 0 0 2 0 67 H. kuda 2 1 0 1 67 0 71 H. spinosissimus 38 41 1 0 15 2 97 H. trimaculatus 117 77 4 1 62 0 261 Total 195 148 5 2 146 2 498  Table 4.5 Percent catch under length at maturity and sex ratio by species.  % sampled catch under length at maturity Sex ratio H. histrix 0 0.5 H. kelloggi 28.3 0.43 H. kuda 66.1 0.56 H. spinosissimus 18.5 0.48 H. trimaculatus 11.4 0.41   103  Figure 4.1 Locations of fisher interviews with commercial and small-scale fishers along Thailand’s coast.    104  Figure 4.2 Vulnerability analysis comparing pressure from fishing and the population’s ability to recover amongst seahorse species observed in port sampling.    105  Figure 4.3 Susceptibility scores by fishing gear for the three seahorse species most vulnerable to fisheries in Thailand.     106 Chapter 5: Developing A Data-poor Species Stock Assessment Using Seahorses As A Case Study 5.1 Summary While more management attention is being paid to the effects of fishing on non-target species, the challenge remains substantial and largely unresolved. Many non-target species are considered data-poor, which poses challenges for effective management.  Data-poor stock assessment methods are available to managers, but how useful are they? I developed the first stock assessment of seahorses – which are commonly landed as non-target bycatch – using existing and newly acquired data, and explored its utility in identifying management measures that might ensure the persistence of these species. My results suggest that seahorse populations can only sustain low fishing mortality rates before declines occur.  I also found that the main seahorse species in Thailand (H. trimaculatus) is overfished, in part because of fisheries-related habitat declines that affected recruitment. Existing recommendations such as a minimum size limit could be important for lowering recruitment to the fishery, a key model parameter. I found that even for these data-poor fisheries, a stock assessment had real benefits in allowing me to evaluate management measures, past and proposed. My analysis conferred confidence that the notable declines in seahorse abundance in Thailand since at least the mid-1990s are real.  I also discovered that a combination of vessel reductions with spatial and temporal closures would be the most effective at slowing future declines.  In doing a data-poor stock assessment for seahorses, I determined that it could also be useful for other non-target, data-poor fishes.    107 5.2 Introduction Ensuring the population health of non-target species that are landed in nonselective fisheries remains an elusive task (Costello et al., 2012). Many non-target species - such as juvenile fish, shellfish and invertebrates - are unintentionally captured in fishing gears (bycatch) around the world (Kelleher, 2005). In addition to direct depletion of species populations, fishing activities put indirect pressure on non-target species by causing habitat damage (Shester and Micheli, 2011), and by leaving abandoned or lost nets to continue ghost fishing (FAO, 2012). The importance of bycatch issues has been highlighted by their incorporation into the various seafood sustainability criteria, to a point where many fisheries have failed to achieve ‘best choice’ status because of problems with bycatch (Pelc et al., 2015). Internationally, numerous regional fisheries management organizations have struggled to prevent both target and non-target species declines (Cullis-Suzuki and Pauly, 2010), in part because difficulties in managing bycatch appropriately (Gilman et al., 2014). While consideration of the impact of fisheries on non-target species is growing in management decisions, it still remains a critical issue in many fisheries today (Carruthers et al., 2014; Diamond, 2004; Pelc et al., 2015).   Developing fisheries management solutions that include non-target species is challenging.  Many non-target species remain un-assessed and unmanaged because they are of low priority and or lack data (Costello et al., 2012; Honey et al., 2010; Reuter et al., 2010). Some progress has been made in managing fisheries to minimize their impact on non-target species, particularly with respect to spatial management (Diamond, 2004; Pelc et al., 2015; Zeeberg et al., 2006). Many bycatch species cluster in space and links have been shown between distribution of fishing effort and quantity of bycatch species (Lewison et al., 2009). Such information has supported various temporal or spatial   108 closures to reduce fishing effort in areas with high bycatch rates (Ardron et al., 2007; Bjorkland et al., 2015; Lewison et al., 2009; Little et al., 2015).  Solutions to reduce quantities of bycatch in various fisheries have also included gear modifications, catch limits or quotas for non-target species (Ardron et al., 2007; Bjorkland et al., 2015; Zeeberg et al., 2006). Many solutions in fisheries management are based on fishery assessment methods, which have advanced in recent years for data-poor species (Carruthers et al., 2014; Honey et al., 2010; Patrick et al., 2009; Reuter et al., 2010).  These data-poor methods include data-poor stock assessment procedures, vulnerability analyses6, and extrapolation techniques7 with the choice of method typically determined by data availability (Honey et al., 2010). The latter two options require the least data, but are not as informative as a stock assessment because they can only highlight which gear poses the most risk, or which species is under the greatest fisheries pressure (Honey et al., 2010).   While the data-poor stock assessments needed for many bycatch species call for fewer data than typical (full) stock assessments, they still require more data than other data-poor methods (Carruthers et al., 2014; Honey et al., 2010). Stock assessments guide the fisheries decision-making process by providing an understanding of fish population dynamics in relation to fishery input (e.g. number of fishing vessels) or output controls (e.g. total allowable catch, quotas) (Sparre and Venema, 1998). Data-poor stock assessments require information in one of three forms (i) a time series of recent catches, (ii) historical catches, or (iii) current estimates of absolute abundance (Carruthers et al., 2014). When data are sparse, more assumptions about fish population dynamics                                                 6 Vulnerability analysis assesses species risk to overfishing associated with various fishing gears and species life history (Stobutzki et al., 2001; Winemiller, 1992) 7 Extrapolation techniques involve the creation of species groups, where management decisions are based on one species of the group with available data (Smith et al., 2009).   109 and their responses to fishing pressure must be made.  If uncertainty around assessments is too high, they will be of little practical use in decision making (Bentley and Stokes, 2009a). However, a comparison of a data-poor stock assessment and vulnerability analyses found that the latter failed to identify the test case species as overfished (FAO, 2010a). Indeed a full stock assessment of the test case species had identified the stock as overfished, causing managers to question the usefulness of vulnerability analysis (FAO, 2010a). Even for species where data are plentiful, stock assessments can make incorrect predictions (Myers et al., 1997; Walters and Maguire, 1996). In many countries, fisheries managers are accustomed to running stock assessments and dealing with uncertainty. For a data-poor species, however, managers may default into a do-nothing mentality until more data are available, even though alternative assessment methods area available (Honey et al., 2010). The question remains: when is running a data-limited stock assessment useful?   Models with minimal data requirements such as age-structured models are preferred for data-poor stocks (Ludwig and Walters, 1985). Age-structured models are well suited for data-poor species because they require little information about the stock itself beyond making some limited assumptions about common fish behavior and life history (Hilborn and Walters, 1992). These models predict catches in the fishery and can be used to explore the impacts of fishing effort based on the following assumptions: (i) all fish gain weight as they age; (ii) tomorrow’s recruitment depends on the reproductive output of today’s fish; and (iii) young fish are less likely to be captured by fishing gear than older fish (Hilborn and Walters, 1992). These models also assume populations are closed8 and that the main sources of mortality are natural mortality and fishing mortality                                                 8 Closed populations are ones that assume there is no immigration or emigration of individuals to/from the population.    110 (Hilborn and Walters, 1992). Because these assumptions describe common relationships in many fish species (growth, longevity, recruitment compensation), age-structured models have formed the basis for numerous investigations related to data-poor species (Brooks et al., 2010; Cope and Punt, 2009; Curtis and Vincent, 2008; Punt et al., 2011).   The development of data-poor stock assessments has strong implications for marine fishes listed under the Convention on International Trade in Endangered Species (CITES). CITES regulates exports of marine fish species by listing them on one of its Appendices9, but very few marine fishes are actually listed on CITES (Vincent et al 2013). Countries that trade in Appendix II species must provide scientific evidence that exports do not harm wild populations, called making a non-detriment finding (NDF), and that the specimens are legally sourced (CITES, 1973). If CITES member countries have difficulty making scientifically defensible NDFs for their exports, they can be issued research and action recommendations by the CITES Animals Committee to guide them with this process. In 2012, Thailand was one of the first two countries required to undergo this enforcement process for a marine fish (Vincent et al., 2013).  Thailand is the world’s largest exporter of seahorses (Hippocampus spp), a CITES-listed group of marine fishes. One of the recommendations issued by the Animals Committee was to model population response to exploitation pressure (CITES, 2012b). Exploring the options for a seahorse data-poor stock assessment should assist Thailand to implement the recommendations by the CITES Animals Committee in support of sustainable trade.                                                  9 Species can be listed on one of three appendices under CITES. Appendix I includes species threatened with extinction that are or may be affected by trade. Appendix II lists species that may become threatened with extinction if trade is not regulated. Appendix III species are taxon of national concern where individual countries ask support from other CITES parties because exploitation is prohibited by national law (CITES, 1973).   111  Seahorses are a non-target marine fish genus, captured by many commercial and small-scale fisheries around the world, and dried specimens are traded internationally for their use in traditional medicines and curios (Lawson et al., in press; Vincent et al., 2011). Because the majority of seahorses are captured as non-target catch in fisheries and because fishers catch a low number of seahorses per night, there are few records of catches, effort, or landings, at national levels (but see Baum and Vincent, 2005; Lawson et al., in press; Perry et al., 2010). Records are available for international seahorse trade volumes because seahorses are listed on CITES Appendix II, but such records are full of discrepancies (Foster et al., 2016). No stock assessments have been conducted for seahorses, although there have been several studies to model seahorse population dynamics (Curtis and Vincent, 2008; Harasti et al., 2012; Mai and Rosa, 2009). Such studies have used either age-structured or Jolly-Seber population models, the latter of which relies heavily on data from wild populations (Curtis and Vincent, 2008; Harasti et al., 2012; Mai and Rosa, 2009). Only one study has evaluated a fishery management measure (minimum size limits) in relation to biological reference points and simulated fishery mortality rates (Curtis and Vincent, 2008). A data-poor fishery method (vulnerability analysis) on seahorse species and fishing gears in Thailand identified the focus for an adaptive management plan on trawling gears and gillnets (Aylesworth et al., 2016b), but little information is available about what type of management actions for these gears may secure the persistence of wild populations.   To explore how a data-poor stock assessment can inform and advance management for data-poor fishes, I developed the first seahorse stock assessment for how these data-poor fishes with unusual life history characteristics (Foster and Vincent, 2004) may respond to fishing pressure. I explored   112 my model to evaluate the sensitivity of life history parameter estimates, and determine if fishing mortality has been sustainable over the course of fisheries growth in Thailand. I reconstructed fishing effort and seahorse catches from 1970 until 2015, and then fit the best of four model variations to my reconstructed data. Finally, I explored the usefulness of a stock assessment to identify what management measures (e.g. reduction in vessel numbers, marine protected areas) would ensure the persistence of my fishes in the future.  5.3 Methods I chose to build and execute models for the seahorse Hippocampus trimaculatus for three reasons.  First, it was the dominant seahorse species observed in both trade and port sampling in Thailand (Foster et al., 2014; Aylesworth et al, under review)(Chapter 4). Second, it is captured by most fishing gears, commercial and small-scale fishing (Aylesworth et al, under review)(Chapter 4). Third, I can access published life history parameters for this species (Appendix D.1). Additionally, seahorses have been documented to meet the assumptions for using a simple age-structured model (Baum et al., 2003; Curtis and Vincent, 2008; Foster and Vincent, 2005, 2004; Lawson et al., 2015).  5.3.1 Building A Seahorse Age-structured Population Model I first built an age-structured population model to explore seahorse population dynamics in relation to uncertainty in the model parameters and simulated fishing effort.  I next used reconstructed data on fishing effort and seahorse catches from Thailand to determine which of four model variations is the optimal for a seahorse stock assessment. I then explored four management scenarios against a ‘do nothing’ option to determine what management actions would help secure sustainable seahorse fisheries into the future.    113  I created an age structured model with exponential population growth (Walters and Martell, 2004) to simulate aspects of seahorse populations and fishery dynamics, and to generate predicted seahorse catches. Parameter definitions and equations used to create such a model are presented in Table 1. I used an age-structured population model based on the life expectancy of H. trimaculatus (three-years), run over monthly time steps (1-36 months) (Equation 1.1-1.4; Table 5.1). A monthly model enabled me to include peaks in seasonal recruitment (Walters and Martell, 2004) as suggested for H. trimaculatus (Truong and Ton, 1995). I ran my model over a 53-year period, from the beginning of fisheries growth in Thailand in 1960 (Chuenpagdee and Pauly, 2003) until 2015, the year of the most recent census of fishing vessels.  I partitioned total mortality into instantaneous fishing mortality (F) and natural mortality (M) (Equations 1.1; 1.2, Table 5.1). I incorporated age-specific fecundity (Equation 1.4, Table 5.1) and annual recruitment of age-1 month fish following a Beverton-Holt stock-recruitment relationship (Equations 2.1-2.5, Table 5.1) parameterized in terms of compensation ratio (k) (Walters and Martell, 2004). A value of 5.0 was used for this parameter and corresponds approximately to 0.5 in terms of the “steepness” parameterization, a common default value for most benthic fishes (Forrest et al., 2013; Myers et al., 1999).  I assumed that seahorses followed a typical length-weight relationship (Equation 1.2, Table 5.1) and maturity schedule for fish (Equation 1.3, Table 5.1). I sourced data on height-at-maturity from recently published literature on H. trimaculatus from Malaysia (Lawson et al., 2015).  For size at vulnerability to the fishery (Equation 1.5, Table 5.1), I used port-sampled data from (Aylesworth et al., 2016b)(Chapter 4) to determine when a seahorse had a 50% chance of being retained when it encountered a fishing net (Sparre and Venema, 1998), which was approximately 9.5 cm.    114 I ran two analyses to gain a better understanding of my seahorse age-structured population model. First, I identified the maximum level of fishing mortality that could be sustained by the seahorse stock to reach a management target of Bt/B010 greater than or equal to 30% from 1960 (the year I equate with B0) until 2015. The ratio of Bt/B0, is a commonly used fisheries management objective for little known stocks (Caddy and Mahone, 1995; Caddy, 2004; Myers et al., 1994). Second, I used a sensitivity analysis to identify which model parameters were most sensitive to estimation error (Appendix D.2).   5.3.2 Reconstructing Fishing Effort And Seahorse Catches Because I lacked data on (i) a time series of recent seahorse catches and (ii) current estimates of absolute abundance, I attempted to reconstruct historical fishing effort and seahorse catches to provide data for my data-poor stock assessment (Carruthers et al., 2014). I reconstructed fishing effort from 1970 until 2015 for both commercial and small-scale fishers.  I gathered data on the number of commercial fishing vessels from four gear types – otter trawls, pair trawls, purse seines and push nets – as reported by the Thai Department of Fisheries (DoF 2015a), and published literature for historic estimates (Chullasorn, 1998; Tokrisna, 2006; Willmann et al., 2000). These gear types have previously been identified to incidentally capture seahorses in Thailand (Aylesworth et al., 2016b; Laksanawimol et al., 2013; Perry et al., 2010). I also estimated the number of historic and current vessels from two types of small-scale fishing gears reported to capture seahorses – gillnets and traps – from published literature for historic values (Janekitkosol et al., 2003)) and the Department of Fisheries (DoF 2015a). For missing yearly estimates of small-                                                10 Bt/B0 represents a ratio in fish biomass. Bt is the current biomass at time t and B0 is the level of unfished biomass.   115 scale fishing vessels, I determined the change over time between the missing and known years and calculated linear estimates to fill in the gaps as per (Teh et al., 2015).  To reconstruct seahorse catches, I used a combination of recently published (2013/2014) seahorse catch rates, the national seahorse catch estimate for Thailand (2015) (Aylesworth et al., 2016b) (Chapter 4), and historic data from fisher interviews (Aylesworth, unpublished). I begin with a summary of how I reconstructed catches, before describing these procedures in detail in the following two paragraphs. Determining historic catches was a two-step process. First I interviewed fishers to identify (i) the historical year used to compare quantities of seahorse bycatch and (ii) how such bycatch quantities had changed over time. Then I generated three datasets of observed historic catches representing the minimum, mean and maximum changes reported by fishers. I assumed that fisher-reported changes in seahorse quantities represented changes in total seahorse catches, as long as my proxy for fishing effort, total vessel numbers, did not increase over the identified time period (Hilborn and Walters, 1992). I anchored historic catches based on the 2015 estimate of 29.4 million seahorses captured per year (Chapter 4), multiplied by the mean weight of 1 dried seahorse (3.16 g) to convert this estimate into biomass (Aylesworth et al., 2016; Kuo, unpublished).  I then back-calculated catches linearly based on fisher reported changes to the mean year of reported historical catch.   To identify how seahorse bycatch had changed over time, I asked fishers questions about historic seahorse bycatch rates, as part of semi-structured fisher interviews (Appendix D.3). Fishers were given the opportunity to comment on how seahorse catches per haul had changed over time (increased, decreased or stable), and then given the chance to quantify reported changes compared   116 to present day catches per haul. As a default, I asked fishers to comment on how seahorse catches per haul had changed compared to when they first started fishing. If fishers had difficulty with this comparison, I invited them to select their own time frame, or gave them suggestions of time frames based on prominent events in Thai history, such as before or after the 2004 tsunami (Stake et al., 2005).   I analyzed the fisher reported changes in catches per haul collectively and by gear type. I selected only the responses where fishers had quantified changes in catch per haul for my analysis (n=45). To address potential reporting error, I determined the minimum, mean and maximum reported declines for all gears in total and by individual gear type. I analyzed otter trawls (n=10), pair trawls (n=14) and gillnets (n=13) individually. I grouped responses from fishers using cages (n=2), compressors (n=3), purse seines (n=2), and pushnets (n=1) together as “other gear” because of low sample sizes. Additionally I calculated the mean number of years of fishing experience per respondent in total and by gear type, and the mean year of reported historical catch comparison to explore the potential for bias based on experience.   Next, I wanted to explore whether fisher reported declines in catches per haul represented true declines in the stock or simply changes in vessel numbers. Fishers may experience perceived declines in catch per unit effort if vessel numbers increase, meaning that the proportional catch of each fisher would decline, but the stock status would remain stable. Since I only had estimates of historic vessel numbers available, I assumed these were a proxy for changes in fishing effort.  To explore this possibility, I looked at the trend of the number of fishing vessels from 1970 to 2015. If the number of vessels increased during the period, fishers-reported declines might not represent   117 changes in stock status. However, if vessel numbers remained steady or declined during this period, fisher reported declines might indeed represent a worsening stock status.   5.3.3 Finding A Model To Fit The Data To conduct a data-poor stock assessment I needed to find a model to predict my data. I created four model variations of my seahorse age-structured model (Walters and Martell, 2004) (Appendix D.4) to compare with the re-constructed data of seahorse catches (“observed data”).  My four models tested the following hypotheses: (i) seahorse catch is best predicted by historic fishing and current fishing effort; (ii) seahorse catch is best predicted by incorporating estimates of illegal, unreported and unregulated fishing (IUU); (iii) seahorse catch is best predicted by habitat decline with fast recovery and weak recruitment compensation; (iv) seahorse catch is best predicted by habitat decline with slow recovery and strong recruitment compensation. I determined these four hypotheses based on a combination of available data (e.g. historic and current fishing vessel numbers), and expert knowledge about Thai fisheries (challenges with IUU) and seahorses (links with habitat).    To test these hypotheses I determined which of my four model variations best predicted the reconstructed catches. I fit the models to the data by varying different parameters, using a sum of squares fitting criterion, attempting to minimize the sum of squared deviations between observed and predicted log catches (Hilborn and Walters, 1992). I deemed the model with the smallest difference to be the best fit and used it to evaluate possible management options. I varied the following parameters for each model to find the best fit: (i) initial biomass; (ii) initial biomass; (iii)   118 initial biomass and recruitment compensation (iv) initial biomass and catchability of habitat by fishing gear  5.3.4 Evaluating Management Options I wanted to know how management might alter the current trend in stock status by 2029, so I compared four possible management scenarios. First I ran a ‘do nothing’ management option with the model that best fit my reconstructed seahorse catches. In this scenario I assumed estimates of vessel numbers and relative catch rates remained the same from 2015 until 2029. I determined the catch trend and the percent of catch decline by 2029 compared to 2015. For the three management scenarios, I compared (1) reductions in the number of vessels, (2) seasonal closures, and (3) marine protected areas to determine the effort adjustments needed to (i) reduce the catch decline rate by 20% corresponding to an improvement from critically endangered (as suggested by declines under the do nothing scenario) to endangered under IUCN Red List criteria (IUCN, 2015), (ii) stabilize catch and (iii) increase catch.   My second intent was to evaluate five current fisheries management efforts in Thailand to determine their possible effectiveness for seahorses in the future. Thailand recently reformed its fisheries policy (2015) in an effort to address IUU fishing efforts (DoF 2015). The scenarios were based on five approaches listed in this new Thai fisheries management plan (DoF 2015): (1) eliminating push nets beginning in 2016; (2) reducing trawl gear by 30%; (3) maintaining three seasonal closures; (4) maintaining the current array of MPAs; and 5) using all these management efforts together.    119 5.4 Results 5.4.1 Building A Seahorse Age-structured Population Model My seahorse age-structured population model suggested that F would need to have been ≤ 0.12 in 1960 (and all years that followed) to maintain 0.3 of the 1960 biomass in 2015.  The results of the sensitivity analysis differed depending on whether I set the sensitivity to 0.1 or 0.5.  When sensitivity was set at 0.1, the length at which seahorses recruited to the fishery was the most influential parameter in determining both (i) how quickly biomass dropped by ≥ 0.3 and (ii) how far the biomass actually dropped (Table B-1; Appendix Table D.2). When I set sensitivity to 0.5, the length at which seahorses recruited to the fishery was still the most influential parameter in determining how quickly biomass dropped by ≥ 0.3.  However, the recruitment compensation parameter mattered most in determining how far the biomass actually dropped (Table B-1; Appendix B).  5.4.2 Reconstructing Fishing Effort And Seahorse Catches The number of both commercial and small-scale fishing vessels increased from 1970 until 1990, then slowly declined until 2015 (Figure 5.1). In the case of commercial fisheries, vessel numbers declined to below 1970 levels, whereas in the case of small-scale fisheries, vessel numbers remained above 1970 levels.    When I asked fishers when they considered the decline in catches per haul to have started, the mean response was 1997 (Table 5.2).  However, only 45 of the 306 fishers I interviewed quantified changes in CPUE over time (Table 5.2). Across all gears, estimates of minimum, mean and maximum catches per haul in 2015 were only 27%, 18% or 14% respectively, of what they had   120 been in 1997. When I summarized these declines by gear type, fishers reported that pair trawls (11%) and gillnets (14%) had the lowest catch per haul in 2015 relative to 1998 and 1999 catches respectively (Table 5.2).  Fishers using ‘other gears’ reported the greatest amount of catch per haul in 2015, at 38% of estimated catches in 1996.    5.4.3 Finding A Model To Fit The Data The model that best matched my reconstructed observed catches included habitat decline, slow habitat recovery and strong recruitment compensation (Figure 5.2, Table 5.3). Neither the model variation with only the 1970 and 2015 catches (variation 1) nor the model with IUU vessel numbers (variation 2) matched the declines from 1997 until 2015 reported by fishers (Fig 5.2). Both habitat recruitment models matched reported declines well, but the model with quick habitat recovery and weak recruitment compensation (variation 3) did not fit as well as the model with slow habitat recovery and strong recruitment compensation (variation 4) (Table 5.3; Figure 5.2).  5.4.4 Evaluating Management Options  The ‘do nothing’ management strategy resulted in continued declines for the next 15 years of model simulation until less than 1% of predicted 2015 catches would remain (Table 5.4). In order to lessen this decline by 20%, the number of vessels using all gear types would need to be reduced by 16% (Table 5.4). Alternatively, a complete seasonal closure (of all gear types) of five and half months or turning 45% of Thai waters into strict no-take MPAs would also achieve the same result (Table 5.4). In order to stabilize catches or have an increasing catch trend, I would need to remove more vessels, enhance seasonal closures or create more MPA coverage (Table 5.4).     121 All Thai management scenarios mooted in 2015 would produce declines in seahorse biomass (Table 5.5). The strategy that combined elimination of pushnet vessels, a 30% reduction in trawling gears, three seasonal closures, and current MPAs would create the best results after 15 years, but this still amounted to only 13.7% of biomass remaining from 2015 levels. The management measure with the poorest results compared to the ‘do nothing’ alternative was the elimination of pushnet vessels (Table 5.5).  5.5 Discussion I found that despite initial uncertainty about life history estimates, historical fishing effort and catches, a data-poor stock assessment had real benefits in allowing me to evaluate management measures past and proposed (Carruthers et al., 2014; Punt et al., 2011; Smith et al., 2009). The work I present here allowed me to refine an adaptive management plan in a way that was otherwise impossible. My results indicate that seahorse populations can only tolerate low fishing mortality rates and that H. trimaculatus is most likely overfished in Thailand. This research highlights that length at recruitment to the fishery is a key model parameter and suggests that the previously identified precautionary approach of setting a minimum size limit may be a beneficial management measure (Foster and Vincent, 2005). Even allowing for the various assumptions I made about seahorse life history and historical depletion, my analysis confirms that there really have been notable declines in seahorse abundance in Thailand since at least the mid-1990s (Perry et al., 2010).  An added benefit of the data-poor stock assessment was that I was able to evaluate management actions that was already prescribed for Thai fisheries (DoF 2015), and suggest whether they may be effective for seahorses in the CITES context (CITES, 2013a). As in other research related to fisheries in Thailand, I found that management action related to vessel reductions in combination   122 with spatial and temporal closures would be the most effective reducing the decline of seahorses in the future (Janekitkosol et al., 2003; Lymer et al., 2008; Phasuk, 1996; Pomeroy, 2012).   With the creation of my model I discovered that decreasing levels of fishing mortality and increasing size at recruitment to the fishery were very important in ensuring that non-target fisheries did not damage H. trimaculatus populations. Both of these factors suggest that non-selective fishing is problematic not only for seahorses, but most likely for other small, non-target fishes (Foster and Vincent, 2010; Parsons and Foster, 2015). With low levels of fishing mortality required to ensure a precautionary reference point of 0.3 Bt / B0, management measures that reduce fishing effort will benefit seahorses (Ahmed et al., 2007; DoF, 2015). However, my study suggests that substantial benefit to seahorse populations may only be achieved through large effort reductions, which may be challenging in the Thai context (Willmann et al., 2000). Minimum size limits for seahorses have already been recommended, and would help to ensure higher recruitment to target fisheries (Foster and Vincent, 2016). Unfortunately for non-target fisheries, such measures may not be effective because fishers make decisions about where to fish and which gears to use based on target species (Daw, 2008; Maina et al., 2011), and such gears do not select for seahorse size.   I argue that the relationship between seahorse abundance and CPUE is probably proportional, such that the large declines reported by fishers represent worrying large declines in seahorse abundance (Hilborn and Walters, 1992).  The capture of seahorses very likely meets a key assumption of this relationship - that fishing effort is randomly distributed – because extensive discussion indicates that seahorse distribution does not guide trawl boat captains’ decisions on where to fish (pers. obs.).  My perspective is further bolstered by the IUCN Red List assessment that H. trimaculatus is   123 ‘Vulnerable’ based on suspected declines of at least 30% (Wiswedel, 2015).  However, my claim of proportional relationships and large declines in CPUE would mean that decline rates of H. trimaculatus are even higher, suggesting that H.trimaculatus in Thailand be placed into the ‘Critically Endangered’ category.   Based on my analysis, populations of H. trimaculatus in Thailand are most likely overfished, in part because of fisheries-related habitat declines that affected the recruitment carrying capacity. I found that an age-structured model linking slow habitat recovery and strong recruitment compensation was best for a seahorse stock assessment, most likely because seahorses are sedentary fishes with strong relationships to habitat (Foster and Vincent, 2004; Vincent et al., 2011). The model predicts that once habitat falls below a certain carrying capacity, there will be a decline in recruitment compensation. Several seahorse species have declined due to associated habitat loss (Vincent et al., 2011), and population declines associated with habitat declines are among the main criteria used to evaluate seahorses as threatened for the IUCN Red List (IUCN, 2015).  This makes sense since H. trimaculatus is known to be a deep-water species caught pre-dominantly in trawling gear in Thailand (Aylesworth et al., 2016b; Lourie et al., 2004) and trawling gears have enormous impacts on the surrounding habitat (Hiddink et al., 2006; Watson et al., 2006).    It logically follows that a model incorporating habitat effects from fishing gear may best represent reconstructed seahorse catches. Such a model may represent a situation with chronic trawling, where initially recruitment compensation is strong in remaining habitat, but becomes weak once continued damage drives residual habitat below a minimum threshold (Kaiser et al., 2002; Steele et al., 2002; Thrush and Dayton, 2012). Other types of models such as those that explicitly account for   124 spatial behavior of fish or fishers (Hilborn and Walters, 1992), or potential ecosystem effects of fishing (Christensen, 1998), required additional data that were unavailable for my study. Future revisions of my stock assessment would include incorporating uncertainty into the model through Bayesian analyses or Markov-chain Monte Carlo simulations to understand how uncertainty affects model outcomes (Jiao et al., 2011; Magnusson et al., 2013; Punt and Hilborn, 1997) or exploring links to the spatial behavior of seahorses or fishers.    I am aware of having made assumptions in two key areas: (i) model type and (ii) reconstructing fishing effort and seahorse catches. First I assumed that an age-structured model was the best type of model to use for my data-poor fishery assessment. This assumption was based on two other inferences: (a) that seahorse life history relationships such as growth, weight, fecundity, mortality followed known relationships in other marine fishes (as per Curtis and Vincent, 2008; Harasti et al., 2012; Mai et al., 2012); and (b) that seahorses in Thailand had seasonal recruitment and reproductive peaks as documented elsewhere in the range of H. trimaculatus (Truong and Ton, 1995). Second, I made three key assumptions in reconstructing Thai fishing effort and seahorse catches, which may have influenced my results. First, with reconstructing fishing effort, I inferred that (where available) the Thailand Department of Fishery estimates of number of fishing vessels were accurate. Second, in reconstructing seahorse catches, I assumed that 2015 catches were the most accurate estimate of seahorse catches in Thailand, which justified anchoring the catch reconstruction in 2015. Third, I assumed that fisher-reported changes in CPUE represented declines in total seahorse catches.    125 Current Thai management measures (DoF, 2015) will not secure stable or increasing seahorse populations in the future. The single measure with the most benefit, a decrease in vessel numbers, coincides well with current documentation that Thai fisheries are over-capacity (DoF 2015). Both commercial and small-scale fishing gears in Thailand have expanded beyond the effort required to achieve maximum sustainable yield for numerous target species (Chuenpagdee and Pauly, 2003; Lymer et al., 2008; Pomeroy, 2012). Despite knowledge of declining resources and over-capacity since the 1980’s, Thailand still struggles with overcapacity challenges today (Christensen, 1998; Chuenpagdee and Pauly, 2003; Pauly, 1988, 1986) (DoF 2015).  A great deal of Thailand’s fisheries research has prioritized gear modifications to reduce the vulnerability of juvenile target species to fisheries, which do little for similarly sized non-target species (Boutson et al., 2009; Meemeskul, 1980; Songrak et al., 2013). While my research did not evaluate gear modifications, most research related to gear modifications has been targeted towards large charismatic mega-fauna (Lewison, R et al., 2004; Lewison et al., 2009; Moore et al., 2009; Zeeberg et al., 2006) or juveniles of commercially important species (Parsons and Foster, 2015) and will do little or nothing for small species such as seahorses. However, my research highlights that combining several measures, reducing vessel numbers in combination with spatial closures, is the most effective, as documented in other research (Anticamara et al., 2011; Berkeley et al., 2004; Grech and Coles, 2011).   My data-poor stock assessment for seahorses, a genus of non-target fishes, is relevant for numerous other data-poor species where the impacts of fisheries have not been assessed, in four ways. First, I was able to obtain information required for such a data-poor stock assessment from historic catches, based on government data on fishing vessel numbers (Carruthers et al., 2014). While I was initially skeptical about the accuracy of vessel number estimates, such information is likely to be available   126 in most countries, and methods are available to account for potential inaccuracies (Teh et al., 2015; Walters and Martell, 2002; Zeller and Pauly, 2015). Second, once historical catch data were generated, I applied expert knowledge to explore hypotheses about historical depletions through model variations (Carruthers et al., 2014). When these analyses yielded similar results (e.g. determining the model of best fit using minimum, mean or maximum fisher-reported declines), I became more confident that H. trimaculatus is most likely over-fished, despite data uncertainty. Fourth, once the model had been built, it required relatively little effort to update as more data became available or as expertise was acquired to incorporate uncertainty (Bentley and Stokes, 2009b; Jiao et al., 2011).   In a CITES context, such as the one confronting Thailand, data-poor stock assessment methods can be useful for listed marine fishes in three ways. First, responding to CITES recommendations requires management action quickly, within 6-24 months (Vincent et al., 2013). Because the timelines are short, it is difficult for Parties to prioritize and dedicate resources to obtain new data. My work shows that Parties can, however, use existing national data about historic vessel numbers to obtain estimates of historic catches that can drive a data-poor stock assessment. Second, deploying a stock assessment approach for CITES work allows Parties to mobilize fisheries ministries on familiar terms; the latter are often uneasy about engaging with CITES but are very much at ease with the language and methods of stock assessments.  Third, a data-poor stock assessment in the CITES context will lead to proposed action that is either complementary to existing national fisheries management or at least within that frame of reference. In this way, global pressure through CITES can help focus the minds of national fisheries mangers on non-target   127 species like seahorses, forcing more of an ecosystem based consideration of non-selective fishing gears and what should be done about them.   My findings that the best response to fisheries that deplete non-target species will be a combination of vessel reduction and spatial management are of broad general value for bycatch species.  Gear modifications such as turtle excluder devices or increasing cod-end mesh sizes, may not be effective for many small non-target species if the latter are not particularly mobile swimmers or are the same size as target species (Brewer et al., 2006; Nguyen and Larsen, 2013) . Temporary closures may be dubious in their success at mitigating fisheries impacts on non-target species (Halliday, 1988; Karras and Agar, 2009). Although attempts to reduce vessel numbers may meet resistance because of the potential socio-economic impacts, they are probably inevitable as fish stocks continue to decline and fishing costs (e.g. fuel prices) increase (DoF, 2015; Panayotou and Jetanavanich, 1987). Given that spatial management is one of the best options for indiscriminate fisheries management, and is required under the internationally agreed Aichi Biodiversity Targets (CBD 2010), its use in fisheries management will likely increase. However pushing for an increase in quantity of spatial management will not achieve conservation goals without enforcement (DeSanto, 2013). To achieve full implementation will require an observing eye – be it with onboard observers (Branch et al., 2006), vessel monitoring systems (Witt and Godley, 2007), or through satellite technology (Al-Abdulrazzak and Pauly, 2013). In essence, then, a low technology approach can be married to high tech tracking to reduce pressures on bycatch species such as seahorses.       128 Table 5.1 Seahorse age-structured model variations, variables, and equations representing life history relationships. Variable Definition Nt,a The number of seahorses alive at age a, and time t Nt+1,a+1 The number of seahorses alive at age a + 1 month, and time t+ 1 month  La Length at age Linf Length of a fish if it were to grow indefinitely vbK The rate (1/year) at which Linf is approached M Natural mortality rate Ct,a The number of individuals of age a, caught during time t Ft Fishing mortality rate for fully vulnerable individuals (exploitation rate at time t) Va An age-specific vulnerability to fishing mortality Et,a Number of eggs produced by each cohort ea Age-specific fecundity Et Total egg production R Annual recruitment Bt Total biomass of the stock / exploitable biomass at time t wa Weight-age relationship E Fishing effort it Ratio of unreported to reported catch per decade It,g Additional number of fishing vessels added per year to account of IUU effort     129 Table 5.1 (con’t) Seahorse age-structured model variations, variables, and equations representing life history relationships. Variable Definition Ht Total unfished habitat  h Habitat remaining rh Habitat regrowth / recovery rate qh Relative catchability coefficient of the habitat itself; relative probability of loss of a sponge /coral when swept over by a single trawl p How habitat recovery rate depends on the current habitat state j Habitat specific value at which juvenile survival rate (pre-recruitment) drops to ½ of its ‘normal’ value; setting this to zero causes the model to predict recruitment rate independent of habitat density; larger values make recruitment fail when habitat is too widely spread Equations Formula Age-length Relationship 1.1 La = Linf*(1-e(-vbK*age/12)) Length-weight relationship 1.2 Wa = aLa b Proportion mature at length 1.3 PmatL = 1/(1+e - (Lt -Lengthmat )/0.5), Fecundity at age 1.4 ea = Wa*PmatL Vulnerability to fishing 1.5  Va = 1/(1+e(-(La-Lvul)/0.5))    130 Table 5.1 (con’t) Seahorse age-structured model variations, variables, and equations representing life history relationships. Equations Formula Survival 1.6 S = e(-M/12) Stock recruitment relationships  2.1 Rt = αEt /(1+βEt) 2.2 Nt+1,1 = Rt 2.3 Et,a = Nt,a ea 2.4 Et = Σ Nt,aPmat,aWa     a=1…..36 2.5  Basic abundance   3.1 Na,1 = R0la 3.2 Nt,a =Nt-1,a-1S(1- e-F Va-1) 3.3 VulBt = Σ Nt,aWaVa 3.4 Ct,a = VulBt(1-e(-F)) Scenario 1:Historical Effort & Catch  4.1 F=Eqt 4.2 Ftot = Q* Σ qgEg,t    for all g,  g = 1-6 Scenario 2: IUU vessels  5.1 Ftot = Q* Σ qg (Eg,t +Ig,t)    for all g,  g = 1-6 5.2 Ig,t = it(Et,g)  01   SSBκ κα βφ−= =  131 Table 5.1 (con’t) Seahorse age-structured model variations, variables, and equations representing life history relationships. Equations Formula Scenario 3: Fast habitat recovery; weak recruitment compensation  6.1 Rt = αEt /(1+β[Et/Ht) 6.2 Ht+1= hrt + rhhrt(1- hrt)p 6.3 hrt = 1e(-Ftqhabitat) Scenario 4: Slow habitat recovery, strong recruitment compensation  7.1 Rt = aEt /(1+bEt/Ht)(j+1)*(Ht/j+Ht) Ht+1= hrt + rhhrt(1- hrt)p hrt = 1e(-Ftqhabitat)      132 Table 5.2 Fisher reported changes in catches per haul collectively and by gear type.     Fraction of historical catch per haul remaining in 2013  Historical years of comparison Fisher experience Gear # fishers (n) Minimum Mean Maximum Mean  year Range Mean years  spent fishing  Otter Trawl 10 0.29 0.23 0.17 1995 1984-2004 32 Pair Trawl 14 0.17 0.14 0.11 1998 1974-2007 26 Gillnet 13 0.23 0.17 0.14 1999 1984-2007 29 Other Gears 8 0.38 0.26 0.15 1996 1974-2004 34 Totals & means 45 0.27 0.19 0.14 1997 1974-2007 29       133 Table 5.3 Model comparisons of habitat and recruitment for observed historical catches  Sum of squares deviation of predicted / observed catches Model Min Decline  Mean decline  Max decline  Null model 47.01  56.7 59.45 IUU vessels 14.5 18.45 20.63 Slow habitat recovery, strong recruitment compensation 0.51 0.71 1.05 Fast habitat recovery, weak recruitment compensation 2.23 2.60 3.31      134 Table 5.4 Evaluation of three management strategies compared to the ‘do nothing’ alternative. Strategies compare 2015 model results to predicted results in 2029.  Management strategy Catch trend from 2015 to 2029 % Catch decline by 2029 (from 2015 levels)  Do nothing decline 99.9    Lessen catch decline rate by  ≥ 20%  Stable catch  Catch increase Overall vessel reduction (%) 16 62 66 Seasonal closure of all gear types (# of months) 5.5 8 8.5  No-take marine protected areas (% of national waters) 45 65 66      135 Table 5.5 Current Thai management measures and their simulated outcome in changing catch trends and decline rates by 2029.  Management Time frame % waters closed % Decline from 2015 catch Do - nothing  ------ 99.9 Vessel reduction No pushnets ------ 99.9 Trawl vessel numbers reduced by 30% ------ 99.4 Seasonal Closure 2 locations: Feb15-May 15   1 location:  Jun 1- Aug 1 9.7    1.5 99.5 National Parks Year-round 23.2 99.1 All four above Year-round ----- 86.3      136 a)  b)  Figure 5.1 Commercial (a) and small-scale (b) fishing vessel numbers from 1970-2015 taken from the Department of Fisheries and published literature.   137  Figure 5.2 Observed (black) vs. predicted catches from model variations representing (i) historic & current fishing vessel numbers (grey); (ii) IUU fishing vessel estimates (yellow); (iii) fast habitat recovery & weak recruitment compensation (blue); (iv) slow habitat recovery and strong recruitment compensation (green).     138 Chapter 6: Taking A Dose Of Our Own Medicine: Implementing Conservation Policy For Marine Fishes 6.1 Summary What happens when I force myself, as a provider of conservation advice, to take policy and management action based on that advice?   Conservation advocates and scientists often initiate regulatory change that has significant implications for government but seldom experience the challenge of responding to such change.  In this case study, I place myself in the role of the government of Thailand, facing its obligations to seahorses (Hippocampus spp.) under the Convention on International Trade of Endangered Species (CITES).  Project Seahorse’s evaluations had led CITES to issue direct recommendations to Thailand for remedial action to ensure that its exports of four seahorse species do not damage wild populations. I ran through a CITES-prescribed framework (that Project Seahorse developed) to evaluate the risk to two of these species from various pressures and determine whether appropriate management was enforced and is currently effective.  Three core questions emerged: 1) what pressures are putting wild populations of the species at risk; 2) is management in place to mitigate the risk or offset those pressures; and 3) are the species responding as hoped to the management policy. I identified risk above a tolerated level for exports of H. trimaculatus because there was a lack of appropriate management in place to mitigate risks.  I determined risk was tolerable for H. kuda but only with monitoring in place to evaluate how the species was responding to management.  Under CITES regulations, exports of both species would be prohibited until more precautionary management emerged.  I found the process of such evaluations to be challenging, even without the obligation to consider social implications of my evaluations.  It became apparent that such evaluations will always be imperfect   139 but this may be acceptable in the context of adaptive management. Conservationists will do well to keep in mind these challenges for their future advice and advocacy.  6.2 Introduction It is time for conservation to take a dose of its own medicine. With increasing pressures on ecosystems worldwide, conservationists are anxious to prevent species decline through the promotion of policy changes. The ideal situation occurs when policy changes lead to changes in human behavior that promote sustainable human species interactions (Meffe and Viederman, 1995). However, the path from policy change to sustainability is long and arduous. Most conservation studies inevitably call for more data (Hamann et al., 2010; Young and Van Aarde, 2011) and such a lack of data can be used to justify management inaction (Johannes, 1998). In fact the real need is not always for more data but for relevant and timely data for both policy makers and government actors tasked with implementation (Buitrago et al., 2008; Meffe and Viederman, 1995). Conservationists are constantly telling resource managers how to do better (Hamann et al., 2010; Young and Van Aarde, 2011). Rarely do conservationists place themselves in the role of policy maker or government actors, people and agencies tasked with implementing these policy changes, to determine if conservationists’ advice is even possible or pragmatic.   For marine wildlife, the challenges in moving from policy change to sustainability are particularly evident in fisheries (Bellmann et al., 2015; Joseph, 1994; Salomon et al., 2011). Marine fishes are typically not considered as part of the wildlife trade (Vincent et al., 2013) despite their substantial contribution to domestic and international commerce (FAO, 2012). However most international disputes about marine fishes typically involve trade disagreements (Bellmann et al., 2015; Gordon   140 et al., 2001; Joseph, 1994; Joyner and Tyler, 2000). Indeed trade policy measures help to shape global patterns in fish supply and demand (Bellmann et al., 2015).  Such influence suggests that there is an opportunity for international trade policies to address issues of sustainability in fisheries (Bellmann et al., 2015).   The Convention on International Trade of Endangered Species (CITES) is a new tool to secure sustainable exports in marine fishes and, by proxy, sustainable fisheries (Bellmann et al., 2015; Vincent et al., 2013). CITES currently regulates the international trade of very few marine fish species, by listing them in its Appendices (Vincent et al., 2013). Countries that trade in Appendix II species11 must prove that exports do not harm wild populations. This is deemed a non-detriment finding (NDF) (CITES, 1973). Parties must overcome two main challenges associated with this process: (1) uncertainties about trade levels, population status and management options and (2) institutional problems associated with stakeholder involvement, financial support, and lack of capacity (Vincent et al., 2013). Failure to declare the sustainability of exports of a species appropriately can lead CITES to propose an evaluation of a Party’s trade (called a Review of Significant Trade), and eventually to suspend the Party’s right to trade the species in question (e.g. queen conch [Strombus gigas] (CITES, 1999)) (UNEP-WCMC, 2012).  There is no prescription for making NDFs and Parties can make NDFs however they wish (CITES, 2013a).  But Parties have been given a great deal of advice, which they can choose to follow or not                                                 11 Species can be listed on one of three appendices under CITES. Appendix I includes species threatened with extinction that are or may be affected by trade. Appendix II lists species that may become threatened with extinction if trade is not regulated. Appendix III species are taxon of national concern where individual countries ask support from other CITES parties because exploitation is prohibited by national law.   141 (Atkinson et al., 2008; CITES, 2013a; Mundy-Taylor et al., 2014; Newton et al., 2008; Rosser and Haywood, 2002).  Some of this advice consists of long lists of the types of data that should be used for making NDFs (Rosser and Haywood, 2002) while other advice has been given in the form of frameworks for making NDFS (CITES, 2013a; Mundy-Taylor et al., 2014). All such guidance is necessarily generic, as it must apply to many Parties, each with different cultural situations, institutional limitations and opportunities (CITES, 2013a; Mundy-Taylor et al., 2014).   The actual value of such guidance is often unclear in practical terms, because seldom do the people who give advice have to implement it.    The history of policy advice for the regulation of marine fish exports under CITES is quite recent, when in 2013, an NDF framework for seahorses (Hippocampus spp.) was created (CITES, 2013a). Indeed seahorses are a pioneer taxon for regulation through CITES and for the international regulation of marine fishes in general. They were among the very first marine fishes to be placed on CITES Appendix II (in 2002), and the first to be subject to a review of how Parties were implementing the convention, dubbed a Review of Significant Trade (RST: Vincent et al 2013). As part of this RST process, Thailand and Vietnam were identified as having exports warranting concern.  Seahorses thus also became the first marine fish species to have experienced a trade suspension, when CITES closed exports of H. kuda from Vietnam because the latter failed to meet its obligations for this species.  Virtually all scientific and technical advice throughout this relationship between CITES and seahorses has come from Project Seahorse, a small group which acts as the IUCN SSC Seahorse, Pipefish and Stickleback Specialist Group.    142 I focus on Thailand because it is the world’s largest exporter of seahorses and because this Party has been a focus of CITES action on behalf of these quirky fishes.  Thailand has undergone the RST process for four species, to determine whether its large volumes of seahorse exports, 3.0-6.5 million (Foster et al., 2016), were harming wild populations. Thailand was unable to make NDFs for its large export volumes, in part due to insufficient data on seahorse populations, fisheries and trade for four of their local seahorse species, Hippocampus kelloggi, Hippocampus kuda, Hippocampus spinosissimus, and Hippocampus trimaculatus. Consequently, Thailand’s trade in these four species was designated as “Urgent Concern”12 (CITES, 2014; UNEP-WCMC, 2012). Therefore, the CITES Animal Committee made ten recommendations to Thailand, based on Project Seahorse’s input, which the country must implement (CITES, 2012b). Given how difficult Thailand has found this process, I decided it behooved me to try assuming Thailand’s responsibility myself.   The goal of my paper is to place myself, the provider of conservation advice, in the role of the national CITES Authorities – who might be asked to take my advice in their implementation of CITES for seahorses – using Thailand as a case study. I decided to begin by following the NDF framework for seahorses. In line with the framework, I first assessed the risk to Thai seahorses from fishing, trade and habitats pressures, and I second evaluated the ability of existing management to mitigate the identified risks. I used this assessment to consider my NDF options, and what actions may be needed to improve management action and/or fill knowledge gaps. In applying the NDF framework I explored the concept of what is ‘enough’ knowledge for Parties to make an NDF under                                                 12 Trade in H. kelloggi, H. kuda, and H. spinosissimus was designated as “urgent concern” in March of 2012. Trade in H. trimaculatus was deemed “urgent concern” in March of 2014 (CITES, 2014; UNEP-WCMC, 2012).   143 CITES, and took a hard look at the implementation process, from the context of initial conservation recommendations given by CITES to Thailand.  6.3 Methods  I used Thailand and its experience with seahorses as a case study.  My focus lay in applying the CITES-prescribed “NDF framework for making Non-Detriment Findings for Seahorses” (version 4.0; (Foster and Vincent, 2016)).  My intention was to use all available information for Thailand’s seahorses to see if I could come to a conclusion about the sustainability of Thai seahorse exports. I then used the results of this evaluation under the NDF framework to explore what Thailand could do better to implement these seahorse listings.  Because non-detriment findings must be made at the species level, I ran through the NDF framework with two of the Thai seahorse species that CITES identified in 2012 as being of “Urgent Concern”, Hippocampus trimaculatus and Hippocampus kuda. These two species are the most vulnerable to fishing pressure (Aylesworth et al., 2016b), and represent the dominant offshore (H. trimaculatus) and inshore (H. kuda) seahorse species in Thai fisheries and trades (Aylesworth et al., 2016b)  In applying the framework I drew on all available data on or related to Thai seahorses as of 31 December 2015, noting that sometimes the data were only available at the genus level.  I described the data requirements of the NDF framework, as well as the data I actually had, and how I evaluated my data in an effort to make a non-detriment finding.   Where I encountered them, I included additional datasets not explicitly requested in the NDF framework as extra inputs to my evaluation.   144  The NDF framework (Figure 6.1) encourages use of all available information on species involved in trade including published literature, grey literature, local knowledge, citizen science contributions, government research and expert opinion (Foster and Vincent, 2016).  I had used my prior knowledge to pre-select wild sourced H. trimaculatus and H. kuda for my evaluation, thus allowing me to skip over Sections 3 and 4.1.    Still under Section 4, the next four steps of the NDF framework describe and then evaluate the risks to the species from fishing and trade pressures, and from the pressures on the seahorse’s habitats (Figure 6.1, Sections 4.2-4.5).  I first gathered all available data on trade, fisheries, and habitats related to seahorses in Thailand (Tables 6.1 and 6.2). This information came from the Thai CITES authorities, including documents submitted to the CITES Secretariat and relevant committees by Thailand in support of the RST process. I also consulted published literature (using Google Scholar) and local experts (n=150), and drew on my own seahorse field research over the last two years in Thailand (much of which is now under peer review).  I then described the pressures facing the two species (as per Section 4.2 of the framework) and assessed the risk of the various pressures to them as guided (as per Sections 4.3-4.5 of the framework). I drew on the framework’s suggestions in assigning the four categories of risk from fishing, trade and habitat pressures: low, moderate, high and unknown (see Appendix A).   Section 5 evaluates the capacity of existing management to mitigate the risks I had just identified by considering whether existing management is appropriate for the risks, whether it is being implemented, and whether it is indeed effective at reducing the identified pressures in support of   145 sustainable seahorse populations and so sustainable trade (Figure 6.1, Section 5).   I based my evaluation on Thai marine management measures in place as of Dec 31, 2015 (DoF, 2015) (Table 6.2 and Appendix B).  I then evaluated the implementation of such management measures, defined as either a) stakeholders following the rules (compliance) or b) authorities taking action to ensure rules are followed (enforcement). The framework infers management effectiveness from evidence of stable or increasing (seahorse) population sizes over time. I then did a second evaluation of appropriate management measures based on spatial overlap of sightings for the two species and known marine management measures (Appendix C). If more than 30% of sightings for either species occurred in management areas, this management was deemed “appropriate” (IUCN, 2015).  Section 6 guides Authorities to make a decision about the NDF (Figure 6.1, Section 6;).  Acting in this context as the government of Thailand, I had to decide on an NDF that was positive, negative or came with conditions.  A positive NDF can be made when all the risks are known and are being managed appropriately and effectively. A negative NDF can be made when risks are not being managed with good results, or are unknown. An NDF with conditions allows for precautionary levels of exports while risks are reduced, gaps in management are addressed, or quality of information is improved.  An NDF with conditions might be assigned when at least one appropriate management measure is in place but improvements on enforcement and data on effectiveness are needed.  Section 7 offers guidance and advice about how to improve management action and/or fill knowledge gaps and uses the framework to inform a national action plan for seahorses (Figure 6.1, Section 7), all in support of adaptive management.   Where risks were not being managed with good   146 results, or were unknown, I identified three key management approaches for each species that were essential to moving forward and creating an action plan.  These were based on which of the many options were (1) the most pressing, (2) the most tractable (likely to succeed), and (3) already required through another policy commitment (such as under an Aichi target).  Such selection criteria will help to focus implementation efforts and encourage pragmatism.   In the current exercise, I did not tackle Section 8, which considers the final steps to take before issuing a permit in situations where a positive NDF can be made for wild populations of seahorses (Figure 6.1, Section 8).  6.4 Results I found a considerable amount of information to describe the pressures on seahorses and evaluate management to inform Section 4 of the NDF framework (Tables 6.1 and 6.2). My literature search yielded five sources of data on fisheries, six sources of data on trade, and 30 sources of data on habitat (Table 6.1).  Information on fisheries and trade primarily emerged from research prompted by the CITES RST recommendations (CITES, 2012a) whereas most information on habitat came from the published literature. I found ten sources of data on appropriate management responses and enforcement (Table 6.2). However, I was unable to find any sources of information on the effectiveness of marine management measures for seahorses as inferred by long-term monitoring of trades, catches or populations.   To inform Section 5 of the NDF framework, I identified six existing management responses that I evaluated as appropriate to address pressures on seahorses in general in Thailand (Table 6.2). None   147 of these has been developed specifically for seahorses, but their implementation should serve to mitigate pressures on seahorses.  I identified appropriate management responses to fisheries pressures (mostly from non-selective gear) that included limited entry, marine protected areas, and spatial and temporal (seasonal closures) gear restrictions (Table 6.2). I found only one management response for pressures from trade, which was the creation of a New Fisheries Management Plan that addresses illegal fisheries and trade (Table 6.2).  I identified marine protected areas, spatial gear restrictions and habitat restoration, as appropriate management responses to habitat pressures (Table 6.2).  There is limited entry for all gears known to catch seahorses (DoF 2015).  I found that declared coverage of marine protected areas in Thailand was adequate - the total is 25% of national waters including 75% of Thailand’s coral reefs and 71% of its seagrass beds (DoF, 2015)- although implementation is a consideration (see below).  Thailand bans trawling within three to five kilometers of land in all coastal provinces, and has implemented three seasonal closures to protect spawning stock and juveniles (DoF, 2015).  Thailand has also developed 96 artificial reefs in Thai waters with the stated aim of restoring fish spawning ground habitats (DoF, 2015)but the trade-off between costs (concentrated populations and fishing) and benefits (more habitat) was hard to determine.  While management measures appropriate to the risks facing H. kuda and H. trimaculatus are in place, evaluating their implementation (i.e. compliance or enforcement) proved challenging because of conflicting data. The majority of data for limited entry, marine protected areas, and spatial and temporal gear restrictions indicated that many fishers did not comply with these measures (Table 6.2), leading me to decide that these were not well implemented. However, unpublished data from the Thai Department of Fisheries enforcement office does indicate that arrests had been made to   148 support these management measures (Table 6.2). In addition, a new fisheries management plan was enacted in late 2015 with the goal of increasing enforcement and compliance (Table 6.2).  It proved easier to confirm the number of artificial reef units deployed and their geospatial locations because of good data coverage on their placement (Table 6.2).  No data were available to evaluate the effectiveness of marine management measures for seahorses (Table 6.2), so I judged the effectiveness of all measures to be “unknown” (Table 6.2).   H. trimaculatus My evaluation of fisheries, trade and habitat pressures on H. trimaculatus with the NDF framework (Section 4) resulted in ten high-risk categories – six in fisheries, three in trade and one in habitat (Table 6.3). I detail my definitions of risk for each category in Appendix A. Five pressures were categorized as moderate risk – two in fisheries and three in habitat. Four categories, three in fisheries and one in habitat, were classified as low risk, and no categories had unknown risk (Table 6.3).   I deemed that no management measures mitigated risks for H. trimaculatus as per Section 5 of the framework (Tables 6.5, 6.6).   First, only 6% of 556 sightings of H. trimaculatus occurred inside marine protected areas, areas with spatial gear restrictions and areas with temporal gear restrictions (Figures 6.2-6.10; Table 6.5.)  Second, just 3% of sightings occurred in the no trawl zones (spatial gear restricted area) for H. trimaculatus (Figure 6.5-6.7).  Third, a mere 2% of sightings occurred within marine protected areas (national parks) (Figure 6.2-6.4) or in areas with seasonal fishing closures (Figure 6.8-6.10).  Moreover, limited entry and habitat restoration are only appropriate   149 when combined with marine protected areas or spatial gear restrictions (as per Appendix B: Table 6.6). The only appropriate means to mitigate the fisheries risks facing H. trimaculatus was the new Fisheries Management Plan, but its enforcement and effectiveness remain unknown (Table 6.3).    After considering the data on risks and management as per Section 6 of the framework, I made a negative NDF for H. trimaculatus.  The dearth of management measures to mitigate the risks for this offshore species meant that I could not support ongoing trade (an NDF with conditions) for this species.   As per Section 7 of the NDF framework, I identified the most pressing issue facing H. trimaculatus as unmanaged and unregulated capture in trawling gears.  In order to address this issue I would recommend portside monitoring – to document changes in catch per unit effort over time – and/or also encourage captains to track seahorse catches in logbooks. From the perspective of Thailand and seahorses, the most tractable response was consistent with a commitment the Thai government had already made, to reduce IUU fishing and trade. Continuing efforts to limit entry and increase enforcement measures will help ensure that non-selective fishing is addressed.  Implementing portside monitoring for seahorses would help to identify adverse effects from fishing over time, and would help prompt efforts to combat IUU fishing.  H. kuda My evaluation of fisheries, trade and habitat pressures on H. kuda with the NDF framework (Section 4) resulted in eight high-risk categories – four in fisheries, three in trade and one in habitat (Table 6.4). Four pressures were categorized as moderate risk – one in fisheries and three in habitat.   150 Six categories, five in fisheries and one in habitat, were classified as low risk, and no categories had unknown risk (Table 6.4).   I deemed areas with spatial and temporal gear restrictions as appropriate to mitigate risks for H. kuda as per Section 5 of the framework (Table 6.4; Table 6.6).  I found that all 38 sightings of H. kuda occurred inside at least one of the marine management areas (Figures 6.2-6.10; Table 6.5).  Just 8% of sightings occurred within marine protected areas (national parks) (Figure 6.2-6.4) but all sightings for H. kuda occurred in the no trawl zones (Figure 6.5-6.7).  As well, 74% of sightings were in areas with seasonal fishing closures (Figure 6.8-6.10).  Since limited entry and habitat restoration were combined with marine protected areas or spatial gear restrictions (as per Appendix B), these measures were also helpful (Table 6.4; Table 6.6).  I also deemed the new Fisheries Management Plan as an appropriate response to fisheries risks for H. kuda (Table 6.5), although information on its enforcement and effectiveness remains unknown.  After considering the data on risks and management as per section 6 of the framework, I made an NDF with conditions for H. kuda. Because many management measures had the potential to mitigate the risks for this inshore species, I felt that trade in H. kuda could continue as long as Thailand maintained annual port sampling at several select sites for both commercial and small-scale gears. I would set trade at precautionary levels – e.g. setting a quota for this species that is capped at the mean volume of the number of exports over the last five years – until such time as results from monitoring become available and can inform any required adjustments in management.     151 I identified the most pressing issue facing H. kuda as capture in gillnets, as per Section 7 of the NDF framework. Encouraging gillnet fishers to record their seahorse catches when documenting other catch would help evaluate this threat as would an outreach program encouraging gillnet fishers to throw seahorses back to the ocean if they are pulled up alive. The most tractable issue was enforcing management measures that would reduce threats to habitat, primarily by increasing enforcement in MPAs and monitoring seahorse populations therein.  As with H. trimaculatus, for H. kuda great benefit could come from Thailand’s existing commitment to reduce IUU fishing and trade through limited entry and increases enforcement.  6.5 Discussion  As conservationists taking a dose of my own medicine, with the implementation of an international wildlife treaty, was difficult. My research supports similar findings for terrestrial species, that there are no clear cut answers as to the best way to implement the CITES treaty for an Appendix II listed species (Castello and Stewart, 2010; Smith et al., 2011). In contrast to other CITES case studies, however, I didn’t spend much time thinking about the uncertainty associated with my data or about data requirements; rather, I moved on with implementation in spite of imperfect data (Smith et al., 2011). Instead I focused on reproducible ways to evaluate the data I had into the various risk and management categories. Despite imperfect datasets and the possibility of perception bias, I was able to complete the NDF process for my two case study species.   I found that the process of making an NDF for my two marine fish species required me to address three main questions: 1) what pressures are putting wild populations of your species at risk; 2) is management in place to mitigate the risk or offset those pressures; and 3) are the species responding   152 as hoped to the management ?  For the first of these, despite imperfect data, I was able to assign risk for all the documented pressures facing my species.  As the framework for making NDFs for seahorses was just that, a framework, it was necessarily not prescriptive about how to categorize risk appropriately based on the information available. My challenge, and the challenge facing others is how much you need to know to label the risk as low/medium/high or unknown. For example I struggled to categorize the risk associated with climate change and marine- and land-based activities, because the information was not clearly documented, not readily available, and/or because the impacts of such activities on seahorses can have varying effects (Vincent et al., 2011). I ended up assigning risk based on my expert opinion, but others could have assigned risk differently, with effects on the results. Dealing with uncertainty is common in conservation assessments (Akçakaya et al., 2000; Gillespie et al., 2011), but the costs, benefits and levels of risk that result from making decisions with minimal data  are not well studied in the NDF context (Smith et al., 2011).   I found that making an NDF for two marine fishes came down to the presence of appropriate management, yet this is seldom documented for CITES Appendix II species (Cheung, 1995; Smith et al., 2011). I made a negative NDF for H. trimaculatus because this offshore, deeper-water species was mostly found outside the trawl exclusion zone, outside MPAs and indeed may only be protectively managed by limited entry (for all vessels/gears) (DoF, 2015). However I made an NDF with conditions for my shallow water, inshore species, H. kuda, because (i) management measures are in place to mitigate risks but (ii) I don’t know whether they are enough to offset the pressures on H. kuda (DoF, 2015). The key to making NDFs with conditions is that the conditions must be clearly defined, with implementing actors identified, and timelines for implementation (CITES,   153 2013a). For H. kuda, conditions should most certainly include the establishment of a long-term monitoring for this species – through regular port sampling or monitoring of populations in management areas (Foster et al., 2014; Loh et al., 2014)– to see how populations are faring under the current management regime.  If populations were found to be increasing or stable then trade could continue, but if they were found to be decreasing then trade should be adjusted. If new management measures for H. trimaculatus were to be developed, such as deep water MPAs or offshore spatial closures, then with similar targets, actors and timelines, precautionary trade could be considered for that species.  I found spatial data critical to evaluating the ability of existing management to offset risk to my species during the NDF process, whereas the importance of such information has been largely overlooked in the CITES NDF literature (Rosser, 2008; Rosser and Haywood, 2002; Smith et al., 2011; Thorson and Wold, 2010). My overlay analysis of species distribution and management was the most informative in terms of identifying whether appropriate management was in place to address the pressures on my species, and led to the different outcomes for each species. I recognize that spatial distribution data may not be available for many data-poor marine fishes but I also emphasize that some data can often be cobbled together, as in my case. Species distribution data often exists in local knowledge (Thornton and Scheer, 2012), and is relatively cheap to generate (Aylesworth et al., 2016a) (Chapter 3). Additionally, spatial data lies at the heart of current conservation and management efforts and many current ocean management strategies are spatial (Chape et al., 2005; NOAA, 2014; Stelzenm et al., 2012). Overlaying spatial data on species distribution and management gets to the core of the NDF process – which boils down to whether management is in place to mitigate species risks (CITES, 2013a; Mundy-Taylor et al., 2014).    154  Working through the NDF process identified clear gaps in management, and I used the framework’s guidance to triage these gaps into next steps for advancing implementation of CITES for seahorses in Thailand.  Identifying actions for each species that addressed the most pressing threat, that were most tractable, and that were already prescribed enabled me to maintain momentum with the NDF process. A lack of data can lead to management inaction (Johannes, 1998), and my results for management enforcement (e.g. conflicting data) and effectiveness (no data available) strongly suggest more data are needed. However, by tackling a conservation triage and identifying the most pressing, tractable and already prescribed issues (Bottrill et al., 2008), I was able to suggest how to move forward without more data. Fishing gears with high risk – trawl gears and gillnets - were the most pressing pressures on both of my marine fish species.   The first steps in addressing these issues would be port sampling to record changes in catch per unit effort over time or encouraging fishers to keep track of catch in log-books. The most tractable management options for my species were those related to fisheries reform (e.g. increasing enforcement) and adding seahorse monitoring into existing underwater surveys or establishing new port sampling protocols (DoF, 2015). I designated Thailand’s fisheries reform to address IUU fishing and trade as the “already prescribed action” (DoF, 2015). By focusing on either the most pressing, tractable or already prescribed actions to address species risk as I did in this NDF exercise, I could prioritize management actions to use available resources effectively (Bottrill et al., 2008).  Similar to many conservation assessment procedures, my research highlighted the importance of expert opinion in the NDF process (Martin et al., 2012; Pitcher et al., 2013; Rodrigues et al., 2006). I identified four steps in the NDF process where expert opinion and the possibility for perception   155 bias played an important role: (i) determining how data fit into the various categories; (ii) determining the risk based on available information; (iii) evaluating conflicting data; and (iv) determining the NDF outcome. The authors were lucky to have worked closely with the CITES Scientific and Management Authorities in Thailand, and therefore had access to a substantial amount of data that may not have otherwise been publicly available. I recognize that other experts might have come up with different results, according to their own experience and knowledge (e.g. Donlan et al., 2010 for sea turtles). Such differences are particularly notable in evaluating conflicting data, such as support for and against enforcement of existing management. While I was a conservationist attempting to implement CITES as if I was a government actor, I recognize that my results might have differed if the socio-economic realities of people’s livelihoods had indeed been my responsibility.   The NDF framework for seahorses suggests Parties consult the recommendations for documentation, research and action – issued by the CITES Animals Committee through the CITES Review of Significant Trade – as they plan remedial action in support of sustainable trade (CITES, 2013a). In the RST process, the CITES Animals Committee gave Thailand ten recommendations to help it move towards positive NDFs (CITES, 2012a); these were actually suggested by the last author on this paper (pers.comm.).  My analysis in this paper showed that some of these recommendations would be more useful than others in assisting Thailand to evaluate their seahorse trade and move forward with sustainable management. Thailand’s obligation to clarify the legal protection and current management measures in place for seahorses was important in identifying what management was appropriate for each species (CITES, 2012a). CITES request that Thailand establish a detailed monitoring program for seahorses, remains critical for evaluating indicators of   156 risk from fishing, and management effectiveness.   The recommended adaptive management program for seahorses would also be important in terms of setting in place measures to ensure enforcement and effectiveness of seahorse management, key factors for making a positive NDF (CITES, 2012a).  Obviously adaptive management requires the manager to embark on action that appears reasonable at the time, and will be adjusted with further research.  In that context, CITES’ recommendation that Thailand implement additional spatial and temporal restrictions on fishing activities was appropriate as an initial set of measures (and have since been shown important, in association with reductions in vessel numbers (Aylesworth et al., 2016c)).  However, two of CITES’ recommendations in the RST might not have been immediate priorities, even if they were useful in the longer-term.  These include conducting a detailed study of life history parameters and modeling population response to exploitation pressure or examining the feasibility of returning live seahorses taken as bycatch to the sea (CITES, 2012a).  My new experience of tackling this NDF process suggests that future RST recommendations focus on identifying the spatial overlap of species, threats and management. This would assist in evaluating whether current management is appropriate to species’ threats, and highlight where to focus management action.   Once an NDF has been made, Parties then must decide how to react to their findings, both in terms of improving sustainability and in a broader context.  In the case of a negative NDF, it might seem to make sense to establish a zero quota, forbidding export for the moment.  The difficulty is that a zero quota may do very little to help a Party move towards eventual sustainable exports, because these seahorses are obtained in bycatch for economically important fisheries.  As well, zero quotas often drive trade underground rather than stopping it (Lemieux and Clarke, 2009).  Instead, a government might be advised immediately to enhance its existing management, particularly in   157 redoubling efforts for management it is already obliged to execute under other obligations.  For all NDFs, parties will need also to meet their additional obligations to ensure that specimens are legally sourced (CITES, 2013a).  My experiment in implementing CITES did not consider this aspect so I again under-represented the challenges a Party faces in implementing CITES.    The path from policy change to sustainability involves finding creative solutions that move societies, spaces and species towards sustainable management (Meffe and Viederman, 1995).  Rarely do conservationists place themselves in the role of policy maker or government actors tasked with implementing policy changes. What became evident from my work, was the importance of moving forward despite imperfect data, in a documented and justified way that could allow for future adaptive management (Meffe and Viederman, 1995). Most conservation studies inevitably call for more data (Hamann et al., 2010; Young and Van Aarde, 2011) in a failed quest for perfect advice (Johannes, 1998).  Yet government does not have the luxury of waiting until such an unlikely scenario emerges and must plunge forward with imperfect knowledge.  It was only when we take off our conservation hats and place ourselves in the role of those responsible for implementation that we realize taking a dose of conservation medicine isn’t as easy as it sounds.  Greater respect for these challenges, mature conversation with managers, and a pause for reflection before issuing recommendations might go a long way towards bridging the gap between science and policy.    158  Figure 6.1 Flow chart describing the Non-Detriment Finding Framework for Seahorses. Section numbers refer to sections of the NDF framework.   159  Figure 6.2 Locations of H. kuda and H. trimaculatus observations from research trawls, scientific surveys and citizen science contributions in relation to national parks on the Andaman Coast.    160  Figure 6.3 Locations of H. kuda and H. trimaculatus observations from research trawls, scientific surveys and citizen science contributions in relation to national parks on the Southern Gulf Coast.    161  Figure 6.4 Locations of H. kuda and H. trimaculatus observations from research trawls, scientific surveys and citizen science contributions in relation to national parks on the Upper Gulf Coast.    162  Figure 6.5 Locations of H. kuda and H. trimaculatus observations from research trawls, scientific surveys and citizen science contributions in relation to no trawl zones on the Andaman Coast.    163  Figure 6.6 Locations of H. kuda and H. trimaculatus observations from research trawls, scientific surveys and citizen science contributions in relation to no trawl zones on the Southern Gulf of Thailand.    164  Figure 6.7 Locations of H. kuda and H. trimaculatus observations from research trawls, scientific surveys and citizen science contributions in relation to no trawl zones on the Upper Gulf of Thailand.    165  Figure 6.8 Locations of H. kuda and H. trimaculatus observations from research trawls, scientific surveys and citizen science contributions in relation to seasonal closures on the Andaman coast.    166  Figure 6.9 Locations of H. kuda and H. trimaculatus observations from research trawls, scientific surveys and citizen science contributions in relation to seasonal closures on the Southern Gulf coast.    167  Figure 6.10 Locations of H. kuda and H. trimaculatus observations from research trawls, scientific surveys and citizen science contributions in relation to seasonal closures on the Upper Gulf coast.    168  Table 6.1 Datasets available to evaluate pressures and risk on H. kuda and H. trimaculatus. Pressure Dataset Fisheries Diversity of fishing methods / gears (Aylesworth et al., 2016b) (CITES, 2013a) (Laksanawimol et al., 2013) (Perry et al., 2010) Fishing mortality (Aylesworth et al., 2016b) Fishing selectivity (Aylesworth et al., 2016b) Discarding practices Aylesworth, unpublished Indicators of fishing impacts Aylesworth, unpublished Trade Diversity of use Aylesworth, unpublished Kuo et al (in prep) (Laksanawimol et al., 2013) (Perry et al., 2010) IUU fishing / trade Aylesworth, unpublished (DoF, 2015) (European Commission, 2015) Kuo et al (in prep) (Perry et al., 2010)    169 Table 6.1 (con’t) Datasets available to evaluate pressures and risk on H. kuda and H. trimaculatus. Pressure Dataset Indicators of trade impacts (domestic & international) Aylesworth, unpublished Kuo et al (in prep) (Laksanawimol et al., 2013) (Perry et al., 2010) Habitat Habitat specialization (Aylesworth et al., 2016a) Aylesworth, unpublished Department of Fisheries, unpublished (Lourie et al., 2004) Marine based activities (Bech, 2002) (Chuenpagdee and Pauly, 2003) (Dearden and Manopawitr, 2011) (FAO, 2010b) (Halpern et al., 2008) (Janekitkosol et al., 2003) (Lian and Bradley, 1986) (Lunn and Dearden, 2006a) (Lymer et al., 2010)  (Wattayakorn, 2012) (Yasue and Dearden, 2009)    170 Table 6.1 (con’t) Datasets available to evaluate pressures and risk on H. kuda and H. trimaculatus. Land based activities (Cheevaporn and Menasveta, 2003) (DoF, 2015) (Grech et al., 2012) (Hughes et al., 2009) (Thampanya et al., 2006) (Thongra-ara and Preeda, 2002) (Wattayakorn, 2012) Climate change (Bennett et al., 2014) (Bennett and Dearden, 2013) (Brown et al., 2002) (Dearden and Manopawitr, 2011) (DoF, 2015) (Grech et al., 2012) (Short et al., 2011)  (Tanzil et al., 2009) (Vincent et al., 2011) (Yeemin et al., 2009)      171 Table 6.1 (con’t) Datasets available to evaluate pressures and risk on H. kuda and H. trimaculatus. Indicators of habitat health (Bennett et al., 2016) (Cheevaporn and Menasveta, 2003) (DoF, 2015) (Grech et al., 2012) (Hughes et al., 2009) (Tanzil et al., 2009) (Thampanya et al., 2006) (Thongra-ara and Preeda, 2002) (Wattayakorn, 2012) (Yeemin et al., 2009)     172 Table 6.2 Datasets available to evaluate management measures related to seahorse species generally.  Appropriate  Management Response Why it is appropriate Dataset  Data on enforcement Data on effectiveness  Limited entry;    Addresses incidental capture from fishing gears because spatial restrictions on fishing gears catching seahorses are also in place Aylesworth, unpublished (fisher) (DoF, 2015) (Teh et al., 2015) (Willmann et al., 2000) Lack of compliance IUU Fishing & enforcement Data on IUU fishing Challenges with limited entry No No No No No Marine Protected Areas  Provides a protected area for species captured incidentally in fishing gears, covers known seahorse habitats Aylesworth, unpublished (fisher) (Bennett and Dearden, 2014) (DoF, 2015) (Lunn and Dearden, 2006a) (Willmann et al., 2000) Lack of compliance Lack of compliance IUU fishing & enforcement Lack of compliance Challenges with enforcement No No No No No    173 Table 6.2 (Con’t) Datasets available to evaluate management measures related to seahorse species generally. Appropriate  Management Response Why it is appropriate Dataset  Data on enforcement Data on effectiveness  Gear restriction-spatial Provides a protected area for species captured incidentally in fishing gears, includes shallow water seahorse habitats (Anuchiracheeva et al., 2003) Aylesworth, unpublished (fisher) (DoF, 2015) (Willmann et al., 2000) Lack of compliance  Lack of compliance Enforcement challenges Enforcement challenges No No No No  Gear restriction-temporal Provides protection when gear restrictions coincide with peak seahorse reproduction periods Aylesworth, unpublished (fisher) (DoF, 2015) (Panjarat and Bennett, 2012) (Willmann et al., 2000) Lack of compliance; Data on IUU fishing Lack of compliance Enforcement challenges No No No No No     174 Table 6.2 (Con’t) Datasets available to evaluate management measures related to seahorse species generally. Appropriate  Management Response Why it is appropriate Dataset  Data on enforcement Data on effectiveness  Thai Fishery Management Plan 2015-2019 Outlines management challenges facing Thailand and details actions and management measures to address overfishing, overcapacity and IUU fishing  (DoF, 2015) IUU fishing & enforcement challenges No Habitat restoration Provides additional habitat when combined with permanent no-take marine protected areas or spatial gear restrictions. Aylesworth, unpublished (fisher) DoF unpublished data (Willmann et al., 2000)  Artificial reef benefits; # of artificial reefs; Restoration began in 1985 No No No   175  Table 6.3 Evaluation of data availability, risk, management response, enforcement and effectiveness of management for H. trimaculatus in Thailand.  Category Risk Reason Appropriate Management Response Enforced? Yes/ no/unknown Effective? Yes/ no/unknown Most pressing issue Most tractable issue Already committed issue Gear diversity High 5 of 6 gears catch H. trimaculatus Limited entry    No  Unknown     Fishing mortality    Catchability in fishing gears High Greater than 50% of port sampled individuals are H. trimaculatus None      Relative fishing gear pressure - otter Trawls High Greater than 50% of port sampled individuals are H. trimaculatus Limited entry   No  Unknown   X      176 Table 6.3 (Con’t) Evaluation of data availability, risk, management response, enforcement and effectiveness of management for  H. trimaculatus in Thailand. Category Risk Reason Appropriate Management Response Enforced? Yes/ no/unknown Effective? Yes/ no/unknown Most pressing issue Most tractable issue Already committed issue Relative fishing gear pressure - pair Trawls High Greater than 50% of port sampled individuals are H. trimaculatus Limited entry   No  Unknown  X   Relative fishing gear pressure - gillnets High Between 30-50% of port sampled individuals are H. trimaculatus but fleet size is > 30,000 fishers Limited entry No  Unknown        177 Table 6.3 (Con’t) Evaluation of data availability, risk, management response, enforcement and effectiveness of management for  H. trimaculatus in Thailand. Category Risk Reason Appropriate Management Response Enforced? Yes/ no/unknown Effective? Yes/ no/unknown Most pressing issue Most tractable issue Already committed issue Relative fishing gear pressure – purse seines  Low Less than 30% of port sampled individuals are H. trimaculatus Limited entry   No  Unknown     Relative fishing gear pressure- pushnets Low  Less than 30% of port sampled individuals are H. trimaculatus Limited entry   No  Unknown         178 Table 6.3 (Con’t) Evaluation of data availability, risk, management response, enforcement and effectiveness of management for  H. trimaculatus in Thailand. Category Risk Reason Appropriate Management Response Enforced? Yes/ no/unknown Effective? Yes/ no/unknown Most pressing issue Most tractable issue Already committed issue Fishing selectivity    Overfishing indicator (catch under length at maturity) Low Less than 30% of port sampled H. trimaculatus are under length at maturity None      Overfishing indicator  (sex bias) High Ratio of males to females caught significantly less than unity None         179 Table 6.3 (Con’t) Evaluation of data availability, risk, management response, enforcement and effectiveness of management for  H. trimaculatus in Thailand. Category Risk Reason Appropriate Management Response Enforced? Yes/ no/unknown Effective? Yes/ no/unknown Most pressing issue Most tractable issue Already committed issue Discarding Moderate Only gillnets – but large # of fishers None      Adverse Fishing impacts High Greater than 50% fisher reported decline since 1997 New Fisheries Management Plan – includes limited entry, and reductions in capacity Unknown Unknown     Uses in trade High Many uses for SH in general None        180 Table 6.3 (Con’t) Evaluation of data availability, risk, management response, enforcement and effectiveness of management for  H. trimaculatus in Thailand. Category Risk Reason Appropriate Management Response Enforced? Yes/no/ unknown Effective? Yes/no/ unknown Most pressing issue Most tractable issue Already committed issue IUU fishing/trade High EU yellow card New Fisheries Management Plan Unknown Unknown  X X Indicators of trade High Increase in price > 50% None      Habitat specialization Low Found in many habitats – seagrass, sandy soft bottoms,  Habitat restoration Yes Unknown    Marine based activities High  Tourism, Shipping, Trawling, Dredging / filling, Gas exploration Habitat restoration Yes Unknown      181 Table 6.3 (Con’t) Evaluation of data availability, risk, management response, enforcement and effectiveness of management for  H. trimaculatus in Thailand. Category Risk Reason Appropriate Management Response Enforced? Yes/no/ unknown Effective? Yes/no/ unknown Most pressing issue Most tractable issue Already committed issue Land based activities Moderate Land based activities known to occur, are on-going but impact is reversible None      Climate change Moderate Some climate change threats have been documented in seahorse habitats; but also documentation of recovery in country None No  Unknown       182 Table 6.3 (Con’t) Evaluation of data availability, risk, management response, enforcement and effectiveness of management for  H. trimaculatus in Thailand. Category Risk Reason Appropriate Management Response Enforced? Yes/no/ unknown Effective? Yes/no/ unknown Most pressing issue Most tractable issue Already committed issue Indicators habitat health Moderate  Declines in habitat health reported for some but not all habitat types None No Unknown         183 Table 6.4 Evaluation of data availability, risk, management response, enforcement and effectiveness of management for H. kuda in Thailand.  Category Risk Reason Management Response Enforced? Yes/ no/unknown Effective? Yes/ no/unknown Most pressing issue Most tractable issue Already committed issue Gear diversity High 4 of 6 gears catch H. kuda Limited entry;  MPAs;  Gear restriction-Spatial;  Gear restriction-Temporal;  No No No   No Unknown Unknown Unknown   Unknown    Fishing mortality    Catchability in fishing gears Low Less than 30% of port sampled individuals are H. kuda None         184 Table 6.4 Evaluation of data availability, risk, management response, enforcement and effectiveness of management for H. kuda in Thailand. Category Risk Reason Management Response Enforced? Yes/no/ unknown Effective? Yes/no/ unknown Most pressing issue Most tractable issue Already committed issue Relative fishing gear pressure - otter trawls Low Less than 30% of port sampled individuals are H. kuda Limited entry;  MPAs;  Gear restriction-Spatial;  Gear restriction-Temporal; No No No   No Unknown Unknown Unknown   Unknown    Relative fishing gear pressure – pair trawls Low Less than 30% of port sampled individuals are  H. kuda Limited entry;  MPAs;  Gear restriction-Spatial;  Gear restriction-Temporal; No No No No Unknown Unknown Unknown Unknown      185 Table 6.4 (Con’t) Evaluation of data availability, risk, management response, enforcement and effectiveness of management for H. kuda in Thailand. Category Risk Reason Management Response Enforced? Yes/no/ unknown Effective? Yes/no/ unknown Most pressing issue Most tractable issue Already committed issue Relative fishing gear pressure - gillnets High Between 30-50% of port sampled individuals are H. kuda but fleet size is > 10,000 fishers MPAs;  Gear restriction-Temporal; No No Unknown Unknown X   Relative fishing gear pressure- pushnets Low  Less than 30% of port sampled individuals are H. kuda Limited entry;  MPAs;  Gear restriction-Spatial;  Gear restriction-Temporal; No No No No Unknown Unknown Unknown Unknown      186 Table 6.4 (Con’t) Evaluation of data availability, risk, management response, enforcement and effectiveness of management for  H. kuda in Thailand. Category Risk Reason Management Response Enforced? Yes/no/ unknown Effective? Yes/no/ unknown Most pressing issue Most tractable issue Already committed issue Fishing selectivity    Overfishing indicator (catch under length at maturity) High Greater than 50% of H. kuda port sampled individuals are under length at maturity None      Overfishing indicator  (sex bias) Low Ratio of males to females caught is not statistically significant (p> 0.1) None         187 Table 6.4 (Con’t) Evaluation of data availability, risk, management response, enforcement and effectiveness of management for  H. kuda in Thailand. Category Risk Reason Management Response Enforced? Yes/no/ unknown Effective? Yes/no/ unknown Most pressing issue Most tractable issue Already committed issue Discarding Moderate Only gillnets – but large # of fishers None      Adverse Fishing impacts High Greater than 50% fisher reported decline since 1997 New Fisheries Management Plan Unknown Unknown     Uses in trade High Many uses for SH in general None      IUU fishing/trade High EU yellow card New Fisheries Management Plan Unknown Unknown   X Indicators of trade High Increase in price > 50% None        188 Table 6.4 (Con’t) Evaluation of data availability, risk, management response, enforcement and effectiveness of management for  H. kuda in Thailand. Category Risk Reason Management Response Enforced? Yes/no/ unknown Effective? Yes/no/ unknown Most pressing issue Most tractable issue Already committed issue Habitat specialization Low Found in many habitats – coral reefs, mangroves, seagrass, sandy soft bottoms, sponges Habitat restoration;  MPAs; Gear restrictions -spatial Yes  No No Unknown  Unknown Unknown    Marine based activities High  At least one activity documented to show severe damage – including loss of habitat that is takes > 1 yr. to regrow and / or threat is on-going Habitat restoration;  MPAs Gear restrictions -spatial Yes  No No Unknown  Unknown Unknown  X     189 Table 6.4 (Con’t) Evaluation of data availability, risk, management response, enforcement and effectiveness of management for  H. kuda in Thailand. Category Risk Reason Management Response Enforced? Yes/no/ unknown Effective? Yes/no/ unknown Most pressing issue Most tractable issue Already committed issue Land based activities Moderate Threat is ongoing but reversible None      Climate change Moderate Some climate change threats have been documented in seahorse habitats; but also documentation of recovery in country; or long term effects of climate change on habitats is uncertain MPAs Gear restrictions-spatial No No Unknown Unknown       190 Table 6.4 (Con’t) Evaluation of data availability, risk, management response, enforcement and effectiveness of management for  H. kuda in Thailand. Category Risk Reason Management Response Enforced? Yes/no/ unknown Effective? Yes/no/ unknown Most pressing issue Most tractable issue Already committed issue Indicators habitat health Moderate  Declines in habitat health reported for some but not all habitat types and information on what habitats are important at various life stages is unknown MPAs Gear restrictions-spatial No No Unknown Unknown  X       191 Table 6.5 Summary of spatial overlap of marine management measures and sightings of H. trimaculatus and H. kuda.  H. trimaculatus H. kuda Number of sightings (n) 556 38 Management % sightings % sightings marine protected areas 2 8 no trawl zones 3 100 seasonal closures 2 74 inside management 6 100      192 Table 6.6 Detailed information on fishing gears & management measures specifically as they relate to H. trimaculatus and H. kuda.    H. trimaculatus H. kuda Risk Gear Appropriate Management Response Enforced? Yes/ no/unknown Effective? Yes/ no/unknown Appropriate Management Response Enforced? Yes/no/ unknown Effective? Yes/no/ unknown High Otter trawl New Fisheries Management Plan Unknown Unknown  Gear restriction-Spatial;  Gear restriction-Temporal; Habitat restoration New Fisheries Management Plan Unknown Unknown Yes Unknown Unknown Unknown Unknown Unknown High Pair trawl New Fisheries Management Plan Unknown Unknown  Gear restriction-Spatial;  Gear restriction-Temporal; Habitat restoration New Fisheries Management Plan Unknown Unknown Yes Unknown Unknown Unknown Unknown Unknown    193 Table 6.6 (Con’t) Detailed information on fishing gears & management measures specifically as they relate to H. trimaculatus  and H. kuda.   H. trimaculatus H. kuda Risk Gear Appropriate Management Response Enforced? Yes/ no/unknown Effective? Yes/ no/unknown Appropriate Management Response Enforced? Yes/no/ unknown Effective? Yes/no/ unknown High Gillnet New Fisheries Management Plan Unknown Unknown Gear restriction-Temporal; New Fisheries Management Plan Unknown Unknown Unknown Unknown Low Purse Seine New Fisheries Management Plan  Unknown Unknown  Gear restriction-Temporal; New Fisheries Management Plan Unknown Unknown Unknown Unknown Low Pushnet New Fisheries Management Plan Unknown Unknown Gear restriction-Spatial;  Gear restriction-Temporal; New Fisheries Management Plan Habitat restoration Unknown Unknown Unknown Yes Unknown Unknown Unknown Unknown   194 Chapter 7: Conclusion 7.1 Overview My thesis explored the most expedient way to develop conservation action for data-poor marine fishes, using seahorses as a case study. In my first two data chapters (Chapter 2 and 3), I focused on how to identify where data-poor marine fishes live because spatial data is essential to marine conservation and management (Chape et al., 2005; Douvere, 2008). In Chapters 4 and 5, I explored the most efficient way to determine fisheries management advice given limited data. Such advice is both timely and relevant because more than 80% of global fish catch comes from un-assessed fisheries (Costello et al., 2012). With my final data chapter I changed perspective from that of a provider of conservation advice to one implementing that advice, as I used the inputs of my previous chapters to evaluate the conservation consequences of seahorse trade in Thailand.  I now review the specific research questions and responses that my thesis developed and explored.  7.2 Research Findings  Below I highlight the key contributions from each of my thesis chapters in terms of developing conservation action for data-poor species. My results from Chapter 2 are helpful in to determine sampling effort by underwater divers in the face of considerable conservation concern. I found in Chapter 3 that fisher knowledge played an invaluable role in building an understanding of species distributions, and demanded relatively less effort and money than other sources of data.  In Chapter 4, I identify a role for vulnerability analysis in evaluating fisheries management challenges for data-poor marine fishes.  In Chapter 5, my use of data-poor stock assessments determined that seahorse populations can only sustain low fishing mortality rates and that size at   195 recruitment to the fishery is a key influence on population dynamics. With Chapter 6, I was forced to confront the reality that it is very difficult to evaluate the threats that trade imposes on wild populations, even with quite tailored tools.    A major strength of my research is the diverse set of approaches I used to solve creatively the challenges related to conservation of data-poor species. In Chapter 2, I trialed an analytical technique not commonly used in the ocean (presence / absence with detection probabilities) (Issaris et al., 2012; Katsanevakis et al., 2012) to understand where to find my cryptic and data-poor fishes of interest. In Chapter 3, I compared fisher knowledge to three external datasets to determine where seahorses live and to identify the most expedient and cost effective method to generate spatial data for data-poor fishes. In Chapter 4, I explored the use of a data-poor fishery assessment method, vulnerability analysis (Honey et al., 2010), to determine whether I could advance management action for my non-target data-poor fishes. With Chapter 5, I examined the use of a data-poor stock assessment (Honey et al., 2010) to understand how such methods may advance management advice for non-target data-poor fishes.  Finally, in Chapter 6, I simulated international policy implementation for two seahorse species (Chapter 6) to explore how to develop and maintain momentum to implement CITES for marine fishes.  For each of my five analytical chapters, I now consider how my findings apply more broadly to both seahorses and marine conservation.    196 7.2.1 Research Question 1: What Is The Most Efficient Way To Search For Cryptic, Rare Or Data-poor Marine Species In The Field? (Chapter 2) My results have important implications for seahorse research because all other seahorse studies have assumed a detection probability = 1, such that they may have underestimated seahorse counts. (e.g. Bell et al., 2003; Curtis and Vincent, 2005; Harasti et al., 2014, 2010; Perante et al., 2002; Rosa et al., 2007; Teske et al., 2007; Vincent et al., 2005) . For example, Yasué et al., (2012) would have done well to incorporate detection probabilities before declaring that marine protected areas had no significant effect on densities of a seahorse species (H. comes) in the Philippines. This is particularly true as I would expect detection probabilities to decrease in more complex environments, such as seagrass beds and coral reefs (Green et al., 2013). Additionally marine protected areas may often be more complex than heavily fished sites (Green et al., 2013).   The results of my research allow me to make five key deductions about how generally to maximize resources appropriately for initial field surveys of cryptic, rare, or data-poor species. First, factoring in detection probabilities changes the conclusions and reduces bias in estimates about locations of marine or terrestrial species of interest (Bailey et al., 2007; Monk, 2013). Deploying a presence / absence with detection probability framework increased my confidence that sites with zeros represented the true absence of species, in line with similar research on land (Kéry and Schmidt, 2008; MacKenzie et al., 2006). This is critical if one is not to commit a Type II error of assuming absence of an organism when it is present, a serious matter when one is considering remedial management action for a species of conservation concern. Second, selecting sites based upon prior knowledge is entirely reasonable if one is trying to figure out where the animals can be found in order to mobilize protection as soon as possible, highlighting   197 the value of local knowledge for research on data-poor species (Thornton and Scheer, 2012). Third, species presence and absence across the land or seascape is useful for conservation planning because such data forms the basis for identifying sites for protection (Weller, 2008) and creating monitoring protocols (Pellet and Schmidt, 2005), both critical steps during the initial phase of management.  Fourth, an emphasis on the presence/absence framework allows meaningful engagement by stakeholders and thus increases dialogue with government in a way that should help generate valuable cooperation for conservation.  Fifth, an appropriate acceptance of a presence/absence framework in rapid or early conservation responses can keep costs manageable as the early initiatives are developed.  Evaluation of relative abundances is important but can come later in the conservation triage, with ongoing collaboration with stakeholders, and after more resources have been marshalled.   7.2.2 Research Question 2: What Is The Most Efficient Way To Generate Spatial Data For A Data-poor Species? (Chapter 3) My results add to the growing body of seahorse literature highlighting the value of fisher knowledge for informing conservation (Meeuwig et al., 2003; O’Donnell et al., 2010; O’Donnell et al., 2010; Perry et al., 2010; Rosa et al., 2005). Most seahorse research is conducted in the field on wild animals (e.g. Harasti et al., 2014; Rosa et al., 2007), in aquaculture (e.g. Hilomen-Garcia et al., 2001; Woods et al., 2003), or through trade surveys and analysis (e.g. Baum and Vincent, 2005; Foster et al., 2016). Of these, only field studies generate spatial data on where seahorses live in the wild, and such studies typically have small spatial coverage (Bell et al., 2003; Curtis and Vincent, 2005; Harasti et al., 2014). For example, if I needed to identify sites to   198 prioritize for seahorse conservation or future research in Mexico, I would be unable to do so based on the trade surveys conducted there by Baum and Vincent, 2005. Since the number of spatial ocean management measures is increasing, a lack of spatial knowledge about basic seahorse distribution in Mexico, in this example, would prevent me from evaluating if any current management may be benefitting seahorses.   My results indicate that different types of data gathering have merit in different contexts or for different concerns, with a need for complementary bottom up and top down approaches.  People embarking on conservation endeavors for data-poor species may wish to begin with local interviews and use them to inform application of scientific surveys, industry research or citizen science programs. If there is a need to design and implement a species-specific management program (NOAA, 2014), my research suggests that local fisher knowledge or citizen scientists can provide an informed starting point for determining species distribution. Such information can be refined to a species level at a later time through more advanced methods. If the objective is to identify key habitat areas as part of large-scale conservation planning (NOAA, 2014), then my research highlights that fisher knowledge is the most expeditious method to do so. If the management objective is to identify species threats, than my research, similar to others (Thornton and Scheer, 2012) supports that stakeholder interviews are the most cost-effective data collection method. While fisher interviews proved beneficial to identify a path forward for management action, they may not be an appropriate starting point in all circumstances because of perception bias, shifting baselines or the risk that respondents are telling researchers what they want to hear (Panjarat and Bennett, 2012; Stake et al., 2005).    199 7.2.3 Research Question 3: How Can We Best Discern The Scale Of The Fisheries Problem? (Chapter 4) My research on these indiscriminate fisheries revealed that they have a larger impact on seahorse populations than previously realized, both in terms of numbers removed and relative gear pressure (Foster et al., 2016; Lawson et al., in press; Vincent et al., 2011). The large annual catch estimate from Thailand calls into question the latest global estimate of seahorse bycatch (37 million individuals) (Lawson et al., in press). My estimate of catch by commercial vessels (19 million individuals) is larger than the one previous provided for all gear in Thailand (~ 9 million individuals: Lawson et al., in press). My results expand previous research on the impact of trawl by-catch on seahorses (Baum et al., 2003; Perry et al., 2010; Vincent et al., 2011) and their capture in small-scale gears (Lawson et al., 2015). I also highlight that volumes of seahorse capture in small-scale gillnets are of a similar scale to volumes from commercial pair trawlers. I also show, for the first time, that not all commercial gears necessarily exerted strong pressures on seahorses, at least in Thailand. Despite documentation of the destructive nature of pushnets operating in coastal waters (Changsang and Poovachiranon, 1994; Tokrisna et al., 1997), the annual capture by pushnets of Thai seahorses was low compared to other gears both in terms of catch rate and number of boats involved in the fishery. Additionally, I found the Thai purse-seine fleet had higher than expected catches, despite the fact that bycatch from this gear type is not well recorded in the seahorse literature (Lawson et al., in press).   The cumulatively high seahorse catches from both commercial and small-scale gears emphasize the vulnerability of non-target marine fishes to indiscriminate gears and the importance of a sound management response. Non-target marine fishes are typically of low priority in data-  200 collection procedures, which results in challenges assessing fisheries-related impacts (Honey et al., 2010). My research confirms there is value to modifying data-poor fishery assessment methods such as vulnerability analyses to match available species-specific data (Brown et al., 2013; Tuck et al., 2011). The procedure I used to generate data for the vulnerability analysis – drawing on available life history characteristics (Froese and Pauly 2016) and data from several months of interviews and port sampling – could be replicated for other groups of non-target fishes that lack sophisticated data.  It seems that vulnerability analysis may be an acceptable compromise in the trade off between gathering data and making effective decisions; it lies between data-rich methods like fishery stock assessments and the least data intensive method, extrapolation from a closely related species (Honey et al., 2010).  7.2.4 Research Question 4: How Can We Best Determine What Fisheries Management Responses Might Work Best? (Chapter 5) The work in this chapter allowed me to provide four insights into appropriate fisheries management for seahorses beyond the model species, H. trimaculatus. First, because size at recruitment to the fishery was a key model parameter, previous recommendations such as a minimum size limit (Foster and Vincent, 2005) were indeed beneficial for precautionary management of seahorses. Unfortunately for non-target fisheries, such measures may not be effective because fishers make decisions about where to fish and which gears to use based on the target species (Tim M Daw, 2008; Maina et al., 2011), and indiscriminate gears do not select for seahorse size. Second, for seahorse species caught in complex fisheries (e.g. by multiple gears targeting different species) I found that no single management measure (e.g. marine protected areas), was as effective as a combination of all management measures to increase future biomass.   201 Third, this research provides evidence of a direct relationship between catch per unit effort and seahorse abundance, which highlights the importance of port or landings sampling as part of a seahorse management plan.  Random fishing effort is one of the key assumptions needed to support the possibility of a direct relationship between changes in catch per unit effort and abundance (Hilborn and Walters, 1992), and fishers in Thailand determined fishing locations  based on the presence of target species (pers.comm.). Monitoring changes in catch per unit effort based on landings or port sampling can inform fisheries managers of changes in abundance of wild seahorse populations. Fourth, future seahorse stock assessments should continue to build off an age-structured model where habitat is linked to recruitment carrying capacity, because seahorses have strong ties to habitat (Foster and Vincent, 2004; Vincent et al., 2011).    The benefits of conducting a data-poor fishery stock assessment could also accrue to other non-target, data-poor fishes, in three ways (Carruthers et al., 2014). First, I was able to obtain the information required for such a data-poor stock assessment (Carruthers et al., 2014) by reconstructing historic catches based on government data on the number of fishing vessels. While I was initially skeptical about the accuracy of vessel number estimates, such information is likely to be accessible in most countries, and methods are available to account for potential inaccuracies (Teh et al., 2015; Walters and Martell, 2002; Zeller and Pauly, 2015). Second, once historical catch data are generated, expert knowledge may be applied to explore assumptions about historical depletions and fish life history through model variations (Carruthers et al., 2014). Third, now that the stock assessment model is built, it will require relatively little effort to update as more data become available or as expertise is acquired to incorporate uncertainty (Bentley and Stokes, 2009b; Jiao et al., 2011).    202  7.2.5 Research Question 5: How Can We Get Practical Movement Implementing CITES At A National Level? (Chapter 6) The focus of Chapter 6 was on implementing CITES for seahorses in Thailand, and I discuss how my findings from this research apply more broadly to seahorses in the next section (7.3). Much of what I learned in Chapter 6 from going from the provider of conservation advice, to the one implementing that advice, is broadly applicable for other marine fishes on CITES. Making a positive NDF for two marine fishes came down to the presence of appropriate management, yet management is seldom actually documented for CITES Appendix II species (Cheung, 1995; Smith et al., 2011). Spatial data are critical to this process, as suggested in Chapters 2 and 3, and yet their importance has also been largely overlooked in the CITES NDF literature (Rosser, 2008; Rosser and Haywood, 2002; Smith et al., 2011; Thorson and Wold, 2010). By working through the NDF process I identified clear gaps in management by using a conservation triage approach (Bottrill et al., 2008), I was able to advance implementation of CITES by identifying which management issues were the most pressing, most feasible or already prescribed. These suggestions are applicable to other marine fishes on CITES and to managers devising conservation measures for data-poor species.   7.3 Practical Applications Of My Research After discussing how my findings apply more broadly to seahorses and conservation, I now turn to the practical applications of my thesis. My research on seahorse distribution (Chapters 2 and 3) contributes to CITES implementation where, as is often the case, a country needs to gather   203 vital data expediently to assess populations and their trajectories under different proposed management or trade regulations (Vincent et al. 2013).   In particular, I addressed the importance of spatial data on distribution, threats and management in supporting CITES implementation for Parties trading in Appendix II species.  The CITES Animals Committee recommended that Thailand undertake studies to identify and monitor areas of high seahorse density and use this information to consider spatial area restrictions for fisheries (CITES, 2012b).  Initially, however, Thailand could not locate areas with seahorses, presenting an impasse to moving forward with CITES implementation.  A practical adaptation of this recommendation would be to gather local knowledge to inform a presence absence study incorporating detection probabilities to generate spatially defined data on wild populations and fishing effort (Johannes, 1998; Mackenzie et al., 2002).   My findings from vulnerability and port sampling analyses (Chapter 4) used minimal data to improve the understanding of the diversity of fishing gears exerting pressure on seahorses and the potential indicators of fisheries impacts (CITES, 2013a). In the spirit of adaptive management (Walters and Holling, 1990), my results indicate the value for Thailand of focusing on commercial trawl and small-scale gillnet gears since they captured the largest numbers and exerted pressure on the most vulnerable species (H. kelloggi, H. kuda, H. trimaculatus). Incidentally, both H. histrix and H. spinosissimus would benefit from management measures aimed at trawling gears, even though these species scored the lowest in terms of pressure from such gears. Given that value of spatial tools for trawl fisheries management, it is fortunate that Thailand has already embraced marine national parks and no trawl zones (CITES, 2013a). Thailand’s marine fisheries landscape is complex because of multi-gear, multi-species fisheries.   204 The analyses I undertook would be both simpler and more tractable in countries where fewer fishing gears are deployed.  My work on data-poor stock assessment methods (Chapter 5) can be useful in three ways. First, implementation of CITES recommendations usually, as with Thailand, requires management action quickly (Vincent et al., 2013). Such tight timelines make it difficult for Parties to conduct original research to generate additional data so they must commonly work off existing knowledge. As with seahorses in Thailand, data regarding historical vessel numbers were already available in the national fishery ministry enabling the estimation of historical catches for a data-poor stock assessment. Second, a recommendation of a data-poor stock assessment may provide common ground for fisheries ministries tasked with implementing CITES but are unfamiliar with how to take appropriate action. In this instance the recommendation of a data-poor stock assessment under CITES provides familiarity to fishery managers in terms of procedure and language with the added benefit of being logistically expeditious. Third, a recommendation of a data-poor stock assessment in the CITES context is complementary to national fisheries management. International pressure can provide the impetus to focus sustainable fisheries management on non-target species like seahorses, and encourage analysis, management and / or fisheries reform that should be on going at national levels.   In terms of implementing CITES for seahorses in Thailand, my assessment of the impact of exports (Chapter 6) differed by species.  I made a negative NDF for H. trimaculatus because this offshore, deeper-water species was mostly found outside the trawl exclusion zone, outside MPAs and indeed may only be protectively managed by limited entry (for all vessels/gears) (DoF,   205 2015). However I made an NDF with conditions for H. kuda because this shallow water, inshore species, has (i) management measures in place to mitigate risks but (ii) it remains unknown whether such measures are enough to offset the pressures on H. kuda (DoF, 2015). The key to making NDFs with conditions is that the terms must be clearly defined, with implementing actors identified, and timelines for implementation (CITES, 2013a). For H. kuda, conditions should most certainly include the establishment of a long-term monitoring for this species – through regular port sampling or monitoring of populations in management areas (Foster et al., 2014; Loh et al., 2014)– to see how populations are faring under the current fisheries and trade management regime, and to adjust export permitting appropriately.   I identified the most tractable conservation options to be increased enforcement of existing marine management and/or creation of a monitoring plan for seahorses in the wild or in fisheries landings. For monitoring of wild populations, efforts should focus on the current marine protected areas where, local knowledge suggests, there may be more seahorses (Aylesworth, unpublished). New monitoring data could help improve parameters relating to marine protected areas in the stock assessment model of Chapter 5. Fisheries landings data could be generated through the continued recording of seahorse catch in the Thai government research trawls, which are mandated to continue.    If new management measures for H. trimaculatus were to be developed, such as deep water MPAs or offshore spatial closures, then the challenge (as in inshore areas) would lie in fisher compliance and government enforcement of such measures (DoF, 2015; Lunn and Dearden, 2006b; Panjarat and Bennett, 2012). Previous research has shown some success with concepts   206 such as freezing the current fishing footprint or eliminating fishing in areas that are not used frequently by fishers (Ardron et al., 2007; Grech and Coles, 2011), but successful implementation depends largely on a consultative process with stakeholders that incorporates an appropriate balance of cajoling and rewarding.  Part of the motivation would be provided by Thailand’s current move to install mandatory vessel monitoring systems in its commercial fishing fleet (P. Suvanachai, pers.comm).  7.4 Broader Implications / Future Directions  While I focused on implementing CITES for seahorses, the results of my research are relevant to implementing other international treaties for marine species such as the Convention on Migratory Species (CMS). The CMS is an international agreement concerned with the conservation of migratory wildlife and their habitats on a global scale (CMS, 1979). As with CITES, various range states can agree to work together to protect listed animals, and their habitats, and to mitigate activities that may threaten them (Miller and Prideaux, 2013).  The lessons I gathered from CITES and seahorses - such as the importance of spatial data, local knowledge, and using conservation triage to prioritize management action- are relevant for those countries party to CMS, where data-poor migratory species require conservation action (Miller and Prideaux, 2013). Indeed many range states in Europe and Pacific Island Countries and Territories have made precautionary agreements under the CMS related to migratory cetacean species, and such advice may promote continued pro-active steps to promote conservation of these marine mammals (Lascelles et al., 2014; Miller and Prideaux, 2013).    207 Increasing global commitments to marine conservation have created an urgent need for practical advice on how to implement international conservation goals at the national level (Smith et al., 2011). While advice is increasingly available about making decisions in the face of imperfect technical knowledge (Akçakaya et al., 2000; Gillespie et al., 2011; Johannes, 2000; Martin et al., 2012), other factors such as political support, organizational structures, and leadership can also hinder or enable effective management and decision making (Greig et al., 2013).  What became evident in my experience for seahorses in Thailand, as in other contexts, was the importance of local leadership in championing conservation action (Greig et al., 2013; Gutiérrez et al., 2011; Martin et al., 2012).  While strong leadership on its own is not sufficient for conservation success, it is increasingly recognized as a notable attribute for effective and successful management (Greig et al., 2013).  This would certainly be true in the CITES context, where national fisheries or oceans agencies are suddenly having to embrace conservation roles that (i) were previously the responsibility of forest or environment agencies, and (ii) often seem quite contrary to the historic fisheries and oceans emphasis on production (Vincent et al 2013).  With numerous governmental agencies tasked with managing different aspects of the marine environment (e.g.DoF, 2015), finding a leader to champion cross-organizational issues of marine resource sustainability appears to be a combination of luck and circumstance.   Advancing species conservation in a complex world is not a simple task. The state of affairs in Thailand as in many other countries, exemplifies that in a crisis driven world, advancing marine conservation requires rapid action (Bottrill et al., 2008). In the Thailand context, issues of social justice (e.g. slavery in the fishing industry), economic hardships associated with overfishing and illegal, unregulated and unreported fishing, are three invidious issues at the forefront of   208 reconciling fisheries and conservation. Government agencies with mandates to manage the marine environment are less than enthusiastic about collaborating with each other and intense distrust between government agencies and non-governmental organizations stifles productive collaboration. Both the Thailand government and the nation’s fisheries are in a state of reform, lending both a sense of urgency and uncertainty to policy changes. All of these issues are urgent and pressing in their own right and it is in this context that I found myself working to advance the conservation agenda of data-poor marine species.   The key to developing action for data-poor marine and terrestrial species alike is figuring out how to use what little information is available to move forward. If I urgently needed to take conservation action for another data-poor species – marine or terrestrial, I would focus on identifying three pieces of information: identifying threats, current management, and metrics of management effectiveness.  I would certainly deploy data-poor fishery assessment methods – which are growing in use (Bentley and Stokes, 2009a, 2009b; Carruthers et al., 2014; Honey et al., 2010; Schroeter et al., 2009; Smith et al., 2009), and can support the CITES process, seafood certification schemes (Pelc et al., 2015) and other instruments intended to reduce unsustainable take.  After reviewing available information and identifying knowledge gaps, I would conduct a conservation triage (Bottrill et al., 2008) as per Chapter 6 to identify the most pressing, most likely to succeed, and already prescribed management issues. I would then use these issues to review and revise (or create) the relevant management plan. Gone are the days when management inaction could be justified because of a lack of data (Johannes, 1998) because paths to move management forward in spite of imperfect data are increasingly available.   209 Bibliography Ahmed, M., Boonchuwongse, P., Dechboon, W., Squires, D., 2007. Overfishing in the Gulf of Thailand: policy challenges and bioeconomic analysis. Environ. Dev. Econ. 12, 145. doi:10.1017/S1355770X06003433 Akçakaya, H.R., Ferson, S., Burgman, M.A., Keith, D.A., Mace, G.M., Todd, C.R., 2000. Making consistent IUCN classifications under uncertainty. Conserv. Biol. 14, 1001–1013. Al-Abdulrazzak, D., Pauly, D., 2013. Managing fisheries from space: Google Earth improves estimates of distant fish catches. ICES J. Mar. Sci. J. du Cons. fst178. Amin, R., Thomas, K., Emslie, R.H., Foose, T.J., Strien, N. Van, 2006. An overview of the conservation status of and threats to rhinoceros species in the wild. Int. Zoo Yearb. 40, 96–117. doi:10.1111/j.1748-1090.2006.00096.x Andrew, N.L., Mapstone, B.D., 1987. Sampling and the description of spatial pattern in marine ecology. Ocean. Mar.Biol. Ann.Rev. 25, 39–90. Anticamara, J.A., Watson, R., Gelchu, A., Pauly, D., 2011. Global fishing effort (1950-2010): Trends, gaps, and implications. Fish. Res. 107, 131–136. doi:10.1016/j.fishres.2010.10.016 Anuchiracheeva, S., Demaine, H., Shivakoti, P., Ruddle, K., 2003. Systematizing local knowledge using GIS: fisheries management in Bang Saphan Bay, Thailand. Ocean Coast. Manag. 46, 1049–1068. doi:10.1016/j.ocecoaman.2004.01.001 Ardron, J.A., Jamieson, G.S., Hangaard, D., 2007. Spatial identification of closures to reduce the by-catch of corals and sponges in the groundfish trawl fishery, British Columbia, Canada. Bull. Mar. Sci. 81, 157–167. Atkinson, M., Kerrigan, B., Roelofs, A., Smith, T., 2008. Non detriment finding for CITES-listed corals in the Queensland coral fishery (No. 9), Working Group. Mexico.   210 Aylesworth, L., Phoonsawat, R., Suvanachai, P., Vincent, A.C.J., 2016a. Generating spatial data for marine conservation and management. under Revis. 1–37. Aylesworth, L., Phoonsawat, R., Vincent, A.C.J., 2016b. Effects of indiscriminate fisheries on small data-poor species. under Rev. Aylesworth, L., Phoonsawat, R., Walters, C.J., Vincent, A.C.J., 2016c. Developing a data-poor species stock assessment using seahorses as a case study. under Rev. 1–21. Bailey, L.L., Hines, J.E., Nichols, J.D., MacKenzie, D.I., 2007. Sampling design trade-offs in occupancy studies with imperfect detection: examples and software. Ecol. Appl. 17, 281–90. Baum, J.K., Meeuwig, J.J., Vincent, A.C.J., 2003. Bycatch of lined seahorses (Hippocampus erectus) in a Gulf of Mexico shrimp trawl fishery. Fish. Bull. 101, 721–731. Baum, J.K., Vincent, A.C.J., 2005. Magnitude and inferred impacts of the seahorse trade in Latin America. Environ. Conserv. 32, 305–319. doi:10.1017/S0376892905002481 Beaudreau, A.H., Levin, P.S., 2014. Advancing the use of local ecological knowledge for assessing data-poor species in coastal ecosystems. Ecol. Appl. 24, 244–256. doi:10.1890/13-0817.1 Beaudreau, A.H., Levin, P.S., Norman, K.C., 2011. Using folk taxonomies to understand stakeholder perceptions for species conservation. Conserv. Lett. 4, 451–463. doi:10.1111/j.1755-263X.2011.00199.x Bech, M., 2002. Imposex and tributyltin contamination as a consequence of the establishment of a marina , and increasing yachting activities at Phuket Island , Thailand 117, 421–429. Bell, E.M., Lockyear, J.F., McPherson, J.M., Marsden, A.D., Vincent, A.C.J., 2003. First field studies of an endangered South African seahorse, Hippocampus capensis. Enviro. Biol.   211 Fishes 67, 35–46. Bellmann, C., Tipping, A., Sumaila, U.R., 2015. Global trade in fish and fishery products: an overview. Mar. Policy 1–8. doi:10.1016/j.marpol.2015.12.019 Bender, L.C., Weisenberger, M.E., Rosas-Rosas, O.C., 2014. Occupancy and habitat correlates of javelinas in the southern San Andres Mountains, New Mexico. J. Mammal. 95, 1–8. doi:10.1644/13-MAMM-A-151 Bennett, E.L., 2015. Legal ivory trade in a corrupt world and its impact on African elephant populations. Conserv. Biol. 29, 54–60. doi:10.1111/cobi.12377 Bennett, N.J., Dearden, P., 2014. Why local people do not support conservation: Community perceptions of marine protected area livelihood impacts, governance and management in Thailand. Mar. Policy 44, 107–116. doi:10.1016/j.marpol.2013.08.017 Bennett, N.J., Dearden, P., 2013. A picture of change: using photovoice to explore social and environmental change in coastal communities on the Andaman Coast of Thailand. Local Environ. Int. J. Justice Sustain. 18, 983–1001. Bennett, N.J., Dearden, P., Murray, G., A., K., 2014. The capacity to adapt?: communities in a changing climate, environment, and economy on the northern Andaman coast of Thailand. Ecol. Soc. 19, 1–5. Bennett, N.J., Dearden, P., Peredo, A.M., 2016. Vulnerability to multiple stressors in coastal communities: a study of the Andaman coast of Thailand. Clim. Dev. 7, 124–141. doi:10.1080/17565529.2014.886993 Bentley, N., 2015. Data and time proverty in fisherires estimaion: potential approaches and solutions. ICES J. Mar. Sci. 72, 186–193. doi:10.1093/icesjms/fsu023 Bentley, N., Stokes, K., 2009a. Contrasting paradigms for fisheries management decision   212 making: how well do they serve data-poor fisheries? Mar. Coast. Fish. 1, 391–401. doi:10.1577/C08-044.1 Bentley, N., Stokes, K., 2009b. Moving fisheries from data-poor to data-sufficient: evaluating the costs of management versus the benefits of management. Mar. Coast. Fish. 1, 378–390. doi:10.1577/C08-045.1 Berkeley, S.A., Hixon, M.A., Larson, R.J., Love, M.S., 2004. Fisheries sustainability via protection of age structure and spatial distribution of fish populations. Fisheries 29, 23–32. Berkes, F., 2009. Evolution of co-management: role of knowledge generation, bridging organizations and social learning. J. Environ. Manage. 90, 1692–702. doi:10.1016/j.jenvman.2008.12.001 Biggs, D., Courchamp, F., Martin, R., Possingham, H.P., 2013. Legal trade of Africa’s rhino horns. Science (80-. ). 339, 1038–1039. doi:10.1126/science.1229998 Bjorkland, R., Dunn, D.C., McClure, M., Jennot, J., Bellman, M.A., Gleason, M., Schiffers, K., 2015. Spatiotemporal patterns of rockfish bycatch in US west coast groundfish fisheries: opportunities for reducing incidental catch of depleted species. Can. J. Fish. Aquat. Sci. 26, 6201–6207. doi:10.1139/cjfas-2014-0242 Bland, L.M., Collen, B., Orme, C.D.L., Bielby, J., 2014. Predicting the conservation status of data-deficient species. Conserv. Biol. 29, 250–259. doi:10.1111/cobi.12372 Boersma, P.D., Parrish, J.K., 1999. Limiting abuse: marine protected areas, a limited solution. Ecol. Econ. 31, 287–304. doi:10.1016/S0921-8009(99)00085-3 Bohensky, E.L., Maru, Y., 2011. Indigenous knowledge, science, and resilience: What have we learned from a decade of international literature on “integration”? Ecol. Soc. 16, 6. Bottrill, M.C., Joseph, L.N., Carwardine, J., Bode, M., Cook, C., Game, E.T., Grantham, H.,   213 Kark, S., Linke, S., McDonald-Madden, E., Pressey, R.L., Walker, S., Wilson, K.A., Possingham, H.P., 2009. Finite conservation funds mean triage is unavoidable. Trends Ecol. Evol. 24, 183–184. doi:10.1016/j.tree.2008.11.007 Bottrill, M.C., Joseph, L.N., Carwardine, J., Bode, M., Cook, C., Game, E.T., Grantham, H., Kark, S., Linke, S., McDonald-Madden, E., Pressey, R.L., Walker, S., Wilson, K.A., Possingham, H.P., 2008. Is conservation triage just smart decision making? Trends Ecol. Evol. 23, 649–54. doi:10.1016/j.tree.2008.07.007 Boutson, A., Mahasawasde, C., Mahasawasde, S., Tunkijjanukij, S., Arimoto, T., 2009. Use of escape vents to improve size and species selectivity of collapsible pot for blue swimming crab Portunus pelagicus in Thailand. Fish. Sci. 75, 25–33. doi:10.1007/s12562-008-0010-z Bowles, E., Trites, A.W., 2013. Faecal DNA amplification in Pacific walruses (Odobenus rosmarus divergens). Polar Biol. 1–5. Bowman, M., 2013. A tale of two CITES: divergent perspectives upon the effectiveness of the wildlife trade convention. Rev. Eur. Community Int. Environ. Law 22, 228–238. doi:10.1111/reel.12049 Bragdon, S.H., 1992. National sovereignty and global environmental responsibility: can the tension be reconciled for the conservation of biological diversity? Harvard Int. Law J. 33, 381–392. Branch, T.A., Hilborn, R., Haynie, A.C.., Fay, G., Flynn, L., Griffiths, J., Marshall, K.N., Randall, J.K., Scheuerell, J.M., Ward, E.J., Young, M., 2006. Fleet dynamics and fishermen behavior: lessons for fisheries managers. Can. J. Fish. Aquat. Sci. 63, 1647–1668. doi:10.1139/F06-072 Braschler, B., 2009. Successfully implementing a citizen-scientist approach to insect monitoring   214 in a resource-poor country. Bioscience 59, 103–104. doi:10.1525/bio.2009.59.2.2 Braunisch, V., Home, R., Pellet, J., Arlettaz, R., 2012. Conservation science relevant to action: A research agenda identified and prioritized by practitioners. Biol. Conserv. 153, 201–210. doi:10.1016/j.biocon.2012.05.007 Brewer, D., Heales, D., Milton, D., Dell, Q., Fry, G., Venables, B., Jones, P., 2006. The impact of turtle excluder devices and bycatch reduction devices on diverse tropical marine communities in Australia’s northern prawn trawl fishery. Fish. Res. 81, 176–188. doi:10.1016/j.fishres.2006.07.009 Britton, J.R., Pegg, J., Gozlan, R.E., 2011. Quantifying imperfect detection in an invasive pest fish and the implications for conservation management. Biol. Conserv. 144, 2177–2181. doi:10.1016/j.biocon.2011.05.008 Brooks, E.N., Powers, J.E., Cortés, E., 2010. Analytical reference points for age-structured models: Application to data-poor fisheries. ICES J. Mar. Sci. 67, 165–175. doi:10.1093/icesjms/fsp225 Brown, E., Clarke, R., Warwick, M., 2002. Serial patterns of biodiversity change in corals across shallow reef flats in Ko Phuket, Thailand, due to the effects of local (sedimentation) and regional (climatic) perturbations. Mar. Biol. 141, 21–29. doi:10.1007/s00227-002-0810-0 Brown, S.L., Reid, D., Rogan, E., 2013. A risk-based approach to rapidly screen vulnerability of cetaceans to impacts from fisheries bycatch. Biol. Conserv. 168, 78–87. doi:10.1016/j.biocon.2013.09.019 Brussard, P.F., 1991. The role of ecology in biological conservation. Ecol. Appl. 1, 6–12. Buitrago, J., Guada, H.J., Doyle, E., 2008. Conservation science in developing countries: an inside perspective on the struggles in sea turtle research and conservation in Venezuela.   215 Environ. Sci. Policy 11, 562–578. doi:10.1016/j.envsci.2008.03.002 Byg, A., Theilade, I., Nielsen, M.R., Lund, J.F., 2012. Local ecological knowledge and its relevance for management and research. Dev. Briefs Policy 16, 1–5. Caddy, J.F., 2004. Current usage of fisheries indicators and reference points, and their potential application to management of fisheries for marine invertebrates. Can. J. Fish. Aquat. Sci. 61, 1307–1324. doi:10.1139/f04-132 Caddy, J.F., Mahone, R., 1995. Reference points for fishery management. Cadrin, S.X., Howe, A.B., Correia, S.J., Currier, T.P., 1995. Evaluating the effects of two coastal mobile gear fishing closures on finfish abundance off Cape Cod. North Am. J. Fish. Manag. 15, 300–315. Cahill, A.J., Walker, J.S., Marsden, S.J., 2006. Recovery within a population of the Critically Endangered citron-crested cockatoo Cacatua sulphurea citrinocristata in Indonesia after 10 years of international trade control. Oryx 40, 1–7. doi:10.1017/S0030605306000366 Carpenter, A.I., Robson, O., Rowcliffe, J.M., Watkinson, A.R., 2005. The impacts of international and national governance changes on a traded resource: a case study of Madagascar and its chameleon trade. Biol. Conserv. 123, 279–287. doi:10.1016/j.biocon.2004.11.015 Carruthers, T.R., Punt, A.E., Walters, C.J., MacCall, A., McAllister, M.K., Dick, E.J., Cope, J., 2014. Evaluating methods for setting catch limits in data-limited fisheries. Fish. Res. 153, 48–68. doi:10.1016/j.fishres.2013.12.014 Carwardine, J., Klein, C.J., Wilson, K.A., Pressey, R.L., Possingham, H.P., 2009. Hitting the target and missing the point: target-based conservation planning in context. Conserv. Lett. 2, 3–10. doi:10.1111/j.1755-263X.2008.00042.x   216 Castellanos-Galindo, G.A., Cantera, J.R., Espinosa, S., Mejía-Ladino, L.M., 2011. Use of local ecological knowledge, scientist’s observations and grey literature to assess marine species at risk in a tropical eastern Pacific estuary. Aquat. Conserv. Mar. Freshw. Ecosyst. 21, 37–48. doi:10.1002/aqc.1163 Castello, L., Stewart, D.J., 2010. Assessing CITES non-detriment findings procedures for Arapaima in Brazil. J. Appl. Ichthyol. 26, 49–56. doi:10.1111/j.1439-0426.2009.01355.x CBD, 2010. Convention on Biological Diversity Aichi Biodiversity Targets [WWW Document]. URL https://www.cbd.int/sp/targets/ (accessed 5.6.16). CBD, 1992. United Nations Conference on Environment and development [WWW Document]. URL https://www.cbd.int/history/ (accessed 2.2.16). Changsang, H., Poovachiranon, S., 1994. The distribution and species compostion of seagrass beds along the Andaman Sea coast of Thailand. Phuket Mar. Biol. Cent. Res. Bull 59, 43–52. Chape, S., Harrison, J., Spalding, M., Lysenko, I., 2005. Measuring the extent and effectiveness of protected areas as an indicator for meeting global biodiversity targets. Philos. Trans. R. Soc. Lond. B. Biol. Sci. 360, 443–55. doi:10.1098/rstb.2004.1592 Cheevaporn, V., Menasveta, P., 2003. Water pollution and habitat degradation in the Gulf of Thailand. Mar. Pollut. Bull. 47, 43–51. doi:10.1016/S0025-326X(03)00101-2 Cheung, J., 1995. Implementation and enforcement of CITES: an assessment of tiger and rhinoceros conservaiton policy in Asia. Pacific Rim Law Policy J. 5, 125–160. Christensen, V., 1998. Fishery-induced changes in a marine ecosystem: insight from models of the Gulf of Thailand. J. Fish Biol. 53, 128–142. Chuenpagdee, R., Pauly, D., 2003. The Gulf of Thailand trawl fisheries, Fisheries.   217 Chullasorn, P., 1998. Status of fishery resources in the Andaman Sea coast of Thailand; Community-based fisheries management in Phang-nga Bay, Thailand, in: National Workshop on Community-Based FIsheries Management. FAO, Regional Office for Asia and the Pacific, Bangkok, pp. 70–83. CITES, 2014. Review of Significant Trade of Appendix II Species, Animals Committee Working Group 7. Veracruz, Mexico. doi:10.1017/CBO9781107415324.004 CITES, 2013a. Building in-country capacity to undertake non-detriment findings with regard to Hippocampus species in Indonesia, Thailand and Viet Nam. Geneva, Switzerland. CITES, 2013b. Sharks and rays listed at CoP 16. https://cites.org/eng/prog/shark/sharks.php. CITES, 2013c. Compliance and Enforcement: Review of Significant Trade. 62nd Meet. Standing Comm. 1–29. CITES, 2012a. Review of significant trade in specimens of Appendix-II species, 26th Meeting of the Animals Committee. doi:10.1017/CBO9781107415324.004 CITES, 2012b. Review of Significant Trade in Specimens of Appendix-II Species, 26th Meeting of the Animals Committee. doi:10.1017/CBO9781107415324.004 CITES, 1999. Implementation of Resolution Conf 8.9: recommendations of the Standing Committee. Geneva. CITES, 1973. Convention on International Trade of Endangered Species of Fauna and Flora. Washington, D.C. CMS, 1979. Convention on Migratory Species. Bonn, Germany. Collier, B.A., Groce, J.E., Morrison, M.L., Newnam, J.C., Campomizzi, A.J., Farrell, S.L., Mathewson, H.A., Snelgrove, R.T., Carroll, R.J., Wilkins, R.N., 2012. Predicting patch occupancy in fragmented landscapes at the rangewide scale for an endangered species: an   218 example of an American warbler. Divers. Distrib. 18, 158–167. doi:10.1111/j.1472-4642.2011.00831.x Cooney, R., Jepson, P., 2012. The international wild bird trade: what’s wrong with blanket bans? Oryx 40, 18–23. doi:10.1017/S0030605306000056 Cooney, R., Lang, A.T.F., 2007. Taking uncertainty seriously: adaptive governance and international trade. Eur. J. Int. Law 18, 523–551. doi:10.1093/ejil/chm030 Cope, J.M., Punt, A.E., 2009. Length-based reference points for data-limited situations: applications and restrictions. Mar. Coast. Fish. 1, 169–186. doi:10.1577/C08-025.1 Costello, C., Ovando, D., Hilborn, R., Gaines, S.D., Deschenes, O., Lester, S.E., 2012. Status and solutions for the world’s unassessed fisheries. Science (80-. ). 338, 517–520. doi:10.1126/science.1223389 Cowling, R.M., Knight, A.T., Faith, D.P., Ferrier, S., Lombard, A.T., Driver, A., Rouget, M., Maze, K., Desmet, P.G., 2004. Nature conservation requires more than a passion for species. Conserv. Biol. 18, 1674–1676. Crall, A.W., Newman, G.J., Stohlgren, T.J., Holfelder, K.A., Graham, J., Waller, D.M., 2011. Assessing citizen science data quality: an invasive species case study. Conserv. Lett. 4, 433–442. doi:10.1111/j.1755-263X.2011.00196.x Cullet, P., Kameri-Mbote, A.P., 1996. Dolphin bycatches in tuna fisheries: a smokescreen hiding the real issues. Ocean Dev. Int. Law 27, 333–348. Cullis-Suzuki, S., Pauly, D., 2010. Failing the high seas: A global evaluation of regional fisheries management organizations. Mar. Policy 34, 1036–1042. doi:10.1016/j.marpol.2010.03.002 Curtis, J.M.R., Vincent, A.C.J., 2008. Use of population viability analysis to evaluate CITES trade-management options for threatened Marine Fishes. Conserv. Biol. 22, 1225–1232.   219 doi:10.1111/j.1523-1739.2008.00994.x Curtis, J.M.R., Vincent, A.C.J., 2005. Distribution of sympatric seahorse species along a gradient of habitat complexity in a seagrass-dominated community. Mar. Ecol. Prog. Ser. 291, 81–91. Davies, R.W.D., Cripps, S.J., Nickson, A., Porter, G., 2009. Defining and estimating global marine fisheries bycatch. Mar. Policy 33, 661–672. Davies, T.D., Baum, J.K., 2012. Extinction risk and overfishing: reconciling conservation and fisheries perspectives on the status of marine fishes. Sci. Rep. 2, 561. doi:10.1038/srep00561 Daw, T.M., 2008. Spatial distribution of effort by artisanal fishers: exploring economic factors affecting the lobster fisheries of the Corn Islands, Nicaragua. Fish. Res. 90, 17–25. doi:10.1016/j.fishres.2007.09.027 Daw, T.M., 2008. How fishers count and management. de Solla, S.R., Shirose, L.J., Fernie, K.J., Barrett, G.C., Brousseau, C.S., Bishop, C.A., 2005. Effect of sampling effort and species detectability on volunteer based anuran monitoring programs. Biol. Conserv. 121, 585–594. doi:10.1016/j.biocon.2004.06.018 Dearden, P., Manopawitr, P., 2011. Climate change–coral reefs and dive tourism in South-east Asia, in: Disappearing Destinations: Climate Change and Future Challenges for Coastal Tourism. UK: CAB International. pp. 144–160. DeSanto, E.M., 2013. Missing marine protected area (MPA) targets: how the push for quantity over quality undermines sustainability and social justice. J. Environ. Manage. 124, 137–146. doi:10.1016/j.jenvman.2013.01.033 Diamond, S.L., 2004. Bycatch quotas in the Gulf of Mexico shrimp trawl fishery: can they work?   220 Rev. Fish Biol. Fish. 14, 207–237. Dickinson, J.L., Zuckerberg, B., Bonter, D.N., 2010. Citizen science as an ecological research tool: challenges and benefits. Annu. Rev. Ecol. Evol. Syst. 41, 149–172. doi:10.1146/annurev-ecolsys-102209-144636 DoF, 2015. Marine fisheries management plan of Thailand. Bangkok, Thailand. Donlan, C.J., Wingfield, D.K., Crowder, L.B., Wilcox, C., 2010. Using expert opinion surveys to rank threats to endangered species: a case study with sea turtles. Conserv. Biol. 24, 1586–1595. doi:10.1111/j.1523-1739.2010.01541.x Douvere, F., 2008. The importance of marine spatial planning in advancing ecosystem-based sea use management. Mar. Policy 32, 762–771. doi:10.1016/j.marpol.2008.03.021 Durso, A.M., Willson, J.D., Winne, C.T., 2011. Needles in haystacks: estimating detection probability and occupancy of rare and cryptic snakes. Biol. Conserv. 144, 1508–1515. doi:10.1016/j.biocon.2011.01.020 Engler, R., Guisan, A., Rechsteiner, L., 2004. An improved approach for predicting the distribution of rare and endangered species from occurrence and pseudo-absence data. J. Appl. Ecol. 41, 263–274. doi:10.1111/j.0021-8901.2004.00881.x European Commission, 2015. EU acts on illegal fishing: Yellow card issued to Thailand while South Korea & Philippines are cleared. Press Release. Fairweather, P.G., 1991. Statistical power and design requirements for environmental monitoring. Aust. J. Mar. Freshw. Res. 42, 555–567. doi:10.1071/MF9910555 FAO, 2012. The state of the world fisheries and aquaculture. Rome. FAO, 2010a. Report of the second workshop on the assessment of fishery stock status in South and Southeast Asia. Rome.   221 FAO, 2010b. FAO activities in relation to CITES and commercially exploited aquatic species. Foran, D.R., Crooks, K.R., Minta, S.C., 1997. Species identification from scat: an unambiguous genetic method. Wildl. Soc. Bull. 25, 835–839. Forrest, R.E., McAllister, M.K., Martell, S.J.D., Walters, C.J., 2013. Modelling the effects of density-dependent mortality in juvenile red snapper caught as bycatch in Gulf of Mexico shrimp fisheries: Implications for management. Fish. Res. 146, 102–120. doi:10.1016/j.fishres.2013.04.002 Foster, S.J., Loh, T.-L., Knapp, C., 2014. iSeahorse landings trends toolkit. Foster, S.J., Vincent, A.. C.J., 2005. Enhancing sustainability of the international trade in seahorses with a single minimum size limit. Conserv. Biol. 19, 1044–1050. doi:10.1111/j.1523-1739.2005.00192.x Foster, S.J., Vincent, A.C.J., 2016. Making non-detriment findings for seahorses-a framework. Vancouver. Foster, S.J., Vincent, A.C.J., 2010. Using life-history information to assess potential effects of shrimp trawling on small fishes. J. Fish Biol. 76, 2434–2454. doi:10.1111/j.1095-8649.2010.02631.x Foster, S.J., Vincent, A.C.J., 2004. Life history and ecology of seahorses: implications for conservation and management. J. Fish Biol. 64, 1–61. doi:DOI 10.1111/j.1095-8649.2004.00429.x Foster, S.J., Wiswedel, S., Vincent, A.C.J., 2016. Opportunities and challenges for analysis of wildlife trade using CITES data - seahorses as a case study. Aquat. Conserv. Mar. Freshw. Ecosyst. 1–19. doi:10.1002/aqc.2493 Fox, H.E., Soltanoff, C.S., Mascia, M.B., Haisfield, K.M., Lombana, A. V., Pyke, C.R., Wood,   222 L., 2012. Explaining global patterns and trends in marine protected area (MPA) development. Mar. Policy 36, 1131–1138. doi:10.1016/j.marpol.2012.02.007 Froese, R., 2004. Keep it simple: three indicators to deal with overfishing. Fish Fish. 5, 86–91. doi:10.1111/j.1467-2979.2004.00144.x Garmestani, A.S., Benson, M.H., 2013. A framework for resilience-based governance of social-ecological systems. Ecol. Soc. 18, 1–11. Geometry Pty Ltd, 2008. iGIS. Hobart, Australia. http://www.geometryit.com/ (accessed 10.17.15) Gerrodette, T., Dayton, P.K., Macinko, S., Fogarty, M.J., 2002. Precautionary management of marine fisheries: moving beyond burden of proof. Bull. Mar. Sci. 70, 657–668. Gilchrist, G., Mallory, M., Merkel, F., 2005. Can local ecological knowledge contribute to wildlife management? Case studies of migratory birds. Ecol. Soc. 10, 20. Gillespie, G.R., Scroggie, M.P., Roberts, J.D., Cogger, H.G., Mahony, M.J., McDonald, K.R., 2011. The influence of uncertainty on conservation assessments: Australian frogs as a case study. Biol. Conserv. 144, 1516–1525. doi:10.1016/j.biocon.2010.10.031 Gilman, E., Passfield, K., Nakamura, K., 2014. Performance of regional fisheries management organizations: Ecosystem-based governance of bycatch and discards. Fish Fish. 15, 327–351. doi:10.1111/faf.12021 Goetze, J.S., Fullwood, L.A.F., 2012. Fiji’s largest marine reserve benefits reef sharks. Coral Reefs 32, 121–125. doi:10.1007/s00338-012-0970-4 Golden, A.S., Naisilsisili, W., Ligairi, I., Drew, J.A., 2014. Combining natural history collections with fisher knowledge for community-based conservation in Fiji. PLoS One 9. doi:10.1371/journal.pone.0098036   223 Gordon, D.V., Hannesson, R., Kerr, W.A., 2001. Of fish and whales: the credibility of threats in international trade disputes. J. Policy Model. 23, 83–98. Grech, A., Chartrand-miller, K., Erftemeijer, P., Fonseca, M., Mckenzie, L., Rasheed, M., Taylor, H., 2012. A comparison of threats , vulnerabilities and management approaches in global seagrass bioregions 024006. doi:10.1088/1748-9326/7/2/024006 Grech, A., Coles, R., 2011. Interactions between a trawl fishery and spatial closures for biodiversity conservation in the Great Barrier Reef World Heritage Area, Australia. PLoS One 6, e21094. doi:10.1371/journal.pone.0021094 Green, S.J., Tamburello, N., Miller, S.E., Akins, J.L., Côté, I.M., 2013. Habitat complexity and fish size affect the detection of Indo-Pacific lionfish on invaded coral reefs. Coral Reefs 32, 413–421. doi:10.1007/s00338-012-0987-8 Greig, L.A., Marmorek, D.R., Murray, C., Robinson, D.C.E., 2013. Insight into enabling adaptive management. Ecol. Soc. 18, 1–11. Guillera-Arroita, G., Lahoz-Monfort, J.J., 2012. Designing studies to detect differences in species occupancy: Power analysis under imperfect detection. Methods Ecol. Evol. 3, 860–869. doi:10.1111/j.2041-210X.2012.00225.x Gullison, R.E., Rice, R.E., Blundell, A.G., 2000. “Marketing” species conservation. Nature 404, 923–924. Gutiérrez, N.L., Hilborn, R., Defeo, O., 2011. Leadership, social capital and incentives promote successful fisheries. Nature 470, 386–389. doi:10.1038/nature09689 Hall, G.B., Moore, A., Knight, P., Hankey, N., 2009. The extraction and utilization of local and scientific geospatial knowledge within the Bluff oyster fishery, New Zealand. J. Environ. Manage. 90, 2055–2070. doi:10.1016/j.jenvman.2007.08.022   224 Halliday, R.G., 1988. Use of seasonal spawning area closures in the management of haddock fisheries in the Northwest Atlantic. NAFO Sci. Counc. Stud. 12, 27–36. Halpern, B.S., Walbridge, S., Selkoe, K.A., Kappel, C.B., Micheli, F., D’Agrosa, C., Bruno, J.F., Casey, K.S., Ebert, C., Fox, H.E., Fujita, R., Heinemann, D., Lenihan, H.S., Madin, E.M.P., Perry, M.T., Selig, E.R., Spalding, M., Steneck, R., Watson, R., 2008. A global map of human impact on marine ecosystems. Science (80-. ). 319, 948–952. doi:10.1126/science.1149345 Hamann, M., Godfrey, M.H., Seminoff, J.A., Arthur, K., Barata, P.C.R., Bjorndal, K.A., Bolten, A.B., Broderick, A.C., Campbell, L.M., Carreras, C., Casale, P., Chaloupka, M., Chan, S.K.F., Coyne, M.S., Crowder, L.B., Diez, C.E., Dutton, P.H., Epperly, S.P., FitzSimmons, N.N., Formia, A., Girondot, M., Hays, G.C., Cheng, I.J., Kaska, Y., Lewison, R., Mortimer, J.A., Nichols, W.J., Reina, R.D., Shanker, K., Spotila, J.R., Tomás, J., Wallace, B.P., Work, T.M., Zbinden, J., Godley, B.J., 2010. Global research priorities for sea turtles: informing management and conservation in the 21st century. Endanger. Species Res. 11, 245–269. doi:10.3354/esr00279 Hamilton, R.J., Giningele, M., Aswani, S., Ecochard, J.L., 2012. Fishing in the dark-local knowledge, night spearfishing and spawning aggregations in the Western Solomon Islands. Biol. Conserv. 145, 246–257. doi:10.1016/j.biocon.2011.11.020 Harasti, D., Glasby, T.M., Martin-Smith, K., Bay, N., 2010. Striking a balance between retaining populations of protected seahorses and maintaining swimming nets. Aquat. Conserv. Mar. Freshw. Ecosyst. 20, 159–166. doi:10.1002/aqc Harasti, D., Martin-Smith, K., Gladstone, W., 2014. Ontogenetic and sex-based differences in habitat preferences and site fidelity of White’s seahorse Hippocampus whitei. J. Fish Biol.   225 85, 1–16. doi:10.1111/jfb.12492 Harasti, D., Martin-Smith, K., Gladstone, W., 2012. Population dynamics and life history of a geographically restricted seahorse, Hippocampus whitei. J. Fish Biol. 81, 1297–1314. doi:10.1111/j.1095-8649.2012.03406.x Harrop, S.R., Pritchard, D.J., 2011. A hard instrument goes soft: The implications of the Convention on Biological Diversity’s current trajectory. Glob. Environ. Chang. 21, 474–480. doi:10.1016/j.gloenvcha.2011.01.014 Heng, L.L., 1999. The implementation of the Convention on International Trade in Endangered Species in Singapore. J. Int. Wildl. Law Policy 2, 46–63. Henry, L.A., Hart, M., 2005. Regeneration from injury and resource allocation in sponges and corals - A review. Int. Rev. Hydrobiol. 90, 125–158. doi:10.1002/iroh.200410759 Hiddink, J.G., Jennings, S., Kaiser, M.J., Queirós, A.M., Duplisea, D.E., Piet, G.J., 2006. Cumulative impacts of seabed trawl disturbance on benthic biomass , production , and species richness in different habitats. North 736, 721–736. doi:10.1139/F05-266 Hilborn, R., Walters, C.J., 1992. Quantitative fisheries stock assessment: choice, dynamics and uncertainty. Rev. Fish Biol. Fish. 2, 177–178. Hilomen-Garcia, G. V, Reyes, R.D., Garcia, C.M.H., Anonymous, 2001. Tolerance and growth of juvenile seahorse Hippocampus kuda exposed to various salinities, Aquaculture 2001, Lake Buena Vista, FL (USA), 21-25 Jan 2001. World Aquaculture Society, 143 J.M Parker Coliseum Louisiana State University Baton Rouge LA 70803 USA. Hind, E.J., Hiponia, M.C., Gray, T.S., 2010. From community-based to centralised national management — A wrong turning for the governance of the marine protected area in Apo Island, Philippines? Mar. Policy 34, 54–62. doi:10.1016/j.marpol.2009.04.011   226 Hines, J., 2006. Presence. Hodgson, G., 2004. Reef Check instruction manual: a guide to reef check coral reef monitoring. Reef Check, Inst. of the Environment. Holmes, J., Clark, R., 2008. Enhancing the use of science in environmental policy-making and regulation. Environ. Sci. Policy 11, 702–711. doi:10.1016/j.envsci.2008.08.004 Honey, K.T., Moxley, J.H., Fujita, R.M., 2010. From rags to fishes: data-poor methods for fishery managers. Manag. Data-Poor Fish. Case Stud. Model. Solut. 1, 159–184. Hughes, A.R., Williams, S.L., Duarte, C.M., Heck Jr., K.L., Waycott, M., 2009. Associations of concern: declining seagrasses and threatened dependent species. Front. Ecol. Environ. 7, 242–246. doi:10.1890/080041 Huntsman, G.R., Schaaf, W.E., 1994. Simulation of the impact of fishing on reproduction of a protogynous grouper, the Graysby. North Am. J. Fish. Manag. 14, 41–52. Hurme, E., Mönkkönen, M., Nikula, A., Nivala, V., Reunanen, P., Heikkinen, T., Ukkola, M., 2005. Building and evaluating predictive occupancy models for the Siberian flying squirrel using forest planning data. For. Ecol. Manage. 216, 241–256. doi:10.1016/j.foreco.2005.05.035 Issaris, Y., Katsanevakis, S., Salomidi, M., Tsiamis, K., Katsiaras, N., Verriopoulos, G., 2012. Occupancy estimation of marine species: dealing with imperfect detectability. Mar. Ecol. Prog. Ser. 453, 95–106. doi:10.3354/meps09668 IUCN, 2015. The IUCN Red List of Threatened Species. [WWW Document]. URL http://www.iucnredlist.org (accessed 11.19.15). Janekitkosol, W., Somchanakij, H., Eiamsa-ard, M., Supongpan, M., 2003. Strategic review of the fishery situation in Thailand. Fish. Res. 915–956.   227 Jerde, C.L., Mahon, A.R., Chadderton, W.L., Lodge, D.M., 2011. “Sight-unseen” detection of rare aquatic species using environmental DNA. Conserv. Lett. 4, 150–157. doi:10.1111/j.1755-263X.2010.00158.x Jiao, Y., Cortes, E., K, A., Guo, F., 2011. Poor-data and data-poor species stock assessment using a Bayesian hierarchical approach. Ecol. Appl. 21, 2691–2708. Johannes, R.E., 2000. Ignore fishers’ knowledge and miss the boat. Fish Fish. 1, 257–271. doi:http://dx.doi.org/10.1046/j.1467-2979.2000.00019.x Johannes, R.E., 1998. The case for data-less marine resource management: examples from tropical nearshore fin fisheries. Trends Ecol. Evol. 13, 243–246. Joseph, J., 1994. The tuna‐dolphin controversy in the eastern pacific ocean: biological, economic, and political impacts. Ocean Dev. Int. Law 25, 1–30. doi:10.1080/00908329409546023 Joyner, C.C., Tyler, Z., 2000. Marine conservation versus international free trade: reconciling dolphins with tuna and sea turtles with shrimp. Ocean Dev. Int. Law 31, 127–150. Kaiser, M.J., Collie, J.S., Hall, S.J., Jennings, S., Poiner, I.R., 2002. Modification of marine habitats by trawling activities: prognosis and solutions. Fish Fish. 2, 114–136. Karras, C., Agar, J., 2009. Cruzan fishers’ perspectives on the performance of the Buck Island Reef National Monument and the red hind seasonal closure. Ocean Coast. Manag. 52, 578–585. Katsanevakis, S., Stelzenmüller, V., South, A., Sørensen, T.K., Jones, P.J.S., Kerr, S., Badalamenti, F., Anagnostou, C., Breen, P., Chust, G., D’Anna, G., Duijn, M., Filatova, T., Fiorentino, F., Hulsman, H., Johnson, K., Karageorgis, A.P., Kröncke, I., Mirto, S., Pipitone, C., Portelli, S., Qiu, W., Reiss, H., Sakellariou, D., Salomidi, M., van Hoof, L.,   228 Vassilopoulou, V., Vega Fernández, T., Vöge, S., Weber, A., Zenetos, A., Hofstede, R.T., 2011. Ecosystem-based marine spatial management: Review of concepts, policies, tools, and critical issues. Ocean Coast. Manag. 54, 807–820. doi:10.1016/j.ocecoaman.2011.09.002 Katsanevakis, S., Weber, A., Pipitone, C., Leopold, M., Cronin, M., Scheidat, M., Doyle, T.K., Buhl-Mortensen, L., Buhl-Mortensen, P., D’Anna, G., de Boois, I., Dalpadado, P., Damalas, D., Fiorentino, F., Garofalo, G., Giacalone, V.M., Hawley, K.L., Issaris, Y., Jansen, J., Knight, C.M., Knittweis, L., Kröncke, I., Mirto, S., Muxika, I., Reiss, H., Skjoldal, H.R., Vöge, S., 2012. Monitoring marine populations and communities: Methods dealing with imperfect detectability. Aquat. Biol. 16, 31–52. doi:10.3354/ab00426 Kelleher, K., 2005. Discards in the world’s marine fisheries: an update. Food & Agriculture Org. Kelly, M.J., Klenzendorf, S., Vaughan, M.R., Hutajulu, M.B., Parakkasi, K., 2013. Threatened predator on the equator: multi-point abundance estimates of the tiger Panthera tigris in central Sumatra. Oryx 47, 211–220. Kéry, M., Schmidt, B.R., 2008. Imperfect detection and its consequences for monitoring for conservation. Community Ecol. 9, 207–216. doi:10.1556/ComEc.9.2008.2.10 Koslow, J.A., 2009. The role of acoustics in ecosystem-based fishery management. ICES J. Mar. Sci. J. du Cons. fsp082. Laksanawimol, P., Damrongphol, P., Kruatrachue, M., 2006. Alteration of the brood pouch morphology during gestation of male seahorses, Hippocampus kuda. Mar. Freshw. Res. 57, 497–502. Laksanawimol, P., Petpiroon, S., Damrongphol, P., 2013. Trade of seahorses, Hippocampus Spp. (Actinopterygii: Syngnathiformes: Syngnathidae), on the East Coast of the Gulf of   229 Thailand. Acta Ichthyol. Piscat. 43, 229–235. doi:10.3750/AIP2013.43.3.07 Lascelles, B., Notarbartolo, G., Sciara, D., Agardy, T., Cuttelod, A., Eckert, S., Glowka, L., Hoyt, E., Llewellyn, F., Louzao, M., Ridoux, V., Tetley, M.J., 2014. Migratory marine species: their status , threats and conservation management needs. Aquat. Conserv. Mar. Freshw. Ecosyst. 24, 111–127. doi:10.1002/aqc.2512 Lawson, J.. M., Foster, S.J., Lim, A.C.O., Chong, V.C., Vincent, A.C.J., 2015. Novel life-history data for threatened seahorses provide insight into fishery effects. J. Fish Biol. 86, 1–15. doi:10.1111/jfb.12527 Lawson, J.M., Foster, S.J., Vincent, A.C.J., in press. Small bycatch of small fishes adds up to big numbers… and big trouble? Fish Fish. Lemieux, A.M., Clarke, R.V., 2009. The international ban on ivory sales and its effects on elephant poaching in Africa. Brit. J. Criminol 49, 451–471. doi:10.1093/bjc/azp030 Lewis, A.E., Hammill, M., Power, M., Doidge, D.W., Lesage, V., 2009. Movement and aggregation of Eastern Hudson Bay beluga whales (Delphinapterus leucas): a comparison of patterns found through satellite telemetry and Nunavik traditional ecological knowledge. Arctic 62, 13–24. Lewison, R, L., Freeman, S.A., Crowder, L.B., 2004. Quantifying the effects of fisheries on threatened species: the impact of pelagic longlines on loggerhead and leatherback sea turtles. Ecol. Lett. 7, 221–231. doi:10.1111/j.1461-0248.2004.00573.x Lewison, R.L., Soykan, C., Franklin, J., 2009. Mapping the bycatch seascape: multispecies and multi-scale spatial patterns of fisheries bycatch. Ecol. Appl. 19, 920–930. Lian, H.M., Bradley, K., 1986. Exploration and Development of Natural Gas, Pattani Basin, Gulf of Thailand.   230 Liljeblad, J., 2008. Finding another link in the chain: international treaties and devolution to local law enforcmenet in the case of the Convention on the International Trade in Endangered Species. South. Calif. Interdiscip. Law J. 18, 527–546. Little, A.S., Needle, C.L., Hilborn, R., Holland, D.S., Marshall, C.T., 2015. Real-time spatial management approaches to reduce bycatch and discards: Experiences from Europe and the United States. Fish Fish. 16, 576–602. doi:10.1111/faf.12080 Loh, T.-L., Knapp, C., Foster, S.J., 2014. iSeahorse trends toolkit: finding and surveying wild seahorse populations in support of conservation. Lorenz, M., 1995. International co-operative programme on assessment and monitoring of air pollution effects on forests-ICP Forests. Water. Air. Soil Pollut. 85, 1221–1226. doi:10.1007/BF00477148 Lorenzen, K., Steneck, R.S., Warner, R.R., Parma, A.M., Coleman, F.C., Leber, K.M., 2010. The spatial dimensions of fisheries: putting it all in place. Bull. Mar. Sci. 86, 169–177. Lourie, S.A., Foster, S.J., Cooper, E.W.T., Vincent, A.C.J., 2004. A guide to the identification of seahorses, Project Seahorse and TRAFFIC North America. TRAFFIC USA, Washington, D.C. Lourie, S.A., Pollom, R.A., Foster, S.J., in press. A global annotated checklist of the seahorses Hippocampus Rafinesque 1810 (Actinopterygii: Syngnathiformes): taxonomy and biogeography with recommendations for further research. Zootaxa 1–64. Ludwig, D., Walters, C.J., 1985. Are age-structured models appropriate for catch-effort data? Can. J. Fish. Aquat. Sci. 42, 1066–1072. Lunn, K.E., Dearden, P., 2006a. Fishers’ needs in marine protected area zoning: a case study from Thailand. Coast. Manag. 34, 37–41. doi:10.1080/08920750600567234   231 Lunn, K.E., Dearden, P., 2006b. Monitoring small-scale marine fisheries: An example from Thailand’s Ko Chang archipelago. Fish. Res. 77, 60–71. doi:10.1016/j.fishres.2005.08.009 Lymer, D., Funge-Smith, S., Khemakorn, P., Naruepon, S., Ubolratana, S., 2008. A review and synthesis of capture fisheries data in Thailand: Large versus small-scale fisheries, RAP Publication 2008/17. Bangkok, Thailand. Lymer, D., Funge-Smith, S., Miao, W., 2010. Status and potential of fisheries and aquaculture in Asia and the Pacific 2010. Rome. Mace, G.M., Cramer, W., Díaz, S., Faith, D.P., Larigauderie, A., Le Prestre, P., Palmer, M., Perrings, C., Scholes, R.J., Walpole, M., Walther, B.A., Watson, J.E., Mooney, H.A., 2010. Biodiversity targets after 2010. Curr. Opin. Environ. Sustain. 2, 3–8. doi:10.1016/j.cosust.2010.03.003 Mackenzie, D.I., Nicholas, J.D., Sutton, N., Kawanishi, K., Bailey, L.L., 2005. Improving inferences in population studies of rare species that are detected imperfectly. Ecology 86, 1101–1113. MacKenzie, D.I., Nichols, J.D., Royle, J.A., Pollock, K.H., Bailey, L.L., Hines, J.E., 2006. Occupancy estimation and modeling: inferring patterns and dynamics of species occurrence. Academic Press. Mackenzie, D.I., Royle, J.A., 2005. Designing occupancy studies: general advice and allocating survey effort. J. Appl. Ecol. 42, 1105–1114. doi:10.1111/j.1365-2664.2005.01098.x Mackenzie, D.I.I., Nichols, J.D., Lachman, G.B., Droege, S., Royle, J.A., Langtimm, C., 2002. Estimating site occupancy rates when detection probabilities are less than one. Ecology 83, 2248–2255. Magnusson, A., Punt, A.E., Hilborn, R., 2013. Measuring uncertainty in fisheries stock   232 assessment: The delta method, bootstrap, and MCMC. Fish Fish. 14, 325–342. doi:10.1111/j.1467-2979.2012.00473.x Mai, A.C.G., Rosa, I.L., 2009. Aspectos ecológicos do cavalo-marinho Hippocampus reidi no estuário Camurupim / Cardoso , Piauí , Brasil , fornecendo subsídios para a criação de uma Área de Proteção Integral Aspectos ecológicos do cavalo-marinho Hippocampus reidi no estuário Camurupim /. Biota Neotrop. 9, 85–91. doi:10.1590/S1676-06032009000300007 Mai, A.C.G., Velasco, G., Otero-Ferrer, F., Molina, L., Socorro, J., Herrera, R., Fernandez-Palacios, H., Izquierdo, M.S., 2012. Population dynamics and reproduction of wild longsnout seahorse Hippocampus reidi. J. Mar. Biol. Assoc. United Kingdom 92, 421–427. doi:10.1017/S0025315411001494 Maina, J., Cinner, J., Robinson, J., Gerry, C., Abunge, C., Thoya, P., Abernethy, K., Cedras, M., Mwaura, J., Ndegwa, S., 2011. The spatial behaviour of artisanal fishers: Implications for fisheries management and development (Fishers in Space ). Mallory, M.L., Gilchrist, H.G., Fontaine, A.J., Akearok, J.A., 2003. Local ecological knowledge of ivory gull declines in Arctic Canada. Arctic 56, 293–298. Martin, J., Runge, M.C., Nichols, J.D., Lubow, B.C., Kendall, W.L., 2009. Structured decision making as a conceptual framework to identify thresholds for conservation and management. Ecol. Appl. 19, 1079–1090. doi:10.1890/08-0255.1 Martin, T.G., Burgman, M.A., Fidler, F., Kuhnert, P.M., Low-Choy, S., Mcbride, M., Mengersen, K., 2012. Eliciting expert knowledge in conservation science. Conserv. Biol. 26, 29–38. doi:10.1111/j.1523-1739.2011.01806.x Martin, T.G., Nally, S., Burbidge, A.A., Arnall, S., Garnett, S.T., Hayward, M.W., Lumsden, L.F., Menkhorst, P., Mcdonald-madden, E., Possingham, H.P., 2012. Acting fast helps   233 avoid extinction. Conserv. Lett. 5, 274–280. doi:10.1111/j.1755-263X.2012.00239.x Maxwell, D., Jennings, S., 2005. Power of monitoring programmes to detect decline and recovery of rare and vulnerable fish. J. Appl. Ichthyol. 42, 25–37. Mazerolle, M.J., Bailey, L.L., Kendall, W.L., Andrew, J., Converse, S.J., Nichols, J.D., 2007. Making great leaps forward: accounting for detectability in herpetological field studies. J. Herpetol. 41, 672–689. McCauley, D.J., Pinsky, M.L., Palumbi, S.R., Estes, J.A., Joyce, F.H., Warner, R.R., 2015. Marine defaunation: animal loss in the global ocean. Science (80-. ). 347, 1–7. doi:10.1126/science.1255641 McClanahan, L., Ferretti, F., Baum, J.K., 2012. From archives to conservation: Why historical data are needed to set baselines for marine animals and ecosystems. Conserv. Lett. 5, 349–359. doi:10.1111/j.1755-263X.2012.00253.x Mcclanahan, T.R., Cinner, J.E., Maina, J., Graham, N.A.J., Daw, T.M., Stead, S.M., Wamukota, A., Brown, K., Ateweberhan, M., Venus, V., Polunin, N.V.C., 2008. Conservation action in a changing climate. Conserv. Lett. 1–24. McMurray, S.E., Blum, J.E., Pawlik, J.R., 2008. Redwood of the reef: Growth and age of the giant barrel sponge Xestospongia muta in the Florida Keys. Mar. Biol. 155, 159–171. doi:10.1007/s00227-008-1014-z McNie, E.C., 2007. Reconciling the supply of scientific information with user demands: an analysis of the problem and review of the literature. Environ. Sci. Policy 10, 17–38. doi:10.1016/j.envsci.2006.10.004 Meemeskul, Y., 1980. Optimum mesh size for the trawl fishery in the Gulf of Thailand, in: Indo-Pacific Fishery Commission. Standing Committee on Resources Research and   234 Development-2. Session, Hong Kong (Hong Kong), 3 Dec 1979. FAO. Meeuwig, J.J., Samoilys, M.A., Erediano, J., Hall, H., 2003. Fisher’s perceptions on the seahorse fishery in the central Philippines: Interactive approaches an an evaluation of results, in: Haggan, N., Brignall, C., Wood, L. (Eds.), Putting Fishers’ Knowledge to Work: Conference Proceedings August 27-30, 2003. Fisheries Centre Research Reports 11(1), Vancouver, BC, pp. 208–223. Meffe, G.K., Viederman, S., 1995. Combining science and policy in conservation biology. Wildl. Soc. Bull. 23, 327–332. Meine, C., Soule, M., Noss, R.F., 2006. “A mission-driven discipline ’’: the growth of conservation biology. Conserv. Biol. 20, 631–651. doi:10.1111/j.1523-1739.2006.00449.x Miller, C., Prideaux, M., 2013. Proactive cetacean conservation in the midst of “Data Deficiency”: Progress of the Convention on Migratory Species Cetacean Agreement in the Pacific Islands Region. J. Int. Wildl. law Policy 16, 41–56. doi:10.1080/13880292.2013.764775 Mobley, K.B., Small, C.M., Jones, A.G., 2011. The genetics and genomics of Syngnathidae: pipefishes, seahorses and seadragons. J. Fish Biol. 78, 1624–1646. doi:10.1111/j.1095-8649.2011.02967.x Moller, H., Berkes, F., 2004. Combining science and traditional ecological knowledge: monitoring populations for co-management. Ecol. Soc. 9, 2. Molloy, P.P., Evanson, M., Nellas, A.C., Rist, J.L., Marcus, J.E., Koldewey, H.J., Vincent, A.C.J., 2013. How much sampling does it take to detect trends in coral-reef habitat using photoquadrat surveys? Aquat. Conserv. Mar. Freshw. Ecosyst. 1–18. doi:10.1002/aqc.2372 Monk, J., 2013. How long should we ignore imperfect detection of species in the marine   235 environment when modelling their distribution? Fish Fish. 1–7. doi:10.1111/faf.12039 Moore, J.E., Wallace, B.P., Lewison, R.L., Cox, T.M., Crowder, L.B., 2009. A review of marine mammal , sea turtle and seabird bycatch in USA fisheries and the role of policy in shaping management. Mar. Policy 33, 435–451. doi:10.1016/j.marpol.2008.09.003 Morgan, A.C., Burgess, G.H., 2005. Fishery-dependent sampling: total catch, effort and catch composition, in: Management Techniques for Elasmobranch Fisheries. p. 182. Mundy-Taylor, V., Crook, V., Foster, S., Fowler, S., Sant, G., Rice, J., 2014. CITES Non-detriment Findings guidance for shark species: a framework to assist Authorities in making Non-detriment Findings ( NDFs ) for species listed in CITES Appendix II. Myers, R.A., Barrowman, N.J., Hutchings, J.A., 1997. Inshore exploitation of Newfoundland Atlantic cod (Gadus morhua) since 1948 as estimated from mark-recapture data. Can. J. Fish. Aquat. Sci. 54, 224–235. Myers, R.A., Bowen, K.G., Barrowman, N.J., 1999. Maximum reproductive rate of fish at low population sizes. Can. J. Fish. Aquat. Sci. 56, 2404–2419. doi:10.1139/f99-201 Myers, R.A., Hutchings, J.A., Barrowman, N.J., 2007. Why do fish stocks collapse? The example of cod in Atlantic Canada. Ecol. Appl. 7, 91–106. Myers, R.A., Rosenberg, A., Mace, P.M., Barrowman, N., Restrepo, V.R., 1994. In search of thresholds for recruitment overfishing. ICES J. Mar. Sci. 51, 191–205. Nakagawa, S., Cuthill, I.C., 2007. Effect size, confidence interval and statistical significance: a practical guide for biologists. Biol. Rev. 82, 591–605. doi:10.1111/j.1469-185X.2007.00027.x Newton, D., Letsela, T., Lijane, T., Mafatle, N., Manyama, P., Naha, S., Ntloko, B., Ntsohi, R., Paetzold, B., Pires, A., Polaki, M., Raimondo, D., Rouget, M., T’Sele, T., Wistebaar, N.,   236 Zimudzi, C., 2008. Development of a non-detriment finding process for Pelargonium sidoides in Lesotho (No. 2), Working Group. Mexico. Nguyen, H.P., Larsen, R.B., 2013. Effect of codend mesh size increases on the size selectivity of commercial species in a small mesh bottom trawl fishery. J. Appl. Ichthyol. 29, 762–768. Nicholson, E., Lindenmayer, D.B., Frank, K., Possingham, H.P., 2013. Testing the focal species approach to making conservation decisions for species persistence. Divers. Distrib. 19, 530–540. doi:10.1111/ddi.12066 Nijman, V., 2010. An overview of international wildlife trade from Southeast Asia. Biodivers. Conserv. 19, 1101–1114. doi:10.1007/s10531-009-9758-4 NOAA, 2014. Indian Ocean-South-East Asian (IOSEA) Marine Turtle Memorandum of Understanding. O’Donnell, K.P., Pajaro, M.. G., Vincent, A.C.J., Aswani, S., 2010a. Improving conservation and fishery assessments with local knowledge: future directions. Anim. Conserv. 13, 526–533. doi:10.1111/j.1469-1795.2010.00377.x O’Donnell, K.P., Pajaro, M.G., Vincent, A.C.J., Aswani, S., 2010b. How does the accuracy of fisher knowledge affect seahorse conservation status? Anim. Conserv. 13, 526–533. doi:10.1111/j.1469-1795.2010.00377.x Pajaro, M.G., Mulrennan, M.E., Vincent, A.C.J., 2010. Toward an integrated marine protected areas policy: connecting the global to the local. Environ. Dev. Sustain. 12, 945–965. Panayotou, T., Jetanavanich, S., 1987. The economics and management of Thai Marine Fisheries. ICLARM Stud. Rev. 14, 82. Panithanarak, T., Karuwancharoen, R., Na-nakorn, U., Thu, T.T., Nguyen, T., 2010a. Population genetics of spotted seahorses (Hippocampus kuda) in Thai waters. Zool. Stud. 49, 564–576.   237 Panithanarak, T., Karuwancharoen, R., Na-nakorn, U., Thu, T.T., Nguyen, T., 2010b. Population genetics of spotted seahorses (Hippocampus kuda) in Thai waters. Zool. Stud. 49, 564–576. Panjarat, S., Bennett, N., 2012. Responses of fishers to a 25-year seasonal closed measure on the Andaman Coast of Thailand. University of Victoria. Parsons, G.R., Foster, D.G., 2015. Reducing bycatch in the United States Gulf of Mexico shrimp trawl fishery with an emphasis on red snapper bycatch reduction. Fish. Res. 167, 210–215. doi:10.1016/j.fishres.2015.02.009 Patrick, W.S., Spencer, P., Ormseth, O., Cope, J., Field, J., Kobayashi, D., 2009. Use of productivity and susceptibility indices to determine stock vulnerability, with example applications to six U.S. fisheries, NOAA Tech. Memo. doi:NMFS-F/SPO-101 Pauly, D., 1988. Some definitions of overfishing relevant to coastal zone management in Southeast Asia. Trop. Coast. Area Manag. 3, 14–15. Pauly, D., 1986. Living resources problems of tropical inshore fisheries: fishery research on tropical soft-bottom communities and the evolution of its conceptual base. Ocean Yearb. 29–37. Peckham, S.H., Diaz, D.M., Walli, A., Ruiz, G., Crowder, L.B., Nichols, W.J., 2007. Small-scale fisheries bycatch jeopardizes endangered Pacific loggerhead turtles. PLoS One 2. Pelc, R.A., Max, L.M., Norden, W., Roberts, S., Silverstein, R., Wilding, S.R., 2015. Further action on bycatch could boost United States fisheries performance. Mar. Policy 56, 56–60. doi:10.1016/j.marpol.2015.02.002 Pellet, J., Schmidt, B.R., 2005. Monitoring distributions using call surveys: estimating site occupancy, detection probabilities and inferring absence. Biol. Conserv. 123, 27–35. doi:10.1016/j.biocon.2004.10.005   238 Perante, N.C., Pajaro, M.G., Meeuwig, J.J., Vincent, A.C.J., 2002. Biology of a seahorse species, Hippocampus comes in the central Philippines. J. Fish Biol. 60, 821–837. doi:10.1006/jfbi.2002.1878 Perry, A.L., Lunn, K.E., Vincent, A.C.J., 2010. Fisheries, large-scale trade, and conservation of seahorses in Malaysia and Thailand. Aquat. Conserv. Mar. Freshw. Ecosyst. 20, 464–475. doi:10.1002/aqc.1112 Petracca, L.S., Ramirez-Bravo, O.E., Hernandez-Santin, L., 2013. Occupancy estimation of jaguar Panthera onca to assess the value of east-central Mexico as a jaguar corridor. Oryx 48, 1–8. doi:10.1017/S0030605313000069 Phasuk, B., 1996. Fishing effort regulation in the coastal fisheries of Thailand. Thai Fish. Gaz. Pitcher, T.J., Lam, M.E., Ainsworth, C., Martindale, A., Nakamura, K., Perry, R.I., Ward, T., 2013. Improvements to Rapfish: a rapid evaluation technique for fisheries integrating ecological and human dimensions. J. Fish Biol. 83, 865–89. doi:10.1111/jfb.12122 Pomeroy, R.S., 2012. Managing overcapacity in small-scale fisheries in Southeast Asia. Mar. Policy 36, 520–527. doi:10.1016/j.marpol.2011.10.002 Punt, A., Smith, D.C., Smith, A.D.M., 2011. Among-stock comparisons for improving stock assessments of data-poor stocks: The “Robin Hood” approach. ICES J. Mar. Sci. 68, 972–981. doi:10.1093/icesjms/fsr039 Punt, A.E., Hilborn, R., 1997. Fisheries stock assessment and decision analysis: the Bayesian approach. Rev. Fish Biol. Fish. 7, 35–63. Rago, P.J., Sosebee, K.A., 2010. Biological reference points for spiny dogfish. US Department of Commerce, National Oceanic and Atmospheric Administration, National Marine Fisheries Service, Northeast Fisheries Science Center.   239 Rands, M.R.., Adams, W.M., Bennun, L., Butchart, S.H.M., Clements, A., Coomes, D., Entwistle, A., Hodge, I., Kapos, V., Scharlemann, J.P.W., Sutherland, W.J., Vira, B., 2010. Biodiversity conservation: challenges beyond 2010. Science (80-. ). 329, 1298–1303. doi:10.1126/science.1189138 Regan, T.J., Taylor, B.L., Thompson, G.G., Cochrane, J.F., Ralls, K., Runge, M.C., Merrick, R., 2013. Testing decision rules for categorizing species’ extinction risk to help develop quantitative listing criteria for the U.S. Endangered Species Act. Conserv. Biol. 27, 821–31. doi:10.1111/cobi.12055 Reid, K., 2007. Monitoring and management in the Antarctic – making the link between science and policy. Antarct. Sci. 19, 267–270. Reijnen, B., van der Meij, S.E.T.T., van Ofwegen, L.P., 2011. Fish, fans and hydroids: host species of pygmy seahorses. Zookeys 103, 1–26. doi:10.3897/zookeys.103.953 Reuter, R.F., Conners, M.E., Dicosimo, J., Gaichas, S., Ormseth, O., Tenbrink, T.T., 2010. Managing non-target, data-poor species using catch limits: Lessons from the Alaskan groundfish fishery, Fisheries Management and Ecology. doi:10.1111/j.1365-2400.2009.00726.x Rivalan, P., Delmas, V., Angulo, E., Bull, L.S., Hall, R.J., Courchamp, F., 2007. Can bans stimulate wildlife trade? Nature 447, 529–530. Robinson, J.G., 2006. Conservation biology and real-world conservation. Conserv. Biol. 20, 658–669. doi:10.1111/j.1523-1739.2006.00469.x Robley, A., Gormley, A.M., Forsyth, D.M., Triggs, B., 2014. Long-term and large-scale control of the introduced red fox increases native mammal occupancy in Australian forests. Biol. Conserv. 180, 262–269. doi:10.1016/j.biocon.2014.10.017   240 Rodrigues, A.S.L., Pilgrim, J.D., Lamoreux, J.F., Hoffmann, M., Brooks, T.M., 2006. The value of the IUCN Red List for conservation. Trends Ecol. Evol. 21, 71–76. doi:10.1016/j.tree.2005.10.010 Rosa, I.L., Alves, R.N., Bonifácio, K.M., Mourão, J.S., Osório, F.M., Oliveira, T.P.R., Nottingham, M.C., 2005. Fishers’ knowledge and seahorse conservation in Brazil. J. Ethnobiol. Ethnomed. 1, 1–12. doi:10.1186/1746-4269-1-12 Rosa, I.L., Oliveira, T.P.R., Castro, A.L.C., De Souza Moraes, L.E., Xavier, J.H.A., Nottingham, M.C., Dias, T.L.P., Bruto-costa, L. V., Araújo, M.E., Birolo, A.B., Mai, A.C.G., Monteiro-neto, C., 2007. Population characteristics, space use and habitat associations of the seahorse Hippocampus reidi (Teleostei: Syngnathidae). Neotrop. Ichthyol. 5, 405–414. Rosser, A., 2008. The CITES-IUCN checklist as an example of a method for making NDFS, and the principles that were deemed important, CITES NDF international Expert Workshop. Mexico. Rosser, A.R., Haywood, M.J., 2002. Guidance for CITES Scientific Authorities: checklist to assist in making non-detriment findings for Appendix II exports. Gland. Rowe, S., Hutchings, J.A., 2003. Mating systems and the conservation of commercially exploited marine fish. Trends Ecol. Evol. 18, 567–572. doi:10.1016/j.tree.2003.09.004 Royle, J.A., Nichols, J.D., 2003. Estimating abundance from repeated presence-absence data or point counts. Ecology 84, 777–790. Salafsky, N., Margoluis, R., Redford, K.H., Robinson, J.G., 2002. Improving the practice of conservation: a conceptual framework and research agenda for conservation science. Conserv. Biol. 16, 1469–1479. Salomon, A.K., Gaichas, S.K., Jensen, O.P., Agostini, V.N., Sloan, N.A., Rice, J., Mcclanahan,   241 T.R., Ruckelshaus, M.H., Levin, P.S., Dulvy, N.K., Babcock, E.A., 2011. Bridging the divide between fisheries and marine conservation science. Bull. Mar. Sci. 87, 251–274. doi:10.5343/bms.2010.1089 Schroeter, S.C., Gutiérrez, N.L., Robinson, M., Hilborn, R., Halmay, P., 2009. Moving from data poor to data rich: a case study of community-based data collection for the San Diego red sea urchin fishery. Mar. Coast. Fish. 1, 230–243. doi:10.1577/C08-037.1 Scotts, D.J., Craig, S.A., 1988. Improved hair-sampling tube for the detection of rare mammals. Wildl. Res. 15, 469–472. Shaffer, M.L., 1981. Minimum population sizes for species conservation. Bioscience 31, 131–134. Shepherd, C.R., Nijman, V., 2008. The trade in bear parts from Myanmar: an illustration of the ineffectiveness of enforcement of international wildlife trade regulations. Biodivers. Conserv. 17, 35–42. doi:10.1007/s10531-007-9228-9 Shepperson, J., Murray, L.G., Cook, S., Whiteley, H., Kaiser, M.J., 2014. Methodological considerations when using local knowledge to infer spatial patterns of resource exploitation in an Irish Sea fishery. Biol. Conserv. 180, 214–223. doi:10.1016/j.biocon.2014.10.013 Shester, G., Micheli, F., 2011. Conservation challenges for small-scale fisheries: bycatch and habitat impacts of traps and gillnets. Biol. Conserv. 144, 1673–1681. doi:10.1016/j.biocon.2011.02.023 Short, F.T., Polidoro, B., Livingstone, S.R., Carpenter, K.E., Sidik, J., Calumpong, H.P., Carruthers, T.J.B., Coles, R.G., 2011. Extinction risk assessment of the world ’ s seagrass species 144. Simberloff, D., 1998. Flagships, umbrellas and keystones: is single species management passe in   242 the landscape era? Biol. Conserv. 83, 247–257. Siriprasertchok, A., 2015. Illegal, unreported and unregulated fishery: an urgent problem of Thailand. Int. J. Public Priv. Manag. 2, 35–42. doi:10.1017/CBO9781107415324.004 Smith, D., Punt, A., Dowling, N., Smith, A., Tuck, G., Knuckey, I., 2009. Reconciling approaches to the assessment and management of data-poor species and fisheries with Australia’s harvest strategy policy. Mar. Coast. Fish. 1, 244–254. doi:10.1577/C08-041.1 Smith, M.J., Benítez-Díaz, H., Clemente-muñoz, M., Donaldson, J., Hutton, J.M., Mcgough, H.N., Medellin, R.A., Morgan, D.H.W., Criodain, C.O., Oldfield, T.E.E., Schippmann, U., Williams, R.J., McGough, H.N., O’Criodain, C., 2011. Assessing the impacts of international trade on CITES-listed species: current practices and opportunities for scientific research. Biol. Conserv. 144, 82–91. doi:10.1016/j.biocon.2010.10.018 Songrak, A., Bodhisuwan, W., Thapanand-chaidee, T., 2013. Selectivity of traps for blue swimming crab in Trang province. Maejo Int. J. Sci. Technol. 7, 36–42. Soulé, M.E. 1985. What is conservation biology? A new synthetic discipline addresses the dynamics and problems of perturbed species, communities, and ecosystems. BioScience. 35(11), 727-734. Sparre, P., Venema, S.C., 1998. Introduction to tropical fish stock assessment-Part 1: Manual. Fao. Stake, R.E., Denzin, N.K., Lincoln, Y.S., 2005. The Sage handbook of qualitative research, The SAGE handbook of qualitative research. Sage Publications ^ eCalifornia. Stanley, D.R., Wilson, C.A., 1990. A fishery-dependent based study of fish species composition and associated catch rates around oil and gas structures off Louisiana. Fish. Bull. 88. Steele, J., Alverson, D.L.E.E., Auster, P., Collie, J., Deegan, L., Coleman, J., Doyle, E., Lee, C.,   243 Michel, J., 2002. Effects of Trawling and Dredging on Seafloor Habitats. Stelzenm, V., Lee, J., South, A., Foden, J.J., Rogers, S.I., 2012. Practical tools to support marine spatial planning: a review and some prototype tools. Mar. Policy 1–14. doi:10.1016/j.marpol.2012.05.038 Stobutzki, I., Miller, M., Brewer, D., 2001. Sustainability of fishery bycatch: a process for assessing highly diverse and numerous bycatch. Environ. Conserv. 28, 167–181. Stuart, S.N., Collen, B., 2013. Conserving Biodiversity in a Target‐Driven World, in: Biodiversity Monitoring and Conservation: Bridging the Gap between Global Commitment and Local Action. Wiley Online Library, pp. 421–438. Tanzil, J.T.I., Brown, B.E., Tudhope, A.W., Dunne, R.P., 2009. Decline in skeletal growth of the coral Porites lutea from the Andaman Sea, South Thailand between 1984 and 2005. Coral Reefs 28, 519–528. doi:10.1007/s00338-008-0457-5 Tear, T.H., Kareiva, P., Angermeier, P.L., Czech, B., Kautz, R., Landon, L., Mehlman, D., Ruckelshaus, M., Scott, J.M., Wilhere, G., 2005. How much is enough? The recurrent problem of setting measurable objectives in conservation. Bioscience 55, 835–849. Teh, L., Zeller, D., Pauly, D., 2015. Preliminary reconstruction of Thailand’s fisheries catches: 1950-2010, Fisheries Centre Working Paper. Vancouver, BC. Teske, P.R., Lockyear, J.. F., Hecht, T., Kaiser, H., 2007. Does the endangered Knysna seahorse, Hippocampus capensis, have a preference for aquatic vegetation type, cover or height? African Zool. 42, 23–30. Thampanya, U., Vermaat, J.E., Sinsakul, S., Panapitukkul, N., 2006. Coastal erosion and mangrove progradation of Southern Thailand. Estuar. Coast. Shelf Sci. 68, 75–85. doi:10.1016/j.ecss.2006.01.011   244 Thongra-ara, W., Preeda, P., 2002. Total mercury concentrations in coastal areas of Thailand: a review. Sci. Asia 28, 301–312. Thornton, T.F., Scheer, A.M., 2012. Collaborative engagement of local and traditional knowledge and science in marine environments: a review. Ecol. Soc. 17, 8. doi:10.5751/es-04714-170308 Thorson, E., Wold, C., 2010. Back to basics: an analysis of the object and purpose of CITES and a blueprint for implementation. Portland, Oregon. Thrush, S.F., Dayton, P.K., 2012. Disturbance to marine benthic habitats by trawling and dredging: implications for marine biodiversity. Ecology 33, 449–473. doi:10.1146/annurev.ecolsys.33.010802.150515 Tisdell, C., Nantha, H.S., Wilson, C., 2007. Biodiversity conservation and public support for sustainable wildlife harvesting: a case study. Int. J. Biodivers. Sci. Manag. 3, 129–144. Tobias, T.N., 2010. Living proof: the essential data collection guide for indigenous use and occupancy map surveys. Ecotrust Canada and Union of British Columbia Indian Chiefs (UBCIC). Tokrisna, R., 2006. Renewal fishery resourced abundance and poverty eradication for coastal fishermen in Thailand, in: IIFET 2006 Portsmouth Proceedings. pp. 1–10. Tokrisna, R., Tokrisna, R., Boonchuwong, P., Boonchuwong, P., Janekarnkij, P., Janekarnkij, P., 1997. A Review on Fisheries and Coastal Community-Based Management Regime in Thailand. Int. Cent. Living Aquat. Resour. Manag. 1–26. Toonen, R.J., Wilhelm, T.A., Maxwell, S.M., Wagner, D., Bowen, B.W., Sheppard, C.R.C., Taei, S.M., Teroroko, T., Moffitt, R., Gaymer, C.F., Morgan, L., Lewis, N., Sheppard, A.L.S., Parks, J., Friedlander, A.M., 2013. One size does not fit all: The emerging frontier   245 in large-scale marine conservation. Mar. Pollut. Bull. 77, 7–10. doi:10.1016/j.marpolbul.2013.10.039 Truong, S.K., Ton, N.M.N., 1995. Reproduction of two species of seahorses Hippocampus histrix and H. trimaculatus in Binhthuan Waters. BÁO CÁO KhOA HỌC Ngay 27, 68. Tuck, G.N., Phillips, R.A., Small, C., Thomson, R.B., Klaer, N.L., Taylor, F., Wanless, R.M., Arrizabalaga, H., 2011. An assessment of seabird – fishery interactions in the Atlantic Ocean. ICES J. Mar. Sci. 68, 1628–1637. doi:10.1093/icesjms/fsr118 Turner, R.A., Polunin, N.V.C., Stead, S.M., 2015. Mapping inshore fisheries: Comparing observed and perceived distributions of pot fishing activity in Northumberland. Mar. Policy 51, 173–181. doi:10.1016/j.marpol.2014.08.005 Turvey, S.T., Fernández-Secades, C., Nuñez-Miño, J.M., Hart, T., Martinez, P., Brocca, J.L., Young, R.P., 2014. Is local ecological knowledge a useful conservation tool for small mammals in a Caribbean multicultural landscape? Biol. Conserv. 169, 189–197. doi:10.1016/j.biocon.2013.11.018 Turvey, S.T., Trung, C.T., Quyet, V.D., Nhu, H. V., Thoai, D. V., Tuan, V.C.A., Hoa, D.T., Kacha, K., Sysomphone, T., Wallate, S., Hai, C.T.T., Thanh, N. V., Wilkinson, N.M., 2015. Interview-based sighting histories can inform regional conservation prioritization for highly threatened cryptic species. J. Appl. Ecol. 52, 422–433. doi:10.1111/1365-2664.12382 Underwood, A.J., 1996. Detection, interpretation, prediction and management of environmental disturbances: some roles for experimental marine ecology. J. Exp. Mar. Bio. Ecol. 200, 1–27. Underwood, A.J., Chapman, M.G., 2003. Power, precaution, Type II error and sampling design in assessment of environmental impacts. J. Exp. Mar. Bio. Ecol. 296, 49–70.   246 doi:10.1016/S0022-0981(03)00304-6 UNEP-WCMC, 2012. Review of Significant Trade: species selected by the CITES Animals Committee following CoP15, CITES Project No. S-380. Cambridge. Usher, P.J., 2000. Traditional ecological knowledge in environmental assessment and management. Arctic 53, 183–193. doi:10.14430/arctic849 van Strien, A.J., Termaat, T., Groenendijk, D., Mensing, V., Kéry, M., 2010. Site-occupancy models may offer new opportunities for dragonfly monitoring based on daily species lists. Basic Appl. Ecol. 11, 495–503. doi:10.1016/j.baae.2010.05.003 Van Strien, A.J., Van Swaay, C.A.M., Termaat, T., 2013. Opportunistic citizen science data of animal species produce reliable estimates of distribution trends if analysed with occupancy models. J. Appl. Ecol. 50, 1450–1458. doi:10.1111/1365-2664.12158 Victor, D.G., Raustiala, K., Skolnikoff, E.B., 1998. The implementation and effectiveness of international environmental commitments: theory and practice. MIT press. Vidthayanon, C., 2005. Thailand Red Data: Fishes. Bangkok, Thailand. Vincent, A.C.J., Evans, K.L., Marsden, A.D., 2005. Home range behaviour of the monogomous Australian seahorse Hippocampus whitei. Enviro. Biol. Fishes 72, 1–12. Vincent, A.C.J., Foster, S.J., Koldewey, H.J., 2011. Conservation and management of seahorses and other Syngnathidae. J. Fish Biol. 78, 1681–1724. doi:10.1111/j.1095-8649.2011.03003.x Vincent, A.C.J., Sadovy de Mitcheson, Y.J., Fowler, S.L., Lieberman, S., 2013. The role of CITES in the conservation of marine fishes subject to international trade. Fish Fish. 1–30. doi:10.1111/faf.12035 Wabnitz, C., Andréfouët, S., Muller-Karger, F.E., 2010. Measuring progress toward global   247 marine conservation targets. Front. Ecol. Environ. 8, 124–129. doi:10.1890/080109 Wade, P.R., 2000. Bayesian methods in conservation biology. Conervation Biol. 14, 1308–1316. Walters, C.J., 2007. Is adaptive management helping to solve fisheries problems? Ambio 36, 304–7. Walters, C.J., Holling, C.S., 1990. Large-scale management experiments and learning by doing. Ecology 71, 2060–2068. Walters, C.J., Maguire, J.J., 1996. Lessons for stock assessment from the northern cod collapse. Rev. Fish Biol. Fish. Biol. Fish. 6, 125–137. Walters, C.J., Martell, S.J.D., 2004. Fisheries ecology and management. Princeton University Press. Walters, C.J., Martell, S.J.D., 2002. Stock assessment needs for sustainable fisheries management. Bull. Mar. Sci. 70, 629–638. Watson, R., Revenga, C., Kura, Y., 2006. Fishing gear associated with global marine catches II. Trends in trawling and dredging. Fish. Res. 79, 103–111. doi:10.1016/j.fishres.2006.01.013 Wattayakorn, G., 2012. Petroleum pollution in the Gulf of Thailand: a historical review. Coast. Mar. Sci. 35, 234–245. Weller, T.J., 2008. Using occupancy estimation to assess the effectiveness of a regional multiple-species conservation plan: Bats in the Pacific Northwest. Biol. Conserv. 141, 2279–2289. doi:10.1016/j.biocon.2008.06.018 Westgate, M., Likens, G., Lindenmayer, D., 2013. Adaptive management of biological systems: a review. Biol. Conserv. 158, 128–139. Williams, R., Thomas, L., 2009. Cost-effective abundance estimation of rare animals: Testing performance of small-boat surveys for killer whales in British Columbia. Biol. Conserv.   248 142, 1542–1547. doi:10.1016/j.biocon.2008.12.028 Willmann, R., Boonchuwong, P., Piumsombun, S., 2000. Fisheries management costs in Thai marine fisheries, IIFET 2000 Proceedings. Winemiller, K.O., 1992. Patterns of live-history diversification in North American fishes: implications for population regulation. Can. J. Fish. Aquat. Sci. 49, 2196–2218. doi:10.1139/f2012-048 Wiswedel, S., 2015. Hippocampus trimaculatus [WWW Document]. IUCN Red List Threat. Species. Witt, M.J., Godley, B.J., 2007. A step towards seascape scale conservation: using vessel monitoring systems (VMS) to map fishing activity. PLoS One 2, e1111. doi:10.1371/journal.pone.0001111 Wittemyer, G., Northrup, J.M., Blanc, J., Douglas-Hamilton, I., Omondi, P., Burnham, K.P., 2014. Illegal killing for ivory drives global decline in African elephants. Proc. Natl. Acad. Sci. U. S. A. 111, 13117–21. doi:10.1073/pnas.1403984111 Woods, C.C., Valentino, F., 2003. Frozen mysids as an alternative to live Artemia in culturing seahorses Hippocampus abdominalis. Aquac. Res. 34, 757–763. Yasue, M., Dearden, P., 2009. Methods to measure and mitigate the impacts of tourism development on tropical beach-breeding shorebirds: the Malaysian plover in Thailand. Tour. Mar. Environ. 5. Yasué, M., Nellas,  a., Vincent, A.C.J., 2012. Seahorses helped drive creation of marine protected areas, so what did these protected areas do for the seahorses? Environ. Conserv. 39, 183–193. doi:10.1017/S0376892911000622 Yates, K.L., Payo, A.P., Schoeman, D.S., 2012. International, regional and national   249 commitments meet local implementation: A case study of marine conservation in Northern Ireland. Mar. Policy 1–11. doi:10.1016/j.marpol.2012.05.030 Yeemin, T., Saenghaisuk, C., Sutthacheep, M., 2009. Conditions of coral communities in the Gulf of Thailand: a decade after the 1998 severe bleaching event. J. Coral Reef Stud. 11, 207–217. Young, K.D., Van Aarde, R.J., 2011. Science and elephant management decisions in South Africa. Biol. Conserv. 144, 876–885. doi:10.1016/j.biocon.2010.11.023 Zar, J.H., 1999. Biostatistical analysis, 4/e. Pearson Education India. Zeeberg, J., Corten, A., de Graaf, E., 2006. Bycatch and release of pelagic megafauna in industrial trawler fisheries off Northwest Africa. Fish. Res. 78, 186–195. doi:10.1016/j.fishres.2006.01.012 Zeller, D., Pauly, D., 2015. Catch reconstruction: concepts, methods, and data. Zeller, K.A., Nijhawan, S., Salom-Pérez, R., Potosme, S.H., Hines, J.E., 2011. Integrating occupancy modeling and interview data for corridor identification: A case study for jaguars in Nicaragua. Biol. Conserv. 144, 892–901. doi:10.1016/j.biocon.2010.12.003 Zipkin, E.F., Royle, J.A., Dawson, D.K., Bates, S., 2010. Multi-species occurrence models to evaluate the effects of conservation and management actions. Biol. Conserv. 143, 479–484. doi:10.1016/j.biocon.2009.11.016    250 Appendices  Appendix A   Supporting Material For Chapter 2  A.1 Total Number Of Sites Surveyed For Seahorses By Search Strategy, Thai Coast And Search Method. Due To Logistical Constraints Not All Methods Were Conducted At Each Site.    Search strategy Thai coast Sites visited Opportunistic sites Sites with timed  swims Sites with transects Relative abundance Andaman 19 0 9 10 Gulf  35 0 35 0 Comparing belt transects & timed swims  Andaman  7  5 12 7 Gulf 3 5 8 4   251 A.2 Rapid Assessment Sites On The Andaman And Gulf Coast Of Thailand. The Number Of Timed Swims And Transects Varied By Location But Total Area Covered For The Majority Of Sites Was Between 1000-2000 m2. Searches Were Conducted In Coral Reef, Seagrass, Mangrove And Sandy Bottom Habitat. Site Name Thai Coast Number of timed swims Number of transects Habitat Seahorse  (Hippocampus spp.) found # Seahorses (Hippocampus spp.) found Reported common by locals Anemone Reef Andaman  2 0 Coral reef Yes 1 Not applicable Ao Jak Gulf 1 0 Sandy bottom No 0 No Ao Khluai Gulf 1 0 Sandy bottom No 0 No Ao Lang Aurn Gulf 1 0 Sandy bottom No 0 No Ao Leuk Gulf 5 0 Sandy bottom No 0 No   252 A.3 (Con’t) Rapid Assessment Sites On The Andaman And Gulf Coasts of Thailand Site Name Thai Coast Number of timed swims Number of transects Habitat Seahorse  (Hippocampus spp.) found # Seahorses (Hippocampus spp.) found Reported common by locals Ao Mudong Andaman 0 4 Seagrass No 0 No Ao Tha Kien Gulf 1 0 Rubble Yes 1 Not applicable Ao Thong Node Gulf 1 0 Coral reef No 0 Yes Ao Yoi Gulf 1 0 Sandy bottom No 0 No Ban Bao Gulf 6 0 Sandy bottom Yes 2 Not applicable Dwarf Reef Andaman 0 5 Coral reef Yes 2 Not applicable   253 A.3 (Con’t) Rapid Assessment Sites On The Andaman And Gulf Coasts of Thailand Site Name Thai Coast Number of timed swims Number of transects Habitat Seahorse  (Hippocampus spp.) found # Seahorses (Hippocampus spp.) found Reported common by locals Hardeep Wreck Gulf 1 0 Shipwreck No 0 No Hat Khlong Chao Gulf 1 0 Sandy bottom No 0 No Hat Khlong Chao Shallow Gulf 3 0 Sandy bottom No 0 No Hin Wong Gulf 3 0 Coral reef No 0 No Kapoe Andaman 0 10 Mangrove No 0 Yes Kata Andaman 1 0 Sandy bottom Yes 1 Not applicable     254 A.3 (Con’t) Rapid Assessment Sites On The Andaman And Gulf Coasts of Thailand Site Name Thai Coast Number of timed swims Number of transects Habitat Seahorse  (Hippocampus spp.) found # Seahorses (Hippocampus spp.) found Reported common by locals Khram Wreck Gulf 1 0 Shipwreck No 0 No Khuraburi Andaman 0 20 Seagrass No 0 Yes Koh Chan Gulf 1 0 Sandy bottom Yes 1 No Koh Chan North Gulf 1 0 Sandy bottom Yes 1 No Koh Hae Andaman 1 4 Coral reef No 0 No Koh Jum Andaman 0 20 Seagrass Yes 1 Not applicable Koh Keh Gulf 1 0 Sandy bottom No 0 No Koh Khai Nai Andaman 0 4 Coral reef No 0 No    255 A.3 (Con’t) Rapid Assessment Sites On The Andaman And Gulf Coasts of Thailand Site Name Thai Coast Number of timed swims Number of transects Habitat Seahorse  (Hippocampus spp.) found # Seahorses (Hippocampus spp.) found Reported common by locals Koh Klung Bedan Gulf 1 0 Sandy bottom Yes 3 Not applicable Koh Kood Artificial Reef Gulf 1 0 Artificial reef No 0 Not applicable Koh Lak Ngam Gulf 1 0 Coral reef No 0 No Koh Lanta Andaman 0 6 Mangrove No 0 No Koh Lan Gulf 1 0 Artificial reef Yes 16 Not applicable Koh Lon Andaman 0 4 Coral reef Yes 1 Not applicable Koh Mai Ton Andaman 0 6 Coral reef No 0 No Koh Maisi Gulf 1 0 Sandy bottom No 0 No   256 A.3 (Con’t) Rapid Assessment Sites On The Andaman And Gulf Coasts of Thailand Site Name Thai Coast Number of timed swims Number of transects Habitat Seahorse  (Hippocampus spp.) found # Seahorses (Hippocampus spp.) found Reported common by locals Koh Ngam Noi Gulf 1 0 Sandy bottom No 0 No Koh Pai Gulf 1 0 Sandy bottom Yes 11 Not applicable Koh Phra Thong Andaman 2 20 Seagrass Yes 1 Not applicable Koh Raed Gulf 1 0 Sandy bottom No 0 No Koh Rin Gulf 1 0 Coral reef No 0 No Koh Sak Gulf 1 0 Artificial reef Yes 2 Not applicable   257 A.3 (Con’t) Rapid Assessment Sites On The Andaman And Gulf Coasts of Thailand Site Name Thai Coast Number of timed swims Number of transects Habitat Seahorse  (Hippocampus spp.) found # Seahorses (Hippocampus spp.) found Reported common by locals Koh Sak East Gulf 1 0 Sandy bottom Yes 1 Not applicable Koh Sed Gulf 1 0 Sandy bottom No 0 Yes Koh Tha Rai Gulf 3 0 Seagrass No 0 Yes Laem Por Mussel Farm Gulf 1 0 Sandy bottom No 0 Yes Mango Bay Gulf 3 0 Sandy bottom Yes 1 Not applicable North Rock Gulf 1 0 Coral Reef No 0 No Palian Andaman 5 0 Seagrass No 0 Yes Phrum Riang Gulf 1 0 Sandy bottom No 0 Yes Racha Yai Andaman 2 0 Coral reef No 0 No Sarasin Andaman 1 14 Seagrass No 0 Yes Shark Point Andaman 1 0 Coral reef No 0 Yes   258 A.3 (Con’t) Rapid Assessment Sites On The Andaman And Gulf Coasts of Thailand Site Name Thai Coast Number of timed swims Number of transects Habitat Seahorse  (Hippocampus spp.) found # Seahorses (Hippocampus spp.) found Reported common by locals Sikao Andaman 3 0 Mangrove Yes 3 Not applicable Tammalang, Satun Andaman 5 0 Mangrove No 0 No Tong Tom Yai Gulf 1 0 Sandy bottom No 0 Yes Twins Gulf 3 0 Sandy bottom No 0 No    259 A.3 Detection Rate (+/- 95% Confidence Intervals) Of Sites With Low And High Abundance Of Seahorses At Sandy Soft Bottom Research sites In Thailand  Fig A.3 Detection rate (+/- 95% Confidence Intervals) of sites with low and high abundance of seahorses at sandy soft bottom research sites in Thailand.   0"0.1"0.2"0.3"0.4"0.5"0.6"0.7"0.8"0.9"10"Sites"""""""""""""""""""""""""""(Low"N=8,"High"=2)" 20"Sites"""""""""""""""""""""""""""""""""""(Low"N="17,""High"N=3)"Detection(Rate(2014(Datasets(Low"Abundance"Sites"High"Abundance"Sites"  260 A.4 Estimated Number Of Survey Replicates Per Site Needed To Obtain A 90% Confidence That Zeros Represent True Absence Of Seahorses (Hippocampus spp.) At The Site.    Fig A.4 Estimated number of survey replicates per site needed to obtain a 90% confidence that zeros represent true absence of seahorses (Hippocampus spp.) at the site.   0"5"10"15"20"25"0.1" 0.2" 0.3" 0.4" 0.5" 0.6" 0.7" 0.8" 0.9"number'of'replicate'surveys'for'90%'con6idence'that'zeros'mean'seahorses'are'absent'from'a'site'Detection'Rate'  261 Appendix B  Supporting Material For Chapter 3 B.1 Defining Characteristics Of Commercial And Small-scale Fishers In Thailand (Lymer et al., 2010) Type of fishery Size of boat Landing sites Distance fished from shore Commercial  > 5 gross tons fishing ports > 3 km + offshore capacity Small-scale < 5 gross tons coastal fishing villages Within 3 km and rarely offshore   B.2 Small-scale And Commercial Fisher Interview Locations Appendix Fig. B2 Interview locations for small-scale and commercial fishers on the Andaman Coast, Thailand.   262 B.3 Sampling Effort By Province Of Fisher Villages And Commercial Ports Along The Andaman Coast  Provinces Ranong Phang-nga Phuket Krabi Trang Satun # fisher villages visited 1 7 5 2 1 3 Total # fisher villages 59 132 66 116 132 116 % fisher villages visited 1.7 5.3 7.6 1.7 0.8 2.6 # ports visited 0 2 1 1 1 2 Total # ports 1 2 1 1 1 2 % ports visited 0 100 100 100 1 100   263 B.4 Department of Fisheries Survey Grid For Research Trawls   Appendix Fig B.4 Thai Department of Fisheries research trawl locations throughout the Andaman Sea. Research trawls conducted in 22 of 44 cells.  264 B.5 Items Included In The Cost Of Generation For The Four Spatial Seahorse Datasets Dataset Salary of staff deployed in Thailand Accommodation  Food Local travel Gas Outreach materials i. Fisher interviews Yes Yes Yes Yes Yes Yes ii. Governmental research trawls Yes No Yes Yes Yes No iii. Scientific diver surveys Yes Yes Yes Yes Yes No iv. Citizen science contributions Yes Yes Yes Yes Yes Yes  B.6 Editing And Verification Procedures Of Fishers’ Maps  Editing of fishers’ maps prior to analysis Prior to analysis, we standardized the maps’ spatial accuracy because fishers’ knowledge varied in how they described the locations of seahorses and their fishing grounds. All spatial analyses were conducted in ArcGIS 10.1 (ESRI 2015). We asked fishers to provide information wherever possible about GPS coordinates, local land marks, distance from shore, and depth ranges for their identified locations. We later standardized these shapefiles based on the following protocol 1) GPS coordinates, 2) local landmarks, 3) distance from shore estimates, and 4) depth ranges based on Thai national bathymetry data. Almost all shapefiles required some editing, with the majority of edits made based on depth ranges from the Thai Royal Navy Bathymetry data, provided by the Fisheries Spatial Analyst Unit of the Thailand Department of Fisheries.   265  Verification of seahorse presence maps We verified our local knowledge maps with two methods: internal consistency and validation interviews (Tobias, 2010). To ensure internal consistency, the authors reviewed the information displayed on maps for any outliers. This method does not involve community feedback but is a useful complement to respondent review (Tobias 2010), and enabled us to ensure saturation was indeed met.   Our second verification method involved validation interviews with both previous respondents and new fishers who had not participated in the initial mapping process. Validation interviews took place in two of the six Andaman coast provinces, at a total of five locations (two commercial ports and three small-scale villages) where the original data were collected. These interviews were conducted in a semi-structured fashion, where respondents were shown the seahorse presence maps and asked to identify areas where they agreed or disagreed about where they could find seahorses. Fishers were also given the opportunity to identify locations where our maps failed to identify where seahorses lived. These locations were compared with the existing presence maps to explore consistencies in verification from interviews (Appendix Fig B.6, Appendix Table B.6).  266   Appendix Fig B.6: Verification of a) commercial and b) small-scale fishing maps based on interviews with commercial and small-scale fishers at two fishing ports and three small-scale fishing villages.     267 Appendix Table B.6: Validation reports from commercial and small-scale fishers on maps reporting seahorse presence. Responses indicate the number of locations identified by fishers as correct, incorrect, failed to identify, and unsure.   Correct Incorrect Failed to identify Unsure Total Commercial fishers 130 16 27 9 182 Small-scale fishers 43 10 19 32 103 Total 173 26 46 41 285  Literature Cited ESRI. (2011) ArcGIS Desktop: Release 10. Redlands, CA: Environmental Systems Research Institute  B.7 Evaluating Effort Discrepancies Between Trawl Fishers And DoF Research Trawls. I compared commercial trawl captains’ knowledge about seahorse locations with the DoF research trawl data. From the commercial fisher interviews, I selected those shapefiles from trawl captains and executed the same analysis as described above to create a seahorse presence map. With the DoF trawl survey data, I created point shapefiles. Next I executed a simple overlay analysis with the trawl captain seahorse maps and the DoF trawl survey data to identify the similarities and differences between the two datasets. I ran an additional analysis of trawling effort to highlight discrepancies in the datasets that were related to effort. I calculated the number   268 of hauls per DoF survey grid area in one year, the number of hauls per day with seahorses, and the number of seahorses per haul. During fisher interviews I gathered information from trawl captains on the number of hauls per day, average length of fishing trips, and estimates of the number of seahorses either per haul, day or fishing trip. Fishers were free to estimate the number of seahorses for the time period in which they were most confident. I then used the responses to estimate the number of hauls per day with seahorses and the number of seahorses per haul and per day. When a range was given, I used the mean for all calculations. To determine the total number of hauls per survey grid per year from the trawl captain responses, I summed the number of hauls per day from all respondents and multiplied it by the mean response of number of days fished per month (25), and then by the number of months per year (12). With the DoF research trawl data, I averaged the number of trawls per grid per year given the three years of data collection. I used the mean number of hauls per day and number of seahorses per haul and per day based on the three years of data.     269 Table B7 Trawl sampling effort from trawl captain interviews and DoF research trawls on the Andaman coast including number of surveyed cells, cells with seahorses, total number of days, fishers interviewed, staff needed and cost per day to generate the dataset  Number of boats surveyed cells where trawling occurred  Mean # hauls over entire grid / year # cells with seahorses out of total sampled Total # of hauls per dataset # Hauls per day Hauls / day with SH  Total # hauls with seahorses / year Mean seahorses / haul Trawl captains 45 44 55,800 44 55,800 186  36.7- 160.4 11,012 – 48,139 10.5  DoF research trawls 1 22 88  22 264 1.5 0.28  16.6 2.03   270 Appendix C  Supporting Material For Chapter 4  C.1  Percent Catch Under Length At Maturity And Sex Ratio By Seahorse Species By Fishing Gear Type.   Otter  trawl Pair trawl Purse  Seine Pushnet Gillnet Cage Total H. histrix % catch under  length at maturity 0 0 0 0 0 0 0 Sex ratio 0 0.5 0 0 0 0 0.5 H. kelloggi % catch under  length at maturity 63.0 37.0 0 0 0 0 28.3 Sex ratio 0.38 0.42 No data No data No data 0.00 0.43 H. kuda % catch under  length at maturity 0 2.0 0 0 98.0 0 66.1 Sex ratio 0.5 0.0 No data 0.0 0.58 No data 0.56 H. spinosissimus % catch under  length at 33.0 50.0 0 0 11.0 6.0 18.5   271 maturity Sex ratio 0.43 0.48 1.0 No data 0.57 0.5 0.48 H. trimaculatus % catch under  length at maturity 3.0 23.0 3.0 0 70.0 0.0 11.4 Sex ratio 0.44 0.37 0.33 1.0 0.43 No data 0.41      272 Appendix D  Supporting Material for Chapter 5 D.1 Life History Indicators For Five Thai Seahorse Species.  All Indicators Taken From Froese and Pauly 2016 Except Length At Reproductive Maturity. (Lawson et al 2015*).   Species N Max L (mm) Linf von Bertalanffy growth coefficient (k) Nat. Mort. (M) Life span   Length at reproductive maturity Hippocampus trimaculatus 261 22.0 23.2 1.01 1.83 2.9  12.1*  D.2 Simulation: Sensitivity Analysis With Fishing Mortality And Life History Variables To understand some of the uncertainty in my parameter estimates, I executed a sensitivity analysis by varying each life history parameter plus and minus 10% (representing minimal error) and 50% (large error) while keeping all other parameters constant, and maintaining fishing mortality at 0.12. I wanted to identify which parameter was most influential in determining (i) how quickly biomass dropped by ≥ 0.3  and (ii) how far the biomass actually dropped. We examined the change in the von Bertalanffy growth parameter, natural mortality, length at maturity, length vulnerable to the fishery, and the recruitment compensation parameter.       273 Appendix Table D.2: Sensitivity analysis of 10% and 50% error in life history parameters for our deterministic age structured seahorse model.   Error (%) Year from start of fishing Ratio  Error (%) Year from start of fishing Ratio VbK +10 11 0.31 +50 8 0.33 -10  10 0.33 -50 19 0.31  M +10 9  0.35  +50 7 0.53 -10  17 0.29 -50 4 0.23 Lmat +10 10 0.29 +50 4 0.23  -10 17 0.34 -50 7 0.44 Lvul +10 11 0.39 +50 6 0.69 -10 7 0.28 -50 3 0.27 recK +10 10 0.36 +50 6 0.46 -10  8 0.29 -50 4 0.22  D.3 Fisher Interviews To identify what gears and seahorse species were involved in seahorse bycatch, I conducted fisher interviews in 11 provinces on both the Andaman (n=6; May-August 2013) and Gulf (n=5;May-August 2014) coasts of Thailand (Fig D.3). I interviewed a total of 306 (137 commercial, 169 small-scale) fishers from both coasts. Since commercial and small-scale   274 fisheries have defining characteristics in Thailand (Lymer et al., 2010) I expected their annual bycatch rates of seahorses to differ. I decided to focus my efforts on four types of commercial (otter trawl, pair trawl, purse seine, pushnet) and two types of small-scale (gillnet, cage) fishing gears based on recommendations of dominant gear types from the DoF.   To select respondents, I targeted two sets of landing locations - fishing ports for commercial fishers and coastal villages for small-scale fishers. Commercial fishing port locations were identified as ports located near government port facilities, whereas landing locations in coastal villages were typically on local beaches or in estuary arms near fishers’ houses. I visited a total of 13 ports (Andaman = 7, Gulf = 6) and 28 coastal villages (Andaman = 21, Gulf = 7) (Figure 1). I determined the number of fishers to interview in each location as either 10% of the estimated total number of fishing boats landing catch at that location or the saturation method (Tobias, 2010). I defined saturation as the point if the 6th, 7th and 8th interview yielded no new information on gear type, fishing grounds, or seahorse locations, then saturation had been reached. I additionally confirmed saturation through our observations of gear type at each landing location. All interviews followed UBC Human Ethics Protocols (H12-02731).   275  Appendix Fig D.3: Locations of fisher interviews with commercial and small-scale fishers along Thailand’s coast.   276 D.4 Model Variations Variation1: I incorporated historic and current fishing effort into my model and linked it to fishing mortality (Equation 4.1-4.2, Table 1). I determined catchability by gear-type based on recent research (Chapter 4). I scaled these values relative to the gear with the highest reported catch rate (Chapter 4).  I then used the sum product of relative catch rate by vessel numbers multiplied by a scalar value of 0.0002 to calculate total fishing mortality by year. I assumed catchability remained constant through time and that changes in total catch were related to changes in fishing effort.   Variation 2: My second model variation incorporated decadal estimates for illegal, unreported and unregulated vessel catches (Teh et al 2015, Janetkitkusul 2003). To simulate IUU vessel numbers, I increased the number of fishing vessels per year according to the mean ratio of unreported to reported catch by decade based on a historical catch reconstruction of Thailand (Teh et al 2015) (Equation 5.1 and 5.2, Table 1).    Variation 3: My third model variation explores a link between habitat decline, fast recovery and weak recruitment compensation. This model variation assumes that habitat declines (e.g. sponge, soft coral) due to increases in trawling pressure (Chuenpagdee and Pauly, 2003) but is able to re-grow relatively quickly (Henry and Hart, 2005; McMurray et al., 2008) (Equations 6.1-6.3, Table 1). However, recruitment compensation is weak in the remaining habitat (Equations 6.1-6.3, Table 1).     277 Variation 4:  The fourth model variation explores a possible link between habitat decline, slow recovery, and an increase in the stock recruitment relationship. This model variation assumes that habitat (e.g. soft coral, sponge) declines and recovery is slow (Shester and Micheli, 2011; Watson et al., 2006), which leads to stronger recruitment compensation within remaining habitat (Equation 7.1, Table 1). The habitat variable is linked to the stock-recruitment carrying capacity parameter, so that only when habitat is very sparse, maximum juvenile survival rate decreases. This scenario represents the long distances for a juvenile seahorse to disperse before finding a place to settle and grow.     278 Appendix E  Supporting Material For Chapter 6 E.1 Risk Assessment For Pressures Facing Seahorse Species. Risk Low Medium High Unknown Gear diversity 0-1 of known gears catch Hippocampus spp. 2-3 of known gears catch Hippocampus spp.  4-6 of known gears catch Hippocampus spp. Unknown how many gears catch Hippocampus spp.  Fishing mortality Vulnerability to capture in fishing gears (port sampling data) (all gears in total) Less than 30% of all port sampled individuals are a given seahorse species Between 30-50% of all port sampled individuals are a given seahorse species > 50% of all port sampled individuals are a given seahorse species No data available on vulnerability to capture in fishing gears       279 E.1 Risk Assessment For Pressures Facing Seahorse Species. Risk Low Medium High Unknown Relative fishing gear pressure (comparison among gears) Less than 30% of port sampled individuals of a given species captured in named fishing gear Between 30-50% of port sampled individuals of a given species captured in named fishing gear and fleet size < 10,000 More than 50% of port sampled individuals of a given species captured in named fishing gear; or between 30-50% of port sampled individuals of a given species captured in named fishing gear and fleet size > 10,000 No data available on relative fishing gear pressure Fishing selectivity Overfishing indicator - % catch under length at maturity (Froese, 2004) Less than 30% of individuals captured under length at maturity Between 30-50% of individuals captured under length at maturity > 50% of individuals captured under length at maturity No data on % catch under length at maturity    280 E.1 Risk Assessment For Pressures Facing Seahorse Species. Risk Low Medium High Unknown Overfishing indicator – sex bias ratio (Rowe & Hutchings, 2003) No sex bias ratio or differences in sex bias ratio not statistically significant (p > 0.1) Differences in sex bias ratio but not statistically significant (0.05 < p < 0.1) Statistically significant sex bias ratio (p<0.05) No data available on sex bias      281 E.1 Risk Assessment For Pressures Facing Seahorse Species. Risk Low Medium High Unknown Discarding No gears throw any seahorses back Some gears throw seahorses back a) based on data from fisher interviews  b)Only gillnets reporting throwing seahorses back. Proportion unspecified but large total # of small-scale gillnet fishers makes this threat medium  All gears throw some proportion of seahorse catch back;   Or proportion discarded identified as > 30% Unknown if any gears throw a proportion of total catch back   Adverse fishing impacts (IUCN 2015) Fisher reported declines < 30%  Fisher reported declines between 30 - 50%  Fisher reported declines > 50% No data available    282 E.1 Risk Assessment For Pressures Facing Seahorse Species. Risk Low Medium High Unknown Uses in trade Zero to one uses in trade  Two to three uses in trade Four or more uses in trade No data available on use in trade Presence of IUU fishing / trade in country Vessel monitoring system available for most fisheries and /or observer system in place; not identified as a top issue in marine fisheries and trade National data available that IUU fishing / trade occurs; concern and awareness of this issue at the national level Internationally identified as a country of concern for IUU fishing No data available on IUU fishing or trade in country Indicators of trade < 30% change in any indicators of adverse trade impact < 50% change in indicators of adverse trade impact > 50% change in indicators of adverse trade impact No data available on changes over time for indicators of adverse trade    283 E.1 Risk Assessment For Pressures Facing Seahorse Species. Risk Low Medium High Unknown Habitat specialization Species is found in more than two habitat types Species is found in only two habitat types or species is found in one habitat type but on many different species- e.g. found in seagrass but on many species of seagrass Seahorse species found on only one species of habitat and is therefore considered a specialist; e.g. H. bargibanti lives on only 1 species of soft coral No information available about seahorse habitat Marine based activities Activities documented to show no or little damage to seahorse habitat  Activities documented show moderate damage – habitat loss can re-grow in 1 year or less and threat is not on-going At least 1 activity documented to show severe damage – including loss of habitat that takes > 1 yr. to regrow and / or threat is on-going Impact from marine activities is conflicting, unknown and/or undocumented    284 E.1 Risk Assessment For Pressures Facing Seahorse Species. Risk Low Medium High Unknown Land based activities Few land-based activities documented to occur in seahorse habitats Land based activities are known to occur in seahorse habitats, and are on-going, but their impact is reversible  Land based activities are known to occur in seahorse habitats, are on-going and impact is irreversible Impact from land based activities is conflicting, unknown and/or undocumented Climate change Climate change threats documented to have little to no impact on seahorse habitats in country Some climate change threats have been documented in seahorse habitats; but also documentation of recovery in country; or long term effects of climate change on habitats is uncertain At least 1 climate change threat documented to have severe damage in seahorse habitat, and no documentation of recovery in country Climate change impacts are conflicting, unknown and/or undocumented      285 E.1 Risk Assessment For Pressures Facing Seahorse Species. Risk Low Medium High Unknown Indicators of habitat health No observed change in any indicators of habitat health  Declines in habitat health reported for some but not all habitat types and information on what habitats are important at various life stages is unknown  Declines in habitat health reported for all habitat types or large declines in habitat health in habitats where species spends the majority of its life cycle Indicators of habitat health are conflicting, unknown and/or undocumented       286 E.2 Potential Management Responses And Their Appropriateness For Mitigating Pressures On Seahorse Populations From Fisheries And Habitat Pressures. Text section Potential management response Appropriate for targeted capture Explanation Appropriate for incidental capture (including both active and static gear types) Explanation Appropriate for additional pressure from habitat loss Explanation 5.2.1 Limited entry YES when combined  Only when used in combination with seahorse catch quotas. YES when combined Only when used in combination with seahorse catch quotas and/or spatial restrictions of gears that catch seahorses. YES when combined Only when used in combination with MPAs or spatial restrictions of gears that catch seahorses. 5.2.2 Permanent, no-take  Marine Protected Areas  (i.e. reserves) YES Where enforced these buffer against all pressures. YES Where enforced these buffer against all pressures. YES Where enforced these buffer against all pressures.    287 E.2 Potential Management Responses And Their Appropriateness For Mitigating Pressures On Seahorse Populations From Fisheries And Habitat Pressures Text section Potential management response Appropriate for targeted capture Explanation Appropriate for incidental capture (including both active and static gear types) Explanation Appropriate for additional pressure from habitat loss Explanation 5.2.3 Gear restrictions - spatial YES Where enforced these buffer against fishing pressures. YES Where enforced these buffer against fishing pressures. YES Where enforced these buffer against gear pressures on habitats. 5.2.4 Gear restrictions - temporal Cautiously Only when temporal gear restrictions coincide with peak seahorse reproduction periods. Cautiously Only when temporal gear restrictions coincide with peak seahorse reproduction periods. NO, usually Not appropriate where habitats are still subject to destructive fishing practices at other times of the year. 5.2.5 Catch quota YES Fishers targeting seahorses limit their catch volumes and so fishing mortality. Cautiously Appropriate only where fishery is completely closed once seahorse bycatch quota is met. Not applicable Output controls do not protect habitats.   288 E.2 Potential Management Responses And Their Appropriateness For Mitigating Pressures On Seahorse Populations From Fisheries And Habitat Pressures Text section Potential management response Appropriate for targeted capture Explanation Appropriate for incidental capture (including both active and static gear types) Explanation Appropriate for additional pressure from habitat loss Explanation 5.2.6 Minimum size limit YES Fishers targeting seahorses are able to be selective, taking only those larger than the agreed minimum size, and leaving smaller individuals where they are found. NO Non-selective fishing gears that catch seahorses cannot be selective for seahorse size – mesh size does not matter. Not applicable Output controls do not protect habitats. 5.2.7 Maximum size limits YES Fishers targeting seahorses are able to be selective, taking only those smaller than the agreed maximum size, and leaving larger individuals where they are found. NO Non-selective fishing gears that catch seahorses cannot be selective for seahorse size – mesh size does not matter. Not applicable Output controls do not protect habitats.    289 E.2 Potential Management Responses And Their Appropriateness For Mitigating Pressures On Seahorse Populations From Fisheries And Habitat Pressures. Text section Potential management response Appropriate for targeted capture Explanation Appropriate for incidental capture (including both active and static gear types) Explanation Appropriate for additional pressure from habitat loss Explanation 5.2.8 Slot size limits YES Fishers targeting seahorses are able to be selective, taking only those that fall between the agreed minimum and maximum size limits, leaving other individuals where they are found. NO Non-selective fishing gears that catch seahorses cannot be selective for seahorse size – mesh size does not matter. Not applicable Output controls do not protect habitats. 5.2.9 Leaving pregnant males YES Fishers targeting seahorses are able to be selective, leaving pregnant males where they are found. NO Non-selective fishing gears that catch seahorses cannot be selective for seahorse reproductive state. Not applicable Output controls do not protect habitats.   290 E.2 Potential Management Responses And Their Appropriateness For Mitigating Pressures On Seahorse Populations From Fisheries And Habitat Pressures Text section Potential management response Appropriate for targeted capture Explanation Appropriate for incidental capture (including both active and static gear types) Explanation Appropriate for additional pressure from habitat loss Explanation 5.2.10 Export quota NO, usually Only where there is a direct feedback loop that generates a catch reduction of seahorses. NO, usually Only where there is a direct feedback loop that generates a catch reduction of seahorses. Not applicable Output controls do not protect habitats. 5.2.11 Reintroduction/ supplementation Not if threat is ongoing There is no evidence that seahorse releases can increase densities of wild seahorse populations. Not if threat is ongoing There is no evidence that seahorse releases can increase densities of wild seahorse populations. Not applicable -       291 E.2 Potential Management Responses And Their Appropriateness For Mitigating Pressures On Seahorse Populations From Fisheries And Habitat Pressures Text section Potential management response Appropriate for targeted capture Explanation Appropriate for incidental capture (including both active and static gear types) Explanation Appropriate for additional pressure from habitat loss Explanation 5.2.12 Habitat restoration YES when combined Only when combined with Permanent, no-take Marine Protected Areas. No sense increasing seahorse habitats if they are going to be targeted by fishers. YES when combined Only when combined with Permanent, no-take Marine Protected Areas. No sense increasing seahorse habitats if they are going to be fished. Cautiously Not if threat that caused habitat decline is ongoing   292 E.3 Identifying Spatial Overlap Of Management Measures And Hippocampus Spp. Observations To identify overlap of existing management measures and observation of seahorses I executed a simple geospatial analysis.  I used three data sources for observations of seahorses by species, DoF research trawls, scientific surveys and citizen science contributions, and three existing management measure datasets- Marine Protected Areas (MPAs), No Trawl Zones and Seasonal Closures. All datasets were collected in the WGS 1984 datum. The marine management datasets were provided by the Thailand Department of Fisheries. The method used to generate each of the seahorse observation datasets is described below:   Department of Fisheries research trawls Trawl surveys were executed by the Thailand Department of Fisheries in 2010, 2012 and 2013 at 22 pre-determined sampling locations throughout the Andaman Sea, originally intended for the purpose of sampling commercially exploited fish species. Each location represented a grid equivalent to 15 x 15 nautical miles (Appendix B.4). The grids were sampled four times per year using a 23.5m otter-board trawl research vessel, with trawl speed set at 2.5 nautical miles / hour. Trawling took place within each gridded area for one hour.  GPS locations were marked at the beginning and ending of each trawl, and the depth of the trawl recorded. All fish from each trawl were sorted and identified by Department of Fisheries staff. Presence or absence of seahorses was noted for each trawl, and seahorses were identified to species level.       293 Scientific diving surveys I used data from Aylesworth et al. 2015 (Chapter 2) to identify locations of seahorses in the Andaman Sea. These data were collected through underwater scientific diving surveys over two three-month field seasons. A total of 26 sites were visited over the two field seasons in coral, seagrass, mangrove and sandy soft bottom habitats (Aylesworth et al. 2015) (Chapter 2). These data included GPS coordinates for each site, and when seahorses were observed, the species, depth, and habitat were recorded (Aylesworth et al. 2015) (Chapter 2). This research was conducted in accordance with UBC Animal Ethics protocols (A12-0288).   Citizen science diver contributions I selected citizen science contributions of seahorse sightings from the global seahorse database, iSeahorse.org. All data available through April 23, 2015 were included for analysis, excluding any data contributed by the authors. The iSeahorse program encourages divers to report sightings of seahorses to the database with the option to include species names, habitat, depth, and any additional information, e.g. behavior. A concerted effort was made during the my time in Thailand to promote the iSeahorse program within Thailand’s diving community through presentations, training workshops, and appeals to divers (English and Thai) on social media. I had no control over who contributed sightings and from which locations, but we initially sourced all sightings spatially located in Thai waters.  Analysis: First I selected only those seahorse observations identified as H. kuda and H. trimaculatus from each of the seahorse observation datasets for inclusion in this analysis (Table E.1). I ran a simple   294 overlay analysis with each of the management shapefiles to identify what % of seahorse sightings was located inside management areas for each species (Main text Table 764). All analyses were run in ArcGIS 10.1 using the WGS 1984 datum.  Appendix Table E.1: Summary of seahorse occurrence datasets by species  DoF research trawls Scientific surveys Citizen science contributions Total H. kuda 0 10 28 38 H.trimaculatus 549 2 5 556 Total 549 12 33 594      295  Appendix F  Fisher Interview Questionnaire Seahorse Fishers: Small Scale & Commercial Interviews  Respondent ID # _______________________ Interviewer: ________________________ Small Or Commercial:  __________________       Date of Interview (mm/dd/yy) ____/____/13  General Information Age: ___________         Year Born: ____________                                               Sex: Male    Female    Is fishing your primary job?   Yes     No   How many years have you been fishing? __________________ Do you own your own boat?  Yes    No Does the boat have a motor? Yes    No If yes, what size is your boat? _______ How many people work on your boat? _______________________________ At what depth do you fish? How deep can your gear go? What kind of fishing gear do you use? Gear(s):_________________________________________________________________________________________________________    296 How does your fishing gear work? _______________________________________________ ____________________________________________________________________________ How long does it take for you to set your net?______________________________________ How many times do you set your net in a night / trip?_______________________________ (If they work on trawler / drag / set nets) How long do you drag / set the nets for at a time? ___________________________________________________________________________ Do you fish in the wet season and dry season?  Yes          No Use the same type of gear year round?     Yes                             No If no, what gear do you use in dry season? ________________________________________________________________________   Wet season? ______________________________________________________________________ Are there any other times when you use a specific type of gear? ______________________________________________________________________ How many months of the year do you fish? _____________________________  How many days / month do you fish?__________________________________ How long is an average fishing trip for you? ___________________________________ What time do you usually go out to sea? _________________________ Do#you#only#fish#on#the#Andaman#Coast?_____________________________________#Where#do#you#keep#your#boat?#____________________________________________#Can#you#show#me#on#this#map?#________________________________________#Where#do#you#fish?#__________________________________________________#Do#you#fish#here#all#the#time?##_____________________________________________#How#many#days#per#month#do#you#go#here?#____________________________________#  297 #What#types#of#gear#do#you#use#here?#____________________________________________#How#deep#is#it#here#______________________________________#Why#do#you#fish#here?#_________________________________________________________#_________________________________________________________________________________#What#is#the#habitat#where#you#fish?###Coral#reef## # ##Seagrass## # ##Mangrove### # ##Muddy#bottom#### ## ##Rocky#area##########Other##______________________________!Do#other#people#fish#in#this#area?#####Yes###################################################No#What#type#of#gear#do#they#use#?#________________________________________#Does#anyone#trawl#in#this#area?####Yes#######################################################No###Are#there#crab#fishers#in#this#area?##Yes###################################################No#Do#people#use#gill#nets#in#this#area?#Yes###################################################No#Do#people#use#pushnets#in#this#area?#Yes#################################################No#Do#people#use#any#other#gear#in#this#area?##___________________________________#Is#this#a#protected#area?#___________________________________________________#How#long#does#it#take#to#travel#to#and#from#this#fishing#grounds?#______________________#Where#else#do#you#fish?#Do#you#ever#catch#seahorses?######Yes###################################################No##What#type#(s)#of#gear#do#you#catch#them#in?#__________________________________#Is#there#any#season#when#you#catch#more#/#find#more#seahorses?#________________________##For#an#average#(*day#or#haul)#in#wet#season,#how#many#seahorses#do#you#catch?#___________#For#an#average#(*week#/#trip*)#in##wet#season,#how#many#seahorses#do#you#catch?#__________#For#an#average#(*day#or#haul)#in#dry#season,#how#many#seahorses#do#you#catch?#___________#For#an#avegage#(*week#/#trip)#in#dry#season#how#many#seahorses#do#you#catch?____________#For#an#average#month#how#many#seahorses#do#you#catch?#_________________________#For#an#average#year?#________________________________##How#many#types#of#seahorses#do#you#catch?#_____________________________##Are#they:#####Smooth## # # # Spiny#  298 What#size#are#most#of#the#seahorses#you#catch?#_______________________#What#is#the#size#of#the#largest#seahorse#you#catch?#_____________________#The#smallest?#__________________#Does#the#amount#of#seahorses#you#catch#change#in#the#wet#season#compared#to#the#dry?#______________________________________________________________________________#In#what#habitat#(s)#do#you#find#seahorses?#______________________________#What#depth(s)#do#you#find#seahorses?#______________________________#What#do#you#do#with#the#seahorses#you#catch?#__________________________#!(if!they!throw!them!back)….Are#they#alive#or#dead#when#you#throw#them#back?#___________________________________________________________________________#Do#you#throw#them#back#in#the#same#general#area#where#they#were#caught?#______________#Does#everyone#throw#them#back,#or#do#you#know#anyone#who#does#something#different?#________________________________________________________________________#Is#there#anything#you#can#use#seahorses#for?#_____________________________________#Compared#with#when#you#first#started#fishing,#is#there########a)#any#difference#in#the#number#of#seahorses#you#catch#(high#v.#low#season)?_________########b)#any#difference#in#types#of#seahorses#you#catch?#__________________________________########c)#any#difference#in#average#size#of#seahorses#you#catch?#______________________________########d)#any#difference#in#price#of#seahorses?#____________________________________________########e)#any#difference#in#number#of#buyers?#____________________________________________###Why#do#you#think#the#change#has#occurred?#_____________________________________________________________________________#For!those!that!sell!seahorses:!!How#many#years#have#you#been#selling#seahorses?#________________________________#How#did#you#get#started?#________________________________________________##Who#do#you#sell#them#to?#(don’t#need#names)#_________________________________#How#many#people##(buy)#are#interested#in#buying#seahorses?#________________________#What#other#things#are#these#people#buying?#______________________________________#  299 Do#they#all#buy#the#same#amount#of#seahorses?#Or#are#some#bigger#buyers#than#others?#__________________________________________________________________________#How#much#do#buyers#pay#for#seahorses?#________________________________________#Does#the#price#vary#on#colour,#size,#smooth#/#spiny?#______________________________#Are#there#any#seahorses#buyers#won’t#take?#______________________________________#Do#you#sell#them#wet#or#dried?#___________________________________________##How#does#the#value#of#seahorses#compare#to#other#marine#products?#____________________#Do#you#know#what#the#buyers#do#with#the#seahorses?#_________________________________#How#long#have#you#been#fishing#with#the#same#type#of#gear#and#boat#in#the#same#place?#_______________________________________________________________________#GENERAL!Questions:!!Is#there#any#time#of#year#/#season#when#the#seahorses#breed?#__________________________#Are#there#any#other#types#of#boats#or#gear#that#catch#seahorses?#_________________________#How#does#the#quantity#of#seahorses#you#catch#compare#with:## a)#other#boats#(fishers)#?#_____________________________________________________## b)#other#gears?#_____________________________________________________________#Will#you#catch#the#same#amount#of#seahorses#5#yrs#from#now#that#you#catch#today?#Why?#Why#not?#________________________________________________________________________________#Do#you#think#your#fishery#in#this#area#is#sustainable?#If#yes#why?#If#not,#why#not?#_______________________________________________________________________________________________________________________________________________________________________________________________________________________________________#What#can#be#done#to#improve#the#sustainability#of#fishing#where#you#fish?##______________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________# #  300 Appendix G  Acknowledgements Continued Thank you to the staff at the Phang-nga (K’Amnat), Phuket (K’Tassapon), Ranong (K’Montri), and Satun (K’Amnuay) Provincial Marine Fisheries Stations and K’Pholphisin at the Fisheries Spatial Analysis Unit. Thank you to the Phuket Marine Biological Center, the Phuket Aquarium and the Department of Marine and Coastal Resources. I am grateful for the support of Kasetsart University (Dr. Yo), the Prince of Songkla University – Phuket (Dr. Chantinee) and Songkla (Dr. James) campuses - and the Thai conservation dive group Save Our Seas. A special thank you to Dr. Parichart Laksanawimol for her generosity, patience, and willingness to collaborate.  Thank you to my research assistants, for having a sense of humor on our various quests for seahorses both in fishing ports and underwater. Thank you to PADI, the Phuket Gazette, and Phuket Today for helping me raise awareness about seahorses in Thailand. I am grateful for the support from numerous dive shops, fishers and community groups who facilitated my research for seahorses. A special thanks for the enduring support from Dive Tribe, KiwiDivers.Com, Laytrang Divers, and the New Heaven Reef Conservation program. Thank you to Sampan, Kung, Nangy and Pi Aey for their support and friendship during my time in Thailand. I am also thankful for the support of Jennie Dean, Tim Foley and Kim Riddick for helping me through the challenges of my PhD.   

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0308076/manifest

Comment

Related Items