@prefix vivo: . @prefix edm: . @prefix ns0: . @prefix dcterms: . @prefix skos: . vivo:departmentOrSchool "Science, Faculty of"@en, "Zoology, Department of"@en ; edm:dataProvider "DSpace"@en ; ns0:degreeCampus "UBCV"@en ; dcterms:creator "Haggarty, Dana Rochelle"@en ; dcterms:issued "2015-11-24T02:06:17"@en, "2015"@en ; vivo:relatedDegree "Doctor of Philosophy - PhD"@en ; ns0:degreeGrantor "University of British Columbia"@en ; dcterms:description "Networks of Marine Protected Areas are implemented to conserve fish populations, yet their effectiveness is rarely comprehensively examined or adaptively managed. In this dissertation, I evaluate a network of Rockfish Conservation Areas (RCAs) implemented to reverse population declines of inshore Pacific rockfishes (Sebastes spp.). First, I used SCUBA surveys to examine patterns of Black Rockfish abundance compared to spatial and temporal variability in recruitment to determine how recruitment influences population density in and around a RCA. Habitat variables such as complexity and rocky substrate predicted adult Black Rockfish abundance while recruitment did not. Next, I surveyed the fish communities of 35 RCAs and adjacent unprotected areas using a Remotely Operated Vehicle (ROV). Habitat features such as percent rocky substrates and depth influenced the density of Quillback (S. maliger), Yelloweye (S. ruberrimus), Greenstriped Rockfishes (S. elongatus), Kelp Greenling (Hexagrammos decagrammus), Lingcod (Ophiodon elongatus) and all inshore rockfishes combined, while reserve status did not. The results give little indication that demersal fish populations have recovered inside the RCA system. I used aerial observations of recreational fishing from surveys before, during and after 77 RCAs were established and found there was no evidence of a change in fishing effort in 83% of the RCAs. Compliance was related to the level of fishing effort around the RCA, the size and perimeter-to-area ratio of RCAs, proximity to fishing lodges and the level of enforcement. Non-compliance in RCAs may be hampering their effectiveness and impeding rockfish recovery. Lastly, I modeled rocky reef habitat using Random Forest Classification to assess habitat in the RCAs. I combined three habitat metrics with data on compliance, RCA size, rockfish bycatch, and connectivity into a single Conservation Score. The Conservation Score is related to the log reserve ratio, a measure of relative abundance, for Quillback Rockfish. RCAs with low Conservation Scores are not likely to be effective and managers should evaluate the reasons for low scores and address reserve shortcomings in an adaptive spatial management framework. Education and enforcement efforts are critical to the recovery of depleted fish stocks. Continued monitoring and evaluation of the RCAs is essential to rockfish conservation."@en ; edm:aggregatedCHO "https://circle.library.ubc.ca/rest/handle/2429/55510?expand=metadata"@en ; skos:note " AN EVALUATION OF THE EFFECTIVENESS OF ROCKFISH CONSERVATION AREAS IN BRITISH COLUMBIA, CANADA by Dana Rochelle Haggarty B.Sc., The University of Victoria, 1997 M.Sc., The University of British Columbia, 2002 A DISSERTATION 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) November 2015 © Dana Rochelle Haggarty, 2015ii Abstract Networks of Marine Protected Areas are implemented to conserve fish populations, yet their effectiveness is rarely comprehensively examined or adaptively managed. In this dissertation, I evaluate a network of Rockfish Conservation Areas (RCAs) implemented to reverse population declines of inshore Pacific rockfishes (Sebastes spp.). First, I used SCUBA surveys to examine patterns of Black Rockfish abundance compared to spatial and temporal variability in recruitment to determine how recruitment influences population density in and around a RCA. Habitat variables such as complexity and rocky substrate predicted adult Black Rockfish abundance while recruitment did not. Next, I surveyed the fish communities of 35 RCAs and adjacent unprotected areas using a Remotely Operated Vehicle (ROV). Habitat features such as percent rocky substrates and depth influenced the density of Quillback (S. maliger), Yelloweye (S. ruberrimus), Greenstriped Rockfishes (S. elongatus), Kelp Greenling (Hexagrammos decagrammus), Lingcod (Ophiodon elongatus) and all inshore rockfishes combined, while reserve status did not. The results give little indication that demersal fish populations have recovered inside the RCA system. I used aerial observations of recreational fishing from surveys before, during and after 77 RCAs were established and found there was no evidence of a change in fishing effort in 83% of the RCAs. Compliance was related to the level of fishing effort around the RCA, the size and perimeter-to-area ratio of RCAs, proximity to fishing lodges and the level of enforcement. Non-compliance in RCAs may be hampering their effectiveness and impeding rockfish recovery. iii Lastly, I modeled rocky reef habitat using Random Forest Classification to assess habitat in the RCAs. I combined three habitat metrics with data on compliance, RCA size, rockfish bycatch, and connectivity into a single Conservation Score. The Conservation Score is related to the log reserve ratio, a measure of relative abundance, for Quillback Rockfish. RCAs with low Conservation Scores are not likely to be effective and managers should evaluate the reasons for low scores and address reserve shortcomings in an adaptive spatial management framework. Education and enforcement efforts are critical to the recovery of depleted fish stocks. Continued monitoring and evaluation of the RCAs is essential to rockfish conservation. iv Preface Chapter 2 is based on research I undertook at the Bamfield Marine Science Centre. I did rockfish surveys while SCUBA diving close to sites that had previously been sampled for fish recruitment by Russel Markel and Katie Lotterhos for their respective PhD’s. Russ and Katie provided their recruitment data to me for use in this chapter. I analyzed the data and wrote the manuscript. The research described in Chapter 3 was part of a large collaborative project between UBC and Fisheries and Oceans Canada (DFO) that was supported by an NSERC Strategic Grant to J. Shurin and E. Taylor and by DFO. I collaborated with Lynne Yamanaka at the Pacific Biological Station to monitor rockfish populations inside and adjacent to RCAs using a Remotely Operated Vehicle (ROV). I planned the surveys and conducted much of the field work. I was the Chief Scientist on 4 of 5 research cruises; Lynne was Chief Scientist on the cruise that took place before I commenced my PhD. We also took the opportunity to collect some additional data on 4 RCAs at the end of two surveys lead by Lynne and Shannon Obradovich (PhD student, UBC Fisheries). Ship time was paid for by DFO Science and funds from the NSERC grant and DFO were used to hire consultants to watch video and to map transects. I analyzed the ROV data and wrote the manuscript. DFO provided financial support during the manuscript preparation. The recreational fishery data I used in Chapter 4 were collected by DFO (South Coast Management) during their creel survey overflights that are used in the management of the v recreational fishery. Raw data were provided to me by Dave O’Brien, Patrick Zeterberg and Andrew Pereboom that were then digitized in a Geographic Information System (GIS) by myself and three assistants. Brenda Wright provided data for the SW Vancouver Island that were already digitized. I analyzed all of the data and wrote the manuscript. A version of this paper has been accepted by the Canadian Journal of Fisheries and Aquatic Sciences. I am the lead author on the paper; Steve Martel provided funding for one of the GIS assistants, suggested the temporal analysis and provided statistical advice; Jon Shurin is the supervisory author who provided statistical advice and manuscript edits. I also collaborated with Lynne Yamanaka of DFO on the habitat model I describe in Chapter 5. Lynne hired me as a biologist at the Pacific Biological Station (DFO) and supported the habitat modeling work. A GIS practicum student from Vancouver Island University, Mesha Sagram, helped us prepare some of the GIS layers I used in the habitat model. The fine resolution bathymetry and backscatter data used in the model were shared with us by the Canadian Hydrographic Service. The intermediate resolution bathymetry datasets were created by and provided to us by Ed Gregr (Scitech Consulting) and Sarah Davies (DFO). I did all of the data analysis and wrote the manuscript. vi Table of Contents Abstract ..................................................................................................................................... ii Preface ....................................................................................................................................... iv Table of Contents ..................................................................................................................... vi List of Tables ............................................................................................................................ xi List of Figures ........................................................................................................................ xiii List of Abbreviations ............................................................................................................. xvi Acknowledgements .............................................................................................................. xvii Dedication ............................................................................................................................... xix Chapter 1: Introduction ........................................................................................................... 1 1.1 Effects of Overfishing on Marine Ecosystems ........................................................... 1 1.2 Marine Protected Areas............................................................................................... 2 1.3 Rockfish Biology ........................................................................................................ 6 1.4 Inshore Rockfish Fishery and Management in BC ................................................... 13 1.5 Rockfish Conservation Areas ................................................................................... 15 1.6 Overview of Dissertation .......................................................................................... 22 1.6.1 Does Post-settlement Recruitment Predict the Adult Abundance of Black Rockfish? .......................................................................................................................... 22 1.6.2 Assessing Population Recovery Inside British Columbia’s Rockfish Conservation Areas with a Remotely Operated Vehicle. ................................................. 22 1.6.3 Lack of Recreational Fishing Compliance May Compromise Effectiveness of Rockfish Conservation Areas in British Columbia. .......................................................... 23 vii 1.6.4 How Do They Score? An Evaluation of Rockfish Conservation Areas Using a Conservation Score that Combines Rockfish Habitat and Key Reserve Features. ........ 24 Chapter 2: Does Post-settlement Recruitment Predict the Adult Abundance of Black Rockfish? ................................................................................................................................. 25 2.1 Introduction ............................................................................................................... 25 2.2 Methods..................................................................................................................... 28 2.2.1 Study System ........................................................................................................ 28 2.2.2 Rockfish Recruitment ........................................................................................... 31 2.2.3 Black Rockfish Dive Survey ................................................................................. 31 2.2.4 Black Rockfish Density and Habitat Analysis ...................................................... 33 2.2.5 Black Rockfish Density and Recruitment Analysis .............................................. 34 2.2.6 Rockfish Conservation Area Evaluation ............................................................... 35 2.3 Results ....................................................................................................................... 35 2.3.1 Black Rockfish Density and Habitat ..................................................................... 36 2.3.2 Black Rockfish Density and Recruitment ............................................................. 36 2.3.3 Fish Density in the Broken Group RCA ............................................................... 36 2.4 Discussion ................................................................................................................. 43 Chapter 3: Assessing Population Recovery Inside British Columbia’s Rockfish Conservation Areas with a Remotely Operated Vehicle ..................................................... 48 3.1 Introduction ............................................................................................................... 48 3.2 Methods..................................................................................................................... 53 3.2.1 Study Area ............................................................................................................ 53 3.2.2 Data Collection ..................................................................................................... 54 viii 3.2.3 Analysis: ............................................................................................................... 57 3.2.3.1 Fish Density .................................................................................................. 57 3.2.3.2 Effects of Protection and Habitat by Transect .............................................. 59 3.2.3.3 Effects of Catch, Age, Area and Shape on RCA Effectiveness .................... 60 3.2.3.4 Length of Fish Inside to Outside of RCAs.................................................... 62 3.3 Results ....................................................................................................................... 62 3.3.1 Fish Density .......................................................................................................... 62 3.3.2 Effects of Protection and Habitat .......................................................................... 63 3.3.3 Effects of Catch, Age, Area and Shape on RCA Effectiveness ............................ 65 3.4 Discussion ................................................................................................................. 80 Chapter 4: Lack of Recreational Fishing Compliance May Compromise Effectiveness of Rockfish Conservation Areas in British Columbia.............................................................. 88 4.1 Introduction ............................................................................................................... 88 4.2 Methods..................................................................................................................... 90 4.2.1 Study Area Context ............................................................................................... 90 4.2.2 Data Collection and Preparation ........................................................................... 91 4.2.3 Temporal Analysis ................................................................................................ 93 4.2.4 Factors Affecting RCA Compliance ..................................................................... 97 4.2.5 Rockfish Catch in RCAs ....................................................................................... 98 4.3 Results ....................................................................................................................... 99 4.3.1 Data Compilation .................................................................................................. 99 4.3.2 Temporal Analysis ................................................................................................ 99 4.3.3 Factors Affecting RCA Compliance ................................................................... 107 ix 4.3.4 Rockfish Catch in RCAs ..................................................................................... 111 4.4 Discussion ............................................................................................................... 111 Chapter 5: How Do They Score? An Evaluation of Rockfish Conservation Areas Using a Conservation Score that Combines Rockfish Habitat and Key Reserve Features. ........ 119 5.1 Introduction ............................................................................................................. 119 5.2 Methods................................................................................................................... 124 5.2.1 Data Sources ....................................................................................................... 124 5.2.2 Random Forest Habitat Model ............................................................................ 126 5.2.3 Conservation Score ............................................................................................. 129 5.2.3.1 Rockfish Conservation Areas Size.............................................................. 130 5.2.3.2 Habitat Scores ............................................................................................. 130 5.2.3.2.1 Habitat Area .......................................................................................... 130 5.2.3.2.2 Habitat Percent ...................................................................................... 130 5.2.3.2.3 Habitat Isolation .................................................................................... 131 5.2.3.3 Bycatch Score ............................................................................................. 131 5.2.3.4 Compliance ................................................................................................. 132 5.2.3.5 Connectivity ................................................................................................ 132 5.2.3.6 Overall Conservation Score ........................................................................ 133 5.3 Results ..................................................................................................................... 138 5.3.1 Random Forest Habitat Model ............................................................................ 138 5.3.2 Conservation Score ............................................................................................. 141 5.4 Discussion ............................................................................................................... 148 5.4.1 Random Forest Habitat Model ............................................................................ 148 x 5.4.2 Conservation Score ............................................................................................. 151 Chapter 6: Conclusion .......................................................................................................... 156 6.1 Synopsis .................................................................................................................. 156 6.1.1 Habitat in RCAs .................................................................................................. 157 6.1.2 Compliance with RCA Regulations .................................................................... 158 6.1.3 Conservation Value of RCAs for Marine Biodiversity ....................................... 159 6.2 Limitations of Research .......................................................................................... 161 6.3 Recommendations ................................................................................................... 162 References .............................................................................................................................. 165 Appendices ............................................................................................................................. 180 Appendix A Supplementary Information from Chapter 2. Does Post-settlement Recruitment Predict the Adult Abundance of Black Rockfish? ......................................... 180 Appendix B Supplementary Information from Chapter 3. Assessing Population Recovery Inside British Columbia’s Rockfish Conservation Areas with a Remotely Operated Vehicle ................................................................................................................ 182 Appendix C Supplementary Information for Chapter 4. Lack of Recreational Fishing Compliance May Compromise Effectiveness of Rockfish Conservation Areas in British Columbia. ................................................................................................................ 197 Appendix D Supplementary Information from Chapter 5. How Do They Score? An Evaluation of Rockfish Conservation Areas Using a Conservation Score that Combines Rockfish Habitat and Key Reserve Features. ..................................................................... 205 Appendix E Species Quantified on ROV Surveys in RCAs. .............................................. 215 xi List of Tables Table 1-1. Life history traits of inshore and common shallow-shelf rockfish species found in Rockfish Conservation Areas in BC (Richards 1986, Love et al. 2002). BC specific figures are given when possible. ................................................................................................... 12 Table 1-2. Fisheries prohibited and permitted in RCAs by sector. ............................................... 18 Table 1-3. Total area and proportion of modeled rockfish habitat protected by the RCAs (Yamanaka and Logan 2010). ....................................................................................................... 19 Table 2-1. Habitat Variable Descriptions. .................................................................................... 32 Table 2-2. Mean and Standard Deviation of Habitat variables (%) by year and Region. ............ 37 Table 2-3. Model results of three Linear Mixed Effects Models with Year/Region/Site as nested random effects.. ................................................................................................................. 41 Table 3-1. The number of transects, mean and Standard Deviation (SD) of Fish Densities (#/100m2) inside and outside of RCAs observed on ROV surveys by region. ............................. 67 Table 3-2. Analysis of Variance Table for fish Density by Region and RCA.............................. 68 Table 3-3. Linear Mixed Effects Model results of averaged model (with shrinkage) of habitat and RCA status on fish density (fish/100m2) on ROV transects (n=298). ....................... 69 Table 3-4. The proportion depth classes sampled on 420 ROV transects and the expected and observed proportion of fish species and 95% Confidence Intervals by depth class. Numbers in bold indicate the preferred depth zones of each species. A Chi-squared analysis indicated that species are not distributed evenly across depth classes. ........................... 73 Table 3-5. Linear Mixed Effects model results of factors affecting Response Ratio with Region as a random variable (n=38). ............................................................................................ 77 xii Table 4-1. Number of images per region and year, positional accuracy (Root Mean Squared (RMS) error) and number of fishing events observed. ................................................. 100 Table 4-2. Results of the preferred linear mixed effects model (AIC=1379) using 462 observations and 77 groups (RCAs) and fit with REML. ........................................................... 101 Table 4-3. Summary of changes in monthly recreational effort with time by RCA. .................. 105 Table 4-4. Explanatory variables that were retained and rejected in the best model .................. 108 Table 5-1. Bathymetric and backscatter (BS) variable descriptions for the MBES and FS models. ........................................................................................................................................ 134 Table 5-2. Model performance statistics for the 20m models by region and the 5m full models (with backscatter (BS) and 5m models without BS at the selected cut-offs. ................. 136 Table 5-3. Comparison of area of habitat modeled at two resolutions in five test areas by model cut-off. 5m models were made with and without backscatter data. ................................. 137 Table 5-4. Total modeled area (km2) and % rocky habitat in RCAs by Pacific Fishery Management Areas (PFMA). Area and % estimates from the old habitat model (Yamanaka and Logan 2010) are shown for comparison. .......................................................... 140 Table 5-5. Summary of RCA feature values and scores, conservation score and key feature score. The age of the RCA was not used in the conservation score. .............................. 143 xiii List of Figures Figure 1-1. Images of the inshore rockfish species in BC. ............................................................. 7 Figure 1-2. Rockfish Conservation Areas in northern British Columbia, shown by green shading, are typically large. .......................................................................................................... 20 Figure 1-3. Rockfish Conservation Areas in southern BC, shown by green shading. One hundred and twenty five of 164 of the RCAs are found in “inside” waters between Vancouver Island and the mainland and are typically smaller than RCAs in “outside” waters. ........................................................................................................................................... 21 Figure 2-1. Sampling sites in Barkley Sound, BC. ....................................................................... 30 Figure 2-2. Black Rockfish Density by % complexity and rocky substrates by region. .............. 38 Figure 2-3. Mean density of Black Rockfish by site in 2010 and 2011 versus mean YOY recruitment in 2006. ...................................................................................................................... 39 Figure 2-4. Mean density of Black Rockfish by site in 2010 and 2011 versus mean YOY recruitment in 2007, 2008 and 2009 and 1-year-old recruits in 2009. .......................................... 40 Figure 2-5. Mean Response Ratio and 95% Confidence Interval of reef fish species in the Broken Group Islands RCA as compared to all sites in Barkley Sound outside of the RCA (n=2). ............................................................................................................................................. 42 Figure 3-1. Rockfish Conservation Areas in southern BC that were sampled using an ROV between 2009 and 2011 are shown in green. ....................................................................... 58 Figure 3-2. The % Relative Variable Importance (RVI), estimate and 95% Confidence Intervals of habitat and protection variables retained in the Linear Mixed Effects Averaged Model............................................................................................................................ 71 xiv Figure 3-3. The mean and SE of the percent of habitat type by region on transects inside and outside of the RCAs. .............................................................................................................. 72 Figure 3-4. The mean log response ratio of Quillback, Yelloweye and Lingcod of habitat-based density inside to outside of all RCAs sampled. Error bars are standard errors. Ratios greater than zero indicate greater densities inside the RCA. ............................................. 74 Figure 3-5. Log Response Ratio (RR) for targeted and non-targeted fish species in RCAs. ....... 75 Figure 3-6. Boxplots of the log Response Ratio (RR) by Species/Species Group and by the years of protection at the time of sampling. ............................................................................ 76 Figure 3-7. Log Response Ratio (RR) versus log reserve area (km2), perimeter-to-area ratio, and the Log+1 Catch (kg) per km2 in a 5 km wide buffer around each RCA. .................... 78 Figure 3-8. Histograms of the length (cm) of fish inside and outside of the RCAs and results of the Kolmogorov-Smirnoff test. ..................................................................................... 79 Figure 4-1. Study area of southern British Columbia showing the Rockfish Conservation Areas (RCAs) on and off of the aerial survey flight path and Pacific Fishery Management Areas (11- 29) (PFMA). ............................................................................................................... 96 Figure 4-2. Boxplot of standardized log +1 fishing effort density inside and outside of RCAs in 2003, 2007, and 2011. ................................................................................................. 102 Figure 4-3. Proportion of recreational effort in RCAs in the Strait of Georgia by management area (PFMA) by year. ............................................................................................ 103 Figure 4-4. Difference in kernel density effort maps between 2007 and 2003 (top) and 2011 and 2007 (bottom). ............................................................................................................. 106 Figure 4-5. Fishing effort probability around Vancouver Island along RCAs on the flight path and 2km-wide buffers, as well as the location of fishing lodges and cities xv (population >5,000). Insets show RCAs in the southern and northern extents of Vancouver Island. ...................................................................................................................... 109 Figure 4-6. Five significant variables were retained in the RCA Compliance Generalized Additive Model. Plots show the smooths of the significant variables. Shaded regions are 2 standard error confidence bands for smooths and the points are partial residuals................... 110 Figure 5-1. Location of RCAs in southern BC the extent of the 20m bathymetry for the Strait of Georgia (SoG), the Queen Charlotte-Johnstone Straits (QCS) and the West Coast of Vancouver Island (WCVI) as well as the extent of the MBES bathymetry data used. The five test areas used for comparisons with multibeam and backscatter models (5x5) are outlined in orange and indicated by letters: A) Goletas Channel, B) Barkley Sound, C) Georgia Strait, D) Gulf Islands, E) Captain Passage. ................................................................. 135 Figure 5-2. Example of four RCAs showing modeled habitat and habitat feature scores (1-3, bad to good). ....................................................................................................................... 144 Figure 5-3 (A–D). Scores (1=bad, 2=medium, 3=good) of features (A) Size, B) Habitat Area, C) % Habitat, D) Habitat Isolation. See Appendix Table 4-1 and 4-2 for values and scores of each RCA. .................................................................................................................... 145 Figure 5-3 (E–H). Scores (1=bad, 2=medium, 3=good) of features (E) Recreational Compliance, F) Bycatch, G) Connectivity and H) overall Conservation Score of RCAs. See Appendix Table 4-1 and 4-2 for values and scores of each RCA. ....................................... 145 Figure 5-4. Response Ratio of Quillback Rockfish, Yelloweye Rockfish and combined inshore rockfish group plotted against the Conservation Score and Key Element Count. ......... 147 xvi List of Abbreviations AUC Area Under the ROC Curve BPI Bathymetric Positioning Index CART Classification and Regression Trees CHS Canadian Hydrographic Service CPUE Catch Per Unit of Effort DFO Fisheries and Oceans Canada GAM Generalized Additive Model GLM General Linear Model LME Linear Mixed Effects Model MBES Multibeam Echosounder MPA Marine Protected Area OOB Out of Bag RCA Rockfish Conservation Area RF Random Forest Classification ROC Receiver Operator Graphs ROV Remotely Operated Vehicle RR log Response Ratio SCUBA Self Contained Underwater Breathing Apparatus YOY Young of the Year Fish Species abbreviations used in some figures: GS Greenstriped Rockfish IRF Inshore Rockfish KG Kelp Greenling LC Lingcod QB Quillback Rockfish YE Yelloweye Rockfish Regions BC British Columbia JS Johnstone Strait QCSt Queen Charlotte Strait SoG Strait of Georgia WCVI West Coast of Vancouver Island xvii Acknowledgements I am grateful to many people who have helped me throughout the course of my PhD. Thank you to my supervisor, Jon Shurin, for being so accessible and supportive despite the geographic distance between us. Mary O’Connor thankfully stepped in to be my co-supervisor and provided good advice and a supportive lab environment. My other committee members, Eric Taylor, Jackie King and Steve Martell provided valuable insight over the years. Although she was not a member of my academic committee, my mentor, Lynne Yamanaka at the Pacific Biological Station, was essential to the success of this work. Thank you for all of your support! This work was funded by a Strategic NSERC grant to J. Shurin, E. Taylor and L. Yamanaka. I was funded by NSERC, a UBC Star Fellowship, a Four-Year-Fellowship (Faculty of Grad Studies), and a West Canadian Universities Marine Science Centre Student Scholarship. I’m grateful to a number of people who assisted me to collect and prepare data. On ROV surveys: the crews of the Neocaligus and the Vector, Rob Flemming, Karina Cooke, Jen Paton, Jen Tyler, Jonathon Martin, Andy Clark, Josh Chernov, Wolfgang Carolsfeld, James Pegg, Stefan Dick and Janine Beckett. SCUBA surveys: Sarah Frioult, Guilia Bernardi, Katie Lotterhos, Jocelyn Nelson, Siobhan Gray, Rebecca Martone and Russel xviii Markel. GIS assistance: Andrea Orantes-Duran, Heloise Chenelot, Elena Chin, and Mesha Sagram. Thank you all for your conscientious work! Over the years, several friends and colleagues have helped me greatly along the way. Ed Gregr, Rob Flemming, Sarah Davies, and Jerry Maedel all helped me as I struggled to become competent at GIS. Tammy Norgard, Russ Markel, Rebecca Martone, Katie Lotterhos, Jenn Yakimishyn, Sarah Dudas, Janine Beckett, Dave Robichaud, Lisa Lacko, Joanne Lessard, Dom Bureau, Shannon Obradovich, Matt Siegle, Michael Folks, Guy and Jodi Le Masurier and Christy and Brice Semmens all provided great support and thoughtful advice. Thank you, my friends! I could not have completed this without the support of my family. Thank you to my parents, Lynn and Dave, my mother-in-law, Sophie, and most of all, my husband, Omi. I love you all. xix Dedication To Eddy 1 Chapter 1: Introduction 1.1 Effects of Overfishing on Marine Ecosystems Fishing is a major human impact to marine ecosystems (Norse 1993, Botsford et al. 1997, Pauly et al. 1998, Jackson 2001, Lotze et al. 2006). Fishing reduces the abundance of the target species and can erode the age and size structure of fish populations (Pauly et al. 2002). The loss of older and larger fish in a population leads to greater variation in recruitment success and increased vulnerability to environmental effects (Pauly et al. 2002, Anderson et al. 2008). There is a growing awareness that the high fecundity of most marine fishes does not make them immune to the threat of extinction and some formerly abundant commercial fish species have been considered for protection under endangered species legislation (Hutchings 2001, Pauly et al. 2002, Dulvy et al. 2003, Dulvy et al. 2004, Hutchings and Reynolds 2004). There has also recently been acceptance that \"fisheries products cannot be extracted from the sea without ecosystem effects\" such as reductions of non-target species through bycatch, changes in predator-prey dynamics that cascade through ecosystems, and the alteration or destruction of habitats by fishing practices (National Research Council 2006). These fishery impacts occur simultaneously with large-scale oceanographic events such as El Niño, decadal-scale oscillation and long-term climate change and the resilience to such effects can be hampered by the effects of fishing, making trends in fish stocks less predictable and more difficult to manage (Pauly et al. 2002, Hsieh et al. 2010, Thrush and Dayton 2010). A combination of major reductions in fishing capacity, reductions in catch, and the use of areas closed to fishing, or Marine Protected Areas (MPAs), is necessary to rebuild fisheries and reverse declines in marine biodiversity (Pauly et al. 2002, Worm et al. 2009). Although MPAs may not be a panacea for marine conservation (Allison et al. 1998), they are considered to be an 2 effective way to conserve rockfish populations (Carr and Reed 1993, Yoklavich 1998, Parker et al. 2000, Hamilton et al. 2010). Many populations of Pacific rockfishes ( Sebastes spp.), have been overfished along the west coast of the United States and British Columbia (BC) (Parker et al. 2000, Love et al. 2002, Williams et al. 2010, Yamanaka and Logan 2010). Marine Protected Areas have been established to protect rockfish and other marine life in California, Oregon and Washington State, and a system of reserves, termed Rockfish Conservation Areas (RCAs), has been established in BC (Yamanaka and Logan 2010). Determining if reserves are an effective means of rebuilding rockfish populations is essential to adaptive management and the conservation of rockfishes (Babcock et al. 2010). 1.2 Marine Protected Areas Marine Protected Areas are defined as \"any area of inter-tidal or sub-tidal terrain, together with its overlying water and associated flora, fauna, historical, or cultural features, which has been reserved by law or other effective means to protect part or all of the enclosed environment\" (Kelleher and Kenchington 1992); while the term marine reserve is usually used for complete \"no-take\" reserves (Council 2001, Agardy et al. 2003). Harvest refugia have been defined as \"a location of restricted harvesting of targeted species for the purpose of replenishing exploited populations through larval recruitment\" (Carr and Reed 1993). Marine Protected Areas are a major component of Ecosystem-Based Management (Halpern et al. 2010, Thrush and Dayton 2010). Marine Protected Areas have been shown to be a successful strategy to increase the size, abundance and diversity of species protected within and sometimes adjacent to them (Allison et al. 1998, Mosqueira et al. 2000, Halpern and Warner 2002, Halpern 2003, Alcala et al. 2005, Claudet et al. 2008, Lester et al. 2009, Babcock et al. 2010). Interconnected networks of reserves 3 may protect species with dispersing larvae as well as promote spill-over effects that support fisheries (Gaines et al. 2010a, Gaines et al. 2010b). Although MPAs can be effective tools for conservation, improperly placed reserves or poorly designed networks of MPAs can lead to a false sense of security and could detract from other forms of management (Allison et al. 1998, Carr and Raimondi 1999, Crowder et al. 2000). The potential economic and ecological implications of MPAs have led to the development of a theory of reserve design (Roberts 2000, Airamé et al. 2003, Gaines et al. 2003, Hastings and Botsford 2003, Leslie et al. 2003, Roberts et al. 2003a, Roberts et al. 2003b, Gaines et al. 2010a, Gaines et al. 2010b). Although the design of an MPA or network of MPAs varies according to the goals of the reserves (i.e. protection of biodiversity versus fishery management) (Carr and Reed 1993, Hastings and Botsford 2003, Gaines et al. 2010a) and the species that are targeted, the design criteria include the following considerations: location, shape, size, spacing, and the proportion of area protected (Yoklavich 1998, Gaines et al. 2003, Roberts et al. 2003a, Roberts et al. 2003b, Gaines et al. 2010a). The most important criteria in situating an MPA are the quality and diversity of habitats protected because a diversity of habitats are required to support different species and ontogenetic stages (Yoklavich 1998, Carr and Raimondi 1999, Crowder et al. 2000, Roberts et al. 2003a, Parnell et al. 2006). Oceanographic factors such as currents and upwelling regimes should also be considered as reserve selection criteria in addition to benthic habitats (Gaines et al. 2003). The placement also depends on the goal of a reserve because boundaries placed within continuous habitat will allow spill-over of adults, benefiting fisheries, while boundaries placed beyond the edge of habitats should retain adults (Chapman and Kramer 2000, Gaines et al. 2010a). The shape of the reserve also affects adult and larval spill-over because greater edges or 4 a smaller area to perimeter ratio increases reserve \"leakiness\" (Yoklavich 1998) and also provides greater opportunity for fishers to exploit the edge of a reserve by \"fishing the line\" (Kellner et al. 2007). The size of reserves is another important criterion. Although positive effects have been shown for reserves of all sizes (Claudet et al. 2008), larger reserves are usually recommended because small reserves may not be large enough to accommodate viable or persistent populations (Crowder et al. 2000, Halpern 2003), larger reserves are required for species with large home ranges (Yoklavich 1998, Carr and Raimondi 1999, Gaines et al. 2010a), and larger reserves may be less susceptible to environmental or human disturbance and can be easier to enforce (Roberts et al. 2003a, Kritzer 2004). The shape of the reserve also affects its enforceability because law enforcement officers must be able to clearly identify and prove if people are fishing within a reserve (Yamanaka and Logan 2010). Networks of MPAs rather than single reserves are recommended because most marine populations are open due to pelagic larval dispersal stages. Benefits from multiple reserves may be synergistic so networks are thought to outperform single reserves (Murray et al. 1999, Botsford et al. 2001, Botsford et al. 2003, Palumbi 2003, Roberts et al. 2003b, National Research Council 2006, Gaines et al. 2010a). The spacing of the reserves in a network is therefore a key factor determining connectivity among reserves in a network (Gaines et al. 2003, Palumbi 2003, Botsford et al. 2009) and should match the larval dispersal distance of the target species (Botsford et al. 2001, Gaines et al. 2003, Largier 2003, Gaines et al. 2010a). The total proportion of marine habitat to be placed in MPAs is another subject of debate. Modeling studies of this question have found that reserve benefits will be maximized when 20-50% of habitat is protected (Roberts et al. 2003b, National Research Council 2006). More than 1600 scientists and conservationists called for the protection of 20% of marine waters to be 5 protected by 2020 (MCBI 1998). Canada signed the Convention on Biological Diversity which includes a commitment of protected a minimum of 10% of coastal and marine areas in MPAs by 2020 (Target 11, Aichi Biodiversity Targets) (Secretatiat of the Convention on Biological Diversity 2011). Some authors caution applying blanket targets to conservation because some rare or vulnerable habitats may require more protection than others, and some abundant habitats, such as deep muddy bottoms, probably do not require as much area in a protected area as would be achieved with a blanket target (Agardy et al. 2003, Carwardine et al. 2009). Reserve siting algorithms, designed to identify networks of reserves with specific proportions of habitats spread over geographic areas have been developed (Sala et al. 2002, Airamé et al. 2003, Leslie et al. 2003, Leslie 2005). The Rockfish Conservation Areas (RCAs) were designated in BC by Fisheries and Oceans Canada as harvest refuges for inshore rockfishes and are not MPAs (Robb et al. 2011). Robb et al. (2010) argued that the RCAs do not meet the IUCN definition of an MPA because they lack permanency because they were not designated through an MPA legislative tool such as Canada’s Ocean Act, but rather through a fishery closure using the Fisheries Act. Rockfish Conservation Areas also have no legislative authority over non-fishing activities. Others, however, consider RCAs to be ‘de-facto’ MPAs (Marliave and Challenger 2009, Lotterhos et al. 2014). RCAs are also not complete \"no-take\" reserves, but restrict fisheries that target or lead to substantial bycatch of inshore rockfishes (Yamanaka and Logan 2010). The RCA’s objective of rebuilding rockfish populations in RCAs by limiting directed or undirected catch is, however, consistent with the idea of rockfish harvest refugia (Carr and Reed 1993, Yoklavich 1998, Yamanaka and Logan 2010). Therefore, the principles of MPA design and evaluation can also be applied to RCAs. Evaluating how the RCAs contribute to rebuilding rockfish populations is critical to 6 rockfish management plans and recovery strategies and is necessary to adaptively manage these species. It is also important to understand their effectiveness in order to assess how they can contribute to a system of MPAs in BC. 1.3 Rockfish Biology Rockfishes of the genus Sebastes, are a diverse group with at least 65 species in the Northeast Pacific between Alaska and Baja California. Thirty-six species are found in BC (Love et al. 2002). Inshore rockfish populations (for which the RCAs were designed) in BC are grouped together for management purposes and are defined as species that aggregate over rocky areas in nearshore waters between about 0-200 meters depth and include Quillback (S. maliger), Yelloweye (S. rubberimus), Copper (S. caurinus), Tiger (S. nigrocinctus), China (S. nebulosus) and Black (S. melanops) Rockfishes (Yamanaka and Logan 2010) (Figure 1-1). Other rockfish species that are often found with Inshore species include two small species that are not targeted by fisheries, Puget Sound, (S. emphaeus) and Greenstriped (S. elongatus), and juvenile Yellowtail (S. flavidus), Canary (S. pinniger), Redstripe (S. proriger), Bocaccio (S. paucispinis), Silvergrey (S. brevispinis), and Vermillion (S. miniatus) Rockfishes. Deacon (S. diaconus) (formerly Blue, S. mystinus) (Frable et al. 2015), Dusky (S. ciliatus) and Brown (S. auriculatus) Rockfishes are relatively uncommon in most nearshore waters in BC. 7 Figure 1-1. Images of the inshore rockfish species in BC. All species are in the genus, Sebastes and are as follows: a) Quillback, b) Yelloweye, c) Copper, d) Tiger, e) China, and f) Black Rockfish. (Photos by Janna Nichols). 8 Rockfish species possess life history characteristics that make them intrinsically vulnerable to effects of fishing (Leaman 1991, Parker et al. 2000, Love et al. 2002). Rockfishes have a closed (physoclistic) gas bladder and therefore suffer from barotrauma when they are brought to the surface from depth (Parker et al. 2006). Therefore, discarded rockfishes suffer high mortality rates (Hannah and Matteson 2007). This characteristic means that catch and release strategies are likely to be ineffective for preventing rockfish mortality as bycatch (Parker et al. 2000, Yamanaka and Logan 2010). Rockfishes also display sedentary behaviours that may also contribute to their vulnerability to overfishing. As a result of low movement rates of many rockfishes, once a localized reef has been fished out, it may take many years for the local population to recover via new recruitment (Parker et al. 2000). In addition, fishers will move from one reef to another and serially deplete the reefs while maintaining high catch rates (hyperstability) making early detection of catch declines difficult (Yoklavich 1998, Kronlund and Yamanaka 2001). Life history characteristics such as maximum age, large size, and late age at maturity also make rockfishes vulnerable to overfishing (Table 1-1). Rockfishes are long-lived and have delayed sexual maturity and slow asymptotic growth for up to 50% of their lives (Leaman 1991). Age at maturity is variable but is typically between 5-7 years and can be as old as 20 years or 50 cm long. Consequently, many species reach marketable size prior to becoming sexually mature (Parker et al. 2000). Delayed maturity and increased lifespan will be adaptive if mortality is primarily on pre-reproductive individuals or if the expectation of successful reproduction is low (Beamish et al. 2006). As a result, rockfishes have little buffering against the effects of reduced lifespan induced by exploitation (Leaman 1991). Although viviparous, rockfishes can be extremely fecund, producing between 10,000 and up to nearly 3 million larvae, depending on the 9 species and size/age of the female (Love et al. 2002) (Table 1-1). However, high fecundity rates do not appear to mitigate risk of extinction or enable more rapid recovery from exploitation (Dulvy et al. 2003). Long-lived species including rockfishes typically undergo numerous years of low recruitment interspersed with occasionally high or extremely good cohorts when oceanographic conditions are favourable. Because years of favourable conditions may be few and far between, longevity has been referred to as the \"storage effect\" (Warner and Chesson 1985, Dulvy et al. 2003). Truncating the age-structure of long-lived fish through fishing therefore increases extinction risk and reduces the recovery potential (Dulvy et al. 2003). Cheung et al. (2005) used life history traits of fishes in an evaluation of intrinsic vulnerability to extinction. Maximum length, age at first maturity, longevity, von Bertalanffy growth parameter K, natural mortality rate, fecundity, strength of spatial behaviour, and geographic range were used as input variables in a fuzzy logic expert system and rated on a scale of 1 to 100 with 100 being the most vulnerable (Cheung et al. 2005). Intrinsic vulnerability for rockfishes was calculated following this methodology and most inshore rockfishes have scores above 60 and as high as 78 for Yelloweye Rockfish (Magnuson-Ford et al. 2009). The Committee on the Status of Endangered Wildlife in Canada (COSEWIC) assessed both inside and outside populations of Yelloweye Rockfish as Special Concern in 2008 and Quillback as Threatened in 2009 (COSEWIC 2008, 2009). Yelloweye have been listed by Canada’s Species at Risk Act (SARA), but Quillback have not yet been listed. None of the other inshore species have been assessed yet and there are likely insufficient data to do so. Yelloweye Rockfish in Puget Sound have been proposed for designation as “Threatened” under American Endangered Species legislation (Williams et al. 2010). COSEWIC designations were based on life history criteria of Yelloweye Rockfish and declining trends in abundance demonstrated 10 through research survey data. Truncation of age/size structure of Yelloweye Rockfish has been observed in some locations (Kronlund and Yamanaka 2001) and local size/age structure change for Quillback Rockfish is also possible but has not been assessed. The loss of older fish in a population has large demographic consequences as larger fish produce greater numbers and quality of larvae (Berkeley et al. 2004b). Berkeley et al. (2004b) showed that older female Black Rockfish not only produced more larvae than younger females, but that the larvae were larger and were provisioned with greater resources in the form of a larger oil globule which enabled larvae to grow faster and to withstand longer periods of starvation. Surviving longer starvation periods increases survival (Berkeley et al. 2004a). Age-related differences in timing and location of spawning may also help to stabilize recruitment (Berkeley et al. 2004b). A study in Oregon found that older Black Rockfish extruded young earlier in the year and produced a greater proportion of recruits (Bobko and Berkeley 2004). Older female Blue, Olive and Yellowtail Rockfish also produce higher quantity and quality larvae that are released earlier in the spawning season (Sogard et al. 2008). The failure to recover in favourable ocean conditions due to truncated age composition and the fact that younger fish are not as productive as older fish has been called \"longevity overfishing\" (Beamish et al. 2006). Beamish et al. (2006) stressed the importance of managing the age structure of long-lived fishes such as rockfishes through the use of interconnected networks of MPAs (Berkeley et al. 2004b). Life history characteristics and ecological traits of fishes also influence the effectiveness of protection in an MPA. Traits include species maximum body size, as a surrogate for age at maturity, growth and reproductive output, habitat type, depth range, schooling behaviour, yearly displacement, home range size, territoriality, and mobility (Jennings 2001, Claudet et al. 2010). Inshore rockfishes possess many of these characteristics and have long been thought to be ideal 11 candidates for protection in MPAs (Carr and Reed 1993, Yoklavich 1998, Parker et al. 2000, Berkeley et al. 2004b). 12 Table 1-1. Life history traits of inshore and common shallow-shelf rockfish species found in Rockfish Conservation Areas in BC (Richards 1986, Love et al. 2002). BC specific figures are given when possible. Groupings are as follows: 1a= common BC inshore species; 1b=less common BC inshore species; 1c=inshore-shallow shelf; 2=shallow or shelf or slope species with shallower juvenile distributions. +100% maturity shown in absence of information on 50%. ), Information specific to BC (indicated with*) is given when possible because length and age at maturity have been found to increase with latitude (Haldorson and Love 1991, Love et al. 2002). Total 50% Maturity Depth (m) Max. Max. Female Male Fecundity Group Common Name Latin Name Range Size (cm) Age Size (cm) Age Size (cm) Age (Larvae/yr) 1a Copper S. caurinus 0-183 66 50 34 6 32 4 16,000-640,000 1a Puget Sound S. emphaeus 3-366 18 22 11 1-2 3,300-58,000 1a Quillback* S. maliger 0-274 61 95 29 11 25 10 ? 1a Black S. melanops 0-366 69 50 41 6-7 36 6-7 125,000-1,200,000 1a China S. nebulosus 3-128 45 79 30+ 6+ 30+ 6+ ? 1a Tiger S. nigrocinctus 18-98 61 116 ? 1a Yelloweye S. ruberrimus 15-49 91 118 46 19 54 22 1,200,000-2,700,000 1b Brown S. auriculatus 0-135 56 34 24-31 4-5 24-31 4-5 55,000-339,000 1b Deacon (formerly Blue) S. diaconus 0-549 33** 44 ? 1c Greenstriped S. elongatus 12-95 43 54 23 7 23 7 11,000-295,000 2 Yellowtail S. flavidus 0-549 66 64 36-45 7 32-44 6 56,900-1,993,000 2 Vermillion S. miniatus 6-436 76 60 37 5 35 5 63,000-2,600,000 2 Canary S. pinniger 0-838 76 84 35-45+ 7-9+ 41+ 7-12+ 260,000-1,900,000 13 1.4 Inshore Rockfish Fishery and Management in BC Inshore rockfishes have been harvested on the coast of BC for millennia by coastal First Nations peoples. A zooarchaelogical study on the west coast of Vancouver Island consistently found rockfish remains in middens that dated back as far as 1500 years ago (McKechnie 2007). Aboriginal people in the Salish Sea also fished and consumed rockfishes and it is thought that they were a staple food that could be harvested at any time of year when seasonally abundant species were not available (Williams et al. 2010). Commercial fishing for inshore rockfishes in BC began in the mid 1800's as they were caught incidentally in the Lingcod (Ophiodon elongatus) fishery. A directed hook-and-line fishery for inshore rockfishes expanded in the 1970's in response to the development of a lucrative live fish market in Vancouver. Quillback Rockfish are the target of the live fishery although the less common Copper, Tiger, China, and Black Rockfishes are also landed. Yelloweye Rockfish are targeted for a fresh, rather than live market. The fishery was unrestricted in the early 1980's until a license (termed \"ZN license\") and logbook system, as well as annual assessment and hook and line surveys, were put in place in 1986. Throughout the 1990's various management actions were taken, including the imposition of total allowable catch (TACs); however, rapid growth in this fishery outpaced measures to limit efforts. Other commercial groundfish fisheries such as trawl, Halibut, Lingcod and Dogfish fisheries also catch inshore rockfishes either as targeted catch or bycatch, therefore measures to limit catch in the directed fishery alone are insufficient to conserve stocks without also reducing bycatch and incidental take in these other fisheries. In addition to incidental catch in other groundfish fisheries, rockfishes are also targeted in recreational and Aboriginal fisheries, and are bycatch in salmon troll, shrimp trawl, and invertebrate trap fisheries (Yamanaka and Logan 2010). 14 Rockfish catch in the \"Inside\" or the protected waters of the Strait of Georgia, Johnston Strait and Queen Charlotte Strait, east of Vancouver Island, peaked in the late 1980's. Catch in the \"Outside\" waters, those waters west of Vancouver Island and the central and north coasts of B.C., peaked in the early 1990's and then sharply declined (Yamanaka and Logan 2010). Although data were insufficient at the time for a comprehensive stock assessment, symptoms of overfishing such as declining catch rates, and anecdotal information about local depletion of stocks and considerable at-sea discarding practices raised concerns for sustainability of stocks (Yamanaka and Logan 2010). The more accessible inside and southern fishing grounds likely experienced serial depletion of local reef areas whereby high catches are maintained by continually moving to new reefs (Kronlund and Yamanaka 2001). Serial depletion can cause hyperstability of catch rates masking declines in stock assessments (Hilborn and Walters 1992) as well as giving fishers a false impression of the population size (Kronlund and Yamanaka 2001). In the case of the ZN fishery, high participant turnover also eroded the historical frame of reference required for the appropriate impression of spatial trends in fishing success (Kronlund and Yamanaka 2001). In 2001, non-governmental organizations (NGOs) lobbied for actions to protect inshore rockfishes, and the American Fishery Society policy statement on the conservation and management of rockfishes (Parker et al. 2000) helped to bring attention to rockfish conservation (Yamanaka and Logan 2010). In November, 2001, the minister of Fisheries and Oceans committed to \"develop a plan to reverse the inshore rockfish decline and ensure stock rebuilding\" (Yamanaka and Logan 2010). DFO Science recommended four specific measures which became the basis of the “Inshore Rockfish Conservation Strategy”: 1) account for all catch; 2) decrease 15 fishing mortality; 3) establish areas closed to all fishing; and 4) improve stock assessment and monitoring (Yamanaka and Lacko 2001). Accounting and managing for total rockfish mortality in all groundfish fisheries henceforth became guiding principles in the integration of groundfish fisheries which brought about major changes to all groundfish fisheries, including 100% at-sea observer or electronic monitoring as well as dockside monitoring of these fisheries. Successful reform and integration of the fisheries has been attributed, in part, to the shared incentive for all fisheries to participate because all fisheries were united in their dependence on access to rockfish quota (Davis 2008). The integration scheme would also allow the transfer of fish quotas across these fisheries to cover the non-directed catch of rockfish, thereby eliminating discarding. Recreational fishing catch monitoring in the form of creel surveys and logbook reporting from guides and lodges was also expanded and enable bi-monthly catch estimates from the creel survey. Daily recreational bag limits for rockfishes were decreased from 10 to 5 rockfish in the north and central coast, from 5 to 3 on the West Coast of Vancouver Island, and from 3 to 1 on the Inside waters. Inshore rockfish total allowable catch (TAC) was reduced up to 50% in outside waters and 75% in inside waters between 2002 and 2005 (Yamanaka and Logan 2010). The value of the inshore rockfish catch between 2002-2005 is estimated at $ 2.3 million annually (Davis 2008). 1.5 Rockfish Conservation Areas A major component of the conservation strategy was the establishment of areas closed to fishing. A 2002 press release describing the strategy explained that \"Extensive inshore rockfish habitat must be protected to provide a buffer against scientific uncertainty and contribute to the protection and rebuilding of rockfish stocks.\" Proposed targets for closure from all fishing were up to 50% of rockfish habitat within the inside area and 20% of the outside area (Yamanaka and 16 Logan 2010). A team made up of DFO managers, scientists, enforcement and communications as well as a member from the province of BC, BC Parks, and Parks Canada Agency was established and stakeholder consultation was subsequently planned and carried out. Activities allowed in the RCAs were reviewed by the team which decided that fishing activities that were likely to incidentally or directly catch inshore rockfishes were prohibited while all other fisheries are still permitted (Table 1-2) (Yamanaka and Logan 2010). The RCA designation process followed three steps, with broad consultation at each step: 1) data gathering for closed-area proposals; 2) internal DFO review of proposals and verification with fishery catch data; and 3) DFO rockfish habitat analysis to meet the closed area targets, spread evenly across statistical areas. 30% of habitat in the waters between Vancouver Island and the mainland, and 20% of the remaining \"Outside\" area were targeted. Additional considerations in the placement, shape and boundaries of the RCAs included ease of description in fishery regulation, recognition by the public and ease of monitoring and enforcement (Yamanaka and Logan 2010). The spatial distribution of rockfish habitats coastwide are unknown; therefore, rockfish habitat was modeled using a bathymetrically derived complexity model (Ardron 2002) and historical rockfish catches (commercial and recreational) (Yamanaka and Logan 2010). This model was used to determine if the conservation targets had been met as well as to try to evaluate candidate RCAs. Broad consultation occurred throughout the designation process that lasted three years. The suite of 164 RCAs was officially implemented in 2007 although they were phased in between 2004-2007. Twenty-eight percent of modeled rockfish habitat was protected on the inside and 15% on the outside (Yamanaka and Logan 2010). Yamanaka and Logan note that “other closed-area initiatives (e.g. national marine conservation areas)” were underway and 17 were expected to result in further fishing closures in the outside area. Only the area of modeled rockfish habitat protected inside the RCAs was included in the final percentages so the total area shown in Table 1-3 is greater than the proportion of protected rockfish habitat that was used to calculate the targets. The mean RCA size is nearly 30 km2, the smallest, Hardy Bay-Five Fathom Rock, is less than a square kilometer in size, and the largest, West Aristazabal Island, is over 500 km2 (Table 1-3). The RCAs in northern and southern BC are shown in Figure 1-2 and Figure 1-3 but this analysis is limited to RCAs in southern BC (Figure 1-3). The potential of the RCAs to rebuild rockfish populations likely depends on the habitat quality found within the RCAs, on the delivery of new recruits to the RCAs, and on compliance with the regulations. The effectiveness of the RCA network has yet to be evaluated. 18 Table 1-2. Fisheries prohibited and permitted in RCAs by sector. Sector Prohibited Fisheries Permitted Fisheries First Nations  Fishing for food, social and ceremonial purposes. Commercial  Groundfish bottom trawl  Groundfish hook and line for Halibut, inside rockfish, outside rockfish, Lingcod, Dogfish  Sablefish by trap  Salmon trolling  Groundfish by mid-water trawl  Invertebrates by hand picking or dive  Crab by trap  Prawn by trap  Scallops by trawl  Salmon by seine or gillnet  Herring by gillnet, seine and spawn-on-kelp  Sardine by gillnet, seine, and trap  Smelt by gillnet  Euphausiid (krill) by mid-water trawl  Opal squid by seine Recreational  Groundfish by hook and line  Salmon trolling, jigging or mooching  Spearfishing  Invertebrates by hand picking or dive  Crab by trap  Shrimp/prawn by trap  Smelt by gillnet 19 Table 1-3. Total area and proportion of modeled rockfish habitat protected by the RCAs (Yamanaka and Logan 2010). RCA Statistics Total Area (km2) Area of Rockfish Habitat (km2) Number of RCAs 164 Total Area 4847.2 2060.3 Total Inside 1518.5 897.4 Total Outside 3328.7 1162.9 Mean/Median Size 29.6/10.8 Standard Deviation 61.2 Minimum Size 0.12 Maximum Size 509.1 20 Figure 1-2. Rockfish Conservation Areas in northern British Columbia, shown by green shading, are typically large. 21 Figure 1-3. Rockfish Conservation Areas in southern BC, shown by green shading. One hundred and twenty five of 164 of the RCAs are found in “inside” waters between Vancouver Island and the mainland and are typically smaller than RCAs in “outside” waters. 22 1.6 Overview of Dissertation In this dissertation, I present four studies that address the following questions concerning the recovery of rockfishes in Rockfish Conservation Areas in BC. 1.6.1 Does Post-settlement Recruitment Predict the Adult Abundance of Black Rockfish? In Chapter 2, I explore the effect of recruitment on Black Rockfish populations in Barkley Sound, BC. Recruitment and connectivity are important criteria for designing effective MPAs as coastal fish populations must be sustained by settling juveniles. I examined patterns of adult Black Rockfish abundance with respect to spatial and temporal variability in recruitment to determine the extent to which adult and juvenile habitats overlap, and how recruitment and post-settlement processes influence population density in and around a RCA. I measured Black Rockfish density on dive surveys at 30 sites where recruitment had previously been monitored. Sites with high recruitment in 2006, a strong recruitment year, had lower adult Black Rockfish density. There was no relationship between adult density and mean young-of-the-year recruitment among sites in 2007, 2008 and 2009; however, adult density increased with the mean abundance of 1-yr old Black Rockfish in 2009. Habitat variables such as topological complexity and the amount of rocky substrate predicted adult Black Rockfish abundance. Post-recruitment processes such as mortality and movement likely play a larger role than recruitment in determining adult density. 1.6.2 Assessing Population Recovery Inside British Columbia’s Rockfish Conservation Areas with a Remotely Operated Vehicle. In Chapter 3, I describe the results of a collaborative project with Fisheries and Oceans Canada to monitor the RCAs using a Remotely Operated Vehicle (ROV). I surveyed the fish communities of 35 RCAs and adjacent unprotected areas in southern BC using a ROV between 23 2009 and 2011. Habitat features such as percent rocky substrates and depth influenced density of Quillback Rockfish, Yelloweye Rockfish, Greenstriped Rockfish, Kelp Greenling, Lingcod and all inshore rockfishes combined, while reserve status did not. I also calculated habitat–based average densities and used the mean log Response Ratio (RR) of the density inside to outside of RCAs to determine if the amount of fishing outside the RCA, previous fishing history, the age, area or perimeter to area ratio influenced population recovery. Few positive reserve effects were apparent for any species/group and there was no relationship among RR and any other factors tested. The results give no indication that demersal fish populations have recovered inside the RCA system. Ongoing monitoring is essential to assess population recovery over time and evaluate the RCAs in terms of criteria such as habitat quality, habitat isolation and the level of compliance in order to enhance their effectiveness. 1.6.3 Lack of Recreational Fishing Compliance May Compromise Effectiveness of Rockfish Conservation Areas in British Columbia. In Chapter 4, I investigate one possible reason for the lack of response to the RCAs identified in Chapter 3. Compliance with spatial fishing regulations (e.g. marine protected areas, fishing closures) is one of the most important, yet rarely measured, determinants of ecological recovery. I used aerial observations of recreational fishing events from surveys before, during and after 77 RCAs were established in BC. There was no evidence of a change in fishing effort in 83% of the RCAs, and effort in five RCAs increased after establishment. Fishing effort in open areas adjacent to the RCAs declined with time and was higher than effort in the RCAs in all three years. I also used compliance data for 105 RCAs around Vancouver Island to examine the drivers of compliance. Compliance was related to the level of fishing effort around the RCA, the size and perimeter-to-area ratio of RCAs, proximity to fishing lodges and the level of 24 enforcement. Non-compliance in RCAs may be hampering their effectiveness and impeding rockfish recovery. Education and enforcement efforts to reduce fishing effort inside protected areas are critical to the recovery of depleted fish stocks. 1.6.4 How Do They Score? An Evaluation of Rockfish Conservation Areas Using a Conservation Score that Combines Rockfish Habitat and Key Reserve Features. Chapter 5 synthesizes the results of the previous chapters to investigate the overall effectiveness of the RCA network. The studies described in Chapters 2 and 3 both demonstrate the importance of habitat to fish densities. Furthermore, I found that the quality of habitat in some RCAs I surveyed with the ROV was low and I recommended a comprehensive habitat analysis be conducted on all of the RCAs. Therefore, I modeled rocky reef habitat using Random Forest Classification and variables derived from a bathymetry with an intermediate resolution of 20 m. Using this habitat model, I calculated the total habitat area, the percent habitat and habitat isolation in 144 RCAs in southern BC. I combined these habitat metrics with my estimate of recreational compliance from Chapter 4, as well as RCA size, commercial compliance, rockfish bycatch in permitted fisheries, and connectivity into a single additive conservation score. The conservation score of the RCA was related to the log reserve ratio (RR) of the density inside to outside of RCAs for Quillback Rockfish, but not Yelloweye Rockfish. RCAs with low conservation scores are not likely to be effective and managers should evaluate the reasons for low scores and address reserve shortcomings in an adaptive spatial management framework. The Conservation Score could be used to select RCAs for further monitoring to ensure rockfish conservation is achieved in the RCAs. 25 Chapter 2: Does Post-settlement Recruitment Predict the Adult Abundance of Black Rockfish? 2.1 Introduction Marine Protected Areas (MPAs) are implemented to restore and conserve depressed populations of marine fishes and to support the sustainability of fisheries (Sale et al. 2005, Gaines et al. 2010a). Worldwide, evidence is mounting that biomass of fishes and invertebrates increases inside well designed reserves (Hamilton et al. 2010, McCook et al. 2010, Russ and Alcala 2011, Edgar et al. 2014). However, the effectiveness of marine reserves varies widely. Reserves need to be evaluated because ineffective reserves can lead to a false sense of security and hamper or prevent other effective management actions (Allison et al. 1998, Sale et al. 2005). Spatial variability in the supply of recruits may contribute to the effectiveness of MPAs (Grorud-Colvert and Sponaugle 2009, Wen et al. 2013) because an MPA must receive new recruits or migratory individuals for populations within the reserve to be sustained (Carr and Reed 1993, Planes et al. 2000, Halpern and Warner 2003, Gaines et al. 2010a). The level of recruitment in an MPA will also determine how quickly populations rebound in response to protection (Grorud-Colvert and Sponaugle 2009). Wen et al. (2013) found that the adult densities of two targeted fish species were two to three times higher in reserves in Australia that included recruitment “hotspots” than reserves that did not include hotspots. Situating reserves in areas receiving large inputs of recruits may be important to their success at promoting population recovery. 26 Although recruitment may limit fish populations, post-settlement density-dependent factors, such as movement, growth, and mortality can dampen the effect of recruitment (Carr and Syms 2006). If recruitment limitation is a major determinant of population variation, then adult population size should track recruitment intensity with some lag in time for growth. Alternatively, if post-settlement processes are most important, then adult population size should be independent of recruitment intensity (Caley et al. 1996). Uncertainty about how recruitment and post-settlement processes influence fish populations complicates the management and conservation of reef fishes. In Hawaii, headlands persistently intercepted and concentrated larval Hawkfish (Paracirrhites arcatus) but post-settlement processes dampened the effects of high recruitment (DeMartini et al. 2013). Reserves in The Florida Keys had different levels of larval supply but similar rates of juvenile recruitment (Grorud-Colvert and Sponaugle 2009). Abundance of larval (Ralston and Howard 1995) or young-of-year rockfishes (Laidig et al. 2007) has been related to strong cohorts evident in the fishery at very broad scales; however, recruitment signals in adult temperate reef fish populations at a finer scale have not been demonstrated (Carr and Syms 2006). Caselle et al. (2010a) found that abundance of adult Kelp Bass (Paralabrax clathratus) was predicted by larval supply at regional levels, but this relationship broke down at finer scales where recruitment and adult survivorship were density-dependent. The relative roles of recruitment and post-settlement processes remain in question. A system of 164 Rockfish Conservation Areas (RCAs) was implemented in British Columbia (BC) between 2004 and 2007 to help conserve overfished populations of inshore rockfishes (genus Sebastes) (Yamanaka and Logan 2010). Rockfishes are long-lived, slow growing and late to mature with sporadic reproduction (Love et al. 2002). In long-lived species like rockfishes, strong recruitment events produce cohorts of adults that survive to reproduce 27 over periods of years. This pattern of reproduction is termed the “storage effect” whereby strong recruitment events are “stored” in the adult population (Warner and Chesson 1985). Capturing spatial and temporal variation in recruitment in space and time may therefore be particularly important to the design of conservation areas for the protection of rockfishes. In this study, I examined the influence of Black Rockfish (S. melanops) recruitment on the abundance of adult and sub-adult Black Rockfish by comparing the density and size structure of fish at sites where recruitment rates had been previously monitored in Barkley Sound, BC. Sites were distributed throughout the sound in six different regions, including inside the Broken Group Islands RCA. Black Rockfish can live to be 50 years old, reach 50% maturity at 6 to 7 years of age and occupy relatively small home ranges of 0.55 km2 (Love et al. 2002, Parker et al. 2007). Larval Black Rockfish settle into nearshore habitats such as kelp forests and eelgrass meadows and move into deeper rocky habitats as they grow (Love et al. 2002). Markel (2011) and Lotterhos and Markel (2012) monitored rockfish settlement at 32 kelp forest sites in Barkley Sound from 2006-2009. Recruitment was highly variable among sites and years, and especially strong in 2006 when mean abundance of recruits was 10 times higher than in other years (Markel 2011). Genetic analysis of the 2006 cohort showed that despite the high abundance of recruits, genetic diversity was low and the incidence of siblings was high, suggesting a “sweepstakes” recruitment event whereby a few adults achieved high reproductive success (Lotterhos and Markel 2012). Sites inside the RCA had lower recruitment rates in 2006 than those outside of the RCA. However, it remains unknown whether the large 2006 cohort remained dominant in subsequent years or if post-settlement processes reduced its impact on the population. In addition, it is unknown whether the spatial pattern of recruitment intensity throughout Barkley 28 Sound corresponds to variability in adult population density, or if the RCA includes areas of both adult and juvenile habitat (Lotterhos and Markel 2012). We examined patterns of Black Rockfish abundance with respect to spatial and temporal variability in recruitment intensity. I tested for spatial co-variation between adult density and settlement rates in Barkley Sound. Finally, I asked how recruitment and adult density vary in and around the Broken Islands RCA in Barkley Sound to determine whether it includes areas of high value as adult and juvenile habitat within its boundaries. 2.2 Methods 2.2.1 Study System This study took place in Barkley Sound (48◦ 50’N, 125◦ 22’W) on the southwest coast of Vancouver Island, Canada. Two groups of islands, the Broken Group Islands to the north and the Deer Group Islands to the south and three wide channels are prominent features of Barkley Sound. The Broken Group Islands are within the Pacific Rim National Park Reserve as well as the Broken Islands Group RCA (Figure 2-1). The village of Bamfield is located on the southern side of Barkley Sound and is home to the Bamfield Marine Sciences Centre where this research and the previous recruitment studies were undertaken. Sites were distributed across Barkley Sound in the following Regions: George Fraser Island, Loudoun Channel, Broken Group Islands (sites are inside and outside of the RCA), Deer Group Islands, and “Prasex” (from Prasiola Point to Execution Rock on Vancouver Island) (Figure 2-1). Regions are separated from each other by at least 2 km. We measured adult rockfish density using SCUBA surveys close to the locations sampled for recruiting rockfish with Standard Monitoring Units for Recruitment of Fishes (SMURFs) in 29 Markel (2011) and Lotterhos and Markel (2012). In Barkley Sound, kelp forests are largely constrained to waters shallower than 6 m. I sampled the rocky reef habitat between 6-18 m deep in closest proximity to each recruitment sampling location. The median distance between pairs of adult survey sites and recruitment sampling locations was 244 m (83-1990 m). Three sites in the Broken Group Islands (Hand, Turret and Chalk) and one in the Deer Group Islands (Kirby) were farther away due to shallow depths and inappropriate adult habitat adjacent to the recruitment sites (Figure 2-1). At three of these locations, different dive sites were surveyed in 2010 and 2011 in order to target better habitat (Hand) or to sample closer to the SMURF site (Kirby and Chalk). Rocky reef habitat was identified based on rocky shoreline characteristics, depth contours on nautical charts and depth sounder, and visual observations of the habitat while SCUBA diving. In this study, recruitment is defined as the mean number of young of the year (YOY) rockfish caught in SMURFs (Ammann 2004). Standard Monitoring Units for Recruitment of Fishes have been used to measure coastal fish recruitment in BC (Markel 2011, Lotterhos and Markel 2012) and California (Ammann 2004, Caselle et al. 2010a, Caselle et al. 2010b). Each SMURF consists of a 1.0 m x 0.35 m cylinder of plastic garden fencing material (2.5 cm x 2.5 cm mesh), stuffed with strips of plastic snow-fence material which serves as habitat for settling juvenile fishes. Sites were sampled approximately every two weeks between June and August using a net and snorkelling equipment to capture all fish occupying each SMURF. 30 Figure 2-1. Sampling sites in Barkley Sound, BC. Recruitment sites (grey) were sampled 2005-2009. Dive sites were sampled in 2010 (red) and 2011(blue). Region Key: GF=George Fraser Island, LO=Loudoun Channel, BG=Broken Group, DG=Deer Group Islands, PE=Prasex. 31 2.2.2 Rockfish Recruitment We calculated the mean recruitment as (Mean Black Rockfish/SMURFSite,Year) for each sampling year between 2006-2009. Standard Monitoring Units for Recruitment of Fishes were sampled approximately every two weeks from late May to September. In 2008, SMURFs were not sampled in August. Not all of the SMURF sites were sampled in all years. In 2006, only 2 sites were sampled inside the BGI RCA. Three different sites were sampled in the RCA in 2007, 2008 and 2009. One-year-old Black Rockfish are caught in addition to the YOY and enumerated in the SMURFs. 2.2.3 Black Rockfish Dive Survey Black Rockfish density was censused on replicate transects by SCUBA divers. I sampled two (in 2010) and four (in 2011) 30 m x 3 m transects per site distributed between 6-18 m of depth. Upon descending, divers searched for rocky habitat within this depth range. Trained divers identified and counted Black Rockfish as well as all rocky-reef fish found within the transect. The second diver followed the transect line to record depth, measured with a dive pressure gauge every 2m, substrate type, relief and habitat complexity according to classifications based on Pacunski and Palsson (2001) (Table 2-1). The divers swam back to the start while reeling in the line and completed a second transect 3 to 7 m shallower following the opposite compass bearing. In 2011, divers completed four transects per dive following the same procedure, except subsequent transects were 1.5 to 3 m shallower and two transects were performed in each compass bearing. To calculate rockfish density, I summed the number of rockfish per transect, divided by the area of the transect (90 m2) and multiplied by 100 to give the density per 100m2. Only fish >8cm in length were included in the density estimates. 32 Table 2-1. Habitat Variable Descriptions. Variable Description Substrate Bedrock Continuous rock or hardpan Boulder Rocks > 20 cm in diameter Cobble Rocks 5-20 cm in diameter Mixed Coarse Gravel (rocks <5 cm in diameter), sand and shell debris Mud Mud Relief R1=None Flat or rolling substrate with vertical relief up to 0.5 m R2=Low Vertical relief from 0.5 m to 2 m R3=High Vertical relief >2 m, slope <45 degrees R4=Wall Vertical relief >2 m, slope >45 degrees Complexity C1=Simple Smooth surfaces, no crevices C2=Low Some irregularity, few crevices (<25% of area) C3=Medium Moderate irregularity, ~25-50% of habitat with crevices C4=High Highly irregular, many crevices (>50% of area with crevices) 33 2.2.4 Black Rockfish Density and Habitat Analysis In order to analyze the habitat data, substrate, complexity and relief observations were summed over each transect, divided by the number of observations (15) and presented as a percent for each transect. Since more than one substrate category could be chosen for the same section of transect, total cover values >100 % were common. I divided the percent of each class by the total percent observed. I combined R1 and R2 into Low Relief and R3 and R4 into High Relief and C1 and C2, and C3 and C4 into Low and High Complexity respectively. I also calculated the mean depth of each transect. I added the % boulder and bedrock into a rock category. I logit transformed all percentages (R package Arm) (Gelman and Su 2013) prior to analysis. To estimate wave exposure at each site, I obtained fetch measurements for every 500 m of coastline in Barkley Sound (Ed Gregr, Scitech Consulting, personal communication). For each point, a full circle of 120 lines spaced every 3 degrees and measuring up to 200 km were drawn and the distance to the nearest land was determined for each line. The distance of all lines was summed as a measurement of fetch (Lessard and Campbell 2007). I used the “near” tool in ArcGIS 10.1 to find the closest fetch estimate to each dive site. The mean distance between the dive site and the sum-of-fetch point was 161 m (SD = 60.9). We used a linear mixed model to investigate the relationship between Black Rockfish density and habitat using the package lmerTest (Kuznetsova et al. 2014) in R (R Development Core Team 2008). I used a square root transformation to normalize all density observations prior to analysis. I used dive site nested within region (Broken Group In the RCA, Broken Group outside of the RCA, Deer Group, George Fraser Island, Loudoun Channel and Prasex), nested within the year the dive surveys took place (2010, 2011) as random variables. I modeled effects 34 of habitat using data from all sites and the logit transformed % rock, % high complexity, and % high relief sampled on transects, the mean depth and fetch as fixed factors. I fit this model with Maximum Likelihood estimation to compare the models with different fixed variables (Zurr et al. 2009). In order to determine which habitat variables affect Black Rockfish Density, I used the R package MuMIn to compare all sub-models and to rank them by their Akaike Information Criterion adjusted for small sample sizes (AICc) (Barton 2013). I refit the best model with the lowest AICc score with Restricted Maximum Likelihood (REML) (Zurr et al. 2009). 2.2.5 Black Rockfish Density and Recruitment Analysis Markel (2011) and Markel and Lotterhos (2012) measured recruitment (Appendix Table 1-1) at different sites in within the BGI RCA in 2006 than they did in 2007, 2008 and 2009 (Appendix Table 1-2); therefore, missing data prevented us from using a single model to investigate the effects of annual recruitment on adult density. Accordingly, I made one model to investigate if Black Rockfish density was related to the strong recruitment event in 2006 and a second model to investigate if Black Rockfish density was related to recruitment in 2007, 2008, 2009 or to the number of 1-year-old Black Rockfish caught in 2009. Density data were square root transformed prior to analysis. For both models, I used dive site nested within region, nested within the year the dive surveys took place (2010, 2011) as random variables. I used the package lmerTest (Kuznetsova et al. 2014) and fit the models with REML. To investigate whether my results were affected by the distance between the dive and SMURF sites, I re-ran the recruitment models without sites where the two were separated by more than 1,000 m. 35 2.2.6 Rockfish Conservation Area Evaluation Marine reserves are often evaluated using a Response Ratio (RR) (Equation 1) that compares the mean density of fish inside to outside of a reserve (Hedges et al. 1999, Russ et al. 2005, Claudet et al. 2010, Hamilton et al. 2010, Edgar et al. 2014). RR = ln⁡(X̅inX̅out) Equation 1 We collected data on all reef fishes in addition to Black Rockfish, so I was able to calculate the mean density of Quillback Rockfish (S. maliger), Copper Rockfish (S. caurinus), Black Rockfish, and all inshore rockfishes combined (Quillback, Copper, Black, China (S. nebulosus), Tiger (S. nigrocinctus) Rockfishes), as well as Lingcod (Ophiodon elongatus) and Kelp Greenling (Hexagrammos decagrammus) observed on all transects inside the BGI RCA and outside of the BGI RCA. (2) 2.3 Results We sampled 65 transects at 32 sites in 2010 and 119 transects at 30 sites in 2011. In 2011, I only sampled three of the five sites at George Fraser Island because the sites were all close together. In both years, Black Rockfish were the most commonly observed Rockfish, followed by Blue and Copper Rockfish. All surveys were conducted in rocky reef habitats and although no mud was observed on any transect, habitat was variable and all other substrate, relief and complexity classes were encountered. Habitat sampled in 2011 had a higher mean percentage of bedrock, high relief and high complexity in most regions than in 2010. The mean 36 depth sampled was consistent between the two years with a mean depth of 10.6 m (SD=2.6) in 2010 and 10.2 m (SD=2.4) in 2011 (Table 2-2). 2.3.1 Black Rockfish Density and Habitat Black Rockfish density was highly variable among sites and years (Figure 2-2). Visual inspection of residual plots of all models revealed no deviations from homoscedactisity or normality. The habitat model with the lowest AICc only retained the percent high complexity and the percent rocky substrates, and dropped percent relief, mean depth, and fetch. There was a strong positive relationship between Black Rockfish density and high complexity. Density also tended to be positively related to the amount of rocky substrates (Figure 2-2, Table 2-3). 2.3.2 Black Rockfish Density and Recruitment Black Rockfish density was negatively related to recruitment in 2006 (Figure 2-3, Table 2-3). Young-of-the-year recruitment in 2007, 2008 and 2009 were not significantly related to adult Black Rockfish density (Figure 2-3) but the sites with the highest recruitment of 1-year-old Black Rockfish in 2009 had higher adult density (Figure 2-3, Table 2-3). Excluding the dive sites that were more than 1,000 m from the SMURF sites did not affect these results. 2.3.3 Fish Density in the Broken Group RCA The average Response Ratio (RR) of all rockfishes (Quillback, Copper, Black and inshore rockfishes combined) was greater than 0, indicating higher densities inside the RCA than outside. The RRs of Kelp Greenling and Lingcod were lower than zero, indicating lower densities of these species inside the RCA (Figure 2-5). 37 Table 2-2. Mean and Standard Deviation of Habitat variables (%) by year and Region. Group GF LO BG-out BG-in DG PE Year 2011 2010 2011 2010 2011 2010 2011 2010 2011 2010 2011 2010 Bedrock 58.9 53.3 66.1 37.3 51.4 58.7 66.0 33.3 78.1 67.8 82.2 50.0 SD 29.0 35.8 44.2 39.4 31.5 37.3 33.9 39.5 28.3 36.3 28.4 42.1 Boulder 77.8 61.3 43.9 68.0 74.3 40.0 50.8 26.7 41.9 34.4 33.7 53.3 SD 23.3 29.6 35.5 37.0 28.3 38.0 32.7 22.2 31.1 35.8 29.3 35.7 Cobble 90.0 45.3 45.3 49.3 60.4 24.0 50.1 64.0 35.4 27.8 21.7 11.1 SD 10.1 26.4 39.4 47.2 18.1 23.3 33.6 38.9 29.5 38.8 22.6 14.3 Mixed Coarse 73.9 48.0 64.7 22.7 56.4 46.7 67.8 70.7 58.6 23.3 63.5 30.0 SD 32.2 37.5 42.8 38.1 26.3 37.4 28.5 41.8 27.4 32.3 32.8 35.2 Low Relief 46.1 88.0 44.7 92.0 53.8 58.7 49.0 78.7 38.5 52.2 23.0 66.7 SD 33.8 21.9 40.6 8.2 35.0 16.3 39.0 41.5 33.1 41.9 23.6 27.9 High Relief 53.9 13.3 55.0 9.3 52.8 41.3 57.9 21.3 66.7 52.2 78.3 35.5 SD 33.8 24.7 40.4 10.5 31.2 16.3 37.5 41.5 31.8 41.3 23.3 27.2 Low Complexity 50.0 40.0 53.9 30.7 45.3 65.3 43.8 89.3 57.8 34.4 41.1 33.3 SD 28.1 27.7 38.6 39.4 34.4 27.5 27.0 19.2 26.5 25.5 26.3 18.0 High Complexity 50.0 57.3 45.6 66.7 54.2 38.9 58.1 10.7 41.4 65.6 60.9 68.9 SD 28.1 29.3 38.6 38.0 34.5 27.6 27.8 19.2 25.4 25.5 23.5 19.1 Depth (m) 9.6 9.7 10.7 10.9 11.0 10.2 9.8 12.1 10.6 11.1 9.5 9.8 SD 2.2 1.8 2.4 2.2 1.9 3.2 2.0 2.4 2.8 2.4 2.8 2.3 Minimum Depth (m) 5.6 6.4 6.4 6.7 6.1 3.7 5.4 7.0 2.7 7.0 3.6 5.2 Maximum Depth (m) 14.5 15.0 16.8 14.6 14.2 15.0 19.2 17.0 16.0 16.1 15.6 13.9 38 Figure 2-2. Black Rockfish Density by % complexity and rocky substrates by region. Regression lines are from the LME model. BG-in =Inside the Broken Group RCA, BG-out =Outside the Broken Group RCA, DG=Deer Group, GF=George Fraser, LO=Loudoun Channel, PE=Prasiola Pt. to Execution Rock. 39 Figure 2-3. Mean density of Black Rockfish by site in 2010 and 2011 versus mean YOY recruitment in 2006. Regression lines are from the LME model. Symbols indicate grouping by region. BG-in =Inside the Broken Group RCA, BG-out =Outside the Broken Group RCA, DG=Deer Group, GF=George Fraser, LO=Loudoun Channel, PE=Prasiola Pt. to Execution Rock. 40 Figure 2-4. Mean density of Black Rockfish by site in 2010 and 2011 versus mean YOY recruitment in 2007, 2008 and 2009 and 1-year-old recruits in 2009. Regression lines are from the LME model. Symbols indicate grouping by region. BG-in =Inside the Broken Group RCA, BG-out =Outside the Broken Group RCA, DG=Deer Group, GF=George Fraser, LO=Loudoun Channel, PE=Prasiola Pt. to Execution Rock. 41 Table 2-3. Model results of three Linear Mixed Effects Models with Year/Region/Site as nested random effects. Degrees of Freedom (df) were calculated with Satterthwaite’s approximation. 95% Confidence intervals (2.5% and 97.5%) are also shown. Significant terms are indicated (*). Variable Estimate SE df t-value p-value 2.5% 97.5% 1. Habitat Model Intercept 1.03 0.53 83.6 1.95 0.05* -0.002 2.16 Complexity 0.34 0.09 165.3 3.93 0.001* 0.17 0.51 Rock 0.01 0.01 173.7 1.92 0.06 -0.001 0.03 2. Recruitment Model 2006 Intercept 2.56 0.36 45.2 7.04 <0.0001* 1.85 3.27 YOY 2006 -0.09 0.04 43.8 -2.18 0.04* -0.17 -0.01 3. Recruitment Model 2007, 2008, 2009 Intercept 0.86 0.43 41.2 2.02 0.05* 0.04 1.66 YOY 2007 0.43 0.41 45.2 1.03 0.31 -0.35 1.22 YOY 2008 -0.1 0.12 41.2 -0.82 0.41 -0.32 1.22 YOY 2009 0.48 0.44 40.5 1.09 0.28 -0.35 1.31 1yr 2009 1.88 0.71 41.2 2.68 0.01* 0.56 3.23 42 Figure 2-5. Mean Response Ratio and 95% Confidence Interval of reef fish species in the Broken Group Islands RCA as compared to all sites in Barkley Sound outside of the RCA (n=2). 43 2.4 Discussion My data indicate very little spatial overlap between habitats of adults and new recruits of Black Rockfish in Barkley Sound. There was even a negative relationship between Black Rockfish densities and the level of recruitment, i.e., the number of juveniles being added to the population, observed in 2006. Spatial variation in adult densities was also not influenced by the level of YOY recruitment observed at sites in other years. These results indicate that post-recruitment processes such as mortality and movement are sufficient in Black Rockfish to dampen any signal of recruitment strength. Post-recruitment density-dependent mortality or ontogenic movements away from these sites may have dampened the signal from this cohort. Post-recruitment migrations and ontogenic movements of fish have been shown to erase the legacy of a strong recruitment, particularly at finer scales (Gillanders et al. 2003). When the nursery habitat is not the same as the adult habitat, as is the case for Black Rockfish, recruitment is actually delayed during a variable time after settlement until the juvenile fish migrate to adult habitats (Planes et al. 2000). Recruitment can also be defined to be when the new fish join the population that is fished (Carr and Syms 2006). Determining the appropriate stage to evaluate recruitment is also important in the assessment of the RCA in the Broken Group Islands and in MPAs in general. Oceanographic conditions, specifically warmer temperatures at parturition and during the larval pelagic phase, and strong upwelling during settlement that affects larval transport, have been related to variation among years in Black Rockfish recruitment in the recruitment dataset analyzed here (Lotterhos and Markel 2012). However, my analysis shows no indication that the recruitment of YOY rockfish was related to my adult abundance data; although 1-year-old abundance was. Adult Black Rockfish densities at my sites were positively related to the 44 abundance of 1-year-old fish caught in the SMURFS in 2009. In contrast, other authors have suggested that year-class strength for rockfishes is primarily set during the larval phase (Ralston and Howard 1995, Yoklavich et al. 1996, Laidig et al. 2007). Laidig et al. (2007) used a 20 year data set of YOY rockfish surveys to show that levels of recruitment of Black, Blue and Yellowtail Rockfishes in northern California varied synchronously and were largely determined by oceanographic conditions in February to March. Furthermore, year-specific adult Yellowtail Rockfish landings at nearby ports were significantly correlated to levels of recruitment. My data indicate that for Black Rockfish populations on the West Coast of Vancouver Island, cohort strength may be set at a later life-history stage. We hypothesize that the winter conditions experienced by YOY fish determine recruitment to the adult population in Barkley Sound. Specifically, the winter conditions experienced by the 2006 cohort negatively affected their abundance. The winter of 2006-2007 was one of the worst storm seasons on record in BC (DFO 2007). A measure of storminess for 2006, the Sea Surface Height (SSH) Anomaly, measured at Point Atkinson, BC and compared to data from 1963 to 2010, was the fourth highest on record (Tinis 2010). Storms may cause higher mortality or movements of juvenile fishes because they move away from exposed sites before or during winter storms (Green and Starr 2011). Kelp abundance, particularly at exposed sites would also be affected by storms. Long-distance movements of sub-adult Black Rockfish have been associated with storms and high wave energy before (Green and Starr 2011). Although adult Black Rockfish have a home range of 0.55 km2 (Parker et al. 2007), Green and Starr (2011) found that tagged immature adult Black Rockfish exhibited greater movement rates than adult fish. Two thirds of tagged fish showed high residency and 0.25 km2 home ranges, but the other third left the study site and moved more than 50 km away, particularly following storms. Carr 45 (1991) also observed a significant shift in habitat by juvenile Black Rockfish following fall storms. Black Rockfish left the kelp and significantly selected high relief rocky substratum in the winter (Carr 1991). Similar juvenile migrations might be responsible for redistributing Black Rockfish recruits throughout the Barkley Sound. Mortality rates may also be affected by winter storms, either by increased predation during movements or by the loss of kelp habitat. Structurally complex habitats such as kelp beds provide refuge space from predators and therefore reduce mortality rates (Johnson 2007). Conversely, favourable winter conditions during the winter of 2008-2009 may have benefited survival for the 2008 cohort. Although 2008 YOY recruitment levels were only slightly above average, the number of 1-year-old fish caught in the SMURFS was much greater than in any other year. That year was the calmest winter on record between 1963 and 2009 (Tinis 2009). Exposure, measured as the wave fetch at my sites, did not, however, influence Black Rockfish density and was not retained in the optimal habitat model. Markel (2011) found exposure to be one of the most important factors influencing recruitment. Exposed sites in Loudoun Channel had the highest recruitment rates in 2006. Conversely, some of the lowest Black Rockfish densities were observed at sites in Loudoun Channel. This drove the negative relationship between adult abundance and the 2006 recruitment. Many sites in Loudoun Channel also had very low levels of habitat complexity. Johnson (2007) found that although recruitment did not vary with habitat complexity, complexity altered the mortality rates of Blue Rockfish (S. mystinus) recruits. At low levels of habitat complexity formed by rocky substrates and kelp, mortality was high and independent of density. As complexity increased both density-dependent and density-independent mortality decreased (Johnson 2007). Mortality of YOY Black Rockfish, at the scale of an entire kelp forest, has also been shown to be density-dependent as predation 46 increases with a shortage of refuge space at high juvenile densities (Johnson 2006). Density-dependent or independent mortality may have similarly been high in the low complexity habitat where many of the 2006 YOY recruited in Loudoun Channel. It seems probable that those sites acted as a sink of recruits due to mortality or sources of migrants to other sites with better habitat for larger fish. We found a significant positive relationship between adult Black Rockfish density and habitat complexity. Proportion of rocky substrate on transects also influenced density. Rocky habitat is known to be important to rockfishes (Love et al. 2002). Black Rockfish have been found on complex rock bottoms, near vertical bedrock walls, and around kelp forests in high- and low-relief rocky terrain in other studies (Leaman 1976, Love et al. 2002, Johnson et al. 2003, Parker et al. 2008). Structurally complex habitat may reduce mortality by providing refuge space from predators (Johnson 2007). We also compared demersal fish densities in the BGI RCA to densities in the other regions in Barkley Sound using the log response ratio (RR). The RR for Black Rockfish was positive in both years, indicating higher densities inside the conservation area than outside, although just slightly above zero (0.14) in 2011. The RR of Quillback and Copper Rockfish as well as an aggregated inshore rockfish group were also positive in both years. The BGI RCA was designated in 2004, so it had been in place for only 6-7 years at the time of sampling (Edgar et al. 2014). Although recruitment sites sampled in the BGI RCA in 2006 showed low levels of recruitment, the sites that were sampled in the RCA in 2007 and 2009 showed levels of recruitment comparable to the rest of the Barkley Sound or, in the case of 2008, above average. Interestingly, the 1-year-old Black Rockfish abundance in 2009 was however, lower in the RCA than in the other regions (See Appendix Table 1-1). Recruitment success of YOY fishes in MPAs 47 in the Mediterranean Sea has been shown to be negatively affected by greater abundance of predators, which reduces the survival of recruits (Planes et al. 2000). Greater predation may also occur in the BGI RCA due to increased densities of adult Rockfishes. However, Lingcod, an important fish predator, as well as Kelp Greenling were less abundant in the RCA. Several authors have recommended the utility of determining physical proxies of larval supply such as headlands that concentrate larvae (DeMartini et al. 2013) and recruitment strength (e.g. indices of upwelling (Caselle et al. 2010a) to inform conservation planning and fisheries management (Bode et al. 2012). The factors affecting Black Rockfish recruitment in Barkley Sound have been shown to be highly complex at the larval and settlement stage (Lotterhos and Markel 2012) and are also likely complex during the post-recruitment phase. Finding suitable indicators of recruitment may, therefore, prove to be challenging in this system. Adult Black Rockfish density is, however, positively related to habitat complexity. Habitat complexity may also affect mortality rates of juvenile Black Rockfish. Habitat complexity would therefore be an important criterion for design or evaluation of RCAs. Habitat complexity and the proportion of rocky substrates are also easier to measure than fish recruitment because it is more static, better understood, and can be modeled using bathymetry and other remotely sensed data (i.e. Iampietro et al. 2008). My results indicate that monitoring recruitment without considering post-recruitment movement, mortality and over-wintering success is not likely to be a good way to predict the eventual effectiveness of RCAs. 48 Chapter 3: Assessing Population Recovery Inside British Columbia’s Rockfish Conservation Areas with a Remotely Operated Vehicle 3.1 Introduction Networks of Marine Protected Areas (MPAs), or reserves that exclude fisheries, are being implemented worldwide to conserve exploited species and sustain fisheries (Gaines et al. 2010a). The use of MPAs has been shown to be a successful strategy to increase the size, abundance and diversity of species protected within them (Allison et al. 1998, Mosqueira et al. 2000, Halpern and Warner 2002, Halpern 2003, Alcala et al. 2005, Claudet et al. 2008, Lester et al. 2009, Babcock et al. 2010, Gaines et al. 2010a, Edgar et al. 2014). Monitoring is critical to the implementation of MPA networks in fisheries management because ineffective reserves can give resource managers a false sense of security and prevent actions that might otherwise help to achieve the goals of MPAs (Allison et al. 1998, National Research Council 2006, Gaines et al. 2010a, Gaines et al. 2010b, Hamilton et al. 2010). Ecosystem Based Management and Adaptive Management require an understanding of which MPAs contribute the most to recovery of over-exploited populations to inform future actions (Hamilton et al. 2010, White et al. 2011). In response to conservation concerns associated with a sharp decline in inshore rockfish catches throughout the 1990’s in the Northeast Pacific, Fisheries and Oceans Canada (DFO) implemented a system of 164 Rockfish Conservation Areas (RCAs) in British Columbia (BC), Canada, as part of a Rockfish Conservation Strategy. Rockfish Conservation Areas were established between 2004 and 2007 and prohibit commercial and recreational hook and line fisheries and bottom trawl fisheries. The RCAs protect almost 30% of rockfish habitats between 49 Vancouver Island and the mainland (inside waters) and approximately 15% of habitat on the rest of the coast (outside waters) (Yamanaka and Logan 2010). Inshore rockfishes include six species of the genus Sebastes (Copper Rockfish S. caurinus, Quillback Rockfish S. maliger, Black Rockfish S. melanops, China Rockfish S. nebulosus, Tiger Rockfish S. nigrocinctus, and Yelloweye Rockfish S. ruberrimus) that are found on shallow (<200m) rocky reefs. Numerous other fish species including Lingcod Ophiodon elongatus, and Kelp Greenling Hexagrammos decagrammus and the Greenstriped Rockfish S. elongatus are also protected in RCAs. Although the RCAs are often not considered to be MPAs because they are managed as fishery closures under the Fisheries Act as opposed to being permanently protected as MPAs by Canada’s Oceans Act (Robb et al. 2011), they are spatially explicit areas where fisheries that target or lead to substantial bycatch of rockfishes are prohibited. Marine Protected Areas may be an effective tool to conserve Pacific rockfishes because rockfishes are long-lived (some > 100 years), and have small home ranges (Yoklavich 1998, Parker et al. 2000). Marine Protected Areas have been effective for conserving rockfishes in California. Two marine reserves in California had significantly larger rockfishes and greater biomass (and therefore greater reproductive output) than non-reserve sites, while a 1-yr old reserve showed no difference from open areas (Paddack and Estes 2000). Five years after the Channel Islands marine reserve network was established, the biomass of targeted fish species, including five rockfish species, was approximately two times higher inside reserves than outside and targeted species biomass trajectories increased with time (Hamilton et al. 2010). Despite overall declining trends in US west coast groundfish CPUE in a fishery independent trawl survey, higher CPUE and larger fish were found for numerous rockfishes in American RCAs that 50 were continuously closed to trawling (Keller et al. 2014). Spatial fisheries closures may therefore be effective for promoting rockfish population recovery. The effectiveness of the RCA network in BC has yet to be evaluated in terms of population recovery. Two studies examined the performance of individual RCAs (Marliave and Challenger 2009, Cloutier 2011). Marliave and Challenger (2009) collected relative abundance and habitat data while SCUBA diving inside and outside of three RCAs in Howe Sound and concluded that the habitat model used to designate the RCAs had not included fine-scale habitat features with the highest rockfish abundances such as a boulder piles. They did not find any evidence of reserve effect in relative abundance of Copper or Quillback Rockfishes in the RCAs compared to open areas, but acknowledged that the early timing of the surveys constitute a baseline condition for comparison with future population trends. Cloutier (2011) completed SCUBA surveys between 8 and 15 m of depth at 15 sites in southern BC and evaluated RCA performance using rockfish presence and density. He found that RCAs had 1.6 times the rockfish density (all rockfish species pooled) than non-RCA sites while accounting for differences in habitat quality. SCUBA surveys are limited in that they can only assess rockfish species or lifestages within shallow depth ranges (<20m) that typically include Copper Rockfish and juvenile Quillback Rockfish (Richards 1987, Love et al. 2002). Quillback Rockfish, a threatened species (COSEWIC 2009), have been observed to 182 m in BC but are most abundant between 20 and 60 meters (Richards 1986). Yelloweye Rockfish, a species of Special Concern (SARA 2011), are caught between 20 and 250 m of depth (COSEWIC 2008) but highest densities have been observed between 40 and 100 m (Richards 1986). Quillback and Yelloweye Rockfishes are the two inshore rockfish species most heavily targeted by fisheries (Yamanaka and Logan 2010). 51 It is therefore important to use methods that can access deeper waters than SCUBA can to assess rockfish abundance in RCAs. Rockfish species segregate niches by habitat type as well as depth (Richards 1986, Matthews 1990a, Anderson and Yoklavich 2007, Laidig et al. 2009). The highest abundance of Quillback Rockfish in the Strait of Georgia was found in complex habitats above 60 m, while Yelloweye were found in similar habitats but in deeper water (Richards 1986). Yelloweye and Greenstriped Rockfish were found in similar depth ranges, but in different habitat types with Greenstriped Rockfish using fine-sediment habitats as opposed to rocky substrates (Richards 1986). The other inshore rockfishes are also associated with complex rocky habitats (Richards 1987, Matthews 1990b, Love et al. 2002). Complex living habitats such as sponge reefs are important rockfish habitats for both juvenile and adult fishes in BC (Cook et al. 2008, Marliave et al. 2009). Habitat is an important source of variability for fish communities; therefore, habitat structure must be accounted for in both the design and assessment of MPAs (Parnell et al. 2006, Claudet and Guidetti 2010, Miller and Russ 2014). Habitat variables should be collected as covariates with fish observations and included models of reserve effectiveness to determine if a positive response is related to protection or to intrinsic structural features of the protected area (Pelletier et al. 2008, Claudet and Guidetti 2010). Remotely Operated Vehicles (ROVs) are one of the most effective non-destructive monitoring tools to sample populations in protected areas (Field et al. 2006, Stoner et al. 2008). Size, operability, and cost of ROVs have all decreased in recent years. In addition to fish abundance and size data, visual surveys also have the ability to collect information on habitat use, behaviour and associations with other species (Yoklavich et al. 2002, Laidig et al. 2009, Love et al. 2009, O'Farrell et al. 2009). Stoner et al. (2008) contend that there is no better way to 52 monitor fishes in structurally complex habitats. ROV surveys targeting deeper-dwelling species were also undertaken in the Channel Islands within the first five years after establishment (Karpov et al. 2012). Karpov et al. (2012) compared the density of fishes on hard-bottom substrates between 20 and 100 m in depth among three pairs of sites inside and outside of reserves (two other site pairs were dropped). Results varied among sites and years; however, effects of protection on some rockfishes were apparent. The authors expected that reserve effects would become more pronounced with additional time for growth and recruitment (Karpov et al. 2012). In this study, I assess the effectiveness of the RCA network in promoting rockfish recovery in BC with ROV surveys. I focus on the following groundfish species: Quillback Rockfish, Yelloweye Rockfish, Greenstriped Rockfish, Lingcod and Kelp Greenling, as well as the inshore rockfish species in a combined group. I aim to address the three questions. 1. Is there any evidence of an increased density or body size of targeted or non-targeted groundfish in RCAs in BC? Species that are targeted by fisheries have been found to show stronger reserve effects than non-targeted species (Cote et al. 2001, Hamilton et al. 2010). If reserves significantly reduce mortality, species targeted by fisheries such as Lingcod, Yelloweye Rockfish and Quillback Rockfish should have higher densities inside reserves while non-targeted species such as Greenstriped Rockfish and Kelp Greenling should show no difference. 2. Does the age of the reserve influence the reserve effect? The effectiveness of MPAs is likely to vary spatially among species with different life histories and with the age of the reserve (Molloy et al. 2009, Babcock et al. 2010). The life-history characteristics that make rockfishes susceptible to overfishing, including slow growth, large size and old age at maturity, and episodic recruitment, also indicate that their recovery will be slow (Hutchings 2001, Dulvy et al. 53 2003, Dulvy et al. 2004, Hutchings and Reynolds 2004). The earliest we are likely to be able to detect any reserve effects for rockfishes is predicted to be between 5 to 10 years (Yoklavich 1998). Reserve effects may be more apparent earlier for fast growing species like Lingcod that are mature at 3 to 5 years (Cass et al. 1990). I expect that RCAs that have been in place for more than five years at time of sampling to show stronger reserve effects than younger RCAs. 3. Does adjacent fishing pressure influence the reserve effect? The intensity of fishing occurring outside of an MPA may also determine the strength of a reserve effect (Mosqueira et al. 2000, Claudet and Guidetti 2010). If fishing is also curtailed adjacent to an RCA, then a reserve effect may not be apparent due to a decreased effect size. In addition to establishing RCAs, Fisheries and Oceans Canada also decreased the fishing mortality in commercial and recreational fisheries by decreasing the total allowable catch and daily limits of the two fishing sectors, respectively. Commercial and recreational catch estimates can be used to measure the fishing effect size adjacent to conservation areas (Claudet and Guidetti 2010). I expect RCAs in areas of the coast with strong fishing pressure to show greater reserve effects than those with low adjacent fishing pressure. 3.2 Methods 3.2.1 Study Area We surveyed RCA on BC’s south coast. Three bodies of water are between Vancouver Island and the mainland: the Strait of Georgia, Johnstone Strait-Discovery Passage and Queen Charlotte Strait (Figure 3-1). These are collectively called “Inside Waters.” The Strait of Georgia has shallow depths, large fresh water inputs from the Fraser River, and numerous islands with high tidal flows between them. Johnstone Strait is characterized by rapid tidal streams and vigorous mixing, steep rocky walls and depths to 500 m. Numerous deep coastal fjords with 54 steep sides are found on the mainland side of Johnstone Strait. The Queen Charlotte Strait is a shallow, island-strewn basin with less intense tidal currents than Johnstone Strait except in passages between islands. The West Coast of Vancouver Island (WCVI) is part of “Outside Waters” along with BC’s north coast. The outer coast is characterized by a rocky shoreline and much greater wave exposure. The WCVI has five main sounds with numerous islands that lead into long coastal inlets (Thomson 1981). DFO manages the “Inside” and “Outside” inshore rockfish fishery separately (Yamanaka and Logan 2010). 3.2.2 Data Collection We surveyed 34 RCAs in four regions of BC (Strait of Georgia, Johnstone Strait, Queen Charlotte Strait, and the West Coast of Vancouver Island) with an ROV on seven research cruises between February 2009 and July 2011 (Figure 3-1). Because there are no comparable survey data from before the RCAs were established, this study employs the Control-Impact design (Underwood 1992) whereby data from inside the RCAs are compared to near-by sites that are open to fishing in order to infer reserve effects (Glasby 1997, Pelletier et al. 2008, Claudet and Guidetti 2010). I identified paired transects 300-900 m long using GIS targeting rockfish habitat inside and outside of RCAs based on depth contours using nautical charts or multibeam bathymetry. Most transect lines were perpendicular to the shore and the ROV traveled from deep to shallow, although transects in Johnstone Strait with very steep walls had to be run parallel to shore. The Saltspring Island N. RCA is adjacent to the Trincomali Passage RCA, so these were pooled for analysis. Similarly, the Dommett Point and Pam Rock, Halibut Bank and McCall Bank, Clio Channel and Viscount Island, and Octopus Island and Read Island RCAs are all close to each other and transects outside of the RCAs could be applied to either RCA. Therefore, I 55 pooled data from these RCAs for analysis, resulting in 30 comparisons of RCAs and unprotected sites in this analysis (Figure 3-1, Table 3-1). ROV surveys were conducted on the Canadian Coast Guard vessels the \"Vector\" and the \"Neocaligus\" using a Deep Ocean Engineering Phantom HD2+2 ROV with a 300 m umbilical. I deployed the ROV from the ship and temporarily fastened the umbilical to a wire with a 225 kg clump-weight to allow for greater control and operability of the ROV and to keep the ROV below the ship to improve ROV tracking. The position of the ROV was determined using “TrackPoint 3” ultra-short baseline acoustic tracking system (Ore International) and ROV and ship navigation data were recorded and mapped using Hypack 2009. The typical speed of the ROV was about 0.4 - 0.8kt, or 1/4 to 1/2 meter per second. Video was captured with a Sony EVI 300 Zoom video camera mounted on the ROV and data were recorded digitally onto hard drives and onto redundant Mini DV tapes. Two green lasers spaced ten cm apart mounted on the ROV were used to estimate the length of fishes as well as the width of the field of view which is taken to be the transect’s width (Haggarty et al. In Preparation). Following the surveys, I reviewed the videos and all fish, habitat, and field of view observations were recorded in a program specially designed for rockfish surveys (AVLog). Viewers counted all fish observed and identified them to the lowest taxonomic level possible. When the entire length of the fish was visible, viewers used a ruler to measure body length when it was in plane with the lasers, as well as the space between the lasers. I found the length of the fish using the ratio of the actual laser width (10 cm) to the laser width measured on screen (Haggarty et al. In Preparation). Fish length was categorized in 5 cm bins to account for the limited accuracy in this method. I found the field of view using the same method. Every 30 56 seconds, the reviewer recorded the distance between the lasers and the width of viewing window on screen and used the laser ratio to find the field of view. Habitat was recorded continuously. Each time the habitat changed, viewers paused the video to record a new habitat record. Primary substrate (substrate with 50% or more cover) was categorized in 9 classes that were subsequently lumped into three classes for analysis: rock, mixed coarse and fine. Similarly, 19 classes of biocover exist in the original dataset but these were grouped into 3 classes: bare, encrusting organisms (e.g. barnacles, tube worms, hydroids) and emergent organisms (e.g. Metridium anemones, sponges, sea pens). Habitat complexity and relief were also recorded in four classes which were re-classified into high and low. Complete habitat descriptions can be found in Haggarty et al. (in Preparation). The ROV transects were mapped in ArcGIS from track points downloaded from the acoustic tracking system. ROV tracking is subject to various errors including bottom characteristics such as steep rock walls that scatter or reflect acoustic energy and affect positional accuracy. Therefore, points that were clear outliers had to be eliminated or edited. Smoothed transect lines were then created from the edited points and calibrated with the date-time. The actual distance the ROV traveled over the ocean floor was then calculated using the z-field (depth) in the ArcGIS tool 3D Analyst to determine the surface length. Area-swept polygons were created using the field of view measurements and the buffer tool with ½ of the field of view measurement as the radius of the buffer. Habitat variables were assigned to each polygon. The date-time field associated with each track point links the field of view measurements, habitat and fish observations and allows them each to be mapped onto the transects. Fish observations from the video analysis were then mapped as points along each transect line. Data are all managed in ArcGIS geodatabases as well as in a master data base maintained at the Pacific Biological 57 Station (PACGFVideo). Complete ROV operations, video analysis and transect mapping methods, including Python Scripts used, are detailed in Haggarty et al. (in Preparation). 3.2.3 Analysis: 3.2.3.1 Fish Density We estimated the densities of the following groundfish species: Quillback Rockfish, Yelloweye Rockfish, Greenstriped Rockfish, Lingcod and Kelp Greenling. Because observations of inshore rockfish species other than Quillback and Yelloweye were less common, I pooled data from all inshore rockfish species, including Quillback and Yelloweye, and analyzed them as a species group. To calculate fish density I summed the number of fish observed, divided by the area surveyed on each transect and expressed it as the number per 100m2. Prior to analysis, I ln(x + 0.1) transformed the densities to normalize variance with the presence of zeros in the data (Pearse and A.H. 1987). I plotted the log density by region and level of protection and used Analysis of Variance (ANOVA) to determine if fish densities differed by region. 58 Figure 3-1. Rockfish Conservation Areas in southern BC. The RCAs that were sampled using an ROV between 2009 and 2011 are shown in green. 59 3.2.3.2 Effects of Protection and Habitat by Transect We constructed a linear mixed effects model (LME) using the package lmertest (Kuznetsova et al. 2014) in R (R Development Core Team 2008) with a nested design to test if fish density was dependent on protection status, habitat and depth. For this analysis, I included only data from the last sampling date for transects that I sampled more than once. I modeled the density of each species/species group separately and fit the final models using the restricted maximum likelihood (REML) (Zurr et al. 2009). Hamilton et al. (2011) found that biogeography was important to include in assessments of networks of protected areas because different regions studied have different physical and ecological characteristics. The regions I studied (the Strait of Georgia, Johnstone Strait, Queen Charlotte Strait, and the West Coast of Vancouver Island), have different physical and ecological characteristics, as well as separate exploitation histories and management. Therefore, I used the identity of the RCAs nested within Region as random variables. First I tested the effect of habitat and depth on fish density. To describe the habitat encountered on the transects, I combined three levels of substrate (rock, mixed coarse and mixed fine) and three levels of biocover (bare, encrusting and emergent organisms) resulting in nine biocover-substrate classes (Bare-Rock, Encrusted-Rock, Emergent-Rock, Bare-Mixed Coarse etc.). I calculated the percent of each transect covered by the nine habitat classes, as well as the percent high complexity and high relief. Prior to analysis, I used the logit transformation on the percentages (Fox and Weisberg 2011). I included all transformed habitat variables and the mean transect depth in a habitat model for each species. I used the package MuMIn (Barton 2013) and maximum likelihood estimation (Zurr et al. 2009) to find the optimum model structure using the second-order Akaike Information Criterion (AICc) for small sample sizes to rank the subsequent models. I found a subset of the models that explain 95% of the model weight and averaged them 60 together. Habitat variables that were significant (p<0.1), were retained and used in the next modeling step. Next, I included the level of protection (RCA or Open) in the model along with habitat variables that were found to influence fish density. Once again, I used MuMIn (Barton 2013) to find the optimum model structure and to perform model averaging on the models that explained 95% of the variance using the full model-averaged coefficients (i.e. with shrinkage). I calculated the relative variable importance (RVI) for each model. In order to determine if I sampled similar habitats inside and outside of RCAs, I plotted the mean and standard error of the percent habitat type and observed on transects and depth surveyed inside and outside of RCAs by Region and by RCA to see if the Standard Errors overlapped. To determine if I sampled the appropriate depth ranges of the fish species, I calculated the total area sampled over all of my transects by depth class and summed the number of fish of each species observed in each depth class. I used a Chi-squared analysis to determine if each species had a preference for certain depth classes by comparing the observed number per class to an expected number assuming no preference. I calculated the 95% confidence interval to determine which classes each species preferred and compared the preferred depth range of each species to the depths sampled. 3.2.3.3 Effects of Catch, Age, Area and Shape on RCA Effectiveness We analyzed data from RCAs to determine what factors influence effectiveness at promoting population recovery using the log response ratio (RR) (Equation 1) as a measure of reserve effectiveness (Hedges et al. 1999, Russ et al. 2005, Claudet et al. 2010, Hamilton et al. 2010, Edgar et al. 2014). 61 𝑅𝑅 = ln⁡(?̅?𝑖𝑛 + 1?̅?𝑜𝑢𝑡 + 1) (Equation 1) If the RR is greater than 0, the species is more abundant inside the reserve; while RRs less than zero indicate the species in more abundant outside the reserve. To control for the effect of habitat variability on species' densities, I used information on habitat-use by each species/group from the LME models to calculate a habitat-specific density. In ArcGIS, I selected the habitat features used by each species/group and then selected the fish observations that intersected with or were within 2 m of the preferred habitat. I calculated the area of the transect that included habitat for each species and excluded other habitats and fish observations from the habitat-based density calculations. Next I calculated the mean fish density on transects inside and outside of each RCA for each sampling trip and calculated a new RR. I plotted the mean and standard error (SE) of the Quillback, Yelloweye and Lingcod by RCA and the RRs for each species/group over all RCAs. I also plotted the log RR of each species/group versus the age of the RCA. To quantify the fishing history and current exploitation levels, I used the commercial catch data from the BC trawl and hook and line fisheries (data courtesy of Fisheries and Oceans Canada, N. Olson, Groundfish Data Unit). I used ArcGIS 10.2 to calculate the total commercial harvest weight (kg) for each of the groundfish species/groups in a buffer area 5 km-wide around each RCA and divided it by the area of the buffer for a total kg/km2. I used trawl data from 1996-2006 and hook and line data from 2002-2006 for the period before the RCAs were established in the RCA buffer as well as in the RCA because these were the years available with the required spatial and species resolution. Data from the period after establishement are from 2007-2011 for 62 both fisheries. I calculated the area (km2), and the area to open-water perimeter ratio of each RCA (i.e. perimeter of the RCA that abuts land was not included in the perimeter-area ratio) in ArcGIS 10.2. The age of the RCA at the time it was sampled was the number of years since establishment. I used a Linear Mixed Effects model of the RRs with Region as the random variable and the fixed effects: log catch + 1 in the RCA before establishment, log catch + 1 outside of the RCA before establishment, log catch + 1 outside of the RCA after establishment, RCA age, log RCA area, RCA perimeter-to-area ratio. I fit the model with REML (Zurr et al. 2009). 3.2.3.4 Length of Fish Inside to Outside of RCAs We compared the length frequency of Quillback, Yelloweye, Greenstriped, and Copper Rockfishes as well as Lingcod and Kelp Greenling found inside and outside of all RCAs pooled together using the Kolmogorov-Smirnov test and length frequency histograms. 3.3 Results 3.3.1 Fish Density We surveyed 199 transects inside 30 RCAs, and 166 outside (Table 3-1). Three RCAs were sampled twice, Brethour was sampled on three occasions and Northumberland was sampled four times. Thirteen of the surveyed RCAs are in the Strait of Georgia, 5 in Johnstone Strait, 5 in Queen Charlotte Strait and 7 in the WCVI. Rockfish Conservation Areas had been designated for between 3 and 7 years at the time of sampling (Appendix Table 2-1). Because the RCAs in the same region were established in the same year, and RCAs close to each other were sampled on the same cruise, RCA age is somewhat confounded with the regions. Quillback Rockfish were the most abundant species of interest observed on my transects (Table 3-1). The density of all of the fish species I studied except for the group of inshore 63 rockfishes differed among regions (Table 3-2). The pattern of abundance varied among species (Appendix Table 2-1). For instance, Quillback Rockfish were most abundant in Johnstone Strait while Lingcod were most abundant in WCVI (Table 3-1). 3.3.2 Effects of Protection and Habitat The LME model for each species revealed habitat features that had positive and negative effects on fish densities (Figure 3-2). The averaged LME model for each fish species/group explained significant variation in densities except for the Kelp Greenling model (Table 3-3). Protection status did not have a significant effect on density on transects for any species/group. The level of Protection (RCA/Open) had one of the lowest variable importance scores for all models. Of the habitat variables, rocky substrates with encrusting and emergent organisms and bare rock often had the highest densities of the species/group studied. Quillback Rockfish density was greatest on rocky substrates as well as mixed coarse and mixed fine substrates with encrusting organisms. Yelloweye Rockfish density was lower on mixed fine and mixed coarse substrates and not sensitive to bare rock. Although all rock categories, bare and encrusted mixed coarse, as well as fine encrusted substrates were retained in the habitat for Kelp Greenling, only encrusted rock was significant in the averaged model (Table 3-3). No substrate variables were retained in the Greenstriped Rockfish habitat model (Table 3-3, Figure 3-2). The Chi-square analysis showed that all fish species had a preferred depth range (Table 3-4). I sampled the preferred depth range of most of my fish species of interest. Ninety-two percent of the area sampled was between 25 and 125 m deep. Quillback, Tiger, Yelloweye and Greenstriped Rockfishes, and Lingcod were found within this depth range, although Yelloweye and Greenstriped Rockfishes were also distributed below this depth (Table 3-4). Kelp Greenling, Copper and China Rockfishes have shallower distributions and only 3% of the area surveyed was 64 above 25m. The Chi-square analysis results were corroborated by the LME model. Mean transect depth was retained in all of the LME models (Table 3-3, Figure 3-2). Depth was the most important variable in the Kelp Greenling and Greenstriped models and was also important in the Yelloweye model. Depth was the variable of lowest importance in the Quillback Rockfish, Lingcod and inshore rockfish models (Table 3-3, Figure 3-2). We successfully targeted similar habitats on transects inside and outside of most RCAs. The average habitat sampled and mean depths were very similar (Figure 3-3). Although habitat sampled was similar inside and outside most RCAs, some (i.e. D’Arcy Island), had different habitat inside vs. out (Appendix Figure 2-1). I sampled more rocky substrates in the RCA. D’Arcy Island is a very large RCA and I had difficulty finding nearby rocky habitat outside of the RCA. Some RCAs such as Halibut Banks, Ballenas I., Saltspring-Trincomali, and Thurston I. had relatively low percentage of rocky substrates. I sampled more rocky substrates outside of the Sarnac I. and Thurston I. RCAs (Appendix Figure 2-1). In order to eliminate the confounding effect of habitat differences in my analysis of the RCAs, I calculated habitat-based fish densities. For inshore rockfishes, Yelloweye Rockfish and Lingcod, I used rock substrates that were bare or had encrusting or emergent biocover (Table 3-3). Quillback Rockfish habitat consisted of all of the rocky substrates as well as mixed coarse and mixed fine with encrusting biocover, which likely indicates the presence of boulders (Table 3-3). Kelp Greenling also used all rocky substrates and bare and encrusting mixed-coarse substrate. Greenstriped Rockfish are limited by depth but no substrates were significant in the habitat LME (Table 3-3). I used all substrates along the transects but only depths greater than 40m because no Greenstriped Rockfish were observed above 40m. 65 Analyzing fish density only within the preferred habitat of each species changed the ranking of RR among the RCAs. For instance, D’Arcy I. had the highest RR using the total area of the transect, and Thurston I. had the lowest RR (Appendix Figure 2-2). When I adjusted fish density to account for habitat sampled, the RRs of these two RCAs were not different from zero (Figure 3-4). The mean habitat-adjusted RR of most (19) RCAs was not different from zero (the standard error bars overlap zero, Figure 3-4) indicating fish densities were not different inside to outside of the RCA. Eight RCAs had mean RRs less than zero, indicating more fish were observed outside and eight had mean RR’s greater than zero, such as Desolation Sound, Bolivar Pass and Prevost Island, indicating more fish were observed inside (Figure 3-4). No fish species/groups showed any difference in RR over all the RCAs (Figure 3-5). Kelp Greenling and Greenstriped Rockfish are not targeted by commercial fisheries and had the narrowest range of RRs (Figure 3-5). 3.3.3 Effects of Catch, Age, Area and Shape on RCA Effectiveness The age of the RCA did not affect the RR; the highest RR values were found in RCAs that had been in place for 5 years when they were sampled (Figure 3-6). The age of the RCA is, however, confounded with Region (Appendix Figure 2-4), with the oldest RCAs found in the WCVI which I sampled last. None of the factors tested in the LME explained any of the variation in RR among RCAs. Commercial catch outside of the RCA or inside the RCA before establishment, catch outside after RCA establishment, RCA age, RCA area, or perimeter-to-area ratio all failed to explain significant variability in RR (Table 3-5, Figure 3-7). Many RCAs, particularly in the Strait of Georgia, had no or very low levels of commercial catch of groundfish species outside of the RCA after RCA implementation (Appendix Figure 2-5). 66 None of the fish species showed any difference in length frequencies by protection status (Figure 3-8). Most of the Yelloweye Rockfish observed were juveniles (less than 50 cm, Figure 3-8) that haven’t yet reached 50% maturity (Table 1-1). 67 Table 3-1. The number of transects, mean and Standard Deviation (SD) of Fish Densities (#/100m2) inside and outside of RCAs observed on ROV surveys by region. SG=Strait of Georgia, JS=Johnstone Strait, QCST=Queen Charlotte Strait, WCVI=West Coast of Vancouver Island. The number of RCAs sampled per region is shown in parentheses. QB YE LC IRF GS KG Region RCA N SD SD SD SD SD SD SG In 122 0.57 0.68 0.10 0.18 0.12 0.19 0.79 0.84 0.26 0.40 0.17 0.27 (13) Out 81 0.70 0.71 0.12 0.21 0.13 0.20 0.89 0.82 0.28 0.38 0.16 0.22 JS In 13 1.01 1.34 0.05 0.10 0.08 0.15 1.06 1.33 0.19 0.19 0.02 0.05 (5) Out 15 1.27 1.14 0.13 0.16 0.06 0.15 1.40 1.09 0.29 0.44 0.04 0.10 QCST In 18 0.86 0.91 0.26 0.36 0.04 0.08 1.31 1.36 0.05 0.10 0.34 0.60 (5) Out 16 0.82 1.20 0.12 0.13 0.10 0.14 1.40 1.84 0.05 0.08 0.34 0.60 WCVI In 46 0.31 0.40 0.20 0.25 0.17 0.21 0.98 0.94 0.01 0.02 0.36 0.52 (7) Out 54 0.37 0.42 0.20 0.26 0.21 0.25 0.99 0.74 0.02 0.07 0.24 0.25 68 Table 3-2. Analysis of Variance Table for fish Density by Region and RCA. DF Sum of Sq. Mean Sq F P Quillback RCA 1 0.05 0.05 0.05 0.8 Region 3 32.3 10.8 12.4 <0.0001 Yelloweye RCA 1 0.2 0.2 0.38 0.5 Region 3 6.1 2 3.9 0.009 Inshore rockfishes RCA 1 0.3 0.3 0.3 0.6 Region 3 3.6 1.2 1..2 0.3 Lingcod RCA 1 0.2 0.2 0.5 0.5 Region 3 8.7 2.9 7.1 0.0001 Kelp Greenling RCA 1 0.1 0.1 0.2 0.6 Region 3 23.9 8 13.2 <0.0001 Greenstriped RCA 1 0.4 0.4 0.8 0.4 Region 3 38.3 12.8 23.2 <0.0001 69 Table 3-3. Linear Mixed Effects Model results of averaged model (with shrinkage) of habitat and RCA status on fish density (fish/100m2) on ROV transects (n=298). RVI=Relative Variable Importance. Term Estimate SE z-value p RVI Quillback Intercept Depth Rock- Bare Rock- Encrusted Rock- Emergent Coarse- Encrusted Fine- Encrusted RCA 1.45 -0.0002 0.17 0.20 0.25 0.14 0.10 -0.002 0.60 0.001 0.07 0.04 0.04 0.07 0.1 0.02 2.42 1.19 2.3 5.25 5.87 1.97 1.01 0.07 0.02 0.85 0.02 <0.0001 <0.0001 0.05 0.31 0.94 0.04 0.92 1 1 0.88 0.58 0.06 Yelloweye Intercept Depth Rock- Bare Rock- Encrusted Rock- Emergent Coarse- Bare Fine- Encrusted RCA -1.49 0.005 0.0006 0.19 0.12 -0.02 -0.003 -0.004 0.28 0.003 0.007 0.03 0.03 0.04 0.02 0.02 5.24 1.62 0.08 6.61 4.01 0.5 0.18 0.17 <0.0001 0.11 0.94 <0.0001 0.0002 0.62 0.85 0.86 0.77 0.02 1 1 0.24 0.06 0.07 Inshore rockfishes Intercept Depth Rock- Bare Rock- Encrusted Rock- Emergent Coarse- Bare Complexity RCA 1.31 -0.00001 0.016 0.29 0.25 -0.02 0.05 -0.01 0.33 0.0001 0.007 0.04 0.04 0.04 0.06 0.07 3.97 0.116 2.17 6.86 5.81 0.44 0.71 0.25 <0.0001 0.91 0.03 <0.0001 <0.0001 0.66 0.48 0.81 0.02 0.90 1 1 0.20 0.39 0.10 70 Table 3-4 (continued). Linear Mixed Effects Model results of averaged model (with shrinkage) of habitat and RCA status on fish density (fish/100m2) on ROV transects (n=298). RVI=Relative Variable Importance. Term Estimate SE z-value p RVI Lingcod Intercept Depth Rock- Bare Rock- Encrusted Rock- Emergent Coarse- Bare Relief RCA -1.07 -0.0001 0.16 0.15 0.05 -0.01 -0.05 0.0002 0.28 0.001 0.06 0.05 0.06 0.02 0.07 0.01 3.82 0.001 2.75 3.32 0.77 0.28 0.67 0.02 0.0001 0.17 0.006 0.001 0.44 0.78 0.5 0.98 0.03 0.97 1 0.43 0.10 0.36 0.04 Kelp Greenling Intercept Depth Rock- Bare Rock- Encrusted Rock- Emergent Coarse- Bare Coarse- Encrusted Fine- Encrusted Complexity RCA -0.25 -0.01 0.05 0.13 0.02 0.004 0.01 -0.04 -0.01 -0.01 0.41 0.002 0.06 0.03 0.04 0.02 0.02 0.06 0.02 0.04 0.60 8.86 0.8 4.30 0.58 0.23 0.23 0.60 0.26 0.27 0.55 <0.0001 0.42 <0.0001 0.56 0.82 0.76 0.55 0.80 0.79 1 0.45 1 0.30 0.07 0.11 0.31 0.08 0.10 Greenstriped Intercept Depth Complexity Relief RCA -2.26 0.01 0.002 -0.002 -0.004 0.22 0.001 0.01 0.01 0.02 10.0 4.46 0.16 0.16 0.20 <0.0001 <0.0001 0.88 0.84 0.85 1 0.04 0.06 0.07 71 Figure 3-2. The % Relative Variable Importance (RVI), estimate and 95% Confidence Intervals of habitat and protection variables retained in the Linear Mixed Effects Averaged Model.1=Bare, 2=Encrusted, 3=Emergent Biocover. 72 Figure 3-3. The mean and SE of the percent of habitat type by region on transects inside and outside of the RCAs. R1=Bare Rock, R2=Rock with encrusting biota, R3=Rock with emergent biota; C1=Bare Coarse substrates, C2=Coarse with encrusting biota, C3=Coarse with emergent biota; F1=Bare Fine substrate, F2=Fine with encrusting biota, F3=Fine with emergent biota; CX=High Complexity, RL=High Relief, D=Depth (in m). 73 Table 3-5. The proportion depth classes sampled on 420 ROV transects and the expected and observed proportion of fish species and 95% Confidence Intervals by depth class. Numbers in bold indicate the preferred depth zones of each species. A Chi-squared analysis indicated that species are not distributed evenly across depth classes. Depth Class (m) 0-25 26-50 51-75 76-100 101-125 126-150 151-200 Habitat Area (m2) 16,517 173,933 160,382 87,062 36,388 13,280 8,260 Proportion by Area 0.03 0.35 0.32 0.18 0.07 0.03 0.02 Copper N=349, χ2=312, df=6, p=<0.0001 Expected 0.03 0.35 0.32 0.18 0.07 0.03 0.02 Upper CI 0.18 0.68 0.23 0.04 0.01 0.01 0.00 Observed 0.15 0.63 0.19 0.03 0.00 0.00 0.00 Lower CI 0.11 0.58 0.15 0.01 0.00 0.00 0.00 China N=530, χ2=347, df=6, p=<0.0001 Expected 0.03 0.35 0.32 0.18 0.07 0.03 0.02 Upper CI 0.11 0.70 0.29 0.01 0.00 0.00 0.00 Observed 0.09 0.65 0.25 0.00 0.00 0.00 0.00 Lower CI 0.06 0.61 0.22 0.00 0.00 0.00 0.00 Kelp Greenling N=1151, χ2=809, df=6, p=<0.0001 Expected 0.03 0.35 0.32 0.18 0.07 0.03 0.02 Upper CI 0.12 0.69 0.22 0.04 0.00 0.00 0.00 Observed 0.10 0.67 0.20 0.03 0.00 0.00 0.00 Lower CI 0.08 0.64 0.18 0.02 0.00 0.00 0.00 Quillback N=2995, χ2=220, p=<0.0001 Expected 0.03 0.35 0.32 0.18 0.07 0.03 0.02 Upper CI 0.04 0.47 0.34 0.16 0.04 0.01 0.003 Observed 0.03 0.45 0.32 0.15 0.04 0.01 0.002 Lower CI 0.03 0.43 0.31 0.13 0.03 0.01 0.000 Lingcod N=869, χ2=66, df=6, p=<0.0001 Expected 0.03 0.35 0.32 0.18 0.07 0.03 0.02 Upper CI 0.037 0.47 0.36 0.184 0.06 0.003 0.006 Observed 0.026 0.44 0.33 0.159 0.04 0.001 0.002 Lower CI 0.016 0.41 0.30 0.134 0.03 -0.001 -0.001 Tiger N=117, χ2=26, p=0.0002 Expected 0.03 0.35 0.32 0.18 0.07 0.03 0.02 Upper CI 0.04 0.33 0.50 0.32 0.07 0.02 0.00 Observed 0.02 0.26 0.42 0.25 0.04 0.01 0.00 Lower CI 0.00 0.19 0.35 0.19 0.01 -0.01 0.00 Yelloweye N=793, χ2=46, df=6, p=<0.0001 Expected 0.03 0.35 0.32 0.18 0.07 0.03 0.02 Upper CI 0.02 0.30 0.44 0.24 0.09 0.03 0.02 Observed 0.01 0.27 0.40 0.21 0.07 0.02 0.01 Lower CI 0.01 0.24 0.37 0.18 0.05 0.01 0.01 Greenstriped N=700, χ2=767, df=6, p=<0.0001 Expected 0.03 0.35 0.32 0.18 0.07 0.03 0.02 Upper CI 0.00 0.01 0.30 0.41 0.26 0.10 0.05 Observed 0.00 0.00 0.27 0.37 0.23 0.08 0.04 Lower CI 0.00 0.00 0.24 0.34 0.20 0.06 0.02 74 Figure 3-4. The mean log response ratio of Quillback, Yelloweye and Lingcod of habitat-based density inside to outside of all RCAs sampled. Error bars are standard errors. Ratios greater than zero indicate greater densities inside the RCA. 75 Figure 3-5. Log Response Ratio (RR) for targeted and non-targeted fish species in RCAs. QB=Quillback, YE=Yelloweye, IRF=Inshore rockfishes, LC=Lingcod, KG=Kelp Greenling, GS=Greenstriped Rockfish. 76 Figure 3-6. Boxplots of the log Response Ratio (RR) by Species/Species Group and by the years of protection at the time of sampling. 77 Table 3-6. Linear Mixed Effects model results of factors affecting Response Ratio with Region as a random variable (n=38). Term Estimate Std. Error DF t-value p Quillback Intercept Log Catch RCA Log Catch Out-B Log Catch Out-A Age Log Area PA 0.02 -0.02 0.04 -0.02 -0.01 0.01 0.01 0.25 0.04 0.05 0.03 0.03 0.05 0.11 31 31 31 31 31 31 31 0.09 -0.36 0.80 -0.63 -0.27 0.19 0.08 0.93 0.72 0.43 0.53 0.79 0.85 0.94 Yelloweye Intercept Log Catch RCA Log Catch Out-B Log Catch Out-A Age Log Area PA -0.03 -0.03 0.02 -0.03 -0.01 0.03 0.15 0.30 0.12 0.16 0.14 0.03 0.07 0.14 31 31 31 31 31 31 31 -0.08 -0.23 0.14 -0.22 -0.45 0.42 1.09 0.93 0.82 0.89 0.83 0.66 0.68 0.29 Inshore Rockfishes Intercept Log Catch RCA Log Catch Out-B Log Catch Out-A Age Log Area PA 1.00 0.08 0.001 -0.06 -0.05 -0.18 -0.03 0.58 0.06 0.07 0.04 0.07 0.09 0.20 31 31 31 31 31 31 31 1.74 1.31 0.01 -1.26 -0.74 -2.08 -0.14 0.11 0.20 0.99 0.22 0.47 0.05 0.89 Lingcod Intercept Log Catch RCA Log Catch Out-B Log Catch Out-A Age Log Area PA -0.11 0.01 0.02 -0.03 -0.04 0.08 0.17 0.31 0.05 0.06 0.05 0.04 0.06 0.13 31 31 31 31 31 31 31 -0.36 0.26 0.30 -0.61 -1.24 1.25 1.30 0.72 0.80 0.77 0.55 0.22 0.22 0.21 Kelp Greenling Intercept Log Catch RCA Log Catch Out-B Log Catch Out-A Age Log Area PA -0.17 -0.05 -0.02 0.07 -0.02 0.06 0.07 0.14 0.05 0.06 0.05 0.01 0.03 0.06 31 31 31 31 31 31 31 -1.28 -1.15 -0.28 1.39 -1.07 1.91 1.11 0.21 0.26 0.78 0.17 0.29 0.07 0.27 Greenstriped Intercept Log Catch RCA Log Catch Out-B Log Catch Out-A Age Log Area PA -0.14 0.37 0.16 -0.17 0.01 0.03 0.02 0.16 1.07 0.49 0.51 0.02 0.04 0.07 31 31 31 31 31 31 31 -0.88 0.34 0.32 -0.34 0.45 0.74 0.23 0.39 0.73 0.75 0.74 0.66 0.47 0.82 78 Figure 3-7. Log Response Ratio (RR) versus log reserve area (km2), perimeter-to-area ratio, and the Log+1 Catch (kg) per km2 in a 5 km wide buffer around each RCA. 79 Figure 3-8. Histograms of the length (cm) of fish inside and outside of the RCAs and results of the Kolmogorov-Smirnoff test. Note that the bars for the level of protection are plotted beside rather than on top of each other. There is no significant difference in the length frequency between fish observed inside and outside of RCAs. 80 3.4 Discussion My survey provides little indication that recovery is underway for demersal fish populations inside BC's Rockfish Conservation Areas three to seven years after their establishment. Fish numbers and sizes were equivalent in visual surveys of most protected and reference sites. Different species showed clear habitat affinities, and inclusion of protection status added no explanatory power in a model to predict fish densities based on depth and bottom cover types. These results have a number of implications for spatial fisheries closures as a management tool for promoting recovery of rockfishes and other groundfish. First, the level of protection afforded by the RCA network may be inadequate to allow populations to recover if illegal fishing or fishing that is allowed within the boundaries of RCAs continues to impose high mortality. Alternatively, sufficient time may not have elapsed for growth and recruitment to replenish depleted stocks inside RCAs. Ongoing visual population assessments are necessary to distinguish these two possibilities, and my data provide a baseline for future comparisons. ROV surveys are an effective way to monitor inshore rockfishes in RCAs. Unlike longline or other fishing surveys, they concurrently collect habitat and fish data and are non-extractive. Surveys do, however, require great technical expertise to run successful surveys and to record data from videos and processes tracking data after the surveys. Depth and habitat were the only variables that explained fish densities in my models. Other studies have shown that rockfishes and life history stages partition habitats by depth and by substrate type (Richards 1986, 1987, Matthews 1990a, b, Yoklavich et al. 2000, Love et al. 2002, Ingram and Shurin 2009). Although depth explained significant variation in the LME model of all species/group the estimate was often close to zero. Fish species often do not show linear relationships with depth, but instead show a humped or skewed distribution centered on 81 their preferred depth range (Richards 1986). Depth was not a very important variable for predicting densities of Quillback Rockfish, Lingcod, or the combined inshore rockfishes group, indicating that these species were evenly distributed across the depth range I sampled. Depth was an important predictor for Kelp Greenling, which have a shallower depth distribution and Yelloweye and Greenstriped Rockfishes that have deeper distributions (Richards 1986, Love et al. 2009). Depth is, therefore, an important covariate in the analysis of RCAs. Habitat features such as substrate type also influenced fish densities. Unlike many habitat characterization schemes that record the two most common substrate types observed e.g.(Love et al. 2009), I only recorded the primary substrate type. However, I also recorded the primary biological cover. Rocky substrates with encrusted or emergent organisms were the most used habitat types by my species of interest. Although this might point to important associations between fish and sessile invertebrates, the presence of certain types of organisms may also indicate the presence of boulders or more complex habitats that many rockfishes prefer (Love et al. 2002). Yelloweye Rockfish density was not associated with bare rock and the biological cover may have indicated more complex boulder habitats that they often prefer (Richards 1986, Oconnell and Carlile 1993). Alternatively, the relationship with biocover types and Yelloweye Rockfish might be a result of differences in the amount of bare rock sampled in the different regions. I sampled more bare rock in the Strait of Georgia, which also had lower Yelloweye Rockfish abundance, than in other regions. Most rocky substrates I sampled on the WCVI and the Queen Charlotte Strait, where Yelloweye Rockfish densities were greatest, were covered by invertebrates. As such, I included bare rock as a habitat type used by Yelloweye Rockfish because they were associated with bare rock in the Strait of Georgia. Encrusting biocover on mixed-coarse and fine substrates probably indicated the presence of boulders as a secondary 82 substrate type. This would explain the importance of these habitats to Quillback Rockfish. I recommend that the secondary substrate be recorded on future surveys. I averaged habitat characteristics over the entire transect; therefore, I likely missed finer-scale habitat associations of all species. My ROV data could also be analyzed for finer-scale use of habitat patches (i.e. Anderson and Yoklavich 2007) including associations with particular types of biocover. My study highlights the importance of collecting fish habitat descriptors concurrently with abundance data in order to assess protected areas using the control-impact method to compare data from within protected areas with open areas (Claudet and Guidetti 2010). Miller and Russ (2014) cautioned that the evaluation of MPAs must include an evaluation of habitat effects, particularly when data from before reserve implementation are not available. Despite my efforts to sample similar habitats inside and outside of reserves, my RRs changed when I accounted for fish habitat encountered along the transects. For instance, I could have concluded a positive reserve effect for the D’Arcy Island RCA; however, differences in fish abundance inside to outside the reserve were a result of better habitat sampled inside the RCA. Other studies have found that controlling for habitat altered the assessment of MPAs on fish densities (Chapman and Kramer 1999, Miller and Russ 2014). I sampled poorer habitat inside some RCAs. When transects that did not sample any appropriate habitat were excluded from analysis, the area of habitat sampled in some RCAs, such as Halibut Bank and Bedwell Sound, was very small. If the habitat I sampled with the ROV is representative of the entire RCA, then some RCAs contain very little high quality rockfish habitat. Although I did not exhaustively sample each RCA or randomly stratify my ROV surveys to assess habitat throughout the RCA, I did target rocky reef habitat both inside and outside of the RCA. It was easier to locate rocky habitat in some RCAs than others. Ensuring appropriate habitat for the target species is represented is one of the most 83 important criteria for MPA effectiveness (Parnell et al. 2006). Habitat in each RCA should be thoroughly assessed. In addition to the amount and quality of habitat, habitat distribution with respect to the boundary of the RCAs should be assessed because spillover from seasonal and day-to-day fish movements across reserve boundaries also affects reserve success (Kramer and Chapman 1999, Babcock et al. 2010, Edgar et al. 2014). Reserve status did not influence the density of any of the groundfish species or the inshore rockfish group collectively. Although the mean response ratio (RR) of the groundfish species I studied for some RCAs was greater than zero, on average RRs were not different from zero, and more fish were observed outside of some RCAs. Furthermore, none of the features I analyzed explained variation in RR among the RCAs or among species, although I may not have had the power to detect significant effects with a sample size of 38 RCAs. Several factors may explain the lack of reserve effects (Hamilton et al. 2010, Keller et al. 2014). First, the RCAs were still relatively young when they were surveyed (three to seven years). In a global meta-analysis of MPA surveys, Babcock et al. (2010) determined time to first detection of a reserve effect on target species was 5.13 ±1.9 years, a surprisingly short time given that many target species shared life history characteristics of longevity and slow growth. Although reserve effects were also found after only five years for some species of rockfishes in California (Hamilton et al. 2010), the rockfish species that responded quickly are relatively short lived (30-44 years) and mature early (4-5 years) (Love et al. 2002). In contrast, Quillback and Yelloweye Rockfishes live to be 95 and 115 years old and 50% maturity isn’t reached until 11 and 15-20 years of age, respectively (COSEWIC 2008, 2009). Few Yelloweye Rockfish observed in this study had reached 50% maturity, indicating that recovery for this species will be particularly slow. The much longer generation time for these species as well as the sporadic recruitment success of 84 rockfish, implies that it will take longer than 3 to 7 years for reserve effects to develop in RCAs as a result of increased growth and reproductive output of these species in RCAs. Similarly, Starr et al. (2015) found strong reserve effects in an MPA that had been closed since 1973 but not in newer reserves in California that were sampled within the first 7 years of protection. They projected that MPAs in Central California may take 20 years or more to show significant changes in response variables (Starr et al. 2015). Most rockfishes distributed along the continental slope did not show a greater proportion of large fish in closed areas of a large RCA in the US, unlike other species with shorter lifespans (Keller et al. 2014). An ROV study of deeper waters of the Channel Islands marine reserve also failed to find significant reserve effects for many species within the first 5 years of protection (Karpov et al. 2012). Although the time since protection of the RCAs varied between 3 and 7 years, I did not find a trend with my RRs and the age of the RCA. The oldest reserves were found in the WCVI region, therefore the age effect was confounded with region. Indirect effects through trophic interactions are also possible in MPAs, although they typically take longer to be detected (Babcock et al. 2010). Lingcod populations in RCAs may increase faster than rockfish populations would as a result of Lingcod’s shorter lifespan (15-20) and earlier age at maturity (2-5 years) (Cass et al. 1990). Higher catch rates and larger Lingcod were found in reserves in the San Juan Islands that had been in place for around 20 years (Beaudreau 2009); however, reserves must be large enough to provide year-round protection of Lingcod from anglers (Martell et al. 2000). It has been hypothesized that increased Lingcod abundance in conservation areas would prevent rockfish from recovering as a result of increased Lingcod predation (Beaudreau and Essington 2007, Tinus 2012). I have no evidence that Lingcod are more abundant or larger in the RCAs. Furthermore, two studies of Lingcod diet 85 preference have shown that rockfishes make up a small proportion of Lingcod’s diet and among rockfishes, the small-bodied Puget Sound Rockfish (S. emphaeus) is the primary prey item (Beaudreau and Essington 2007, Tinus 2012). It is therefore unlikely that predation by Lingcod is hindering the recovery of rockfishes in RCAs. Species that are directly targeted by fisheries typically respond to protection more quickly to protection than species that are not directly targeted (Molloy et al. 2009, Babcock et al. 2010, Claudet et al. 2010, Hamilton et al. 2010). Quillback and Yelloweye Rockfishes, Lingcod and the combined inshore rockfish group are all targeted by commercial and recreational fisheries in BC. Greenstriped Rockfish and Kelp Greenling are also taken as bycatch in both commercial and recreational fisheries; however, they are typically not targeted and the catch of these species is much lower than the other groundfish species I surveyed (Figure 3-7). Although none of the species/groups mean RR differed from zero, the mean RR for Greenstriped Rockfish and Kelp Greenling fell right on zero. The intensity of commercial fishing outside of the RCAs did not influence the RR. In order to improve the assessment of marine reserves, Claudet and Guidetti (2010) recommend that the actual fishing pressure outside of an MPA be quantified, rather than assumed when the assessment of effectiveness is assessed relative to external controls. Other management measures, coincident with the creation of conservation areas complicate the evaluation of RCAs (Keller et al. 2014, Starr et al. 2015). In addition to establishing RCAs, DFO also greatly reduced the directed catch (Total Allowable Catch, TAC) for inshore rockfishes. Commercial TAC was reduced by 50% for outside populations and 75% for inside populations, and recreational limits decreased from five to one fish per day on the inside and from five to three fish on the WCVI (Yamanaka and Logan 2010). In addition to the changes associated with the 86 Rockfish Conservation Strategy, all of the commercial groundfish fisheries in BC were integrated under the Commercial Groundfish Initiative. A guiding principle for this initiative was to account for all rockfish catch (targeted and bycatch) and individual transferable quotas (ITQs) and 100% at sea observer or electronic monitoring were instated (Davis 2008). As a result of the decline in commercial rockfish catch and effort, I measured the catch for my species of interest around each RCA and hypothesized it would greatly influence the RR. Our analysis did show that areas open to fishing adjacent to many RCAs, particularly in the Strait of Georgia, are currently not fished. However, the RR varies greatly even among RCAs that are not fished and the level of catch did not influence the RR for any of my species. Catch inside or outside the RCA from before they were established was also unrelated to RR; however, the time-period of the historical data I used (1996-2006) likely did not adequately represent the exploitation history because landings had already declined considerably by 1990 (Yamanaka and Logan 2010).We did not consider recreational catches in this study; however, recreational non-compliance may also impede population recovery inside the RCAs (Marliave and Challenger 2009, Haggarty et al. Accepted). Although commercial compliance is known to be high, recreational compliance has not yet been assessed. Compliance with regulations is a key feature that influences the effectiveness of MPAs, and illegal fishing may be preventing population recovery inside RCAs (Edgar et al. 2014). The area and shape of MPAs are other design criteria that may affect reserve effectiveness (Halpern 2003, Gaines et al. 2010a); however empirical evidence is mixed (White et al. 2011). I found no relationship between RR and RCA size or perimeter-to-area ratio. Rockfish Conservation Areas effectiveness may be determined by a number of potentially interacting factors that may be difficult to detect by regression analyses. Identifying the factors 87 that influence reserve success is, however, critical in order to inform adaptive management to improve the effectiveness of future actions (White et al. 2011). A thorough analysis of the habitat quality, habitat isolation, compliance and other key features of the entire RCA network are badly needed. Continued monitoring of the RCAs is essential given the demographic time-lag associated with long-lived rockfishes. Our study shows that ROV surveys are an effective way to monitor RCAs and my data can be used to represent the initial conditions of the RCAs and be compared to in the future. At present, I see little indications of population recovery inside RCAs. 88 Chapter 4: Lack of Recreational Fishing Compliance May Compromise Effectiveness of Rockfish Conservation Areas in British Columbia 4.1 Introduction Jurisdictions around the world are increasingly using spatial management (e.g., marine reserves, fishing closures) to conserve and restore overfished populations (Yoklavich 1998, Parker et al. 2000, Hamilton et al. 2010, Yamanaka and Logan 2010). The effectiveness of marine reserves and marine protected areas (MPAs) depends on reducing or eliminating fishing pressure within their boundaries (Kritzer 2004, Sethi and Hilborn 2008, Gaines et al. 2010a, Ainsworth et al. 2012). Reserves with high rates of non-compliance show limited recovery of fish communities (Kritzer 2004, McClanahan et al. 2009, Ainsworth et al. 2012, Campbell et al. 2012), and compliance information can predict reserve response ratio of fish biomass without incorporating any reserve design factors (Bergseth et al. 2013). Furthermore, a global study of marine reserves identified enforcement as one of five key features that influenced effectiveness (Edgar et al. 2014). Similarly, compliance levels reported by resource users and population density were the best predictors of reserve effectiveness in a study of 56 tropical marine reserves (Pollnac et al. 2010). However, despite their importance for recovery, compliance rates are rarely quantified (Sethi and Hilborn 2008, Smallwood and Beckley 2012), and when compliance is measured, most data are qualitative (anecdotal or expert opinion) and direct empirical observations are rare (Bergseth et al. 2013). Non-compliance is prevalent where it has been studied. For example, Smallwood and Beckley (2012) found that 8 to 12% of recreational boats were fishing in closed zones in Australian MPAs. Ten percent of recreational fishers involved in a questionnaire using the 89 random response technique in Australia admitted to fishing in closed areas (Arias and Sutton 2013). Williamson et al. (2014) used the density of derelict fishing line as a proxy of recreational compliance in no-take zones in the Great Barrier Reef Marine Park and found that no take areas had 30% of the fishing effort of surrounding open areas (Williamson et al. 2014). Compliance can also change over time. For instance, in Mexico, travel to marine reserves by fishermen declined shortly after a reserve was implemented, but compliance declined within four years as fishermen learned that there was no enforcement (Fujitani et al. 2012). The environmental and social drivers that influence compliance in marine reserves are a critical research frontier for the implementation of more effective MPAs (Bergseth et al. 2013). Reserve design features such as size and shape may influence compliance (Kritzer 2004, Gaines et al. 2010a, Read et al. 2011). Fishing is often concentrated around the edge of a reserve; therefore larger MPAs with lower perimeter-to-area ratios should have lower fishing rates due to lower straying rates along their perimeters (Read et al. 2011). The location of reserves may also influence compliance as fishing often decreases with the distance to port and is greater with closer proximity to towns, fishing piers and boat ramps (Stelzenmüller et al. 2008, Read et al. 2011). Involving stakeholders in the planning process of conservation areas may also affect compliance and enhance MPA effectiveness (Pollnac et al. 2010). However, few studies have actually tested these assumptions with empirical measurements of compliance (Pollnac et al. 2010, Bergseth et al. 2013). Here I examine recreational fishing compliance in Rockfish Conservation Areas (RCAs), in British Columbia (BC), Canada. In response to conservation concerns associated with a sharp decline in inshore rockfish catches throughout the 1990s in the Northeast Pacific, Fisheries and Oceans Canada (DFO) implemented a system of 164 Rockfish Conservation Areas. Rockfish 90 Conservation Areas were established between 2004 and 2007 and prohibit commercial and recreational hook and line fisheries and bottom trawl fisheries (Yamanaka and Logan 2010). The BC commercial fishery exhibits good compliance with the RCAs as a result of onboard and electronic fishery monitoring by Global Positioning System (GPS) technology and observers (unpublished data, N. Olsen, Fisheries and Oceans Canada). Compliance by the recreational fishery has yet to be assessed, although anecdotal observations indicate that fishing in closed areas persists and that recreational compliance might be low (Marliave and Challenger 2009). The purpose of my study was to assess and identify drivers of recreational compliance with RCAs. I aimed to answer two questions: 1. Has the spatial pattern of recreational fishing effort changed in the RCAs since they were established? and 2. What geographic factors affect variation in compliance among RCAs? 4.2 Methods 4.2.1 Study Area Context Rockfish Conservation Areas were implemented in BC as part of a larger Rockfish Conservation Strategy. Inshore rockfishes include six species of the genus Sebastes (S. caurinus, S. maliger, S. melanops, S. nebulosus, S. nigrocinctus, and S. ruberrimus) that are found on shallow (>200m) rocky reefs. Although RCAs are not considered MPAs because they are fisheries closures rather than permanently legislated protected areas (Robb et al. 2011), they do prohibit commercial and recreational fisheries that target or lead to a significant bycatch of rockfishes (Yamanaka and Logan 2010). Recreational (e.g., through the Sport Fish Advisory Board) and commercial fishers as well as conservationists, First Nations and the public were consulted in the designation of the RCAs (Yamanaka and Logan 2010). Although the protection targets were ambitious – 30% of rockfish habitat in “inside” waters east of Vancouver Island, 91 and 20% of outer coast habitats – attempts were made to minimize socio-economic impacts on recreational and salmon troll fisheries. Other considerations in the designation of the RCAs included rockfish habitat, the ease of monitoring and enforcement (Yamanaka and Logan 2010). Rockfishes are targeted by recreational fishers and caught as bycatch while angling for salmon or bottom fish such as Halibut and Lingcod (Zetterberg et al. 2012b). The contribution of recreational fisheries to rockfish mortality varies by region. In 2011, recreational fishers caught 90% of the estimated 35,000 inshore rockfishes caught in the Strait of Georgia by commercial and recreational fisheries, compared to 35% of 60,000 fish and 8% of 93,000 on the west coast and northeast coast of Vancouver Island, respectively (recreational data from D. O’Brien, Fisheries and Oceans Canada; commercial data from N. Olsen, Fisheries and Oceans Canada). In addition to designating RCAs, Fisheries and Oceans Canada reduced recreational daily bag limits from ten to five rockfish in outside waters and from five to one rockfish on the inside as part of the Rockfish Conservation Strategy (Yamanaka and Logan 2010). I focused my analysis on Vancouver Island and the Strait of Georgia. Vancouver Island is separated from the mainland coast by the inland waters of the Strait of Georgia, Johnstone Strait and Queen Charlotte Strait, locally called “inside” waters. British Columbia’s two largest population centers, Vancouver and Victoria, as well as numerous smaller towns border the Strait of Georgia. The northeast and west coasts of Vancouver Island are much more sparsely populated but have numerous recreational fishing lodges. One hundred and forty four of the 164 RCAs are found in southern BC and 129 of these are in inside waters (Figure 4-1). 4.2.2 Data Collection and Preparation Fisheries and Oceans Canada monitors the marine recreational fishery in BC using Creel Surveys (English et al. 2002). The creel survey has two components: (1) dockside interviews, 92 where fishers are asked where they fished, what they caught, kept and released, and how long they fished; and (2) effort counts via aerial surveys along pre-defined flight paths timed to cover major periods of fishing activity. Planes fly at an altitude of 150-300 m to allow for a broad range of vision and easy identification of vessel type and activity. Between 6 and 10 flights per month are completed during the peak fishing season (Zetterberg et al. 2012a). An observer counts all boats and uses binoculars to determine if they are actively fishing (lines in the water) or not (traveling or engaged in other activities such as trap fishing). The observer marks the estimated geographic location of the boats on maps of the study area. The number of boats are summed by management area and used in conjunction with dockside interviews to estimate the total sport-fish effort and the number of salmon and groundfish caught in the sport fishery (i.e. Zetterberg, Watson et al. 2012). However, the data have not previously been digitized or geo-referenced with the RCA network to assess compliance. We used the spatial information from the creel survey over-flights to assess recreational compliance in the RCAs. I digitized and georeferenced creel survey maps for the northern and southern Strait of Georgia, Queen Charlotte Sound and the northwest coast of Vancouver Island using major coastal features and management area boundaries in ArcGIS 10.1 (ESRI 2011). I recorded the root mean squared (RMS) error (ESRI 2008), a measure of the difference between known locations and points that have been interpolated or digitized, for each image. The fishing observations were then manually digitized for each survey date between May and September, when 85% of the sport-fish effort in BC occurs (Zetterberg et al. 2012b). To avoid bias, RCAs were not displayed in the GIS project while I digitized fishing observations. I digitized data for the Strait of Georgia for 2003, 2007 and 2011, while data for the northeast and west coasts of Vancouver Island were only available for 2011. Only RCAs that fall within view of the survey 93 route were used in my analyses (Figure 4-1). The Desolation Sound RCA was dropped from the temporal analysis because the flight pattern near it changed between 2007 and 2011. 4.2.3 Temporal Analysis We used data from 2003 to represent fishing effort prior to establishment because the RCAs were designated between 2004 and 2006. The creel survey estimates of effort for 2003, 156,670 boat trips, was equal to the median value from 1999-2003 (Zetterberg et al. 2012b, In Preparation). The first year all RCAs were in place was 2007 and 2011 is at least five years after RCA establishment. The temporal analysis is limited to the Strait of Georgia. We compared the effort in 77 RCAs in the Strait of Georgia between 2003, 2007 and 2011. Recreational Fishing effort is higher on weekends than on weekdays so the creel survey is stratified by type of day (English et al. 2002). Following the methods used in English et al. (2002) I normalized effort in and around RCAs by the type of day (weekday vs. weekend) as well as the number of flights per month in order to compare effort between years (Equation 1). The number of flights per month varied among years. To normalize the sampling effort for each RCA in each year, the total boat observations by day were multiplied by the total number of each day of that type in each month (i.e. 19, 20 or 21 weekdays and 9, 10 or 11 weekend days) to yield the number of boat-days. I calculated the average monthly boat observations for each day type by dividing by the number of surveys taken on that day type and then added the average of the weekday and weekend together for an estimate of the monthly boat observations (per RCA, per year). I then summed the observations for all months (May-September) to get an estimate of fishing effort in boats/year during the periods of observation (Equation 1). 94 ?̅?𝑀𝑅 =∑((∑𝐵𝑀𝑡R ∗ 𝐷𝑀𝑡)𝐹𝑀𝑡)2𝑡 (Equation 1) Where: B = Boats (effort) M = Month t = Day type (1 = weekday, 2 = weekend) D = Days per month R = RCA F = Flights (surveys) In order to control for overall changes in fishing effort in the vicinity of the RCAs, I also calculated the effort in a 1-km wide buffer strip around each RCA. In GIS, I intersected the fishing observations with the RCAs and a 1-km-wide buffer strip around each RCA. To compare effort among RCAs and between RCAs and buffers, I calculated an effort density by dividing the effort by the area (km2) of the RCA and the buffer. To analyze the relationship between fishing effort and protection status among years, I used a linear mixed effects model with a Gaussian distribution in the program R (R Development Core Team 2008) and the package nlme (Pinheiro et al. 2012). Because my data were highly skewed, I used a log(x+1) data transformation to normalize variance (Zar, 1986). Year (2003, 2007 and 2011), treatment (RCA vs. Buffer), and the interaction of year and treatment were entered as fixed effects and RCA was a random factor. Visual inspection of the residual plots did not reveal any obvious deviations from homoscedactisity or normality after transformation. The model was estimated using the log-likelihood method so that models could be compared using the Aikake Information Criterion (AIC) (Zurr et al. 2009). The full model was compared to 95 reduced models with only Year, Treatment, and Year + Treatment without the interaction. An interaction between year and treatment indicates that the fishing effort differential between protected and unprotected areas varied over time. A contrast between 2003 vs. 2007 and 2011 indicates an effect of fisheries closure on effort. The final model was fit using restricted maximum likelihood estimation (REML) (Zurr et al. 2009) and the package lmerTest (Kuznetsova et al. 2014). Next, I identified which RCAs had experienced a change in fishing effort over time. I used analysis of variance on the monthly log(x + 1) effort density (May-September) for each RCA with year (2003, 2007, 2011) as the independent variable. I adjusted the critical P value for multiple comparisons with the Bonferroni correction. When significant differences among years were found, I used a pairwise Tukey HSD test to determine which years differed. To examine changes in effort on even finer spatial scales, I plotted spatial fishing effort using the Kernel Density tool in ArcGIS 10.1 (ESRI 2011) with a grid cell size of 100 x 100m and a one kilometer search radius. The Kernel Density tool calculates a magnitude effort per unit area from point features using a cubic function to fit a smoothly tapered surface to each point (ESRI 2012). To turn these kernel density plots into a probability density, I divided effort in each cell in the kernel density plot in each year by the total number of observations used to calculate the plot using the Raster Calculator tool (Abbott and Haynie 2012). Once the density plots have been normalized using this method, the volume under the entire density plot equals one and the value in each grid cell is the probability of fishing effort. Abbott and Haynie (2012) used this method to compare fishing effort from trawling in and around two fishing closures in Alaska. I then compared the density probabilities among time periods using the raster calculator to subtract the effort between years and mapped the results. 96 Figure 4-1. Study area of southern British Columbia showing the Rockfish Conservation Areas (RCAs) on and off of the aerial survey flight path and Pacific Fishery Management Areas (11- 29) (PFMA). 97 I also calculated the standardized effort for each Pacific Fisheries Management Area (PFMA, Figure 4-1) using Equation 1. I compared the proportion of the summed effort in all RCAs in a Pacific Fishery Management Area (PFMA) to the total effort in each area and compared this proportion among years with a Chi-square contingency table analysis. 4.2.4 Factors Affecting RCA Compliance I explored the geographic factors that affect recreational compliance with a Generalized Additive Model (GAM) using the package mgcv (Wood 2011) in R (R Development Core Team 2008). I used a Gaussian distribution and Identity link function to predict fishing effort in 2011 in 105 RCAs around Vancouver Island that could be observed from the over-flights. I omitted one extremely small RCA (Hardy Bay) (0.1 km2). I calculated the fishing effort probability using the normalized kernel density method described above. I made separate probability plots for each region surveyed in 2011. Because I did not compare among years in this analysis, I did not need to standardize the data using Equation 1. I summed the fishing probability over the entire RCA as well as a two kilometer-wide buffer area around each RCA. The dependent variable in the GAM model was the total probability of effort in each RCA divided by the area of the RCA. Potential explanatory variables included the distance from each RCA to the closest city (population >5,000) and to the closest fishing lodge; the fishing effort in the buffer; the estimated rockfish catch in each PFMA; the human population within a 25 km radius of each RCA; RCA size and the perimeter-to- area ratio; the number of hours patrolled by conservation officers by geographic region; and geographic region as a factor (Appendix Table 3-1). I did not include the distance to other coastal features such as the closest community or boat ramps because of the low variability among RCAs in these features. The mean distance between RCAs and the closest community and boat ramp was 7.1 km (SD 6.7) and 6.8 km (SD 6.7), respectively. I used a rank 98 transformation on the dependent variable as well as effort outside, enforcement, population and RCA Area (Appendix Table 3-1). In order to identify the explanatory variables that have a significant influence on the fishing effort density, I used the package MuMIn (Barton 2013) in R (R Development Core Team 2008) to compare all sub-models of the full model and to rank the sub-models using the AICc. The model with the lowest AICc value is the best model and the explanatory variables retained in this model can be assumed to have a significant influence on the fishing-effort density in the RCAs (Stelzenmüller et al. 2008). Finally, I tested the effort density in RCAs among regions and among park type (i.e., RCAs that are in a National Park, Provincial Park, Provincial Ecological Reserve, or not a park) using a Kruskal-Wallis test. 4.2.5 Rockfish Catch in RCAs It is not possible to directly measure the rockfish catch taken from RCAs from the spatial data because creel interviews are conducted over regions at broader spatial resolution. In order to estimate the fisheries take from the RCAs, I calculated the total recreational effort for each Pacific Fisheries Management Area (PFMA, Figure 4-1) using Equation 1. I then calculated the fraction of fishing effort in each PFMA that occurred in RCAs and applied this proportion to the creel survey estimates of total rockfish caught per PFMA derived from the overflight data and dockside interviews (data from D. O’Brien, South Coast Management, DFO). I assumed that rockfish could be caught anywhere in the PFMA, regardless of habitat features such as depth or bottom cover, and that catch rates are equivalent between RCAs and unprotected areas. This is a conservative estimate of take from the RCAs because it is likely that a greater proportion of the rockfish catch would be caught in suitable rockfish habitat, and therefore in the RCAs. 99 4.3 Results 4.3.1 Data Compilation The number of images georeferenced by region and year, along with the positional accuracy (Root Mean Squared (RMS) error) is shown in Table 4-1. The fishing events from the southwest coast of Vancouver Island in 2011 were supplied digitally by Fisheries and Oceans Canada without an estimate of positional accuracy. The dataset contains over 65,000 recreational fishing observations. 4.3.2 Temporal Analysis Effort in the one km-wide buffers around the 77 RCAs declined with time between 2003, 2007 and 2011, while effort in the RCAs remained largely unchanged (Figure 4-2). This decline corresponds with the declining trend that is seen for the fishing effort in the whole Strait of Georgia (Appendix Figure 3-1). The linear mixed-effect model with the lowest AIC value was the full model with fixed effects for year, protection status, and the interaction of year and protection status (Table 4-2). The treatment effect is the strongest, with greater effort outside of the RCAs than inside in all time periods, including 2003 before the RCAs were established. The interaction between year and treatment indicates that the difference between effort in and outside of the RCAs declined with time (Figure 4-2). I also compared the proportion of effort in the RCAs to the total effort in each management area (PFMA) by year (Figure 4-3). I did not show PFMA 15 due to a change in the flight pattern over one large RCA. A Chi-square test showed no evidence of change in the proportional effort in the RCAs by year (Figure 4-3). 100 Table 4-1. Number of images per region and year, positional accuracy (Root Mean Squared (RMS) error) and number of fishing events observed. Area/Year # Images Mean RMS Error (m) # Fishing Events Strait of Georgia 2003 575 20.2 18,778 Strait of Georgia 2007 581 15.9 11,734 Strait of Georgia 2011 727 30.1 14,551 N. Vancouver I. 2011 55 16.9 4,660 SW. Vancouver I. 2011 NA NA 15,499 Total 1938 65,222 101 Table 4-2. Results of the preferred linear mixed effects model (AIC=1379) using 462 observations and 77 groups (RCAs) and fit with REML. Degrees of freedom calculated with Satterthwaite approximations. Fixed Effects Value Std. Error DF t-value p-value 2.5% 97.5% Intercept 356.6 35.7 382.3 10.0 <0.001* 286.6 1.3 Year -0.2 0.02 382.3 -9.9 <0.001* -0.2 -0.1 Treatment -280.8 50.6 382.3 -5.5 <0.001* -379.8 -181.9 Year*Treatment 0.1 0.03 382.3 5.5 <0.001* 0.1 0.2 *significant effect 102 Figure 4-2. Boxplot of standardized log +1 fishing effort density inside and outside of RCAs in 2003, 2007, and 2011. Two extreme outliers are not shown. 103 Figure 4-3. Proportion of recreational effort in RCAs in the Strait of Georgia by management area (PFMA) by year. PFMA 15 is not shown due to a change in the flight pattern over one large RCA. 104 I used estimates of monthly effort per year for each RCA (Table 4-3, Appendix Figure 3-2, Appendix Table 3-2) and the kernel density effort maps (Figure 4-4) to examine effort in individual RCAs. There was no evidence of a change in effort in most RCAs (83%) amongst the three years. Thirty RCAs had moderate or high effort in all years. Effort in two RCAs increased between 2003 to 2007, and effort increased in three different RCAs between 2007 and 2011. Deepwater Bay and Oyster Bay had greater effort 2007 than 2003 and Copeland and Darcy Island had greater effort in 2011 than in 2007. Effort in Deepwater Bay declined in 2011 as compared to 2007 (Figure 4-4). Effort declined significantly in five RCAs in both time periods and in three other RCAs in 2007 or 2011 (Table 4-3). Rockfish Conservation Areas south of Victoria generally showed good compliance although there was some increase in effort between 2007 and 2011 (Figure 4-4). Not all RCAs are fished. Fourteen RCAs had no effort in any year and 20 others had only very low effort in all three years. Offshore RCAs, Ajax, Halibut and McCall banks, had no effort in any year (Figure 4-4). 105 Table 4-3. Summary of changes in monthly recreational effort with time by RCA. Change in Effort # of RCAs % No Change, Effort in all years 30 39.0 No Change, No Effort 14 18.2 No Change, Low Effort† 20 26.0 Increase in 2007* 2 2.6 Increase in 2011* 3 3.9 Decrease in 2007* 2 2.6 Decrease in 2011* 1 1.3 Decrease in 2007 and 2011* 5 6.5 Total 77 100.0 *Monthly effort in some RCAs increased or decreased (p<0.05). †RCAs with 0.01-0.5 boats per km2 were characterized with low effort. 106 Figure 4-4. Difference in kernel density effort maps between 2007 and 2003 (top) and 2011 and 2007 (bottom). Blue colors indicate lower effort in the later year while warmer colors indicate higher effort. Most RCAs are shown in grey which indicates no change in effort. The analysis is limited to the Strait of Georgia. 107 4.3.3 Factors Affecting RCA Compliance The best fit GAM model to predict effort in the RCAs retained the fishing effort density in the buffer, the distance to fishing lodges, the amount of enforcement effort (patrol hours), the area of the RCA and the perimeter-to-area ratio. Distance to cities, population within 25 km and the level of rockfish catch in the PFMA were not included in the best model (Table 4-4). Effort in the RCA generally increased with the amount of fishing effort in the buffer. Fishing effort also increased in proximity to fishing lodges. Fishing effort declined with distance to a lodge up to around 15-20 km, beyond which the relationship was flat. Many RCAs, particularly in the southern Strait of Georgia, are not near any fishing lodge (Figure 4-5). Larger RCAs also experienced greater fishing effort per unit area than smaller RCAs. Effort also increased with the ratio of perimeter-to-area, although the relationship reaches a maximum and flattens out. The influence of enforcement is complex and non-linear, perhaps because the patrol hours spent in a geographic region is a coarse estimate that may not realistically represent the actual amount of time an RCA in a region was enforced. Region was not a significant factor and was not retained in the model despite a significant difference in effort between Mid-Strait of Georgia (MSOG), where many RCAs are not fished, and the Campbell River (CR) and Queen Charlotte Strait regions (Figure 4-6). I found no difference in effort density among park types with a Kruskal-Wallis test (K = 1.4, df = 3, p = 0.7), although the power to detect a difference is low. 108 Table 4-4. Explanatory variables that were retained and rejected in the best model (deviation explained 75.7%, adjusted R2 of 0.67, n=105). Model Variable Retained in Model F p Fishing Effort Density in 2km Buffer Yes 8.0 <0.0001 Size (km2) Yes 31.6 <0.0001 Enforcement (Patrol hrs/Stat Area) Yes 6.1 <0.0001 Distance to Fishing Lodge (km) Yes 3.6 0.001 Perimeter-Area Ratio (PA) Yes 4.0 0.006 Distance to City (km) No Rockfish Catch (per Stat Area) No Population within 25 km radius No Region (factor) No 109 Figure 4-5. Fishing effort probability around Vancouver Island along RCAs on the flight path and 2km-wide buffers, as well as the location of fishing lodges and cities (population >5,000). Insets show RCAs in the southern and northern extents of Vancouver Island. 110 Figure 4-6. Five significant variables were retained in the RCA Compliance Generalized Additive Model. Plots show the smooths of the significant variables. Shaded regions are 2 standard error confidence bands for smooths and the points are partial residuals. Region (lower right) was not retained in the model despite an overall significant difference among regions; however, a multiple comparison test showed that Effort Density in RCAs was only lower in the Mid-Strait of Georgia (MSOG) than in the Campbell River (CR) and Queen Charlotte Strait (QCS) regions. 111 4.3.4 Rockfish Catch in RCAs The fishing effort in the RCAs was quite low compared to the effort in each management area (PFMA) (<5% in most cases). One large RCA on NW Vancouver Island (PFMA 27), Topknot, experiences 10% of the fishing effort in the region (Appendix Figure 3-3). Rockfish Conservation Areas near Nanaimo (PFMA 17) and in the Southern Gulf Islands (PFMA 18) experienced 8% and 6% of the effort respectively. I estimated that RCAs in PFMA 17 and 18 contributed almost 720 and 370 rockfish, respectively, to the fishery in 2011, although this catch was spread over numerous RCAs. I estimated that 175 rockfish were fished from the Topknot RCA in PFMA 27 in 2011 (Appendix Figure 3-3). 4.4 Discussion My data strongly suggest that compliance with recreational fishing regulations in the RCAs in BC is low. Almost 40% of 77 RCAs studied in the Strait of Georgia had moderate or high fishing effort after the establishment of the RCAs and effort in five RCAs even increased. Compliance was influenced by the amount of fishing effort adjacent to the RCA, the proximity to fishing lodges, the size and perimeter-to-area ratio and enforcement. A lack of compliance in MPAs may inhibit population recovery and impair reserve effectiveness (Fujitani et al. 2012, Bergseth et al. 2013, Edgar et al. 2014); however, the effectiveness of the RCA network remains unknown. Our results indicate that the effectiveness of the RCA network at promoting population recovery may be compromised by lack of compliance. Only eight of 77 RCAs decreased in one or both years relative to 2003 before the RCAs were established, although effort adjacent to the RCAs declined with time. Recreational fishing effort in the Strait of Georgia is driven by Chinook Salmon (Oncorhynchus tshawytscha) and Coho Salmon (O. kisutch) catches and has steadily declined since the 1980s when over 500,000 112 boat trips per summer were estimated by the creel survey. Recreational effort in the whole Strait of Georgia declined between 2003 and 2011 as a result of declining salmon stocks (Zetterberg et al. 2012b). Therefore, a decline in recreational effort cannot be attributed to the creation of the RCAs. If the recreational fishery increases with improvements in salmon stocks, RCAs may experience greater levels of fishing if compliance is not improved. Fishing effort was lower in the RCAs than in unprotected reference buffer areas; however, effort in the area of the RCAs was also lower before the RCAs were established. Some RCAs were not fished recreationally in any year and a lack of effort in those RCAs is not evidence of a change in fishing behaviour as a result of spatial management. The RCAs were designated in consultation with the sport fishing community and the boundaries of the RCAs were designed to minimize effects on the sport fishery by leaving popular salmon fishing locations open (Granek et al. 2008, Yamanaka and Logan 2010). Involving the resource users in the implementation of management actions is thought to enhance conservation success and fishery management (Granek et al. 2008). Consultation with recreational fishers may not have been effective in this case as awareness about the RCAs appears to be low. If the most popular fishing spots are also the prime rockfish habitat, then the RCAs may not include the most important areas for promoting population recovery. The greatest factor influencing compliance among the RCAs is the amount of fishing adjacent to the RCA, followed by the size of the RCA, with larger RCAs having greater effort per unit area than smaller RCAs. This contrasts with results of Kritzer (2004) who found greater compliance in large than small MPAs. Kritzer’s data highlighted the importance of the perimeter-to-area ratio as people tend to fish near the boundaries of MPAs (Kritzer 2004). Rockfish Conservation Areas with greater perimeter-to-area ratios are more at risk from fishing; 113 however, this relationship levels off at intermediate perimeter-to-area ratios. Rockfish Conservation Areas in close proximity to fishing lodges (within 15 to 20 km) have lower compliance. Although fishing effort is often higher closer to towns (Stelzenmüller et al. 2008, Read et al. 2011) the proximity to cities and the population within 25 km of the RCA were not significant factors. My model indicates that non-compliance is more likely if there is greater effort in the region, and if the RCAs are a large “target” with a long boundary. Fishers are likely incidentally fishing in RCAs rather than actively targeting them, particularly because the rockfish catch did not influence compliance rates. Recreational fishing patterns in BC are not driven by rockfish because the primary target of most recreational fishers in BC is salmon and the most sought-after groundfish are Halibut and Lingcod (Fisheries and Oceans Canada 2012). Fishers may not, therefore, be aware of the regulations or locations of RCA, despite the fact that rockfish are caught incidentally in these other fisheries which are prohibited in RCAs. The number of patrol hours by region was also a significant predictor in the model, although the patterns are complex and non-linear. Rockfish Conservation Areas near Victoria are relatively well patrolled and have good compliance, despite high fishing effort adjacent to the RCAs. Conversely, one of the most heavily fished RCAs, Topknot, on the remote northwest coast of Vancouver Island, is close to fishing lodges and is rarely patrolled. These patterns might reflect the dense population near Victoria vs. the sparsely populated northwestern coast of Vancouver Island, although population within 25 km of an RCA did not influence compliance. Pollnac et al. (2010) found that reserves close to dense populations were more effective than those near lower populations, perhaps due to greater vigilance by reserve neighbours. Their findings did, however, suggest that compliance is related to a range of conditions such as education, formal consultation, monitoring by community, and clearly defined boundaries rather 114 than to just the level of enforcement per se (Pollnac et al. 2010). Additional enforcement would likely increase compliance in RCAs because fishers adjust their behaviour with levels of enforcement (Fujitani et al. 2012). For example, effort in one RCA, increased between 2003 and 2007, but declined between 2007 and 2011, perhaps as a result of targeted outreach and enforcement by Fishery Officers in Campbell River (Personal Communication, Joe Knight, Conservation and Protection, DFO). Other factors that were not included in the model, such as local stewardship, may also affect compliance. A local newspaper has printed information about the Lion’s Bay RCA and residents with waterfront properties bordering the RCA actively promote it. The Lion’s Bay RCA is the only RCA in Howe Sound that had lower effort in 2007 and 2011 than in 2003. In another case, a resident with waterfront property adjacent to the Mayne Island North RCA reportedly uses a bullhorn to inform people fishing in that RCA of the fishing regulations and reports observations to Fisheries and Oceans Canada. This RCA had less effort in 2011, while a near-by RCA, the Bell Chain Islets, saw an increase in effort. Park type did not, however, affect compliance. There was no difference in fishing effort among RCAs in National or Provincial Parks, or Provincial Ecological Reserves and those with no additional protection. Education and enforcement by park staff could improve compliance in the RCA within parks. Lack of awareness about the location and regulations in the RCAs likely contributes to noncompliance. No physical markers demark the boundaries of the RCAs. After an initial outreach campaign associated with the Rockfish Conservation Strategy, DFO promoted the RCAs with references in the Sport Fish Guide, information on its website, and signs at some boat ramps. The hard copy of a booklet published about the RCAs is not widely available and fishers are often directed to the internet to download an electronic copy (Fisheries and Oceans Canada 115 2008). The location of the RCAs are not readily available in any electronic form or on mapping software used by sport fishers so it is difficult for a fisher to determine when they are in one. This is particularly a problem in inside waters where there are numerous RCAs. Awareness of the RCAs could be ameliorated through the use of new technologies such as GPS and smart phones and the development of tools (apps) that help fishers to locate RCAs as well as educate them about rockfishes and regulations. No-take reserves are more effective than reserves that allow some fishing (Edgar et al. 2014). Therefore, increasing the level of protection to no-take reserves might also simplify the regulations, increase compliance, and increase effectiveness. Positional accuracy of boat observations represent the greatest source of error in identifying whether boats are fishing in RCAs, especially near their boundaries. Although the error associated with the georeferencing process was low (root mean squared error of 16 to 30 m), the positional accuracy of the placement of the boat observations by the aerial survey observer cannot be assessed in this study. Bias was avoided for most of the study because the boundaries of the RCA were not included on the maps used in the aerial survey or during digitizing, with the exception of the West Coast of Vancouver Island (WCVI) where the observer had the locations of the RCAs. This region had lower instances of boats fishing in RCAs than most other regions. The difference was not significant so observer bias likely had little effect on my analysis. Although aerial surveys can cover much larger areas, boat-based surveys might have greater positional accuracy. Smallwood and Beckley (2012) measured compliance in an Australian MPA using aerial and boat-based surveys. They found that aerial and coastal surveys yielded similar results; however, they identified more boats as being engaged in fishing during the boat-based survey. Aerial survey found that 8% of recreational boats were fishing in closed 116 zones while coastal surveys found a slightly higher proportion, 12%, were fishing in closed zones. Our estimates of fishing effort might therefore be conservative if some of the boats observed as “not fishing” were actually fishing. The results of this study could be used to target outreach and enforcement activities. For instance, outreach with fishing lodges could decrease effort in RCAs. Education and outreach activities have been shown to reach a broader population than enforcement and therefore have greater effects on compliance (Alder 1996, Leisher et al. 2012). An Australian study found that education and outreach activities such as newspaper articles and flyers, to TV and radio ads, displays at boat and outdoor shows, signage at boat ramps, and education courses for tourism operators and school children, had a wider impact than enforcement. The per-capita cost of education was also lower than enforcement, although total education costs were greater because education reaches a wider audience (Alder 1996). Leisher et al. (2012) also found that education and outreach activities resulted in greater knowledge of conservation issues and a more positive attitude towards MPAs in Indonesia, particularly among people who were initially undecided about their attitudes towards MPAs. Education and outreach did not change the attitudes of people who started with negative attitudes about MPAs; therefore, a combination of education and enforcement are required to change compliance levels (Leisher et al. 2012). Despite the lack of compliance in many RCAs, the proportional effort in RCAs in most management areas is quite low. However, some of the more heavily fished RCAs in the southern Strait of Georgia may be impacted and the Topknot RCA on the northwest coast of Vancouver Island experiences at least 10% of effort in that region. Continued fishing may affect the performance of such RCAs. In an age-structured model of Black Rockfish, Sethi and Hilborn (2008) found that high rates of poaching negated the biological and fishery benefits of 117 implementing reserves. High rates of non-compliance in reserves have been found to show limited recovery of fish communities empirically as well (McClanahan et al. 2009, McCook et al. 2010, Pollnac et al. 2010, Campbell et al. 2012, Edgar et al. 2014). In an Indonesian marine reserve, Campbell et al. (2012) found that low compliance with no-take zones resulted in decreases in fish biomass in all reserve zones despite an observed recovery in coral cover. In the Great Barrier Reef, McCook at al. (2010) found that no-entry zones had a higher abundance of coral trout than no-take and fishing zones. They concluded that this implied some non-compliance in the less strictly enforced no-take zones. In reserves in the Indian Ocean, McClanahan et al. (2009) found that zones with high enforcement showed significant trends in fish size and age, whereas zones with less enforcement and implied weaker compliance showed lower responses. In the Mediterranean, no-take zones that were accessible by car had a lower abundance of harvested sea urchins than inaccessible no-take zones (Ceccherelli et al. 2011). The impacts of non-compliance may be much greater for long-lived species with low recruitment rates such as rockfish (Bergseth et al. 2013). Non-compliance in the RCAs may therefore affect the performance of RCAs and may be one reason why most RCAs have not shown reserve effects in an accompanying study of fish density measured with remotely operated vehicle surveys inside and outside of RCAs (Chapter 3, Haggarty In Prepartion). Compliance is critical to spatial fisheries management to promote recovery of over-exploited populations (Pollnac et al. 2010, Bergseth et al. 2013, Edgar et al. 2014). Recreational fishing compliance in the RCAs needs to be improved for the RCAs to conserve rockfish populations. Several approaches should be taken including greater enforcement, developing a communication and outreach plan (Grorud-Colvert et al. 2010), and developing tools to assist with locating the RCAs. Fisheries and Oceans Canada should continue to use the fine-scale 118 spatial information in the creel survey over-flight data to monitor compliance in the RCAs and other spatial management actions. In addition to assessing management actions, the model results can be used to inform policy and to target enforcement and education to enhance compliance. The results indicate that efforts to improve compliance with RCAs are urgently needed if they are to meet their goal of contributing to recovery of rockfish populations. 119 Chapter 5: How Do They Score? An Evaluation of Rockfish Conservation Areas Using a Conservation Score that Combines Rockfish Habitat and Key Reserve Features. 5.1 Introduction Networks of Marine Protected Areas (MPAs), or reserves that exclude fisheries, are being implemented worldwide to conserve exploited species and sustain fisheries (Gaines et al. 2010a). Marine Protected Areas have been shown to be a successful strategy to increase the size, abundance and diversity of species protected within them (Allison et al. 1998, Mosqueira et al. 2000, Halpern and Warner 2002, Halpern 2003, Alcala et al. 2005, Claudet et al. 2008, Lester et al. 2009, Babcock et al. 2010, Gaines et al. 2010a, Edgar et al. 2014). Monitoring is critical to the implementation of MPA networks in fisheries management, because ineffective reserves can give resource managers a false sense of security and prevent actions that might otherwise help to achieve the goals of MPAs (Allison et al. 1998, National Research Council 2006, Gaines et al. 2010a, Hamilton et al. 2010). Ecosystem Based Management and Adaptive Management require an understanding of which MPAs contribute the most to recovery of over-exploited populations to inform future actions (Hamilton et al. 2010, White et al. 2011). In response to conservation concerns associated with a sharp decline in inshore rockfishes catches throughout the 1990s in the Northeast Pacific, Fisheries and Oceans Canada (DFO) implemented a system of 164 Rockfish Conservation Areas (RCAs) in British Columbia (BC), Canada, as part of a Rockfish Conservation Strategy. Rockfish Conservation Areas were established between 2004 and 2007 and prohibit commercial and recreational fisheries hook and 120 line and bottom trawl fisheries. Inshore rockfishes include six species of the genus Sebastes (Copper Rockfish S. caurinus, Quillback Rockfish S. maliger, Black Rockfish S. melanops, China Rockfish S. nebulosus, Tiger Rockfish S. nigrocinctus, and Yelloweye Rockfish S. ruberrimus) that are found on shallow (>200m) rocky reefs. Although the RCAs are managed as fishery closures under the Fisheries Act as opposed to being permanently protected as MPAs by Canada’s Oceans Act (Robb et al. 2011), they are designated areas where fisheries that target or lead to substantial bycatch of Rockfishes are prohibited. Rockfish Conservation Areas may be effective strategies to conserve Pacific rockfishes because they are long-lived (some > 100 years) and have small home ranges (Yoklavich 1998, Parker et al. 2000). Marine Protected Areas have been effective for conserving rockfish in California (Paddack and Estes 2000, Hamilton et al. 2010, Keller et al. 2014). Spatial fisheries closures, such as the RCAs, may therefore be effective for promoting rockfish population recovery (Yamanaka and Logan 2010). Inshore rockfishes are associated with complex nearshore rocky habitats (Richards 1987, Matthews 1990b, Love et al. 2002). The highest abundance of Quillback Rockfish in the Strait of Georgia were found in complex rocky habitats above 60m, while Yelloweye Rockfish were found in similar habitats but in deeper waters with greatest abundance 40-100m (Richards 1986). Complex living habitats such as sponge bioherms are also important habitats for both juvenile and adult rockfishes in BC (Cook et al. 2008, Marliave et al. 2009). Habitat structure is one of the most important criteria in the design and assessment of MPAs (Margules and Pressey 2000, Airamé et al. 2003, Roberts et al. 2003a , Parnell et al. 2006, Smith et al. 2009, Gaines et al. 2010a , Claudet and Guidetti 2010, Miller and Russ 2014). A rockfish habitat model was used to designate and assess the conservation targets of the RCAs. A complexity analysis (second derivative of the slope) based on low-resolution (100 m) 121 bathymetry data combined with spatial Catch per Unit of Effort (CPUE) data were used to identify rockfish habitat (Yamanaka and Logan 2010). This model was used to designate the closed area targets as 30% of identified rockfish habitat in “inside waters” between Vancouver Island and the mainland, and 20% of habitat in outside waters on the rest of the coast. The realized habitat protected was 28% and 13% of modeled habitat on inside and outside waters, respectively (Yamanaka and Logan 2010). The quality of habitat inside the RCAs has, however, been questioned (Marliave and Challenger 2009, Cloutier 2011). Marliave and Challenger (2009) studied some RCAs and concluded that the habitat model used to designate the RCAs did not include fine-scale habitat features with the highest rockfish abundance such as a boulder piles. Haggarty (Chapter 3) conducted Remotely Operated Vehicle (ROV) surveys of 35 RCAs and found that although some RCAs contained an abundance of rockfish habitat, rocky reef habitat was sparse in others. The density of Quillback and Yelloweye Rockfishes observed was dependent on the percent of rock substrates observed on ROV transects. SCUBA surveys of rockfish in Barkley Sound, BC, also showed that Black Rockfish density was related to the habitat complexity and the proportion of rocky substrates (Haggarty, Chapter 2). These patterns indicate that lack of suitable habitat may impede rockfish population recovery in some RCAs. Seafloor habitat maps are produced by interpreting a continuous digital bathymetry using biological or geological ground-truthing observations of the seabed. The ground-truthing process samples only a small portion of the seafloor; therefore, the seafloor map is inferred from the association between the remotely sensed environmental data, such as bathymetry and bathymetry derivatives (e.g. slope, rugosity) and the in situ sample data (Brown et al. 2012). High resolution (2-5m) multibeam echosounder (MBES) data are often used to model substrates and habitats (Brown et al. 2011, Lucieer et al. 2013, Diesing et al. 2014, Hill et al. 2014), including rockfish 122 habitats (Yoklavich et al. 2000, Iampietro et al. 2005, Iampietro et al. 2008, Young et al. 2010, Yamanaka et al. 2012, Yamanaka and Flemming 2013). Iampietro et al. (2008) used bathymetric rugosity (a measure of benthic roughness), slope, aspect, depth, and the bathymetric positioning index (BPI) and a General Linear Model (GLM) to effectively predict Yellowtail Rockfish (S. flavidus) habitat in one MPA in California. Models using MBES data have not, to date, been applied at a broader regional scale. The coastline of BC, with its numerous islands and many long fjords, is over 27,000 km long (Thomson 1981). Although the Canadian Hydrographic Service (CHS) has collected MBES data on much of the coast, many areas have not yet been surveyed, particularly in water shallower than 50m (an area termed the “white strip”) (Gregr et al. 2013). Although low-resolution (90-100m) data have been used to model hard-bottom substrates (Dunn and Halpin 2009) and rockfish habitat (Yamanaka and Logan 2010) at regional scales, these data have not been compared to finer-resolution models. We used a new digital bathymetry model available at intermediate-resolution (20m) that was created using original data from the CHS field sheets and nearest neighbour interpolation (unpublished data, E. Greg, SciTech Consulting and S. Davies, DFO) to model rocky reefs using Random Forests (Breiman 2001). Random Forests (RF) (Breiman 2001) is a supervised classification algorithm that has been used in machine learning, data-mining and large-scale predictions (Prasad et al. 2006) and is designed to produce accurate predictions without overfitting the data (Breiman 2001, Cutler et al. 2007). It is related to classification and regression tree (CART) analysis that uses predictor variables to construct a number of decision trees by recursively partitioning the data into smaller homogeneous groups based on a single predictor (Prasad et al. 2006). Unlike CART, RF combines the results of many classification 123 trees that are created by resampling with replacement (bootstrapping). The number of predictors used to find the best split at each node of the tree is also randomly chosen (hence “random”) and a large number of trees (500-2,000) are generated (hence “forest”). Not all of the observations are used to create the bootstrapped sample used to generate the classification. These randomly selected withheld or “out of bag” samples can be used to assess the accuracy of predictions and to compare the relative importance of each variable (Prasad et al. 2006). Random Forest models have been used to classify terrestrial (Prasad et al. 2006, Cutler et al. 2007, Freeman et al. 2012) and marine benthic habitat (Che Hasan et al. 2012, Lucieer et al. 2013, Diesing et al. 2014). A comparison of different supervised algorithms for classifying benthic substrates found that RF achieved the highest accuracy and the best predictive capability (Lucieer et al. 2013, Diesing et al. 2014). We used RF to model rocky substrates above 200m in depth for the entire south coast of BC. The large geographic extent of my study area required a large number of observations to ground-truth the model. I used substrate observations to model rocky reef habitat and then used my rocky reef model as an abiotic surrogate for rockfish habitat (Brown et al. 2011). I applied the model of rocky substrates in an assessment of 144 RCAs in southern BC. In addition to assessing habitat in the RCAs, I explored the cumulative effects of a number of reserve features. Edgar et al. (2014) found that cumulative effects of the following five key features influenced the effectiveness of global MPAs: the degree of fishing permitted within the MPA; the level of enforcement; MPA age (time since designation); MPA size; and habitat isolation, or “the presence of continuous habitat allowing unconstrained movement of fish across MPA boundaries.” Mora et al. (2006) combined some of the same reserve features (MPA size, regulations on the level of extraction, and level of enforcement) as well as external 124 risks and reserve connectivity into a weighted Conservation Score to assess the protection of coral reefs in global MPAs. Similarly, I developed a Conservation Score to assess the RCAs that combines three aspects of habitat (total area of habitat, percent of the RCA, and isolation), RCA size, commercial and recreational compliance, level of extraction (bycatch of rockfishes in the prawn fishery), and connectivity. The Conservation Score of the RCAs is designed to be used as an adaptive spatial management tool (Pressey et al. 2013, Mills et al. 2015) to improve the effectiveness of RCAs to conserve inshore rockfish populations. Adaptive spatial management incorporates new information to improve conservation plans (Pressey et al. 2013, Mills et al. 2015). Adaptive management (Walters 1986) can be an active process whereby conservation plans are treated as formal experiments, or where plans are progressively updated or “fine-tuned” through a less formal “learning by doing” approach (Mills et al. 2015). The need to start adaptive spatial planning can be triggered by the collection of monitoring data and conservation area assessments (Mills et al. 2015). Our Conservation Score provides a basis for adaptive management decisions about expanding or contracting the RCA network in different areas to improve its representation of rockfish habitat and other features that may be related to population recovery. 5.2 Methods 5.2.1 Data Sources The 20m bathymetry raster layer was built using the point soundings data from the original CHS field sheets, as well as electronic nautical charts (S57). Natural neighbour interpolation was then used in ArcMap 10.2.2 to create the 20 x 20m raster (unpublished data, E. Greg, SciTech Consulting and S. Davies, DFO). Separate raster datasets exist for the West Coast 125 of Vancouver Island (WCVI), the Queen Charlotte Strait-Johnstone Strait (QCSt-JS), and the Strait of Georgia (SoG) (Figure 5-1). The multibeam echosounder (MBES) bathymetry and backscatter data were provided for this study under a data-sharing agreement with the Canadian Hydrographic Service (CHS) (Figure 5-1). The MBES bathymetry data were resolved to 5m and output as an XYZ grid from Caris (Geospatial Software Solutions www.caris.com) and then converted to a raster in ArcGIS 10.2.2. Seven terrain variables were derived from each bathymetry data set (5 and 20m resolution): Bathymetric Positioning Index (BPI) at three scales (Broad, Medium and Fine); Slope; Standard Deviation of the Slope; Curvature and Rugosity (Table 5-1). Acoustic backscatter data are produced by the acoustic signal from the MBES that is scattered by the seabed as a function of the angle of incidence of the beam and hardness of the bottom. The strength of the signal and the textural information that it contains relates to the hardness of the seabed (Che Hasan et al. 2014); however, the capacity of standardized backscatter data to provide useful information is just being developed (Lucieer et al. 2013) and was formerly only interpreted through “expert interpretation” (Brown et al. 2012). Backscatter data need to be processed to remove noise before they are useful in substrate classification. Processed backscatter data were not available for the whole extent of the MBES bathymetry data, so I chose five regions with processed backscatter data as study areas (Figure 5-1). Prior to analysis, I used ArcGIS to filter and perform focal statistics on the backscatter raster to remove any remaining noise in the dataset and saved the backscatter raster as a TIFF file. Next I used the package GLCM (Zvoleff 2015) in R (Yoklavich et al. 2007, R Development Core Team 2008) to calculate a Grey Level Co-occurrence Matrix (GLCM) from the TIFF files and to derive the following texture features from the GLCM: Mean, Variance, Homogeneity, Entropy, Correlation 126 and Dissimilarity (Table 5-1). These GLCM textures from backscatter data have been used in other studies to predict substrate type (Lucieer et al. 2013, Che Hasan et al. 2014). The Gini Index, a measure of variable importance calculated by RF showed that only the Mean and Variance of the GLCM contributed information to my classifications, so the other textures were dropped from the analysis. We used a historical dataset of substrate samples from the Canadian Hydrographic Service and ROV observations as my in situ substrate data. The ROV surveys were collected in and adjacent to the RCAs (Chapter 3). Primary substrate observations from the ROV transects were mapped as polygons (Haggarty et al. Accepted). I used the “polygon to point” tool in ArcGIS 10.2.2. to convert the polygons to points each 20m (for the 20m raster) or 5m (for the 5m raster) along the transect. I reclassified substrate points into a binary classification of Rock (bedrock and boulder) and not rock (all other substrates). Rocky Reef Model Using Random Forests 5.2.2 Random Forest Habitat Model In order to model rocky reef habitat, I ran numerous RF models on data from BC’s south coast. I used RF to perform a supervised classification of rocky substrates using the bathymetry data with a 20m resolution and modeled the following regions separately: WCVI, Queen Charlotte Strait-Johnston Strait, and the Strait of Georgia. Next, I used RF to model rocky substrates using MBES bathymetry and backscatter data in five test areas (Figure 5-1). I also modeled the test areas using only MBES bathymetry (no backscatter variables). Last, I used MBES bathymetry data mosaicked for the Strait of Georgia, Queen Charlotte Strait-Johnston Strait, and the WCVI and modeled the three regions using only bathymetry data (no backscatter) (Figure 5-1). 127 The first step in my modeling process was to align the substrate observations (points) with the terrain-variable rasters. I exported data from ArcGIS and used the R packages “Maptools” (Bivand and Lewin-Koh 2015) and “Raster” (Hijmans 2015) to create a data frame of my dependent and independent variables. Next, I used the Marine Geospatial Ecology Tools (MGET) within ArcGIS 10.2.2 to randomly split my data frame into a training and test set, reserving a third of the data as test data (Roberts et al. 2010). The MGET utility provides tools that link ArcGIS to R (R Development Core Team 2008) in order to perform complicated statistical analysis from within the ArcGIS platform (Roberts et al. 2010). Using the training dataset, MGET and the R package “randomForest” (Liaw and Wiener 2002), I fit the RF classification model, and tested the model predictions with the reserved test data. I looked at the model performance statistics as well as the importance of each predictor variable using the Gini Index (Breiman 2001). I dropped any variables with low importance and re-fit the RF model. Finally, I mapped the predicted classification using the predictor variable rasters using MGET and “randomForests” (Liaw and Wiener 2002, Roberts et al. 2010). I mapped the rocky reef model predictions as a probability surface (i.e. the value in each raster cell is the probability that it is rock). In order to translate the probability surface into a binary habitat classification that indicates the presence and absence of habitat, a cut-off threshold needs to be selected. The commonly used default is a threshold of 0.5, but this does not necessarily result in the highest prediction accuracy (Freeman and Moisen 2008). The selection of a cut-off threshold has implications for the prediction accuracy of the classification as well as for the amount of area classified (i.e. a lower cut-off will classify a larger area with lower accuracy while a high cut-off will classify a small area with high accuracy). Cohen’s kappa is a frequently used statistic to compare the performance of classifiers that measure the proportion of 128 correctly classified locations after accounting for the expected agreement by random chance (Freeman and Moisen 2008, Stephens and Diesing 2014). Freeman et al. (2008) found that the least biased results were obtained when threshold cut-offs were chosen to maximize kappa. For each model, I calculated model performance statistics with the test data set for the following threshold range: 0.5, 0.6, 0.7, 0.8 and 0.9. For the 5m resolution models, I selected a cut-off of 0.6 because it produced the highest kappa values for all of the 5m models (Table 5-2). I calculated the total area classified as rock in my test areas for the 5m models using this cut-off and compared the total area classified as rock from the 20m models with a range of cut-offs. In most test areas, the 20m resolution model estimated greater area of rock substrates at a 0.6 threshold (Table 5-3). In order to produce more realistic estimates of area of habitat using the broader resolution data, I classified the probability surface for the 20m models using higher thresholds. I used 0.7 as the cut-off for the Queen Charlotte-Johnstone Straits and the Strait of Georgia and 0.8 for the West Coast of Vancouver Island (Table 5-2). I overlaid the resulting classifications in ArcGIS and inspected the agreement of the models. Once cut-offs were determined, I reclassified the predicted probability surface to show the presence and absence of rock (1, 0) and converted the raster to polygons that indicate the presence of rock (Figure 5-2). In my comparison of the 20m and 5m models (see Results), I found that the 20m models underestimated habitat in inlets with steep sides. This model deficiency meant that RCAs in inlets had unrealistically low habitat estimates. 45 RCAs are found in inlets so this was a significant problem. In order to rectify this problem, I combined the resulting 20m rock polygon layer and the 5m rock polygon layer created with the mosaicked MBES bathymetry data without backscatter (see Figure 5-1 for extent) using the “union” tool in 129 ArcGIS so that any area classified by rock by either layer appears as rock in the resulting polygon layer. I called this final model the 20 + 5 rocky reef model. Only 24 RCAs in inlets had MBES data, so I needed to determine how much habitat was under-estimated by the 20m model so that I could correct the remaining 21 RCAs without MBES data. To do so, I calculated the total habitat area in those 24 RCAs using the 20m and 20 + 5 models. I divided the 20 + 5 habitat area by 20m habitat area. The 20 + 5 habitat area was an average of 4 times higher in those RCAs. For my assessment of the habitat in the RCAs in inlets without MBES (indicated in bold print in Appendix Table 4-1), I multiplied the modeled area of habitat in the 21 RCAs with no MBES data by 4. We calculated the area of predicted rocky habitat using the 20 + 5 m model in each RCA and by Pacific Fishery Management Area (PFMA) and compared them to the amounts estimated in the original model used to designate the RCAs (Yamanaka and Logan 2010). 5.2.3 Conservation Score In order to create a system for evaluating the RCAs in terms of habitat quality and protection afforded to fish, I calculated a Conservation Score for each of the 144 RCAs in southern BC following a similar method used in Mora et al. (2006). Our Score includes the following features that have been linked to reserve performance (Mora et al. 2006, Edgar et al. 2014): area of the RCA, area of habitat, percent habitat, habitat isolation, rockfish bycatch, recreational and commercial compliance, and connectivity. For each RCA, I ranked each component of the Conservation Score between 1 and 3 (bad to good). I determined the rankings of each feature using information found in the literature or used a relative ranking as described below. For the final score, I added each of the feature scores without weighting (see below). 130 5.2.3.1 Rockfish Conservation Areas Size The overall size of reserves is one of the key features of a successful reserve, with reserves larger than 100 km2 being optimal (Edgar et al. 2014). I applied the same thresholds as Edgar et al. (2014): Low (1): <1 km2; Medium (2): 1-100 km2; High (3): >100 km2. 5.2.3.2 Habitat Scores I used my final rocky reef model (20+5) to represent rockfish habitat in the RCAs. Our Conservation Score uses three habitat metrics: the total area of habitat protected in the RCA, the percent habitat of the RCA, and habitat “isolation.” The California Marine Life Protection Act Size and Spacing Analysis used guidelines for both the percent and total area of habitat protected because a small MPA may protect a large fraction of habitat but an insignificant amount of habitat. Similarly, a large MPA may protect a low fraction but large amount of habitat. Edgar et al. (2014) identified habitat “isolation” as a key MPA feature. 5.2.3.2.1 Habitat Area The California MLPA Advisory Team (2006) determined that habitat 5 km2 of habitat within any MPA was a sufficient amount based on adult fish movement patterns. I used 5 km2 as the upper threshold to indicated high area. I used the measured home range size the Black Rockfish, 0.55 km2 (Parker et al. 2007) as the minimum area of habitat. The thresholds used for area of habitat were: Low (1): 0-0.55 km2; Medium (2): 0.55-5 km2; and High (3): >5 km2. 5.2.3.2.2 Habitat Percent I also used recommendations in the California MLPA to determine the rankings for the percent habitat score. The California MLPA considered a rare habitat such as kelp forests to be present in an MPA if it covered at least 10% of the reserve (California MLPA Advisory Team 131 2006). I used this as my middle threshold for percent habitat score and arbitrarily set 50% as my upper threshold. Percent Habitat Score: Low (1): 0-10%; Medium (2): 10-50%; High (3): >50%. 5.2.3.2.3 Habitat Isolation Edgar et al. (2014) identified habitat isolation, defined as “the presence of continuous habitat allowing unconstrained movement of fish across MPA boundaries” as a key feature determining reserve effectiveness. Conversely, Mora et al. (2006) used ‘isolation’ in terms of reserve connectivity. I adopted Edgar et al.’s (2014) terminology but use the modifier “habitat” in association with it. I found habitat isolation to be difficult to measure quantitatively, so I used the following qualitative assessment: Low (1): many rocky reef features breach the RCA boundary; Medium (2): some rocky reef features breach boundary; High (3): few habitat features cross the RCA boundary. 5.2.3.3 Prawn Bycatch Score Another key planning and management feature identified by Edgar et al. (2014) is “the degree of fishing permitted within the MPA.” The RCAs are not no-take MPAs, but they do prohibit fisheries that directly target inshore rockfishes as well as those that lead to substantial levels of rockfish bycatch (Yamanaka and Logan 2010). Commercial and recreational prawn fishing by trap is permitted in RCAs. Research into rockfish bycatch in the prawn fishery showed that rockfish bycatch was low, 0.015 rockfishes per trap per day, but could cumulatively impact rockfish populations in areas of high fishing pressure. Juvenile Quillback Rockfish (S. maliger) were the most frequently caught vertebrate in the research prawn survey and all fish suffered severe symptoms of barotrauma (Favaro et al. 2010). I obtained commercial prawn-fishing effort (number of traps) in 10 km2 blocks for 2006-2009 from DFO (unpublished data, L. Barton, Shellfish Data Unit, Pacific Biological Station). Recreational prawn-fishing data were not 132 available. In ArcGIS, I intersected the prawn-fishery effort data with the RCAs and totaled the number of traps fished per RCA per year. I then multiplied the number of traps by the rockfish bycatch rate given in Favaro et al. (2010) and estimated the mean number of rockfishes caught per RCA over the four years. I subjectively set the bycatch thresholds as: Low (3): <10 rockfish/year; Medium (2): 10-100 rockfish/year; and High (1): >100 rockfish/year. 5.2.3.4 Compliance The level of enforcement is another key feature influencing reserve effectiveness (Mora et al. 2006, Edgar et al. 2014). The commercial groundfish fisheries in BC are either monitored electronically with video and GPS technology (hook and line fisheries) or have 100% observer monitoring (bottom-trawl fishery) (Davis 2008). The level of commercial compliance with the RCA regulations is therefore known to be high (unpublished data, N. Olsen, Fisheries and Oceans Canada) and all RCAs were given a score of 3. Recreational compliance in many RCAs is, however, low (Haggarty et al. Accepted). I used estimates of recreational effort per RCA from aerial surveys of recreational effort in 2011 given in Haggarty et al. (Accepted, see also Haggarty Chapter 4). I assumed that RCAs that were not visible on from the aerial survey route had no effort because the survey routes were planned to capture most of the areas that are recreationally fished (English et al. 2002). To determine thresholds, I calculated the median effort (4.7) as well as the first (0) and third quartiles (18). Recreational compliance was then scored as follows: Low (3): >18 boats; Medium (2): 4.7-18 boats; High (1): 0-4.7. 5.2.3.5 Connectivity Networks of MPAs may be more effective if reserves within them are situated close enough to allow for connectivity through larval dispersal (Mora et al. 2006, Botsford et al. 2009, Gaines et al. 2010a). Mean larval dispersal for nearshore rockfish species that have been studied 133 is estimated to be 1-40 km (Gunderson et al. 2008). To determine the shortest distance among RCAs, I used ArcGIS 10.2 to create a line shapefile and drew lines or multipart lines using the “snap to nearest feature” tool to find the closest edge of the RCA. The length of each line was then calculated in kilometers using the “calculate geometry” tool (Lotterhos et al. 2014). I used the shortest distance between adjacent RCAs and scored connectivity following criteria: Low (1): >40 km; Medium (2): 10-40 km; High (3): 1-10 km. 5.2.3.6 Overall Conservation Score We added the score of all eight features to give a Conservation Score for each RCA. Unlike Mora et al. (2006), I did not weight any of the features because I have no clear way to determine which are more important to population recovery. The lowest possible score is 8 and the highest is 24. I also calculated the number of key features (no-take, old, enforced, large, and isolated) that Edgar et al. (2014) found to be significant determinants of reserve success. I calculated summary statistics for each feature. We compared the Conservation Score as well as the number of key features to the log Response Ratio (RR), an indicator of population recovery, for inshore rockfishes, Quillback Rockfish and Yelloweye Rockfish inside to outside of RCAs sampled using an ROV (Haggarty, Chapter 3) and by SCUBA surveys (Haggarty, Chapter 2). I performed a linear regression of the RR and the Conservation Score and the number of key features. 134 Table 5-1. Bathymetric and backscatter (BS) variable descriptions for the MBES and FS models. Variable Description Parameters Software MBES 5 m FS 20 m Fine BPI Bathymetric Positioning Index (BPI) is a measure of location relative to landscape that compares the elevation of a cell to the mean elevation of surrounding cells. Locations higher than surrounding cells are positive and depressions are negative (Wright et al. 2012). BPI is scale-dependent and was calculated at 3 scales. Inner and outer neighbourhoods are specified in map-units and are given to the right for each model resolution. 3 x 25 15 x 125m 3 x 25 6 x 500m BTM Toolbox1 Medium BPI 10 x 100 50m x 500m 10 x 100 200 x 2000m BTM Toolbox1 Broad BPI 25 x 125 250 x 1250m 25 x 125 500 x 5000m BTM Toolbox1 Slope Slope is the maximum change in depth between the cell and the cells in a neighbourhood. Output is in degrees from horizontal. 3 x 3 3 x 3 Spatial Analyst ArcGIS 10.2.22 SD of Slope Standard deviation of slope in a neighbourhood. 3 x 3 3 x 3 Focal Stats, Spatial Analyst ArcGIS 10.2.22 Curvature The slope of slope (the rate of change of the slope). 3 x 3 3 x 3 Spatial Analyst, ArcGIS 10.2.22 Rugosity An index of surface roughness using the arc-chord ratio defined as the contoured area of the surface divided by the area of the surface orthogonally projected onto a plane of best fit (Du Preez 2015). Arc-chord ratio3 BS GLCM Mean Mean backscatter from grey-level co-occurrence matrix (GLCM) in a neighbourhood. 3 x 3 NA GLCM4 BS GLCM Variance Variance of backscatter from GLCM in a neighbourhood. 3 x 3 NA GLCM4 1(Wright et al. 2012), 2(ESRI 2011), 3(Du Preez 2015), 4(Zvoleff 2015) 135 Figure 5-1. Location of RCAs in southern BC the extent of the 20m bathymetry for the Strait of Georgia (SoG), the Queen Charlotte-Johnstone Straits (QCS) and the West Coast of Vancouver Island (WCVI) as well as the extent of the MBES bathymetry data used. The five test areas used for comparisons with multibeam and backscatter models (5x5) are outlined in orange and indicated by letters: A) Goletas Channel, B) Barkley Sound, C) Georgia Strait, D) Gulf Islands, E) Captain Passage. 136 Table 5-2. Model performance statistics for the 20m models by region and the 5m full models (with backscatter (BS) and 5m models without BS at the selected cut-offs. SoG=Strait of Georgia, QCSt=Queen Charlotte Strait, WCVI=West Coast of Vancouver Island. Res. m Area Cut-off OOB Error Rate AUC Accuracy Error rate True positive rateTrue negative rate Kappa 20 QCSt 0.7 0.26 0.84 0.73 0.27 0.50 0.94 0.45 5 Goletas 0.6 0.15 0.97 0.90 0.10 0.86 0.93 0.79 5 No BS 0.6 0.13 0.98 0.93 0.07 0.96 0.91 0.86 20 WCVI 0.8 0.18 0.89 0.75 0.25 0.61 0.94 0.51 5 Barkley Sound 0.6 0.12 0.95 0.91 0.09 0.77 0.96 0.76 5 No BS 0.6 0.13 0.95 0.91 0.10 0.76 0.97 0.76 20 SoG 0.7 0.21 0.86 0.75 0.25 0.50 0.93 0.46 5 Georgia Strait 0.6 0.07 0.87 0.92 0.08 0.43 0.99 0.53 5 No BS 0.6 0.07 0.86 0.91 0.09 0.29 1.00 0.41 5 Gulf Islands 0.6 0.18 0.92 0.83 0.17 0.81 0.86 0.63 5 No BS 0.6 0.20 0.92 0.81 0.19 0.79 0.86 0.61 5 Captain Passage 0.6 0.14 0.92 0.84 0.16 0.81 0.88 0.67 5 No BS 0.6 0.15 0.91 0.82 0.18 0.81 0.83 0.63 137 Table 5-3. Comparison of area of habitat modeled at two resolutions in five test areas by model cut-off. 5m models were made with and without backscatter data. Resolution Model Test Area Cut-off Habitat Area km2 Difference km2 % of Total Area Captain Passage 5 With BS 0.6 0.2 3.4 5 No BS 0.6 0.4 0.2 6.0 20 0.6 0.8 0.6 11.8 20 0.7 0.5 0.3 7.3 20 0.8 0.2 0.0 3.3 20 0.9 0.1 -0.2 0.9 Gulf Islands 5 With BS 0.6 23.4 43.3 5 No BS 0.6 22.4 -0.9 41.6 20 0.6 19.1 -4.3 35.4 20 0.7 12.3 -11.1 22.8 20 0.8 6.3 -17.0 11.7 20 0.9 1.6 -21.8 3.0 Georgia Strait 5 With BS 0.6 4.1 0.8 5 No BS 0.6 4.0 0.0 0.8 20 0.6 11.3 7.2 2.3 20 0.7 4.2 0.1 0.9 20 0.8 1.7 -2.4 0.3 20 0.9 0.5 -3.6 0.1 Goletas 5 With BS 0.6 4.1 4.3 5 No BS 0.6 6.2 2.1 6.5 20 0.6 5.7 1.6 6.0 20 0.7 2.3 -1.8 2.4 20 0.8 0.8 -3.3 0.9 20 0.9 0.2 -3.9 0.2 Barkley Sound 5 With BS 0.6 6.7 4.6 5 No BS 0.6 6.9 0.2 4.8 20 0.6 32.1 25.4 22.2 20 0.7 22.6 15.9 15.7 20 0.8 13.7 7.0 9.5 20 0.9 5.2 -1.5 3.6 138 5.3 Results 5.3.1 Random Forest Habitat Model Random Forests allowed us to predict rocky substrates with high accuracy rates even when using bathymetry data with an intermediate resolution of 20m. Receiver Operator Graphs (ROC) plot the false positive rate versus the True Positive Rate and are one way of visualizing the performance of a classifier (Fawcett 2006). Area Under the ROC Curve (AUC) is a good measure of overall model performance, with good models having an AUC close to 1 (Freeman and Moisen 2008). Our 20m models had AUC values between 0.84 and 0.89, which places them in the “excellent discrimination” category, while most of the 5m models had AUC values between 0.91-0.98, giving them “outstanding discrimination” ability (Dunn and Halpin 2009). No realistic classifier should have an AUC less than 0.5 because random guessing produces a diagonal line between (0,0) and (1,1) on the ROC plot which has an area of 0.5 (Fawcett 2006). The AUC is independent of the cut-off chosen and was all high (minimum 0.84, maximum 0.98) indicating good overall performance (Table 5-2). The “out-of-bag” (OOB) error rate of my models ranged from 13-26 percent and was similar to the error rate from the test data set (Table 5-2). Model Accuracy was higher for the 5m resolution models than for the 20m models, but even the 20m models had accuracies higher than 70%. Model performance of the 5m models with and without backscatter was very similar (Table 5-2). There was very good agreement in rocky reef areas around islands, such as in the Barkley Sound and Captain Pass test areas (Appendix Figure 4-1). In the Goletas Channel and the Gulf Islands test areas that have very steeply sloping rocky bottoms, the amount of rocky habitat classified in the 20m models at these cut-offs was much lower than at 5m resolution. The 20m resolution model underestimates rocky habitat in steep habitat (Appendix Figure 4-1). 139 The total area of modeled rocky habitat from my final 20 + 5 model is considerably lower than the total area estimated by the original habitat model Yamanaka and Logan (2010) used to designate the RCAs (Table 5-4). The spatial resolution of the original model (100m) likely over-estimated habitat (Appendix Figure 4-2). I also found that my model estimates for the overall amount of habitat protected in RCAs inside waters (between Vancouver Island and the mainland) was lower than the original estimate (22% vs. 28%), although it was equal or higher in two management areas (PFMA 18 and 19) (Table 5-4). Our model estimated that slightly more habitat is protected in RCAs on the WCVI (14% vs 13%). 140 Table 5-4. Total modeled area (km2) and % rocky habitat in RCAs by Pacific Fishery Management Areas (PFMA). Area and % estimates from the old habitat model (Yamanaka and Logan 2010) are shown for comparison. Area estimates by sub-area on the WCVI were not given in Yamanaka and Logan (2010). Our Model Yamanaka and Logan (2010) Area Total Habitat Area Habitat in RCAs % Habitat in RCAs Total Habitat Area Habitat in RCAs % in RCA Inside 1223.9 267.4 22 3159.2 897.4 28 12 456.5 105.7 23 1154.0 314.1 27 13 123.1 22.6 18 454.4 133.2 29 14 65.1 10.4 16 144.3 40.6 28 15 106.0 20.6 19 242.6 72.5 30 16 128.8 29.0 22 262.6 77.6 30 17 73.2 18.7 26 282.8 82.2 29 18 65.8 19.6 30 217.2 58.6 27 19 74.4 22.1 30 186.0 55.4 30 20 50.6 2.8 6 0.0 28 37.8 9.7 26 132.3 38.2 29 29 42.7 6.2 14 83.0 25.1 30 Outside (WCVI) 953.8 132.4 14 3660.5 471.7 13 11 161.7 25.2 16 21/121 12.8 0.7 6 23/123 204.8 27.8 14 24/124 88.6 4.6 5 25/125 123.1 8.7 7 26/126 158.0 20.8 13 27/127 204.7 44.6 22 Grand Total 2177.8 399.8 18 141 5.3.2 Conservation Score The mean Conservation Score for the RCAs was 18.6 (SD 1.4) (Table 5-4). No RCA received the lowest or highest possible score. Although the minimum and maximum possible scores were 8 and 24, the range observed was 15-22. The total amount of rocky habitat and percent habitat per RCA ranged greatly (Table 5-4). Nineteen RCAs in inlets without multibeam had less than 10 percent habitat even after correcting for under-estimates. Twenty-nine other RCAs, including six in inlets with MBES data, had less than 10% rocky habitats. Race Rocks had the highest percent at 98% rocky habitat, but only accounts for 2.7 km2 (Figure 5-2). Race Rocks also scored low (1) for habitat isolation because most of the modeled habitat crosses the RCA boundary. Rockfish Conservation Areas that scored high for percent habitat often scored low for habitat isolation while RCAs with medium percent and area habitat often scored high for habitat isolation (Figure 5-2C,D), unless the RCA boundaries crossed the primary habitat features (Figure 5-2A). With a mean area of 18 km2, most RCAs fell into the medium category for RCA size (Table 5-4, Figure 5-3). Only one RCA scored a 1 for size (not visible on Figure 5-3) and three RCAs on the WCVI scored a 3. Large RCAs often also contain a large amount of habitat, but have a low percent habitat (Figure 5-3). For instance, the largest RCA on the S. Coast, Scott Islands (338 km2), includes only 9.8% rocky habitat, but due to its large extent, protects the maximum area of habitat (33.3 km2). Recreational compliance is affected by many factors (Haggarty et al. Accepted) but large RCAs, with long open-water perimeters often score low, indicating high recreational fishing pressure (1) (Figure 5-3). 142 Many RCAs are not affected by bycatch associated with the prawn fishery because most of the prawn effort is located in parts of the Strait of Georgia and in inlets (Figure 5-3). Most RCAs are likely to be well connected and scored a 3 for connectivity, indicating proximity to other RCAs that is within the expected range of larval dispersal. Only two RCAs at the heads of inlets score a 1 for low connectivity (Figure 5-3). We had data ROV or SCUBA data on fish populations for 36 RCAs (Haggarty, Chapter 2 and 3). I found a significant linear relationship between the Quillback Rockfish log Response Ratio (RR) (the log density of inside to outside of the RCA) and the Conservation Score (Figure 5-4). The linear regression was not significant for the combined inshore rockfish group or Yelloweye Rockfish (Figure 5-4). The number of key features identified by Edgar et al. (2014) when the RCA was sampled ranged from 0 to 3 and did not relate to the RR. None of the RCAs possessed all five key features identified in Edgar et al. (2014) and the mean number was quite low (1.7) (Table 5-5). Although some of the RCAs have now been in place for 10 years (Table 5-5), I did not score them as “old” (>10 years) because the monitoring data I compared to were collected when the RCAs had been in place for less than 10 years. 143 Table 5-5. Summary of RCA feature values and scores, conservation score and key feature (Edgar et al. 2014) score. The age of the RCA was not used in the conservation score. Value Score Feature Mean SD Mean SD Age (Years) 9.7 0.9 – – RCA Size (km2) 18.1 36.0 2.0 0.3 Habitat Area (km2) 3.1 5.1 1.8 0.7 % Habitat 20.1 17.7 1.7 0.6 Habitat Isolation 2.2 0.7 Compliance- Commercial 3.0 0.0 Compliance-Recreational (Effort) 23.2 63.8 2.4 0.8 Bycatch (Rockfish/RCA/Year) 20.4 37.1 2.6 0.6 Connectivity (km) 5.8 8.6 2.9 0.4 Conservation Score 18.6 1.4 # Key Feature 1.7 0.9 144 Figure 5-2. Example of four RCAs showing modeled habitat and habitat feature scores (1-3, bad to good). Area is given in km2. 145 Figure 5-3 (A–D). Scores (1=bad, 2=medium, 3=good) of features (A) Size, B) Habitat Area, C) % Habitat, D) Habitat Isolation. See Appendix Table 4-1 and 4-2 for values and scores of each RCA. 146 Figure 5-3 (E–H). Scores (1=bad, 2=medium, 3=good) of features (E) Recreational Compliance, F) Bycatch, G) Connectivity and H) overall Conservation Score of RCAs. See Appendix Table 4-1 and 4-2 for values and scores of each RCA. 147 Figure 5-4. Response Ratio of Quillback Rockfish, Yelloweye Rockfish and combined inshore rockfish group plotted against the Conservation Score and Key Element Count. The RR for Quillback Rockfish is significantly related to the Conservation Score (r2=13.8, F=7.3, df=1, 38, p=0.01). 148 5.4 Discussion 5.4.1 Random Forest Habitat Model Our model of potential rockfish habitat has finer spatial resolution than the original rockfish habitat model used to designate the RCAs (Yamanaka and Logan 2010). Fisheries and Oceans Canada scientists have been working to improve rockfish habitat models (Yamanaka et al. 2012, Yamanaka and Flemming 2013) but these models have been limited by the extent of existing high resolution MBES bathymetry and backscatter data. Spatial planning and fisheries management is often limited by a lack of high-resolution habitat data available at a regional scale (Dunn and Halpin 2009, Le Pape et al. 2014, Rinne et al. 2014). Using the intermediate-resolution data allowed us to model rocky reef habitat above 200m over the whole south coast of BC. Our assessment of habitat in the RCAs showed that habitat in some RCAs is limited. Furthermore, when the cumulative impacts of low habitat, high continued recreational fishing and bycatch of rockfish in the prawn fishery are considered, the recovery potential of rockfish in some RCAs may be poor. Other RCAs scored well, have abundant habitats and have good recovery potential. In most of my test areas, the amount of habitat predicted was greater with the intermediate resolution data when using the same cut-off values on the MBES models. Broader resolution data are likely to overestimate the area of habitat because they do not have the spatial resolution to capture the true heterogeneity of habitats present in the grid cell (Rinne et al. 2014). Similarly, my estimate of the total area of potential rockfish habitat (2180 km2) is roughly a third of the amount estimated by the low-resolution (100m) model used to designate the RCAs (6820 km2) (Yamanaka and Logan 2010). Selecting an appropriate threshold to turn the probability surface into a binary habitat classification is difficult but has major implications for the accuracy 149 of the predictions as well as the total area of predicted habitat (Freeman and Moisen 2008). Choosing a higher threshold to classify the 20m data (0.7 in inside waters and 0.8 on the WCVI) allowed us to realistically estimate the amount of potential rockfish habitat. The intermediate (20m) resolution models performed well in areas such as Barkley Sound and Captain Pass that have small islands surrounded by rocky reefs. Rinne et al. (2014) used bathymetry data with a similar resolution (25m) and BPI at two scales to model rocky reefs in an island archipelago and found that their model successfully predicted 81% of the ground-truthed reefs. Our 20m model performed worse in Goletas Channel and on the steeply sloping eastern side of the Southern Gulf Islands. This presented a problem for estimating habitat within other steeply sloping areas such as the RCAs in coastal fjords and Johnstone Strait. Combining the 20m resolution predictions with predictions from MBES data with a 5m resolution improved my habitat estimates in areas with steeply sloping bottoms such as in coastal fjords. Future work could also explore joining bathymetry data to terrestrial elevation data to determine if this improves the nearshore classification abilities. Area estimates in these steeply sloping habitats are also limited because they are a two-dimensional representation of a habitat with three dimensions. Although this is an issue for all of my area estimates, it is exaggerated in inlets with steep vertical walls. 3-D habitat mapping methods should be explored (Yamanaka and Flemming 2013). RF model performance is also limited by the number and distribution of substrate observations used to train the classification. The MBES test area in the central Strait of Georgia had relatively sparse substrate data (ntrain=229) and a low proportion of observations of rock (9%). The MBES models in this test area with and without backscatter data had the lowest AUC values (0.87 and 0.86, respectively) of the MBES model, although their discriminatory ability is 150 still considered “excellent” (Dunn and Halpin 2009). I was surprised by the similarity in the model performance statistics as well as the area estimates produced by the MBES models with and without backscatter because I expected that the acoustic backscatter data, which relates to the hardness of benthic substrates (Lucieer et al. 2013, Che Hasan et al. 2014), would be useful in discriminating between soft mounds and rocky outcrops (Yamanaka and Flemming 2013). If this had been the case, the RF model with backscatter in the central Strait of Georgia, which is close to the Fraser River Estuary and therefore has an abundance of soft substrates, would have found a lower area of rocky habitat than the model without backscatter. Conversely, the two models found almost the same area of rock (Table 5-3, 4.1 km2 vs. 4.0 km2), indicating that MBES bathymetry data without backscatter can be used to adequately model substrate using RF. The 20m model in this test area did predict rocky substrates along these areas of soft sediment but often with a probability less than the cut-off of 0.7. Future work could include adding water energy data (current or exposure) (Gregr et al. 2013) into the 20m RF model to improve the classification of rock versus soft mounds. Our model of rocky reefs identified potential rockfish habitat. The realized habitat corresponds to the portion of the potential habitat that is occupied at a given time (Le Pape et al. 2014). Further work should be done to explore realized rockfish habitat as well as the segregation of habitats among inshore rockfish species. Observations of inshore rockfishes from ROV or long line surveys could be used with the RF model results of rocky substrates and combined with depth, complexity, exposure and currents to build habitat suitability models for the inshore rockfish community (Le Pape et al. 2014) using methods such as generalized linear or non-linear regression analysis (Young et al. 2010), boosted regression trees (Hill et al. 2014), or gradient forests (Ellis et al. 2012). The RF model could also be expanded to predict other 151 substrate types. Habitat evaluations of the RCAs could also be extended to include data on kelp forests, eelgrass beds, and sponge reefs (Marliave et al. 2009). Incorporating these biogenic habitats in my estimate might change the amount of rockfish habitat present in some RCAs. 5.4.2 Conservation Score We found the overall RCA Conservation Score was significantly related to the response ratio of Quillback Rockfish found inside to outside of 36 for which I had data. The number of the five key features identified by Edgar et al. (2014) present in the RCAs did not relate to the Response Ratio for Quillback; therefore, I feel that my score is a better index of RCA effectiveness. The RCA Conservation Score can be used to assess RCAs for which biological data are lacking. I recommend using my Conservation Score to identify RCAs that may be underperforming and to examine the feature scores to gain an understanding of why particular reserves may be underperforming (White et al. 2011). For instance, 9 RCAs had scores of 15 or 16. The cumulative effects of low habitat scores and high recreational non-compliance and high or moderate bycatch rates may undermine the success of the Dinner Rock, Northumberland Channel, Trincomali Channel and Copeland Islands RCAs, which all scored 15, as well as Maud Island and Galiano Island North that scored 16. West of Bajo Reef and Lasqueti Island-Young Point, both with a score of 16, had moderate or high scores for compliance and bycatch, but had low total area (0.5 and 0.4 km2, respectively) and less than 5% of the RCAs are rocky habitat. Topknot scored low at 16 despite a high total amount of habitat (6.6 km2) as a result of low percent habitat (7%), low habitat isolation, and high continued recreational fishing. All of the top scoring RCAs, Numas Islands, Goletas Channel and Dickson-Polkinghorne Islands with a score of 22, are in the Queen Charlotte Strait. 152 We estimated that 18% (22% inside, 14% outside) of rockfish habitat is protected on BC’s South Coast. The representation of suitable habitat in a marine reserve is one of the most important criteria of reserve effectiveness (Roberts et al. 2003a, Gaines et al. 2010a). The amount or proportion of suitable habitat in an MPA is not included in the list of five key features that determined the effectiveness of MPAs studied by Edgar et al. (2014). However, the Reef Life Survey dataset of shallow reef fish densities used by Edgar et al. (2014) probably pre-selected for MPAs with suitable habitat. The total area of habitat and the percent habitat are important metrics of RCA success. Further assessment of RCAs that have both a low percentage of modeled rockfish habitat as well as a low area of habitat, such as West of Bajo Reef and Lasqueti Island-Young Point, should be undertaken to determine if the socio-economic cost of RCAs with little or no rockfish habitat are warranted (Lancaster et al. In press). Many of the RCAs that scored high for habitat area and percent, scored low for Habitat Isolation. The Broken Group Islands RCA, with a score of 21, scored high or moderate in all features except habitat Isolation. Race Rocks had the highest percent habitat (98%) of any RCA, but scored low for isolation because the boundary is not beyond the edge of the reef. Habitat isolation, or the placement of the reserve boundary with respect to the edge of habitat features was an important feature with successful reserves having a depth- or sand-barrier around reefs (Edgar et al. 2014). Although the placement of a boundary across a habitat may facilitate spillover which could support fisheries (Chapman and Kramer 2000, Russ and Alcala 2011), conservation benefits will be greatest if the reserve edge is matched with a habitat boundary (Gaines et al. 2010a). Rockfish Conservation Areas that scored low for habitat isolation may be ineffective for the conservation of rockfish, particularly if the reef outside of the RCA, such as 153 Topknot, is also heavily fished. Small adjustments to the boundaries of some RCAs may greatly improve their effectiveness. Almost all RCAs in the network scored moderate or high for size and connectivity. The overall size of most RCAs was either moderate (10-100 km2) or high (>100km2). Although I assessed the size following the criteria used in Edgar et al. (2014); intermediate reserves 10-100 km2 that are within networks can be effective (Halpern and Warner 2003). The spacing of the RCAs (mean=5.8 km) also indicates a high likelihood that most RCAs will be well connected through larval dispersal since this is well below the estimated mean dispersal distance of most nearshore rockfishes, 40 km (Gunderson et al. 2008). Connectivity was the second highest scoring feature. A genetic study of RCAs on the outer coast of BC found that RCAs are well connected for Black Rockfish (Lotterhos et al. 2014). Only two RCAs in southern BC were greater than 40km apart from their nearest neighbour (Bute Inlet North and Holberg Inlet). Although these RCAs far up inlets are a great distance from coastal reserves, oceanography of the inlet may limit larval exchange between RCAs in inlets and the open coast because Copper Rockfish populations in inlets have been shown to be genetically divergent from coastal populations (Dick et al. 2014). Despite the low connectivity of RCA in inlets, they may be very important for conserving local populations within inlets. All RCAs scored a 3 for commercial compliance because electronic or observer monitoring is a requirement of commercial groundfish fishery licenses (Davis 2008). Recreational compliance in the RCAs, on the other hand, needs to be improved through outreach programs designed to educate fishers about the location of RCAs, the regulations, and the importance of rockfish conservation (Haggarty et al. Accepted) because awareness about RCAs among the recreational fishing community has been found to be low (Lancaster et al. Submitted-154 b). The level of enforcement has been related to the effectiveness of marine reserves worldwide (Mora et al. 2006, Bergseth et al. 2013, Edgar et al. 2014). Continued recreational fishing in the RCAs and rockfish bycatch in prawn traps are the two most likely causes of continued rockfish fishing mortality in the RCAs. High levels of bycatch in the prawn fishery are, however, limited to nine RCAs. Rockfish Conservation Area management might be amended to exclude prawn fishing in the RCAs or to require bycatch reduction devices be used when fishing in RCAs (Favaro et al. 2013). The best way to address the uncertainty in our knowledge concerning the design of MPAs is to use the existing science to adaptively manage the design and implementation of networks of reserves (Sale et al. 2005, Mills et al. 2015). The new information presented in this study could be used as a trigger for adaptive planning (Mills et al. 2015). Adaptive management may take the form of altering regulations (i.e. prawn fishery), increasing awareness (i.e. recreational fishing compliance), or changing boundaries (i.e. to address spillover and habitat isolation). Rockfish Conservation Areas with very low areas and proportions of habitat should be explored further and possibly modified, although I acknowledge that habitat in RCAs in inlets is likely still underestimated in my model. If low levels of realized rockfish habitat are found in those RCAs, and small rockfish populations are found within them, managers should consider alterations to their boundaries. This would, however, require monitoring and assessment to continue. Currently, DFO has no plans to continue monitoring the RCAs; nor does it have a mandate to review or adaptively manage the RCAs. These have both been acknowledged to be impediments to adaptive spatial management (Mills et al. 2015) and the mandated review of spatial plans is rare (Pressey et al. 2013). Stakeholder support is often also lacking for adaptive spatial management (Mills et al. 2015). Recreational, Commercial and First Nations Fishers interviewed 155 about the RCAs all felt strongly that the RCAs should be monitored and the effectiveness of the RCAs reviewed (Haggarty 2014). The RCAs in BC have now been in place for 9 to10 years. We are reaching the timeline when we should expect to see changes in the rockfish population size in effective RCAs. Our Conservation Score could be used to select RCAs for further monitoring to ensure rockfish conservation is achieved in the RCAs. 156 Chapter 6: Conclusion 6.1 Synopsis Rockfishes are inherently vulnerable to being overfished due to life history characteristics such as extreme longevity, sporadic recruitment success, small home range size, and poor survival after being discarded (Parker et al. 2000). In British Columbia (BC), over harvesting has resulted in COSEWIC designations of Special Concern and Threatened for Yelloweye and Quillback Rockfish respectively. Systems of MPAs and RCA have been developed to conserve Pacific rockfishes in Canada (Yamanaka and Logan 2010) and the US (Karpov et al. 2012, Gleason et al. 2013, Keller et al. 2014, Starr et al. 2015). Response to protection is, however, expected to be slow for rockfishes because of the same life history traits that make them vulnerable to overfishing (Karpov et al. 2012, Starr et al. 2015). Starr et al. (2015) predict that it could take as long as 20 years for reserve responses in MPAs to be detectable. However, Halpern and Warner (2002) found that reserve effects are often detectable after only 3 years of protection. At the outset of this work, I expected that few significant reserve effects would be apparent as a result of the newness of the RCAs. Early monitoring and evaluation are; however, essential to establish initial conditions in RCAs as well as to identify potential short comings or problems with RCA networks that should be remedied so that population recovery is more likely. In this dissertation, I found little evidence that rockfish populations have started to recover in the RCAs. Habitat features and depth explained the fish densities I observed with the ROV while the level of protection did not. Once habitat sampled was accounted for, few Response Ratios (RRs) were different from zero indicating little difference between reserves and reference sites. However, this should not be taken as evidence that the RCAs are “not working.” As Karpov et al. (2012) explained in their analysis of fish densities in the Channel Island MPA 157 “with additional time, the full dimensions of protection should become evident.” I did, however, identify issues that should be addressed now, rather than waiting for another decade to pass and to then find out that some RCAs have flaws that are impeding their success. Specifically, RCAs should be re-evaluated and adaptively managed now that finer-resolution habitat data and more refined models are available. Recreational compliance and overall awareness of rockfish conservation and the RCAs must be bolstered. The positive aspects of the RCAs should also be recognized because, despite some shortcomings, they have great potential to aid not only in rockfish conservation, but in the conservation of marine biodiversity more generally. 6.1.1 Habitat in RCAs Rockfishes are so named because it is well established that are closely associated with rocky reefs (Love et al. 2002). My SCUBA data and ROV data, described in Chapters 2 and 3, showed positive associations between Black Rockfish, Quillback Rockfish, Yelloweye Rockfish and all inshore rockfishes combined and rocky substrates. This result suggests that the fundamental attribute that will determine the success of any RCA is the quality of habitat they contain. Until recently, our ability to model habitat at a broad scale in BC has been limited by the poor resolution of data available. The habitat model I described in Chapter 5 could be extended to the north coast of BC so that all 164 RCAs could be re-evaluated. As finer-resolution multibeam echosounder data are collected, the habitat model could be improved. My modeling results showed that the original data used to plan the RCAs overestimated the amount of habitat, especially in inside waters. It also showed that some RCAs do not contain much habitat and/or have a low percent rocky habitat. These RCAs should be targeted for multibeam echosounder surveys to determine if the habitat model is correct. If those RCAs do have low quality habitat, then they should be moved or altered. For instance, my model results indicate that the placement 158 of the “Topknot” and the “Off Bajo Reef” RCAs on the West Coast of Vancouver Island may have missed the primary reef structures. “Off Bajo Reef” could be moved south to be “On Bajo Reef” and “Topknot” could be extended into deeper water to encompass the excellent reef structure that is only partially contained in the RCA. Other RCA boundaries should also be adjusted to address low habitat isolation. Although some spillover of adult fish from the reserve may enhance fisheries, too many fish straying out of the conservation area will negatively affect population recovery if many fish are caught outside of the reserve (Gaines et al. 2010a, Edgar et al. 2014). The analysis I presented in Chapter 5 showed that many RCAs with a high total amount and high percent habitat, such as “Race Rocks” and “Broken Group Islands” scored low for habitat isolation. If the boundaries of these RCAs were expanded so that it did not bisect the rocky reef habitat, these RCAs would be exceptional. The finer resolution habitat model I presented in Chapter 5 should be used for fine-tuning the RCA network. 6.1.2 Compliance with RCA Regulations In Chapter 4, I evaluated recreational compliance to show that recreational effort in many of the RCAs did not change and that fishing has continued in many, sometimes with high frequency. The drivers of non-compliance I found to be significant: the amount of fishing outside of the RCA, the size and perimeter-to-area ratio, the amount of enforcement and the proximity to fishing lodges, indicated that non-compliance is likely as a result of ignorance about the RCA boundaries and regulations rather than directed poaching. Lancaster (2015) followed up on my research by placing trail cameras in RCAs and by surveying recreational fishers about their knowledge of the RCAs and her results corroborate mine. Her trail camera work showed similar rates of non-compliance in RCAs in the Southern Gulf Islands (Lancaster et al. Submitted-b). In Lancaster’s social survey, 16% of respondents admitted to accidentally fishing in an RCA while 159 7% admitted to intentionally fishing in one. She also showed that despite a high belief in the need for rockfish conservation (77%), 25% had never heard of the RCAs and 60% were unsure of their locations. Fisher’s knowledge of the regulations was also low as 38% and 23% believed that salmon and halibut fishing, respectively was permitted (Lancaster et al. Submitted-a). In Chapter 5, I showed that the cumulative effects of low compliance, low habitat scores and high bycatch may undermine the success of some RCAs. Greater outreach and education about the locations of the RCAs and a clarification about the regulations are essential to their success. 6.1.3 Conservation Value of RCAs for Marine Biodiversity Over the course of this research, I made observations about the RCAs that are difficult to show quantitatively. I observed excellent habitat, abundant rockfishes, including gravid females and numerous juveniles in many RCAs. Rockfishes can also be considered “umbrella species” whose protection will confer protection on other species and habitats that are associated with them. Many species other than rockfishes are also protected in RCAs both in the rockfish habitat as well as adjacent habitats. I identified and counted 90 taxa of fishes on my ROV transects in RCAs (Appendix 5). Other rocky-reef dwelling fishes likely to benefit from RCAs include often caught species such as Lingcod (Ophiodon elongatus), Kelp Greenling (Hexagrammos decagrammus), and Cabezon (Scorpaenichthys marmoratus). Shiner Surfperch (Cymatogaster aggregata), are not commercial or recreationally important, but may be very significant ecologically as prey for numerous species. Non-reef habitats often form the majority of the habitat features in an RCA and include sand, mud, gravel, and mixed-substrates. Species very common to these habitats are juvenile and adult flatfishes, Spotted Ratfish (Hydrolagus colliei), Pacific Cod (Gadus macrocephalus), Hake (Merluccius productus) and Walleye Pollock (Theragra chalcogramma), numerous species of sculpins, skates, and other odd bottom-dwelling 160 creatures like Blackbelly Eelpouts (Lycodes pacificus). All of these species may benefit from reduced fishing mortality and habitat destruction as a result of RCA regulations. Many commercially caught flatfish and cod species found in waters deeper than 200m utilize shallower habitats as juveniles. Likewise, juvenile rockfishes that are found in deeper waters as adults such as Canary, Yellowtail, Widow, Boccacio and Sharpchin Rockfishes were observed in RCAs during the ROV surveys (Appendix E). Rockfish Conservation Areas may therefore play a role in protecting critical juvenile habitats for other commercially important groundfish species. In addition to quantifying the fishes that were observed on ROV transects common, abundant, and habitat-forming invertebrates were also recorded during video review. Seventy-five invertebrate taxa were observed in RCAs (Appendix E). This is not a comprehensive list of the invertebrates present or their abundance along the transects because not all invertebrates that were recorded were counted or identified, and many others were not visible. Nonetheless, this list provides a sense of the diversity of species that are found in the RCAs. Many of these species form important habitat for rockfishes and other fishes and other invertebrates and support biodiversity (Marliave et al. 2009). Habitat forming species observed during the ROV surveys include gorgonian corals (Gorgonacea), glass and boot sponges (Hexactinellida), sea pens (Ptilosarcus gurneyi) and whips (Virgularia), and anemones (Metridium and Anthopleura). Habitat forming invertebrates are often at risk from bottom contact fishing methods such as bottom trawl, traps, and longlines and therefore may benefit from protection in RCAs. Other invertebrate species that were observed in RCAs are commercially fished species such as giant red sea cucumbers (Parastichopus californicus), prawns (Pandalus platyceros), Dungeness crabs (Cancer magister) and Pacific octopus (Enteroctopus dofleini). Although fisheries for these species are still permitted in the RCAs, they may also benefit indirectly from habitat protection. 161 Rockfish Conservation Areas may therefore play an important role in the protection of marine biodiversity more generally. 6.2 Limitations of Research This evaluation of the RCAs is limited to BC’s South Coast. Monitoring data have not been collected in many RCAs on the north coast and habitat data and recreational fishing observations are much more limited. Many of the same issues may, however, exist and the north coast RCAs should be evaluated. The BC coast is enormous and complex, and monitoring and evaluation of RCAs requires a concerted effort by many more government and academic scientists than is possible in the course of a Ph.D. thesis. Although I surveyed 35 RCAs with an ROV, these represent just 20% of the RCAs. My evaluation of the RCAs might be different if other RCAs were sampled. Fish densities on ROV surveys may be over or under estimated if fishes either avoid or are attracted to the ROV (Stoner et al. 2008). The detectability of fishes may also be lower in highly complex habitats such as around sponges. Further work should be completed to measure the detectability of rockfishes in various habitats. Future ROV surveys could also be planned to test hypotheses about RCA effectiveness. For instance, data could be collected in RCAs with various scores for habitat isolation or recreational compliance. The aerial survey data I presented in Chapter 4 to examine recreational compliance in the RCAs are only snap-shot views of the fishing effort. Although I standardized the effort to be able to compare among RCAs and years, the metric of effort is a relative, not absolute amount. I confirmed which RCAs were visible from flight route with the Creel observers, but I did not account for different levels of visibility encountered on particular surveys. Although DFO only conducts a survey if the weather allows for acceptable visibility, they do record the 162 environmental conditions for each survey. I did not include any weather conditions in my estimates of effort. The habitat model presented in Chapter 5 has many limitations. Low performance in steep habitats can be rectified in the future with additional MBES data and by developing three dimensional models. The model of rocky substrates should also be extended to other substrates types and combined with depth and other environmental data to develop habitat suitability models for the inshore rockfishes. Rockfish Conservation Areas could then be reviewed in terms of the habitat suitability for each species. The RCAs may not perform equally well for all species. 6.3 Recommendations British Columbia now has almost a decade of experience with spatial protection in the network of RCAs. Recommendations from this analysis of the RCAs could be applied to the network of MPAs that Canada and other counties are developing to meet their Aichi Biodiversity Targets (Secretatiat of the Convention on Biological Diversity 2011). 1. Long term monitoring of reserves is essential to determine their effectiveness. Assessments of which reserves are performing well and which reserves are under-performing are necessary ingredients for adaptive management (Hamilton et al. 2010, White et al. 2011). Monitoring data and reserve assessments are also necessary to gain and retain buy-in from the fishing communities. Funding for long-term monitoring is lacking due to the short period of federal research grants and the lack of resources or mandate for the DFO to undertake such efforts. The need for monitoring should therefore be mandated in the establishment of MPAs. 163 2. Outreach, education and enforcement plans must also be developed and maintained for networks of MPAs. Commercial compliance with the RCAs is high because electronic fishery monitoring was put in place shortly after they were established. Recreational compliance, on the other hand, was found to be quite low. Greater education and outreach of RCA boundaries and regulations as well as why they are important is desperately needed. Modern tools such as smart phone applications that employ GPS technology should be explored. These tools could both educate people about conservation initiatives as well as to help people navigate in our increasingly complex world of spatially-explicit management regulations. Compliance monitoring should also be built into monitoring plans to assess if regulations are having their desired effects. Enforcement must also be made a priority and supported with funding. 3. Although many RCAs protect good rockfish habitat and contain good rockfish populations, not all of the RCAs are likely to be effective. Some RCAs were simply not well located. Rockfish Conservation Areas identified by my Conservation Score as suboptimal should be examined further and adapted. The habitat modeling work described here should be extended into habitat suitability for inshore rockfishes. 4. Rockfish Conservation Areas should be spatially adaptively managed (Pressey et al. 2013, Mills et al. 2015). The ongoing development of a system of MPAs to achieve the Aichi Biodiversity target for marine biodiversity conservation (Secretatiat of the Convention on Biological Diversity 2011) could provide an opportunity for the RCAs to be reviewed, spatially adapted, and upgraded into MPAs. White et al. (2011) very elegantly put it: “Now that networks of reserves have been implemented worldwide, the time is ripe for the implementation of adaptive management. …Questions need to evolve 164 from “Do reserves work?” to “When and why do marine reserves work, how long does it take, and what should we be measuring?” 165 References Abbott, J. K. and A. C. Haynie. 2012. What are we protecting? Fisher behavior and the unintended consequences of spatial closures as a fishery management tool. Ecological Applications 22:762-777. Agardy, T., P. Bridgewater, M. P. Crosby, J. Day, P. K. Dayton, R. Kenchington, D. Laffoley, P. McConney, P. A. Murray, J. E. Parks, and L. Peau. 2003. Dangerous targets? Unresolved issues and ideological clashes around marine protected areas. Aquatic Conservation: Marine and Freshwater Ecosystems 13:353-367. Ainsworth, C. H., H. N. Morzaria-Luna, I. C. Kaplan, P. S. Levin, and E. A. Fulton. 2012. Full compliance with harvest regulations yields ecological benefits: Northern Gulf of California case study. Journal of Applied Ecology 49:63-72. Airamé, S., J. E. Dugan, K. D. Lafferty, H. Leslie, D. A. McArdle, and R. R. Warner. 2003. Applying ecological criteria to marine reserve design: a case study from the California Channel Islands. Ecological Applications 13:170-184. Alcala, A. C., G. R. Russ, A. P. Maypa, and H. P. Calumpong. 2005. A long-term, spatially replicated experimental test of the effect of marine reserves on local fish yields. Canadian Journal of Fisheries and Aquatic Sciences 62:98-108. Alder, J. 1996. Costs and effectiveness of education and enforcement, Cairns Section of the Great Barrier Reef Marine Park. Environmental Management 20:541-551. Allison, G. W., J. Lubchenco, and M. H. Carr. 1998. Marine reserves are necessary but not sufficient for marine conservation. Ecological Applications 8:S79-S92. Ammann, A. J. 2004. SMURFs: standard monitoring units for the recruitment of temperate reef fishes. Journal of Experimental Marine Biology and Ecology 299:135-154. Anderson, C. N. K., C.-h. Hsieh, S. A. Sandin, R. Hewitt, A. Hollowed, J. Beddington, R. M. May, and G. Sugihara. 2008. Why fishing magnifies fluctuations in fish abundance. Nature 452:835-839. Anderson, T. J. and M. M. Yoklavich. 2007. Multiscale habitat associations of deepwater dernersal fishes off central California. Fishery Bulletin 105:168-179. Ardron, J. A. 2002. A GIS recipe for determining benthic complexity: an indicator of species richness.in J. Breman, editor. Marine geography: GIS for the oceans and seas. ESRI Press, Redlands, California. Arias, A. and S. G. Sutton. 2013. Understanding Recreational Fishers' Compliance with No-take Zones in the Great Barrier Reef Marine Park. Ecology and Society 18. Babcock, R. C., N. T. Shears, A. C. Alcala, N. S. Barrett, G. J. Edgar, K. D. Lafferty, T. R. McClanahan, and G. R. Russ. 2010. Decadal trends in marine reserves reveal differential rates of change in direct and indirect effects. Proceedings of the National Academy of Sciences 107:18256-18261. Barton, K. 2013. MuMIn: Multi-model inference. R package version 1.9.11. http://CRAN.R-project.org/package=MuMIn. Beamish, R. J., G. A. McFarlane, and A. Benson. 2006. Longevity overfishing. Progress In Oceanography 68:289-302. Beaudreau, A. H. 2009. The Predatory Role of Lingcod (Ophiodon elongatus) in the San Juan Archipelago. University of Washington, Seattle. 166 Beaudreau, A. H. and T. E. Essington. 2007. Spatial, temporal, and ontogenetic patterns of predation on rockfishes by lingcod. Transactions of the American Fisheries Society 136:1438-1452. Bergseth, B. J., G. R. Russ, and J. E. Cinner. 2013. Measuring and monitoring compliance in no-take marine reserves. Fish and Fisheries:n/a-n/a. Berkeley, S. A., C. Chapman, and S. M. Sogard. 2004a. Maternal age as a determinant of larval growth and survival in a marine fish, Sebastes melanops. Ecology 85:1258-1264. Berkeley, S. A., M. A. Hixon, R. J. Larson, and M. S. Love. 2004b. Fisheries Sustainability via protection of age structure and spatial distribution of fish populations. Fisheries 29:23-32. Bivand, R. and N. Lewin-Koh. 2015. maptools: Tools for Reading and Handling Spatial Objects. R package version 0.8-34. Bobko, S. J. and S. A. Berkeley. 2004. Maturity, ovarian cycle, fecundity, and age-specific parturition of black rockfish (Sebastes melanops). Fishery Bulletin 102:418-429. Bode, M., P. R. Armsworth, H. E. Fox, and L. Bode. 2012. Surrogates for reef fish connectivity when designing marine protected area networks. Marine Ecology Progress Series 466:155-166. Botsford, L. W., D. R. Brumbaugh, C. Grimes, J. B. Kellner, J. Largier, M. R. O'Farrell, S. Ralston, E. Soulanille, and V. Wespestad. 2009. Connectivity, sustainability, and yield: bridging the gap between conventional fisheries management and marine protected areas. Reviews in Fish Biology and Fisheries 19:69-95. Botsford, L. W., J. C. Castilla, and C. H. Peterson. 1997. The Management of Fisheries and Marine Ecosystems. Science 277:509-515. Botsford, L. W., A. Hastings, and S. D. Gaines. 2001. Dependence of sustainability on the configuration of marine reserves and larval dispersal distance. Ecology Letters 4:144-150. Botsford, L. W., F. Micheli, and A. Hastings. 2003. Principles for the design of marine reserves. Ecological Applications 13:25-31. Breiman, L. 2001. Random forests. Machine Learning 45:5-32. Brown, C. J., J. A. Sameoto, and S. J. Smith. 2012. Multiple methods, maps, and management applications: Purpose made seafloor maps in support of ocean management. Journal of Sea Research 72:1-13. Brown, C. J., S. J. Smith, P. Lawton, and J. T. Anderson. 2011. Benthic habitat mapping: A review of progress towards improved understanding of the spatial ecology of the seafloor using acoustic techniques. Estuarine, Coastal and Shelf Science 92:502-520. Caley, M. J., M. H. Carr, M. A. Hixon, T. P. Hughes, G. P. Jones, and B. A. Menge. 1996. Recruitment and the local dynamics of open marine populations. Annual Review of Ecology and Systematics 27:477-500. California MLPA Advisory Team. 2006. California Marine Life Protection Act Size and Spacing Analyses. Campbell, S. J., A. S. Hoey, J. Maynard, T. Kartawijaya, J. Cinner, N. A. J. Graham, and A. H. Baird. 2012. Weak Compliance Undermines the Success of No-Take Zones in a Large Government-Controlled Marine Protected Area. Plos One 7. Carr, M. H. 1991. Habitat selection and recruitment of an assemblage of temperate zone reef fishes. Journal of Experimental Marine Biology and Ecology 146:113-137. 167 Carr, M. H. and P. T. Raimondi. 1999. Marine protected areas as a precautionary approach to management. California Cooperative Oceanic Fisheries Investigations Reports 40:71-76. Carr, M. H. and D. C. Reed. 1993. Conceptual Issues Relevant to Marine Harvest Refuges: Examples from Temperate Reef Fishes. Canadian Journal Fisheries Aquatic Sciences 50:2019-2028. Carr, M. H. and C. Syms. 2006. Recruitment. Pages 411-427 in L. G. Allen, D. J. Pondella II, and M. H. Horn, editors. The Ecology of Marine Fishes. University of California Press, Berkeley, CA. Carwardine, J., C. J. Klein, K. A. Wilson, R. L. Pressey, and H. P. Possingham. 2009. Hitting the target and missing the point: target-based conservation planning in context. Conservation Letters 2:3-10. Caselle, J. E., M. H. Carr, D. P. Malone, J. R. Wilson, and D. E. Wendt. 2010a. Can we predict interannual and regional variation in delivery of pelagic juveniles to nearshore populations of rockfishes (Genus Sebastes) using simple proxies of ocean conditions? California Cooperative Oceanic Fisheries Investigations Reports 51:91-105. Caselle, J. E., B. P. Kinlan, and R. R. Warner. 2010b. Temporal and spatial scales of influence on nearshore fish settlement in the Southern California Bight. Bulletin of Marine Science 86:355-385. Cass, A. J., R. J. Beamish, and G. A. McFarlane. 1990. Lingcod (Ophiodon elongatus). Canadian Special Publication of Fisheries and Aquatic Sciences 109. Ceccherelli, G., A. Pais, S. Pinna, N. Sechi, and L. A. Chessa. 2011. Human impact on Paracentrotus lividus: the result of harvest restrictions and accessibility of locations. Marine Biology 158:845-852. Chapman, M. R. and D. L. Kramer. 1999. Gradients in coral reef fish density and size across the Barbados Marine Reserve boundary: effects of reserve protection and habitat characteristics. Marine Ecology-Progress Series 181:81-96. Chapman, M. R. and D. L. Kramer. 2000. Movements of fishes within and among fringing coral reefs in barbados. Environmental Biology of Fishes 57:11-24. Che Hasan, R., D. Ierodiaconou, L. Laurenson, and A. Schimel. 2014. Integrating Multibeam Backscatter Angular Response, Mosaic and Bathymetry Data for Benthic Habitat Mapping. Plos One 9:e97339. Che Hasan, R., D. Ierodiaconou, and J. Monk. 2012. Evaluation of Four Supervised Learning Methods for Benthic Habitat Mapping Using Backscatter from Multi-Beam Sonar. Remote Sensing 4:3427-3443. Cheung, W. W. L., T. J. Pitcher, and D. Pauly. 2005. A fuzzy logic expert system to estimate intrinsic extinction vulnerabilities of marine fishes to fishing. Biological Conservation 124:97-111. Claudet, J. and P. Guidetti. 2010. Improving assessments of marine protected areas. Aquatic Conservation-Marine and Freshwater Ecosystems 20:239-242. Claudet, J., C. W. Osenberg, L. Benedetti-Cecchi, P. Domenici, J. A. Garcia-Charton, A. Perez-Ruzafa, F. Badalamenti, J. Bayle-Sempere, A. Brito, F. Bulleri, J. M. Culioli, M. Dimech, J. M. Falcon, I. Guala, M. Milazzo, J. Sanchez-Meca, P. J. Somerfield, B. Stobart, F. Vandeperre, C. Valle, and S. Planes. 2008. Marine reserves: size and age do matter. Ecology Letters 11:481-489. 168 Claudet, J., C. W. Osenberg, P. Domenici, F. Badalamenti, M. Milazzo, J. M. Falcon, I. Bertocci, L. Benedetti-Cecchi, J. A. Garcia-Charton, R. Goni, J. A. Borg, A. Forcada, G. A. de Lucia, A. Perez-Ruzafa, P. Afonso, A. Brito, I. Guala, L. Le Direach, P. Sanchez-Jerez, P. J. Somerfield, and S. Planes. 2010. Marine reserves: Fish life history and ecological traits matter. Ecological Applications 20:830-839. Cloutier, R. N. 2011. Direct and Indirect Effects of Marine Protection: Rockfish Conservation Areas as a Case Study. Simon Fraser University. Cook, S. E., K. W. Conway, and B. Burd. 2008. Status of the glass sponge reefs in the Georgia Basin. Marine Environmental Research 66:S80-S86. COSEWIC. 2008. COSEWIC assessment and status report on the Yelloweye Rockfish Sebastes ruberrimus, Pacific Ocean inside waters population and Pacific Ocean outside waters population, in Canada. Committee on the Status of Endangered Wildlife in Canada, Ottawa. COSEWIC. 2009. COSEWIC assessment and status report on the Quillback Rockfish Sebastes maliger in Canada. Committee on the Status of Endangered Wildlife in Canada, Ottawa. Cote, I. M., I. Mosqueira, and J. D. Reynolds. 2001. Effects of marine reserve characteristics on the protection of fish populations: a meta-analysis. Journal of Fish Biology 59:178-189. Council, N. R. 2001. Marine Protected Areas: tools for sustaining ocean ecosystems. Committee on the evaluation, design and monitoring of marine reserves and protected areas in the United States., Ocean Studies Board, National Research Council, Washington, DC. Crowder, L. B., S. L. Lyman, W. Figueira, and J. Priddy. 2000. Source-sink population dynamics and the problem of siting marine reserves. Bulletin of Marine Science 66:799-820. Cutler, D. R., T. C. Edwards, K. H. Beard, A. Cutler, K. T. Hess, J. Gibson, and J. J. Lawler. 2007. Random forests for classification in Ecology. Ecology 88:2783-2792. Davis, N. A. 2008. Evaluating collaborative fisheries management planning: A Canadian case study. Marine Policy 32:867-876. DeMartini, E. E., J. L. K. Wren, and D. R. Kobayashi. 2013. Persistent spatial patterns of recruitment in a guild of Hawaiian coral reef fishes. Marine Ecology Progress Series 485:165-U200. DFO. 2007. State of the Pacific Ocean 2006. . Can. Sci. Advis. Sec. Sci. Advis. Rep., DFO. Dick, S., J. B. Shurin, and E. B. Taylor. 2014. Replicate divergence between and within sounds in a marine fish: the copper rockfish (Sebastes caurinus). Molecular Ecology 23:575-590. Diesing, M., S. L. Green, D. Stephens, R. M. Lark, H. A. Stewart, and D. Dove. 2014. Mapping seabed sediments: Comparison of manual, geostatistical, object-based image analysis and machine learning approaches. Continental Shelf Research 84:107-119. Du Preez, C. 2015. A new arc–chord ratio (ACR) rugosity index for quantifying three-dimensional landscape structural complexity. Landscape Ecology 30:181-192. Dulvy, N. K., J. R. Ellis, N. B. Goodwin, A. Grant, J. D. Reynolds, and S. Jennings. 2004. Methods of assessing extinction risk in marine fishes. Fish and Fisheries 5:255-276. Dulvy, N. K., Y. Sadovy, and J. D. Reynolds. 2003. Extinction vulnerability in marine populations. Fish and Fisheries 4:25-64. Dunn, D. C. and P. N. Halpin. 2009. Rugosity-based regional modeling of hard-bottom habitat. Marine Ecology-Progress Series 377:1-11. Edgar, G. J., R. D. Stuart-Smith, T. J. Willis, S. Kininmonth, S. C. Baker, S. Banks, N. S. Barrett, M. A. Becerro, A. T. F. Bernard, J. Berkhout, C. D. Buxton, S. J. Campbell, A. 169 T. Cooper, M. Davey, S. C. Edgar, G. Forsterra, D. E. Galvan, A. J. Irigoyen, D. J. Kushner, R. Moura, P. E. Parnell, N. T. Shears, G. Soler, E. M. A. Strain, and R. J. Thomson. 2014. Global conservation outcomes depend on marine protected areas with five key features. Nature 506:216-220. Ellis, N., S. J. Smith, and C. R. Pitcher. 2012. Gradient forests: calculating importance gradients on physical predictors. Ecology 93:156-168. English, K., G. F. Searing, and D. A. Nagtegaal. 2002. Review of the Strait of Georgia Recreational Creel Survey, 1983-1999. Fisheries and Oceans Canada. ESRI. 2008. Georeferencing a raster dataset. ArcGIS Resources. ESRI. 2011. ArcMap 10.2.2. ESRI Inc. ESRI. 2012. How Kernel Density Works. ArcGIS Resources. Favaro, B., S. D. Duff, and I. M. Côté. 2013. A trap with a twist: evaluating a bycatch reduction device to prevent rockfish capture in crustacean traps. ICES Journal of Marine Science: Journal du Conseil 70:114-122. Favaro, B., D. T. Rutherford, S. D. Duff, and I. M. Côté. 2010. Bycatch of rockfish and other species in British Columbia spot prawn traps: Preliminary assessment using research traps. Fisheries Research 102:199-206. Fawcett, T. 2006. An introduction to ROC analysis. Pattern Recogn. Lett. 27:861-874. Field, J. C., A. E. Punt, R. D. Methot, and C. J. Thomson. 2006. Does MPA mean 'major problem for assessments'? Considering the consequences of place-based management systems. Fish and Fisheries 7:284-302. Fisheries and Oceans Canada. 2008. Rockfish Conservation Areas Booklet. Fisheries and Oceans Canada. 2012. Survey of Recreational Fishing in Canada 2010. DFO/2012-1804. Fox, J. and S. Weisberg. 2011. An {R} Companion to Applied Regression, Second Edition. Sage, Thousand Oaks CA. Frable, B. W., D. Wagman, T. N. Frierson, A. Aguilar, and B. L. Sidlauskas. 2015. A new species of Sebastes (Scorpaeniformes: Sebastidae) from the northeastern Pacific, with a redescription of the blue rockfish, S. mystinus (Jordan and Gilbert, 1881). Fishery Bulletin 113:355-377. Freeman, E. A. and G. G. Moisen. 2008. A comparison of the performance of threshold criteria for binary classification in terms of predicted prevalence and kappa. Ecological Modelling 217:48-58. Freeman, E. A., G. G. Moisen, and T. S. Frescino. 2012. Evaluating effectiveness of down-sampling for stratified designs and unbalanced prevalence in Random Forest models of tree species distributions in Nevada. Ecological Modelling 233:1-10. Fujitani, M. L., E. P. Fenichel, J. Torre, and L. R. Gerber. 2012. Implementation of a marine reserve has a rapid but short-lived effect on recreational angler use. Ecological Applications 22:597-605. Gaines, S., C. White, M. Carr, and S. R. Palumbi. 2010a. Designing marine reserve networks for both conservation and fisheries management. Proceedings of the National Academy of Sciences 107:18286-18293. Gaines, S. D., B. Gaylord, and J. L. Largier. 2003. Avoiding current oversights in marine reserve design. Ecological Applications 13:32-46. 170 Gaines, S. D., S. E. Lester, K. Grorud-Colvert, C. Costello, and R. Pollnac. 2010b. Evolving science of marine reserves: New developments and emerging research frontiers. Proceedings of the National Academy of Sciences 107:18251-18255. Gelman, A. and Y.-S. Su. 2013. arm: Data Analysis Using Regression and Multilevel/Hierarchical Models. R package version 1.6-10. Gillanders, B. M., K. W. Able, J. A. Brown, D. B. Eggleston, and P. F. Sheridan. 2003. Evidence of connectivity between juvenile and adult habitats for mobile marine fauna: an important component of nurseries. Marine Ecology Progress Series 247:281-295. Glasby, T. M. 1997. Analysing data hom post-impact studies using asymmetrical analyses of variance: A case study of epibiota on marinas. Australian Journal of Ecology 22:448-459. Gleason, M., E. Fox, S. Ashcraft, J. Vasques, E. Whiteman, P. Serpa, E. Saarman, M. Caldwell, A. Frimodig, M. Miller-Henson, J. Kirlin, B. Ota, E. Pope, M. Weber, and K. Wiseman. 2013. Designing a network of marine protected areas in California: Achievements, costs, lessons learned, and challenges ahead. Ocean & Coastal Management 74:90-101. Granek, E. F., E. M. P. Madin, M. A. Brown, W. Figueira, D. S. Cameron, Z. Hogan, G. Kristianson, P. De Villiers, J. E. Williaims, J. Post, S. Zahn, and R. Arlinghaus. 2008. Engaging Recreational Fishers in Management and Conservation: Global Case Studies. Conservation Biology 22:1125-1134. Green, K. M. and R. M. Starr. 2011. Movements of small adult black rockfish: implications for the design of MPAs. Marine Ecology Progress Series 436:219-230. Gregr, E. J., J. Lessard, and J. Harper. 2013. A spatial framework for representing nearshore ecosystems. Progress In Oceanography 115:189-201. Grorud-Colvert, K., S. E. Lester, S. Airamé, E. Neeley, and S. D. Gaines. 2010. Communicating marine reserve science to diverse audiences. Proceedings of the National Academy of Sciences 107:18306-18311. Grorud-Colvert, K. and S. Sponaugle. 2009. Larval supply and juvenile recruitment of coral reef fishes to marine reserves and non-reserves of the upper Florida Keys, USA. Marine Biology 156:277-288. Gunderson, D. R., A. M. Parma, R. Hilborn, J. M. Cope, D. L. Fluharty, M. L. Miller, R. D. Vetter, S. S. Heppell, and H. G. Greene. 2008. The Challenge of Managing Nearshore Rocky Reef Resources. Fisheries 33:172-179. Haggarty, D. R. 2014. Rockfish Conservation Areas in B.C.: Our current state of knowledge. Haggarty, D. R., R. Flemming, and K. L. Yamanaka. In Preparation. Remotely Operated Vehicle Surveys of Rockfish Conservation Areas in British Columbia, February 2009-July 2011. Canadian Technical Report of Fisheries and Aquatic Sciences. Haggarty, D. R., S. J. D. Martell, and J. B. Shurin. Accepted. Lack of recreational fishing compliance may compromise effectiveness of Rockfish Conservation Areas in British Columbia. Canadian Journal Fisheries Aquatic Sciences. Haldorson, L. J. and M. Love. 1991. Maturity and fecundity in the rockfishes, Sebastes spp., a review. Marine Fisheries Review 53:25-31. Halpern, B. S. 2003. The impact of marine reserves: do reserves work and does reserve size matter? Ecological Applications 13:117-137. Halpern, B. S., S. E. Lester, and K. L. McLeod. 2010. Placing marine protected areas onto the ecosystem-based management seascape. Proceedings of the National Academy of Sciences 107:18312-18317. 171 Halpern, B. S. and R. R. Warner. 2002. Marine reserves have rapid and lasting effects. Ecology Letters 5:361-366. Halpern, B. S. and R. R. Warner. 2003. Matching marine reserve design to reserve objectives. Proceedings of the Royal Society B-Biological Sciences 270:1871-1878. Hamilton, S. L., J. E. Caselle, C. A. Lantz, T. L. Egloff, E. Kondo, S. D. Newsome, K. Loke-Smith, D. J. Pondella, K. A. Young, and C. G. Lowe. 2011. Extensive geographic and ontogenetic variation characterizes the trophic ecology of a temperate reef fish on southern California (USA) rocky reefs. Marine Ecology Progress Series 429:227-244. Hamilton, S. L., J. E. Caselle, D. P. Malone, and M. H. Carr. 2010. Incorporating biogeography into evaluations of the Channel Islands marine reserve network. Proceedings of the National Academy of Sciences 107:18272-18277. Hannah, R. W. and K. M. Matteson. 2007. Behavior of nine species of Pacific rockrish after hook-and-line capture, recompression, and release. Transactions of the American Fisheries Society 136:24-33. Hastings, A. and L. W. Botsford. 2003. Comparing designs of marine reserves for fisheries and for biodiversity. Ecological Applications 13:S65-S70. Hedges, L. V., J. Gurevitch, and P. S. Curtis. 1999. The meta-analysis of response ratios in experimental ecology. Ecology 80:1150-1156. Hijmans, R. J. 2015. raster: Geographic data analysis and modeling. R package version 2.3-24. Hilborn, R. and C. Walters. 1992. Quantitative Fisheries Stock Assessment: Choice, Dynamics and Uncertainty. Chapman and Hall, New York. Hill, N. A., V. Lucieer, N. S. Barrett, T. J. Anderson, and S. B. Williams. 2014. Filling the gaps: Predicting the distribution of temperate reef biota using high resolution biological and acoustic data. Estuarine, Coastal and Shelf Science 147:137-147. Hsieh, C. H., A. Yamauchi, T. Nakazawa, and W. F. Wang. 2010. Fishing effects on age and spatial structures undermine population stability of fishes. Aquatic Sciences 72:165-178. Hutchings, J. A. 2001. Conservation biology of marine fishes: perceptions and caveats regarding assignment of extinction risk. Canadian Journal of Fisheries and Aquatic Sciences 58:108-121. Hutchings, J. A. and J. D. Reynolds. 2004. Marine Fish Population Collapses: Consequences for Recovery and Extinction Risk. Bioscience 54:297-309. Iampietro, P. J., R. G. Kvitek, and E. Morris. 2005. Recent advances in automated genus-specific marine habitat mapping enabled by high-resolution multibeam bathymetry. Marine Technology Society Journal 39:83-93. Iampietro, P. J., M. A. Young, and R. G. Kvitek. 2008. Multivariate prediction of rockfish habitat suitability in Cordell Bank National Marine Sanctuary and Del Monte Shalebeds, California, USA. Marine Geodesy 31:359-371. Ingram, T. and J. B. Shurin. 2009. Trait-based assembly and phylogenetic structure in Northeast Pacific rockfish assemblages. Ecology 90:2444-2453. Jackson, J. B. C. 2001. What was natural in the coastal oceans? Proceedings of the National Academy of Sciences of the United States of America 98:5411-5418. Jennings, S. 2001. Patterns and prediction of population recovery in marine reserves. Reviews in Fish Biology and Fisheries 10:209-231. Johnson, D. W. 2006. Predation, habitat complexity, and variation in density-dependent mortality of temperate reef fishes. Ecology 87:1179-1188. 172 Johnson, D. W. 2007. Habitat complexity modifies post-settlement mortality and recruitment dynamics of a marine fish. Ecology 88:1716-1725. Johnson, S. W., M. L. Murphy, and D. J. Csepp. 2003. Distribution, habitat, and behavior of rockfishes, Sebastes spp., in nearshore waters of southeastern Alaska: observations from a remotely operated vehicle. Environmental Biology of Fishes 66:259-270. Karpov, K., M. Bergen, and J. Geibel. 2012. Monitoring fish in California Channel Islands marine protected areas with a remotely operated vehicle: the first five years. Marine Ecology Progress Series 453:159-172. Kelleher, G. and R. Kenchington. 1992. Guidelines for establishing Marine Protected Areas. IUCN, Gland, Switzerland. Keller, A., W. Wakefield, C. Whitmire, B. Horness, M. Bellman, and K. Bosley. 2014. Distribution of demersal fishes along the US west coast (Canada to Mexico) in relation to spatial fishing closures (2003-2011). Marine Ecology Progress Series 501:169-190. Kellner, J. B., I. Tetreault, S. D. Gaines, and R. M. Nisbet. 2007. Fishing the line near marine reserves in single and multispecies fisheries. Ecological Applications 17:1039-1054. Kramer, D. L. and M. R. Chapman. 1999. Implications of fish home range size and relocation for marine reserve function. Environmental Biology of Fishes 55:65-79. Kritzer, J. P. 2004. Effects of noncompliance on the success of alternative designs of marine protected-area networks for conservation and fisheries management. Conservation Biology 18:1021-1031. Kronlund, A. R. and K. L. Yamanaka. 2001. Yelloweye rockfish (Sebastes ruberrimus) life history parameters assessed from areas with contrasting fishing histories. Pages 257-280 in G. H. Kruse, N. Bez, A. Booth, M. W. Dorn, S. Hills, R. N. Lipcius, D. Pelletier, C. Roy, S. J. Smith, and D. Witherell, editors. Spatial Processes and Management of Marine Populations. Kuznetsova, A., P. Bruun Brockhoff, and R. Haubo Bojesen Christensen. 2014. lmerTest: Tests in Linear Mixed Effects Models R package version 2.0-20. . Laidig, T. E., J. R. Chess, and D. F. Howard. 2007. Relationship between abundance of juvenile rockfishes (Sebastes spp.) and environmental variables documented off northern California and potential mechanisms for the covariation. Fishery Bulletin 105:39-48. Laidig, T. E., D. L. Watters, and M. M. Yoklavich. 2009. Demersal fish and habitat associations from visual surveys on the central California shelf. Estuarine Coastal and Shelf Science 83:629-637. Lancaster, D. 2015. Conservation and compliance: A quantitative assessment of recreational fisher compliance in Rockfish Conservation Areas. School of Environmental Studies. Lancaster, D., P. Dearden, and N. C. Ban. Submitted-a. Drivers of recreational fisher compliance in temperate marine conservation areas: A study of RCAs in British Columbia, Canada. Global Ecology and Conservation. Lancaster, D., D. R. Haggarty, and N. C. Ban. In press. Pacific Canada's Rockfish Conservation Areas: using Ostrom's design principles to assess management effectiveness. Ecology and Society. Lancaster, D., D. R. Haggarty, J. Volpe, P. Dearden, and N. C. Ban. Submitted-b. Effectiveness of shore-based remote camera monitoring for quantifying recreational fisher compliance in marine conservation areas. Conservation Biology. 173 Largier, J. L. 2003. Considerations in estimating larval dispersal distances from oceanographic data. Ecological Applications 13:71-89. Le Pape, O., J. Delavenne, and S. Vaz. 2014. Quantitative mapping of fish habitat: A useful tool to design spatialised management measures and marine protected area with fishery objectives. Ocean & Coastal Management 87:8-19. Leaman, B. M. 1976. Association between the black rockfish Sebastes melanops Girard and beds of the giant kelp Macrocystis integrifolia Bory in Barkley Sound, British Columbia. University of British Columbia. Leaman, B. M. 1991. Reproductive styles and life-history variables relative to exploitation and management of Sebastes stocks. Environmental Biology of Fishes 30:253-271. Leisher, C., S. Mangubhai, S. Hess, H. Widodo, T. Soekirman, S. Tjoe, S. Wawiyai, S. N. Larsen, L. Rumetna, A. Halim, and M. Sanjayan. 2012. Measuring the benefits and costs of community education and outreach in marine protected areas. Marine Policy 36:1005-1011. Leslie, H., M. Ruckelshaus, I. R. Ball, S. Andelman, and H. P. Possingham. 2003. Using siting algorithms in the design of marine reserve networks. Ecological Applications 13:S185-S198. Leslie, H. M. 2005. A Synthesis of Marine Conservation Planning Approaches Síntesis de Estrategias de Planificación de Conservación Marina. Conservation Biology 19:1701-1713. Lessard, J. and A. Campbell. 2007. Describing Northern Abalone, Haliotis kamtchatkana, habitat: focusing rebuilding efforts in British Columbia, Canada. Journal of Shellfish Research 26:677-686. Lester, S. E., B. S. Halpern, K. Grorud-Colvert, J. Lubchenco, B. I. Ruttenberg, S. D. Gaines, S. Airame, and R. R. Warner. 2009. Biological effects within no-take marine reserves: a global synthesis. Marine Ecology-Progress Series 384:33-46. Liaw, A. and M. Wiener. 2002. Classification and Regression by randomForest. R News 2:18-22. Lotterhos, K. E., S. J. Dick, and D. R. Haggarty. 2014. Evaluation of rockfish conservation area networks in the United States and Canada relative to the dispersal distance for black rockfish (Sebastes melanops). Evolutionary Applications 7:238-259. Lotterhos, K. E. and R. W. Markel. 2012. Oceanographic drivers of offspring abundance may increase or decrease reproductive variance in a temperate marine fish. Molecular Ecology 21:5009-5026. Lotze, H. K., H. S. Lenihan, B. J. Bourque, R. H. Bradbury, R. G. Cooke, M. C. Kay, S. M. Kidwell, M. X. Kirby, C. H. Peterson, and J. B. C. Jackson. 2006. Depletion, degradation, and recovery potential of estuaries and coastal seas. Science 312:1806-1809. Love, M., M. Yoklavich, and D. Schroeder. 2009. Demersal fish assemblages in the Southern California Bight based on visual surveys in deep water. Environmental Biology of Fishes 84:55-68. Love, M., M. Yoklavich, and L. Thorsteinson. 2002. The Rockfishes of the Northeast Pacific. University of California Press, Los Angeles. Lucieer, V., N. A. Hill, N. S. Barrett, and S. Nichol. 2013. Do marine substrates 'look' and 'sound' the same? Supervised classification of multibeam acoustic data using autonomous underwater vehicle images. Estuarine Coastal and Shelf Science 117:94-106. 174 Magnuson-Ford, K., T. Ingram, D. W. Redding, and A. Ø. Mooers. 2009. Rockfish (Sebastes) that are evolutionarily isolated are also large, morphologically distinctive and vulnerable to overfishing. Biological Conservation 142:1787-1796. Margules, C. R. and R. L. Pressey. 2000. Systematic conservation planning. Nature 405:243-253. Markel, R. W. 2011. Rockfish Recruitment and trophic dynamics on the west coast of Vancouver Island: fishing, ocean climate, and sea otters. University of British Columbia. Marliave, J. and W. Challenger. 2009. Monitoring and evaluating rockfish conservation areas in British Columbia. Canadian Journal of Fisheries and Aquatic Sciences 66:995-1006. Marliave, J. B., K. W. Conway, D. M. Gibbs, A. Lamb, and C. Gibbs. 2009. Biodiversity and rockfish recruitment in sponge gardens and bioherms of southern British Columbia, Canada. Marine Biology 156:2247-2254. Martell, S. J. D., C. J. Walters, and S. S. Wallace. 2000. The use of marine protected areas for conservation of lingcod (Ophiodon elongatus). Bulletin of Marine Science 66:729-743. Matthews, K. R. 1990a. A comparative-study of habitat use by young-of-the-year, subadult, and adult rockfishes on 4 habitat types in Central Puget Sound. Fishery Bulletin 88:223-239. Matthews, K. R. 1990b. An experimental study of the habitat preferences and movement patterns for copper, quillbak, and brown rockfishes (Sebastes spp.). Environmental Biology of Fishes 29:161:178. MCBI. 1998. Troubled waters: a call for action. Marine Conservation Biology Institute. McClanahan, T. R., N. A. J. Graham, S. K. Wilson, Y. Letourneur, and R. Fisher. 2009. Effects of fisheries closure size, age, and history of compliance on coral reef fish communities in the western Indian Ocean. Marine Ecology Progress Series 396:99-109. McCook, L. J., T. Ayling, M. Cappo, J. H. Choat, R. D. Evans, D. M. De Freitas, M. Heupel, T. P. Hughes, G. P. Jones, B. Mapstone, H. Marsh, M. Mills, F. J. Molloy, C. R. Pitcher, R. L. Pressey, G. R. Russ, S. Sutton, H. Sweatman, R. Tobin, D. R. Wachenfeld, and D. H. Williamson. 2010. Adaptive management of the Great Barrier Reef: A globally significant demonstration of the benefits of networks of marine reserves. Proceedings of the National Academy of Sciences 107:18278-18285. McKechnie, I. 2007. Investigating the complexities of sustainable fishing at a prehistoric village on western Vancouver Island, British Columbia, Canada. Journal for Nature Conservation 15:208-222. Miller, K. I. and G. R. Russ. 2014. Studies of no-take marine reserves: Methods for differentiating reserve and habitat effects. Ocean & Coastal Management 96:51-60. Mills, M., R. Weeks, R. L. Pressey, M. G. Gleason, R. L. Eisma-Osorio, A. T. Lombard, J. M. Harris, A. B. Killmer, A. White, and T. H. Morrison. 2015. Real-world progress in overcoming the challenges of adaptive spatial planning in marine protected areas. Biological Conservation 181:54-63. Molloy, P. P., I. B. McLean, and I. M. Cote. 2009. Effects of marine reserve age on fish populations: a global meta-analysis. Journal of Applied Ecology 46:743-751. Mora, C., S. Andrèfouët, M. J. Costello, C. Kranenburg, A. Rollo, J. Veron, K. J. Gaston, and R. A. Myers. 2006. Coral Reefs and the Global Network of Marine Protected Areas. Science 312:1750-1751. Mosqueira, I., I. M. Cote, S. Jennings, and J. D. Reynolds. 2000. Conservation benefits of marine reserves for fish populations. Animal Conservation 3:321-332. 175 Murray, S. N., R. F. Ambrose, J. A. Bohnsack, L. W. Botsford, M. H. Carr, G. E. Davis, P. K. Dayton, D. Gotshall, D. R. Gunderson, M. A. Hixon, J. Lubchenco, M. Mangel, A. MacCall, D. A. McArdle, J. C. Ogden, J. Roughgarden, R. M. Starr, M. J. Tegner, and M. M. Yoklavich. 1999. No-take reserve networks: Sustaining fishery populations and marine ecosystems. Fisheries 24:11-25. National Research Council. 2006. Dynamic Changes in Marine Ecosystems: Fishing, Food Webs, and Future Options. National Academies Press, Washington (DC). Norse, E. 1993. Global Marine Biological Diversity A Strategy for Building Conservation into Decision Making. Island Press, Washington, D.C. O'Farrell, M. R., M. M. Yoklavich, and M. S. Love. 2009. Assessment of habitat and predator effects on dwarf rockfishes (Sebastes spp.) using multi model inference. Environmental Biology of Fishes 85:239-250. Oconnell, V. M. and D. W. Carlile. 1993. Habitat-specific density of adult yelloweye rockfish Sebastes-ruberrimus in the Eastern Gulf of Alaska. Fishery Bulletin 91:304-309. Pacunski, R. E. and W. A. Palsson. 2001. Macro- and micro-habitat relationships of adult and sub-adult Rockfish, Lingcod, and Kelp Greenling in Puget Sound.in Puget Sound Research 2001, Olympia, Washington. Paddack, M. J. and J. A. Estes. 2000. Kelp forest fish populations in marine reserves and adjacent exploited areas of central California. Ecological Applications 10:855-870. Palumbi, S. R. 2003. Population genetics, demographic connectivity, and the design of marine reserves. Ecological Applications 13:S146-S158. Parker, S. J., S. A. Berkeley, J. T. Golden, D. R. Gunderson, J. Heifetz, M. A. Hixon, R. Larson, B. M. Leaman, M. S. Love, J. A. Musick, V. M. O'Connell, S. Ralston, H. J. Weeks, and M. M. Yoklavich. 2000. Management of Pacific rockfish. Fisheries 25:22-30. Parker, S. J., H. I. McElderry, P. S. Rankin, and R. W. Hannah. 2006. Buoyancy regulation and barotrauma in two species of nearshore rockfish. Transactions of the American Fisheries Society 135:1213-1223. Parker, S. J., J. M. Olson, P. S. Rankin, and J. S. Malvitch. 2008. Patterns in vertical movements of black rockfish Sebastes melanops. Aquatic Biology 2:57-65. Parker, S. J., P. S. Rankin, J. M. Olson, and R. W. Hannah. 2007. Movement patterns of black rockfish (Sebastes melanops) in Oregon coastal waters. Pages 39-57 in J. Heifetz, J. Dicosimo, A. J. Gharrett, M. S. Love, V. M. Oconnell, and R. D. Stanley, editors. Biology, Assessment, and Management of North Pacific Rockfishes. Alaska Sea Grant Coll Program, Fairbanks. Parnell, P. E., P. K. Dayton, C. E. Lennert-Cody, L. L. Rasmussen, and J. J. Leichter. 2006. Marine Reserve Design: Optimal Size, Habitats, Species Affinities, Diversity, And Ocean Microclimate. Ecological Applications 16:945-962. Pauly, D., V. Christensen, J. Dalsgaard, R. Froese, and F. Torres. 1998. Fishing down marine food webs. Science 279:860-863. Pauly, D., V. Christensen, S. Guenette, T. J. Pitcher, U. R. Sumaila, C. J. Walters, R. Watson, and D. Zeller. 2002. Towards sustainability in world fisheries. Nature 418:689-695. Pearse, J. S. and H. A.H. 1987. Long-term population dynamics of sea urchins in a central California kelp forest: rare recruitment and rapid decline. Marine Ecology Progress Series 39:275-283. 176 Pelletier, D., J. Claudet, J. Ferraris, L. Benedetti-Cecchi, and J. A. Garcia-Charton. 2008. Models and indicators for assessing conservation and fisheries-related effects of marine protected areas. Canadian Journal of Fisheries and Aquatic Sciences 65:765-779. Pinheiro, J., D. Bates, S. DebRoy, D. Sarkar, and R. D. C. Team. 2012. nlme: Linear and Nonlinear Mixed Effects Models. Planes, S., R. Galzin, A. G. Rubies, R. Goni, J. G. Harmelin, L. Le Direach, P. Lenfant, and A. Quetglas. 2000. Effects of marine protected areas on recruitment processes with special reference to Mediterranean littoral ecosystems. Environmental Conservation 27:126-143. Pollnac, R., P. Christie, J. E. Cinner, T. Dalton, T. M. Daw, G. E. Forrester, N. A. J. Graham, and T. R. McClanahan. 2010. Marine reserves as linked social–ecological systems. Proceedings of the National Academy of Sciences 107:18262-18265. Prasad, A., L. Iverson, and A. Liaw. 2006. Newer Classification and Regression Tree Techniques: Bagging and Random Forests for Ecological Prediction. Ecosystems 9:181-199. Pressey, R. L., M. Mills, R. Weeks, and J. C. Day. 2013. The plan of the day: Managing the dynamic transition from regional conservation designs to local conservation actions. Biological Conservation 166:155-169. R Development Core Team. 2008. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. Ralston, S. and D. F. Howard. 1995. On the development of year-class strength and cohort variability in 2 northern California rockfishes. Fishery Bulletin 93:710-720. Read, A. D., R. J. West, M. Haste, and A. Jordan. 2011. Optimizing voluntary compliance in marine protected areas: A comparison of recreational fisher and enforcement officer perspectives using multi-criteria analysis. Journal of Environmental Management 92:2558-2567. Richards, L. J. 1986. Depth and habitat distributions of 3 species of rockfish (Sebastes) in British-Columbia - observations from the submersible Pisces-iv. Environmental Biology of Fishes 17:13-21. Richards, L. J. 1987. Copper rockfish (Sebastes-caurinus) and quillback rockfish (Sebastes-maliger) habitat in the Strait of Georgia, British-Columbia. Canadian Journal of Zoology-Revue Canadienne De Zoologie 65:3188-3191. Rinne, H., A. Kaskela, A.-L. Downie, H. Tolvanen, M. von Numers, and J. Mattila. 2014. Predicting the occurrence of rocky reefs in a heterogeneous archipelago area with limited data. Estuarine, Coastal and Shelf Science 138:90-100. Robb, C. K., K. M. Bodtker, K. Wright, and J. Lash. 2011. Commercial fisheries closures in marine protected areas on Canada's Pacific coast: The exception, not the rule. Marine Policy 35:309-316. Roberts, C. M. 2000. Selecting marine reserve locations: optimality versus opportunism. Bulletin of Marine Science 66:581-592. Roberts, C. M., S. Andelman, G. Branch, R. H. Bustamante, J. C. Castilla, J. Dugan, B. S. Halpern, K. D. Lafferty, H. Leslie, J. Lubchenco, D. McArdle, H. P. Possingham, M. Ruckelshaus, and R. R. Warner. 2003a. Ecological criteria for evaluating candidate sites for marine reserves. Ecological Applications 13:S199-S214. Roberts, C. M., G. Branch, R. H. Bustamante, J. C. Castilla, J. Dugan, B. S. Halpern, K. D. Lafferty, H. Leslie, J. Lubchenco, D. McArdle, M. Ruckelshaus, and R. R. Warner. 177 2003b. Application of ecological criteria in selecting marine reserves and developing reserve networks. Ecological Applications 13:S215-S228. Roberts, J. J., B. D. Best, D. C. Dunn, E. A. Treml, and P. N. Halpin. 2010. Marine Geospatial Ecology Tools: An integrated framework for ecological geoprocessing with ArcGIS, Python, R, MATLAB, and C plus. Environmental Modelling & Software 25:1197-1207. Russ, G. R. and A. C. Alcala. 2011. Enhanced biodiversity beyond marine reserve boundaries: The cup spillith over. Ecological Applications 21:241-250. Russ, G. R., B. Stockwell, and A. C. Alcala. 2005. Inferring versus measuring rates of recovery in no-take marine reserves. Marine Ecology Progress Series 292:1-12. Sala, E., O. Aburto-Oropeza, G. Paredes, I. Parra, J. C. Barrera, and P. K. Dayton. 2002. A general model for designing networks of marine reserves. Science 298:1991-1993. Sale, P. F., R. K. Cowen, B. S. Danilowicz, G. P. Jones, J. P. Kritzer, K. C. Lindeman, S. Planes, N. V. C. Polunin, G. R. Russ, Y. J. Sadovy, and R. S. Steneck. 2005. Critical science gaps impede use of no-take fishery reserves. Trends in Ecology & Evolution 20:74-80. Secretatiat of the Convention on Biological Diversity. 2011. Strategic Plan for Biodiversity 2011-2020 and the Aichi Targets. Sethi, S. A. and R. Hilborn. 2008. Interactions between poaching and management policy affect marine reserves as conservation tools. Biological Conservation 141:506-516. Smallwood, C. B. and L. E. Beckley. 2012. Spatial distribution and zoning compliance of recreational fishing in Ningaloo Marine Park, north-western Australia. Fisheries Research 125:40-50. Smith, R. J., P. D. Eastwood, Y. Ota, and S. I. Rogers. 2009. Developing best practice for using Marxan to locate Marine Protected Areas in European waters. ICES Journal of Marine Science: Journal du Conseil 66:188-194. Sogard, S. M., S. A. Berkeley, and R. Fisher. 2008. Maternal effects in rockfishes Sebastes spp.: a comparison among species. Marine Ecology-Progress Series 360:227-236. Starr, R. M., D. E. Wendt, C. L. Barnes, C. I. Marks, D. Malone, G. Waltz, K. T. Schmidt, J. Chiu, A. L. Launer, N. C. Hall, and N. Yochum. 2015. Variation in Responses of Fishes across Multiple Reserves within a Network of Marine Protected Areas in Temperate Waters. Plos One 10:e0118502. Stelzenmüller, V., F. Maynou, G. Bernard, G. Cadiou, M. Camilleri, R. Crec'hriou, G. Criquet, M. Dimech, O. Esparza, R. Higgins, P. Lenfant, and A. Perez-Ruzafa. 2008. Spatial assessment of fishing effort around European marine reserves: Implications for successful fisheries management. Marine Pollution Bulletin 56:2018-2026. Stephens, D. and M. Diesing. 2014. A Comparison of Supervised Classification Methods for the Prediction of Substrate Type Using Multibeam Acoustic and Legacy Grain-Size Data. Plos One 9. Stoner, A. W., C. H. Ryer, S. J. Parker, P. J. Auster, and W. W. Wakefield. 2008. Evaluating the role of fish behavior in surveys conducted with underwater vehicles. Canadian Journal of Fisheries and Aquatic Sciences 65:1230-1243. Thomson, R. E. 1981. Oceanography of the British Columbia Coast. Thrush, S. F. and P. K. Dayton. 2010. What Can Ecology Contribute to Ecosystem-Based Management? Annual Review of Marine Science 2:419-441. Tinis, S. 2009. Storm Surge Forecasting for Southwestern British Columbia: Fall/Winter 2008-2009. Fisheries and Oceans Canada. 178 Tinis, S. 2010. Storm surge climatology for Southwestern British Columbia: fall/winter 2010-2011. Fisheries and Oceans Canada. Tinus, C. A. 2012. Prey preference of lingcod (Ophiodon elongatus), a top marine predator: implications for ecosystem-based fisheries management. Fishery Bulletin 110:193-204. Underwood, A. J. 1992. Beyond BACI - the detection of environmental impacts on populations in the real, but variable, world. Journal of Experimental Marine Biology and Ecology 161:145-178. Walters, C. 1986. Adaptive management of renewable resources. Macmillan, New York. Warner, R. R. and P. L. Chesson. 1985. Coexistence Mediated by Recruitment Fluctuations: A Field Guide to the Storage Effect. The American Naturalist 125:769-787. Wen, C. K. C., G. R. Almany, D. H. Williamson, M. S. Pratchett, T. D. Mannering, R. D. Evans, J. M. Leis, M. Srinivasan, and G. P. Jones. 2013. Recruitment hotspots boost the effectiveness of no-take marine reserves. Biological Conservation 166:124-131. White, J. W., L. W. Botsford, M. L. Baskett, L. A. K. Barnett, R. J. Barr, and A. Hastings. 2011. Linking models with monitoring data for assessing performance of no-take marine reserves. Frontiers in Ecology and the Environment 9:390-399. Williams, G. D., P. S. Levin, and W. A. Palsson. 2010. Rockfish in Puget Sound: An ecological history of exploitation. Marine Policy 34:1010-1020. Williamson, D. H., D. M. Ceccarelli, R. D. Evans, J. K. Hill, and G. R. Russ. 2014. Derelict Fishing Line Provides a Useful Proxy for Estimating Levels of Non-Compliance with No-Take Marine Reserves. Plos One 9. Wood, S. N. 2011. Fast stable restricted maximum likelihood and marginal likelihood estimation of semiparametric generalized linear models. Journal of the Royal Statistical Society. Series B (Methodological) (B) 73:3-36. Worm, B., R. Hilborn, J. K. Baum, T. A. Branch, J. S. Collie, C. Costello, M. J. Fogarty, E. A. Fulton, J. A. Hutchings, S. Jennings, O. P. Jensen, H. K. Lotze, P. M. Mace, T. R. McClanahan, C. Minto, S. R. Palumbi, A. M. Parma, D. Ricard, A. A. Rosenberg, R. Watson, and D. Zeller. 2009. Rebuilding Global Fisheries. Science 325:578-585. Wright, D. J., M. Pendleton, J. Boulware, S. Walbridge, B. Gerlt, D. Eslinger, D. Sampson, and E. Huntley. 2012. ArcGIS Benthic Terrain Modeler (BTM), v. 3.0, Environmental Systems Research Institute, NOAA Coastal Services Center, Massachusetts Office of Coastal Zone Management. . Available online at http://esriurl.com/5754. Yamanaka, K. and G. Logan. 2010. Developing British Columbia’s Inshore Rockfish Conservation Strategy. Marine and Coastal Fisheries: Dynamics, Management, and Ecosystem Science 2:28-46. Yamanaka, K. L. and R. Flemming. 2013. Development of spatial management tools to address fisheries conservation concerns for quillback rockfish (Sebastes maliger) in British Columbia Canada. Page 246 in T. Nishida and A. E. Caton, editors. GIS/Spatial Analyses in Fishery and Aquatic Sciences. International Fishery GIS Society, Saitama, Japan. Yamanaka, K. L. and L. Lacko. 2001. Inshore rockfish (Sebastes ruberrimus, S. maliger, S. caurinus, S. melanops, S. nigrocinctus and S. nebulosus) stock assessment for the West Coast of Canada and recommendations for management., Canadian Science Advisory. Yamanaka, K. L., K. Picard, K. W. Conway, and R. Flemming. 2012. Rock Reefs of British Columbia, Canada: Inshore Rockfish Habitats.in P. T. Harris and E. K. Baker, editors. Seafloor Geomorphology as Benthic Habitat. Elsevier. 179 Yoklavich, M. 1998. Marine harvest refugia for West Coast rockfish: a workshop. NOAA-TM-NMFS-SWFSC-255, NOAA Technical Memorandum NMFS, Pacific Grove, California. Yoklavich, M., G. Cailliet, R. N. Lea, H. G. Greene, R. Starr, J. De Marignac, and J. Field. 2002. Deepwater habitat and fish resources associated with the Big Creek Marine Ecological Reserve. California Cooperative Oceanic Fisheries Investigations Reports 43:120-140. Yoklavich, M. M., H. G. Greene, G. M. Cailliet, D. E. Sullivan, R. N. Lea, and M. S. Love. 2000. Habitat associations of deep-water rockfishes in a submarine canyon: an example of a natural refuge. Fishery Bulletin 98:625-641. Yoklavich, M. M., V. J. Loeb, M. Nishimoto, and B. Daly. 1996. Nearshore assemblages of larval rockfishes and their physical environment off central California during an extended El Nino event, 1991-1993. Fishery Bulletin 94:766-782. Yoklavich, M. M., M. S. Love, and K. A. Forney. 2007. A fishery-independent assessment of an overfished rockfish stock, cowcod (Sebastes levis), using direct observations from an occupied submersible. Canadian Journal of Fisheries and Aquatic Sciences 64:1795-1804. Young, M. A., P. J. Iampietro, R. G. Kvitek, and C. D. Garza. 2010. Multivariate bathymetry-derived generalized linear model accurately predicts rockfish distribution on Cordell Bank, California, USA. Marine Ecology-Progress Series 415:247-261. Zar, J.H. 1996. Biostatistical Analysis. 3rd Edition. Prentice Hall, Upper Saddle River, New Jersey. Zetterberg, P. R., N. M. Watson, and D. S. O'Brien. 2012a. Strait of Georgia Recreational Fishery Statistics for Salmon and Groundfish, 2009. Zetterberg, P. R., N. M. Watson, and D. S. O'Brien. 2012b. Strait of Georgia Recreational Fishery Statistics for Salmon and Groundfish, 2010. Zetterberg, P. R., N. M. Watson, and D. S. O'Brien. In Preparation. Strait of Georgia Recreational Fishery Statistics for Salmon and Groundfish, 2011. Zurr, A. F., E. N. Leno, N. Walker, A. A. Savaliev, and G. M. Smith. 2009. Mixed Effects Models and Extensions in Ecology with R. Springer, New York. Zvoleff, A. 2015. glcm: Calculate textures from grey-level co-occurrence matrices (GLCMs) in R. 180 Appendices Appendix A Supplementary Information from Chapter 2. Does Post-settlement Recruitment Predict the Adult Abundance of Black Rockfish? Appendix Table 1-1. YOY and 1 yr Black Rockfish recruitment by region and year (data from R. Markel and K. Lotterhos). 2006 2007 2008 2009 Region Mean SD Mean SD Mean SD Mean SD Black Rockfish YOY BG-in 1.15 1.77 0.67 1.28 2.00 2.83 0.27 0.55 BG-out 3.00 3.46 0.59 1.33 2.43 2.70 0.47 0.74 DG 4.64 4.50 0.74 1.37 0.55 0.87 0.49 1.71 GF 5.38 7.17 0.84 1.44 0.38 0.74 0.20 0.46 LO 10.30 17.80 1.00 1.68 0.76 1.54 0.82 2.04 PE 4.96 8.98 0.76 1.37 2.06 5.09 0.50 1.18 All 6.27 12.22 0.80 1.44 1.05 2.50 0.48 1.38 Black Rockfish 1-Yr BG-in 0.00 0.00 0.00 0.00 0.00 0.00 0.05 0.21 BG-out 0.23 0.44 0.05 0.21 0.29 0.76 0.33 1.05 DG 0.00 0.00 0.05 0.22 0.00 0.00 0.21 0.61 GF 0.00 0.00 0.03 0.16 0.10 0.30 0.37 0.86 LO 0.00 0.00 0.00 0.00 0.00 0.00 0.07 0.25 PE 0.04 0.20 0.35 1.25 0.00 0.00 0.36 0.90 All 0.02 0.16 0.08 0.54 0.04 0.23 0.23 0.70 181 Appendix Table 1-2. Number of samples per region and site by year. Recruitment (SMURF) Survey Dive Transects Site 2006 2007 2008 2009 2010 2011 Broken Group-In 13 19 9 22 10 20 Chalk 7 2 4 Faber 6 3 7 2 4 Howell 6 2 7 2 4 Pressure 7 4 8 2 4 Turrett 6 2 4 Broken Group-Out 13 19 7 15 8 15 Austin 5 3 2 4 Clarke 6 7 3 7 2 3 Effingham 7 4 5 2 4 Hand 7 2 4 Deer Group 16 39 29 39 12 24 Diana 3 5 5 5 2 4 EK1 3 6 5 7 2 4 EK2 3 7 5 8 2 4 Kirby 2 7 5 6 2 4 Seppings 3 7 4 8 2 4 Taylor 2 7 5 5 2 4 George Fraser 18 38 21 41 10 12 GF1 3 6 2 5 2 4 GF2 3 7 4 8 2 0 GF3 3 6 4 7 2 4 GF4 3 6 3 6 2 0 GF5 3 7 4 8 0 0 GF6 3 6 4 7 2 4 Loudoun 39 43 25 44 12 24 Ally 6 6 4 7 2 4 Bocco 6 7 4 8 2 4 Chrow 7 7 4 5 2 4 GreatBear 6 8 4 8 2 4 LoudounChannel 7 8 5 8 2 4 Page 7 7 4 8 2 4 Prasex 18 34 17 36 12 24 Blowhole 3 6 2 6 2 4 Bluestone 3 6 3 7 2 4 Execution 3 4 2 5 2 4 Nudibranch 3 7 3 6 2 4 Prasiola 3 6 3 5 2 4 Whittlestone 3 5 4 7 2 4 182 Appendix B Supplementary Information from Chapter 3. Assessing Population Recovery Inside British Columbia’s Rockfish Conservation Areas with a Remotely Operated Vehicle Appendix Figure 2-1a. The mean and SE of the percent of habitat type on by region and inside and outside of the RCAs of the RCAs with positive RRs. R1=Bare Rock, R2=Rock with encrusting biota, R3=Rock with emergent biota; C1=Bare Coarse substrates, C2=Coarse with encrusting biota, C3=Coarse with emergent biota; F1=Bare Fine substrate, F2=Fine with encrusting biota, F3=Fine with emergent biota; CX=High Complexity, RL=High Relief, D=Depth (in m). 183 Appendix Figure 2-1b. The mean and SE of the percent of habitat type on by region and inside and outside of the RCAs with negative RRs. R1=Bare Rock, R2=Rock with encrusting biota, R3=Rock with emergent biota; C1=Bare Coarse substrates, C2=Coarse with encrusting biota, C3=Coarse with emergent biota; F1=Bare Fine substrate, F2=Fine with encrusting biota, F3=Fine with emergent biota; CX=High Complexity, RL=High Relief, D=Depth (in m). 184 Appendix Figure 2-1c. The mean and SE of the percent of habitat type on by region and inside and outside of the RCAs with neutral RRs. R1=Bare Rock, R2=Rock with encrusting biota, R3=Rock with emergent biota; C1=Bare Coarse substrates, C2=Coarse with encrusting biota, C3=Coarse with emergent biota; F1=Bare Fine substrate, F2=Fine with encrusting biota, F3=Fine with emergent biota; CX=High Complexity, RL=High Relief, D=Depth (in m). 185 Appendix Figure 2-1d. The mean and SE of the percent of habitat type on by region and inside and outside of the RCAs with neutral RRs. R1=Bare Rock, R2=Rock with encrusting biota, R3=Rock with emergent biota; C1=Bare Coarse substrates, C2=Coarse with encrusting biota, C3=Coarse with emergent biota; F1=Bare Fine substrate, F2=Fine with encrusting biota, F3=Fine with emergent biota; CX=High Complexity, RL=High Relief, D=Depth (in m). 186 Appendix Figure 2-2. The mean log RR (response ratio) of Quillback, Yelloweye and Lingcod of density on total area of the transects inside to outside of all RCAs sampled. Error Bars are Standard Errors. Ratios greater than zero indicate greater densities inside the RCA. 187 Appendix Figure 2-3. Log RR (Response Ratio) by RCA for each species/species group. 188 Appendix Figure 2-4. Boxplots of the log Response Ratio by Species/Species Group and by Region (JS=Johnstone Strait, QCST=Queen Charlotte Strait, SG=Strait of Georgia, WCVI=West Coast of Vancouver Island). 189 Appendix Figure 2-5. Log Response Ratio (RR) versus log commerical catch +1 in the RCAs before establishment and outside of each RCA before and after establishement for each species/group by Region. Commercial data are the total catch (kg) of each species/group from the trawl and hook and line fisheries. Data from the period before the RCAs were established are from 1996-2006 for the trawl fishery and 2002-2006 from the hook and line fisheries; data from the period after establishement are from 2007-2011 for all fisheries. Region key: Johnstone Strait=Red, Queen Charlotte Strait=Blue, Strait of Georgia=Green, West Coast of Vancouver Island=Purple. 190 Appendix Table 2-1. Mean and Standard Error of Fish Densities (#/100m2) inside and outside of RCAs observed on ROV surveys. QB YE IR LC Name Age RCA N SE SE SE SE SG 2009 Brethour 3 In 6 0.12 0.08 0.01 0.01 0.27 0.10 0.09 0.04 Out 3 0.57 0.19 0 0 0.61 0.17 0.16 0.16 Domett-Pam 4 In 9 0.21 0.06 0.01 0.01 0.27 0.07 0.04 0.03 Out 5 0.26 0.10 0.07 0.05 0.33 0.14 0 0 Halibut-McCall 3 In 8 0.15 0.09 0.04 0.03 0.19 0.12 0.20 0.15 Out 4 0.12 0.02 0.02 0.02 0.15 0.02 0.10 0.04 Northumberland 3 In 3 0.55 0.12 0.17 0.07 0.78 0.19 0.09 0.02 Out 6 0.53 0.07 0.08 0.04 0.70 0.13 0.51 0.19 Saltspring-Trincomali 3 In 14 0.31 0.12 0 0 0.53 0.19 0.10 0.03 Out 3 1.91 1.18 0.06 0.06 2.13 1.27 0.09 0.09 Brethour 3 In 6 0.75 0.11 0.08 0.08 1.27 0.43 0.10 0.06 Out 4 0.53 0.16 0 0 0.70 0.13 0.26 0.06 Darcy 3 In 3 1.03 0.40 0.10 0.05 1.34 0.37 0.31 0.17 Out 3 0 0 0 0 0.06 0.06 0.17 0.08 Desolation 3 In 13 0.97 0.22 0.26 0.06 1.27 0.25 0.06 0.03 Out 10 0.71 0.21 0.28 0.12 0.98 0.25 0.02 0.01 Dinner 3 In 4 0.31 0.07 0.04 0.02 0.36 0.07 0.17 0.02 Out 3 1.21 0.60 0.01 0.01 1.24 0.60 0.09 0.09 Halibut-McCall 3 In 6 0.15 0.07 0.05 0.03 0.19 0.09 0.06 0.06 Out 5 0.25 0.11 0.11 0.06 0.36 0.12 0.04 0.02 Hardy 3 In 2 0.79 0.18 0.24 0.24 1.03 0.42 0.08 0.08 Out 2 1.14 0.50 0.45 0.10 1.59 0.59 0.09 0.09 Northumberland 3 In 3 0.62 0.11 0.29 0.06 0.92 0.17 0.12 0.07 Out 3 1.45 0.59 0.15 0.02 1.75 0.61 0.23 0.09 Prevost 3 In 7 0.57 0.09 0.08 0.06 1.10 0.12 0.14 0.06 Out 5 0.53 0.21 0.07 0.07 0.70 0.20 0.13 0.06 Sabine 3 In 3 0.74 0.70 0.36 0.13 1.10 0.82 0.05 0.05 Out 3 0.63 0.58 0.36 0.20 0.99 0.64 0.09 0.05 Saltspring-Trincomali 3 In 15 0.40 0.14 0.02 0.01 0.55 0.15 0.15 0.05 Out 5 0.92 0.29 0.20 0.12 1.31 0.44 0.03 0.03 Valdes 3 In 3 2.00 1.06 0.31 0.31 2.39 0.84 0.42 0.24 Out 3 1.14 0.06 0 0 1.19 0.10 0.13 0.08 SG 2010 Ballenas 6 In 3 0.14 0.10 0.01 0.01 0.15 0.11 0.07 0.04 Out 2 0.04 0.04 0 0 0.04 0.04 0.11 0.11 Northumberland 4 In 3 0.77 0.13 0.03 0.03 0.86 0.12 0.13 0.09 Out 3 1.27 0.36 0.26 0.09 1.56 0.45 0.21 0.13 191 QB YE IR LC Name Age RCA N SE SE SE SE Valdes 4 In 2 0.53 0.27 0.04 0.04 0.57 0.23 0.08 0.08 Out 2 0.53 0.03 0 0 0.53 0.03 0.12 0.12 Northumberland 4 In 7 1.59 0.34 0.28 0.09 2.00 0.46 0.17 0.10 Out 5 1.02 0.18 0.13 0.04 1.18 0.17 0.18 0.02 SG 2011 Brethour 5 In 2 0.38 0.22 0.04 0.04 0.50 0.24 0.03 0.03 Out 2 0.47 0.27 0 0 1.08 0.82 0.09 0.03 JS 2010 Chancellor 4 In 2 0.52 0.52 0.07 0.07 0.62 0.62 0 0 Out 2 0.26 0.26 0.07 0.07 0.33 0.33 0.07 0.07 Clio-Viscount 4 In 3 2.10 1.60 0 0 2.10 1.60 0.16 0.16 Out 5 2.26 0.67 0.01 0.01 2.27 0.66 0 0 Dickson 4 In 2 0.78 0.40 0 0 0.78 0.40 0.15 0.03 Out 2 1.01 0.02 0.07 0.01 1.07 0.02 0.07 0.07 Octopus-Read 4 In 3 0.62 0.11 0.15 0.10 0.77 0.22 0.08 0.08 Out 4 0.71 0.27 0.31 0.05 1.02 0.23 0 0 Thurston 6 In 3 0.80 0.24 0 0 0.80 0.24 0 0 Out 2 1.15 0.10 0.20 0.20 1.40 0.04 0.28 0.28 QCST 2011 Bate 5 In 3 1.99 0.77 0.75 0.33 3.22 1.07 0.03 0.03 Out 3 2.89 0.88 0.29 0.08 3.78 0.94 0.17 0.09 Bolivar 5 In 6 0.29 0.07 0.19 0.12 0.50 0.09 0.04 0.02 Out 4 0.28 0.06 0.03 0.02 0.31 0.08 0.03 0.02 Browning 5 In 5 1.07 0.38 0.14 0.07 1.49 0.50 0.08 0.06 Out 5 0.30 0.09 0.11 0.06 0.46 0.13 0.08 0.03 Goletas 7 In 2 0.90 0.56 0.26 0.26 1.43 1.09 0 0 Out 2 0.69 0.41 0.12 0.12 3.47 2.01 0.24 0.24 Shelter 5 In 2 0.30 0.22 0 0 0.30 0.22 0 0 Out 2 0.21 0.14 0.04 0.04 0.30 0.23 0.04 0.04 WCVI 2011 BedwellS In 5 0.06 0.04 0.12 0.04 0.46 0.15 0 0 Out 5 0.04 0.04 0 0 0.16 0.11 0 0 Folger 7 In 6 0.14 0.04 0.03 0.02 0.38 0.07 0.19 0.05 Out 6 0.26 0.08 0.05 0.02 0.59 0.14 0.23 0.08 Saranac 7 In 4 0.11 0.11 0 0 0.82 0.63 0 0 Out 6 0.25 0.19 0.07 0.03 0.94 0.22 0.07 0.05 Brooks 7 In 6 0.40 0.15 0.25 0.10 1.33 0.45 0.16 0.07 Out 7 0.49 0.12 0.19 0.03 1.08 0.12 0.16 0.03 192 QB YE IR LC Name Age RCA N SE SE SE SE Checleset 6 In 10 0.60 0.21 0.34 0.09 1.77 0.37 0.32 0.10 Out 11 0.58 0.16 0.33 0.09 1.54 0.26 0.25 0.07 Scott 7 In 10 0.25 0.09 0.14 0.06 0.61 0.17 0.12 0.05 Out 12 0.21 0.06 0.22 0.11 0.88 0.23 0.15 0.06 Topknot 7 In 5 0.32 0.11 0.44 0.12 1.09 0.25 0.25 0.08 Out 7 0.60 0.23 0.35 0.06 1.22 0.27 0.56 0.11 Appendix Table 2-1 continued. Mean and Standard Error of Fish Densities (#/100m2) inside and outside of RCAs observed on ROV surveys. GS KG Name RCA N SE SE SG 2009 Brethour In 6 0 0 0.28 0.17 Out 3 0 0 0.53 0.09 Domett-Pam In 9 0.09 0.05 0 0 Out 5 0.07 0.04 0 0 Halibut-McCall In 8 0.40 0.21 0.10 0.09 Out 4 0.37 0.11 0 0 Northumberland In 3 0.21 0.06 0.03 0.03 Out 6 0.74 0.05 0.09 0.02 Saltspring-Trincomali In 14 0.05 0.03 0.13 0.04 Out 3 0.26 0.17 0.20 0.11 Brethour In 6 0 0 0.34 0.12 Out 4 0 0 0.55 0.21 Darcy In 3 0.11 0.11 0.50 0.35 Out 3 0 0 0.10 0.05 Desolation In 13 0.51 0.12 0.04 0.02 Out 10 0.41 0.16 0.01 0.01 Dinner In 4 0.31 0.24 0.19 0.07 Out 3 0.27 0.14 0.14 0.09 Halibut-McCall In 6 0.83 0.20 0.01 0.01 Out 5 0.93 0.22 0 0 Hardy In 2 0.72 0.04 0 0 Out 2 0.73 0.09 0 0 Northumberland In 3 0.21 0.10 0.11 0.06 Out 3 0.14 0.06 0.37 0.11 Prevost In 7 0 0 0.59 0.17 Out 5 0 0 0.23 0.08 Sabine In 3 0.93 0.26 0 0 193 GS KG Name RCA N SE SE Out 3 0.44 0.26 0.22 0.11 Saltspring-Trincomali In 15 0.02 0.01 0.22 0.08 Out 5 0 0 0.11 0.07 Valdes In 3 0.62 0.48 0.29 0.16 Out 3 0.05 0.05 0.09 0.09 SG 2010 Ballenas In 3 0.16 0.10 0 0 Out 2 0 0 0.15 0.15 Northumberland In 3 0.16 0.13 0.15 0.12 Out 3 0.35 0.01 0.26 0.15 Valdes In 2 0.61 0.01 0.22 0.05 Out 2 0 0 0.12 0.12 Northumberland In 7 0.48 0.18 0.14 0.08 Out 5 0.30 0.10 0.27 0.12 SG 2011 Brethour In 2 0 0 0.29 0.29 Out 2 0 0 0.38 0.08 JS 2010 Chancellor In 2 0.15 0.09 0 0 Out 2 0.29 0.10 0 0 Clio-Viscount In 3 0 0 0 0 Out 5 0 0 0 0 Dickson In 2 0.13 0.13 0.09 0.09 Out 2 0 0 0.22 0.15 Octopus-Read In 3 0.31 0.13 0.04 0.02 Out 4 0.64 0.30 0.03 0.03 Thurston In 3 0.32 0.11 0 0 Out 2 0.58 0.47 0.06 0.06 QCST 2011 Bate In 3 0 0 1.31 0.64 Out 3 0 0 1.24 0.59 Bolivar In 6 0 0 0.09 0.04 Out 4 0.01 0.01 0.08 0.05 Browning In 5 0.12 0.06 0.21 0.11 Out 5 0.06 0.04 0.11 0.08 Goletas In 2 0.10 0.10 0.05 0.05 Out 2 0 0 0.38 0.10 Shelter In 2 0 0 0.25 0.01 Out 2 0.21 0.06 0.07 0.07 194 GS KG Name RCA N SE SE WCVI 2011 BedwellS In 5 0 0 0.05 0.04 Out 5 0 0 0.09 0.05 Folger In 6 0.03 0.02 0.18 0.07 Out 6 0.08 0.06 0.16 0.06 Saranac In 4 0 0 0.04 0.04 Out 6 0.07 0.05 0.10 0.04 Brooks In 6 0 0 0.90 0.47 Out 7 0 0 0.31 0.07 Checleset In 10 0 0 0.53 0.11 Out 11 0.01 0.01 0.41 0.10 Scott In 10 0 0 0.32 0.09 Out 12 0.01 0.01 0.29 0.08 Topknot In 5 0.01 0.01 0.21 0.05 Out 7 0.01 0.01 0.14 0.03 195 Appendix Table 2-2. Short and full names of the RCAs sampled and the region they are found in. Names of RCAs that were pooled for analysis are separated by a semi-colon. Region RCA RCA full name SG Ballenas Ballenas Island SG Brethour Brethour,Domville,Forrest,Gooch Islands SG Brethour Brethour,Domville,Forrest,Gooch Islands SG Brethour Brethour,Domville,Forrest,Gooch Islands SG Darcy D'arcy Island to Beaumont Shoal SG Desolation Desolation Sound SG Dinner Dinner Rock SG Domett-Pam Domett Point; Pam Rock SG Halibut-McCall Halibut Bank; McCall Banks SG Halibut-McCall Halibut Bank SG Hardy Hardy Island SG Northumberland Northumberland Channel SG Northumberland Northumberland Channel SG Northumberland Northumberland Channel SG Northumberland Northumberland Channel SG Prevost Prevost Island North SG Sabine Sabine Channel-Jervis-Jedediah Islands SG Saltspring-Trincomali Saltspring Island North; Trincomali Channel SG Saltspring-Trincomali Saltspring Island North; Trincomali Channel SG Valdes Valdes Island East SG Valdes Valdes Island East JS Chancellor Chancellor Inlet West JS Clio-Viscount Lower Clio Channel; Viscount Island JS Dickson Dickson - Polkinghorne Islands JS Octopus-Read Octopus Islands to Hoskyn Channel; Read - Cortes Islands JS Thurston Thurston Bay 196 Region RCA RCA full name QCST Bate Bate - Shadwell Passage QCST Bolivar Bolivar Passage QCST Browning Browning Passage - Hunt Rock QCST Goletas Goletas Channel QCST Shelter Shelter Bay WCVI BedwellS Bedwell Sound WCVI Brooks Brooks Bay WCVI Checleset Checleset Bay WCVI Folger Folger Passage WCVI Saranac Saranac Island WCVI Scott Scott Islands WCVI Topknot Topknot 197 Appendix C Supplementary Information for Chapter 4. Lack of Recreational Fishing Compliance May Compromise Effectiveness of Rockfish Conservation Areas in British Columbia. Appendix Figure 3-1. Recreational Fishing Effort and estimated total Inshore Rockfish catch (number of fish) in the Strait of Georgia 1999-2011. Data shown are from Zetterberg et al. (2012) and Zetterberg et al. (In preparation). The years addressed in this study, 2003, 2007 and 2011 are highlighted. Inshore rockfish recreational catch, which is estimated from the Creel Survey effort data and dockside interviews closely tracks the effort. 198 Appendix Table 3-1. Model Variable data sources and GIS calculations. Variable Data GIS Calculation Transformation City Towns with Population >5000 Distance Nearest to RCA “Near function” None Fishing Lodge Google Earth Fishing lodge locations. Distance Nearest to RCA-Near Function None Rockfish catch (rfcatch) Creel survey estimate of the total number of rockfish caught per creel area, May-September 2011 (data from D. O’Brien, South Coast Management, DFO). Spatial Join with RCAs None Effort Density Adjacent to RCA (2 km Buffer) Normalized Kernel Density analysis (2011) Sum per 2 km buffer strip around each RCA using Zonal Statistics tool, divided by area of buffer. Rank Enforcement (patrol hours) Numbers of hours each Conservation and protection field unit spent on the water in each statistical area in 2011 (data from C. Todd, Conservation and Protection, DFO). Spatial Join with RCA Rank Population within 25 km radius of each RCA 2011 Census data joined to BC place names Spatial Join to calculate the sum of population within 25 km of each RCA Rank RCA Area (size) RCA Calculate geometry (km2) Rank Perimeter to Area Ratio (PA) RCA shapefile I created a 1 cm-buffer around each RCA, erased the RCA from it, to have an RCA perimeter. I then erased the shoreline so that I would only have the perimeter of the RCA that was fronted by water. I divided the open perimeter by the area of the RCA. None Regions 9 regions RCA assigned to the factor region None 199 Appendix Figure 3-2. Changes in monthly recreational effort with time by RCA. Increases and decreases of effort (*) are RCAs with significant differences in effort while the other RCAs showed no significant difference in effort with time. RCAs with <0.5 boats per km2 were characterized with low effort while RCAs with >0.5 boats per km2 were characterized as having effort. 200 Appendix Table 3-2. Mean monthly fishing effort per RCA by Year with ANOVA test statistic (F) and probability using the bonferroni correction (df=2). For years with different effort, Tukey multiple comparison test results are shown. RCA 2003 2007 2011 F p Direction Tukey Multiple Comparisons Ajax 0.0 0.0 0.0 Zero all years Davie 0.0 0.0 0.0 Zero all years Halibut 0.0 0.0 0.0 Zero all years Kanish 0.0 0.0 0.0 Zero all years Mariners 0.0 0.0 0.0 Zero all years McCall 0.0 0.0 0.0 Zero all years McNaughton 0.0 0.0 0.0 Zero all years Navy Channel 0.0 0.0 0.0 Zero all years Nelson 0.0 0.0 0.0 Zero all years Read 0.0 0.0 0.0 Zero all years Sinclair 0.0 0.0 0.0 Zero all years Trial 0.0 0.0 0.0 Zero all years Upper Centre 0.0 0.0 0.0 Zero all years West Bay 0.0 0.0 0.0 Zero all years Lasqueti 0.2 0.2 0.0 3.7 0.06 low effort all years Malaspina 0.0 0.1 0.0 3.3 0.07 low effort all years Gabriola 0.3 0.0 0.0 2.7 0.11 low effort all years Pasley 0.3 0.0 0.1 2.2 0.16 low effort all years Saltspring 0.0 0.1 0.4 2.2 0.16 low effort all years Sabine 0.2 0.3 0.0 1.8 0.20 low effort all years Saturna 0.2 0.4 0.1 1.5 0.26 low effort all years Menzies 0.5 0.0 0.3 1.3 0.30 low effort all years Portland 0.0 0.2 0.0 1.0 0.39 low effort all years Chrome 0.1 0.0 0.0 1.0 0.40 low effort all years 201 RCA 2003 2007 2011 F p Direction Tukey Multiple Comparisons Danger Reefs 0.3 0.0 0.0 1.0 0.40 low effort all years Heriot 0.4 0.0 0.0 1.0 0.40 low effort all years Patey 0.5 0.0 0.0 1.0 0.40 low effort all years Savoie 0.3 0.0 0.0 1.0 0.40 low effort all years Sisters 0.0 0.0 0.0 1.0 0.40 low effort all years Woolridge 0.0 0.3 0.0 1.0 0.40 low effort all years Young 0.0 0.0 0.1 1.0 0.40 low effort all years Trincomali 0.1 0.3 0.4 0.8 0.48 low effort all years Coal 0.1 0.3 0.0 0.6 0.58 low effort all years Hardy 0.2 0.2 0.1 0.3 0.72 low effort all years Darcy 0.0 0.0 0.2 19.0 0.00 increase in 2011 2011>2003, 2007 (0.007,0.003) Walken 0.1 0.0 1.9 5.6 0.02 increase in 2011 2011>2003,2007 (0.04,0.03) Pam Rock 0.0 3.1 0.4 25.7 0.00 increase in 2007 2007>2003 (0.00005), 2011<2007 (0.004), 2011=2003 Bell Chain 0.2 1.3 0.6 6.3 0.01 increase in 2007 2007>2003 (0.01), 2011=2003,2007 De Courcy 0.0 0.3 0.0 5.8 0.02 increase in 2007 2007>2003,2011 (0.03,0.03) Coffin 2.2 1.4 0.2 3.6 0.06 effort Brethour 0.1 1.0 0.4 3.6 0.06 effort Ruxton 0.4 0.7 0.1 3.0 0.09 effort Bowyer 4.3 1.1 1.1 2.9 0.10 effort Reynolds 0.0 0.8 0.0 2.8 0.10 effort Russell 0.0 0.0 1.3 2.6 0.12 effort 202 RCA 2003 2007 2011 F p Direction Tukey Multiple Comparisons Brentwood 3.8 2.0 1.3 2.6 0.12 effort West Van 2.8 1.5 0.4 2.5 0.12 effort Mid Finlayson 0.2 1.4 0.2 2.3 0.14 effort Nanoose 1.1 0.5 0.9 2.0 0.18 effort Thetis 1.0 1.7 0.8 1.9 0.19 effort Thormanby 0.9 0.0 0.0 1.9 0.19 effort Domett 0.0 0.6 0.2 1.5 0.27 effort Prevost 0.6 0.1 0.2 1.0 0.39 effort Departure 0.9 0.0 0.6 1.0 0.39 effort Duntze Head 1.8 0.0 0.0 1.0 0.40 effort Galiano 1.0 0.4 1.0 1.0 0.40 effort Passage 0.0 0.0 1.2 1.0 0.40 effort Maple 1.4 0.1 0.4 1.0 0.40 effort Deepwater 11.8 55.5 47.8 0.9 0.42 effort Maud 0.7 0.8 3.7 0.8 0.46 effort Northumber-land 0.7 0.4 0.3 0.7 0.50 effort Copeland 0.6 0.2 0.4 0.7 0.51 effort Teakerne 0.2 0.7 0.2 0.7 0.52 effort Ballenas 1.0 1.5 0.8 0.6 0.58 effort Oyster 0.2 0.8 0.3 0.6 0.58 effort Bentinck 0.8 0.0 2.2 0.5 0.60 effort Valdes 1.4 1.4 1.4 0.3 0.78 effort Bedwell 0.8 0.5 0.5 0.1 0.94 effort Dinner Rock 0.6 0.4 0.8 0.0 0.99 effort Mayne 4.0 1.4 0.3 4.9 0.03 decrease in 2011 2011<2003 (0.02) 203 RCA 2003 2007 2011 F p Direction Tukey Multiple Comparisons Race Rocks 5.6 0.7 2.1 5.8 0.02 decrease in 2007 2007<2003 (0.01), 2011=2003,2007 Becher 33.7 1.0 4.6 4.6 0.03 decrease in 2007 2007<2003 (0.03), 2011=2003,2007 Sooke 13.1 1.1 0.4 21.7 0.00 decrease 2011(0.0002), 2007(0.0004)<2003 Lions Bay 10.4 1.2 0.6 18.9 0.00 decrease 2007,2011<2003 (0.001,0.0003) Mitlenatch 1.0 0.0 0.0 14.1 0.00 decrease 2007,2011<2003 (0.002, 0.001) Burgoyne 4.4 1.0 0.8 5.5 0.02 decrease 2011, 2007(0.04,0.03)<2003 Discovery 1.0 0.0 0.0 5.3 0.02 decrease 2007,2011<2003 (0.04,0.04) 204 Appendix Figure 3-3. Relative fishing effort (%) in RCAs by Pacific Fishery Management Areas (PFMA). This proportion was applied to the estimated rockfish catch by PFMA in 2011 to estimate the number of rockfish that could have been taken in the recreational fishery in RCAs in each PFMA. Areas 17 and 18 (near Nanaimo and the Gulf Islands), has the highest estimated rockfish catch, followed by area 27, NW Vancouver Island. 205 Appendix D Supplementary Information from Chapter 5. How Do They Score? An Evaluation of Rockfish Conservation Areas Using a Conservation Score that Combines Rockfish Habitat and Key Reserve Features. Appendix Figure 4-1. Examples of classified habitat 20m model compared with the 5m models with and without backscatter (BS) in the five test areas. The black outline indicates the limit of the MBES extent. 206 Appendix Figure 4-2. Examples of the 20 m model compared to the 20+5 m model used in the analysis overlain on the 100m model used to designate the RCAs (Old Model) showing the effects the spatial resolution of the bathymetry data. 207 Appendix Table 4-1. Values of RCA Features used to determine the Conservation Score and Number of Key Features. Habitat Areas shown in Bold print are in inlets without 5m resolution data. RCAs marked with an * were sampled using an ROV and with ** by SCUBA diving. RCA Age RCA Area km2 Habitat Area km2 % Habitat Isolation Recreational Compliance Boats/yr Bycatch Estimate Connectivity Km Ajax / Achilles Bank 11 73.9 2.5 3.4 H 0.0 181.7 4.3 Ballenas Island* 11 5.8 1.7 29.4 L 26.8 41.8 2.4 Bate-Shadwell* 9 17.8 2.2 12.4 L 5.0 0.0 3.2 Baynes Sound 9 2.5 0.03 1.0 H 0* 0.3 4.5 Becher Bay East 9 1.0 0.6 55.9 L 28.0 0.0 4.7 Bedwell Harbour 11 2.5 0.8 30.4 L 8.4 42.4 2.2 Bedwell Sound* 11 15.3 1.1 7.1 L 0* 0.0 1.2 Bell Chain Islets 9 13.0 5.2 39.8 L 49.4 20.1 2.9 Belleisle Sound 9 5.1 0.1 1.4 H 0* 0.3 13.0 Bentinck Island 11 0.5 0.5 89.1 L 6.0 0.0 0.9 Bolivar Passage* 9 16.7 11.3 67.2 L 24.8 0.0 4.2 Bond Sound 9 3.8 0.03 0.7 M 0* 10.3 4.3 Bowyer Island 11 3.1 0.6 18.4 M 18.6 7.9 1.1 Brentwood Bay 9 3.4 0.6 19.0 L 31.0 4.2 5.5 Brethour* 9 18.7 6.7 35.4 L 48.4 47.5 2.7 Broken I. Group** 11 39.7 23.9 51.7 L 0.0 0.0 1.8 Brooks Bay* 11 72.3 9.6 13.0 L 0.0 0.0 9.9 Browning to Raynor 9 16.6 4.1 23.3 L 16.0 0.2 3.6 Browning-Hunt Rock* 9 10.0 3.9 39.0 L 15.0 0.0 3.5 Burgoyne Bay 11 2.6 0.3 11.0 L 11.5 7.2 3.1 Burley Bay 9 10.7 0.2 1.8 M 0* 5.4 2.9 Bute Inlet North 9 46.2 4.8 10.3 M 0* 2.7 77.4 Carmanah 11 8.2 0.3 3.7 M 0.0 0.0 3.0 Chancellor Inlet E 9 3.5 0.02 0.6 M 0* 0.7 2.8 Chancellor Inlet W* 9 13.9 2.4 17.5 M 0* 0.6 2.8 Checleset Bay* 10 147.4 21.0 14.1 L 5.0 0.0 9.9 Chrome Island 9 3.9 0.1 2.9 M 0.0 6.3 3.0 Coal Island 9 3.1 0.8 25.9 L 0.0 1.5 2.2 Coffin Point 9 4.3 0.2 5.2 M 4.0 34.7 0.8 Copeland 11 15.3 4.4 25.8 L 51.5 130.5 3.8 Cracroft Point South 9 2.7 0.8 30.1 L 9.0 0.0 4.9 Danger Reefs 11 1.5 1.0 66.9 L 0.0 47.0 0.7 D'Arcy to Beaumont* 9 53.8 11.2 20.8 M 62.2 50.3 3.3 Dare Point 11 3.5 0.1 2.3 M 0.0 0.0 3.0 Davie Bay 9 10.2 1.6 15.6 M 0.0 71.4 5.7 De Courcy N 9 4.0 0.6 15.8 M 0.0 8.4 1.2 Deepwater Bay 11 1.8 0.02 1.1 M 480.8 0.0 6.8 208 RCA Age RCA Area km2 Habitat Area km2 % Habitat Isolation Recreational Compliance Boats/yr Bycatch Estimate Connectivity Km Departure Bay 9 2.7 0.3 11.7 M 8.0 1.0 3.1 Desolation Sound* 9 60.0 11.3 18.8 L 5.0 111.6 3.8 Dickson - Polkinghorne* 9 15.9 8.4 53.0 L 0* 8.8 3.8 Dinner Rock* 9 6.6 0.2 3.4 M 37.5 110.9 7.5 Discovery-Chatham 11 3.2 2.2 59.4 L 0.0 0.0 3.3 Domett Point* 9 2.1 0.2 8.3 M 2.0 3.0 3.7 Drury Inlet 9 11.6 2.2 19.0 M 0* 3.0 20.9 Duntze Head 11 0.9 0.3 30.6 L 0.0 0.0 8.8 Eastern Burrard Inlet 9 2.7 0.1 3.5 M 0* 0.0 5.9 Eden-Bonwick 9 68.7 30.2 44.0 L 54.5 6.2 1.6 Estevan Point 10 186.0 11.3 6.1 M 0.0 0.0 23.2 Folger Passage* 11 17.0 4.5 26.6 M 2.4 0.0 1.8 Forward Harbour 9 3.3 0.002 0.05 H 0* 0.0 1.6 Gabriola Passage 11 2.6 0.5 20.1 L 0.0 5.7 1.2 Galiano Island N 11 9.7 0.4 4.6 M 63.1 21.1 9.4 Goletas Channel* 11 36.7 7.2 19.7 L 0.0 0.0 1.4 Greenway Sound 9 15.1 5.7 31.8 M 0* 28.6 13.7 Haddington 9 2.5 0.2 6.9 M 0.0 0.0 10.3 Halibut Bank* 9 33.0 1.6 4.9 M 0.0 28.6 4.6 Hardy Island* 9 16.0 3.5 22.0 M 5.0 0.0 4.0 Hardy-Five Fathom 9 0.1 0.02 13.0 M 15.3 36.7 15.5 Havannah Channel 9 32.0 5.9 18.5 L 0* 34.4 3.8 Heriot Bay 11 5.1 1.1 20.9 M 0.0 21.6 4.4 Holberg Inlet 11 22.5 11.9 53.1 L 0.0 17.5 52.8 Hotham Sound 9 22.4 5.9 26.6 L 0* 16.3 9.0 Crocker Island 9 8.9 1.2 13.9 M 0* 0.6 3.7 Twin Islands 9 2.9 0.5 16.8 L 0* 0.0 3.7 Kanish Bay 11 8.0 0.4 5.4 L 0.0 0.0 7.4 Kwatsi Bay 9 3.4 0.1 1.6 M 0* 4.3 4.3 Lasqueti Island South 9 18.5 4.8 25.7 L 0.0 73.0 6.2 Lasqueti-Young Pt 9 9.3 0.4 4.3 H 10.0 36.6 2.4 Lions Bay 9 4.8 0.6 12.3 M 15.0 10.6 1.1 Loughborough Inlet 9 37.1 0.5 1.3 L 0* 34.4 12.8 Lower Clio Channel* 9 13.9 2.4 17.0 L 0* 6.7 4.7 Mackenzie - Nimmo 9 4.0 0.01 0.4 H 0* 0.0 2.9 Malaspina Strait 9 28.3 3.3 11.8 M 1.8 50.9 3.6 Maple Bay 11 3.2 0.4 12.5 L 6.0 8.1 2.5 Mariners Rest 9 1.9 0.2 9.0 M 0.0 9.4 3.6 Maud Island 9 3.1 0.3 9.5 L 68.1 0.0 1.9 Mayne Island North 11 7.0 4.0 57.3 L 12.2 3.9 0.0 209 RCA Age RCA Area km2 Habitat Area km2 % Habitat Isolation Recreational Compliance Boats/yr Bycatch Estimate Connectivity Km McCall Bank* 9 13.4 0.8 6.2 M 0.0 19.3 4.6 McNaughton Point 11 2.2 0.9 42.7 L 0.0 50.2 3.0 Menzies Bay 11 3.9 0.1 3.6 M 9.0 0.0 1.9 Mid Finlayson 9 1.9 0.3 17.9 M 2.0 0.3 5.5 Mitlenatch Island 9 24.9 2.3 9.4 M 0.0 71.5 5.0 Nanoose-Schooner 9 12.0 2.1 17.9 L 63.4 17.4 2.4 Navy Channel 9 8.3 0.8 9.6 L 0.0 7.9 2.9 Nelson Island 11 8.7 1.5 16.4 L 0.0 39.4 3.6 Northumberland* 9 14.8 1.4 9.7 M 29.2 72.7 3.1 Nowell Channel 9 12.4 4.9 39.0 L 0* 5.5 1.6 Numas Islands 11 28.9 5.7 19.7 M 0.0 0.3 7.7 Octopus to Hoskyn* 9 35.8 4.5 12.5 L 21.7 144.7 4.5 Oyster Bay 9 9.1 0.2 2.5 M 0* 0.0 5.0 Pachena Point 11 19.1 0.5 2.7 M 0.0 0.0 12.4 Pam Rock* 11 5.7 1.2 21.0 M 13.4 20.8 1.2 Pasley Island 9 12.0 3.6 30.2 L 6.0 19.3 5.3 Passage Island 9 0.8 0.2 23.6 L 4.4 12.2 0.4 Patey Rock 11 0.9 0.4 41.8 L 0.0 14.3 9.3 Pendrell Sound 9 15.2 2.9 19.2 L 0* 19.3 6.7 Port Elizabeth 9 6.0 0.3 4.7 L 0* 0.7 12.4 Portland Island 11 3.0 1.5 49.7 L 0.0 3.5 2.2 Prevost* 9 9.0 1.9 21.0 L 10.7 4.6 2.5 Princess Louisa Inlet 9 6.2 0.5 8.4 0* 0.0 1.4 Queen's Reach E 9 4.5 0.8 18.3 M 0* 0.0 1.4 Queen's Reach W 9 3.5 0.6 17.1 M 0* 0.5 3.5 Race Rocks 11 2.7 2.7 97.9 L 36.8 0.0 0.9 Read-Cortes* 9 30.3 5.8 19.0 L 0.0 104.3 4.4 Reynolds Point 9 4.3 0.3 7.0 L 0.0 0.0 5.7 Russell Island 9 2.4 0.2 8.4 M 16.0 9.8 2.5 Ruxton - Pylades 9 6.8 1.7 25.1 M 3.3 21.1 1.6 Sabine Channel* 9 22.3 6.1 27.4 L 0.0 133.2 2.4 Salmon Channel 11 14.1 4.6 32.0 L 0.0 0.3 2.1 Salmon Inlet 9 17.5 2.7 15.2 M 0* 0.0 8.7 Saltspring Island North* 11 8.4 1.4 16.6 L 21.4 42.8 0.0 Saranac Island* 11 10.9 0.9 8.0 M 0* 0.0 1.2 Savoie Rocks 9 1.7 0.1 3.6 M 0.0 3.4 3.0 Scott Islands* 11 338.4 33.3 9.8 M 0* 0.0 24.0 Shelter Bay* 9 15.5 4.8 30.7 L 0.0 0.0 1.4 Sinclair Bank 9 19.2 2.3 11.9 M 0.0 60.1 3.8 Sisters Islets 9 10.7 2.1 19.5 H 0.0 33.7 4.3 210 RCA Age RCA Area km2 Habitat Area km2 % Habitat Isolation Recreational Compliance Boats/yr Bycatch Estimate Connectivity Km Skookumchuck 9 13.1 6.0 45.3 M 0* 14.3 8.7 Sooke Bay 11 3.4 0.2 6.7 M 8.0 0.0 10.8 South Saturna 9 30.8 4.4 14.2 M 13.9 9.3 2.2 Storm Islands 11 37.2 18.3 49.0 L 54.5 1.9 5.5 Susquash 11 8.1 1.1 13.2 M 34.3 0.0 11.7 Teakerne Arm 11 8.4 1.5 17.8 M 11.8 37.7 8.6 Thetis-Kuper 11 25.5 5.7 22.3 L 119.7 251.0 0.7 Thompson Sound 9 13.9 0.4 2.6 M 0* 10.9 6.1 Thormanby Island 11 3.2 0.9 29.2 L 0.0 36.9 3.0 Thurston Bay* 11 6.6 1.0 15.8 L 0.0 0.0 2.2 Topknot* 11 96.1 6.6 6.9 M 398.2 0.0 21.2 Trial Island 11 0.8 0.4 46.0 L 0.0 0.0 4.2 Trincomali* 9 21.6 1.8 8.3 L 55.5 22.7 0.9 Upper Call Inlet 9 21.0 0.2 1.1 H 0* 46.1 3.8 Upper Centre Bay 9 1.1 0.2 21.2 L 0.0 9.9 3.0 Valdes Island East* 9 10.0 1.7 16.5 M 82.8 23.2 7.2 Vargus to Dunlap 11 2.8 0.5 18.9 L 0.0 0.0 2.8 Viscount Island 9 21.9 2.2 10.1 L 0* 65.2 1.7 Wakeman Sound 9 12.3 0.2 1.9 M 0* 0.8 13.0 Walken to Hemming 9 13.6 3.8 27.7 L 151.6 0.0 2.2 Wellborne 9 22.9 3.9 17.0 M 0* 8.4 1.6 West Bay 9 1.1 0.3 23.9 L 0.0 16.9 3.0 West Cracroft 9 3.6 2.0 54.5 L 0* 0.0 4.9 West of Bajo Reef 10 41.8 0.5 1.2 M 5.3 0.0 23.2 West Vancouver 9 2.8 0.5 18.1 L 7.2 27.6 0.4 Weynton Passage 11 17.6 8.4 47.8 L 75.3 0.0 7.6 Woolridge 9 3.8 0.7 19.2 M 0.0 10.0 3.6 211 Appendix Table 4-2. RCA Feature Scores, Conservation Score, number of Key Features of each RCA ranked by Conservation Score (best to worst). RCAs marked with an * were sampled using an ROV and with ** by SCUBA diving. Rank RCA Size Habitat Area % Habitat Isolation Compliance Commercial Compliance Recreational Bycatch Connectivity Conservation Score Key Feature s 1 Numas Islands 2 3 2 3 3 3 3 3 22 4 1 Goletas Channel* 2 3 2 3 3 3 3 3 22 4 1 Dickson - Polkinghorne* 2 3 3 2 3 3 3 3 22 3 2 Queen's Reach W 2 2 2 3 3 3 3 3 21 4 2 Queen's Reach E 2 2 2 3 3 3 3 3 21 4 2 Lower Clio Channel* 2 2 2 3 3 3 3 3 21 4 2 Chancellor Inlet W* 2 2 2 3 3 3 3 3 21 4 2 Salmon Inlet 2 2 2 3 3 3 3 3 21 4 2 Wellborne 2 2 2 3 3 3 3 3 21 4 2 Hotham Sound 2 3 2 3 3 3 2 3 21 3 2 Broken I.Group** 2 3 3 1 3 3 3 3 21 3 2 Skookumchuck 2 3 2 3 3 3 2 3 21 3 3 Gabriola Passage 2 1 2 3 3 3 3 3 20 4 3 Scott Islands* 3 3 1 2 3 3 3 2 20 4 3 Bedwell Sound* 2 2 1 3 3 3 3 3 20 4 3 Drury Inlet 2 2 2 3 3 3 3 2 20 4 3 De Courcy N 2 2 2 2 3 3 3 3 20 3 3 Ruxton - Pylades 2 2 2 3 3 3 2 3 20 3 3 West Cracroft 2 2 3 1 3 3 3 3 20 3 3 Crocker Island 2 2 2 2 3 3 3 3 20 3 3 Woolridge 2 2 2 2 3 3 3 3 20 3 3 Coal Island 2 2 2 2 3 3 3 3 20 3 3 Thurston Bay* 2 2 2 2 3 3 3 3 20 3 3 Sisters Islets 2 2 2 3 3 3 2 3 20 3 3 Discovery-Chatham 2 2 3 1 3 3 3 3 20 3 3 Pendrell Sound 2 2 2 3 3 3 2 3 20 3 3 Folger Passage* 2 2 2 2 3 3 3 3 20 3 3 Shelter Bay* 2 2 2 2 3 3 3 3 20 3 3 Brooks Bay* 2 3 2 1 3 3 3 3 20 3 3 Checleset Bay* 3 3 2 1 3 2 3 3 20 3 3 Greenway Sound 2 3 2 3 3 3 2 2 20 3 3 Holberg Inlet 2 3 3 3 3 3 2 1 20 3 3 Havannah Channel 2 3 2 2 3 3 2 3 20 2 4 Savoie Rocks 2 1 1 3 3 3 3 3 19 4 4 Dare Point 2 1 1 3 3 3 3 3 19 4 212 Rank RCA Size Habitat Area % Habitat Isolation Compliance Commercial Compliance Recreational Bycatch Connectivity Conservation Score Key Feature s 4 Carmanah 2 1 1 3 3 3 3 3 19 4 4 Estevan Point 3 3 1 1 3 3 3 2 19 4 4 Bute Inlet North 2 2 2 3 3 3 3 1 19 4 4 Baynes Sound 2 1 1 3 3 3 3 3 19 4 4 Eastern Burrard Inlet 2 1 1 3 3 3 3 3 19 4 4 Domett Point* 2 1 1 3 3 3 3 3 19 4 4 Oyster Bay 2 1 1 3 3 3 3 3 19 4 4 Duntze Head 1 1 2 3 3 3 3 3 19 4 4 Forward Harbour 2 1 1 3 3 3 3 3 19 4 4 Mackenzie - Nimmo 2 1 1 3 3 3 3 3 19 4 4 Chancellor Inlet E 2 1 1 3 3 3 3 3 19 4 4 Bond Sound 2 1 1 3 3 3 3 3 19 4 4 Kwatsi Bay 2 1 1 3 3 3 3 3 19 4 4 Burley Bay 2 1 1 3 3 3 3 3 19 4 4 Princess Louisa Inlet 2 1 1 3 3 3 3 3 19 4 4 Vargus to Dunlap 2 1 2 2 3 3 3 3 19 3 4 Salmon Channel 2 2 2 1 3 3 3 3 19 3 4 Nowell Channel 2 2 2 1 3 3 3 3 19 3 4 Saranac Island* 2 2 1 2 3 3 3 3 19 3 4 Upper Centre Bay 2 1 2 2 3 3 3 3 19 3 4 Mid Finlayson 2 1 2 2 3 3 3 3 19 3 4 Twin Islands 2 1 2 2 3 3 3 3 19 3 4 Navy Channel 2 2 1 2 3 3 3 3 19 3 4 Departure Bay 2 1 2 3 3 2 3 3 19 3 4 Bowyer Island 2 2 2 3 3 1 3 3 19 3 4 McCall Bank* 2 2 1 3 3 3 2 3 19 3 4 Portland Island 2 2 2 1 3 3 3 3 19 3 4 Halibut Bank* 2 2 1 3 3 3 2 3 19 3 4 Mitlenatch Island 2 2 1 3 3 3 2 3 19 3 4 Danger Reefs 2 2 3 1 3 3 2 3 19 2 4 Nelson Island 2 2 2 2 3 3 2 3 19 2 4 Storm Islands 2 3 2 2 3 1 3 3 19 2 4 Teakerne Arm 2 2 2 3 3 2 2 3 19 2 4 Viscount Island 2 2 2 2 3 3 2 3 19 2 4 Lions Bay 2 2 2 3 3 2 2 3 19 2 4 Heriot Bay 2 2 2 2 3 3 2 3 19 2 4 Davie Bay 2 2 2 2 3 3 2 3 19 2 4 Prevost* 2 2 2 2 3 2 3 3 19 2 4 Bate-Shadwell* 2 2 2 2 3 2 3 3 19 2 4 Sinclair Bank 2 2 2 2 3 3 2 3 19 2 213 Rank RCA Size Habitat Area % Habitat Isolation Compliance Commercial Compliance Recreational Bycatch Connectivity Conservation Score Key Feature s 4 Malaspina Strait 2 2 2 2 3 3 2 3 19 2 4 Browning-Hunt Rock* 2 2 2 2 3 2 3 3 19 2 4 Mayne Island North 2 2 3 1 3 2 3 3 19 2 4 South Saturna 2 2 2 2 3 2 3 3 19 2 4 Read-Cortes* 2 3 2 2 3 3 1 3 19 2 4 Sabine Channel* 2 3 2 2 3 3 1 3 19 2 4 Bolivar Passage* 2 3 3 1 3 1 3 3 19 2 4 Desolation Sound* 2 3 2 3 3 2 1 3 19 2 4 Eden-Bonwick 2 3 2 2 3 1 3 3 19 2 5 Pachena Point 2 1 1 3 3 3 3 2 18 4 5 Port Elizabeth 2 1 1 3 3 3 3 2 18 4 5 Belleisle Sound 2 1 1 3 3 3 3 2 18 4 5 Wakeman Sound 2 1 1 3 3 3 3 2 18 4 5 Chrome Island 2 1 1 2 3 3 3 3 18 3 5 Menzies Bay 2 1 1 3 3 2 3 3 18 3 5 Reynolds Point 2 1 1 2 3 3 3 3 18 3 5 Kanish Bay 2 1 1 2 3 3 3 3 18 3 5 Mariners Rest 2 1 1 2 3 3 3 3 18 3 5 Trial Island 1 1 2 2 3 3 3 3 18 3 5 Ajax / Achilles Bank 2 2 1 3 3 3 1 3 18 3 5 Upper Call Inlet 2 1 1 3 3 3 2 3 18 3 5 Thompson Sound 2 1 1 3 3 3 2 3 18 3 5 Susquash 2 2 2 3 3 1 3 2 18 3 5 Becher Bay East 2 2 3 1 3 1 3 3 18 2 5 Browning to Raynor 2 2 2 1 3 2 3 3 18 2 5 Cracroft Point South 2 2 2 1 3 2 3 3 18 2 5 Walken to Hemming 2 2 2 2 3 1 3 3 18 2 5 West Bay 2 1 2 2 3 3 2 3 18 2 5 Burgoyne Bay 2 1 2 2 3 2 3 3 18 2 5 Maple Bay 2 1 2 2 3 2 3 3 18 2 5 West Vancouver 2 1 2 2 3 2 3 3 18 2 5 Brentwood Bay 2 2 2 2 3 1 3 3 18 2 5 McNaughton Point 2 2 2 1 3 3 2 3 18 2 5 Thormanby Island 2 2 2 1 3 3 2 3 18 2 5 Race Rocks 2 2 3 1 3 1 3 3 18 2 5 Lasqueti Island South 2 2 2 1 3 3 2 3 18 2 5 Weynton Passage 2 3 2 1 3 1 3 3 18 2 5 D'Arcy to Beaumont* 2 3 2 2 3 1 2 3 18 1 5 Bedwell Harbour 2 2 2 2 3 2 2 3 18 1 5 Pam Rock* 2 2 2 2 3 2 2 3 18 1 214 Rank RCA Size Habitat Area % Habitat Isolation Compliance Commercial Compliance Recreational Bycatch Connectivity Conservation Score Key Feature s 5 Pasley Island 2 2 2 2 3 2 2 3 18 1 5 Bell Chain Islets 2 3 2 2 3 1 2 3 18 1 5 Brethour* 2 3 2 2 3 1 2 3 18 1 6 Haddington 2 1 1 2 3 3 3 2 17 3 6 Sooke Bay 2 1 1 3 3 2 3 2 17 3 6 Deepwater Bay 2 1 1 3 3 1 3 3 17 3 6 Hardy-Five Fathom 1 1 2 3 3 2 3 2 17 3 6 Loughborough Inlet 2 1 1 3 3 3 2 2 17 3 6 Coffin Point 2 1 1 2 3 3 2 3 17 2 6 Bentinck Island 1 1 3 1 3 2 3 3 17 2 6 Passage Island 1 1 2 2 3 3 2 3 17 2 6 Russell Island 2 1 1 2 3 2 3 3 17 2 6 Patey Rock 1 1 2 2 3 3 2 3 17 2 6 Octopus to Hoskyn* 2 2 2 3 3 1 1 3 17 2 6 Saltspring N* 2 2 2 2 3 1 2 3 17 1 6 Valdes Island East* 2 2 2 2 3 1 2 3 17 1 6 Ballenas Island* 2 2 2 2 3 1 2 3 17 1 6 Nanoose-Schooner 2 2 2 2 3 1 2 3 17 1 6 Hardy Island* 2 2 2 1 3 2 2 3 17 1 6 Thetis-Kuper 2 3 2 2 3 1 1 3 17 1 7 West of Bajo Reef 2 1 1 2 3 2 3 2 16 2 7 Maud Island 2 1 1 2 3 1 3 3 16 2 7 Galiano Island N 2 1 1 3 3 1 2 3 16 2 7 Topknot* 2 3 1 1 3 1 3 2 16 2 7 Lasqueti-Young Pt 2 1 1 2 3 2 2 3 16 1 8 Dinner Rock* 2 1 1 3 3 1 1 3 15 2 8 Northumberland* 2 2 1 1 3 1 2 3 15 1 8 Trincomali* 2 2 1 1 3 1 2 3 15 1 8 Copeland 2 2 2 1 3 1 1 3 15 1 215 Appendix E Species Quantified on ROV Surveys in RCAs. Scientific Name Common Name Number Fishes Sebastinae Rockfishes 7334 Hydrolagus colliei Spotted Ratfish 1563 Cymatogaster aggregata Shiner Perch 1366 Sebastes maliger Quillback Rockfish 1344 Sebastes emphaeus Puget Sound Rockfish 1250 Gadidae Codfishes 1227 Theragra chalcogramma Walleye Pollock 1143 Pleuronectiformes Flatfishes 973 Zoarcidae Eelpouts 852 Unknown fish Unknown Fish 747 Hexagrammos decagrammus Kelp Greenling 562 Sebastes pinniger Canary Rockfish 558 Sebastes flavidus Yellowtail Rockfish 527 Sebastes ruberrimus Yelloweye Rockfish 378 Sebastes elongatus Greenstriped Rockfish 367 Ophiodon elongatus Lingcod 346 Sebastes nebulosus China Rockfish 262 Agonidae Poachers 245 Sebastes entomelas Widow Rockfish 201 Sebastes caurinus Copper Rockfish 193 Cottidae Sculpins 155 Sebastes helvomaculatus Rosethorn Rockfish 143 Lipariscus nanus Pygmy Snailfish 136 Sebastes zacentrus Sharpchin Rockfish 108 Lepidopsetta bilineata Southern Rock Sole 99 Sebastes nigrocinctus Tiger Rockfish 80 Microstomus pacificus Dover Sole 78 Merluccius productus Pacific Hake 73 Sebastes miniatus Vermilion Rockfish 71 Lumpenus sagitta Snake Prickleback 70 Sebastes mystinus Blue Rockfish 59 Stichaeidae Pricklebacks 59 Rhacochilus vacca Pile Perch 49 Gadus macrocephalus Pacific Cod 44 Myoxocephalus polyacanthocephalus Great Sculpin 43 Bathymasteridae Ronquils 30 Parophrys vetulus English Sole 30 Clupea pallasii Pacific Herring 27 Embiotocidae Surfperches 24 Sebastes brevispinis Silvergray Rockfish 18 Glyptocephalus zachirus Rex Sole 17 Chirolophis decoratus Decorated Warbonnet 15 216 Scientific Name Common Name Number Ronquilus jordani Northern Ronquil 14 Hippoglossus stenolepis Pacific Halibut 13 Sebastes melanops Black Rockfish 10 Hemilepidotus hemilepidotus Red Irish Lord 9 Sebastes wilsoni Pygmy Rockfish 9 Podothecus accipenserinus Sturgeon Poacher 7 Raja rhina Longnose Skate 7 Percidae Perches 6 Scorpaenichthys marmoratus Cabezon 6 Lycodes pacificus Blackbelly Eelpout 5 Microgadus proximus Pacific Tomcod 5 Osmeridae Smelts 5 Sebastes diploproa Splitnose Rockfish 5 Squalus suckleyi North Pacific Spiny Dogfish 5 Anarrhichthys ocellatus Wolf Eel 4 Enophrys bison Buffalo Sculpin 4 Liparidae Snailfishes 4 Lyopsetta exilis Slender Sole 4 Rajidae Skates 4 Salmonidae Salmonids 4 Clupeidae Herrings 3 Porichthys notatus Plainfin Midshipman 3 Raja binoculata Big Skate 3 Ruscarius meanyi Puget Sound Sculpin 3 Ammodytidae Sand Lances 2 Gobiidae Gobies 2 Hexagrammos stelleri Whitespotted Greenling 2 Hyperprosopon argenteum Walleye Surfperch 2 Malacocottus kincaidi Blackfin Sculpin 2 Nautichthys oculofasciatus Sailfin Sculpin 2 Platichthys stellatus Starry Flounder 2 Sebastes crameri Darkblotched Rockfish 2 Sebastes paucispinis Bocaccio 2 Sebastes proriger Redstripe Rockfish 2 Sebastes variegatus Harlequin Rockfish 2 Brosmophycis marginata Red Brotula 1 Citharichthys stigmaeus Speckled Sanddab 1 Dasycottus setiger Spinyhead Sculpin 1 Hemilepidotus spinosus Brown Irish Lord 1 Hexagrammidae Greenlings 1 Icelinus borealis Northern Sculpin 1 Icelinus filamentosus Threadfin Sculpin 1 Isopsetta isolepis Butter Sole 1 Jordania zonope Longfin Sculpin 1 Leptocottus armatus Pacific Staghorn Sculpin 1 217 Scientific Name Common Name Number Limanda aspera Yellowfin Sole 1 Odontopyxis trispinosa Pygmy Poacher 1 Reinhardtius stomias Arrowtooth Flounder 1 Invertebrates Asteroidea Starfish 3165 Parastichopus californicus Giant Red Sea Cucumber 2864 Metridium Metridium 1262 Crinoidea Sea Lilies And Feather Stars 809 Hexactinellida Glass Sponges 652 Munida quadrispina Squat Lobster 606 Pennatulacea Sea Pens 335 Pycnopodia helianthoides Sunflower Starfish 297 Pandalus platyceros Prawn 255 Pachycerianthus fimbriatus Pachycerianthus Fimbriatus 249 Strongylocentrotus purpuratus Purple Sea Urchins 193 Anthopleura Anthopleura 188 Chlamys hastata Spiny Scallop 156 Balticina septentrionalis Sea Whip 150 Pisaster brevispinus Pink Short-Spined Star 65 Nudibranchia Seaslugs 55 Dendrobranchiata Shrimp 52 Euryalina Basket Stars 44 Actiniaria Anemone 42 Ptilosarcus gurneyi Sea Pen 40 Hydrozoa Hydroid 29 Demospongiae Bath Sponges 27 Chionoecetes Tanner Crabs 26 Mediaster aequalis Vermillion Starfish 23 Strongylocentrotus franciscanus Red Urchin 22 Virgularia Sea Whip 22 Ceramaster patagonicus Cookie Star 19 Strongylocentrotus droebachiensis Green Urchin 19 Sedentaria Tube Worms 18 Echinacea Sea Urchins 14 Enteroctopus dofleini Giant Pacific Octopus 14 Gorgonacea Gorgonian Corals 14 Pectinidae Scallop 14 Bivalvia Bivalve Molluscs 13 Cancer magister Dungeness Crab 13 Octopoda Octopus 12 Palaeotaxodonta Protobranchia 11 Cancer productus Red Rock Crab 10 Lopholithodes Box Crabs 9 Paragorgia pacifica Pink Gorgonian Coral 9 Solaster stimpsoni Striped Sun Starfish 9 218 Scientific Name Common Name Number Aeolidiidae Aeolidiidae 8 Brachiopoda Lampshells 8 Cirripedia Barnacles 8 Paguridae Right-Handed Hermits 8 Phrynophiurida Phrynophiurida 8 Henricia leviuscula annectens Blood Star 7 Luidia foliolata Sand Star 6 Ctenodiscus crispatus Mud Star 5 Porifera Sponges 5 Rossia pacifica Pacific Bobtail Squid 5 Calcarea Calcareous Sponges 3 Teuthida Squids 3 Balanophyllia Balanophyllia 2 Cancridae Cancer Crabs 2 Cephalopoda Cephalopods 2 Dermasterias imbricata Leather Star 2 Lithodinae Lithodinae 2 Pandalidae Pandalid Shrimp 2 Aoridae Aoridae 1 Ascidiacea Ascidians And Tunicates 1 Balanophyllia elegans Balanophyllia Elegans 1 Brachyura True Crabs 1 Crossaster papposus Rose Starfish 1 Dorididae Dorididae 1 Euphausiacea Euphausiids 1 Euspira lewisii Lewis Moonsnail 1 Evasterias troschelii Mottled Star 1 Fusitriton oregonensis Oregontriton 1 Lithodes Lithodes 1 Oregonia gracilis Graceful Decorator Crab 1 Orthasterias koehleri Long-Armed Sea Star 1 Oxyrhyncha Spider Crabs 1 Pteraster tesselatus Cushion Star 1 Solasteridae Solasteridae 1 "@en ; edm:hasType "Thesis/Dissertation"@en ; vivo:dateIssued "2016-02"@en ; edm:isShownAt "10.14288/1.0216564"@en ; dcterms:language "eng"@en ; ns0:degreeDiscipline "Zoology"@en ; edm:provider "Vancouver : University of British Columbia Library"@en ; dcterms:publisher "University of British Columbia"@en ; dcterms:rights "Attribution-NonCommercial-NoDerivs 2.5 Canada"@* ; ns0:rightsURI "http://creativecommons.org/licenses/by-nc-nd/2.5/ca/"@* ; ns0:scholarLevel "Graduate"@en ; dcterms:title "An evaluation of the effectiveness of Rockfish Conservation Areas in British Columbia, Canada"@en ; dcterms:type "Text"@en ; ns0:identifierURI "http://hdl.handle.net/2429/55510"@en .