University of Florida Thesis Or Dissertation Formatting Template

LIFE HISTORY AND ECOLOGY OF INVASIVE LIONFISH POPULATIONS IN THE NORTHERN GULF OF MEXICO: IMPACTS TO NATIVE REEF FISH COMMUNITIES AND THEIR POTENTIAL MITIGATION

KRISTEN ANN DAHL

A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA

2019

To Mom & Dad, Mema, and Uncle Steven

ACKNOWLEDGMENTS

I could not have accomplished my PhD without the backing of my family and friends, who have always encouraged my career aspirations. To Mema (Velma Collier) and Uncle Steven (Dahl), two of the strongest people I have ever known, who are no longer here to see this achievement but inspire me to this day. I am grateful for my parents, sisters, Lana and Shea, my grandparents, and Tom TinHan, who make life worth living. They kept me level-headed, never failed to believe in me, and helped me move from Florida to Alabama, and back again. I cherish the trips they took to visit in

Alabama for Mardi Gras parades and balls, and in Gainesville for football games at The

Swamp, and trips to Ichetucknee and Cedar Key. Tom, thank you for always understanding the struggles of this degree, stemming any self-doubt, and tolerating our distance over the years. Thanks to Brian Klimek, Whitney Scheffel, and Steve Garner for being the best roommates I could ask for over the last four years. Thanks to Matt and Brittany Knee and the Anderson family for your encouragement over many more years. Your friendship and support have lifted me through uncertain times. I also must thank my dogs, Kiah and Dill, for forcing me to stay active, and always giving me something to laugh about.

I would not be here and could not have done this project without the support and guidance of my advisor, Dr. Will Patterson. From the time I first met Will during my undergraduate degree, he has mentored me and allowed me to realize my own potential as a scientist. Will has always led by example and I strive to be as driven, accomplished, and respected as he is. I thank him for providing me endless support to obtain my degree and attend professional opportunities at local, state, national, and international conferences. I thank him for his humor and compassion in difficult times. I

consider myself quite lucky to have had him as a teacher and mentor and have learned so much during this experience. After this, I plan to “keep my foot on the gas” and to make him proud.

I thank my committee members, Dr. Tom Frazer, Dr. Mike Allen, and Dr. Dave

Portnoy, for agreeing to take me on as an advised student, and for their guidance and expertise during this process. I consider myself fortunate to have had such an invested, yet easy-going supervisory committee. I thank Dave for welcoming me into his lab at

Texas A&M-Corpus Christi to learn the molecular skills necessary to apply microsatellite genotyping. I thank Mike for his support, instruction, and excitement about my project over the years beginning at the NMFS RTR Lionfish Ecosystem Modeling Workshop. I thank Tom for his help navigating my transfer to UF, as he has been a member of my committee from the start. I appreciate the time he took to explain what to expect as I neared various hurdles of the PhD. I thank other mentors from Dauphin Island Sea Lab,

Alison Robertson, Alice Ortmann, and Ken Heck. Special thanks to Alice and Alison for assisting with molecular methodology that became an important and novel part of this project. Thanks to Dick Snyder for being a fantastic teacher and mentor, and for his contributions to my lionfish removal experiment. I thank Derek Hogan, John Johnson and John Gold for welcoming me at TAMU-CC and for contributions to the development of my microsatellite genotyping work. I thank Morgan Edwards for her assistance and experience in aging lionfish otoliths.

Next, I thank Dalton Kennedy, Clint Retherford, Scott Bartel, Bryan Clark & Anna

Clark, Josh & Joe Livingston, Andy Ross, Alex Fogg, Bob & Carol Cox, Grayson

Shepherd, Michael Day, Jeremy Porter, Kylie Gray and Meaghan Faletti for help in

scuba diving and collecting thousands of lionfish. I thank charter vessel captains Johnny

Greene, Gary Jarvis, Josh Livingston, Sean Kelley, and Seth Wilson and their crews for aid in ROV sampling. You all have made enormous contributions to this work, and I admire your abilities and sheer determination to fight this invasive species. Above all, you are great people who I enjoyed spending time with over the years of this project.

I am indebted to a great deal of people that helped shape my graduate school journey. I have worked at two labs in pursuit of my PhD - Dauphin Island Sea Lab and here at the University of Florida. When I started out in the lab, I was the only woman in a lab of men. I will always remember times spent debating endless subjects in the lab and vying for who would stay working the latest. I thank Joe Tarnecki and Steve Garner for their help and incredible skill with ROV operating. I thank Joe, again, for help learning video analysis, and diet analysis, and Brian Klimek, Miaya Glabach, Steve

Garner, Michael Norberg, Justin Lewis, Jordan Bajema, Beverly Barnett, Erin Bohaboy,

Holden Harris, Joe Kuehl, Lauren Still, Gracie Barnes and Sarah Friedl for help in the field and laboratory. I thank Amanda Barker, Lei Wang, Pearce Cooper, Natalie Ortell,

Pavel Dimens, Shannon O’Leary, Dannielle Kulaw, Liz Hunt, Dom Swift, and Stuart

Willis for support in molecular labs at DISL and TAMU-CC. The thousands of hours spent aboard offshore vessels, processing whole lionfish, reading ROV videos, pipetting, cutting otoliths, and writing was made faster, more efficient, and much more fun with all of your help. Finally, thanks to all the other friends I have made from DISL and UF during my PhD, who are now spread out at STEM jobs across the country. To all those that gave me camaraderie, feedback, motivation, and charity, I thank you.

TABLE OF CONTENTS

page

ACKNOWLEDGMENTS ...... 4

LIST OF TABLES ...... 10

LIST OF FIGURES ...... 12

LIST OF ABBREVIATIONS ...... 15

ABSTRACT ...... 17

CHAPTER

1 INTRODUCTION AND JUSTIFICATION FOR STUDY...... 19

2 HABITAT-SPECIFIC DENSITY AND DIET OF RAPIDLY EXPANDING POPULATIONS OF INVASIVE RED LIONFISH, PTEROIS VOLITANS, POPULATIONS IN THE NORTHERN GULF OF MEXICO ...... 29

Materials and Methods...... 32 Lionfish Density Estimates ...... 32 Sampling Lionfish Tissues ...... 34 Diet Analysis ...... 35 Muscle Stable Isotope Analysis ...... 36 Results ...... 37 Lionfish Density and Size ...... 37 Lionfish Diet Analysis ...... 38 Muscle Stable Isotope Analysis ...... 41 Discussion ...... 42 Lionfish Density Trends ...... 42 Lionfish Trophic Ecology in the nGOM ...... 44 Conclusions and Implications ...... 47

3 DNA BARCODING SIGNIFICANTLY IMPROVES RESOLUTION OF INVASIVE LIONFISH DIET IN THE NORTHERN GULF OF MEXICO ...... 60

Materials and Methods...... 63 Study Location and Specimen Collection ...... 63 Visual Gut Content Analysis ...... 63 DNA Barcoding Preparation and Analysis ...... 64 Barcoding Sequence Analysis ...... 66 Incorporating Barcode Information into Diet ...... 67 Results ...... 68 Discussion ...... 73 Potential Cannibalism ...... 74

Trends in Diet with Unidentified Fish Resolved ...... 78 DNA Barcoding Efficacy ...... 80

4 GENOTYPING CONFIRMS SIGNIFICANT CANNIBALISM IN NORTHERN GULF OF MEXICO INVASIVE RED LIONFISH, PTEROIS VOLITANS ...... 101

Methods and Materials...... 104 Sample Collection ...... 104 DNA Barcoding of Unidentified Lionfish Prey ...... 105 Microsatellite Analysis ...... 106 Results ...... 109 Discussion ...... 110 Lionfish Cannibalism: Causes and Consequences ...... 113 DNA Barcoding and Predator-Prey Interactions ...... 116

5 DENSITY-DEPENDENT CONDITION AND GROWTH OF INVASIVE LIONFISH IN THE NORTHERN GULF OF MEXICO ...... 124

Materials and Methods...... 127 Study Location and Specimen Collection ...... 127 Weight-Length Relationships and Condition...... 128 Age Estimation ...... 128 Results ...... 131 Density Trends ...... 131 Population Demographics ...... 132 Weight-Length Relationships and Body Condition ...... 132 Age Estimates, Precision, and Margin Condition ...... 133 Population Dynamics and Growth ...... 134 Discussion ...... 136 Density-Dependent Growth and Condition ...... 136 Age and Growth in the nGOM ...... 141 Conclusions and Implications ...... 143

6 EXPERIMENTAL ASSESSMENT OF LIONFISH REMOVALS TO MITIGATE REEF FISH COMMUNITY SHIFTS ON NORTHERN GULF OF MEXICO ARTIFICIAL REEFS...... 159

Materials and Methods...... 163 Study Region and Experimental Reefs ...... 163 ROV Video Sampling and Analysis ...... 163 Targeted Removals of Lionfish ...... 165 Data Analysis ...... 166 Pre-Removal Experiment ...... 166 Removal Experiment ...... 167 Results ...... 169 Pre-Removal Experiment ...... 169 Removal Experiment ...... 170

Discussion ...... 174

7 CONCLUSIONS AND STUDY IMPACT ...... 199

Population Trends in the nGOM ...... 200 Invasive Lionfish ...... 200 Native Reef Fish Communities ...... 202 Lionfish Feeding Ecology in the nGOM...... 203 Age and Growth of Lionfish in the nGOM ...... 208 Potential Mitigation of Lionfish ...... 210 Future Directions ...... 212

LIST OF REFERENCES ...... 216

BIOGRAPHICAL SKETCH ...... 238

LIST OF TABLES

Table page

2-1 Prey taxa observed in red lionfish stomachs sampled in the northern Gulf of Mexico ...... 50

2-2 PERMANOVA table of factors affecting lionfish diet by prey mass ...... 52

3-1 Fish prey taxa observed in red lionfish (Pterois volitans) stomachs sampled in the northern Gulf of Mexico...... 84

4-1 Microsatellite loci used to determine individual identities for consumers and prey to determine cannibalism events ...... 119

4-2 Results from micro-checker analyses (van Oosterhout et al. 2004) showing the number of expected homozygotes (He) and observed homozygotes (Ho), and presence/absence of stuttering, scoring errors and null alleles ...... 120

5-1 Three-factor analysis of variance (ANOVA) results for model testing the effect of sex, habitat, time (early versus late invasion), and interactions on relative condition factor, Kn, of northern Gulf of Mexico lionfish ...... 145

5-2 Parameters of von Bertalanffy growth models estimated for all lionfish, and lionfish grouped by sex and habitat combinations sampled from the northern Gulf of Mexico ...... 145

5-3 Three-factor analysis of variance (ANOVA) results for model computed to test the effects of sex, habitat, and integer age on total length (mm) of northern Gulf of Mexico lionfish ...... 146

6-1 Taxonomic groupings for small (<100 mm TL) demersal reef fishes, exploited fishery species (>200 mm TL) and small (<150 mm TL) pelagic planktivores. . 182

6-2 PERMANOVA results of model computed to test for differences in reef fish community structure (species composition and relative abundance) between samples collected in 2009-10 versus 2011-12 estimated via ROV ...... 184

6-3 One-way repeated measures ANOVA results for models computed to test the effect of timing, 2009-10 versus 2011-12, on reef fish diversity indices and number of individuals ...... 185

6-4 Post hoc pairwise multiple comparisons (Tukey) for significant main test results from one-way repeated measures ANOVA testing effect of time (2009-10 versus 2011-12 versus 2013-14) on reef fish diversity indices ...... 186

6-5 Mean density (fish 100 m–2) and percent change in the 25 most abundant fishes observed at study artificial reef sites in 2009-10 prior to lionfish presence versus in 2011-12 after lionfish presence was confirmed ...... 186

6-6 PERMANOVA results for models computed to test the effects of lionfish removal treatment and sample timing on reef fish community structure (species composition and relative abundance) estimated via ROV ...... 187

6-7 Two-way repeated measures ANOVA results for models computed to test the effects of lionfish removal treatment (control, clear-once, maintain-clear) and sample timing on reef fish diversity indices and number of individuals...... 188

LIST OF FIGURES

Figure page

2-1 Location of study sites for lionfish density and diet sampling ...... 53

2-2 Mean densities of lionfish 100m-2 on natural and artificial reefs from fall 2010 to fall 2013 ...... 54

2-3 Length to weight relationship of speared lionfish predicts lionfish mass from ROV length observations ...... 55

2-4 Habitat-specific total length distributions of lionfish samples by season from April 2013 to March 2014 ...... 56

2-5 Indices of lionfish diet by habitat, season and size class. Stacked bar plots of A) mean percent diet by number of prey items, B) mean percent diet by prey mass, and C) mean percent index of relative importance by prey category ...... 57

2-6 Bi-plot of lionfish muscle δ15N versus δ13C values from samples collected at northern Gulf of Mexico natural (NR) and artificial (AR) reefs in spring 2013 and winter 2014 ...... 59

3-1 Location of study natural and artificial reefs where red lionfish, P. volitans, were sampled for visual and DNA barcoding diet analyses ...... 88

3-2 PCR and sequencing primers used in the DNA barcoding of unidentified fish prey from invasive red lionfish (Pterois volitans) in the northern Gulf of Mexico...... 89

3-3 Size distribution of COI barcoding sequence read lengths in number of nucleotide base pairs for unidentified red lionfish (Pterois volitans) fish prey items, n=696...... 90

3-4 Red lionfish (Pterois volitans) prey DNA samples with initial sequence length >300 bp selected for duplicate DNA barcoding analysis...... 91

3-5 Red lionfish (Pterois volitans) prey DNA samples with initial sequence length <300 bp selected for duplicate DNA barcoding analysis...... 93

3-6 Examples of prey fish visually identified as P. volitans from lionfish stomach contents ...... 97

3-7 Stacked bar plots of mean percent red lionfish (P. volitans) diet by mass including visually identified prey items only, visual + unidentifiable prey items, and prey items identified visually and with DNA barcoding ...... 98

3-8 Mass distribution of 696 prey samples of red lionfish (P. volitans) identified via DNA barcoding to species, genus, family level, or to red lionfish or vermilion snapper ...... 99

3-9 Species accumulation curves of red lionfish (P. volitans) prey taxa identified during this study ...... 100

4-1 Microsatellite genotypes of consumers and prey pairs analyzed to evaluate potential cannibalism ...... 121

5-1 Maps of the northern Gulf of Mexico indicating A) the study region and B) natural (triangles) and artificial (circles) reefs where lionfish were sampled for age and growth analyses ...... 147

5-2 Invasive lionfish density 100m-2 estimated via ROV at northern Gulf of Mexico natural and artificial reef locations reported in Chapter 2 and updated through 2017...... 148

5-3 Summary of lionfish and otolith sample sizes relative to sampling years, and habitat type...... 149

5-4 Length-weight relationship for lionfish (n = 3,266) sampled in the northern Gulf of Mexico between 2013 and 2017...... 149

5-5 Total length and age distributions of lionfish sampled in the northern Gulf of Mexico during 2013-2017...... 150

5-6 Mean (±95% CI) relative condition factor, Kn, of northern Gulf of Mexico lionfish by habitat, sex and invasion timing ...... 151

5-7 Post-hoc pairwise multiple comparisons for significant main test results from three-factor ANOVA testing effects of sex, habitat and time on relative condition factor, Kn...... 152

5-8 Distribution of opaque zone (i.e., annuli) count differences between otolith readers for a subset of randomly selected otoliths (n = 1,000)...... 152

5-9 Trend in lionfish otolith marginal condition (n = 3,082) among months for fish sampled in the northern Gulf of Mexico between 2013 and 2017 ...... 153

5-10 Sex- and habitat-specific scatterplots of lionfish total length versus age. Plotted lines are von Bertalanffy growth function fits to the data (t0 fixed at 0.072 y), with functions given on each panel...... 154

5-11 Mean (±95% CI) size-at-age of lionfish by sex and habitat for ages 1 to 6 years. Asterisks indicate significant (α < 0.05) effect of habitat within sex and year class ...... 155

5-12 P-values of post-hoc pairwise multiple comparisons for significant main test results from three-factor ANOVA testing effects of sex, habitat and age on mean total length (i.e., size-at-age)...... 156

5-13 Sex-specific scatterplots of mean size-at-age versus lionfish density at sampled artificial reefs for ages 2-4 years ...... 157

5-14 Scatterplot of relative condition factor, Kn, versus lionfish density at sampled artificial reefs ...... 158

6-1 Map of the northern Gulf of Mexico indicating the locations of the Escambia East-Large Area Artificial Reef Site (EE-LAARS) and the 27 experimental reefs under study ...... 189

6-2 Species diversity indices and number of individual fish across taxa on study reefs during spring 2009-winter 2010, fall 2011-summer 2012, and then computed during the removal experiment from fall 2013-summer 2015 ...... 190

6-3 Mean (±SE) density (fish 100 m–2) of fish taxa observed in remotely operated video samples at artificial reef study sites during spring 2009-winter 2010, fall 2011-summer 2012, and during the lionfish removal experiment ...... 191

6-4 Mean (±SE) lionfish density estimated from counts made with a remotely operated vehicle (ROV) and then scaled (x1.29) to correct for incomplete detectability...... 193

6-5 Two-factor repeated measures ANOVA results for model computed to test the effects of lionfish removal treatment and sample timing on lionfish density (fish 100 m–2) estimates at study artificial reefs ...... 194

6-6 Total length distributions of lionfish estimated with a red laser scaler and remotely operated vehicle at control (no lionfish removal) artificial reef study sites from fall 2013 through summer 2015...... 195

6-8 Post hoc pairwise comparisons for significant main test results from PERMANOVA testing effect of removal treatment and sample timing on reef fish community structure of nGOM removal experimental artificial reefs ...... 196

6-9 Post hoc Tukey tests from two-way repeated measures ANOVA testing effect of removal treatment and sample timing on reef fish diversity indices and number of individuals on nGOM removal experimental artificial reefs...... 197

6-10 Scatterplot of estimated lionfish density versus days after lionfish removal for clear-once experimental artificial reef sites and the line fit to the significant fixed effect of days after removal ...... 198

LIST OF ABBREVIATIONS

ANOVA Analysis of variance

APE Average percent error

AR Artificial reef

BOLD Barcode of Life Database bp Base pairs

CI Confidence interval

COI Cytochrome c oxidase subunit I

DWH Deepwater Horizon Oil Spill

EE-LAARS Escambia-East Large Area Artificial Reef Site

GOM Gulf of Mexico

He Expected heterozygosity

Ho Observed heterozygosity

Hs Unbiased gene diversity

H’ Shannon-Wiener diversity index

J’ Pielou’s evenness index

MtDNA Mitochondrial DNA

NCBI National Center for Biotechnology Information

Ne Effective number of alleles nGOM Northern Gulf of Mexico

NR Natural reef

PCR Polymerase chain reaction

PERMANOVA Permutational multivariate analysis of variance

ROV Remotely operated vehicle

SAB South Atlantic Bight

SD Standard deviation

SE Standard error

TL Total length

UNID Unidentified

WFS West Florida Shelf

%F Percent frequency of occurrence

%IRI Percent index of relative importance

%M Percent mass

%N Percent number

Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy

LIFE HISTORY AND ECOLOGY OF INVASIVE LIONFISH POPULATIONS IN THE NORTHERN GULF OF MEXICO: IMPACTS TO NATIVE REEF FISH COMMUNITIES AND THEIR POTENTIAL MITIGATION

Kristen Ann Dahl

May 2019

Chair: William F. Patterson III Major: Interdisciplinary Ecology

Species invasions have been increasing in frequency and severity in recent decades, furthering the need for research on ecological impacts from invasions and means to mitigate them. The spread of Indo-Pacific lionfish (Pterois spp.) into the northern Gulf of Mexico (nGOM) causes concern, given potential negative impacts on native fish and invertebrate communities.

My first objective was to document the progression of the lionfish invasion in the nGOM by monitoring densities among natural and artificial reef habitats. Lionfish density in both habitats exponentially increased from 2010 to 2014, and afterwards stabilized or slightly declined. Density at artificial reefs was two orders of magnitude greater than at natural reefs throughout the study.

To characterize trophic impacts of lionfish on native reef fish communities of the nGOM, I report a comprehensive assessment of feeding ecology using traditional diet, stable isotope, and DNA barcoding analyses. Results indicate lionfish are generalist mesopredators that become more piscivorous at larger sizes. However, lionfish exhibit a more varied diet at artificial reefs, where they forage on open substrates away from reef

structure. DNA barcoding of unidentified fish prey significantly increased diet resolution and exposed potential cannibalism occurrence. Cannibalism was later confirmed using microsatellite genotyping, and increased in frequency through time, mirroring increases in lionfish density.

Next, I estimated age, growth, and condition of lionfish sampled over five years of invasion, with the objective to test for density-dependent effects. Significant declines in mean size-at-age and condition as a function of lionfish density indicated density- dependent effects that were likely due to inter- and intra-specific competition. The increase in these effects through time likely explains the plateauing of nGOM populations in latter years of study.

Finally, I conducted a 2-year experiment to examine the effectiveness and ecological benefits of lionfish removals. Removals reduced lionfish densities, but juveniles and adults quickly recruited to cleared reefs, thus removal efforts were insufficient to achieve native reef fish recovery. This work has important implications for invasive lionfish population dynamics and carrying capacity in the nGOM, and data herein will be used to parameterize models estimating the removal effort necessary to mitigate their impacts effectively.

CHAPTER 1 INTRODUCTION AND JUSTIFICATION FOR STUDY

The rise of globalization in the latter half of the 20th century facilitated the transfer of ideas and cultures around the world and fostered economic growth in many nations

(Castells 2010). However, one unintended consequence of globalization has been the translocation of species across biogeographic barriers that otherwise would naturally restrict their ranges and potential for exchange (Crawley 1987). This has led to an exponential increase in introduced species around the world, which represent a diverse array of taxonomic groups and biogeographic origins (Elton 1958, Mack et al. 2000). In some cases, introduced species may become invasive, with a biological invasion being defined as the arrival and establishment of these organisms into, and subsequent diffusion from, a community in which they did not previously exist (Carlton 1989). While the vast majority of species introductions do not result in populations established in novel habitats, some do flourish and become invasive, often to the detriment of native communities (Elton 1958, Lodge 1993). These organisms can significantly alter ecological functioning of native communities by reducing biodiversity, displacing native species, altering community structure, introducing pathogens, or initiating trophic cascades (Ruiz et al. 1997, Chapin et al. 2000, Grosholz et al. 2000, Mack et al. 2000,

Olden et al. 2004). Biological invasions contribute to a growing number of anthropogenic stressors on marine ecosystems in recent decades, such as climate change, habitat degradation, and overfishing and have far-reaching implications for native biota, ecosystem functioning, and human economies (Pimentel et al. 2000, Crain et al. 2008). Ongoing species introductions and their resulting invasions are considered to be one of the leading causes of ecosystem disturbance and human-caused change

around the world and are implicated as one of the greatest causes of global biodiversity loss (Chapin et al. 2000, Barnes and Milner 2005).

Species invasions in marine systems have been increasing in frequency and severity in recent decades (Semmens et al. 2004, Molnar et al. 2008), furthering the need for research on the ecological and socioeconomic impacts from invasions, as well as on means to mitigate these impacts (Mooney and Cleland 2001, Pimentel et al.

2005). The growing body of literature on invasion ecology suggests invasion success and the severity of ecological impacts depends on the life history and trophic dynamics of the invader, as well as the ecology of the invaded community (Levine 2000, Kolar and

Lodge 2001). For example, species possessing traits such as rapid reproduction, high dispersal potential, toxicity, and close association with humans are more likely to become successful invaders (Holway and Suarez 1999, Bax et al. 2003). Biodiversity, the status of upper trophic level organisms, and the degree of habitat degradation within recipient communities also can greatly influence invasion success (Lodge 1993, Moyle and Light 1996). Invasions by predators (i.e., upper trophic level organisms) are expected to have the most damaging impact on native ecosystems given that predator- prey interactions greatly influence community assemblages in both terrestrial and marine systems (Paine 1966, Hixon and Carr 1997, Grosholz et al. 2000, McDonald et al. 2001, Caut et al. 2008). It is essential to understand how invasive predators and native prey interact to effect change in the structure and function of the marine communities in which they invade (Rilov 2009).

The invasion of a marine predator is currently ongoing in the western Atlantic

Ocean. Indo-Pacific lionfishes, Pterois volitans/miles complex (hereafter lionfish), have

exhibited an invasion so extensive and rapid in this region that they are considered the most successful marine fish invader to date (Morris and Whitfield 2009, Albins and

Hixon 2013, Côté et al. 2013a). Lionfish (Family: Scorpaenidae) were first introduced into the western Atlantic region off southeastern Florida in the late 1980s (Schofield

2009). Due to their ornate fins, aposematic coloration, and hardiness in captivity, lionfish are highly valued in the aquarium trade, in which fish are captured in their native region and then sold throughout the US (Whitfield et al. 2002). Accidental release from the aquarium trade is considered to be the most likely vector of initial lionfish introduction into the western Atlantic and may continue to be a source of propagule pressure

(Whitfield et al. 2002, Semmens et al. 2004). Invasive lionfish populations spread throughout the US South Atlantic Bight (SAB) in the 1990s, and their invasive range expanded throughout the Caribbean Sea in the 2000s (Whitfield et al. 2006, Schofield

2010). The Gulf of Mexico (GOM) is the most recently invaded basin, where lionfish were first reported in 2009 off the northern Yucatan Peninsula, Mexico (Aguilar-Perera and Tuz-Sulub 2010), in the Florida Keys, and along the west Florida shelf (Schofield

2010). By late 2010, lionfish had been observed in eastern, northern, and western regions of the GOM (Schofield 2010, Fogg et al. 2013, Dahl and Patterson 2014, Nuttall

2014). At the time of this writing, lionfish have established an invaded area of over 7 million km2, recruiting to a diversity of habitat types including mangroves (Barbour et al.

2010), seagrass beds (Claydon et al. 2012), mesophotic reefs (Nuttall 2014), and artificial reefs (Smith and Shurin 2010) across the US southeast Atlantic coast,

Caribbean Sea, and GOM (Schofield 2010, Côté et al. 2013a, Schofield et al. 2014).

Numerous life history and behavioral traits of lionfish are believed to facilitate their continued spread and population growth. In the western Atlantic, lionfish are generalist mesopredators that consume a broad range of fish and invertebrate prey

(Morris and Akins 2009, Muñoz et al. 2011), including the juvenile stages of ecologically and economically important species (Lesser and Slattery 2011). Lionfish are highly efficient predators consuming 1-2 prey items h-1 on average (Côté and Maljković 2010), a distinction that has been attributed to unique predatory behaviors and prey naïveté, wherein prey fishes do not recognize lionfish as a potential predator (Fishelson 1997,

Hornstra and Herrel 2004, Albins and Lyons 2012, Cure et al. 2012). Lionfish are thought to experience little to no predation themselves, in part due to the presence of venomous dorsal, pelvic, and anal spines (Maljković et al. 2008, Sih et al. 2010, Albins and Hixon 2013, Benkwitt et al. 2017). As opportunistic life history strategists, lionfish grow rapidly and reproduce at young (<1 year) ages (Ahrenholz and Morris 2010, Morris et al. 2011b). The success with which lionfish have invaded their introduced Atlantic range suggest native communities exert little biotic resistance to invasion, resulting in reduced interspecific competition (Albins 2013), few constraints on growth (Darling et al.

2011), and few effective, novel parasites (Sikkel et al. 2014, Tuttle et al. 2017).

Furthermore, predation by native predators does not appear to be regulating lionfish populations (Hackerott et al. 2013), which have reached higher densities and body sizes than are observed in their native Indo- Pacific (Darling et al. 2011, Kulbicki et al. 2012,

Pusack et al. 2016).

Lionfish have the potential to cause direct and indirect negative impacts on native reef fish communities in the western Atlantic. Established invasive lionfish populations

have been shown to cause dramatic shifts in native reef fish community structure and trophic dynamics in coral reef and other hardbottom ecosystems (Albins and Hixon

2008, Darling et al. 2011, Lesser and Slattery 2011, Green et al. 2012). The direct impact of lionfish predation has caused substantial declines in the abundance of small adult reef fishes, as well as juvenile recruits of larger reef fish species, such as parrotfishes (Albins and Hixon 2008). Lionfish have caused significant declines in prey fish biomass (Green et al. 2012) and species richness (Albins and Hixon 2008) over brief spans of time following their arrival on small patch reefs in The Bahamas. Lionfish may also act to destabilize native predator-prey dynamics given they have been shown to cause nearly three times greater prey mortality relative to native mesopredators

(Albins 2013). Native communities may also be impacted via indirect processes such as competition for prey resources or space. Dietary overlap of lionfish with native mesopredators may ultimately lead to decreases in the abundances of those species

(Layman and Allgeier 2012, Chagaris et al. 2017). Additionally, lionfish may alter reef fish behavior by their mere presence on reef structures via competition for space and shelter with native species (Curtis-Quick et al. 2013, Raymond et al. 2015). Finally, the biomass of potential lionfish predators (i.e., large piscivorous reef fishes) have been fished to historically low levels (Barbour et al. 2011, Morris et al. 2011a, Chagaris et al.

2017), negating any potential natural control mechanism for the invader.

The rapidity and expanse of the lionfish invasion coupled with the profound negative impacts to recipient ecosystems has motivated researchers to work towards best management practices to mitigate impacts to native species. Besides the ecological impacts from invasions, managers also must consider socio-economic

impacts to human health, tourism, and commercial fisheries. Nearly all lionfish management strategies hinge on the goal of a reduction of lionfish populations, thus their corresponding impacts. Targeted removal of lionfish (i.e., culling) has gained considerable attention in recent years and has been shown to reduce both the numbers and mean size of individuals (Frazer et al. 2012). Unfortunately, many removal efforts are disorganized and too small-scale to cause meaningful reductions in populations.

Lionfish populations have shown an ability to recover quickly from removals, thus requiring repeated removals from any given site to maintain low abundances (Arias-

González et al. 2011, Barbour et al. 2011). In some cases, partial culling has been effective where loss of native prey fish biomass ceased with lower effort than would be required for complete culls (Green et al. 2014); however, others still contend that all lionfish must be removed to see substantial conservation gains (Benkwitt 2015).

Promotion of the species as a food fish has gained recent popularity and could be a means to increase the scale of lionfish removal efforts (Ferguson and Akins 2010,

Morris et al. 2011c, Côté et al. 2013a).

Research is essential to predict the direct impacts of lionfish predation on regional community structure and reef fish abundance, given the taxa of highest importance to lionfish diet may vary greatly by habitat type or prey availability (Morris and Whitfield 2009). The majority of lionfish impact assessments to date have come from western Atlantic and Caribbean hardbottom (Muñoz et al. 2011) or coral reef habitats (Côté and Maljković 2010, Albins and Hixon 2013) that are quite different ecologically from the northern GOM (nGOM). This is especially true for artificial reefs which have been deployed throughout the nGOM. Artificial reef habitat has been

deployed over large (1,000s km2) areas of the nGOM and consists of manmade structures either specifically created for the purpose of housing reef-associated fishes for harvest (Bohnsack et al. 1994), or are inadvertent marine habitats such as offshore petroleum platforms (Scarborough Bull and Kendall 1994). Artificial reef habitats may differ ecologically from natural hardbottom reef habitat within the same geographic area

(Hixon and Carr 1997, Perkol-Finkel et al. 2006), and differences in community structure and habitat complexity have been shown to exist (Lingo and Szedlmayer 2006, Dance et al. 2011). Notably, artificial reef fish communities in the nGOM have much lower densities of small demersal fishes (e.g., damselfishes, blennies, gobies, and wrasses)

(Patterson et al. 2014, Tarnecki and Patterson 2015), that have been shown to be the predominate prey among lionfish sampled at natural reefs throughout the Caribbean

(Morris and Akins 2009).

Life history information of invasive species at regional and local scales is critical for successful population management. Characterization of lionfish growth in different habitats, or in response to varying levels of density, can provide information on the potential degree of impacts to the invaded community, and these life history parameters are especially important inputs for population and ecosystem models employed to predict population and community dynamics across time (Barbour et al. 2011, Morris et al. 2011a, Cerino et al. 2013, Chagaris et al. 2017). Lionfish age and growth parameters have been estimated from several locations within the western Atlantic; however, datasets are often truncated due to being sampled within too few years following colonization (Potts et al. 2010, Edwards et al. 2014, Johnson and Swenarton 2016,

Fogg 2017). Population dynamics of lionfish may also change as the invasion

progresses (Bøhn et al. 2004, Gutowsky and Fox 2012). This is because the population growth of most invasive species follows a predictable trajectory which starts with a lag period of low densities, increases to exponential growth and high densities, and eventually peaks near carrying capacity (Crooks and Soule 1999, Sakai et al. 2001).

Thus, through the process of invasion and establishment a species may experience both density-independent and density-dependent factors based on the demographic and environmental conditions present, resulting in phenotypic changes in life history traits (Bøhn et al. 2004). As the invasion of lionfish in the nGOM progresses, density- dependent processes, including decreased growth, may begin to regulate lionfish populations in the region via increased inter- and intra-specific competition for prey resources.

Evidence suggests that lionfish are likely to remain members of western Atlantic,

Caribbean, and GOM fish communities in the future, thus, there is consensus among researchers and managers that lionfish control using targeted removals be implemented to mitigate their negative effects on marine ecosystems and economies, (Morris and

Whitfield 2009, Arias-González et al. 2011). However, the potential benefits of targeted lionfish removal programs remain unclear. Baseline reef fish community information acquired from nGOM ecosystems prior to the lionfish invasion make this system an ideal model to quantify the effects of lionfish on native ecosystem structure and function, as well as to examine the efficacy of lionfish removals as a means to mitigate, and potentially reverse, their negative effects on native reef communities.

The full ecological implications of the lionfish invasion, among multiple stressors to nGOM reef ecosystems, are not well understood. The majority of recent research on

invasive lionfish comes from regions outside of the nGOM, yet regional differences in lionfish life history traits and ecological interactions with novel communities are likely to exist. Understanding how lionfish numbers are changing over time across different habitats (i.e., rate of invasion, degree of colonization) is a crucial objective. In Chapters

2 and 5, I tracked the progression of lionfish invasion by monitoring densities and size distributions of lionfish in the nGOM over time with a remotely operated vehicle (ROV).

A comprehensive understanding of lionfish trophic ecology on both natural and artificial reefs is necessary to predict impacts on native reef fish communities, especially fishery species, in the nGOM.

To address data gaps in lionfish trophic and feeding ecology in the region, I used a novel, interdisciplinary approach that comprise Chapters 2, 3, and 4. In Chapter 2, I characterized lionfish trophic ecology in the nGOM region using traditional diet and stable isotope analyses. These analyses indicated habitat-specific and ontogenetic patterns in lionfish diet. However, overall diet estimates may be biased or incomplete given high proportions of prey contents unable to be identified visually due to digestion.

Thus, in Chapter 3, I employed the molecular method of DNA barcoding to identify previously unidentifiable prey contents from Chapter 2. Objectives of this research were to determine whether the method was effective at identifying degraded prey items, and whether diet estimates were enhanced when DNA barcoded prey items were included in diet analyses. Work I conducted in Chapter 4 built on the observation of potential cannibalism in lionfish that was revealed from DNA barcoding. Using nuclear DNA microsatellites, I was able to directly compare predator and prey genotypes to test for

cannibalism. Results confirmed density-dependent cannibalism, which has the potential to limit lionfish populations.

In Chapter 5, I used samples collected over five years of invasion to estimate age, growth and condition of lionfish in the nGOM region. Specific objectives of this life history work were to test for the presence of density-dependent feedbacks and to study how different stages of invasion may impact population dynamics of an invasive species. Finally, in Chapter 6, I discuss a field experiment of targeted removals conducted over two years in which I assessed the effectiveness of removals to control lionfish densities. Through extensive baseline data on nGOM reef fish communities, I was also able to assess the effectiveness of lionfish removal to elicit community recovery following initial population declines. In the final chapter, I synthesize the findings of my dissertation research and consider their implications for native reef fish communities in the nGOM, as well as the insights they provide for the future lionfish population dynamics and management in the region.

CHAPTER 2 HABITAT-SPECIFIC DENSITY AND DIET OF RAPIDLY EXPANDING POPULATIONS OF INVASIVE RED LIONFISH, PTEROIS VOLITANS, POPULATIONS IN THE NORTHERN GULF OF MEXICO

Introduction of exotic species to marine ecosystems has been increasing in frequency and severity around the globe, which has also led to an increase in species invasions (Carlton 1989, Molnar et al. 2008). Invasive species can significantly transform recipient communities where they reduce biodiversity, displace native species, alter community structure, introduce pathogens, or initiate trophic cascades

(Ruiz et al. 1997, Chapin et al. 2000, Mack et al. 2000, Olden et al. 2004).

Anthropogenic activities, namely trade, commerce, and aquaculture, are escalating the rate of species introductions, furthering the need for research on the prevention and mitigation of potential ecological and socioeconomic impacts from invasions (Pimentel et al. 2005, Lockwood et al. 2007). The severity of ecological impacts depends on the life history and trophic dynamics of the invader, as well as the ecology of the invaded community (Levine 2000, Kolar and Lodge 2001). Predator-prey interactions are known to shape community assemblages in both terrestrial and marine systems (Paine 1966,

Hixon and Carr 1997), thus predator invasions are expected to have the most damaging impact on native ecosystems (Grosholz et al. 2000, McDonald et al. 2001, Caut et al.

2008).

Predator invasions in marine ecosystems are atypical, yet the invasion of Indo-

Pacific lionfishes (Pterois volitans/miles) complex in the western Atlantic Ocean has

Reprinted by permission from Kristen Dahl. PLOS. PLOS ONE. Dahl KA, Patterson WF III. Habitat- Specific Density and Diet of Rapidly Expanding Invasive Red Lionfish, Pterois volitans, Populations in the Northern Gulf of Mexico. 2014.

been so extensive and rapid that lionfish are considered the most successful marine fish invaders to date (Morris and Akins 2009, Green et al. 2011, Albins 2013). Lionfish have established an expansive invaded area that includes the US Southeast Atlantic coast, the Caribbean Sea and portions of the Gulf of Mexico (GOM) (Schofield 2010,

Ruttenberg et al. 2012). The GOM is the most recently invaded of these basins, where lionfish were not reported until 2009 off the northern Yucatan peninsula, Mexico

(Aguilar-Perera and Tuz-Sulub 2010). Red lionfish, Pterois volitans (hereafter lionfish), were first reported from northern GOM (nGOM) in summer 2010 and have since been observed in the western GOM as well (Curtis-Quick et al. 2013, Fogg et al. 2013, Nuttall

2014).

Several life history and behavioral traits of lionfish are thought to facilitate their continued spread and population growth. For example, lionfish are voracious, novel predators that consume a wide variety of naïve prey in the western Atlantic, but experience little to no predation themselves, in part due to the presence of large, venomous dorsal, pelvic and anal spines (Morris and Whitfield 2009, Sih et al. 2010,

Albins and Hixon 2013). Lionfish can reach sexual maturity within one year (Morris

2009) and have a high reproductive output (Morris and Whitfield 2009, Morris et al.

2011b). Therefore, lionfish populations in their invaded range have the potential to reach far greater densities than those reported in the Indo-Pacific (Whitfield et al. 2006,

Green and Côté 2009, Darling et al. 2011, Kulbicki et al. 2012). Among invaded western

Atlantic reef communities, lionfish appear to have damaging effects on native fishes due to the direct consumption of a broad array of native fishes, including some economically important reef fishes (Albins and Hixon 2008, Green 2012). However, the majority of

lionfish impact assessments to date have come from south Atlantic hardbottom (Muñoz et al. 2011) or coral reef habitats (Albins and Hixon 2008, Côté and Maljković 2010) that are quite different ecologically from nGOM reef habitats. This is especially true for artificial reefs which have been deployed throughout the nGOM and whose reef fish communities have much lower densities of small demersal fishes (e.g., damselfishes, blennies, gobies, and wrasses) (Dance et al. 2011, Patterson et al. 2014), that have been shown to be the preferred prey among lionfish sampled at natural reefs throughout the Caribbean (Morris and Akins 2009).

Local research is essential to estimate the direct impacts of invasive lionfish on native fishes, thus an understanding of lionfish diet on both natural and artificial reefs is necessary to predict their impacts in the nGOM. The first objective of this study was to document the progression of the lionfish invasion in an area of the nGOM by monitoring lionfish densities among natural and artificial reef habitats. I also characterized lionfish feeding ecology in the region to recognize potential direct and indirect impacts of lionfish on native reef fish communities of the northern GOM. Stomach content analysis was employed to test for seasonal and ontogenetic effects on lionfish diet between natural and artificial reefs. Traditional diet analysis relies on recently ingested prey, thus was complimented by stable isotope analysis of lionfish white muscle tissue which provided information on the trophic ecology of lionfish revealed by isotopic dietary signals integrated over the previous weeks to months. Results discussed below have clear implications for predicting both direct and indirect effects of invasive lionfish on natural and artificial reefs in the nGOM.

Materials and Methods

Lionfish Density Estimates

Northern GOM natural (n = 16) and artificial (n = 22) reefs were surveyed with a micro remotely operated vehicle (ROV) each fall (October to December) from 2009 through 2013 to examine changes in lionfish density and size distribution over time (Fig.

2-1 B). Reefs were randomly selected from a larger sample frame of regional reefs

(Dance et al. 2011, Patterson et al. 2014), and ranged in depth between 17 and 73 m.

Sampling was conducted with a VideoRay Pro4 ROV (dimensions: 36 cm long, 28 cm tall, 22 cm wide; mass = 4.8 kg). The ROV has a depth rating of 170 m, a 570-line color camera with wide angle (116º) lens, and was equipped with a red laser scaler to estimate fish size. The laser scaler consisted of two 5-mw @ 635 nm (red) class IIIa lasers mounted in a fixed position 75 mm apart. The ROV was tethered to the surface where it was controlled by a pilot via an integrated control box that contains a 38-cm video monitor to observe and capture digital video captured by the ROV’s camera during sampling.

Video sampling was conducted at study reefs with either a point-count or transect method, depending on habitat type and dimensions. The point-count method, which is described by Patterson et al. (2009), was used to sample a 15-m cylinder around isolated reef habitat, such as single artificial reef modules. In that method, the ROV was positioned 1 m above the seafloor and approximately 5 m away from a given reef. The

ROV was slowly pivoted 360º and then moved to the opposite side of the reef. Once there, it was again positioned 1 m above the seafloor and approximately 5 m away from the reef and pivoted 360º. The ROV then was flown to 1 m directly above the reef and pivoted 360º to video fishes in the water column above the reef. Next, the ROV was

flown to 10 m above the reef and pivoted 360º. Once all sample segments were completed, the ROV was flown back down to the reef to observe fishes located on the reef’s surface or inside the reef structure.

A transect sampling method was utilized for reef habitat that was more broadly distributed, such as was characteristic of natural reef habitat examined in this study. In this method, a 5-m wide transect was video sampled as the ROV moved forward at a rate of approximately 0.5 m s-1 along a 25-m long transect. The width of the transect was controlled by flying the ROV with a camera angle of 45º approximately 1 m above the seabed given the 116º viewing angle of the camera (Patterson et al. 2014). Four orthogonal transects were flown over natural reef habitats, thus a total area of approximately 500 m2 of reef habitat was surveyed. The distance covered on a given transect was controlled by flying the ROV with a fixed scope of tether away from a 5-kg clump weight attached in-line to the tether. Transect distance was confirmed with a

Tritech MicroNav ultrashort baseline acoustic positioning system deployed with the

ROV.

Analysis of video samples was performed with a Sony DVCAM DSR-11 digital

VCR and a Sony LMD-170 high resolution LCD monitor. When the point-count method was employed, lionfish counts were summed among all sampling segments and then divided by the sample area (176.7 m2) to estimate fish density. Lionfish density for transect samples was computed by summing counts and then dividing by the total area estimated to have been sampled among transects. Total length (TL) was estimated for lionfish struck by the red dots of the laser scaler by first multiplying the length of a fish measured in a video frame by the known distance between lasers (75 mm), and then

dividing that product by the distance measured between lasers in the frame. Patterson et al. (2009) estimated a mean negative bias of 3% (SD = 0.6) resulted from this method, thus estimated lionfish TL was bias-corrected based on a random probability draw and normally distributed bias with mean equal to 3 % and a standard deviation of

0.6%.

The difference in lionfish density between natural versus artificial habitats and among years was tested with a two-factor analysis of variance (ANOVA) model, with

Tukey’s multiple comparison procedure computed to test all pairwise comparisons. Too few TL estimates were available to test the habitat effect, thus TL was pooled between artificial and natural reefs and the effect of year was tested with a single-factor ANOVA model, with Tukey’s multiple comparison procedure computed to test all pairwise comparisons. A priori, α was set to 0.05 for all statistical tests.

Sampling Lionfish Tissues

Lionfish were sampled by divers with spears to examine lionfish trophic ecology at nGOM reefs. Dive trips were made seasonally from April 2013 through March 2014 to both natural and artificial reefs, with sampling reefs ranging in depth from 24 to 35 m

(Fig. 2-1 C). Spearing of lionfish was localized immediately posterior to the head which severed their spinal column. Fish were dead upon arrival to the surface where carcasses were placed in mesh bags in an ice-slurry. Once on land, fish were ranked by size and systematic random sampling was employed to sample every nth fish such that approximately 100 fish were sampled per habitat type per season. Lionfish samples were weighed to the nearest 0.1 g and measured to the nearest mm TL. Approximately

30 g of white muscle tissue was dissected from each fish above its pectoral fin. Muscle tissue was placed in plastic bags and frozen at -80º C. Stomachs were dissected after

inspecting gills for regurgitated prey and their contents placed in plastic bags, fixed in

100% ethanol.

A non-linear regression (mass = aTLb) was fit to lionfish mass and TL data. The fitted equation was then employed to predict mass of lionfish scaled with the red laser scaler in ROV video samples. The difference in mean predicted mass among years was tested with ANOVA. Tukey’s multiple comparison procedure also was computed to test for differences in predicted mass between each pair of years.

Diet Analysis

Prey items in stomach samples were sorted to the lowest taxonomic level possible, counted, and then dried at 60º C for at least 48 h to obtain dry mass. Prey taxa were grouped into seven prey categories: shrimps, crabs, other benthic invertebrates, pelagic invertebrates, reef fishes, non-reef benthic fishes, and pelagic fishes. Percent mass and percent number by prey category were computed for each sample that had prey items present. Percent frequency of occurrence (%F) was calculated among fish captured in a given season and from a given habitat type as the number of stomachs containing a particular prey category divided by the number of stomachs with prey present (Bowen 1996). The index of relative importance (IRI) was then computed as IRI

= (%M + %N) x %F (Hacunda 1981), and %IRI was calculated by dividing the IRI value for each prey category by the sum of the IRI values among all prey categories and multiplying by 100.

The effects of habitat type (natural versus artificial reef), season, and size class

(small: <200 mm, medium: 200-250 mm, and large: >250 mm TL) on lionfish diet by %M and %N were tested with three-factor permutational multivariate analysis of variance

(PERMANOVA) models computed with the Primer statistical package (ver. 6; Anderson

et al. 2008). Data were square-root transformed and a dummy variable (value = 1) was added to each sample prior to computing the Bray-Curtis similarity measure among all pairs of samples. Then, PERMANOVA models were computed with 10,000 permutations to test if the pattern observed in Bray-Curtis similarity between habitats, among seasons, or among fish size classes was significantly different from random. All significant effects or interactions observed in PERMANOVA results were further examined via pair-wise PERMANOVA tests.

Muscle Stable Isotope Analysis

Stable isotope analysis was conducted for lionfish muscle samples collected in spring 2013 and winter 2014. Samples from 8 fish from each habitat in each of these seasons were selected with systematic random sampling for analysis. Muscle tissue samples were dried in an oven at 60º C for at least 48 h, and then ground in a tissue grinder prior to being pulverized with a glass mortar and pestle. Mortars and pestles were rinsed with deionized water and air-dried between samples, while the tissue grinder was wiped free of dried tissue remnants with a lint-free laboratory tissue. Muscle samples were analyzed for δ13C and δ 15N with a Thermo-Finnigan MAT Delta+

Advantage stable isotope ratio-mass spectrometer (SIR-MS) equipped with an elemental analyzer at the Marine Science Institute of the University of California Santa

Barbara. Values of δ 13C are reported relative to the international standard Vienna

Peedee Belemnite, and δ 15N values are reported relative to atmospheric nitrogen, which is isotopically homogenous. Isotope ratios for both C and N are reported in the

13 15 standard delta notation: δX = [(Rsample/Rstandard) – 1] x 1000, where X = C or N and R

= 13C:12C or 15N:14N. Check standards run periodically during the analysis included US

Geological Survey standard reference materials 40 and 41 (glutamic acid).

Values of δ 13C were corrected for %lipid with the regression equation reported by Post et al. (2007), for aquatic animals: ∆δ13C = –3.32 + (0.99 x C:N), where ∆δ13C is the correction applied to δ13C to account for %lipid and C:N is a proxy for %lipid.

Correlation analyses were computed to test the relationship between fish TL and δ15N, and between TL and δ13C. Differences in TL, δ13C, and δ15N were tested between natural and artificial reefs and between spring 2013 and winter 2014 were tested with two-factor ANOVA models. In the case of a significant interaction term, all pairwise multiple comparisons were computed with Holm-Sidak tests.

Results

Lionfish Density and Size

Lionfish density increased rapidly from fall 2010, when no fish were observed at study reefs (no ROV surveys conducted at artificial reefs in fall 2010), through fall 2013, when mean density was 0.49 fish 100 m–2 on natural reefs and 14.7 fish 100 m–2 on artificial reefs (Fig. 2-2). Among all samples, lionfish density ranged from 0 to 1.8 fish

100 m–2 on natural reefs and from 0 to 38.5 fish 100 m–2 on artificial reefs. Habitat- specific lionfish density estimates violated parametric assumptions, and no transformation was successful in meeting the assumption of normality. ANOVA is robust to violations of normality (Brown and Forsythe 1974), thus the two-factor model testing the effect of habitat and year on lionfish density was computed with ln-transformed data.

Both habitat (ANOVA, F1,112 = 44.60, p < 0.001) and year (ANOVA, F2,112 = 9.56, p <

0.001) were significant in the model, but their interaction was significant as well

(ANOVA, F2,112 = 8.35, p < 0.001). The significant interaction was due to more rapid population growth at artificial reefs for which lionfish densities were two orders of magnitude higher than on natural reefs in fall 2013 (Fig. 2-2).

Mean ± 95% confidence intervals of lionfish TL estimated with the ROV’s laser scale in video samples increased from 204.7 ± 16.9 mm in fall 2011 to 242.9 ± 7.8 in fall

2013 among all study reefs (Fig. 2-3 A). Data were ln-transformed prior to computing the ANOVA testing if fish size was significantly different among years. The model was significant (ANOVA, F2,187 = 9.11, p < 0.001), and pairwise comparisons were significant between 2011 and 2013 (Tukey’s, p < 0.001) and 2012 and 2013 (Tukey’s, p = 0.007), but not between 2011 and 2012 (Tukey’s, p = 0.438). The non-linear regression relating mass to TL for fish (n = 934) sampled by divers with spears was statistically significant

(p < 0.001) with an adjusted R2 of 0.98 (Fig. 2-3 B). Given that fish mass increased faster than TL, the percent increase in predicted mass (69.8%) of fish measured with the ROV’s laser scale was greater than the percent increase in TL (18.7%) (Fig. 2-3 C).

Predicted lionfish mass was significantly different among years (ANOVA, F2,187 = 3.42, p

= 0.035), but pairwise comparisons revealed predicted mass was only significantly different between years 2011 and 2013 (Tukey’s, p = 0.033) (Fig. 2-3 C).

Lionfish Diet Analysis

There were 934 lionfish sampled by divers with spears from natural and artificial reefs. Among the 8 habitat-season combinations, sample size ranged from 88 fish at natural reefs in spring 2013 to 157 fish at artificial reefs in summer 2013, with a mean sample size of 117 fish among the combinations. Total length ranged from 67 to 377 mm, with distinct modes in size distributions indicating multiple year classes were likely present among samples (Fig. 2-4). Among natural reef samples, 85% (361 of 426) had prey present in their stomachs, as did 81% (409 of 508) of artificial reef samples.

Among all samples, 43% of lionfish prey by mass was unidentifiable. Identifiable prey consisted of 77 taxa, 39% of which were identified to species (Table 2-1). Prey taxa

unique for the northern GOM included two fish families (Family: Pleuronectidae, right eye flounders and Family: Paralichthyidae, large tooth flounders) and notable invertebrate taxa, for example, squid (Loligo sp.), slipper lobster (Family: Scyllaridae) and Florida stone crab (Menippe mercenaria). By mass, the diet of lionfish collected at natural reefs predominantly consisted of reef-associated prey (98.2%), most which

(89.5%) consisted of small (< 5 cm) demersal reef fishes, such as damselfishes, twospotted cardinalfish (Apogon pseudomaculatus), blennies, and wrasses. In contrast, reef-associated prey constituted only 24.4% of lionfish diet at artificial reef sites, which principally consisted of juvenile vermilion snapper (Rhomboplites aurorubens), and bank seabass, (Centropristis ocyurus). Non-reef benthic fishes, such as lizardfishes, flounders, and searobins, constituted the highest percentage (42.6%) of lionfish diet at artificial reefs, but pelagic jacks and scads (16.3%) and benthic invertebrates (12.3%) were also well-represented.

Similar patterns were observed in lionfish diet among prey categories whether

%M, %N, or %IRI was considered (Fig. 2-5). This was due to the fact that there was not a wide range in the sizes of prey items observed. For example, crabs were often similar in size to benthic fishes, and no zooplankton or similar-sized prey were observed in lionfish stomach samples. Habitat type, season, and size class all were significant (p <

0.001) in PERMANOVA models that tested for diet differences by %M or %N (Table 2-

2). However, the interactions between habitat type and season (PERMANOVA, p <

0.001), as well as between season and size class (PERMANOVA, p = 0.020), were significant in the %M model, and the interactions between habitat type and season

(PERMANOVA, p < 0.001) and habitat type and size class (PERMANOVA, p = 0.022) were significant for %N.

When the habitat type x season interaction in the %M model was sliced by season, there were significant differences in diet between artificial and natural reefs in each season (PERMANOVA, p < 0.016). However, when the same interaction was sliced by habitat, there were significant differences in diet for artificial reef samples among all seasons (PERMANOVA, p < 0.001) except for between fall and winter

(PERMANOVA, p = 0.180). The pattern was different for natural reef samples, where diet was only significantly different between winter and the other seasons

(PERMANOVA, p < 0.011). The same results were observed when the habitat type x season interaction in the %N model was sliced by habitat or season, although the p- values were slightly different than for %M. Overall, these patterns reflect a more constant diet of small demersal reef fishes displayed by lionfish sampled at natural reefs, versus a more varied diet at artificial reefs. The significant difference observed for natural reef samples in winter versus other months reflects a higher percentage of shrimps in that season (Fig. 2-5).

The season x size class interaction sliced by season for the %M model revealed significant differences among all size classes in fall and winter (PERMANOVA, p <

0.036) but no differences in spring or summer (PERMANOVA, p > 0.058). When the same interaction was sliced by size class, there were significant differences in diet between all seasons (PERMANOVA, p < 0.022) except fall and winter (PERMANOVA, p

> 0.112) for small and medium sized fish (PERMANOVA, p < 0.022). However, the only differences for large fish occurred between winter and spring (PERMANOVA, p = 0.008)

and fall and winter (PERMANOVA, p = 0.005). The differences observed for small and medium fish among seasons were mostly due to fluctuations in the percentage of their diets constituted by small demersal reef fishes, while the differences observed for large fish were mostly due to an increase in shrimp consumption in winter.

The habitat x size class interaction sliced by habitat for the %N model revealed significant differences among all size class at artificial reefs (PERMANOVA, p < 0.004) but no differences among size classes at natural reefs (PERMANOVA, p > 0.083).

There were also significant differences between reef types for each of the three size classes (PERMANOVA, p < 0.001). Overall, these results reflect the more variable diet observed for lionfish sampled at artificial versus natural reefs.

Muscle Stable Isotope Analysis

Lionfish sampled for muscle isotope analysis ranged in size between 78 and 363 mm TL (mean TL ± 95% CI = 224.7 ± 28.9 mm). Total length of these samples was not significantly different between seasons (ANOVA, F1,28 = 1.305, p = 0.263) or habitats

15 (ANOVA, F1,28 = 0.150, p = 0.701). Correlations between TL and δ N (Pearson’s r =

0.79, p <0.001) and TL and δ13C (Pearson’s r = 0.41, p = 0.006) were both significant.

Therefore, the effect of TL on both δ15N and δ13C was removed by subtracting the slope of the linear relationship between each variable and TL; hereafter, δ15N and δ13C refer to TL-corrected values.

Both habitat type (ANOVA, F1,28 = 6.29, p = 0.018) and season (ANOVA, F1,28 =

16.28, p <0.001) significantly affected lionfish muscle δ15N, but their interaction also was significant (ANOVA, F1,28 = 4.29, p = 0.050). Results of Holm-Sidak tests indicated the habitat effect was significant in spring (p = 0.003) but not winter (p = 0.750), and the season effect was significant for natural reefs (p < 0.001) but not artificial reefs (p =

13 0.172). The effect of sampling season was significant for muscle δ C (ANOVA, F1,28 =

11.80, p = 0.002), but neither the habitat effect (ANOVA, F1,28 = 0.06, p = 0.808) nor the interaction between habitat and season (ANOVA, F1,28 = 4.70, p = 0.499) was significant.

Discussion

Lionfish Density Trends

The time series of invasive lionfish density estimates reported here indicates an exponential increase in their population size since first being observed in the nGOM in summer 2010. In fact, by fall 2013 mean lionfish density at study artificial reef sites was among the highest reported in the western Atlantic (Whitfield et al. 2006, Green and

Côté 2009, Hackerott et al. 2013). For example, Hackerott et al. (2013), reported lionfish densities between 0 and 52 fish 100 m–2 among reefs in Belize, Cuba, and The

Bahamas. However, mean density among all their samples (n = 71) was only 4.4 fish

100 m–2. Mean lionfish density was several fold higher on nGOM artificial reefs reported here, but even more remarkable is the fact that those densities were reached in only 3 y since lionfish were first observed in this region. Furthermore, the rapid growth of individuals indicates that lionfish biomass, hence prey demand, is increasing even more rapidly in the region than their population growth in numbers.

Factors that have facilitated the expansion of invasive lionfish throughout the western Atlantic are well documented, and include the presence of venomous spines, a voracious appetite, fast growth, early maturity, pelagic egg masses, and historically low abundances of native piscivores in much of the invaded range (Morris 2009, Ahrenholz and Morris 2010, Fogg et al. 2013, Chagaris et al. 2017). It is unclear what mechanisms have facilitated the extremely rapid increase in lionfish densities in the nGOM or why

lionfish densities on artificial reefs are two orders of magnitude higher than on natural reefs, although high lionfish densities have been reported on manmade structures in other systems (Smith and Shurin 2010, Jud et al. 2011). The highest lionfish densities reported among Caribbean reefs surveyed by Hackerott et al. (2013) occurred at patch reefs in The Bahamas, and artificial reefs in the nGOM likely mimic attributes of patch reefs versus more expansive types of natural reefs. Artificial reefs deployed in the nGOM tend to be isolated concrete modules, as were the bulk of reefs examined in the current study, or other sunken manmade materials, like shipwrecks or surplussed military vehicles (Dance et al. 2011, Patterson et al. 2014). Similar to coralline patch reefs (Hackerott et al. 2013), nGOM artificial reefs also tend to have a footprint on the scale of 102-103 m2 and vertical relief that is substantially higher (typically 2-3 m) than the surrounding seabed (Patterson et al. 2014). Multiple large (>200 km2) artificial reef permit areas on the shallow (<50 m) nGOM shelf facilitate widespread artificial reef deployment in the region, but reef modules or other types of manmade structure tend to occur on sandy or muddy substrates with adjacent artificial reefs often being >500 m apart (Turpin and Bortone 2002, Strelcheck et al. 2005, Dance et al. 2011). In areas of the shelf lacking natural reef structure, settling larval fishes cue to high vertical relief

(Hornstra and Herrel 2004, Smith and Shurin 2010), thus the patchy distribution of artificial reefs may serve to concentrate settling juvenile lionfish.

Predator-prey dynamics also may affect the distribution of lionfish on the nGOM shelf, although perhaps in unexpected ways. Circumstantial and direct evidence exists that large (>10 kg) piscivores, such as sharks or groupers, may consume adult lionfish in some parts of their invaded range (Maljković et al. 2008, Mumby et al. 2011, Diller et

al. 2014). However, Patterson et al. (2014) reported that large piscivores actually had higher densities on artificial versus natural reefs in the nGOM. Artificial reef communities in the region are dominated (up to 25% by number and 40% by biomass) by red snapper, Lutjanus campechanus (Dance et al. 2011), but no lionfish have been observed in red snapper stomach samples (Tarnecki and Patterson 2015). Groupers

(family: Serranidae) tend to be twice as abundant on natural versus artificial reefs in the system, but their density is an order of magnitude lower than that of snappers (family:

Lutjanidae) (Patterson et al. 2014). Perhaps lionfish population control by native piscivores would be more likely to occur via predation on early life stages versus adult lionfish (Carr and Hixon 1995, Lorenzen 1996), but no direct observation of that has been reported to date.

Lionfish Trophic Ecology in the nGOM

Throughout their invaded range, diet analyses have demonstrated invasive lionfish to be generalist mesopredators with a preference for small (<5 cm) demersal reef fish prey (Albins and Hixon 2008, Muñoz et al. 2011). Among our samples, crustaceans and other invertebrates were more important contributors to the diet of smaller (<200 mm) lionfish, which has been reported from other regions (Morris and

Akins 2009). Piscivory clearly increased as lionfish grew, although the contribution of fish to lionfish diet was lower on artificial versus natural reefs. The ontogenetic shift to greater piscivory observed among diet samples was corroborated via δ15N analysis of muscle tissue, given the positive correlation between δ 15N and TL and the fact that δ15N increases with trophic position due to trophic fractionation (Post 2002). Furthermore, the highest muscle δ 15N values occurred among the largest (>250 mm TL) lionfish sampled

at natural reefs in spring 2013 which corresponded to the highest degree of piscivory observed among diet samples.

Small demersal fishes, such as damselfishes (family: Pomacentridae), blennies

(family: Blenniidae), gobies (family: Gobiidae), and wrasses (family: Labridae) are among the more numerically dominant taxa on nGOM natural reefs but are nearly absent from artificial reef communities (Patterson et al. 2014). Therefore, higher abundances of small demersal reef fishes on natural versus artificial reefs also would seem to favor higher densities of lionfish on natural reefs, but the opposite pattern was observed. Lionfish sampled at nGOM artificial reefs tended to have more varied diets than on natural reefs, with fish foraging on a higher proportion of non-reef and invertebrate prey. Therefore, the lack of small demersal reef fish prey on artificial reefs did not seem to be a limiting factor with respect to lionfish density.

Clearly, the generalist nature of lionfish foraging can extend well beyond simply feeding on a variety of small reef fishes. That was especially true of lionfish sampled at artificial reefs in fall and winter when invertebrate prey constituted >50% of their diet, even for the largest (>250 mm TL) fish. Those trends were supported by muscle δ15N values that were lower in winter when fish were feeding on lower trophic level prey, and also lower for lionfish samples collected at artificial versus natural reefs. Values of muscle δ13C were not significantly different between reef types, but there was a significant season effect in which δ13C was lower in winter than spring samples. Again, this corroborates diet data in that benthic invertebrates are likely to have lower δ13C values due to benthic microalgae being depleted in 13C relative to phytoplankton

(Moncreiff and Sullivan 2001).

Invasive lionfish have the potential to cause substantial ecosystem impacts in the nGOM given their density, feeding ecology, and growth rates (Green et al. 2012,

Ruttenberg et al. 2012, Albins and Hixon 2013). However, another concern for resource managers is their potential impact on exploited species. Few exploited fishes were observed within lionfish stomach samples, but among them were flounders (families:

Paralichthyidae and Pleuronectidae) and vermilion snapper, Rhomboplites aurorubens.

While few reef fish taxa were observed in lionfish stomachs at artificial reef sites, vermilion snapper was present in summer, fall, and winter samples and constituted

10.5% of lionfish diet by mass at artificial reefs. Many of the exploited species common to both natural and artificial reefs in the nGOM initially settle out of the plankton in other habitats and then recruit to reefs later in life (Lindeman et al. 2000). Fishes that settle out of the plankton directly onto reefs, such as vermilion snapper, likely will be much more vulnerable to direct lionfish impacts than species that recruit to reefs as older individuals.

Being ecological generalists, in terms of habitat and/or dietary preferences, is a characteristic shared among the most successful of fish invaders (Moyle and Marchetti

2006). Although lionfish had a broad diet among all habitats, those sampled from artificial reefs fed on a wider variety of prey resources, the majority of which were non- reef associated prey inhabiting nearby sandy substrates. This pattern likely stems from higher lionfish densities on artificial reefs that may have depleted available reef prey

(Albins and Hixon 2008), or increased intra-specific competition (Jones 1987) forcing individuals away from reefs to forage. The likelihood of food limitation for lionfish would be inherently greater at artificial reefs if their diet was restricted to small demersal reef

fishes, given that these fishes are less abundant and diverse on local artificial reefs than on natural reefs (Dance et al. 2011, Patterson et al. 2014). Ultimately, the high densities of lionfish at artificial reef sites, coupled with abundant non-reef associated taxa in their stomachs, demonstrate their ability to forage on open substrates away from reefs.

Movement of lionfish with respect to foraging behavior may vary widely depending on the characteristics of a given site. For example, lionfish have been observed traveling away from coral patch reefs during foraging bouts in the Bahamas (Green et al. 2011), yet in an estuarine system, researchers found that lionfish display high site fidelity (Jud and Layman 2012). Consistent with earlier investigations of lionfish trophic ecology, our results suggest lionfish are ecological generalists in the nGOM and illustrate their adaptability to a range of habitat (Kimball et al. 2004, Jud et al. 2011) and foraging conditions (Morris and Akins 2009, Muñoz et al. 2011).

Conclusions and Implications

Northern GOM lionfish populations likely have not yet reached their peak, as estimates of density, TL, and body mass increased throughout the study period, and lionfish density on artificial reefs is two orders of magnitude higher than on natural reefs.

Juvenile lionfish have been shown to exhibit density-dependent growth on artificial patch reefs in The Bahamas (Benkwitt 2013), but it is unknown whether lionfish densities are sufficiently high relative to food resources to cause a similar negative feedback in the nGOM. Sample sizes from ROV sampling were insufficient to test for differences in lionfish TL between natural and artificial reefs, but there was a significant difference in fish size among years. Total length distributions of lionfish sampled by spear also shifted to larger sizes from spring 2013 to winter 2014, which is consistent with larger fish becoming more predominant in the system. The presence of multiple TL

modes among lionfish sampled with spears also indicates the presence of multiple year classes in each of the sampled habitats and, the increase in the number of individuals

<150 mm TL in winter 2014 may indicate local self-recruitment is occurring. Self- recruitment would imply that invasive lionfish populations now clearly established in the nGOM would not require recruitment from other regions to ensure population persistence or growth. Furthermore, entrainment of eggs and larvae in the Gulf Loop

Current from nGOM spawning events may implicate this region as a source of recruits to downstream regions such as the west Florida shelf or the Florida Keys (Lee et al.

1992, Lee and Williams 1999).

The potential for negative ecological impacts is likely to increase as lionfish populations expand in the nGOM. Although predation on adult lionfish by large piscivores has been inferred or observed in some regions (Maljković et al. 2008, Mumby et al. 2011, Diller et al. 2014), native predator density has not impacted lionfish colonization or population density in the Caribbean region (Hackerott et al. 2013).

Lionfish have enormous potential to negatively affect native communities either by consuming fauna directly or competing with native predators for the same forage base or space on reefs. Native groupers and snappers have habitat preferences similar to those of lionfish, thus examining reef fish behavior and movement on reefs with respect to lionfish presence could reveal indirect effects on these native faunas. The infrequency with which empty lionfish stomachs were encountered in this study implies highly successful feeding, and may indicate the naivety of native prey species to lionfish presence (Côté and Maljković 2010). Reductions in the abundance of reef and non-reef associated small demersal fishes due to lionfish predation may have far reaching

impacts on nGOM reef ecosystems. In such cases that prey resources become depleted, it would be useful for researchers to monitor changes in lionfish foraging behavior and site fidelity, both of which have the potential to impact the effectiveness of lionfish mitigation efforts. In addition, future research should also be focused on tracking changes in lionfish density over time and examining their bioenergetic demands, direct consumption of native reef fishes, and growth rates.

Table 2-1. Prey taxa observed in red lionfish stomachs sampled in the northern Gulf of Mexico. The overall percent diet by mass is given for natural (NR) and artificial reef (AR) samples. Taxon Common name Habitat NR %mass AR %mass Crabs Anasimus latus Silt spider crab 0.00 0.02 Brachyura Crabs 0.07 0.45 Calappidae Box crabs 0.00 0.37 Calappa sulcata Yellow box crab Reef 0.00 0.38 Majidae Spider crabs 0.05 0.07 Menippe mercenaria Florida stone crab Reef 0.14 0.06 Pagurus sp. Hermit crab 0.00 0.00 Parthenopidae Elbow crab 0.00 0.01 Porcellana sigsbeiana Striped porcelain crab Reef 0.01 0.00 Portunidae Swimming crab 0.05 0.22 Portunus sayi Sargassum swimming crab Non-reef 0.09 0.75 Portunus spinicarpus Longspine swimming crab Non-reef 0.70 0.19 Stenorhynchus seticornis Yellowline arrow crab Reef 0.00 0.24 Xanthidae Mud crabs Non-reef 0.08 0.43 Total Crabs 1.19 3.18

Shrimps Decapod shrimp Shrimps 1.39 2.75 Alpheidae Snapping shrimps Reef 0.03 0.04 Caridea Snapping shrimps Non-reef 0.12 0.31 Penaeidae Penaeid shrimps Non-reef 2.28 4.12 Litopenaeus setiferus White shrimp Non-reef 0.27 0.55 Farfantepenaeus duorarum Pink shrimp Non-reef 0.09 0.18 Trachypenaeus similis Roughneck shrimp Non-reef 0.03 0.00 Stenopodidae Cleaner shrimps Reef 0.00 0.02 Total Shrimps 4.21 7.97

Other Benthic Invertebrates Benthic Gastropod Sea snails 0.00 0.16 Eumunida sp. Squat lobster Non-reef 0.51 0.14 Scyllaridae Slipper lobsters Reef 0.00 0.13 Hippoidea Mole crabs Non-reef 0.00 0.11 Axiidae Thalassinidean shrimp 0.02 0.11 Decopoda 0.09 0.10 Cumacea Hooded shrimps Non-reef 0.03 0.00 Squillidae Mantis shrimps Non-reef 0.21 0.39 Squilla empusa Mantis shrimp Non-reef 0.37 0.00 Octopoda Octopus Reef 0.01 0.00 Mollusca 0.01 0.00 Pleocyemata 0.07 0.00 Total Other Benthic Invertebrates 1.31 1.14

Pelagic Invertebrates Achelata phyllosoma Larval lobster Non-reef 0.00 0.01 Larval shrimp 0.00 0.01 Euphausiacea Euphausiid Non-reef 0.00 0.01 Gammaridae Amphipod 0.00 0.02 Loligo sp. Squid Non-reef 0.00 4.51

Table 2-1. Continued Taxon Common name Habitat NR %mass AR %mass Lophogastrida Pelagic shrimp Non-reef 0.00 0.07 Mysida Mysid shrimps Non-reef 0.08 0.02 Total Pelagic Invertebrates Non-reef 0.08 4.67

Reef Fishes Apogon sp. Cardinalfishes Reef 0.01 0.00 Apogon pseudomaculatus Twospotted cardinalfish Reef 10.35 0.28 Blennidae Blennies Reef 7.35 0.17 Gobiidae Gobies Reef 4.08 0.55 Labridae Wrasses Reef 1.19 0.06 Halichoeres bathyphilus Greenband wrasse Reef 11.11 0.00 Halichoeres bivattutus Slippery dick Reef 1.82 1.68 Haemulon aurolineatum Tomtate Reef 0.48 0.00 Monacanthus sp. Filefish Reef 0.09 0.00 Pomacentridae Damselfish Reef 14.47 0.24 Chromis enchrysurus Yellowtail reeffish Reef 0.77 0.00 Chromis scotti Purple chromis Reef 14.52 0.00 Stegastes fuscus Dusky damselfish Reef 9.10 0.00 Rhomboplites aurorubens Vermilion snapper Reef 0.78 10.49 Scorpaenidae Scorpionfishes Reef 2.38 0.95 Serranidae Sea basses Reef 2.82 0.00 Baldwinella vivanus Red barbier Reef 0.01 0.00 Centropristis ocyurus Bank seabass Reef 7.86 9.41 Centropristis sp. Seabass Reef 0.00 0.14 Serranus subligarus Belted sandfish Reef 0.27 0.00 Total Reef Fishes 89.48 23.96

Non-Reef Benthic Fishes Diplectrum formosum Sand perch Non-reef 0.00 6.19 Diplectrum sp. Sand perch Non-reef 0.00 3.60 Paralichthys albigutta Gulf flounder Non-reef 0.00 0.14 Paralichthyidae Large-tooth flouders Non-reef 0.57 2.78 Pleuronectidae Righteye flounders Non-reef 0.00 0.05 Pleuronectiformes Flatfishes Non-reef 0.00 0.02 Prionotus sp. Searobin Non-reef 0.00 0.08 Synodontidae Lizardfishes Non-reef 1.61 22.22 Synodus synodus Diamond lizardfish Non-reef 0.00 1.28 Triglidae Searobins Non-reef 0.30 0.44 Bellator brachychir Shortfin searobin Non-reef 0.00 0.27 Xyrichthys novacula Pearly razorfish Non-reef 0.00 5.76 Total Non-Reef Benthic Fishes

Pelagic Fishes Carangidae Jacks Non-reef 1.27 1.59 Decapterus macarellus Mackerel scad Non-reef 0.00 3.43 Decapterus sp. Scad Non-reef 0.00 3.62 Trachurus lathami Rough scad Non-reef 0.00 7.64 Total Pelagic Fishes 1.27 16.28

Table 2-2. PERMANOVA table of factors affecting lionfish diet by prey mass. Results from 3-factor permutational analysis of variance model testing the effect of habitat type (natural versus artificial reefs), season, or fish size class (<200, 200-250, or >250 mm total length) on lionfish diet by percent mass. Abbreviations: df = degrees of freedom, SS = type III sums of squares, MS = mean square error, Hab = habitat type, Seas = Season, Size = size class. Source Df SS MS pseudo-F p-value Habitat 1 101,040 101,040 38.14 <0.001 Season 3 66,913 22,304 8.42 <0.001 Size 2 35,809 17,905 6.76 <0.001 Hab x Seas 3 33,589 11,196 4.27 <0.001 Hab x Size 2 9.595 4,798 1.81 0.075 Seas x Size 6 22,807 4,635 1.75 0.02 Hab x Seas x Size 6 23,314 3,886 1.47 0.073 Residual 435 115,220 2,649

Figure 2-1. Location of study sites for lionfish density and diet sampling. A) Map of the Gulf of Mexico with study region indicated. B) Natural (circles) and artificial (triangles) reef sites sampled with a remotely operated vehicle in fall 2010- 2013 to estimate invasive lionfish density; water bodies: MB = Mobile Bay, PrB = Perdido Bay, PB = Pensacola Bay, and CB = Choctawhatchee Bay. C) Natural (circles) and artificial (triangles) reef sites sampled by divers who speared lionfish for stomach content and muscle stable isotope analysis.

Figure 2-2. Mean densities of lionfish 100m-2 on natural and artificial reefs. Mean (95% CI) density of lionfish in fall 2010-2013 estimated with micro remotely operated vehicle-based video sampling at northern Gulf of Mexico natural (A, B) and artificial reef sites (C, D). Reef images were taken via ROV camera where red arrows indicate lionfish. No video sampling occurred at artificial reef sites in fall 2010.

Figure 2-3. Length to weight relationship of speared lionfish predicts lionfish mass from ROV length observations. A) Mean (95% CI) total length of red lionfish (n = 190) observed in remotely operated vehicle video (ROV) samples at northern Gulf of Mexico reef sites and measured in video images with a red laser scale attached to the ROV; F = fall, 11 = 2011, 12 = 2012, and 13 = 2013. B) Non- linear regression computed to predict red lionfish mass from total length from fish (n = 934) captured by spearfishing. C) Mean (95% CI) predicted mass of lionfish (n = 190) observed in ROV video samples and measured with a red laser scale.

Figure 2-4. Habitat-specific total length distributions of lionfish samples by season. Season-specific total length distributions of lionfish sampled by spear at northern Gulf of Mexico natural and artificial reef sites from April 2013 through March 2014.

Figure 2-5. Indices of lionfish diet by habitat, season and size class. Stacked bar plots of A) mean percent diet by number of prey items, B) mean percent diet by prey mass, and C) mean percent index of relative importance for seven prey categories (PI = pelagic invertebrates, BI = other benthic invertebrates, Cr =

crabs, Sh = shrimps, PF = pelagic fishes, nrF = non-reef fishes, and RF = reef fishes) observed in lionfish stomach samples. Lionfish were sampled with spears in the northern Gulf of Mexico during spring 2013 through winter 14. Size categories: S = <200 mm total length (TL), M = 200-250 mm TL, and L = >250 mm TL. Habitat types: natural = natural reefs, artificial = artificial reefs.

Figure 2-6. Bi-plot of lionfish muscle δ15N versus δ13C values. Plot of total length- corrected δ15N and δ13C values from lionfish white muscle samples collected at northern Gulf of Mexico natural (NR) and artificial (AR) reefs in spring 2013 and winter 2014. Mean values of δ15N and δ13C are depicted with 95% CI by the four combinations of season and habitat.

CHAPTER 3 DNA BARCODING SIGNIFICANTLY IMPROVES RESOLUTION OF INVASIVE LIONFISH DIET IN THE NORTHERN GULF OF MEXICO

Indo-Pacific red lionfish, Pterois volitans (Linnaeus, 1758) (hereafter lionfish) are piscivorous scorpionfish that have exhibited an unprecedented invasion in the western

Atlantic Ocean over the past 30 years (Whitfield et al. 2002, Schofield 2009). While invasions by marine predators are atypical, lionfish have exhibited an invasion so extensive and rapid in this region that they are considered to be the most successful marine fish invader to date (Morris and Akins 2009, Côté et al. 2013a). Lionfish (Family:

Scorpaenidae) were first introduced into the waters off southeastern Florida in the late

1980s (Schofield 2009), then spread throughout the US South Atlantic Bight (SAB) in the 1990s, and the Caribbean Sea in the 2000s (Whitfield et al. 2006, Schofield 2010).

The Gulf of Mexico (GOM) is the most recently invaded basin, where lionfish were first reported in 2009 off the northern Yucatan Peninsula, Mexico (Aguilar-Perera and Tuz-

Sulub 2010), in the Florida Keys, and along the west Florida shelf (Schofield 2010). By late 2010, lionfish had been observed in eastern, northern and western regions of the

GOM (Schofield 2010, Fogg et al. 2013, Dahl and Patterson 2014, Nuttall 2014). To date, lionfish have established populations in diverse habitats covering over 7 million km2 across the US southeast Atlantic coast, Caribbean Sea, and portions of the GOM

(Schofield 2010, Côté et al. 2013a, Schofield et al. 2014). Throughout their invaded range, lionfish pose a threat to fisheries resources, native fish communities, reef ecosystems, and human health (Morris and Akins 2009, Morris and Whitfield 2009).

Reprinted by permission from Kristen Dahl: Springer Nature. Biological Invasions. Dahl KA, Patterson WF III, Robertson A, Ortmann AC. DNA barcoding significantly improves resolution of invasive lionfish diet in the northern Gulf of Mexico. 2017.

Invasive lionfish have been described as generalist mesopredators that consume a broad range of fish and invertebrate prey (Morris and Akins 2009, Muñoz et al. 2011,

Dahl and Patterson 2014), including the juvenile stages of ecologically and economically important species (Lesser and Slattery 2011). Their distinction as highly efficient predators has been attributed to unique predatory behaviors and prey naïveté

(Fishelson 1997, Hornstra and Herrel 2004, Albins and Lyons 2012, Cure et al. 2012).

These factors, along with fast growth, early maturity, high fecundity, and a lack of native predators have resulted in higher lionfish densities and larger body sizes than are observed in their native range (Darling et al. 2011, Hackerott et al. 2013, Dahl and

Patterson 2014).

Invasions by predators are expected to have the most damaging impacts on native ecosystems given that predator-prey dynamics can greatly influence community assemblages in both terrestrial and marine systems (Paine 1966, Hixon and Carr 1997,

Grosholz et al. 2000, McDonald et al. 2001, Caut et al. 2008). Thus, it is essential to understand how invasive predators and native prey interact in a trophic context to effect change in the structure and function of invaded marine communities (Rilov 2009). Visual diet (i.e., gut content) analysis is conventionally used to examine such trophic relationships (Hyslop 1980), but digestion often results in substantial (>40%) portions of the stomach contents being unidentifiable (Morris and Akins 2009, Dahl and Patterson

2014). Unidentifiable portions of diet represent lost information given that visually identifiable portions may not fully characterize diet. Thus, traditional visual diet analyses may provide incomplete data on trophic relationships between invasive and native species. Furthermore, unidentifiable diet portions are often excluded when describing

taxonomic contributions to diet, which can heavily bias results if differential digestion rates cause some species or life stages to become visually unidentifiable more quickly than others (Hyslop 1980, Collis et al. 2002). More complete diet information could also be useful in understanding the transfer and bioaccumulation of natural toxins, such as those associated with ciguatera fish poisoning, in ciguatera-endemic regions of the

Caribbean (Robertson et al. 2013).

In the case of invasive lionfish, as much as 70% of prey items observed in their stomachs cannot be visually identified to species with a dissecting microscope (Valdez-

Moreno et al. 2012, Côté et al. 2013b, Dahl and Patterson 2014). However, molecular tools can be employed to identify otherwise unidentifiable prey items. For example, mitochondrial DNA (mtDNA) barcoding offers an approach to genetically identify prey based on a target fragment (~650 base pair region) of the cytochrome c oxidase subunit

I (COI) gene (Hebert et al. 2003, Ivanova et al. 2007, Valdez-Moreno et al. 2012, Côté et al. 2013b). COI sequences are species-specific, highly conserved, and have high reliability and resolution for species identification in fish, even in partially digested or archived samples (Carreon-Martinez et al. 2011). Valdez-Moreno et al. (2012) employed barcoding to examine invasive lionfish diet in the Mexican Caribbean, while

Côté et al. (2013b) employed barcoding to examine lionfish diet diversity (i.e., richness) in The Bahamas. In both of those studies, barcoding greatly increased the power of the diet analysis, thus the ability of researchers to evaluate the trophic ecology and impacts of lionfish in those systems.

I previously reported results from visual analysis of lionfish (n = 934) stomach contents sampled in the northern GOM (nGOM), a region with lionfish densities among

the highest in the western Atlantic (Dahl and Patterson 2014, Dahl et al. 2016). Despite the relatively large sample size of that study, the potential exists that diet estimates were incomplete or biased given 43% by mass of observed prey could not be identified visually due to digestion (Dahl and Patterson 2014). In the current study, I applied DNA barcoding to identify previously unidentifiable fish prey items reported by Dahl and

Patterson (2014). Specific objectives of this work were to 1) evaluate the effectiveness of DNA barcoding to identify visually unidentifiable lionfish prey, and 2) determine whether lionfish diet estimates are enhanced when DNA-barcoded prey items are included in diet analysis.

Materials and Methods

Study Location and Specimen Collection

Lionfish sampling was performed seasonally from April 2013 through March 2014 on nGOM natural and artificial reefs ranging in depth from 24 to 35 m (Fig. 3-1).

Individuals were captured by divers via spearing immediately posterior to the spinal column and were placed in a saltwater ice slurry upon surfacing. Speared lionfish were ranked by size and randomly sampled to retain every nth fish so that approximately 100 lionfish per habitat type per season were sampled for diet analyses. Each lionfish was weighed (nearest 0.1 g) and measured (nearest mm total length, TL). All applicable institutional and/or national guidelines for the care and use of animals were followed during the course of this study.

Visual Gut Content Analysis

Red lionfish (Pterois volitans) stomachs were extracted from fish after inspection of esophagus and gills for regurgitated prey. Stomach contents of each sampled lionfish were fixed in 100% (200 proof) ethanol in plastic bags. Preserved prey items were

visually identified to the lowest taxonomic level possible, counted, and dried for at least

48 h at 70° C to obtain dry mass (Hoese and Moore 1998, Kells and Carpenter 2011).

Visual identification was performed either with the naked eye or under an Olympus

SZX12 dissecting microscope. Identifiable prey taxa were grouped into seven categories: shrimps, crabs, other benthic invertebrates, pelagic invertebrates, reef fishes, non-reef benthic fishes, and pelagic fishes, while unidentifiable prey items were grouped into fish and invertebrates. To assess dietary contribution of each visually identified prey category, percent mass (%M) was computed by dividing the dry mass of each prey taxa by the total dry mass of all prey from each individual, including visually unidentified prey. Diet was assessed by habitat type (natural versus artificial reefs) and lionfish size class (small: <200mm TL, medium: 200–250mm TL, large: >250mm TL).

DNA Barcoding Preparation and Analysis

Any prey item that was identified as a fish but could not be identified to a taxonomic level lower than family was considered unidentifiable and retained for DNA barcoding analysis. These prey items were categorized into two digestion stages, whole fish (i.e., most of carcass was present) or fish remains, and weighed wet to the nearest

0.01 g. Samples were processed for barcoding by first removing any external layer of tissue that had been in contact with the lionfish sample’s stomach wall and/or fluids with sterile scalpels and forceps, and then a small (approximately 1 mm3, 15–25 mg) plug of muscle tissue was excised from each unidentifiable prey item. Tissue samples were blotted dry and placed in a 1.5 ml sterile microcentrifuge tube. To prevent cross- contamination between tissue extractions, tools were rinsed with 70% ethanol and then flame sterilized. Tissue plugs were frozen at -80°C until DNA extraction.

DNA extractions were performed following the optimized standardized protocols of Handy et al. (2011) with DNeasy Blood & Tissue Kits (Qiagen, CA). Tissue was placed in a solution of lysis buffer and proteinase K and homogenized with sterile disposable polypropylene pestles (Thermo Fisher Scientific, MA), and then digested for

1–2 h at 56°C until lysed. DNA extractions were quantified on a Nanodrop 1000 spectrophotometer (Thermo Fisher Scientific, MA) for DNA concentration and purity

(i.e., ratio of absorbance at 260 nm and 280 nm) and then visualized on 1% agarose gels to verify these estimates. Samples with high DNA yield (>100 ng/ul) were diluted

10X with ultrapure water.

Amplification of the 650 bp barcode region of the COI gene was performed with universal M13 tailed fish primer cocktails C_FishF1t1and C_FishR1t1 developed by

Ivanova et al. (2007) for barcoding (Fig. 3-2). The 20 µl polymerase chain reaction

(PCR) mixes included 10 µl of 10% trehalose, 4.4 µl ultrapure water, 2 µl 10X PCR buffer, 1 µl 50 mM MgCl2, 0.2 µl each 10 µM primer cocktail, and 0.1 µl each of 10 mM dNTPs and Platinum® Taq polymerase (Thermo Fisher Scientific, MA) and 1.6 µl DNA template. DNA template concentration ranged from approximately 1–40 ng/µl. Thermal conditions for COI PCR amplification included initial denaturation at 95°C for 2 min, followed by 35 cycles of 94°C for 40 s, 53.5°C for 40 s, and 72°C for 1 min, with a final extension at 72°C for 5 min and a hold at 4°C. Positive and negative controls were used for each 96-well test-plate of amplification reactions to test for PCR reaction quality and contamination in reagents. Template DNA was replaced with ultrapure water in negative controls, and template DNA extracted from Pterois volitans muscle tissue was utilized for positive controls.

PCR products were visualized on 1% agarose gels and positive reactions were identified by a clear band aligned to the 650 bp fragment of the TrackIt™ 1 Kb Plus

DNA ladder (Thermo Fisher Scientific, MA). All positive PCR products, along with positive and negative controls, were sent to Beckman Coulter Genomics (Danvers, MA) where they underwent PCR cleanup and bidirectional sequencing using M13F and

M13R primers (Fig. 3-2).

Barcoding Sequence Analysis

The COI barcode sequences of unidentified prey items were analyzed to assess length and quality using Geneious (version 8.1.6; Kearse et al. 2012). Forward and reverse sequences were trimmed to remove ambiguous and/or low-quality bases and remnant primer from amplification or sequencing reactions and then aligned using

ClustalW to produce consensus sequences. Sequences were then visualized and edited in Geneious to resolve less reliable base calls. Final contigs were ranked by length as long (>500 bp), medium (300–500 bp) and short (<300 bp). Long and medium ranked sequences were submitted to nucleotide BLAST searches through the NCBI

GenBank database to classify unidentified prey item sequences to the closest match in the COI database. BLAST searches returned the closest ten sequences in the reference databases based on values of sequence similarity (i.e., % identity), E-value (i.e., number of expected hits by chance) and grade (i.e., percentage combining % identity,

E-value and query coverage). The species-level identification function of the Barcode of

Life Database was used to resolve any potential database gaps in GenBank (BOLD, www.boldsystems.org, Ratnasingham and Hebert 2007). Collection and sequence data are available from GenBank (http://www.ncbi.nlm.nih.gov).

I adopted three criteria to establish accurate fish identification from BLAST results. A species match was identified for long sequences with >99% grade. A genus level match was identified for long sequences with a 95–99% grade. Finally, a family level match was identified by either long sequences with <95% grade or medium sequences with >95% grade. Sequences with genus and family level identifications from

BLAST were also submitted into the BOLD ID function which assigned species level identification based on less than 1% sequence divergence and genus level identification based on less than 3% divergence with a reference sequence. Classification to genus and/or family level was necessary in some cases due to incomplete taxonomic coverage of GOM fishes in the reference databases. Sequences of less than 300 bp were not included in further analyses. These criteria for sequence inclusion were considered to be conservative given recent studies have utilized COI sequences as short as 100 bp if they had a high (90%) probability of being accurately assigned to species level (Meusnier et al. 2008, Valentini et al. 2009).

To improve sequence quality or verify the accuracy of the extraction and barcoding process, replicates were performed by extracting new tissue plugs and repeating the entire DNA barcoding process. This was performed for all samples that had poor sequence quality from the original sequencing and also had sufficient tissue mass remaining. Replicates were also randomly chosen among a subset of samples that had high initial quality, as well as among those that returned species of interest

(e.g., new or exploited species) as the closest COI match.

Incorporating Barcode Information into Diet

To incorporate new species identified by barcoding into existing %M estimates of lionfish diet, wet masses of unidentified prey items were converted to simulated original

dry masses using the digestion stage (i.e., whole fish or fish remains) and a wet-dry mass conversion factor from visually identified fish prey. Prey identified by barcoding were then sorted into the original prey categories for further analysis. The proportion of visually unidentifiable fish prey that was not able to be resolved through barcoding remained in the unidentifiable fish category.

Sample-based prey taxa accumulation curves were computed in Primer (v6;

Clarke and Gorley 2006) to assess diet richness in nGOM lionfish as a function of sample size, and whether estimates of prey diversity changed when DNA barcoded prey were added. The order in which samples were successively incorporated was randomly resampled by permutation (999 permutations) to satisfy the assumption of random sample order. Given the number of individuals (n = 934) was equal between methods, the shapes of the curves (i.e., rate of species accumulation, asymptote reached) and 95% confidence intervals (i.e., overlapping or not) could be compared between methods to assess whether DNA barcoding added significant information to diet estimates (Colwell et al. 2004). Curves were computed for lowest taxonomic level of unique prey taxa as well as higher order identification (i.e., fish family and invertebrate infraorder) to examine trends in diet diversity.

Results

Lionfish (n = 934) examined in this study ranged in size from 67 to 377 mm TL and were collected from natural and artificial reefs over the course of one year (mean n

= 117 fish per treatment combination). With traditional visual diet identification, 1,485

(64.8% by number, 57% by mass) prey items were identified in lionfish stomach samples, while 807 (35.2% by number, 43% by mass) were not visually identifiable.

Unidentified fish prey items (n = 696) were processed for barcoding as described above, but unidentified invertebrate prey items (n = 73 items) were not barcoded.

DNA extracts from the 696 visually unidentified fish prey items yielded 575

(82.6%) readable sequences that could be identified to species, genus or family levels.

Mean length (± SD) of barcode sequences identified to species was 627 ± 26 bp (Fig. 3-

3). Mean length of barcode sequences identified to genus and family level were 622 ±

24 bp and 455 ± 115 bp, respectively. Among readable COI sequences, 444 (77.2%) were positively identified to species level (63.8% of total visually unidentified fish prey items). The remaining 131 sequences were identified to either genus (n = 73) or family level (n = 58) based our classification criteria. With the addition of barcoding the number of identifiable fish prey nearly doubled (932 versus 499 prey items), resulting in 78.0% of total fish prey being identified. Whole fish prey had a slightly higher probability (85%) of being identified using barcoding than more highly digested fish remains (79%). As many as five unique species were identified by DNA barcoding from a single lionfish stomach. For all test plates, positive controls produced a COI barcode with a correct species match. Negative controls did not produce any COI barcodes, indicating satisfactory quality control across all test plates.

Dahl and Patterson (2014) previously reported 11 species, 3 genera, 16 families and 8 orders or infraorders among the visually-identified invertebrate prey, along with 18 fish prey species from 15 families (Table 3-1). Prey diversity increased greatly when

DNA barcoding was incorporated into prey identification, including 24 species, 19 genera, and 9 families not previously identified with visual methods (Table 3-1). Of these, 30 are novel prey taxa unreported from other lionfish diet studies (Table 3-1).

Notable species found in lionfish diet for the first time via DNA barcoding were red snapper (Lutjanus campechanus), wenchman (Pristipomoides aquilonaris), roughtongue bass (Pronotogrammus martinicensis) and red porgy (Pagrus pagrus), which are all ecologically or economically important reef fishes in the region. Potential cannibalism also was revealed by DNA barcoding, with red lionfish being the most frequently identified prey taxon (n=100 of 696 or 14.4%) via barcoding. Among red lionfish sequences in the GenBank database, our sequences matched voucher specimens from the western and southern Caribbean, and Brazil (Accession Numbers:

KJ739816, KM488633, and KP641132, respectively) most closely (>99.7% similarity).

Prey items barcoded as red lionfish occurred in lionfish of all size classes, but was found to be most prevalent in medium to large fish size categories.

A subset of prey samples (n = 150) was selected for duplicate DNA barcoding of new dissections of prey tissue. Samples were chosen to represent a range of species and tissue quality levels, and where further species verification was deemed essential

(e.g., red lionfish, exploited species) (Fig. 3-4). Duplication of barcoding for the 39 initially short sequences (< 300bp) and 72 initially poor sequences (<100 bp) resulted in

67 sequences >300bp in length that were sufficient for inclusion in further analyses (Fig.

3-5). Out of 39 duplicates run from initially short sequences, 30 resulted in similar or longer length sequences than initial runs (Fig. 3-4). For these samples, all but one returned identical matches from reference databases, and this sample was excluded from further analyses (Figs. 3-4, 3-5). Species identity was confirmed for duplicate analysis of the prey samples that originally were barcoded red snapper, as was true for

15 other species of interest (Figs. 3-4, 3-5).

Species identification was confirmed for the 16 duplicates of samples that originally were barcoded as Pterois volitans (Figs. 3-4, 3-5). It should be noted that while cannibalism was not documented visually among the samples collected in this study, Dahl and Patterson (2014) visually identified 28 partially digested individual prey as family: Scorpaenidae. Furthermore, subsequent stomach content analysis of lionfish sampled in the nGOM has produced clearly identifiable juvenile P. volitans in the stomachs of larger conspecifics; however, such observations have been infrequent (Fig.

3-6).

Differences in diet among size classes, habitats and seasons were apparent from visually identified prey, with piscivory increasing with fish size. Diet breadth was greatest on artificial reefs due in part to consumption of non-reef associated fishes, and invertebrates contributing increasingly to diet in winter (Fig. 3-7 A). Dahl and Patterson

(2014) originally characterized nGOM lionfish diet based on visually identified prey only, thus excluding unidentified prey. When visually unidentified prey are included in estimates of diet percentages by prey category, changes in mean %M were apparent, especially with respect to contributions of invertebrate prey relative to fish prey (Fig. 3-7

B). Percent mass contribution of invertebrates to lionfish diet considering only visually- identified prey was 6.8% on natural reefs and 17.0% on artificial reefs. With the addition of barcoded fish prey these contributions declined to 4.5% and 8.0%, respectively (Fig.

3-7 C). Prey categorized as unidentified fishes (Fig. 3-7 B) was reduced from >40% of diet by mass to ~7% after DNA barcoding (Fig. 3-7 C, Table 3-1). The importance of fishes to the diet of lionfish was more pronounced in lionfish collected from natural reef habitat; however, the trend of increased foraging on invertebrates in winter was

observed in both habitats (Fig. 3-7 C, Table 3-1). Prey identified as reef-associated fish was greatest in spring and summer seasons on artificial reefs, due to large numbers of vermilion snapper (~16% diet by mass) identified via barcoding (Fig. 3-7 C, Table 3-1).

An ontogenetic shift toward piscivory with increasing lionfish size remained apparent with the addition of barcoded prey, where juvenile lionfish consume more invertebrates by mass across all seasons (Fig. 3-7 C).

Prey exhibited a wide range in estimated dry mass, from 0.01 to 11.20 g, with a mean (± SD) of 0.72 ± 1.1 g (Fig. 3-8). Mean (± SD) mass (0.71 ± 1.46 g) of fish prey that remained unidentified following barcoding was similar to the mean of the overall distribution. Masses of prey identified to family were larger on average (mean ± SD =

1.0 ± 1.59 g) than of either species- or genus-level (mean ± SD = 0.68 ± 0.84 g and

0.64 ± 0.57 g, respectively) identification. High prevalence of relatively large jawfishes

(family: Opistognathidae) and seabasses (family: Serranidae) that could be identified only to family with barcoding explains this trend. The size of fish prey that were barcoded as lionfish (i.e., potential cannibalism) was much smaller than the majority of the samples that underwent barcoding. Half of the prey samples identified by barcoding as P. volitans had a mass <0.25 g and >90% were under 1 g. Vermilion snapper were similarly small in size, with nearly 80% of samples identified via barcoding being under 1 g in mass.

The number of unique prey taxa contributing to visually-identified lionfish diet accumulated rapidly with increasing numbers of lionfish sampled (Fig. 3-9 A). At a sample size of 100 individuals, already thirty prey taxa were observed, which continued to increase through all 934 individuals, although the curve begins to reach an asymptote

after approximately 800 samples. When taxa identified with DNA barcoding were included, that curve did not reach an asymptote (Fig. 3-9 A). At 100 individuals sampled with DNA barcoding, the number of prey taxa increased to forty-seven unique taxa (Fig.

3-9 A). Overall, this represents an addition of 36 unique prey taxa via DNA barcoding.

Prey richness is significantly different with the addition of barcoded fish prey given that

95% CIs do not overlap the curve for visually-identified prey alone. When curves were fit only to fish family and invertebrate infraorder, prey richness was significantly higher when DNA barcoding results were included (Fig. 3-9 B).

Discussion

DNA barcoding greatly enhanced the ability to describe lionfish diet, and together with visual identification this study reveals the richest diversity of prey reported to date for this invasive species. At the family level alone, these diet data nearly double the highest estimates of prey diversity reported for lionfish (Morris and Akins 2009, Valdez-

Moreno et al. 2012, Côté et al. 2013b). Including visually identified prey, I recorded a total of 41 fish species, 37 genera, and 24 families among lionfish prey. DNA barcoding revealed lionfish diet in the nGOM to be far broader than previously reported from traditional visual identification results where I was able to identify 24 species, 19 genera and nine families of fishes that were previously undetected by visual gut content analysis (Dahl and Patterson 2014).

Our estimates of prey diversity remain high even when compared to other studies in which DNA barcoding was utilized to describe lionfish diet (Valdez-Moreno et al.

2012, Côté et al. 2013b). Valdez-Moreno et al. (2012) reported fish prey from 14 families, 22 genera and 34 species, while Côté et al. (2013b) reported fish prey from 16 families, 27 genera and 37 species. While species richness of prey was comparable

among all studies, the diversity of prey constituting lionfish diet at the level of family and genus were far greater in the current study. Our study was conducted outside of the

Caribbean, thus prey communities lionfish encountered were inherently different than where lionfish diet was previously explored with DNA barcoding. Côté et al. (2013b) estimated the occurrence of 90 potential prey fish species that were small enough (<13 cm) for lionfish to ingest in the region of The Bahamas in which they conducted their barcoding diet study. In our study region, Patterson et al. (2014) reported 91 reef fishes present on nGOM natural and artificial reefs, but <70 of these species had some life stage present on reefs which would meet Côté et al.'s (2013c) 13-cm threshold.

Therefore, the higher diversity of lionfish prey observed in the current study occurred despite lower diversity of potential reef fish prey.

Potential Cannibalism

DNA barcoding not only increased our ability to characterize lionfish diet, but also revealed potential cannibalism in the nGOM, a relatively recently invaded region. While it is certain that lionfish mitochondrial DNA was associated with tissue samples isolated from these prey samples in lionfish stomachs, unequivocal evidence of lionfish cannibalism does not exist just because some prey samples were barcoded as lionfish.

There are three possible explanations for lionfish identifications via DNA barcoding: 1) prey tissue may have been contaminated with DNA from the consumer lionfish during initial dissections, 2) prey DNA was degraded due to digestion to the point that the only viable DNA present and amplified was that from the consumer lionfish, or, 3) lionfish

DNA was present in prey tissue because the prey was a cannibalized lionfish.

A rigorous sample processing protocol was put in place to guard against contaminating prey tissue samples with DNA from consumer lionfish, but contamination

cannot be completely discounted as the source of lionfish DNA in barcoded samples.

Great effort was expended to ensure sterile techniques, perform positive and negative barcoding controls, and extract tissue from prey items after removing surface layers of tissue. Despite these techniques, it remains a possibility that DNA from the stomach lumen of consumers could have been shed onto prey fragments and not removed prior to PCR amplification. However, the mass of potentially contaminating consumer lionfish stomach tissue, cells, or DNA would have been orders of magnitude lower than the mass, hence DNA, of prey tissue sampled for barcoding. Therefore, it seems unlikely that contamination alone could have resulted in the degree to which lionfish DNA was present in barcoded prey samples, especially given the techniques utilized in this study.

Amplification of a species’ DNA via PCR is not only a function of the abundance of that species’ DNA versus other DNA present in a given sample, but also the quality of the DNA present (Gonzalez et al. 2012). While the mass of any consumer lionfish DNA present in prey tissue dissected for DNA amplification was likely orders of magnitude lower than the mass of prey DNA, it is possible that highly digested prey items had DNA of such poor quality that most of the DNA capable of being amplified was that of the consumer lionfish. Clearly, the likelihood of this phenomenon occurring would increase with more advanced states of digestion. In future barcoding studies, qualitative scoring of the degree of prey digestion would be one way to test whether the incidence of apparent cannibalism increases with increasing stage of prey digestion.

The final possibility for why lionfish mitochondrial DNA was so pervasive among barcoded prey samples is that at least some of those samples were in fact cannibalized lionfish. The likelihood of that being the case might be somewhat diminished by the lack

of direct observation of cannibalized lionfish for samples examined here. However, I have observed evidence of lionfish cannibalism in subsequent sampling of lionfish stomach contents (see Fig. 3-6), and several small (<5 cm) prey items were reported as

Family: Scorpaenidae when lionfish prey items were originally identified visually (Dahl and Patterson 2014). One potential reason for the lack of visual evidence for cannibalism in the current study could be the size of lionfish being consumed by larger conspecifics. Prey samples that were barcoded as lionfish were smaller than the majority of prey samples identified with barcoding. This would suggest that any cannibalism that did occur resulted primarily from larger adults cannibalizing small, recently settled juveniles, which is similar to the pattern reported by Valdez-Moreno et al. (2012). Conversely, these results could also be explained by tissue of small individuals of other species being digested so quickly that consumer lionfish DNA amplified rather than degraded DNA of non-lionfish prey.

Very little is known about lionfish larval and early juvenile stages, particularly the size at which lionfish settle onto reefs following their planktonic larval stage (Ahrenholz and Morris 2010). The late post-flexion larval stage for the species is reported to be between 9.5 and 11 mm standard length (SL) (Imamura and Yabe 1996), and a newly settled (i.e., not fully pigmented) juvenile collected in the nGOM was 15 mm SL (Byron et al. 2014). This is similar to the most often encountered size range for prey samples that barcoded as P. volitans in this study. The next largest component of diet revealed by DNA barcoding was vermilion snapper, and these prey items were also most likely newly settled recruits given their relatively small size. Early juvenile prey fishes lose most or all identifiable characters rapidly (<60 min) after ingestion by other fishes

(Schooley et al. 2008, Legler et al. 2010). Therefore, predation on these early life stages is very difficult to detect visually and may be missed completely without molecular techniques.

While highly useful in identifying species, DNA barcoding alone cannot conclusively detect cannibalism within a species. Since predator specimens were not subjected to DNA barcoding, it was not possible to directly compare lionfish prey sequence haplotypes to those of their consumer. However, even if this had been done, the COI haplotype diversity in nGOM lionfish populations is likely too low to detect differences between predator and prey (Hamner et al. 2007, Betancur-R. et al. 2011,

Toledo-Hernández et al. 2014). As stated above, one of the possible explanations for lionfish mitochondrial DNA associated with prey tissue samples is contamination by

DNA from consumers, which then amplified during PCR. Molecular blocking can be used to prevent amplification of consumer (or otherwise erroneous) DNA (De Barba et al. 2014), but was not used here.

Prey samples from 89 adult lionfish barcoded as P. volitans, which occurred across small, medium, and large size classes from both natural and artificial reefs. The frequency of this potential cannibalism increased from spring to winter and occurred most frequently in larger adult lionfish. The highest incidence of P. volitans prey barcodes were observed in winter when lionfish were at their highest densities and their largest mean size (Dahl and Patterson 2014). If cannibalism was in fact occurring, these results would suggest size-structured and density-dependent cannibalism in nGOM lionfish. Intraspecific competition for prey resources has most likely increased with the rapid population growth observed in invasive lionfish in the nGOM, with all size classes

competing for the same pool of resources and space on reefs. On nGOM reefs, fish diversity declines during winter (Dance et al. 2011), which corresponds to a higher proportion of invertebrates in lionfish diets (Dahl and Patterson 2014). Therefore, the higher incidence of potential cannibalism during winter may have reflected lower availability of preferred prey (Rudolf 2008). However, any inferences drawn about potential cannibalism at this stage cannot be substantiated given uncertainties about the likelihood of contamination of prey samples with consumer lionfish DNA. Genotyping of consumer lionfish and prey samples that barcoded as lionfish could provide an unequivocal test of whether cannibalism is occurring in the nGOM, and could also be employed to quantify the extent to which it occurs (Primmer et al. 2000, Guichoux et al.

2011).

Trends in Diet with Unidentified Fish Resolved

Including prey taxa that were identified via DNA barcoding among lionfish stomach samples clearly expands our knowledge of invasive lionfish diet in the nGOM.

However, re-examining visually-identified prey taxa previously reported highlighted an issue with how the data were originally presented. In Dahl and Patterson (2014), percent diet composition by mass was only reported for taxa that had been visually identified. However, unidentified fish prey constituted a larger percentage of the diet than did unidentified invertebrate prey (by mass, 43% to <1%). Therefore, by only considering visually identifiable prey when computing diet composition, the importance of fish, in general, to lionfish diet was understated. While lionfish clearly are generalist mesopredators, they are more piscivorous in the nGOM than was previously reported

(see Fig. 3-7).

With the inclusion of DNA barcoding results, general conclusions reported earlier about lionfish diet still hold, but some differences are apparent. Again, the overall breadth of lionfish diet in the nGOM is substantially higher than previously reported.

Reef-associated fish prey contributed >10% more (by mass) to lionfish diet on artificial reefs, which is mostly attributable to juvenile snappers (e.g., vermilion snapper, red snapper, wenchman) and potential cannibalism on conspecific juveniles. Reef- associated fish prey on natural reefs declined nearly 10% with the addition of non reef- associated fishes such as flounders (family: Bothidae), jawfishes (family:

Opistognathidae), and lizardfishes (family: Synodontidae). Contribution of pelagic fishes to the diet remained similar for lionfish captured on natural reefs, but declined by nearly

5% on artificial reefs. These differences taken together do not affect the overall trends that lionfish on natural reefs predominantly consume reef-associated fishes, and that lionfish on artificial reefs have broader diets with higher contributions from non reef- associated and pelagic fishes.

Several exploited fish species were identified in invasive lionfish diet via DNA barcoding, which may be of particular interest to resource managers. Vermilion snapper

(Rhomboplites aurorubens) was the second most commonly barcoded prey species with 98 prey items from 61 adult lionfish, and constituted nearly 16% of lionfish diet by mass on artificial reefs across the study. There were 24 lionfish that had multiple juvenile vermilion snapper identified with barcoding, and in 7 cases more than three vermilion snapper juveniles were found in the stomach of an individual lionfish. Red snapper and red porgy are other notable fishery species found in lionfish diet for the first time via DNA barcoding. Predation on juvenile red snapper by lionfish was not expected

given red snapper typically do not recruit to reef habitat until their second year of life when they are >250 mm TL (Workman et al. 2002, Patterson et al. 2005). However, precocious recruitment to adult habitat is sometimes observed (Bailey et al. 2001).

DNA Barcoding Efficacy

DNA barcoding was highly effective in identifying lionfish fish prey that were not identifiable visually. Our optimized DNA barcoding methodology yielded >80% efficiency for producing reads with high enough quality for prey taxon identification. Combining

DNA barcoding with traditional visual identification methods resulted in identification of

>90% of all lionfish prey by mass. Furthermore, the addition of fish prey identified with barcoding resulted in approximately 250% more fish species identified in diets relative to visual diet analysis. Digestion level may have contributed to DNA barcoding success, where lightly digested whole prey accounted for approximately 60% of molecular identifications and more highly digested prey only accounted for approximately 40%.

Validation studies on the reliability of COI as a taxonomic tool are numerous across a diversity of taxa (Hebert et al. 2003, Clare et al. 2007, Dawnay et al. 2007) including fishes (Ward et al. 2005) and reproducibility of COI barcodes based on novel DNA extractions and amplification from prey samples was high in this study.

Higher-level taxonomic identifications (i.e., genus and family level) from barcoding were necessitated in part due to poor sequence quality but also incomplete taxonomic coverage for COI in reference databases. For example, I visually identified purple reeffish (Chromis scotti) in the stomachs of lionfish (Dahl and Patterson 2014), but it was not possible to obtain a species level match with DNA barcoding in the

GenBank or BOLD databases. Sequences from 24 prey specimens most closely matched Chromis scotti, however, a verified barcode does not currently exist for the

species given that fewer than three voucher sequences have been submitted to BOLD

(Ratnasingham and Hebert 2007). Thus, these individuals were classified as Chromis sp. It is highly likely that these prey were actually C. scotti given high sequence quality and complete taxonomic coverage of remaining Chromis spp. that occur in the nGOM.

Similar issues with reference database gaps occurred for genera within the families

Apogonidae, Gobiidae, Labridae, Opistognathidae, Serranidae, and Triglidae. For example, the presence of a cryptic species of cardinalfish (Astrapogon sp.) in the nGOM is inferred by the lack of a species match from taxonomically complete databases for this genus, which was also reported by Valdez-Moreno et al. (2012). Ultimately, this study highlights that gaps exist in COI databases with respect to nGOM reef fishes.

Global barcoding initiatives (e.g. FISH-BOL, CBOL) that aim to obtain COI records of all fishes may result in more nGOM fishes being barcoded in the near future (Ward et al.

2009).

DNA barcoding is often used to study diet in fishes with sample size limitations as a result of rare occurrence or logistical challenges in sampling. However, this method can be applied more broadly if enough information is gained by its use. Our study demonstrates that even for a diet study with a relatively large sample size, for which visual diet analysis might be expected to have detected most prey taxa, DNA barcoding added a significant amount of new dietary information. This was clearly evident from prey accumulation curves comparing lionfish prey richness between visual identification and barcoding methodologies. Using visual identification, 30 prey taxa were observed at a sample size of 100, which is comparable to the richness that Côté et al. (2013b) observed at a similar sample size using DNA barcoding in The Bahamas. In contrast, I

found 47 unique taxa at a sample size of 100 individuals combining visual identification with DNA barcoding. Despite the high proportion of prey identified with DNA barcoding, an asymptote in prey richness was not reached in this study. The slope of the curve for visually identified prey noticeably flattened after sampling 400 individuals, but no obvious asymptote was reached. This is a notable finding given the large sample size

(nearly 1,000 fish) and prior studies indicating diet saturation at ~700 individuals (Morris and Akins 2009).

Lionfish prey diversity appears to be higher in the nGOM region than in other regions, which was not evident from visual diet analysis alone. The broad diet of invasive lionfish diet in the nGOM supports the inference that lionfish are generalists that will opportunistically prey on small demersal or benthic species or early life stages, suites of which differ between regions of the invaded range (Morris and Akins 2009,

Muñoz et al. 2011, Green et al. 2011, Green and Côté 2014). The diversity of prey identified within lionfish stomachs in the nGOM indicates they are a substantial threat to numerous taxa in this region. As lionfish population size, mean body size, and biomass continue to increase on both natural and artificial reefs these interactions could only be expected to intensify (Dahl et al. 2016).

Our ability to determine with high resolution exactly what lionfish are eating on nGOM reefs will allow for more accurate assessment of direct lionfish impacts on native reef fish communities of the nGOM and shed light on potential indirect impacts to other species. Of particular importance is the detection of several regionally exploited reef fish species as lionfish prey, which has important implications for fishery managers. The diversity of non-reef and reef fishes alike in lionfish stomach samples may portend

population declines due to increasing lionfish predation. Furthermore, lionfish predation also may indirectly impact reef fishes that also depend on those species for food.

Lionfish feeding ecology data at regional scales are important inputs for ecosystem modeling efforts (e.g., Chagaris et al. 2017). Such efforts are critical to understanding current and predicted future direct and indirect impacts of lionfish in the nGOM ecosystem, as well as for simulating potential management strategies to mitigate these impacts.

Table 3-1. Fish prey taxa observed in red lionfish (Pterois volitans) stomachs sampled in the northern Gulf of Mexico. The overall percent diet by mass is given for natural (NR) and artificial reef (AR) samples. Asterisks (*) indicate novel prey taxa identified by DNA barcoding in this study. See Dahl and Patterson (2014) for invertebrate prey taxa. Visually Visual + Visual + Identified Unidentified DNA Barcoding NR AR NR AR NR AR Family Taxon Common name %mass %mass %mass %mass %mass %mass Reef Fishes Antennariidae* Fowlerichthys radiosus* singlespot frogfish 0 0 0 0 0 0.13 Apogonidae Apogon affinis* bigtooth cardinalfish 0 0 0 0 0.55 0.03 Apogon aurolineatus* bridle cardinalfish 0 0 0 0 1.94 0.04 Apogon maculatus flamefish 0 0 0 0 1.81 0.32 twospotted Apogon pseudomaculatus cardinalfish 10.35 0.28 8.46 0.22 6.18 0.12 Apogon sp. cardinalfish 0 0 0 0 0.83 0.63 Astrapogon sp. cardinalfish 0 0 0 0 0.56 0 Phaeoptyx pigmentaria dusky cardinalfish 0 0 0 0 0.06 0 Blenniidae Parablennius marmoreus* seaweed blenny 0 0 0 0 0.59 0 Blenniidae blennies 7.35 0.17 6.01 0.13 4.5 0.07 Chlorophthalmidae* Chlorophthalmidae greeneyes 0 0 0 0 0.01 0 Gobiidae Coryphopterus sp. goby 0 0 0 0 0.93 0 Microgobius carri* seminole goby 0 0 0 0 0 0.10 Gobiidae gobies 4.08 0.55 3.33 0.42 2.70 0.23 Haemulidae Haemulon aurolineatum tomtate 0.48 0 0.40 0 0.70 0.37 Labridae Halichoeres bathyphilus greenband wrasse 11.11 0 9.09 0 6.63 0

Halichoeres bivattatus slippery dick 1.82 1.68 1.49 1.30 1.09 0.71 Halichoeres sp. wrasse 0 0 0 0 0.95 0 Labridae wrasses 1.19 0.06 0.97 0.05 3.43 0.02 Lutjanidae Lutjanus campechanus* red snapper 0 0 0 0 0.02 0.10 Pristipomoides aquilonaris* wenchman 0 0 0 0 0.07 3.32

Table 3-1. Continued Visually Visual + Visual + Identified Unidentified DNA Barcoding NR AR NR AR NR AR Family Taxon Common name %mass %mass %mass %mass %mass %mass Rhomboplites aurorubens vermilion snapper 0.78 10.49 0.64 8.12 0.84 15.94 Lutjanidae snappers 0 0 0 0 0 0.07 Microdesmidae* Ptereleotris calliura* blue dartfish 0 0 0 0 0.11 0 Moncanthidae Monacanthus sp. filefish 0.09 0 0.07 0 0.05 0 Pomacentridae Chromis enchrysurus yellowtail reeffish 0.77 0 0.63 0 0.49 0.08 Chromis scotti purple chromis 14.52 0 11.87 0 8.67 0 Chromis sp. damselfish 0 0 0 0 2.01 0 Pomacentridae damselfishes 14.47 0.24 11.83 0.19 8.98 0.10 Stegastes fuscus dusky damselfish 9.10 0 7.44 0 5.43 0 Stegastes variabilis cocoa damselfish 0 0 0 0 0.06 0.03 Scorpaenidae Pterois volitans red lionfish 0 0 0 0 5.36 4.38 Scorpaena brasiliensis* barbfish 0 0 0 0 0.19 0 Scorpaenidae scorpionfishes 2.38 0.95 1.95 0.74 1.94 1.09 Serranidae Centropristis ocyurus bank seabass 7.86 9.41 6.42 7.28 8.86 6.25 Centropristis sp. seabass 0 0.14 0 0.1 0 0.06 Pronotogrammus martinicensis* roughtongue bass 0 0 0 0 0 0.04 Serraniculus sp.* pygmy sea bass 0 0 0 0 0.75 0.07 Serranus subligarius belted sandfish 0.27 0 0.22 0 0.16 0 Serranidae groupers 2.82 0 2.30 0 3.02 0.26 Sparidae* Pagrus pagrus* red porgy 0 0 0 0 0 0.01 Total Reef-Fishes 89.46 23.96 73.12 18.55 80.45 34.59

Non-Reef Fishes Bothidae* Bothus robinsi* twospot flounder 0 0 0 0 0 0.07 Bothidae left-eye flounders 0 0 0 0 0 0.02

Table 3-1. Continued Visually Visual + Visual + Identified Unidentified DNA Barcoding NR AR NR AR NR AR Family Taxon Common name %mass %mass %mass %mass %mass %mass Dactyloscopidae* Dactyloscopidae sand stargazers 0 0 0 0 0 0.17 Labridae Xyrichtys novacula pearly razorfish 0 5.76 0 4.45 0 4.24 Opistognathidae* Opistognathus robinsi* spotfin jawfish 0 0 0 0 0.05 0 Opistognathidae jawfishes 0 0 0 0 1.37 3.52 Paralichthyidae Cyclopsetta fimbriata* spotfin flounder 0 0 0 0 0 0.07 Paralichthys albigutta gulf flounder 0 0.14 0 0.10 0 0.06 Syacium papillosum* dusky flounder 0 0 0 0 0.30 0.26 Syacium sp. large-tooth flounder 0 0 0 0 0.45 0 Paralichthyidae large-tooth flounders 0.57 2.78 0.47 2.15 0.49 1.18 Pleuronectidae Pleuronectidae righteye flounders 0 0.05 0 0.04 0 0.02 Serranidae Diplectrum formosum sand perch 0 6.19 0 4.79 0.77 5.24 Diplectrum sp. sand perch 0 3.60 0 2.79 0 1.72 Synodontidae Saurida brasiliensis* Brazilian lizardfish 0 0 0 0 0 0.05

Synodus macrostigmus* largespot lizardfish 0 0 0 0 0.08 0 Synodus poeyi* offshore lizardfish 0 0 0 0 0.25 9.53 Synodus synodus diamond lizardfish 0 1.28 0 0.99 0 0.61 Synodontidae lizardfishes 1.61 22.22 1.32 17.19 2.24 9.43 Triglidae Bellator brachychir shortfin searobin 0 0.27 0 0.21 0 0.11 Bellator militaris* horned searobin 0 0 0 0 0 0.02 Prionotus sp. searobins 0 0.08 0 0.07 0 0.04 Triglidae searobins 0.30 0.44 0.24 0.34 0.25 1.15 Uranoscopidae* Uranoscopidae stargazers 0 0 0 0 0 0.29 Total Non-Reef Fishes 2.48 42.81 2.03 33.12 6.24 37.79

Table 3-1. Continued Visually Visual + Visual + Identified Unidentified DNA Barcoding NR AR NR AR NR AR Family Taxon Common name %mass %mass %mass %mass %mass %mass Pelagic Fishes Carangidae Decapterus punctatus mackerel scad 0 3.43 0 2.66 0.34 4.63 Decapterus sp. scad 0 3.62 0 2.80 0 1.54 Trachurus lathami rough scad 0 7.64 0 5.91 0 4.91 Carangidae jacks 1.27 1.59 1.04 1.23 0.76 0.68 Dussumieriidae* Etrumeus teres* round herring 0 0 0 0 0 0.17 Total Pelagic Fishes 1.27 16.28 1.04 12.6 1.10 11.92

Benthic invertebrates 1.31 1.14 1.07 0.88 0.78 0.48 Pelagic invertebrates 0.08 4.65 0.07 3.60 0.05 1.97 Shrimps 4.21 7.97 3.44 6.16 2.51 3.38 Crabs 1.19 3.19 0.98 2.46 0.71 1.35 Unidentified invertebrates - - 0.20 0.30 0.47 0.80 Unidentified fishes - - 18.05 22.34 7.68 7.72 Total Prey Mass 100 100 100.0 100.0 100.0 100.0

Figure 3-1. Natural (circles) and artificial (triangles) reefs where red lionfish, P. volitans, were sampled for diet analyses. Samples were acquired with divers during spring 2013 through winter 2014. Water bodies are Mobile Bay (MB), Perdido Bay (PrB), Pensacola Bay (PB) and Choctawhatchee Bay (CB).

Figure 3-2. PCR and sequencing primers used in the DNA barcoding of unidentified fish prey from invasive red lionfish (Pterois volitans) in the northern Gulf of Mexico.

Figure 3-3. Size distribution of COI barcoding sequence read lengths in number of nucleotide base pairs for unidentified red lionfish (Pterois volitans) fish prey items, n=696.

Figure 3-4. Red lionfish (Pterois volitans) prey samples with initial sequence length >300 bp selected for duplicate DNA barcoding analysis. Initial and duplicate BLAST results from GenBank. Sequences with base pair (bp) lengths <100 were not submitted to BLAST.

Figure 3-4. Continued

Figure 3-5. Red lionfish (Pterois volitans) prey samples with initial sequence length <300 bp selected for duplicate DNA barcoding analysis. Initial and duplicate BLAST results from GenBank. Sequences with base pair (bp) lengths <100 were not submitted to BLAST. Samples sorted by bp length.

Figure 3-5. Continued

Figure 3-6. Photos courtesy of and reproduced with permission from author, Kristen Dahl. Examples of prey fish visually identified as P. volitans from lionfish stomach contents. A) A red lionfish prey item (30 mm in length) visually identified from the stomach contents of a red lionfish measuring 127 mm TL, collected from an artificial reef south of Pensacola, FL, and B) a red lionfish prey item (37 mm in length) visually identified from the stomach contents of a red lionfish measuring 218 mm TL, collected from an artificial reef south of Mexico Beach, FL

Figure 3-7. Stacked bar plots of mean percent red lionfish (P. volitans) diet by mass for prey categories observed across habitat types, seasons, and size class for: A) visually identified prey items, excluding unidentifiable prey, B) visually identified prey items, including unidentifiable prey, and C) prey items identified visually and with DNA barcoding. Size categories: S = <200 mm total length (TL), M = 200-250 mm TL, and L = >250 mm TL and habitat types: natural = natural reefs, artificial = artificial reefs

Figure 3-8. Mass distribution of 696 prey samples of red lionfish (P. volitans) identified via DNA barcoding to species, genus, family level, or to red lionfish or vermilion snapper

Figure 3-9. Species accumulation curves of red lionfish (P. volitans) prey taxa identified during this study. A) Cumulative number of unique prey identified to species or lowest taxonomic level of identification, and B) fish families and invertebrate infraorders, as a function of number of lionfish sampled and analyzed with visual gut content analysis (solid line) or visual identification plus DNA barcoding (dashed line). Every 5th 95% confidence interval is plotted to allow comparison of curves between visual identification and DNA barcoding methods

100

CHAPTER 4 GENOTYPING CONFIRMS SIGNIFICANT CANNIBALISM IN NORTHERN GULF OF MEXICO INVASIVE RED LIONFISH, PTEROIS VOLITANS

A range of genetic techniques have been developed to identify organisms when visual identification is problematic (Symondson 2002, Hebert et al. 2003). One of the more widely used of these molecular approaches is DNA barcoding, a technique based upon a highly conserved 650 base pair region of the mitochondrially-encoded cytochrome c oxidase subunit I (COI) gene (Hebert et al. 2005, Ivanova et al.

2007)(Hebert et al. 2005, Ivanova et al. 2007). This molecular marker is species-specific and can be used to identify species with great accuracy if voucher sequences exist in globally available databases (Frézal and Leblois 2008, Ward et al. 2009). Given its utility, DNA barcoding is employed by ecologists, as well as taxonomists and forensic scientists, to investigate biodiversity, food safety, illegal wildlife trade, and predator-prey interactions (Teletchea et al. 2008, Valentini et al. 2009). This last application is particularly useful in marine fish ecology given that prey items are often too degraded from digestion to identify visually (Sheppard and Harwood 2005, Ward et al. 2005, Dahl et al. 2017).

Confounding factors can arise when applying DNA barcoding to identify prey items due to the sensitivity of PCR amplification, the use of universal primers, and the relative lack of COI diversity within species. When a prey sample is identified as the same species as the consumer (i.e., self-DNA with respect to the consumer), it is not possible with COI sequences alone to distinguish this result as a true indication of

101

cannibalism versus a false positive (Sheppard and Harwood 2005, O’Rorke et al. 2012,

Jo et al. 2014). Across a range of marine consumer taxa, authors of diet studies applying DNA barcoding frequently report the amplification of consumer species DNA among prey items (e.g., Sheppard and Harwood 2005, Jo et al. 2014). Some authors have discarded results that indicate prey are the same species as the consumer given potential issues with contamination, ignoring potential cannibalism (Bartley et al. 2015,

Moran et al. 2015), while others have reported all detections of consumer DNA as the existence of cannibalism, potentially overestimating the true rate (Valdez-Moreno et al.

2012, Braid et al. 2012, Côté et al. 2013b, Arroyave and Stiassny 2014).

In a recent study, results from DNA barcoding of visually unidentifiable invasive red lionfish, Pterois volitans (hereafter lionfish), fish prey (n = 696) from the northern

Gulf of Mexico (nGOM) indicated thirty-four prey species including potential instances of cannibalism (Dahl et al. 2017, Chapter 3). Lionfish have exhibited an extensive invasion across the tropical and subtropical western Atlantic since the late 1980s and first entered the Gulf of Mexico in 2009 (Schofield 2010). As highly effective, generalist predators that consume a wide variety of fishes and invertebrates (Albins and Hixon

2008, Morris and Akins 2009, Muñoz et al. 2011, Dahl and Patterson 2014), lionfish are capable of directly altering community and trophic structure of native reef fishes across a variety of western Atlantic ecosystems (Lesser and Slattery 2011, Albins 2015, Dahl et al. 2016), potentially causing extirpations (Ingeman 2016). The success with which lionfish have invaded the introduced Atlantic range suggest native communities exert

Reprinted by permission from Kristen Dahl: Springer Nature. Biological Invasions. Dahl KA, Portnoy DS, Hogan JD, Johnson J, Gold JR, Patterson WF III. Genotyping confirms significant cannibalism in northern Gulf of Mexico invasive red lionfish, Pterois volitans. 2018.

102

little biotic resistance to invasion, resulting in reduced interspecific competition (Albins

2013), few constraints on growth (Darling et al. 2011), and few effective, novel parasites

(Sikkel et al. 2014). Furthermore, predation by native predators does not appear to be regulating lionfish populations (Hackerott et al. 2013), which have reached higher densities and body sizes than are observed in their native Indo-Pacific (Darling et al.

2011, Kulbicki et al. 2012, Dahl and Patterson 2014, Pusack et al. 2016). Concurrently, lionfish have experienced ecological release from natural population control mechanisms (e.g., predators, diseases, parasites) within their native range (Albins and

Hixon 2013, Sikkel et al. 2014, Tuttle et al. 2017). While recently invaded, the nGOM region has some of the higher lionfish densities in its invasive, western Atlantic range

(Dahl et al. 2016), and characterizing diet composition and potential cannibalism are important for understanding the impacts of lionfish on local reef fish communities as well as factors that may substantially limit lionfish populations.

Though 100 lionfish were detected as prey via DNA barcoding, indicating potential cannibalism on juveniles in the northern GOM, lionfish consumers were not barcoded by Dahl et al. (2017) to compare with the prey, and so, contamination of degraded prey tissue with consumer DNA could not be ruled out. Further, given low haplotype diversity documented in western Atlantic lionfish populations, it is unlikely there would be detectable differences in COI sequences between predators and prey, especially in the nGOM (Betancur-R. et al. 2011, Toledo-Hernández et al. 2014,

Johnson et al. 2016). Cannibalism in invasive lionfish populations has seldom been reported from visual inspection of gut contents (Valdez-Moreno et al. 2012, Villaseñor-

Derbez and Herrera-Pérez 2014, Dahl et al. 2017), and if confirmed, DNA barcoding

103

might suggest that cannibalism in this species has been underestimated in the invaded range.

Therefore, to reexamine potential lionfish cannibalism, consumers and prey were genotyped with previously reported nuclear DNA microsatellites. Microsatellites are short sequence repeats that exhibit high levels of allele diversity and when assayed across multiple loci can provide a unique genotype profile for each individual examined, providing a high resolution way to distinguish between cannibalism and contamination

(Chistiakov et al. 2006). While microsatellite genotyping is widely used in fisheries and aquaculture to address questions about the relatedness of individuals, genetic diversity, and population association, it has also been applied in studies of predator-prey interactions (Kvitrud et al. 2005, Sundqvist et al. 2008), including an examination of filial cannibalism (DeWoody et al. 2001). Here, lionfish microsatellites were employed to 1) test whether lionfish DNA in prey samples is unique from that of consumer lionfish, and

2) determine the degree of cannibalism among previously barcoded samples. Results of this study have implications for interpreting self-DNA detections from DNA barcoding analysis of diet, as well as for management of invasive lionfish.

Methods and Materials

Sample Collection

Lionfish were sampled for diet analyses (Dahl and Patterson 2014, Dahl et al. 2017) by scuba divers from April 2013 through March 2014 on nGOM natural and artificial reefs at depths of 24 to 35 m. Individuals were captured by spearing immediately posterior to the spinal column and then placed in a saltwater ice slurry upon surfacing. Each lionfish was weighed to the nearest 0.1 g and measured to the nearest mm total length (TL). Lionfish samples were categorized into small: <200 mm

104

TL, medium: 200-250 mm TL, and large: >250 mm TL size classes (e.g., Fig. 4-1).

White muscle tissue (~5 g) was dissected from each lionfish at the time of capture and frozen at -80 °C until DNA extraction. Stomachs and all prey contents were removed from each sample lionfish and fixed in 100% (200 proof) molecular grade ethanol in plastic bags (Dahl and Patterson 2014) All applicable institutional and/or national guidelines for the care and use of animals were followed during the course of this study.

DNA Barcoding of Unidentified Lionfish Prey

Previously, DNA barcoding was performed for lionfish prey items that were identified as fish but could not be identified to a taxonomic level lower than family (n =

696) (Dahl et al. 2017). Samples were processed by first removing any external layer of tissue that had been in contact with the lionfish’s stomach wall or fluids with sterile scalpels and forceps, and then muscle tissue was excised from each unidentifiable prey item. To prevent cross-contamination between tissue extractions, tools were rinsed with

70% ethanol and flame sterilized. For detailed information about DNA barcoding protocols, see Dahl et al. (2017).

Potential cannibalism was revealed by DNA barcoding in 100 consumers, where

Pterois volitans (i.e., self-DNA) was the most frequently identified prey taxon (14.4%) among unidentified fish prey examined. These prey sequences matched voucher specimens of red lionfish from the western and southern Caribbean, and Brazil

(Accession Numbers: KJ739816, KM488633, and KP641132, respectively) most closely

(≥99.7% pairwise similarity). From the samples in which potential cannibalism was detected, 80 had sufficient DNA material remaining to undergo microsatellite genotyping.

105

Microsatellite Analysis

Original DNA extractions of prey tissue (n = 80) identified as lionfish via DNA barcoding (Dahl et al. 2017) were stored frozen (-20 °C) and then secondarily subjected to microsatellite genotyping. Genomic DNA of lionfish consumers was extracted from 15 to 25 mg of muscle tissue with DNeasy blood & tissue kits (Qiagen, CA). All DNA extractions were diluted 10x with ultrapure water prior to PCR.

All fish were genotyped at four nuclear microsatellite loci (Table 4-1). The four microsatellite loci (PVM12, PVM14, PVM31, and PVM42) were chosen from a previously published primer note (Schultz et al. 2013) after testing for consistent amplification. All loci were amplified using primers developed by Schultz et al. (2013), with PCR conditions being modified to obtain strong amplification. Microsatellite genotyping was conducted via PCR amplification in 15 μl reactions containing up to 2 ng of DNA template, 1x Colorless GoTaq Flexi PCR Buffer, 3 mM MgCl2, 0.2 mM dNTPs each, 0.75 U GoTaq Flexi DNA polymerase, 0.12 mM forward labeled primer, and 0.3 mM reverse primer (Table 4-1). The forward primer from each primer pair was labelled with a fluorescent label of either VIC®, FAMTM, or NEDTM dye (G5 dye set,

Applied Biosystems). All PCR reactions were performed under the same cycling conditions consisting of initial denaturation at 94 ºC for 4 min, followed by 35 cycles of

94 ºC for 15 seconds, annealing at 62 ºC for 15 seconds, 72 ºC for 30 seconds, followed by 10 cycles of 94 ºC for 15 seconds, 53 ºC for 15 seconds, and 72 ºC for 30 seconds, and a final extension 72 ºC for 5 minutes.

PCR amplicons were analyzed at the TAMU–CC Core Genomics Lab, electrophoresed on an ABI 3730XL automated capillary sequencer (Applied

Biosystems) with the GeneScanTM 600 LIZTM Size Standard (Applied Biosystems) in

106

each lane. Size fragments were scored using GeneMarker® software and visual verification of all allele sizes was made to ensure correct calls. For a subset of pilot samples, PCR amplicons were electrophoresed on 6% polyacrylamide gels with an ABI

Prism 377 sequencer (Applied Biosystems) and the GeneScanTM 400HD ROXTM Size

Standard (Applied Biosystems) in each lane following the methods of Renshaw et al.

(2013). Size fragments were scored manually, using Genescan v. 3.1.2 (Applied

Biosystems) and Genotyper v. 2.5 (Perkin Elmer). In all cases, consumer DNA and that of its prey were analyzed on the same machine with the same size standard and the allele calls were made with the same software to ensure comparability of allele calls.

Any reactions that failed were repeated up to three times.

To assess the genetic diversity of microsatellite markers employed for this study, the number of alleles, unbiased gene diversity, and effective number of alleles were estimated for each microsatellite locus from consumers with GenoDive v. 2.0 (Nei 1987,

Meirmans and Van Tienderen 2004). Unbiased gene diversity (Hs), a corrected expected heterozygosity measure, is simply the probability that two sampled alleles will be different within a population (Nei 1987). The effective number of alleles (Ne) is a measure of the number of alleles in the sampled population weighted by their frequencies, thus accounts for alleles that are more common than others (Kimura and

Crow 1964).

Visual inspection of matching and non-matching alleles at individual loci allowed for direct microsatellite genotype comparisons between prey and consumer individuals.

Prey template DNA had the potential to be of low quality, especially if extensive digestion had occurred within the consumer lionfish’s stomach prior to sampling.

107

Therefore, sources of genotyping error (Taberlet 1996, Hoffman and Amos 2005) were considered when developing protocols for determining whether microsatellite data supported cannibalism. For example, allelic dropout, or the failure of one allele of a heterozygous individual to be amplified via PCR, can lead to incorrect genotyping of that individual as a homozygote (Gagneux et al. 1997, Soulsbury et al. 2007), and null alleles, or alleles that do not amplify by PCR, can lead to blank or incorrectly identified genotypes (Shaw et al. 1999, Van Oosterhout et al. 2004). Another source of genotyping error stems from PCR artifacts (i.e., stutters) in which amplification products are generated that can be misinterpreted as true alleles (Taberlet 1996, Goossens et al.

1998, Bradley and Vigilant 2002). Thus, micro-checker v. 2.2.3 (Van Oosterhout et al.

2004) was used to screen the data for the presence of null alleles, stuttering, and scoring errors by comparing expected (He) versus observed (Ho) homozygotes for all loci.

Cannibalism was considered to have occurred when at least two different alleles were observed between a lionfish and its prey across the four loci tested. When no allele differences were observed and all loci were amplified, the pair was scored as indicative of no cannibalism. Lastly, two types of results were deemed inconclusive with respect to cannibalism or the lack thereof. The first occurred when no allele differences were observed between consumer and prey but one or more loci failed to amplify. A single-allele difference between consumer and prey was also deemed inconclusive, regardless of the number of loci amplified successfully. The proportion of cannibalism observed from genotyping was then calculated using two approaches to obtain minimum and maximum estimates of cannibalism. I calculated the proportion of

108

cannibalism occurring across the total number of samples, regardless of amplification success, as well as among only the sample pairs with conclusive results from genotyping.

Results

PCR amplification was successful in 76- 93% of samples depending on the locus. The observed number of alleles in consumers sampled from the nGOM was lower for all loci except PVM42 compared to those results previously reported by

Schultz et al. (2013), and ranged from 4 to 9 alleles (Table 4-1). The effective number of alleles was lower than those observed from consumers sampled in the nGOM and ranged from 2.06 to 5.34 alleles. Gene diversity ranged from 0.52 to 0.82 across all loci

(Table 4-1). Results from micro-checker showed evidence of null alleles (i.e., general excess of homozygotes) at PVM14, but not for other loci (Table 4-2). There was no evidence of stutter peaks or scoring errors at any locus (Table 4-2).

Conclusive genotypes were obtained for 50 of the 80 consumer-prey pairs in which DNA barcoding indicated the prey may be a lionfish (Dahl et al. 2017; Fig. 4-1).

Twenty-one of the paired samples had two or more different alleles between lionfish and prey DNA across the assessed microsatellite loci (Fig. 4-1), thus cannibalism was confirmed in 26.3% (21/80) of the total consumers and in 42% (21/50) of consumers for which the data were considered conclusive. Among those samples, there were as many as seven different alleles observed between lionfish consumers and cannibalized prey

(Fig. 4-1). Cannibalism was found to occur in lionfish collected from both natural (n =

10) and artificial (n = 11) reef habitats, and across all size classes of lionfish, but was more frequently detected in medium (n = 10) and large size classes (n = 9) (Fig. 4-1).

Cannibals ranged in length from 138 to 316 mm TL (Fig. 4-1). Cannibalism was

109

documented during all four seasons, but more instances occurred in fall (n = 5) and winter (n = 10). When placed in the broader context of all 934 lionfish sampled for the visual diet study by Dahl and Patterson (2014), cannibalism was confirmed in 2.7% of samples that had prey in their stomachs, and in 2.2% of fish overall. The 100 lionfish prey detected via DNA barcoding reported by Dahl et al. (2017) corresponded to 4.87% of all lionfish diet by percent mass (%M); thus, confirmed cannibalism constituted as much as 2.01% diet by mass. Notably, for consumers in which cannibalism was detected via barcoding, cannibalized prey constituted a significant proportion of the diet by mass (mean %M=71.1%).

Beyond the 50 consumer-prey pairs that were successfully genotyped, results from the remaining pairs of samples (n = 30) were deemed inconclusive. This was because some loci did not amplify in either the prey or consumer and exhibited zero or single-allele differences (n = 23), or because there was only a single-allele difference between the consumer and the prey from complete genotypes (n = 7). Notably, in two such cases of single-allele differences between consumer and prey, the difference was seen at PVM42, a locus with only four possible alleles, two of which are rare (Table 4-1;

Fig. 4-1), which may be indicative of cannibalism but was below our conservative threshold.

Discussion

Microsatellite genotyping of lionfish consumers and their prey that had been previously identified as lionfish via DNA barcoding provides definitive evidence of cannibalism in the nGOM. While the invasion is relatively recent in this region, lionfish densities have increased exponentially and those reported from nGOM artificial reefs are among the highest in the western Atlantic (Hackerott et al. 2013, Dahl and Patterson

110

2014, Dahl et al. 2016). Lionfish are opportunistic feeders, consuming a wide diversity of native, reef-dwelling organisms in this region (Dahl and Patterson 2014, Dahl et al.

2017). Their increasingly high densities appear to be forcing them to switch to other prey besides reef fishes, such as non-reef associated fishes, pelagic fishes, and invertebrates (Dahl and Patterson 2014, Dahl et al. 2017). Cannibalism reported here may also be a response to growing lionfish densities and increasingly limited prey supply (Polis 1981). Notably, cannibalized P. volitans in this study were generally small, most weighing under 0.3 g wet mass (Dahl et al. 2017), indicating that juveniles are most commonly cannibalized.

Cannibalism was confirmed in consumers of all size classes, on both natural and artificial reefs, and across all seasons. However, cannibalism frequency of occurrence increased with increasing consumer size and from spring to winter but was observed only slightly more frequently on artificial reef habitats compared to natural reefs. The high frequency of lionfish cannibalism observed in winter coincides with the period when lionfish were at their highest density and their largest mean size during the study period

(Dahl and Patterson 2014). The patterns observed here confirm those reported in Dahl et al. (2017) for prey identified as lionfish via DNA barcoding, and align with studies that consider cannibalism to be an asymmetric interaction, where larger individuals consume smaller individuals (Polis 1981, Pereira et al. 2017).

Estimates of cannibalism from this study should be considered conservative due to cautionary assignment criteria. Our cannibalism assignment criterion of at least two allele differences between consumer and prey genotypes provides unequivocal evidence of cannibalism for consumer-prey pairs that met the criterion, yet one allele

111

difference may also indicate cannibalism. If the instances of single-allele difference between consumer and prey samples were considered to be sufficient evidence of cannibalism, then our upper estimate of confirmed cannibalism would increase from

42% to 49% of barcoding samples that indicated self-DNA was present in prey.

While an objective, conservative criteria for detecting cannibalism was established a priori, low allelic diversity in western Atlantic lionfish populations (Johnson et al. 2016), and the greater difficulty in amplifying nuclear microsatellites versus mitochondrial barcodes from partially digested prey samples (Broquet et al. 2006,

Oliveira and Duarte 2013) may have precluded cannibalism detection for some of the consumer-prey pairs. Western Atlantic lionfish populations have a well-described genetic founder effect, and genetic diversity is especially low in GOM populations

(Johnson et al. 2016). Estimated allele diversity in this study was generally lower as compared to populations originally sampled from North Carolina (Schultz et al. 2013), and the effective number of alleles was fairly reduced relative to the observed number of alleles due to skewed allele frequencies. Thus, while I employed high-diversity microsatellite loci, unique consumer and prey lionfish may not have been genetically distinct at the loci used in this study. Finally, while nuclear microsatellites are relatively stable in degraded DNA and were able to be amplified from most of the digested prey samples in this study, differential digestion likely affected amplification success

(Schneider et al. 2004). In degraded or low concentration samples, PCR amplification success is reduced for nuclear DNA microsatellites compared to mitochondrial DNA markers because hundreds more copies of mitochondria are present in a given cell

(Broquet et al. 2006, Oliveira and Duarte 2013). Thus, degraded DNA quality may have

112

prevented the detection of more cannibalism events in this study and may partially explain the higher detection of self-DNA via barcoding.

The frequency of cannibalism reported here for the nGOM is high when compared to observations in other regions of the western Atlantic, where cannibalism has been reported infrequently (Valdez-Moreno et al. 2012, Côté et al. 2013b,

Villaseñor-Derbez and Herrera-Pérez 2014). Authors of DNA barcoding studies in the

Bahamas and the Mexican Caribbean reported fewer than twenty instances of self-

DNA, which were inferred to indicate cannibalism (Valdez-Moreno et al. 2012, Côté et al. 2013b), notwithstanding the inability of barcoding alone to confirm. Rarer yet are reports of cannibalism observed in visual diet studies (Valdez-Moreno et al. 2012,

Villaseñor-Derbez and Herrera-Pérez 2014, Dahl et al. 2017). Interestingly, there has been no lionfish DNA observed in DNA barcoding diet studies in some other regions of the western Atlantic, such as Belize in the western Caribbean and the Flower Garden

Banks in the western GOM (J.D. Hogan, unpubl. data).

Lionfish Cannibalism: Causes and Consequences

Cannibalism is commonly observed in size-structured predator populations

(Claessen et al. 2004, Rudolf 2008) and is particularly evident in fishes, where it has been recorded in more than 36 teleost families, including Scorpaenidae (Polis 1981,

Smith and Reay 1991, Morte et al. 2001). In most fishes that exhibit cannibalism in nature, conspecific prey provide occasional diet supplementation, but the behavior can also be influenced by exogenous factors (Pereira et al. 2017). For example, cannibalism is often observed to be an inverse function of the availability of alternate prey, increasing when other prey are either absent or unavailable (Polis 1981, Juanes 2003).

Cannibalism may also result from high conspecific density in combination with low prey

113

diversity or abundance (Pereira et al. 2017). The potential benefits of cannibalism are largely governed by density-dependent processes, including increased survival and growth (Babbitt and Meshaka 2000) or reduced competition (Persson et al. 2000).

In some cases, cannibalism can be triggered or exacerbated by unnatural conditions, such as biological invasion (Polis 1981). Lionfish in their invaded range reach densities far greater than those seen in their native range, stemming from a high reproductive output coupled with a lack of effective predators, competitors, and parasites (Green and Côté 2009, Darling et al. 2011, Kulbicki et al. 2012, Albins 2013,

Hackerott et al. 2013). This is especially true on artificial reef habitats in the nGOM, where lionfish mean density had already reached more than 30 fish 100 m−2 by spring

2014 (Dahl and Patterson 2014, Dahl et al. 2016). Individuals of many species maintain fixed spaces or territories in which they are intolerant of conspecifics, and high densities increase the frequency of conspecifics violating this intraspecific space (Polis 1981,

Bailey et al. 2001). Ultimately, unnaturally high densities stemming from predation release in invasive lionfish populations may promote cannibalism on juveniles.

In the nGOM region, little is known about lionfish larval and early juvenile stages, particularly where lionfish settle following their planktonic larval stage. It has been hypothesized that lionfish may settle preferentially in shallow water nursery habitats, such as seagrasses and mangroves, and shift habitat preferences with ontogeny to deeper reefs (Barbour et al. 2010, Biggs and Olden 2011). Perhaps one reason for the lack of cannibalism documented in other regions is that juvenile and adult life stages naturally occupy different habitats. For example, juveniles may move from mangrove habitats to reef habitats as they mature (Claydon et al. 2012). Such ontogenetic habitat

114

shifts may be an adaptive trait to reduce adult antagonism towards and cannibalism on juveniles (Claydon et al. 2012). In the nGOM, juvenile lionfish occupy the same offshore reef habitats as adults sampled in this study because reefs are typically distant (>15 km) from estuaries supporting seagrasses, and mangroves are not currently established in these regions (Stevens et al. 2006). Furthermore, inshore water temperatures may limit lionfish distributions in the nGOM, as they can drop below lionfish critical thermal minima (10° C) in winter, while offshore waters remain warmer (Kimball et al. 2004).

Therefore, even if lionfish were to settle preferentially in shallow water nursery habitats, they may not survive low winter temperatures. This may result in a lack of separation between juvenile and adult habitat in the nGOM, leading to higher encounter rates of densely settled adults and juveniles, leading to higher rates of cannibalism.

Cannibalism has the potential to influence population dynamics of lionfish through density-dependent regulation of population size (Ricker 1954, Polis 1981,

Claessen et al. 2004). To date, there is little evidence for predation on invasive lionfish by native reef fishes in the western Atlantic, whether due to predator naiveté or deterrence from venomous spines (Hackerott et al. 2013, Diller et al. 2014).This apparent lack of biotic resistance to lionfish from native communities has led to unchecked populations of lionfish in the western Atlantic (Albins and Hixon 2013).

However, the degree of cannibalism reported herein may provide regulation for lionfish populations that appear to be plateauing in the nGOM (Dahl et al. 2016). Across the 934 fish sampled for diet analyses and from which cannibalism was detected via DNA barcoding, cannibalism was confirmed in 2.2% of fish via genotyping, and cannibalized lionfish constituted high proportions of the diet when consumed (Dahl et al. 2017). While

115

this is a conservative estimate for reasons stated above, it is notable that even when cannibalism accounts for a small proportion of a species' diet, it may still be a significant source of mortality for the species in question (Polis 1981, Pereira et al. 2017). This may be especially true for invasive lionfish, a species that has escaped natural population control mechanisms (Sikkel et al. 2014, Tuttle et al. 2017). Scant information exists on the frequency of cannibalism reported in wild fish populations with which to compare the case of invasive lionfish (Polis 1981, Pereira et al. 2017). While no information exists on other Scorpaeniform fishes, for flounders, the frequency of juvenile cannibalism is also reported to be relatively low (i.e., frequently <5%, rarely <25%)

(Tanaka et al. 1989, Pereira et al. 2017). Evidence suggests that cannibalism is a major mortality factor in the regulation of many populations (Polis 1981), and, in some cases, cannibalism appears more common in species residing outside of natural geographic ranges (Gomiero and Braga 2004, Fugi et al. 2008, Pereira et al. 2017). Therefore, the incidence of cannibalism observed in lionfish may not be insignificant, although it remains unknown to what extent cannibalism may regulate invasive lionfish populations.

What is known is that nGOM lionfish densities approximately doubled between 2014 and present (Dahl et al. 2016), so the rate of cannibalism may increase further and play an increasing role in population regulation in the region. Recent evidence indicating lionfish population declines in The Bahamas, another region of high lionfish densities, may indicate cannibalism as a potential density-dependent feedback on populations

(Benkwitt et al. 2017).

DNA Barcoding and Predator-Prey Interactions

DNA barcoding is being increasingly used to investigate predator-prey interactions. One of the frequently cited reasons to apply DNA barcoding in diet studies

116

is to resolve bias in diet characterization by extracting high-resolution, species-specific information. Potential issues with employing DNA barcoding to identify trophic interactions, such as secondary predation (i.e., prey within a predator, then eaten by a second predator) (Harwood et al. 2001, Sheppard and Harwood 2005) and scavenging

(Symondson 2002), have been reported in the literature, but the issue of how to treat self-DNA results has been largely unexplored to date.

While it is a commonly held perception that cannibalism is widespread in fishes, relatively few reports exist that describe cannibalism in nature (Smith and Reay 1991,

Pereira et al. 2017). This may be due in part to diet study methodologies, such as DNA barcoding, that lack the ability to discern it. This study demonstrates how using DNA barcoding to characterize predator-prey interactions (i.e., diet) may be biased toward ignoring or overreporting potential cannibalism. The amplification and identification of

DNA barcodes among prey items that match the consumer (i.e., self-DNA) is a frequent occurrence, but results are handled differently among researchers (Sheppard and

Harwood 2005). A false-positive for cannibalism can occur when prey samples are handled with non-sterile techniques, however, even with rigorous sterilization procedures in place, trace amounts of consumer DNA may amplify preferentially over prey DNA if the quality of prey DNA is poor due to digestion (Gonzalez et al. 2012, Dahl et al. 2017). Blocking primers that are used in many DNA barcoding diet studies to prevent the amplification of consumer DNA during polymerase chain reaction (PCR) amplification (e.g., Sousa et al. 2016) may be useful for consumer species known not to exhibit cannibalism (Vestheim and Jarman 2008, De Barba et al. 2014). However,

117

inhibiting amplification of consumer DNA would result in lost information in cases where actual cannibalism is occurring.

Future research opportunities exist to examine how cannibalism may influence invasive species population dynamics, and native community structure. Cannibalism confirmed here for nGOM lionfish via microsatellite genotyping is a step towards better documentation and understanding of cannibalism in wild fish populations. I know of only one other study where microsatellite genotyping was applied to address questions about cannibalism in wild fishes (DeWoody et al. 2001). The results here suggest the approach has wide applicability and that detections of self-DNA from consumers under study should be investigated more closely, which should be more straightforward for species such as lionfish for which polymorphic microsatellite markers and primers have already been developed. Ultimately, a greater understanding of cannibalism in invasive fishes would serve to improve our understanding of their population dynamics, sources of mortality, and potential mitigation.

118

Table 4-1. Microsatellite loci used to determine individual identities for consumers and prey to determine cannibalism events. Loci were isolated and primers designed by Schultz et al. (2013). ABI dye set G5 dye colors are indicated for each primer set, where forward primers were dye labeled, reverse primers were unlabeled. The number of alleles (Na) observed for each locus in this study are reported alongside those reported by Schultz et al. (2013) in parentheses. Unbiased gene diversity (Hs) and effective number of alleles (Ne) are also shown.

Locus Repeat Motif G5 Dye Color Size Range (bp) Primers Na Hs Ne F: TGGTTGGGACTATGCAGACA PVM12 (ACAG)11 VIC 190-246 9(20) 0.810 5.041 R: CCCACACTCAATACCAGCAC F: GGATTCTTTCAGGGCAGGTT PVM14 (AGAT)12 FAM 256-302 6(12) 0.821 5.335 R: TTGTGACCATGACAGCATCA F: TTGGTCCTCCATTTCTGAGG PVM31 (ACT)9 NED 176-221 5(9) 0.747 3.883 R: AGCCTCACTGAGTCCACCAT F: GTGTGTCAGACGCTGAAGGA PVM42 (ATC)11 NED 227-236 4(3) 0.519 2.064 R: ACGTACAGCGGGTTAGGATG

119

Table 4-2. Results from micro-checker analyses (van Oosterhout et al. 2004) showing the number of expected homozygotes (He) and observed homozygotes (Ho), and presence/absence of stuttering, scoring errors and null alleles. Results from probability tests for null alleles are also shown. Evidence of Evidence of Evidence of Significant Locus H H e o Stuttering Scoring Error Null Alleles Probability Test

PVM12 11.3 18 none none Possible No

PVM14 11.6 28 none none Possible Yes

PVM31 19.8 16 none none none No

PVM42 36.8 34 none none none No

120

Figure 4-1. Microsatellite genotypes of consumers and prey pairs analyzed to evaluate potential cannibalism. Size class: S= ≤200 mm, M = 200–250 mm, L= ≥250 mm TL; season: S= spring, U=summer, F=fall, W=winter; and, habitat: natural reef = NR, artificial reef = AR. Consumer genotypes are compared against their respective prey genotypes at each allele (e.g. a and b). A lack of data is indicated by dashes (---). Number of allele differences between consumers and prey are identified, as well as whether a complete genotype was obtained (i.e., all loci amplified for consumer and prey). Cannibalism inference was based on a priori criteria and was “inconclusive” if amplification failures occurred or only one allele difference was observed between genotypes. Consumer-prey pairs in bold indicate cannibalism confirmed by genotyping, where at least two alleles differed between a lionfish and its prey

121

Figure 4-1. Continued

122

Figure 4-1. Continued

123

CHAPTER 5 DENSITY-DEPENDENT CONDITION AND GROWTH OF INVASIVE LIONFISH IN THE NORTHERN GULF OF MEXICO

Indo-Pacific red lionfish, Pterois volitans (Linnaeus, 1758) have extensively invaded the tropical and subtropical western Atlantic Ocean, including waters of the

Caribbean Sea and Gulf of Mexico, over the last 30 years (Schofield 2010, Schofield et al. 2014). Part of their success in the invaded range is an ecological release from natural population control mechanisms otherwise present in their native range, such as predation, disease, and parasitism (Albins and Hixon 2013, Tuttle et al. 2017). For example, predators do not appear to limit lionfish populations in the western Atlantic, thus resulting in greater densities and larger body sizes than those reported for lionfish in the Pacific (Darling et al. 2011, Hackerott et al. 2013). Broad environmental tolerances have led to lionfish recruiting to a diversity of habitat types, including mangroves (Barbour et al. 2010), seagrass beds (Claydon et al. 2012), low relief hardbottom reefs (Muñoz et al. 2011), mesophotic reefs (Lesser and Slattery 2011) and artificial reefs (Dahl and Patterson 2014), and as generalist predators themselves they are able to adapt to a variety of locally abundant prey (Côté and Maljković 2010, Dahl et al. 2017). These attributes have contributed to lionfish being arguably the most successful marine fish invader recorded (Morris and Akins 2009, Côté et al. 2013a), which poses long-term threats to native communities by directly altering community and

Content reprinted with permission from Kristen Dahl. Inter-Research. Marine Ecology Progress Series. Dahl KA, Patterson WF III, Edwards MA. Density-dependent condition and growth of invasive lionfish in the northern Gulf of Mexico. In Review

124

trophic structure (Lesser and Slattery 2011, Dahl et al. 2016), as well as by reducing prey fish biomass and species richness (Green et al. 2012, Dahl et al. 2016).

Lionfish occurrence in the northern Gulf of Mexico (nGOM) is relatively recent, where first sightings were reported off the southwest Florida coast in early 2010

(Schofield 2010). Lionfish were reported throughout the entire GOM basin less than a year later (Fogg et al. 2013, Dahl and Patterson 2014, Nuttall 2014). In the nGOM, lionfish densities increased exponentially in the following years on both natural and artificial reefs, with their densities on nGOM artificial reefs being among the highest in the western Atlantic (Dahl and Patterson 2014). Lionfish in this region consume a broad diversity of fish and invertebrate prey, and exhibit habitat-specific and ontogenetic trends in feeding ecology (Dahl and Patterson 2014, Dahl et al. 2017). Most recently, extensive density-dependent cannibalism also has been documented in the region

(Dahl et al. 2018).

Given the high abundance and wide distribution of lionfish in the invaded range, eradication of the species is thought to be unachievable (Côté et al. 2013a). Targeted lionfish removal programs are currently the best management option available to reduce lionfish biomass and body size in regions where lionfish are already having negative impacts (Barbour et al. 2011, Frazer et al. 2012). Population and ecosystem models may be used to estimate the removal effort necessary to reduce lionfish biomass and mitigate negative impacts (Morris et al. 2011a, Chagaris et al. 2017), but these approaches rely heavily on age and growth estimates to parameterize and model population dynamics across time (Kolar and Lodge 2001, Morris et al. 2011a).

125

Lionfish age and growth parameters have been estimated from several locations within the western Atlantic; however, datasets often exhibit a truncated age structure due to being sampled within too few years following colonization (Potts et al. 2010,

Edwards et al. 2014, Johnson and Swenarton 2016, Fogg 2017). Furthermore, lionfish population dynamics are likely to change as the invasion progresses (Bøhn et al. 2004,

Gutowsky and Fox 2012). This is because the population growth of most invasive species follows a predictable trajectory which starts with a lag period of low densities, increases to exponential growth and high densities, and eventually peaks near carrying capacity (Crooks and Soule 1999, Sakai et al. 2001). Thus, through the process of invasion and establishment a species may experience both density-independent and density-dependent factors based on the demographic and environmental conditions present, resulting in phenotypic changes in life history traits (Bøhn et al. 2004). As the invasive lionfish population reaches carrying capacity in the nGOM, density-dependent processes, including decreased growth, may begin to regulate lionfish populations in the region via increased inter- and intra-specific competition for prey resources.

Here, I report age and growth estimates for lionfish sampled during 2013-2017 at natural and artificial reefs in an approximately 25,000 km2 region of the nGOM.

Specific objectives were to test for the presence of density-dependent effects on lionfish condition (mass at length relative to the population) and growth. The data presented herein are unique given their comprehensive nature and the time series over which sampling occurred, thus enabling the examination of density-dependent feedbacks as the nGOM invasive lionfish population reached its apparent peak.

126

Materials and Methods

Study Location and Specimen Collection

All applicable institutional and/or national guidelines for the care and use of animals were followed during the course of this study. Lionfish density estimation occurred at offshore (>10 km) nGOM artificial and natural reefs across the Florida shelf between 2013 and 2017. Remotely operated vehicle (ROV) sampling was conducted within this region following the methods and at the reef locations reported by Dahl and

Patterson (2014). The purpose of ROV sampling was to track regional habitat-specific lionfish densities and update through 2017 the 2010-2013 time series reported by Dahl and Patterson (2014).

Lionfish were sampled for age and growth analyses via spearfishing at nearby artificial and natural reefs in the same system (Fig. 5-1), with sampling reefs ranging in depth from 24 to 54 m. Artificial reef study sites were primarily concrete modules consisting of single pyramid, paired tetrahedron, or paired reef balls deployed in 2003 by the Florida Fish and Wildlife Conservation Commission (FWC) within the Escambia-

East and Okaloosa Large Area Artificial Reef Sites (LAARS) south of Pensacola and

Destin, FL, USA, respectively (Dance et al. 2011). Other artificial habitats sampled included larger concrete pyramids, decommissioned oil platforms and small shipwrecks.

Natural reef habitat sampled by divers off northwest Florida included carbonate or sandstone outcrops with vertical relief ≤3 m, and moderately sloping ridges of rock rubble and shell hash with little vertical relief (Thompson et al. 1999). Lionfish spearing was localized immediately posterior to the head which severed their spinal column. At the surface, fish were placed in mesh bags in an ice-slurry. Each lionfish was weighed to the nearest 0.1 g, measured to the nearest mm total length (TL), and sex was

127

determined by macroscopic examination of gonads. Both sagittal otoliths were extracted, cleaned and stored dry in 0.5mL microcentrifuge tubes for age determination.

Weight-Length Relationships and Condition

A weight-length non-linear regression was computed for all lionfish collected in the study following the power relationship:

푊 = 푎퐿푏, (3-1) where W is wet mass (kg), L is total length (mm), a is the allometric relationship coefficient between mass and length, and b is the exponent of that relationship.

Relative condition factor (Kn; Le Cren 1951) was computed to examine condition relative to fish in the population from which the individual was sampled:

푊 퐾 = , 푛 푎퐿푏 (3-2) where W and L are variables as defined above, and a and b parameters were derived from the non-linear regression of lionfish weight and length. A three-factor analysis of variance (ANOVA) was computed to test the effects of sex, habitat, timing [i.e., early

(2013-2014) versus late 2015-2017)], and their interactions on Kn. In this analysis, timing serves as a proxy for lionfish density given lionfish density increased across study years but could not be estimated for all reefs where lionfish were sampled.

Age Estimation

Left sagittal otoliths were embedded in epoxy and then sectioned in a transverse plane with a diamond-bladed low-speed saw to approximately 0.4 mm thickness.

Opaque zones were counted along the sulcus under a dissecting microscope with transmitted light and a polarizing light filter at a 20x to 64x magnification and without knowledge of date or location of capture, morphometric data, or opaque zone counts of

128

a second reader (see below). Otolith margins were scored as being either opaque or translucent. Timing of opaque zone formation was evaluated by plotting marginal condition by month for all data pooled among years.

Otolith readability scores were assigned from 1 (unreadable) to 4 (excellent) to determine otoliths suitable for final analysis. A randomly selected subset (n = 1,000) of sectioned otoliths was read by a second (MAE) independent reader without knowledge of fish size or the first (KAD) reader’s age estimate. Average percent error (APE) was calculated to assess precision of age estimates between readers (Beamish and

Fournier 1981). Primary reader counts were utilized to estimate age and estimate growth.

Integer age estimates were used in all analyses with the exception of von

Bertalanffy growth models. Lionfish integer age (in years) was assigned from the number of opaque zones present in otolith sections under the later-verified assumption that opaque zones form annually (Edwards et al. 2014). For samples collected close to the timing of opaque zone formation (i.e., March-April), integer age was sometimes adjusted ±1 year based on margin condition (Beckman et al. 1991). A year was added for fish collected between March and May with a translucent margin >2/3 the thickness of the previous translucent margin (i.e., late forming opaque zone). A year was subtracted for fish collected in December through February with an opaque margin <1/3 the thickness of the previous opaque margin (i.e., early forming opaque zone).

Fractional age was estimated from the number of opaque zones, assumed birthdate, timing of opaque zone formation, and capture date. A mean birthdate of July 1 was based on peak lionfish spawning in the nGOM (Fogg et al. 2017). It was assumed

129

that opaque zone formation was annual and began on March 1 based on previous studies (Edwards et al. 2014, Fogg 2017), as well as data reported below. Fractional age was estimated by first subtracting one opaque zone from the total count for a given otolith and then multiplying the difference by 365 d. Next, 274 days were added to account for the first calendar year of life (July 1 to March 1). Finally, the day of the year that a fish was sampled (days since March 1) was added to account for the number of days in the sampling year that the fish was alive since it began forming its last opaque zone (or since birth in the case of otoliths with zero opaque zones). This result was divided by 365 d to estimate fractional age in years. Similar to integer age estimation methods above, adjustments were made based on margin condition to assign correct age class for some fish. For fish sampled in December-February that had early opaque zone formation, two was subtracted from the total number of opaque zones before multiplying by 365 d. For fish sampled in March-May that had not yet begun forming an opaque zone, zero was subtracted from the total number of opaque zones before multiplying by 365 d.

Lionfish growth was estimated by fitting the von Bertalanffy growth function (von

Bertalanffy 1957) to observed TL-at-age data using fractional age estimates:

퐿 = 퐿 (1 − e−퐾[푡−푡0]), 푡 ∞ (3-3) where Lt = predicted total length (mm) at age t, L∞ = asymptotic total length, k = Brody’s growth coefficient, t = age (y), and t0 = hypothetical age at zero length.

Separate models were fit using all lionfish pooled, as well as for each sex by habitat type using the Solver function in MS Excel to fit the L∞, k, and t0 parameters by minimizing the negative log-likelihood of each model; t0 was fixed to -0.0717 y (Barbour

130

et al. 2011), which corresponds to the pelagic larval duration (26.2 d) of lionfish

(Ahrenholz and Morris 2010). Sex-specific models were calculated with all juveniles included in each model.

A three-factor analysis of variance (ANOVA) was computed to test the effect of sex and habitat on mean size-at-age of the more common ages (1–6 years). Main effects in the model were habitat, sex, and integer age; significant interactions with integer age were sliced by age to test for significant habitat, sex, or habitat x sex effects.

Lionfish density (fish 100 m–2) was estimated on LAARS artificial reefs (n = 16) where fish were sampled with spears from the number of lionfish removed and known sampling area (176 m2). The effect of lionfish density on growth was tested with linear regressions of mean size-at-age for ages 2-4 years versus lionfish density. Age-specific models were computed for males and females separately. The effect of lionfish density on condition (Kn) was tested with linear regressions of mean Kn versus lionfish density, where mean Kn represents all individuals at a given artificial reef.

Results

Density Trends

Densities estimated by ROV sampling on individual study reefs ranged from 0.0 to 1.9 fish 100 m–2 on natural reefs and from 0.0 to 90.4 fish 100 m–2 on artificial reefs. An exponential increase in lionfish density at both natural and artificial reefs was observed beginning in 2011 through 2014, after which mean lionfish density on both reef types reached an apparent peak (Fig. 5-2). By 2013, mean densities on artificial reefs (14.7 fish 100 m–2) were two orders of magnitude higher than on natural reefs

(0.49 fish 100 m–2), a trend which continued through 2017 (Fig.5-2). The highest mean density on artificial reefs (32.98 fish 100 m–2) was observed in 2014, after which

131

densities declined through 2016 to 20.45 fish 100 m–2. By 2017, mean density recovered to peak values observed on artificial reefs in 2014. Densities on natural reefs displayed similar temporal trends with the highest mean density (0.57 fish 100 m–2) observed in 2014, followed by a decline in 2015 (0.34 fish 100 m–2), and stabilizing (i.e., increasing) through 2017 (Fig. 5-2).

Population Demographics

Between 2013 and 2017, divers collected a total of 3,296 lionfish samples for this study, with 2,066 fish from artificial habitats and 1,230 from natural habitats (Fig. 5-3).

Sample sizes were relatively similar between habitats among years; however, no fish were collected from natural reef habitats in 2015, or from artificial habitats in 2017 (Fig.

5-3). Lionfish samples consisted of 1,338 males and 1,609 females (1:1.2 ratio). Sex was not determined for 349 fish, 73.3% of which were smaller than mean size at maturity (<180 mm TL; Morris et al. 2009).

Weight-Length Relationships and Body Condition

Lionfish ranged from 67 to 410 mm in total length (TL) and 0.004 to 0.99 kg in mass (Fig. 5-4). Growth was rapid for both sexes, but male lionfish reached larger maximum sizes compared to females (Fig. 5-5). Males attained 180 mm TL on average within the first year but grew as large as 265 mm TL; females attained on average 168 mm TL but grew as large as 256 mm TL. Mean TL (±SE) was 272.7 (±1.52) mm for males, 238.5 (±1.17) mm for females and 157.6 (±2.55) mm for juveniles. Length frequency distributions reveal steady increases in lionfish body size over time, coupled with a noticeable lack of smaller, younger cohorts in 2016 and 2017 (Fig. 5-5). For females, mass (kg) related to total length (p < 0.001, R2 = 0.95) according to: 푊 =

3.36 푥 10−9퐿3.25, and for males, this relationship (p < 0.001, R2 = 0.96) was: 푊 =

132

2.75 푥 10−9퐿3.28. Log-transformed weight-length relationships were not significantly different between males and females (ANCOVA test for equal slopes; F1,3,976 = 3.02, p =

0.083), thus parameters fit to pooled data were used to calculate relative condition of lionfish. For pooled data from juveniles, males, and females, mass related to TL (p <

0.001, R2 = 0.96, Fig. 5-4) according to: 푊 = 3.09 푥 10−9퐿3.27.

Visual analysis of relative condition data and residual plots revealed that the assumptions of normality and homoscedasticity were met and were appropriate given a large sample size (n = 2,909). Relative condition factor was significantly different by habitat (ANOVA, F1, 2,908 = 272.85, p < 0.001), time (ANOVA, F1, 2,908 = 4.59, p = 0.032), and their interaction (ANOVA, F1, 2,908 = 11.65, p < 0.001) (Table 5-1, Fig. 5-5).

Condition was higher on natural reefs compared to artificial reefs for both early (p <

0.001) and late invasion (p < 0.001) time periods (Figs. 5-6, 5-7). Condition was higher in the early invasion time period compared to later years on natural reefs (p < 0.001), but was not significantly different between time periods on artificial reefs (p = 0.290)

(Figs. 5-6, 5-7). There was no significant effect of sex on relative condition of lionfish in this study (Table 5-1, Fig. 5-6).

Age Estimates, Precision, and Margin Condition

Of the 3,145 sagittal otoliths undamaged and prepared for age estimation, age could be determined for 3,081 individual fish. Otoliths with readability scores >1 were able to be aged, with 22.4% of otolith sections scored as 2 (difficult), 61.3% scored as 3

(readable), and 16.3% scored as 4 (excellent). The overall APE between independent readers was 8.5% among 1,000 otoliths. Agreement between readers was high

(60.1%), and 93% of disagreements were within one opaque zone. Disagreements in

133

counts did not exceed 2 zones and there was no evidence of systematic aging bias (i.e., directionality; Fig. 5-8). The frequency of disagreements increased with lower readability scores, but readability scores for twice-read sections were not proportionally different from those of the full sample.

Marginal condition analysis demonstrated a trend in opaque zone completion occurring during winter/spring (Fig. 5-9). Otolith opaque zones generally began forming in February, with a peak in complete otolith opaque margins (>80%) in March and April.

Translucent margins were dominant (>50%) in samples collected between August and

January, with the highest proportion of translucent margins occurring in January.

Population Dynamics and Growth

Lionfish age demographics in the nGOM were followed over 5 years of invasion.

Otolith opaque zones revealed integer ages from age 0 to age 7 years, with each age class comprising, on average, 1.3%, 11.1%, 29.7%, 29.8%, 17.3%, 8.1%, 2.3%, and

0.3% of samples, respectively (Fig. 5-5). Among all years, lionfish had a mean age

(±SE) of 2.9 (±0.02) years, and 88% of individuals were between the ages of 1 and 4 years (Fig. 5-5). Fractional ages were estimated to be as young as 0.2 years old. The oldest estimated fractional ages came from a 408 mm male estimated to be 7.3 years old, and a 304 mm female estimated to be 7.7 years old. Estimated ages corresponded to birth years and settlement between 2008 and 2016.

The four Von Bertalanffy growth functions computed separately for females and males by habitat type exhibited differential growth and resulted in different parameter values (Table 5-2, Fig. 5-10). All lionfish exhibited rapid growth in the first 1-3 years of life, after which growth slowed towards L∞ (Fig. 5-10). Females were predicted to grow

-1 faster towards L∞ than males on both artificial (k, 0.460 versus 0.359 y ) and natural (k,

134

0.558 versus 0.375 y-1) reefs (Fig. 5-10). Males ultimately grew larger than females on both artificial (L∞, 387.5 versus 315.5 mm) and natural (L∞, 393.0 versus 300.8 mm) habitats, demonstrating sexually dimorphic growth (Fig. 5-10). The model fit to data from all lionfish resulted in intermediate parameter values of L∞ (344.8 mm) and k (0.431 y-1; Table 5-2). Simple t-tests confirmed that the standardized residual means were not significantly different than zero (i.e., all p-values = 1.000), validating the method of fixing t0 for fitting growth models. Furthermore, standardized residuals were rarely above ± 2

(5.4-6.1%).

Differences in nGOM lionfish growth were also demonstrated by the significant interaction between the effects of age, sex and habitat on mean size-at-age (ANOVA, p

< 0.001) of lionfish aged 1–6 years old (Table 5-3, Fig. 5-11). Males grew larger compared to females across all age classes and in both habitats (p < 0.001, Fig. 5-12), with the exception of age-1 fish on natural reefs (p < 0.356, Fig. 5-12). Differences in size-at-age by habitat depended on both sex and age class (Fig. 5-11). Females from natural reefs were significantly larger than those from artificial reefs for age-1 (p <

0.001), age-2 (p < 0.001), and age-3 fish (p = 0.002), but those differences were not statistically significant for age-4, age-5, or age-6 fish (Figs. 5-11, 5-12). A similar trend was seen in males, where fish from natural reefs were significantly larger than those from artificial reefs for age-2 (p = 0.004), age-3 (p < 0.001), and age-4 fish (p < 0.001), but not for age-1, age-5 or age-6 fish (Figs. 5-11, 5-12). A lack of sufficient sample sizes precluded the ability to include 0- and 7-year age classes in size-at-age analyses, and

‘time’ was not included in the model, because main effects and interactions were not significant.

135

Densities estimated by the number of lionfish removed on study artificial reefs ranged from 6.8 to 78.4 fish 100 m–2. Linear regressions of mean size-at-age for ages 2-

4 years versus lionfish density revealed a negative correlation between lionfish size-at- age and lionfish density, where size decreased linearly with increasing lionfish density

(Fig. 5-13). Statistically significant relationships were observed in all female age

2 2 classes: age-2 (F1:14 = 5.46, p = 0.036, R = 0.30), age-3 (F1:14 = 10.52, p = 0.006, R =

2 0.45), age-4 (F1:9 = 19.32, p = 0.002, R = 0.71). Statistically significant relationships

2 were also observed in age-3 (F1:9 = 11.44, p = 0.010, R = 0.59) and age-4 males (F1:8 =

2 8.18, p = 0.024, R = 0.54). This relationship was not significant for age-2 males (F1:9 =

3.04, p = 0.119, R2 = 0.27), however low power of the test (0.34) may have precluded our ability to detect a significant difference. The linear regression of mean relative

2 condition factor, Kn, versus lionfish density was significant (F1:13 = 5.21, p = 0.041, R =

0.30), where Kn decreased linearly with increasing lionfish density (Fig. 5-14).

Discussion

Density-Dependent Growth and Condition

Age and growth estimates reported here yield key insights into the life history and population dynamics of invasive lionfish from both artificial and natural reef habitats across the Florida shelf of the nGOM over a five-year period of invasion, which in turn has important implications for regional carrying capacity. Clear evidence exists to indicate density-dependent effects on growth and condition of invasive lionfish in the nGOM. Lionfish inhabiting densely populated artificial reefs exhibited slower growth and lower body condition suggesting that habitat effects are likely due to differences in lionfish density. On natural reefs, lionfish displayed lower body condition in later years

(2015-2017) of the invasion as lionfish biomass continued to increase steadily. Declines

136

in mean size-at-age and condition with increasing lionfish density on artificial reefs provided additional evidence of a density-dependent feedback on growth in invasive lionfish in the region. While reported experimentally in juvenile (≤71 mm TL) lionfish

(Benkwitt 2013), this is the first evidence of density-dependent growth occurring in adults of this invasive species.

Released from typical constraints on population growth, such as predation, competition, and habitat availability, lionfish populations have reached high density and biomass in their introduced range (Darling et al. 2011, Kulbicki et al. 2012), but all invasive species must eventually reach carrying capacity (Lockwood et al. 2007).

Artificial structures tend to hold high densities of lionfish (Smith and Shurin 2010), and this is especially true in the nGOM where mean densities were nearly 10 fish 100 m–2 within 2 years of colonizing the region and were as high as 32.98 fish 100 m–2 by 2014

(Dahl and Patterson 2014, Dahl et al. 2016). Alternatively, mean densities on natural reefs remained <1 fish 100 m–2 across this study, which is two orders of magnitude lower than those recorded on artificial reefs. Lionfish abundance trends across habitats showed exponential increases in abundance through 2014, followed by a plateau or slight decline in recent years. An oscillating pattern may suggest density-dependent feedbacks which drive populations towards carrying capacity, or the biomass of individuals the environment can support at any one time (Lorenzen and Enberg 2002).

Recent evidence of lionfish population declines on patch reefs in The Bahamas, another region with previously high (> 5 fish m-2) densities, may also be the result of density- dependent feedbacks on those populations (Benkwitt et al. 2017).

137

Direct density dependence in demographic rates (e.g., growth, fecundity, mortality) is responsible for regulating population growth (Jenkins et al. 1999, Post et al.

1999) and may act strongly on invasive species over time as populations establish and grow rapidly towards carrying capacity (Bøhn et al. 2004, Gutowsky and Fox 2012). At high densities, density dependence regulates growth of juveniles and adults via increased inter- and intra-specific competition and a scarcity of resources available to sustain individuals (Jenkins et al. 1999, Smith and Smith 2001, Lorenzen and Enberg

2002). Potential consequences of density-dependent growth include changes in population level fecundity and mortality which are both strongly related to body size

(Lorenzen 1996, Lorenzen and Enberg 2002).

Artificial reefs in the nGOM have significantly different reef fish communities compared to natural hard-bottom reefs, notably with much lower densities of small demersal fishes (e.g., damselfishes, blennies, gobies, and wrasses) (Dance et al. 2011,

Dahl et al. 2016), that have been shown to be the predominant prey among lionfish sampled at natural reefs throughout the nGOM (Dahl and Patterson 2014, Dahl et al.

2017). High lionfish densities observed on artificial reefs are more likely to drive local depletions in prey species forcing lionfish to increase foraging distance from reefs or time spent hunting prey. Indeed, lionfish collected from nGOM artificial reefs in 2013 and 2014 were seen to supplement their diets with non-reef associated prey (Dahl and

Patterson 2014) indicating strong competition for local prey resources and conveying consequences of slower growth and lower body condition (Coulter et al. 2018, this study). Body condition on natural reefs declined in later years of the invasion as lionfish biomass increased, thus low condition observed on artificial reefs across the study may

138

point to density-dependent feedbacks acting on nGOM lionfish populations as early as

2013, as densities were consistently high during sampling years for age and growth analyses.

Lionfish condition and size-age-age were negatively related to density on nGOM artificial reefs, suggesting the overall effect of habitat on lionfish growth in the region was due to stark differences in density between artificial and natural reefs. For each additional lionfish per 100 m2, lionfish were 0.36 to 0.45 mm TL shorter at age on nGOM artificial reefs. These trends translated into large differences in mean total length

(approximately 30 to 70 mm) between the lowest and highest density sites sampled.

While density does not explain all of the variance observed, it is clearly a strong predictor of mean TL. Linear relationships between density and size-at-age were significant for all but age-2 males, though a declining trend remained apparent. A reduction in the mean size of lionfish is a desirable management outcome given that body size is a major determinant of lionfish energetic demands and diet. Larger, mature individuals have higher energetic demands and consume prey at higher rates than smaller sized individuals (Cerino et al. 2013). Fish also represent a higher proportion of lionfish diet for larger individuals; therefore, a population of mostly smaller lionfish may reduce cumulative predation on vulnerable prey fishes (Morris and Akins 2009, Dahl and Patterson 2014).

An implicit assumption of relating condition or size-at-age to lionfish density is that densities estimated at reefs were experienced over the lifetime of lionfish sampled there, thus that lionfish remain resident at a given reef. Studies describing the post- settlement movements of lionfish have shown that lionfish generally have high site

139

fidelity (Akins et al. 2014, Bacheler et al. 2015),but it can be lower under high density conditions (Tamburello and Côté 2015). In the nGOM region, experimental removals performed in the same system revealed adult movement onto cleared reefs, where reefs were separated by 0.3 to 1 km and were distant (>5 km) from nearby natural habitat (Dahl et al. 2016). One aim of ongoing research in this respect is to explicitly test for potential density-dependent dispersal among reefs with acoustic telemetry.

Additional evidence of intense intra-specific competition comes from recent work documenting density-dependent cannibalism of juveniles in nGOM lionfish (Dahl et al.

2018). Cannibalism was confirmed in lionfish sampled from both habitats between 2013 and 2014 which increased in frequency through time, mirroring increases in lionfish density. Incidence of cannibalism is influenced by a variety of factors, namely prey availability and conspecific density, and benefits of the behavior include increased growth and survival, and decreased intra-specific competition (Pereira et al. 2017). In this system, juvenile lionfish appear to recruit to the same offshore reefs inhabited by adults, leading to increased rates of cannibalism via a lack of habitat separation and high encounter rates between different life stages (Claydon et al. 2012, Dahl et al.

2018). Cannibalism acts as another density-dependent feedback mechanism by which populations may persist through time, where lionfish may feed on conspecifics when preferred prey are depleted or unavailable (Polis 1981). While it remains unknown to what extent cannibalism may regulate invasive lionfish populations, smaller and younger fish were infrequently observed in later years of our study which may indicate higher rates of cannibalism across time as prey demand increased with increasing regional lionfish biomass (Dahl et al. 2018).

140

Density-dependent processes may also be compensatory such that increases in population growth rate can occur when populations are at low densities (Rose et al.

2001). Management efforts that directly remove lionfish temporarily lower their densities

(Frazer et al. 2012, Dahl et al. 2016), but remaining individuals may be released from density-dependent mechanisms. Compensatory demographic rates such as increased growth, survival, or recruitment at lower lionfish density following culling events may unintentionally hinder management efforts that aim to reduce lionfish impacts. Among fish, longevity and high fecundity predict strong compensatory responses in population growth (Rose et al. 2001). It is predicted that lionfish may live upwards of 10 years in the wild (Potts et al. 2010), and fecundity is relatively high (Fogg et al. 2017).

Therefore, future studies should be designed to monitor for compensatory changes in growth, survival, or recruitment of lionfish following targeted removals.

Age and Growth in the nGOM

The nGOM lionfish sampled during this study comprised males and females aged 0.2 to 7.7 years, with the majority of fish aged 1-4 years old. Although lionfish were first reported on nGOM habitats in fall 2010, back-calculation of birth dates from individual ages estimated in this study indicates that lionfish may have started colonizing the region as early as 2008, but were at densities too low to be detected. Our

-1 overall (joint-sex) estimates of L∞ (344.8 mm) and k (0.43 y ) are generally consistent with results from previous studies on lionfish in the western Atlantic (Potts et al. 2010,

Barbour et al. 2011, Fogg 2017). Lionfish in North Carolina reached larger maximum sizes (i.e., L∞ = 455mm), where they have the oldest established populations (Potts et al. 2010), and lionfish were reported to grow fastest (k = 0.56 y-1) immediately following initial colonization (2012-2014) in the GOM (Fogg 2017), as compared to pooled data

141

from the current study. Variation in growth estimates between this study and Fogg

(2017) could be explained by a number of different factors including model fitting methods (e.g., t0), sampling period, elapsed time since colonization, environmental factors (e.g., temperature regimes), sampling methods (e.g., divers, tournaments, traps, hook-and-line fishing), or regional differences in lionfish density or prey availability.

Von Bertalanffy growth models are useful for comparison among regions or studies given the range of studies in which lionfish VBGFs have been computed.

However, VBGF parameters are known to be influenced by sample size, especially for datasets that lack smaller or larger fish (Gwinn et al. 2010, Wilson et al. 2015).

Therefore, a more useful means to compare growth between sexes or habitats sampled in the current study is to examine differences in size-at-age directly. For example, lionfish on natural reefs were significantly larger at age than those from artificial reefs across the study, but significance varied by sex and year class. Significant differences were identified at younger ages between 1-3 years for females, and at slightly older ages between 2-4 years for males. Given females reach maximum reproductive output at larger body sizes (i.e., older ages), a greater amount of energy diverted from somatic growth into gamete production would likely explain similarities in size-at-age of older females between habitats (Fogg 2017). Slower growth seen in older males on artificial reefs may reflect higher prey demands required for larger body sizes to maintain the same rate of growth coupled with lower prey diversity and abundance on artificial reefs

(Dahl et al. 2016). An alternative explanation is that older males on high density artificial reefs expend more energy on behaviors related to mating such as agonistic behaviors of competing males (Fishelson 1975, Fogg and Faletti 2018).

142

Finally, sex-specific size-at-age for nGOM lionfish indicate sexually dimorphic growth across the study, where mean size-at-age of males was significantly larger than females. This pattern was apparent across all ages (i.e., 1–6 years) on both natural and artificial reefs with the exception of age-1 fish on natural reefs. Overall, female lionfish in the nGOM achieved a maximum length that was only 0.84x that of males but reached their asymptotic maximum length 1.3x faster. Lionfish growth was rapid for both sexes across the study with male and female fish reaching 50% of their asymptotic length (L∞, averaged across habitats) within 1.9 and 1.5 years, respectively. Growth slowed significantly after 4 years of age in females, but steady growth continued in males through older age classes. Sexually dimorphic growth has been reported for lionfish in the invaded range (Edwards et al. 2014, Fogg 2017), as well as for a number of other

Scorpaenids (Kelly et al. 1999, Bilgin and Çelik 2009, La Mesa et al. 2010), where it is recognized that females grow slower in older age classes due to the higher energy expenditure of gamete (i.e., egg) production compared to males (Cerino et al. 2013).

Conclusions and Implications

The comprehensive lionfish size-at-age dataset presented herein and collected over a five-year period of invasion in the nGOM provides evidence of density-dependent feedbacks on lionfish condition and growth, which may be due to either intra- or interspecific competition. Lionfish populations regulated more by intra-specific competition for limited resources describe a worst-case invasion scenario of invaders released from natural sources of population control and an invaded community exhibiting little biotic resistance to invasion (Albins and Hixon 2013). Fortunately, a rise in biotic resistance to previously unchecked lionfish populations in the western Atlantic may be occurring.

Inter-specific competition (Chagaris et al. 2017) and emerging pathogens (Harris et al.

143

2018) may negatively impact lionfish populations in the region and provide additional population control mechanisms beyond intra-specific competition. Taken together with density-dependent effects on growth reported here, trends in population density signal that lionfish likely are reaching regional carrying capacity in population growth in the nGOM (Simberloff and Gibbons 2004). While this study may not have captured the full extent of density-dependent effects possible as the invasion continues to mature in the nGOM, temporal trends in age and growth of lionfish observed in this study illustrate the importance of the stage of invasion in influencing the population dynamics of invasive species (Bøhn et al. 2004). In the years immediately following invasion, lionfish exhibited traits of rapid population growth, but in later years these traits shifted to reflect density-dependent limitations on lionfish which highlights the role of density in structuring invasive lionfish populations. Future studies should examine other potential density-dependent demographic rates (e.g., movement, fecundity, mortality) in invasive lionfish populations. Ultimately, sex- and habitat-specific growth coefficients from this study may inform previously parameterized models (e.g., Barbour et al. 2011, Chagaris et al. 2017) to assess the level of removal effort required to induce recruitment overfishing and to incorporate potential compensatory growth that may hamper future removal efforts.

144

Table 5-1. Three-factor analysis of variance (ANOVA) results for model computed to test the effect of sex, habitat, time (early versus late invasion), and interactions on relative condition factor, Kn, of northern Gulf of Mexico lionfish samples. Factor levels: time = early invasion (2013-2014), late invasion (2015-2017); sex = female, male; habitat = artificial reef, natural reef. Source df SS MS F p-value Time 1 0.084 0.084 4.59 0.032 Sex 1 0.004 0.004 0.23 0.634 Habitat 1 4.963 4.963 272.85 <0.001 Time x Sex 1 0.001 0.001 0.01 0.926 Time x Habitat 1 0.212 0.212 11.65 <0.001 Sex x Habitat 1 0.001 0.001 0.02 0.901 Time x Sex x Habitat 1 0.060 0.060 3.30 0.069 Residual 2,901 52.763 0.018 Total 2,908 57.898 0.020

Table 5-2. Parameters of von Bertalanffy growth models estimated for all lionfish, and lionfish grouped by sex and habitat combinations sampled from the northern Gulf of Mexico. Von Bertalanffy growth model parameters: are L∞ =

asymptotic maximum length; k = Brody’s growth coefficient; t0 = theoretical

age when total length equals 0. Following Barbour et al. (2011), the t0 parameter in each model was fixed at -0.0717 y. 2 Model group n L∞ (mm) k t0 R All lionfish 3,070 344.79 0.431 -0.0717 0.68 Female AR 1,139 315.45 0.460 -0.0717 0.73 Female NR 678 300.82 0.558 -0.0717 0.70 Male AR 1,018 387.52 0.359 -0.0717 0.74 Male NR 552 392.98 0.375 -0.0717 0.75

145

Table 5-3. Three-factor analysis of variance (ANOVA) results for model computed to test the effects of sex, habitat, and integer age on total length (mm) of northern Gulf of Mexico lionfish. Factor levels: sex = female, male; habitat = artificial reef, natural reef; and age = 1-5 years. Source df SS MS F p-value Age 5 3,866,909 773,382 852.30 <0.001 Habitat 1 23,718 23,718 26.14 <0.001 Sex 1 301,271 301,271 332.02 <0.001 Age x Habitat 5 4,741 948 1.05 0.389 Age x Sex 5 80,561 16,112 17.76 <0.001 Habitat x Sex 1 587 587 0.65 0.421 Age x Habitat x Sex 5 24,645 4,929 5.43 <0.001 Residual 2,715 2,463,596 907 Total 2,738 7,697,847 2,811

146

Figure 5-1. Maps of the northern Gulf of Mexico indicating A) the study region and B) natural (triangles) and artificial (circles) reefs where lionfish were sampled. Isobaths are indicated from 10 to 200 m.

147

Figure 5-2. Invasive lionfish density estimated from remotely operated vehicle video samples at northern Gulf of Mexico natural (n = 16) and artificial (n = 22) reef locations reported by Dahl and Patterson (2014) and updated through 2017.

148

Figure 5-3. Summary of lionfish and otolith sample sizes relative to sampling years, and habitat type.

Figure 5-4. Length-weight relationship for lionfish (n = 3,266) sampled in the northern Gulf of Mexico between 2013 and 2017.

149

Figure 5-5. Total length and age distributions of lionfish sampled in the northern Gulf of Mexico during 2013-2017.

150

Figure 5-6. Mean (±95% CI) relative condition factor, Kn, of northern Gulf of Mexico lionfish by habitat, sex and invasion timing. AR = artificial reef; NR = natural reef; early = 2013-2014; and, late = 2015-2017.

151

Figure 5-7. Post-hoc pairwise multiple comparisons for significant main test results from three-factor ANOVA testing effects of sex, habitat and time on relative condition factor, Kn.

Figure 5-8. Distribution of opaque zone (i.e., annuli) count differences between otolith readers for a subset of randomly selected otoliths (n = 1,000).

152

Figure 5-9. Trend in lionfish otolith marginal condition (n = 3,082) among months for fish sampled in the northern Gulf of Mexico between 2013 and 2017. Numbers above bars indicate aggregate monthly sample sizes.

153

Figure 5-10. Sex- and habitat-specific scatterplots of lionfish total length versus age. Data for juveniles with unassigned sex were allocated to both sexes. Plotted lines are von Bertalanffy growth function fits to the data (t0 fixed at 0.072 y), with functions given on each panel. Symbols denote females (circles) and males (triangles). AR = artificial reef; NR = natural reef.

154

Figure 5-11. Mean (±95% CI) size-at-age of lionfish by sex and habitat for ages 1 to 6 years. Asterisks indicate significant (α < 0.05) effect of habitat within sex and year class (ANOVA, Table 3-2). AR = artificial reef; NR = natural reef.

155

Figure 5-12. P-values of post-hoc pairwise multiple comparisons for significant main test results from three-factor ANOVA testing effects of sex, habitat and age on mean total length (i.e., size-at-age).

156

Figure 5-13. Sex-specific scatterplots of mean size-at-age versus lionfish density at sampled artificial reefs for ages 2-4 years. Plotted lines are linear regressions fit to the data. Each data point represents the mean of all individuals sampled at a given site.

157

Figure 5-14. Scatterplot of relative condition factor, Kn, versus lionfish density at sampled artificial reefs. Plotted line is a linear regression fit to the data. Each data point represents the mean of all individuals sampled at a given site.

158

CHAPTER 6 EXPERIMENTAL ASSESSMENT OF LIONFISH REMOVALS TO MITIGATE REEF FISH COMMUNITY SHIFTS ON NORTHERN GULF OF MEXICO ARTIFICIAL REEFS

Indo-Pacific lionfishes, Pterois volitans/miles complex (hereafter lionfish), have exhibited an extensive and rapid invasion in the western Atlantic Ocean, thus earning the species the distinction of being the most successful marine fish invader to date

(Morris et al. 2009, Albins and Hixon 2013). Lionfish are so abundant and broadly distributed in their invaded range that their complete eradication is thought to be unachievable (Côté et al. 2013a). At the time of this writing, lionfish have established an invaded range of over 7 million km2, in diverse habitat types across the tropical and subtropical western Atlantic Ocean, including waters of the Caribbean Sea and Gulf of

Mexico (GOM) (Schofield 2010, Côté et al. 2013a, Schofield et al. 2014).

The GOM is the most recently invaded basin, where lionfish were first reported in

2009 off the northern Yucatan Peninsula, Mexico (Aguilar-Perera and Tuz-Sulub 2010), in the Florida Keys, USA (Ruttenberg et al. 2012), and along the west Florida shelf

(Schofield 2010). By late 2010, lionfish had been observed in eastern, northern and western regions of the GOM (Schofield 2010, Fogg et al. 2013, Dahl and Patterson

2014, Nuttall 2014). In the short span of time since initial observations, lionfish populations in the northern GOM (nGOM) have increased exponentially and have reached high densities (>20 fish 100 m–2) on artificial reefs, yet their densities on natural reefs remain two orders of magnitude lower (Dahl and Patterson 2014).

Content reprinted by permission from Kristen Dahl: Inter-Research. Marine Ecology Progress Series. Dahl KA, Patterson WF III, Snyder RA. Experimental assessment of lionfish removals to mitigate reef fish community shifts on northern Gulf of Mexico artificial reefs. 2016.

159

The first sightings of lionfish in the nGOM coincided with another significant event in the region: the Deepwater Horizon Oil Spill (DWH). The spill released more than 200 million gallons (~7.6 x 108 L) of oil over several months beginning in April 2010. Effects of the DWH on various biological communities have been documented, yet it is unclear how resilient the nGOM ecosystem will be to this large-scale disturbance (Graham et al.

2010, DeLaune and Wright 2011, Williams et al. 2011, Whitehead et al. 2012). Among reef fishes, reported DWH impacts include changes in diet and trophic position (Norberg

2015, Tarnecki and Patterson 2015), and shifts in community structure (NOAA-NRDA

2015) following exposure to toxic petroleum compounds (Murawski et al. 2014). Recent ecosystem modeling simulations have indicated that depleted reef fish stocks in the region could have contributed to the rapid increase in lionfish density and biomass

(Chagaris et al. 2017). While the DWH may not be the singular factor initiating fish declines, the negative effects of disturbance on native reef fish communities may have similarly increased the system’s vulnerability to lionfish invasion.

While the full extent of chronic impacts of the DWH on reef fishes in the nGOM remains unclear, the literature on invasive lionfish in the western Atlantic suggests they pose a clear long-term threat to nGOM reef fishes. Lionfish impacts on native communities have been reported from invaded regions, suggesting that lionfish alter reef fish community and trophic structure in regions where they have become abundant

(Lesser and Slattery 2011, Albins and Hixon 2013, Albins 2015). Lionfish are novel predators that consume a broad range of fish and invertebrate prey (Albins and Hixon

2008, Morris and Akins 2009, Muñoz et al. 2011), including the juvenile stages of ecologically and economically important fishes (Lesser and Slattery 2011, Dahl and

160

Patterson 2014). Lionfish predation has been correlated with substantial declines in the abundances of small adult reef fishes, as well as juvenile recruits of larger reef fish species (Albins and Hixon 2008, Green 2012, Albins 2015, Benkwitt 2015).

Furthermore, lionfish have caused significant and rapid declines in prey fish biomass

(Green et al. 2012) and species richness (Albins and Hixon 2008) following their arrival on both continuous reefs and patch reefs. Native predator-prey dynamics may also be destabilized in the presence of lionfish (Ingeman and Webster 2015). The invaders can cause nearly 3-fold greater prey mortality when compared with native mesopredators

(Albins 2013). Larger native reef fish species may also be affected via indirect processes such as competition for resources. Dietary overlap of lionfish with native mesopredators, or even apex predators, may lead to decreases in the abundances of those species (Layman and Allgeier 2012). Additionally, lionfish may alter the behavior of native reef fish and invertebrates via competition for space and shelter (Curtis-Quick et al. 2013, Raymond et al. 2015).

The speed of the lionfish invasion coupled with negative impacts to recipient ecosystems has motivated researchers to work towards developing best management practices to mitigate impacts to native communities. There is consensus among researchers and managers that lionfish control is desirable to mitigate their negative effects on marine ecosystems and economies, given that lionfish may remain members of western Atlantic fish communities into the future (Morris and Whitfield 2009, Arias-

González et al. 2011). However, the potential benefits as well as the costs of targeted lionfish removal programs remain unclear. All lionfish management strategies hinge on the goal of a reduction of lionfish populations and thus their corresponding impacts.

161

Targeted removals of lionfish have gained considerable attention in recent years and in some cases have reduced both the numbers and mean size of individuals (Frazer et al.

2012, de León et al. 2013). However, lionfish populations have shown an ability to recover quickly from removal efforts, requiring repeated and substantial harvesting effort to maintain low abundances (Arias-González et al. 2011, Barbour et al. 2011). Partial culling has been effective in some cases for stopping the loss of native prey fish biomass with lower effort than would be required for complete lionfish removal (Green et al. 2014). However, other cases have reported that all lionfish must be removed to see substantial conservation gains (Benkwitt 2015). Promotion of the species as a food fish is also gaining popularity and could be a way to increase the geographical scale of lionfish removals (Ferguson and Akins 2010, Morris et al. 2011c, Côté et al. 2013a).

Here, I report results of a lionfish removal experiment conducted at artificial reef sites in the nGOM. The objectives of the study were to evaluate the effectiveness of targeted lionfish removals as a means to control lionfish densities, as well as to evaluate the effectiveness of lionfish removal for native reef fish community recovery. Pre- invasion community structure data enabled us to examine shifts in native reef fish communities that occurred after lionfish were observed on study reefs in 2010, and then to examine whether lionfish removal efforts facilitated recovery of native fishes.

However, the occurrence of the DWH in 2010 presented a confounding factor for initial changes in reef fish communities, and also patterns seen in reef fish communities following lionfish removals. Therefore, I interpret study results with respect to experimental treatments, as well as within the context of potential effects of the DWH on nGOM reef fishes.

162

Materials and Methods

Study Region and Experimental Reefs

Study sites consisted of 27 artificial reefs within the Escambia East-Large Area

Artificial Reef Site (EE-LAARS; 260 km2), which is located approximately 32 km south of

Pensacola, FL, USA (Fig. 6-1). The same reefs were used for both the observational and the experimental component of this study. Reefs were originally deployed on the seabed (depth range 27-41 m) by the Florida Fish and Wildlife Conservation

Commission in 2003 and consist of 3 different design types: single pyramid, paired tetrahedrons, and paired cylinders with rounded tops (Dance et al. 2011). The composition of all reefs was principally concrete, although pyramid reefs had sides composed of steel rebar in a lattice configuration. Reef volume ranged from 4.09 to 5.68 m3.

Three reefs of each design type were randomly selected for inclusion in one of two lionfish removal treatments or a control group during the removal experiment. Nine reefs were selected for a single lionfish removal event (clear-once treatment) in early

2014, and 9 additional reefs were selected to be repeatedly cleared of lionfish via triannual removal events (maintain-clear treatment) through May 2015. The remaining 9 sites were selected for a control treatment with no lionfish removed over the study.

However, one of the clear-once reefs was mistakenly not cleared of lionfish in winter

2014; thus, there was one more control reef (n = 10), and one less clear-once site (n =

8), than originally planned.

ROV Video Sampling and Analysis

Study reefs were sampled with a VideoRay Pro4 remotely operated vehicle

(ROV) to estimate reef fish community structure for both components of the study, albeit

163

on different time scales. Video sampling was conducted quarterly in 2009-2010, which provided baseline data on reef fish community structure prior to lionfish being observed in the nGOM, and again in 2011-2012, the year following the start of the lionfish invasion. During the removal experiment, triannual (i.e., approximately every 4 mo) video sampling was conducted from December 2013 through August 2015. Specifically,

ROV sampling for the removal experiment was performed in December 2013 (fall 2013),

March 2014 (spring 2014), July 2014 (summer 2014), December 2014 (fall 2014), May

2015 (spring 2015), and August 2015 (summer 2015).

The VideoRay Pro4 ROV (dimensions: 36 cm long, 28 cm tall, 22 cm wide; mass

= 4.8 kg) has a depth rating of 170 m, a 570-line color camera with wide angle (116º) lens, and is equipped with a red laser scaler to estimate fish size. The laser scaler consists of two 5-mW, 635 nm (red) class IIIa lasers mounted in a fixed position 75 mm apart, allowing for estimation of fish size using the ratio of the distance between lasers to the distance between snout and fork length of fishes observed onscreen (Patterson et al. 2009, Dance et al. 2011). The ROV was tethered to the surface and controlled by a pilot via an integrated control box that contains a 38 cm video monitor to observe and capture the digital video feed from the ROV’s camera. Additionally, a GoPro Hero4 high definition (1080p at 120 fps) digital camera was mounted to the forward view of the

ROV to provide high definition video for reef fish community surveys.

The ROV-based point-count sampling method described by Patterson et al.

(2009) was employed to estimate reef fish community structure in a 15 m wide cylinder with reefs at the center of the cylinder’s base. High definition video samples were viewed in a darkroom on a Sony LMD-2110W high-resolution LCD monitor to

164

enumerate and identify reef fishes to the lowest taxonomic level possible. A second video reader independently analyzed randomly selected video samples (n = 16) to estimate reader agreement. Differences between reader estimates were evaluated by computing the average percent error (APE) for each taxon in a given sample, following

(Beamish and McFarlane 1983). The mean of site-specific APEs across all taxa was computed to produce an overall APE between readers.

Total length (TL) was estimated for lionfish struck by the laser scaler if lionfish orientation was estimated to be less than 20° from perpendicular to lasers in order to minimize measurement error (Patterson et al. 2009). Fish size was estimated by first multiplying the length of a fish measured in a video frame by the known distance between lasers (75 mm), and then dividing that product by the distance measured between lasers in the frame. Patterson et al. (2009) estimated a mean negative bias of

3% (SD = 0.6) from this method, thus our estimated lionfish TL was bias-corrected based on a random probability draw and normally distributed bias with a mean equal to

3% and a standard deviation of 0.6%. Total length distributions of lionfish on control sites were gathered from ROV estimates of TL. A one-factor ANOVA was computed to test whether estimates of mean lionfish length estimated with the ROV laser scaler were different among removal treatments at the start of the study in the fall of 2013.

Targeted Removals of Lionfish

Divers removed lionfish from study reefs via spearfishing. Divers were able to capture and remove all lionfish present from reef structure and surrounding seabed during removal events. Initial removals on clear-once and maintain-clear sites were performed in January and February 2014, and then repeated on maintain-clear sites in

July and August 2014, and in February and May 2015. Poor weather conditions

165

prevented February and May 2015 removal events from occurring closer in time.

Lionfish were speared immediately posterior to the skull-spinal column juncture and then placed in a saltwater ice slurry to euthanize. Each lionfish removed was weighed to the nearest 0.1 g and measured to nearest mm total length (TL). The growth function reported by Barbour et al. (2011) for lionfish in USA waters was then solved for age and used to predict age distributions from the TL data obtained from culled fish. A linear regression was computed between lionfish counts from the ROV video samples and lionfish subsequently removed by divers at clear-once or maintain-clear reefs to test for bias in the ROV-derived lionfish counts.

Data Analysis

Pre-Removal Experiment

Permutational analysis of variance (PERMANOVA) models were computed with the Primer statistical package (ver. 6; Anderson et al. 2008) to test for differences in reef fish community structure. Taxa-specific fish densities (fish 100 m–2) were the dependent variables in PERMANOVA models, which were computed with standardized (by total sample abundance) untransformed fish density data, using Bray-Curtis dissimilarities with 10,000 permutations. Models tested whether the pattern in the similarity matrices between levels of factors was significantly different from random. Single-factor

PERMANOVA models were computed to test for differences in reef fish community structure between samples collected in 2009–2010 versus 2011–2012 for all fishes, as well as separately for exploited species (e.g., snappers, groupers, porgies, triggerfish and jacks; Table 6-1) and small demersal reef fishes (e.g., damselfishes, cardinalfishes, blennies, wrasses, gobies; Table 6-1). In the model, reef surveys were nested within reefs across time to account for repeated sampling of individual reefs over time. This

166

partitioning of variance accounted for differences within individual reefs and resulted in a residual error term inappropriate to test the effect of differences among individual reefs (Hinkle et al. 2003, Zar 2010), nor was this effect of primary interest to the study.

Thus, F ratios and p-values were not interpreted for effects of individual reefs in all repeated measures analyses.

Species-specific contrasts were performed for the 25 most abundant species with one-factor repeated measures ANOVAs computed in R, using the stats package

(version 3.1.1; R Development Core Team 20016) to test for differences in fish density between 2009–2010 and 2011–2012 time periods. One-factor repeated measures

ANOVAs also were computed to test the effect of time on reef fish diversity indices of species richness (number of species present), diversity (Shannon-Wiener Hʹ), and evenness (Pielou’s J’), as well as number of individuals (individuals from all species).

For all ANOVA models, assumptions of normality and equal variances were assessed with Shapiro-Wilks (stats package) and Levene’s (Fox and Weisberg 2005, “car” package) tests, respectively, within R. Data met the assumption of equal variances in all models but normality was occasionally violated. Given ANOVA is robust to minor departures from normality (Schmider et al. 2010), models were computed with untransformed data.

Removal Experiment

Two-factor PERMANOVA models were computed to test the effect of removal treatment, sample timing, and their interaction on reef fish community structure among all fishes, as well separately for exploited species and small demersal reef fishes (Table

6-1). Reef sites were nested within treatment to account for repeated sampling of reefs over time. Given only one sample (reef survey) occurred at each site during each time

167

period, the highest order interaction, ‘site(treatment) x time’, was excluded from the model (Anderson et al. 2008). For any significant main effect (at α = 0.05) post-hoc pairwise tests were computed with 10,000 permutations.

Two-factor repeated measures ANOVAs were computed to test the effect of removal treatment, sample timing, and their interaction on reef fish diversity indices of species richness (number of species present), Shannon-Wiener diversity (Hʹ), and

Pielou’s evenness (J’), as well as number of individuals (individuals from all species) and lionfish density. Pairwise multiple comparison procedures (Tukey tests) were used to test which levels were different when a main effect was detected. One-factor repeated measures ANOVAs also were computed to test whether reef fish diversity indices (species richness, Hʹ and J’) and the number of individuals (individuals from all species) were different among all time periods (2009–2010, 2011–2012, and 2013–

2014). The 2013–2014 time period included data from all reefs prior to removals and only control sites following removals.

Video samples collected with ROV at clear-once reefs following lionfish removal in January and February 2014 enabled the estimation of lionfish recolonization rate over the remainder of the study. A linear mixed-effects regression was fit using restricted maximum likelihood to estimate the relationship between lionfish density and time since removal (Pinhiero et al. 2016, “nlme” package). In the model, estimated lionfish density was the response variable, with days since removal as a fixed effect and reef site as a random effect to account for non-independence among repeat samples of the same reefs. The model formula in R is therefore: estimated lionfish density ~ days since removal + (1 | reef), where “1” assumes different intercepts for each reef (i.e., multiple

168

responses dependent on reef). R2 was calculated to describe the proportion of variance explained by both fixed and random factors (Nakagawa and Schielzeth 2013).

Confidence intervals (95%) were calculated for model estimates of intercept and slope.

Results

A total of 299 video samples were collected at study reefs, with 221,965 fish observed among 109 taxa (96.0% identified to species). Of these samples, 137 were collected in 2009–2010 and 2011–2012 (83,967 fish among 85 taxa), and 162 were collected during the removal experiment (137,998 fish among 80 taxa). Fish counts were compared between readers for 16 samples, which produced 191 taxa-specific paired comparisons. The overall APE between readers was 5.7% among these 16 video samples.

Pre-Removal Experiment

No lionfish were observed in 2009-10 video samples, but were observed when sampling resumed following the DWH event. Thus, lionfish first appeared on study reefs sometime between winter 2010 and fall 2011. There was a significant difference in reef fish community structure between 2009-10 and 2011-12 (PERMANOVA, p = 0.015;

Table 6-2). There also were differences in community structure of fishery species

(PERMANOVA, p = 0.042; Table 6-2) and small demersal species (PERMANOVA, p <

0.001; Table 6-2) between 2009–2010 and 2011–2012, as well as in species richness and diversity (ANOVA, p ≤ 0.002; Tables 6-3, 6-4). Higher diversity and evenness, as well as approximately 50% more species, were observed at study reefs in 2009–2010 than in 2011-12; however, there was an increase in number of individuals across all taxa in the latter time period (Fig. 6-2). The increase in the mean number of individuals observed during 2011–2012 is mostly attributed to increases in small (<150 mm TL)

169

pelagic planktivores (e.g., mackerel scad, Decapterus macarellus) and tomtate,

Haemulon aurolineatum (Table 6-2, Fig. 6-3). Declines observed in species richness, diversity and evenness between 2009–2010 and 2011–2012 were beginning to reverse to pre-invasion values by 2013–2014, but trends were not statistically significant for richness or evenness (Fig. 6-2, Table 6-4). Declines in mean density were observed for

19 of the 25 most abundant reef fish species from 2009–2010 to 2011–2012 (Table 6-

5). Out of these, statistically significant declines were observed in bank sea bass,

Centropristis ocyurus (p = 0.033), red porgy, Pagrus pagrus (p < 0.001), slippery dick,

Halichoeres bivittatus (p = 0.002), seaweed blenny, Parablennius marmoreus (p <

0.001), lane snapper, Lutjanus synagris (p = 0.008), yellowtail reeffish, Chromis enchrysura (p < 0.001), gulf flounder, Paralichthys albigutta (p = 0.007), and lesser amberjack, Seriola fasciata (p = 0.009) (Table 6-5). While declines were observed for some larger predatory reef fishes (e.g., snappers, jacks, triggerfish) during 2011–2012, the biggest declines were seen in small (<100 mm) demersal planktivores and invertivores (Table 6-1, e.g., cardinalfishes, blennies, gobies, damselfishes, wrasses).

The density of many of these small demersal species declined by >90% between the

2009–2010 and 2011–2012 time periods (Table 6-5, Fig. 6-3).

Removal Experiment

Six triannual ROV sampling events were conducted at study reefs from fall 2013 to summer 2015 for the lionfish removal experiment. A linear regression relating numbers of lionfish removed and lionfish numbers counted in ROV samples was

2 statistically significant (F1,34 = 283.4, p < 0.001, R = 0.90, lionfish removed = 1.47 +

1.29 x lionfish count). The slope of 1.29 indicates that on average 29% more lionfish were removed from study reefs during removal events than had been estimated to exist

170

on those reefs from ROV video samples. Therefore, ROV-based lionfish counts were scaled upward by a factor of 1.29 to account for incomplete detectability in ROV samples.

Unscaled lionfish counts in video samples ranged from 1 to 184 during the study, which translates to a density range of 0.7 to 103 fish 100 m–2. Estimated initial mean

±SE lionfish density was not different among control, clear-once, and maintain-clear reefs and ranged from 28.9 ± 12.3 to 31.8 ± 5.7 fish (Figs. 6-4, 6-5). Divers removed

1,575 individual lionfish from clear-once and maintain-clear study reefs, including 564 fish during the two follow-up removals at maintain-clear reefs. There was a significant interaction between the effects of removal treatment and sample timing on lionfish density (ANOVA, p < 0.001) (Figs. 6-4, 6-5). Densities of lionfish on control reefs significantly increased over the study period from 31.1 ± 5.7 fish 100 m–2 in fall 2013 to

49.2 ± 7.9 fish 100 m–2 in summer 2015 (p = 0.042) (Fig. 6-5) despite a brief decline in abundance in December 2014 (Fig. 6-4). Across all sample timing periods, control reefs held higher densities of lionfish than maintain-clear (p ≤ 0.027) (Fig. 6-5) reefs following the initial removal event. Control reefs held higher densities than clear-once sites for only two surveys immediately following lionfish removal (p ≤ 0.004) (Figs. 6-4, 6-5). On maintain-clear reefs, lionfish density was only significantly different between December

2013 and March 2014, the time periods immediately prior to and following removal (p =

0.003) (Figs. 6-4, 6-5), and May 2015 (p = 0.020). Following their removal from clear- once and maintain-clear sites in January and February 2014, lionfish densities increased to 4.1 ± 2.0 fish 100 m–2 at maintain-clear and 5.2 ± 1.6 fish 100 m–2 at clear- once reefs by March 2014 (Fig. 6-4). Estimates of lionfish density on maintain-clear

171

sites averaged 9.7 ± 1.5 fish 100 m–2 among all ROV sampling events that on average occurred 2.3 months after lionfish removal events (Fig. 6-4). Lionfish density steadily increased in the year following the single removal event on clear-once reefs, and densities recovered to pre-removal levels by July 2014 (p = 0.153) (Figs. 6-4, 6-5). The initial, pre-removal mean lionfish density on clear-once reefs was surpassed by the end of the study (Fig. 6-4).

There was no difference among treatments in the initial fall 2013 estimates of mean lionfish size estimated with the ROV laser scaler (ANOVA, F2,18=2.77, p = 0.089) which ranged from 207 mm on maintain-clear reefs to 242 mm on control reefs.

Estimated TL of lionfish (n = 222) on control reefs over the study period ranged from

134 to 456 mm (Fig. 6-6), but the size distribution of fish mostly fell between 150 and

350 mm TL. Length frequency distributions from removal events at clear-once and maintain-clear treatments were computed from the 1,575 culled individual lionfish.

Among all sites, total lengths of removed lionfish ranged from 74 to 376 mm (Fig. 6-7).

Length distributions for clear-once and maintain-clear reefs from the initial removal event were similar and skewed toward larger sized lionfish (>200 mm TL; Fig. 6-7 A,B).

Age distributions estimated with the growth function reported by Barbour et al. (2011) indicated the majority of lionfish from both treatments were 1 and 2 year-old fish (Fig. 6-

7 F,G). The size distribution of lionfish removed from maintain-clear reefs in July 2014 had two distinct modes, with the predominant mode centered on 150 mm TL (Fig. 6-7

C). Therefore, fish that recruited to cleared reefs following the initial removal events in

January and February 2014 likely consisted of a large percentage of age-0 fish, as well as individuals as old as 4 years old (Fig. 6-7 G). The final removal events conducted at

172

maintain-clear reefs in February-May 2015 had fewer small (<200 mm TL) fish than the previous removal event, but more than were originally removed in early 2014 (Fig. 6-7

E).

Reef fish communities were not different between treatments at the beginning of the experiment prior to removals. Removal treatment had significant effects on reef fish community structure. For the PERMANOVA model containing all reef fish taxa, both treatment (p = 0.021) and sample timing (p = 0.001) were significant, but their interaction was not (p = 0.254) (Table 6-6). Reef fish communities on control reefs were significantly different from both removal treatments (PERMANOVA, p ≤ 0.034), but removal treatments were not significantly different from each other (PERMANOVA, p=0.333, Fig. 6-8). Models computed for exploited species and small demersal fishes produced different results wherein the effect of sample timing was significant, but treatment and the interaction between the main effects were not (Table 6-6). Substantial gains in abundance were not observed for most taxa regardless of lionfish removal effort (Fig. 6-3). Modest increases in mean density were seen for bank sea bass

(Centropristis ocyurus), pelagic planktivores (e.g., scads, sardines), small demersal fishes (e.g., damselfishes, cardinalfishes) and slippery dick (Halichoeres bivittatus) from targeted lionfish removals (Fig. 6-3). Removal treatment did not affect any of the reef fish diversity indices measured (Table 6-7, Fig. 6-2). The effect of sample timing was significant for species richness, but not for diversity or evenness (Table 6-7, Figs. 6-2,

6-9). Differences in numbers of individuals across taxa stemmed from low numbers in spring and summer 2014 compared to increases in numbers in summer 2015 (Figs. 6-2,

173

6-9). There were no significant interactions between removal treatment and sample timing on any reef fish diversity index (Table 6-7).

The linear mixed model regression fit to lionfish density versus experiment day for clear-once reefs was significant with a slope (±95% CI) of 0.063 (± 0.011) lionfish

100 m–2 d–1 (Fig. 6-10). Therefore, the density of lionfish on cleared reefs was estimated to increase by 1 fish 100 m–2 approximately every 16 days.

Discussion

Extensive baseline data on reef fish community structure at both natural and artificial reefs in the nGOM (e.g., Dance et al. 2011, Patterson et al. 2014) have enabled us to track the lionfish invasion in this region (Dahl and Patterson 2014). Results presented here for artificial reef study sites off northwest Florida clearly demonstrate that shifts in reef fish community structure occurred between 2009–2010 and 2011–

2012, time periods which bracket the appearance of lionfish in the nGOM region (Dahl and Patterson 2014). Taxa-specific differences were most pronounced for small demersal reef fishes, such as damselfishes, cardinalfishes, wrasses, blennies, and gobies, which have been documented as predominant prey of lionfish (Albins and Hixon

2008, Morris and Akins 2009, Dahl and Patterson 2014). In fact, local depletion or extirpation of these taxa due to lionfish predation has been reported in other systems

(Green et al. 2012, Albins 2015). Lionfish densities on nGOM artificial reefs were among the highest in the western Atlantic by fall 2013 (Dahl and Patterson 2014), and mean density had already reached nearly 10 fish 100 m–2 on our study reefs by fall

2012. This is significant in that such densities are above threshold values where ecological impacts have been predicted to occur on Caribbean reefs (Green et al. 2014,

Benkwitt 2015).

174

Observed shifts in reef fish community structure following the arrival of lionfish in the nGOM seem like compelling evidence of lionfish effects, especially given similar declines attributed to lionfish in other parts of their invaded range. However, the occurrence of the DWH in summer 2010 is a confounding factor in drawing inference about potential ecological impacts of lionfish in this region. Estimates of the spatial extent of DWH surface oil extended over our study area periodically from April to August

2010 (Goni et al. 2015). There is clear evidence that some nGOM reef fishes were exposed to toxic petroleum compounds released during the spill (Murawski et al. 2014), with documented impacts on fishes including genetic effects, shifts in trophic position, declines in size at age, and changes in community structure (Whitehead et al. 2012,

NOAA-NRDA 2015, Norberg 2015, Tarnecki and Patterson 2015). Therefore, it is possible that reef fish declines observed at study reefs in 2011–2012 were initially driven by the DWH. Declines observed in larger species, such as snappers and gray triggerfish, during 2011–2012 versus 2009–2010 could have resulted from mortality due to the spill or emigration from spill-affected areas. Few of these species settle directly on reefs, but instead recruit to reefs following months to years in intermediate nursery habitats, such as Sargassum wracks, seagrass beds, or shell rubble reefs. Therefore, it is unlikely that lionfish directly consumed these groups on our study reefs, a conclusion supported by diet data (Dahl and Patterson 2014). Small demersal fishes, such as damselfishes, cardinalfishes, wrasses, blennies, and gobies, are obligate reef residents, settle directly from the plankton onto reef habitat, and are much more site attached than the larger taxa described above. Therefore, localized effects of lionfish were more likely

175

to have affected small demersal fishes directly as opposed to larger and more mobile species.

Our inability to definitely state that reef fish community shifts predated lionfish becoming well-established on study reefs partly stems from the fact that no data on fish community structure were collected at the study reefs during the year immediately following the DWH when mean lionfish density (~5 fish 100 m–2) was less than predicted threshold values from the Caribbean (Dahl and Patterson 2014, Green et al. 2014).

However, declines in the number of species and lower species diversity observed in

2011–2012 relative to 2009–2010 showed signs of stabilizing by 2013, with the possibility of a reversing trend. This occurred while mean lionfish density on study reefs increased to over 30 fish 100 m–2, and mean mass of individuals had nearly doubled over what was observed in fall 2011 (Dahl and Patterson 2014). Therefore, despite an increasing lionfish population in the region, and specifically on our study reefs, fish communities had somewhat stabilized from declines observed following 2009–2010.

One group that did not experience density or diversity increases during 2013, however, was small demersal fishes. Given these taxa are the predominant prey of lionfish in other systems, exponentially increasing lionfish populations after 2012 may have then suppressed any resiliency these groups may have otherwise shown in recovery from the DWH event or limited lionfish presence.

Regardless of the ultimate cause(s) of reef fish community structure shifts observed between 2009–2010 and 2011–2012, a central goal of this study was to conduct lionfish removal events to examine what level of effort would be required to facilitate recovery in affected communities. Targeted removals from nGOM artificial

176

reefs did significantly reduce lionfish density. However, reductions were short-lived as juvenile and adult lionfish rapidly recruited to cleared reefs. Lionfish were observed on cleared reefs as early as a week after removing all lionfish, and more than 500 individuals were removed from maintain-clear reefs during follow-up removal events during the year following initial culling. One year after lionfish removal, clear-once sites had lionfish densities comparable to those of control reefs, and mean lionfish density on clear-once reefs eventually surpassed the levels initially observed in fall 2013. When accounting for incomplete detectability, our estimates of lionfish density illustrate the extent to which the nGOM region is invaded. Mean densities from our control sites throughout the study, and clear-once sites at the conclusion of the study, were 8 to 10- fold higher than the mean density (4.4 fish 100 m–2) reported by Hackerott et al. (2013) in a meta-analysis of lionfish densities on Caribbean reefs. This may explain why I failed to see lasting lionfish reductions in both population numbers and size. Indeed, results reported here are consistent with models that predict sustained removal efforts are required to control lionfish populations (Arias-González et al. 2011, Barbour et al. 2011,

Morris et al. 2011a), perhaps at intensities greater than has been performed elsewhere in the invaded range (Frazer et al. 2012, de León et al. 2013, Benkwitt 2015).

177

lionfish onto previously cleared reefs was high following the first removal event in early

2014, effectively lowering the mean size of individuals at both maintain-clear and clear- once reefs. However, the size distribution of lionfish present on maintain-clear reefs from the final removals in February and May 2015 had shifted back toward larger adults, thus mostly negating early reductions in mean size of lionfish.

The early life history and recruitment dynamics of invasive lionfish are not well understood. Therefore, little information exists to evaluate whether juveniles that recruited to cleared reefs were more likely to have local or distant sources. Adult lionfish also quickly recruited to cleared reefs, which means they had to swim long (>300 m) distances over open substrate to study reefs that were isolated from any natural reef habitat (>5 km) and located between 300 m and 1 km from adjacent artificial reefs. This inference contrasts with reports of limited adult or post-settlement movement in estuarine (Jud and Layman 2012), southeast Atlantic natural hardbottom (Bacheler et al. 2015), and Caribbean patch and continuous coral reef ecosystems (Akins et al.

2014), where site fidelity of lionfish has been reported to be high, and may explain the higher degree of success of targeted removal efforts in such areas (Frazer et al. 2012).

Our findings support recent work that has indicated that lionfish display lower site fidelity under high density conditions (Tamburello and Côté 2015). Lionfish densities observed on control reefs throughout our study represent the highest values reported across their invaded range; thus, intraspecific competition for prey resources may be prompting movement on greater scales than has been reported previously (Benkwitt 2015,

Tamburello and Côté 2015). Consistent with that hypothesis is the fact that non-reef benthic fishes (e.g., lizardfishes, flounders, sea robins) and invertebrates constituted

178

significantly greater proportions of lionfish diet at nGOM artificial reefs versus lionfish recovered from natural reefs (Dahl and Patterson 2014). Therefore, lionfish associated with artificial reefs are clearly spending time away from reefs foraging on non-reef associated prey. The extent of these movements and the area over which lionfish are utilizing prey resources is currently unknown, but conventional or acoustic tagging approaches could be employed to examine those questions.

The rapid recolonization rate of juvenile lionfish settling from the plankton and/or adults immigrating from nearby habitat onto cleared reefs resulted in lionfish densities that were rarely below thresholds proposed by others to mitigate ecological impacts to native fishes despite substantial removal effort (Green et al. 2014, Benkwitt 2015). This may be why our lionfish removals did not translate into significant gains for most fish taxa, though previous studies were mostly focused on small fishes likely to be consumed by lionfish. For larger species included in our analyses, the effects of lionfish were not apparent but may have been undetectable on the timescale studied, especially if impacts are indirect, competitive trophic interactions resulting in reduced growth or reproduction (Albins 2015). Additionally, this study differs from other removal experiments in that small demersal reef fishes (e.g., damselfishes, cardinalfishes, wrasses), which constitute high proportions of lionfish diet in systems or habitats where they are abundant, were already nearly absent from study reefs at the start of the experiment. While these species did increase somewhat following lionfish removals in this study, their densities remained less than 25% of the values observed in 2009–2010, or those reported by Dance et al. (2011) for even earlier time periods. Benkwitt (2015) reported that even single lionfish were able to negate substantial gains in lionfish prey

179

species on small (1 m3) patch reefs in The Bahamas, and Green et al. (2014) reported that approximately 90% lionfish removal was required to foster ecological resiliency for native prey fish communities on larger (100-150 m2) Bahamian patch reefs. No such estimate yet exists for the nGOM region of a threshold lionfish density necessary to mitigate lionfish effects and foster ecosystem resiliency, but recolonization rates of lionfish following experimental removals at study reefs could be used hereafter to predict the level of harvesting effort that would be required to keep lionfish densities suppressed below some threshold. Indeed, our results predict that to maintain lionfish densities <5 fish 100 m–2, all lionfish must be harvested from reefs approximately every two months, about twice the frequency performed in this study.

The extraordinary and continued success of invasive lionfish in the nGOM may be attributable to mechanisms of decreased biotic resistance or resilience. Disturbed ecosystems, regardless of causation, have been shown to be more vulnerable to invasion (Stachowicz et al. 2002). Indeed, recent trophic dynamic ecosystem simulations computed with an Ecopath with Ecosim model of the west Florida Shelf ecosystem indicate that depleted biomass of top predators (e.g., groupers, snappers) can influence the relative invasion success of lionfish (Chagaris et al. 2017). Evidence of native western Atlantic species preying on lionfish is rare; thus, top predators in the

Chagaris et al. (2017) model were assumed not to prey upon lionfish. The model also assumed no lionfish cannibalism; thus, no direct lionfish control was present in the model. Despite this, their results suggest that lionfish invasion success can be influenced through competitive trophic interactions. Historic overexploitation (i.e., overfishing) of top predators in the nGOM region, coupled with declines following the

180

DWH, may have compromised ecosystem resistance to the initial invasion success of lionfish and contributed thereafter to their exponential increases in abundance and biomass.

Localized lionfish removal efforts in this study did not result in substantial gains in native reef fish abundance, but sustained removal efforts were somewhat effective at limiting lionfish densities to relatively low levels on nGOM artificial reefs. Unfortunately, regionally high lionfish densities may require more frequent removal efforts than was attempted, or on much larger spatial scales, to effect meaningful reductions in lionfish density and biomass. If expansive lionfish culling efforts could be accomplished on the shallow (<40 m depth) shelf, lionfish populations associated with mesophotic reefs on the outer shelf and upper continental slope (i.e., below traditional recreational diving limits, 40 m), or other areas that receive little to no control efforts, might still serve as constant sources of new lionfish recruits. Efforts to reduce lionfish biomass at those depths will be logistically challenging and expensive. Therefore, to see beneficial effects on local reef fish communities, lionfish removals going forward will require an effort high enough to offset recolonization from difficult to reach source populations. Ongoing ecosystem modeling efforts that are aimed at evaluating the ecological impacts of lionfish in the nGOM should be coupled with economic models to estimate the expense and feasibility of lionfish removal or harvesting efforts that will be required to accomplish the goal of minimizing lionfish impacts in the northern Gulf of Mexico.

181

Table 6-1. Taxonomic groupings for small (<100 mm TL) demersal reef fishes, exploited fishery species (>200 mm TL) and small (<150 mm TL) pelagic planktivores. Common Name Taxon Small Demersal Reef Fishes Bank Butterflyfish Chaetodon aya Bank Sea Bass Centropristis ocyurus Barred Blenny Hypleurochilus bermudensis Beaugregory Stegastes leucostictus Bicolor Damselfish Stegastes partitus Blackbar Drum Pareques iwamotoi Blue Angelfish Holacanthus bermudensis Cocoa Damselfish Stegastes variabilis Cubbyu Equetus umbrosus Doctorfish Acanthurus chirurgus Gray Angelfish Pomacanthus arcuatus Greenband Wrasse Halichoeres bathyphilus Highhat Equetus acuminatus Honeycomb Cowfish Acanthostracion polygonius Jacknife Fish Equetus lanceolatus Leopard Toadfish Opsanus pardus Painted Wrasse Halichoeres caudalis Purple Reeffish Chromis scotti Queen Angelfish Holacanthus ciliaris Reef Butterflyfish Chaetodon sedentarius Saddle Bass Serranus notospilus Sand Perch Diplectrum formosum Scrawled Cowfish Lactophrys quadricornis Seaweed Blenny Parablennius marmoreus Sharpnose Puffer Canthigaster rostrata Slippery Dick Halichoeres bivittatus Spitlure Frogfish Antennarius scaber Spotfin Butterflyfish Chaetodon ocellatus Spotfin Hogfish Bodianus pulchellus Spotted Batfish Ogcocephalus Pantostictus Squirrelfish Holocentrus adscensionis Striated Frogfish Antennarius striatus Striped Burrfish Chilomycterus schoepfi Twospot Cardinalfish Apogon pseudomaculatus Unidentified Blennies Blennidae

182

Table 6-1. Continued Common Name Taxon Unidentified Cardinalfishes Apogonidae Unidentified Damsels Pomacentridae Unidentified Gobies Gobiidae Unidentified Wrasses Labridae Whitespotted Soapfish Rypticus maculatus Wrasse Bass Liopropoma eukrines Yellowtail Reeffish Chromis enchrysura

Exploited Fishery Species Almaco Jack Seriola rivoliana Cobia Rachycentron canadum Gag Mycteroperca microlepis Goliath Grouper Epinephelus itajara Gray Snapper Lutjanus griseus Gray Triggerfish Balistes capriscus Graysby Epinephelus cruentatus Greater Amberjack Seriola dumerili Jolthead Porgy Calamus bajonado King Mackerel Scomberomorus cavalla Lane Snapper Lutjanus synagris Lesser Amberjack Seriola fasciata Littlehead Porgy Calamus proridens Red Grouper Epinephelus morio Red Porgy Pagrus pagrus Red Snapper Lutjanus campechanus Saucereye Porgy Calamus calamus Scamp Mycteroperca phenax Sheepshead Porgy Calamus penna Snowy Grouper Epinephelus niveatus Unidentified Porgies Sparidae Vermilion Snapper Rhomboplites aurorubens Whitebone Porgy Calamus leucosteus

Pelagic Planktivores Blue Runner Caranx crysos Mackerel Scad Decapterus macarellus Bigeye Scad Selar crumenophthalmus Round Sardinella Sardinella aurita Unidentified Jacks Carangidae

183

Table 6-2. PERMANOVA results of model computed to test for differences in reef fish community structure (species composition and relative abundance) between samples collected in 2009-10 versus 2011-12 estimated from video samples collected with a remotely operated vehicle at study reefs. Model Source df Type III SS MS pseudo-F p-value

All Fishes Time 1 21,114 21,114 3.46 0.015 Site (Time) 34 2.13 x 105 6,257 Residual 101 1.86 x 105 1,838 Total 136 4.20 x 105

Exploited Time 1 14,049 14,049 2.42 0.042 Reef Fishes Site (Time) 34 2.01 x 105 5,927 Residual 101 2.50 x 105 2,479 Total 136 4.67 x 105

Small Time 1 27,318 27,318 4.37 <0.001 Demersal Reef Fishes Site (Time) 34 2.17 x 105 6,393 Residual 101 2.63 x 105 2,604 Total 136 5.09 x 105

184

Table 6-3. One-way repeated measures ANOVA results for models computed to test the effect of timing, 2009-10 versus 2011-12, on reef fish diversity indices and number of individuals. Index Source df Type III SS MS F p-value Species Richness Between Subjects 17 154.91 9.11 Between Treatments 1 130.34 130.34 65.10 <0.001 Residual 17 30.04 2.00 Total 35 319.29

Shannon-Wiener Between Subjects 17 3.414 0.201 Diversity Hʹ Between Treatments 1 0.766 0.766 13.462 0.002 Residual 17 0.967 0.056 Total 35 5.148

Pielou’s Between Subjects 17 0.339 0.020 Evenness J’ Between Treatments 1 0.019 0.019 1.284 0.168 Residual 17 0.158 0.009 Total 35 0.517

Number of Between Subjects 17 3.29 x 106 1.94 x 105 Individuals Between Treatments 1 7.75 x 104 7.75 x 104 1.267 0.276 (Across Taxa) Residual 17 1.04 x 106 6.12 x 104 Total 35 4.41 x 106

185

Table 6-4. Post hoc pairwise multiple comparisons (Tukey) for significant main test results from one-way repeated measures ANOVA testing effect of time (2009- 10 versus 2011-12 versus 2013-14) on reef fish diversity indices on nGOM artificial reefs. Test Species Richness Diversity Hʹ 2009-10 vs 2011-12 <0.001 0.001 2009-10 vs 2013-14 <0.001 0.153 2011-12 vs 2013-14 0.652 0.851

Table 6-5. Mean density (fish 100 m–2) and percent change in the 25 most abundant fishes observed at study artificial reef sites in 2009-10 prior to lionfish presence versus in 2011-12 after lionfish presence was confirmed. Significant differences in mean density between time periods indicated with asterisks (*). 2009-10 2011-12 Species Common Name Mean Mean % Density SE Density SE Change Decapterus macarellus* mackerel scad 82.1 33.8 236.9 69.7 188.7 Haemulon aurolineatum tomtate 82.1 20.3 102.9 29.6 25.4 Lutjanus campechanus red snapper 21.1 3.1 15.7 2.6 -25.8 Rhomboplites aurorubens vermilion snapper 19.2 5.6 10.2 3.7 -46.8 Apogon pseudomaculatus twospot cardinalfish 11.1 6.2 0.2 0.1 -98.5 Caranx crysos blue runner 7.8 4.1 0.1 0.1 -99.2 Centropristis ocyurus* bank sea bass 7.4 1.9 2.8 0.8 -62.4 Pagrus pagrus* red porgy 7 1.4 0.5 0.2 -93.3 Halichoeres bivittatus* slippery dick 6.9 1.9 0.5 0.2 -92.6 Seriola dumerili greater amberjack 5 0.9 6.8 2.1 36.3 Balistes capriscus gray triggerfish 5 0.9 3.4 0.6 -31.4 Parablennius marmoreus* seaweed blenny 3.7 0.9 0.1 --- -97.7 Lutjanus griseus gray snapper 3.3 0.7 3.3 1 1.8 Lutjanus synagris* lane snapper 3.3 1 0.5 0.2 -84.6 Rypticus maculatus whitespotted soapfish 2.8 0.8 1.3 0.2 -53.8 Chromis enchrysura* yellowtail reeffish 2 0.4 0 --- -99.6 Equetus lanceolatus jacknife fish 1.7 0.5 1.2 0.3 -30 Apogon sp. unidentified cardinalfishes 0.8 0.6 0 --- -100 Harengula jaguana scaled sardine 0.7 0.4 0 --- -100 Canthigaster rostrata sharpnose puffer 0.5 0.1 0.8 0.2 62 Paralichthys albigutta* gulf flounder 0.5 0.2 0 --- -100 Seriola fasciata* lesser amberjack 0.4 0.1 0 --- -100 Mycteroperca phenax* scamp 0.4 0.1 0.8 0.1 88.8 Chaetodon ocellatus spotfin butterflyfish 0.4 0.1 0.3 0.1 -15.8 Epinephelus morio red grouper 0.3 0.1 0.2 --- -42.4

186

Table 6-6. PERMANOVA results for models computed to test the effects of lionfish removal treatment and sample timing on reef fish community structure (species composition and relative abundance) estimated from video samples collected with a remotely operated vehicle at study reefs. Model Source df Type III SS MS pseudo-F p-value

All Fishes Treatment 2 51,242 25,621 2.70 0.021 Timing 5 16,587 3,317 2.24 0.001 Site(Treatment) 24 2.27 x 105 9,474 Trt x Timing 10 16,770 1,677 1.13 0.254 Residual 120 1.77 x 105 1,482 Total 161 4.90 x 105

Exploited Treatment 2 10,677 5,338 0.71 0.641 Reef Fishes Timing 5 27,020 5,404 4.42 0.001 Site(Treatment) 24 1.83 x 105 7,621 Trt x Timing 10 13,149 1,315 1.07 0.356 Residual 116 1.42 x 105 1,224 Total 157 3.76 x 105

Small Treatment 2 19,348 9,674 0.88 0.489 Demersal Timing 5 20,130 4,026 3.37 0.001 Reef Fishes Site(Treatment) 24 2.62 x 105 10,948 Trt x Timing 10 9,874 987 0.83 0.805 Residual 119 1.42 x 105 1,195 Total 160 4.54 x 105

187

Table 6-7. Two-way repeated measures ANOVA results for models computed to test the effects of lionfish removal treatment (control, clear-once, maintain-clear) and sample timing on reef fish diversity indices and number of individuals. Index Source df Type III SS MS F p-value

Species Treatment 2 8.34 4.17 0.214 0.809 Richness Site(Treatment) 24 468.44 19.52 Timing 5 149.33 29.87 7.252 <0.001 Trt x Timing 10 26.83 2.68 0.652 0.767 Residual 120 494.17 4.12 Total 161 1147.61

Shannon-Wiener Treatment 2 1.35 0.68 0.877 0.429 Diversity Hʹ Site(Treatment) 24 18.5 0.77 Timing 5 1.64 0.33 2.015 0.081 Trt x Timing 10 1.56 0.16 0.962 0.481 Residual 120 19.5 0.16 Total 161 42.41

Pielou’s Treatment 2 0.31 0.16 1.36 0.275 Evenness J’ Site(Treatment) 24 2.77 0.12 Timing 5 0.28 0.06 1.911 0.097 Trt x Timing 10 0.28 0.03 0.951 0.49 Residual 120 3.52 0.03 Total 161 7.13

Number of Treatment 2 2.38 x 106 1.19 x 106 0.399 0.676 Individuals Site(Treatment) 2 7.15 x 107 2.98 x 106 (Across Taxa) Timing 5 1.54 x 107 3.09 x 106 3.216 0.009 Trt x Timing 10 6.29 x 106 6.29 x 105 0.655 0.764 Residual 120 1.15 x 108 1.30 x 106 Total 161 2.11 x 108

188

Figure 6-1. Map of the northern Gulf of Mexico indicating the locations of the Escambia East-Large Area Artificial Reef Site (EE-LAARS) and the 27 experimental reefs examined in the current study. Symbols (triangles, squares, circles) denote different reef types.

189

Figure 6-2. Species diversity indices and number of individual fish across taxa on study reef sites during spring 2009-winter 2010, fall 2011-summer 2012, and then computed during the lionfish removal experiment from fall 2013-summer 2015. Values are mean values ± SE. The diversity index Hʹ is Shannon- Wiener diversity, and the evenness index J’ is Pielou’s evenness. Removals occurred between fall 13 and spring 14 (maintain-clear and clear-once), summer 14 and fall 14 (maintain-clear), and fall 14 and spring 15 (maintain- clear).

190

Figure 6-3. Mean (±SE) density (fish 100 m–2) of fish taxa observed in remotely operated video samples at artificial reef study sites during spring 2009-winter 2010, fall 2011-summer 2012, and during the lionfish removal experiment. Tomtate (Haemulon aurolineatum), red snapper (Lutjanus campechanus), jacks, Family: Carangidae, gray triggerfish (Balistes capriscus), groupers, Family: Serranidae and small demersal fishes, pelagic planktivorous fishes, twospot cardinalfish, (Apogon psuedomaculatus), gobies and blennies, Superfamily: Gobioidea, damsefishes, Family: Pomacentridae, slippery dick (Halichoeres bivittatus) and bank seabass (Centropristis ocyurus). Values for tomtate at control and maintain-clear reefs in summer 2015 were 484 (±464) and 520 (±523), respectively.

191

Figure 6-3. Continued

192

Figure 6-4. Mean (±SE) lionfish density estimated from counts made with a remotely operated vehicle (ROV) and then scaled (x1.29) to correct for incomplete detectability. Arrows indicate timing of lionfish removal efforts between ROV samples.

193

Figure 6-5. Two-factor repeated measures ANOVA results for model computed to test the effects of lionfish removal treatment and sample timing on lionfish density (fish 100 m–2) estimates at study artificial reefs. Post hoc pairwise multiple comparisons (Tukey) for significant main test results follow. Sample timing is as follows: December 2013 (T1), March 2014 (T2), July 2014 (T3), December 2014 (T4), May 2015 (T5), and August 2015 (T6). Items in bold type indicate significant pairwise comparisons.

194

Figure 6-6. Total length distributions of lionfish estimated with a red laser scaler and remotely operated vehicle at control (no lionfish removal) artificial reef study sites from fall 2013 through summer 2015.

Figure 6-7. Total length distributions of lionfish removed from A) clear-once and B-D) maintain-clear study artificial reefs. Removals for panels A and B were made in February and March 2014, for panel C in July and August 2014, and for panel D between February and May 2015. Age distributions were estimated via the von Bertalanffy growth function reported by Barbour et al. (2011) for lionfish from US Atlantic Ocean waters. Age distributions in panels E-H correspond to size distributions in panels A-D.

195

Figure 6-8. Post hoc pairwise comparisons for significant main test results from PERMANOVA testing effect of removal treatment and sample timing on reef fish community structure of nGOM removal experimental artificial reefs. Items in bold type indicate significant pairwise comparisons.

196

Figure 6-9. Post hoc pairwise multiple comparisons (Tukey) for significant main test results from two-way repeated measures ANOVA testing effect of removal treatment and sample timing on reef fish diversity indices and number of individuals on nGOM removal experimental artificial reefs. Significant main test results were found only for sample timing for A) species richness and B) number of individuals. Items in bold type indicate significant pairwise comparisons.

197

Figure 6-10. Scatterplot of estimated lionfish density (scaled upward by a factor of 1.29 for incomplete detectability) versus days after lionfish removal for clear-once experimental artificial reef sites and the line fit to the significant fixed effect of days after removal. The intercept (±95% CI) is the average of coefficients from individual reefs in the model. The slope (±95% CI) is the recolonization rate of lionfish to all cleared reefs, taking individual reef variation into account.

198

CHAPTER 7 CONCLUSIONS AND STUDY IMPACT

A manipulative field experiment and a series of observational and molecular studies conducted across a range of temporal scales constitute my dissertation research on invasive lionfish populations and their interactions with native communities in the northern Gulf of Mexico (nGOM). Focusing on natural and artificial reefs in the nGOM, my work on population trajectories, life history, trophic ecology, and mitigation of lionfish reveals the significant role that lionfish now play in invaded nGOM ecosystems and provides evidence that the invader poses a substantial threat to numerous taxa in the region. Overall, my dissertation research provides much needed and previously lacking information on lionfish trophic ecology and life history on natural and artificial reefs in the nGOM, as well as on how lionfish affect associated native reef fish communities. The existence of historical data on nGOM reef fish community structure has enabled me to examine shifts in these communities following the lionfish invasion, as well as estimate effectiveness of removal efforts to mitigate those impacts. This research is a novel contribution to invasive species ecology in general, but more specifically to lionfish research, given that few data sets exist with which to compare post-invasion communities with those that existed prior to lionfish being present in a given system. Other novel aspects of my research include the findings of density- dependent effects on cannibalism, body condition, and growth. These factors, combined with plateauing regional lionfish densities point to lionfish in this system reaching a regional carrying capacity. As a direct result of my research, much clearer pictures of lionfish ecology, their direct and indirect effects on native reef fishes, and the likelihood of lionfish removal efforts to mitigate those impacts have emerged. Furthermore, results

199

of these studies provide invaluable empirical data which will readily inform population and ecosystem models parameterized to examine current lionfish impacts and project likely future invasion and population control scenarios.

Population Trends in the nGOM

Invasive Lionfish

Invasive lionfish were first observed on nGOM study reefs in the summer of

2010. In the three years following, lionfish densities exponentially increased on both natural and artificial reef sites in the nGOM (2011 to 2014). Increases in mean TL and predicted body mass were observed during this period, thus lionfish biomass was steadily increasing in the system along with lionfish densities. By 2013, mean lionfish density on artificial reefs (~15 fish 100 m-2) was already among the highest reported among invaded western Atlantic regions, and were two orders of magnitude higher than densities observed on area natural reefs. These discrepancies in lionfish density between habitats persisted for the duration of the study (i.e., through 2017). The highest mean densities across all study years were observed in 2014, with ~33 fish 100 m-2 on artificial reefs and ~0.5 fish 100 m-2 on natural reefs, after which mean lionfish density on both reef types declined. Stabilizing patterns in density between 2015 and 2017 likely indicate lionfish are nearing carrying capacity in the study area, a conclusion that is supported by density-dependent cannibalism and feedbacks on growth reported in

Chapters 3, 4, and 5.

It remains unclear what mechanisms facilitated the extremely rapid increase in lionfish densities documented in the nGOM (Chapter 2). Undoubtedly, a number of traits and ecological mechanisms contribute to rapid population growth in invasive lionfish such as early maturation, high fecundity, and a lack of effective native predators and

200

competitors (Morris 2009, Albins 2013, Diller et al. 2014). The generalist feeding ecology of lionfish under varying conditions of prey availability (Chapters 2-4) may also have contributed to their observed exponential population growth. Additionally, anthropogenic stressors and large-scale ecosystem disturbances, such as historic overfishing of native piscivores, or the Deepwater Horizon Oil Spill (DWH) may have played a role. While overfishing and the DWH are clearly not the sole factors initiating fish declines, the negative effects of disturbance on native reef fish communities may have increased the system’s vulnerability to lionfish invasion.

Differences in fish community structure and habitat complexity between natural and artificial reef habitats likely affected the distribution of lionfish on the nGOM shelf.

Anecdotal and direct evidence exists that large (>20 kg) piscivores, such as sharks or groupers, consume adult lionfish in their invaded range, and groupers (family:

Serranidae) tend to be twice as abundant on natural versus artificial reefs in the system.

However, large snappers (family: Lutjanidae) reside on area artificial reefs in high densities, and no evidence exists that snappers or groupers are consuming lionfish in the nGOM region. Perhaps lionfish population control by native piscivores would be more likely to occur via predation on early life stages versus adult lionfish. However, predator-prey dynamics of juvenile lionfish and potential native predators remain uncharacterized in the region and the western Atlantic as a whole.

Widespread artificial reef deployment has occurred in the region, with reefs typically positioned on sandy or muddy substrates. The bulk of artificial reefs examined in this study were isolated modules with a small footprint (<5 m2) and vertical relief that is substantially higher than the surrounding seabed. Natural reef habitat sampled in this

201

study included carbonate or sandstone outcrops with moderate (≤3 m) vertical relief, and in contrast to isolated artificial reefs, exhibits greater complexity and has a much larger footprint. In areas lacking natural reef habitat, settling larval fish cue to high vertical relief, thus the patchy distribution of artificial reefs may serve to concentrate settling juvenile lionfish in the region, leading to record densities. Furthermore, it is possible that widespread artificial reef habitat, in the absence of suitable natural reef habitat in some parts of the nGOM continental shelf, has facilitated the spread of invasive lionfish in the region. However, the relatively low densities of lionfish on nGOM natural reefs should not be discounted, as natural reefs hold far higher numbers of lionfish across the region given their habitat coverage is several orders of magnitude higher than the area covered by artificial reefs across the region (Chagaris et al. 2017).

Furthermore, densities of lionfish on natural reefs estimated via ROV transects may be biased low due to habitat complexity and cryptic behavior of lionfish. Removal experiment data from Chapter 6 indicated that the ROV undercounted lionfish compared to the numbers of fish removed with divers. However, this discrepancy does not fully account for the sizeable differences in density between habitats.

Native Reef Fish Communities

Disturbed ecosystems, regardless of causation, have been shown to be more vulnerable to invasion (Stachowicz et al. 2002). Baseline data on reef fish community structure at artificial reefs in the nGOM (e.g. Dance et al. 2011, Patterson et al. 2014) have enabled tracking of reef fish community structure and diversity indices over time and in response to disturbance events (Chapter 6). Substantial declines in reef fishes were observed at nGOM artificial reef sites between 2009−2010 and 2011−2012, a period that bracketed the appearance of invasive lionfish in this ecosystem. Small

202

demersal reef fishes, the predominant prey of lionfish in other systems (Albins and

Hixon 2008, Morris and Akins 2009), displayed the greatest declines. However, the first sightings of lionfish in the nGOM coincided with another significant disturbance in the region: the DWH, which is a confounding factor is drawing inference on the cause of native fish declines. The spatial extent of estimated DWH surface oil extended over our study area periodically in 2010, and there is clear evidence that some nGOM reef fishes were exposed to toxic petroleum compounds released during the spill (Goni et al. 2015,

Murawski et al. 2014, Patterson et al. 2019). Mean lionfish density in the year following the DWH was less than threshold values predicted to mitigate native species decline in the Caribbean (Green et al. 2014), and several functional groups were seen to stabilize or reverse density and diversity declines alongside increasing lionfish populations in the nGOM in 2013. However, one functional group that did not experience density or diversity increases during this time was small demersal fishes. Thus, reef fish declines observed at study reefs in 2011−2012 may have been initially driven by the DWH, making the region more susceptible to rapid invasion, and lionfish presence may continue to inhibit recovery of vulnerable groups. It is likely that exponentially increasing lionfish populations following 2012 may have negated any resiliency small demersal fishes may have otherwise shown in response to large-scale disturbance (i.e., DWH or lionfish presence).

Lionfish Feeding Ecology in the nGOM

Feeding ecology of lionfish in the nGOM region was investigated comprehensively using a multidisciplinary approach of traditional stomach content analysis, DNA barcoding analysis, stable isotope analysis, and nuclear DNA microsatellite genotyping. The implications of comprehensive feeding ecology

203

information on lionfish in the newly invaded nGOM are substantial. This work greatly benefits attempts to understand and control rapidly expanding lionfish populations, as well as inform management of important reef fish fisheries in the region. Lionfish feeding ecology data at regional scales are critical data inputs for bioenergetics and ecosystem modeling (Cerino et al. 2013, Chagaris et al. 2017). Such efforts are critical to understanding current and predicted future direct and indirect impacts of lionfish in the nGOM ecosystem, as well as for simulating potential management strategies to mitigate these impacts.

Overall, I found lionfish in the nGOM to be generalist mesopredators with a broad diet, with differences among seasons, habitats, and body size. Lionfish diet was strongly influenced by prey availability via differences in community composition between habitats and among seasons. Small demersal reef fishes, such as damselfishes (family:

Pomacentridae), blennies (family: Blenniidae), gobies (family: Gobiidae), and wrasses

(family: Labridae) were among the more numerically dominant taxa on nGOM natural reefs, but were nearly absent from artificial reef communities after the arrival of lionfish

(Chapter 6). Given high lionfish densities on artificial reefs, I expected to find lionfish there consuming fewer prey due to prey species becoming locally depleted under a high prey demand. However, the lack of prey on artificial reefs appeared to drive lionfish to increase foraging areas to open sandy substrate, resulting in the consumption of species that are not associated with reef habitat. The significance of this finding has implications not only for assessing the direct impacts of lionfish through consumptive effects on native reef fishes, but for assessing indirect impacts to native species through interspecific competition for the same forage base. For example, several nGOM reef

204

fish species, including the economically important red snapper Lutjanus campechanus, are known to forage in sandy substrate habitat surrounding artificial reefs instead of directly on the reef itself (McCawley and Cowan 2007, Tarnecki and Patterson 2015).

Competitive trophic interactions between lionfish and other piscivores have been predicted from ecosystem models on the West Florida Shelf showing intense competition for shared food resources (Chagaris et al. 2017), thus dietary overlap of lionfish with native mesopredators, or even apex predators, may ultimately lead to declines in abundance of those species.

Lionfish became more piscivorous as they grew larger, which was observed in both habitats and across all seasons of study. This ontogenetic shift to greater piscivory observed among diet samples was corroborated via stable isotope analysis of muscle tissue. DNA barcoding revealed that while lionfish are generalist mesopredators, they are more piscivorous in the nGOM than was previously reported (Chapter 3). When

DNA-barcoded prey items were included in estimates of diet, contributions of invertebrates to the overall diet were reduced, yet the ontogenetic shift remained, whereby juvenile lionfish consumed more invertebrates across all seasons.

Employing a multidisciplinary approach to examine lionfish feeding ecology, I was able to determine with high resolution exactly what lionfish consume on nGOM reefs. DNA barcoding greatly enhanced the ability to describe lionfish diet, and together with visual identification, this research revealed the richest diversity of prey reported to date across the invaded range. This molecular method was highly effective at identifying partially digested fish to a high level of resolution, and, most often, sequences were identified to species level (77%). Estimates of prey diversity remain high even when

205

compared to other studies in which DNA barcoding was utilized to describe lionfish diet

(Valdez-Moreno et al. 2012, Côté et al. 2013b), indicating lionfish are a substantial threat to numerous taxa in this region. Of particular importance is the detection of juveniles of several regionally exploited reef fish species (e.g., vermilion and red snappers) as lionfish prey, which has important implications for fishery managers. As lionfish population size, mean body size, and biomass increased on both natural and artificial reefs after data collection ceased for diet analyses (see Chapters 2, 5), these interactions may be stronger now. Overall, diet research presented herein will allow for a more accurate assessment of direct lionfish impacts on native reef fish communities and shed light on potential indirect impacts to other species.

DNA barcoding not only increased our ability to characterize lionfish diet, but also revealed potential cannibalism in the nGOM. Self-DNA (i.e., DNA of the consumer under study) amplification via DNA barcoding could have resulted from either contamination, DNA degradation, or cannibalized lionfish. Given the inability of DNA barcoding to discriminate individuals within a species, thus, differentiate between lionfish self-DNA and a cannibalized lionfish, I utilized nuclear DNA microsatellites to test whether cannibalism was truly occurring in nGOM lionfish populations. Cannibalism was found to be true based on genotyping results, and the highest incidence corresponded to larger sized individuals from areas with high lionfish densities, suggesting cannibalism in the region lionfish is size- and density-dependent.

Increasingly high densities in the region appear to be forcing lionfish to switch to other prey besides reef fishes, such as non-reef associated fishes, pelagic fishes,

206

invertebrates, and themselves, thus cannibalism reported here may have been a response to growing lionfish densities and increasingly limited prey supply.

Cannibalism has the potential to influence population dynamics of lionfish through density-dependent regulation of population size (Ricker 1954, Polis 1981,

Claessen et al. 2004). For a species that appears to have escaped population control via native predators, cannibalism may be a significant source of mortality even when it accounts for a small proportion of a species’ diet (Polis 1981, Pereira et al. 2017). The degree of cannibalism I report, coupled with evidence for density-dependent effects on growth, may provide regulation for lionfish populations that appear to be plateauing in the nGOM in recent years (see Chapter 5). The current rate of cannibalism may be higher than is reported here, given the time in which sample collection occurred, and cannibalism may be playing an increasingly important role in population regulation in the region.

The implications of my multidisciplinary research described in Chapters 2, 3, and

4 extend beyond invasive lionfish feeding ecology. DNA barcoding is often used to study diet in cases where there are sample size limitations. However, DNA barcoding added a significant amount of new information to the characterization of lionfish diet in the region, even though a substantial sampling effort was conducted for visual diet analysis. This result suggests that other diet studies may also benefit from the addition of DNA barcoding to analyses of diet, especially in cases describing trophic interactions of other invasive species. Furthermore, issues surrounding self-DNA results from DNA barcoding of diet were revealed by this research. Authors of diet studies that apply DNA barcoding frequently report self-DNA among prey items (Sheppard and Harwood 2005,

207

Jo et al. 2014). Some authors ignored potential cannibalism by discarding results given potential issues with contamination, (Bartley et al. 2015, Moran et al. 2015), while others have reported all detections of self-DNA as cannibalism, potentially overestimating the true rate. Therefore, these results have important implications for interpreting DNA barcoding analysis of diet in other predatory species where cannibalism may be underreported.

Age and Growth of Lionfish in the nGOM

Age and growth estimates reported here yielded key insights into the life history and population dynamics of invasive lionfish from both artificial and natural reef habitats across the nGOM shelf over a five-year period of invasion, which has important implications for regional carrying capacity. Overall, I found clear evidence indicating density-dependent effects on lionfish growth and condition in the nGOM. Lionfish inhabiting densely populated artificial reefs exhibited slower growth and lower body condition suggesting habitat effects are likely due to differences in lionfish density. On natural reefs, lionfish displayed lower body condition in later years (2015-2017) of the invasion as lionfish biomass continued to increase steadily. Declines in mean size-at- age and condition with increasing lionfish density on artificial reefs provided additional evidence of a density-dependent feedback on growth in invasive lionfish in the region.

This is the first evidence of density-dependent growth occurring in adults of this invasive species. However, recent evidence of lionfish population declines on patch reefs in The

Bahamas, another region with previously high (> 5 fish m-2) densities, may also be the result of density-dependent feedbacks on those populations (Benkwitt et al. 2017).

Although lionfish were first reported on nGOM habitats in fall 2010, back- calculation of birth dates from individual ages estimated in this study indicates that

208

lionfish may have started colonizing the region as early as 2008, but were at densities too low to be detected. There were significant differences in growth and size-at-age between sexes and habitats, with males attaining larger size-at-age than females and fish growing faster at natural reefs. Sexually dimorphic growth has been reported for lionfish elsewhere in the invaded range (Edwards et al. 2014, Fogg 2017), as well as for a number of other Scorpaenids (Kelly et al. 1999, Bilgin and Çelik 2009, La Mesa et al.

2010), where it is recognized that females grow slower in older age classes due to the higher reproductive energy expenditure as compared to males (Cerino et al. 2013).

There are important implications for comprehensive age and growth information on lionfish in the newly invaded nGOM region. The work will greatly benefit attempts to understand and control these rapidly expanding invaders by providing the best regional estimates of growth coefficients possible to inform bioenergetics and ecosystem models

(Barbour et al. 2011, Cerino et al. 2013, Chagaris et al. 2017). Thus far, published growth estimates have come from North Carolina, the Cayman Islands, northeast

Florida, and the southern GOM. However, these datasets suffered from truncated age distributions or did not consider how dynamics of an invader may change over longer temporal scales (Potts et al. 2010, Edwards et al. 2014, Johnson and Swenarton 2016,

Fogg 2017). Growth coefficients from this study may be input into previously parameterized models (e.g., Barbour et al. 2011, Chagaris et al. 2017) to assess the level of removal effort required to induce recruitment overfishing. Finally, lionfish growth rates may be compared to those of similarly sized native mesopredators in the system to infer potential competitive displacement of native species.

209

Potential Mitigation of Lionfish

The rapidity and expanse of the lionfish invasion in the western Atlantic has motivated researchers to work towards best management practices to mitigate impacts to native species. Nearly all lionfish management strategies hinge on the goal of a reduction of lionfish populations, thus their corresponding impacts, as complete eradication of the invasion is unlikely with resources available to management (Barbour et al. 2011). Targeted lionfish removals via spearfishing have the potential to reduce the numbers and mean size of individuals in local populations (Frazer et al. 2012), and were utilized on nGOM artificial reefs to examine their effectiveness and potential to recover native reef fish declines to pre-disturbance levels.

Targeted removals from nGOM artificial reefs significantly reduced lionfish density, but total extirpation was not achieved and juvenile and adult lionfish rapidly recruited to recolonize cleared reefs. While repeated removals kept densities limited to relatively low levels for regional artificial reefs, rapid recolonization resulted in densities that were rarely below thresholds predicted to mitigate ecological impacts to native fishes in the Caribbean, despite substantial removal efforts (Green et al. 2014).

Estimates of lionfish density on maintain-clear sites averaged twice that of the threshold

(~5 fish 100m–2) values predicted by Green et al. (2014). Results from this study suggest that removals should be conducted about twice as frequently as was performed, at least in areas with comparable densities as was observed on nGOM artificial reefs. Recolonization rates reported herein could be used in future studies to predict the level of harvesting effort required to keep lionfish densities suppressed below some threshold.

210

A reduction in the mean size of lionfish present in the system would be a desirable management outcome as it could reduce cumulative predation on vulnerable reef fishes given that lionfish diet shifts with ontogeny and proportionally more fish are consumed at larger sizes (Chapters 2, 3). Larger, mature individuals also have higher energetic demands and consume prey at higher rates than smaller sized fish (Cerino et al. 2013). Unfortunately, targeted removals reduced the mean body size of lionfish only temporarily, when juveniles initially recruited to cleared reefs, but by the final removal cleared reefs had shifted back to holding larger individuals. Compensatory growth dynamics following lionfish removals may be to blame (see Chapter 5), which is deserving of further study with respect to the design of targeted removal programs in the future.

Pre-invasion community structure data enabled me to examine shifts in native reef fish communities and diversity indices that occurred after lionfish were observed on study reefs in 2010, and then to examine whether lionfish removal efforts facilitated recovery of native fishes. Notably, the occurrence of the DWH in 2010 presented a confounding factor for initial changes in reef fish communities, and also patterns seen in reef fish communities following lionfish removals. Overall, increases community in abundance or diversity indices in response to lionfish removals were weak, with few ecological gains experienced by taxa under study. This could be in part due to lasting chronic impacts of DWH in the system, but also due to the temporal duration of the study. For example, two years may not be sufficient time to detect the indirect effects of competitive tropic interactions with lionfish (Albins 2015). For larger, longer-lived species, such as snappers and groupers, negative impacts in the form of decreased

211

reproduction or slower growth may not be detected for a number of years later. For species expected to have direct negative interactions with lionfish (e.g., small demersal fish), lionfish removals resulted in modest gains in abundance, but these taxa did not fully recover to pre-invasion values and gains occurred towards the end of the experiment.

Future Directions

While a substantial amount of work has been done to advance our understanding of lionfish trophic ecology, life history, and mitigation strategies, questions remain as lionfish persist as members in nGOM reef communities. In combination with density- dependent effects on condition, growth, and cannibalism, trends in population density signal that lionfish are reaching a regional carrying capacity in the nGOM. Temporal trends observed in age and growth of nGOM lionfish illustrate the importance of the stage of invasion and the role of density in influencing the population dynamics of invasive species (Bøhn et al. 2004). Lionfish exhibited traits of rapid population growth in the years immediately following invasion, but in later years these traits shifted to reflect density-dependent limitations on lionfish. Future studies should monitor density- dependent effects on growth as the lionfish invasion continues to mature in the nGOM, and also examine other potential density-dependent demographic rates (e.g., fecundity, mortality) in invasive lionfish populations.

Targeted removals via spearfishing remain the most feasible control option for reducing densities of lionfish below carrying capacity or some threshold value, but often only provide local population control in relatively shallow (<40 m) depths. In areas with high densities, such as the nGOM, removals may need to be attempted at much larger spatial scales to exact meaningful reductions in lionfish density and biomass. Lionfish

212

populations in neighboring habitats, mesophotic reefs (i.e., below traditional recreational diving limits), or other areas that receive little to no control efforts, complicate the possibility of such efforts as they may serve as constant sources of new lionfish recruits.

For example, in a recent review of the ecological significance of mesophotic reefs,

Andradi-Brown et al. (2017) proposed invasive lionfish populations derive resilience benefits from mature mesophotic lionfish populations that allow lionfish to invade new sites at shallow depths via egg dispersal and larval settlement. They suggest source- sink dynamics are at play in invasive lionfish populations whereby juveniles and adults subsequently migrate to deeper depths that offer refuge from lionfish control programs.

Localized control of lionfish may only be effective when considering the effects of immigration or recruitment from neighboring or distant habitats, where removal effort must be high enough to offset the recolonization of new individuals. Thus, recruitment dynamics of invasive lionfish populations should be characterized in future studies, as this information data gap exists both regionally and across the entire invaded range.

My findings of large lionfish on newly cleared reefs during the targeted removal experiment in Chapter 6 suggest managers also need to consider the movement of adults from neighboring habitats when planning control efforts. Lionfish in the region are clearly capable of long-range movements which supports recent work that indicates that lionfish display lower site fidelity under high-density conditions (Tamburello & Côté

2015). Lionfish densities observed on nGOM artificial reefs represent the highest values reported across their invaded range, thus intense intra-specific competition for prey resources may be prompting movement on greater scales than has been reported previously (e.g., Green et al. 2011a, Jud and Layman 2012). Indeed, lionfish have been

213

shown to spend time foraging on non-reef associated prey items at distances away from reefs (Chapters 2, 3). The extent of these movements and the area over which lionfish are utilizing prey resources is currently unknown, but conventional or acoustic tagging approaches could be employed to examine those questions. Together with life history information on growth rates, animal movement information is an essential input into ecological models that are the basis for population management of marine species and current lionfish population models do not account for movement of fish following their recruitment to reef habitats (Barbour et al. 2011, Chagaris et al. 2017).

Managers should also consider the impacts that density-dependent factors may have on lionfish populations following targeted removal efforts. Density-dependent processes that limit lionfish populations may also result in increases in population growth rates when populations are at low densities (Rose et al. 2001). Future studies should be conducted to monitor for compensatory changes in population dynamics

(e.g., growth, reproduction) following targeted removal efforts, given that remaining individuals following lionfish removal efforts may be released from density-dependent mechanisms. Compensatory rates (e.g., increased growth, survival, or recruitment) may unintentionally hinder management efforts that aim to reduce lionfish impacts, thus managers would benefit to understand the degree of compensatory growth or recruitment that occurs in populations undergoing removal efforts. This information should increase the accuracy of models that predict population dynamics of lionfish and the amount of removal effort needed to maintain reduced or target densities of this invasive species.

214

DNA barcoding and subsequent genotyping indicated that cannibalism occurred primarily on small juvenile or newly settled lionfish, of which very little is known. Given that early juvenile prey fishes lose most or all identifiable characters rapidly after ingestion, predation on these early life stages is very difficult to detect visually and may be missed completely without molecular techniques. As lionfish persist in western

Atlantic ecosystems, future studies should include DNA barcoding or other molecular techniques to enhance conventional diet studies of native predators and examine whether lionfish are being consumed as juveniles or larvae in the invaded range.

215

LIST OF REFERENCES

Aguilar-Perera, A., and A. Tuz-Sulub. 2010. Non-native, invasive red lionfish (Pterois volitans [Linnaeus, 1758]: Scorpaenidae), is first recorded in the southern Gulf of Mexico, off the northern Yucatan Peninsula, Mexico. Aquatic Invasions 5:S9– S12.

Ahrenholz, D. W., and J. a. Morris. 2010. Larval duration of the lionfish, Pterois volitans along the Bahamian Archipelago. Environmental Biology of Fishes 88:305–309.

Akins, J. L., J. A. Morris, and S. J. Green. 2014. In situ tagging technique for fishes provides insight into growth and movement of invasive lionfish. Ecology and Evolution 4:3768–3777.

Albins, M. A. 2013. Effects of invasive Pacific red lionfish Pterois volitans versus a native predator on Bahamian coral-reef fish communities. Biological Invasions 15:29–43.

Albins, M. A. 2015. Invasive Pacific lionfish Pterois volitans reduce abundance and species richness of native Bahamian coral-reef fishes. Marine Ecology Progress Series 522:231–243.

Albins, M. A., and M. A. Hixon. 2008. Invasive Indo-Pacific lionfish Pterois volitans reduce recruitment of Atlantic coral-reef fishes. Marine Ecology Progress Series 367:233–238.

Albins, M. A., and M. A. Hixon. 2013. Worst case scenario: Potential long-term effects of invasive predatory lionfish (Pterois volitans) on Atlantic and Caribbean coral-reef communities. Environmental Biology of Fishes 96:1151–1157.

Albins, M. A., and P. J. Lyons. 2012. Invasive red lionfish Pterois volitans blow directed jets of water at prey fish. Marine Ecology Progress Series 448:1–5.

Anderson, M., R. Gorley, and K. Clarke. 2008. PERMANOVA+ for PRIMER: Guide to Software and Statistical Methods.

Andradi-Brown, D. A., M. J. A. Vermeij, M. Slattery, M. Lesser, I. Bejarano, R. Appeldoorn, G. Goodbody-Gringley, A. D. Chequer, J. M. Pitt, C. Eddy, S. R. Smith, E. Brokovich, H. T. Pinheiro, M. E. Jessup, B. Shepherd, L. A. Rocha, J. Curtis-Quick, G. Eyal, T. J. Noyes, A. D. Rogers, and D. A. Exton. 2017. Large- scale invasion of western Atlantic mesophotic reefs by lionfish potentially undermines culling-based management. Biological Invasions 19:939–954.

Arias-González, J. E., C. González-Gándara, J. Luis Cabrera, and V. Christensen. 2011. Predicted impact of the invasive lionfish Pterois volitans on the food web of a Caribbean coral reef. Environmental Research 111:917–925.

216

Arroyave, J., and M. L. J. Stiassny. 2014. DNA barcoding reveals novel insights into pterygophagy and prey selection in distichodontid fishes (Characiformes: Distichodontidae). Ecology and Evolution 4:4534–4542.

Babbitt, K. J., and W. E. Meshaka. 2000. Benefits of eating conspecifics: effects of background diet on survival and metamorphosis in the Cuban Treefrog (Osteopilus septentrionalis). Copeia 2000:469–474.

Bacheler, N. M., P. E. Whitfield, R. C. Muñoz, B. B. Harrison, C. A. Harms, and C. A. Buckel. 2015. Movement of invasive adult lionfish Pterois volitans using telemetry: Importance of controls to estimate and explain variable detection probabilities. Marine Ecology Progress Series 527:205–220.

Bailey, H. K., J. H. Cowan, and R. L. Shipp. 2001. Experimental evaluation of potential effects of habitat size and presence of conspecifics on habitat association by young-of-the-year red snapper. Gulf of Mexico Science 19:119–131.

De Barba, M., C. Miquel, F. Boyer, C. Mercier, D. Rioux, E. Coissac, and P. Taberlet. 2014. DNA metabarcoding multiplexing and validation of data accuracy for diet assessment: Application to omnivorous diet. Molecular Ecology Resources 14:306–323.

Barbour, A. B., M. S. Allen, T. K. Frazer, and K. D. Sherman. 2011. Evaluating the potential efficacy of invasive lionfish (Pterois volitans) removals. PLoS ONE 6:e19666.

Barbour, A., M. Montgomery, A. Adamson, E. Díaz-Ferguson, and B. Silliman. 2010. Mangrove use by the invasive lionfish Pterois volitans. Marine Ecology Progress Series 401:291–294.

Barnes, D. K. A., and P. Milner. 2005. Drifting plastic and its consequences for sessile organism dispersal in the Atlantic Ocean. Marine Biology 146:815–825.

Bartley, T., K. S. Mccann, T. J. Bartley, H. E. Braid, K. S. Mccann, N. P. Lester, B. J. Shuter, and R. H. Hanner. 2015. DNA barcoding increases resolution and changes structure in Canadian boreal shield lake food webs. DNA Barcodes 3:30–43.

Bax, N., A. Williamson, M. Aguero, E. Gonzalez, and W. Geeves. 2003. Marine invasive alien species: A threat to global biodiversity. Marine Policy 27:313–323.

Beamish, R. J., and D. A. Fournier. 1981. A method for comparing the precision of a set of age determinations. Canadian Journal of Fisheries and Aquatic Sciences 38:982–983.

Beamish, R. J., and G. A. McFarlane. 1983. The forgotten requirement for age validation in fisheries biology. Transactions of the American Fisheries Society 112:735–743.

217

Beckman, D. W., A. L. Stanley, J. H. Render, and C. A. Wilson. 1991. Age and growth- rate estimation of sheepshead Archosargus probatocephalus in Louisiana waters using otoliths. Fishery Bulletin 89:1–8.

Benkwitt, C. E. 2013. Density-dependent growth in invasive lionfish (Pterois volitans). PLoS ONE 8:e66995.

Benkwitt, C. E. 2015. Non-linear effects of invasive lionfish density on native coral-reef fish communities. Biological Invasions 17:1383–1395.

Benkwitt, C. E., M. A. Albins, K. L. Buch, K. E. Ingeman, T. L. Kindinger, T. J. Pusack, C. D. Stallings, and M. A. Hixon. 2017. Is the lionfish invasion waning? Evidence from The Bahamas. Coral Reefs 36:1255–1261. von Bertalanffy, L. 1957. Quantitative laws in metabolism and growth. The Quarterly Review of Biology 32:217–231.

Betancur-R., R., A. Hines, A. Acero P., G. Ortí, A. E. Wilbur, and D. W. Freshwater. 2011. Reconstructing the lionfish invasion: insights into Greater Caribbean biogeography. Journal of Biogeography 38:1281–1293.

Biggs, C. R., and J. D. Olden. 2011. Multi-scale habitat occupancy of invasive lionfish (Pterois volitans) in coral reef environments of Roatan, Honduras. Aquatic Invasions 6:347–353.

Bilgin, S., and E. Ş. Çelik. 2009. Age, growth and reproduction of the black scorpionfish, Scorpaena porcus (Pisces, Scorpaenidae), on the Black Sea coast of Turkey. Journal of Applied Ichthyology 25:55–60.

Bøhn, T., O. T. Sandlund, P. A. Amundsen, and R. Primicerio. 2004. Rapidly changing life history during invasion. Oikos 106:138–150.

Bohnsack, J. A., D. E. Harper, D. B. McClellan, and M. Hulsbeck. 1994. Effects of reef size on colonization and assemblage structure of fishes at artificial reefs off southeastern Florida, U.S.A. Bulletin of Marine Science 55:796–823.

Bowen, S. H. 1996. Quantitative description of the diet. Page 513–532 of 732 Fisheries Techniques.

Bradley, B. J., and L. Vigilant. 2002. False alleles derived from microbial DNA pose a potential source of error in microsatellite genotyping of DNA from faeces. Molecular Ecology Notes 2:602–605.

Braid, H. E., J. Deeds, S. L. DeGrasse, J. J. Wilson, J. Osborne, and R. H. Hanner. 2012. Preying on commercial fisheries and accumulating paralytic shellfish toxins: a dietary analysis of invasive Dosidicus gigas (Cephalopoda Ommastrephidae) stranded in Pacific Canada. Marine Biology 159:25–31.

218

Broquet, T., N. Ménard, and E. Petit. 2006. Noninvasive population genetics: A review of sample source, diet, fragment length and microsatellite motif effects on amplification success and genotyping error rates. Conservation Genetics 8:249– 260.

Brown, M. B., and A. B. Forsythe. 1974. Robust tests for the equality of variances. Journal of the American Statistical Association 69:364–367.

Byron, D., K. L. Heck, and M. A. Kennedy. 2014. Presence of juvenile lionfish in a northern Gulf of Mexico nursery habitat. Gulf of Mexico Science 32:8.

Carlton, J. T. 1989. Man’s role in changing the face of the ocean: Biological invasions and implications for conservation of near-shore environments. Conservation Biology 3:265–273.

Carr, M. H., and M. A. Hixon. 1995. Predation effects on early post-settlement survivorship of coral- reef fishes. Marine Ecology Progress Series 124:31–42.

Carreon-Martinez, L., T. B. Johnson, S. A. Ludsin, and D. D. Heath. 2011. Utilization of stomach content DNA to determine diet diversity in piscivorous fishes. Journal of Fish Biology 78:1170–1182.

Castells, M. 2010. The Information Age: Economy, Society and Culture Volume I: The Rise of the Network Society. 2nd edition. Wiley Blackwell, West Sussex.

Caut, S., E. Angulo, and F. Courchamp. 2008. Dietary shift of an invasive predator: rats, seabirds and sea turtles. Journal of Applied Ecology 45:428–437.

Cerino, D., A. S. Overton, J. A. Rice, and J. A. Morris. 2013. Bioenergetics and trophic impacts of the invasive Indo-Pacific lionfish. Transactions of the American Fisheries Society 142:1522–1534.

Chagaris, D., S. Binion-Rock, A. Bogdanoff, K. Dahl, J. Granneman, H. Harris, J. Mohan, M. B. Rudd, M. K. Swenarton, R. Ahrens, W. F. Patterson, J. A. Morris, and M. Allen. 2017. An ecosystem-based approach to evaluating impacts and management of invasive lionfish. Fisheries 42:421–431.

Chapin, F. S., E. S. Zavaleta, V. T. Eviner, R. L. Naylor, P. M. Vitousek, H. L. Reynolds, D. U. Hooper, S. Lavorel, O. E. Sala, S. E. Hobbie, M. C. Mack, and S. Díaz. 2000. Consequences of changing biodiversity. Nature 405:234–242.

Chistiakov, D. A., B. Hellemans, and F. A. M. Volckaert. 2006. Microsatellites and their genomic distribution, evolution, function and applications: A review with special reference to fish genetics. Aquaculture 255:1–29.

Claessen, D., A. M. de Roos, and L. Persson. 2004. Population dynamic theory of size- dependent cannibalism. Proceedings of the Royal Society B: Biological Sciences 271:333–340.

219

Clare, E. L., B. K. Lim, M. D. Engstrom, J. L. Eger, and P. D. N. Hebert. 2007. DNA barcoding of Neotropical bats: species identification and discovery within Guyana. Molecular Ecology Notes 7:184–190.

Clarke, K., and R. Gorley. 2006. PRIMER version 6: User manual/tutorial. Plymouth.

Claydon, J. A. B., M. C. Calosso, and S. B. Traiger. 2012. Progression of invasive lionfish in seagrass, mangrove and reef habitats. Marine Ecology Progress Series 448:119–129.

Collis, K., D. D. Roby, D. P. Craig, S. Adamany, J. Y. Adkins, and D. E. Lyons. 2002. Colony size and diet composition of piscivorous waterbirds on the Lower Columbia River: Implications for losses of juvenile salmonids to avian predation. Transactions of the American Fisheries Society 131:537–550.

Colwell, R. K., C. X. Mao, and J. Chang. 2004. Interpolating, extrapolating, and comparing incidence-based species accumulation curves. Ecology 85:2717– 2727.

Côté, I. M., S. J. Green, and M. A. Hixon. 2013a. Predatory fish invaders: Insights from Indo-Pacific lionfish in the western Atlantic and Caribbean. Biological Conservation 164:50–61.

Côté, I. M., S. J. Green, J. A. Morris, J. L. Akins, and D. Steinke. 2013b. Diet richness of invasive Indo-Pacific lionfish revealed by DNA barcoding. Marine Ecology Progress Series 472:249–256.

Côté, I., and A. Maljković. 2010. Predation rates of Indo-Pacific lionfish on Bahamian coral reefs. Marine Ecology Progress Series 404:219–225.

Coulter, D. P., R. MacNamara, D. C. Glover, and J. E. Garvey. 2018. Possible unintended effects of management at an invasion front: Reduced prevalence corresponds with high condition of invasive bigheaded carps. Biological Conservation 221:118–126.

Crawley, M. J. 1987. What makes a community invasible? Pages 429–453 Colonization, Succession and Stability.

Le Cren, E. D. 1951. The length-weight relationship and seasonal cycle in gonad weight and condition in the perch (Perca fluviatilis). The Journal of Animal Ecology 20:201–219.

Crooks, J. A., and M. E. Soule. 1999. Lag times in population explosions of invasive species: Causes and implications. Pages 103–125 in O. T. Sandlund, P. J. Schei, and A. Viken, editors. Invasive Species and Biodiversity Management. Kluwer Academic Publishers, London.

220

Cure, K., C. E. Benkwitt, T. L. Kindinger, E. A. Pickering, T. J. Pusack, J. L. McIlwain, and M. A. Hixon. 2012. Comparative behavior of red lionfish Pterois volitans on native Pacific versus invaded Atlantic coral reefs. Marine Ecology Progress Series 467:181–192.

Curtis-Quick, J., E. Underwood, S. Green, L. Akins, A. Harborne, and I. Côté. 2013. Interactions between the Caribbean spiny lobster, Panulirus argus, and invasive lionfish, Pterois volitans: Who displaces whom? Proceedings of the 66th Gulf and Caribbean Fisheries Institute.

Dahl, K. A., and W. F. Patterson. 2014. Habitat-specific density and diet of rapidly expanding invasive red lionfish, Pterois volitans, populations in the northern Gulf of Mexico. PLoS ONE 9:e105852.

Dahl, K. A., W. F. Patterson, A. Robertson, and A. C. Ortmann. 2017. DNA barcoding significantly improves resolution of invasive lionfish diet in the Northern Gulf of Mexico. Biological Invasions 19:1917–1933.

Dahl, K. A., W. F. Patterson, and R. A. Snyder. 2016. Experimental assessment of lionfish removals to mitigate reef fish community shifts on northern Gulf of Mexico artificial reefs. Marine Ecology Progress Series 558:207–221.

Dahl, K. A., D. S. Portnoy, J. D. Hogan, J. E. Johnson, J. R. Gold, and W. F. Patterson. 2018. Genotyping confirms significant cannibalism in northern Gulf of Mexico invasive red lionfish, Pterois volitans. Biological Invasions 20:3513–3526.

Dance, M. A., W. F. Patterson III, and D. T. Addis. 2011. Fish community and trophic structure at artificial reef sites in the northeastern Gulf of Mexico. Bulletin of Marine Science 87:301–324.

Darling, E. S., S. J. Green, J. K. O’Leary, and I. M. Côté. 2011. Indo-Pacific lionfish are larger and more abundant on invaded reefs: A comparison of Kenyan and Bahamian lionfish populations. Biological Invasions 13:2045–2051.

Dawnay, N., R. Ogden, R. McEwing, G. R. Carvalho, and R. S. Thorpe. 2007. Validation of the barcoding gene COI for use in forensic genetic species identification. Forensic Science International 173:1–6.

DeLaune, R. D., and A. L. Wright. 2011. Projected impact of Deepwater Horizon Oil Spill on U.S. Gulf Coast wetlands. Soil Science Society of America Journal 75:1602.

DeWoody, J. a, D. E. Fletcher, S. D. Wilkins, and J. C. Avise. 2001. Genetic documentation of filial cannibalism in nature. Proceedings of the National Academy of Sciences of the United States of America 98:5090–5092.

221

Diller, J. L., T. K. Frazer, and C. A. Jacoby. 2014. Coping with the lionfish invasion: Evidence that naïve, native predators can learn to help. Journal of Experimental Marine Biology and Ecology 455:45–49.

Edwards, M. A., T. K. Frazer, and C. A. Jacoby. 2014. Age and growth of invasive lionfish (Pterois spp.) in the Caribbean Sea, with implications for management. Bulletin of Marine Science 90:953–966.

Elton, C. S. 1958. The ecology of invasions by animals and plants. Methuen, London, UK.

Ferguson, T., and J. Akins. 2010. The lionfish cookbook: The Caribbean’s new delicacy. REEF Environmental Education Foundation, Key Largo, Florida.

Fishelson, L. 1975. Ethology and reproduction of pteroid fishes found in the Gulf of Agaba (Red Sea), especially Dendrochirus brachypterus. Pubbl. Staz. Zool. Napoli 39:635–656.

Fishelson, L. 1997. Experiments and observations on food consumption, growth and starvation in Dendrochirus brachypterus and Pterois volitans (Pteroinae, Scorpaenidae). Environmental Biology of Fishes 50:391–403.

Fogg, A. Q. 2017. Life history of the non-native invasive red lionfish (Pterois volitans) in the northern Gulf of Mexico. MS Thesis, University of Southern Mississippi, Ocean Springs.

Fogg, A. Q., N. J. Brown-Peterson, and M. S. Peterson. 2017. Reproductive life history characteristics of invasive red lionfish (Pterois volitans) in the northern Gulf of Mexico. Bulletin of Marine Science 93:791–813.

Fogg, A. Q., and M. E. Faletti. 2018. Invasive lionfish (Pterois sp.) agonistic behavior observations. Bulletin of Marine Science 94:1–2.

Fogg, A. Q., E. R. Hoffmayer, W. B. Driggers, III, M. D. Campbell, G. J. Pellegrin, and W. Stein. 2013. Short communication: Distrbution and length frequency of invasive lionfish (Pterois sp.) in the northern Gulf of Mexico 25:111–115.

Fox, J., and S. Weisberg. 2005. An R Companion to Applied Regression. 2nd edition. Sage, Thousand Oaks, CA.

Frazer, T. K., C. A. Jacoby, M. A. Edwards, S. C. Barry, and C. M. Manfrino. 2012. Coping with the lionfish invasion: Can targeted removals yield beneficial effects? Reviews in Fisheries Science 20:185–191.

Frézal, L., and R. Leblois. 2008. Four years of DNA barcoding: Current advances and prospects. Infection, Genetics and Evolution 8:727–736.

222

Fugi, R., K. D. G. Luz-Agostinho, and A. A. Agostinho. 2008. Trophic interaction between an introduced (peacock bass) and a native (dogfish) piscivorous fish in a Neotropical impounded river. Hydrobiologia 607:143–150.

Gagneux, P., C. Boesch, and D. S. Woodruff. 1997. Microsatellite scoring errors associated with noninvasive genotyping based on nuclear DNA amplified from shed hair. Molecular Ecology 6:861–868.

Gomiero, L. M., and F. M. S. Braga. 2004. Cannibalism as the main feeding behaviour of tucunares introduced in southeast Brazil. Brazilian Journal of Biology 64:625– 32.

Goni, G. J., J. A. Trinanes, A. MacFadyen, D. Streett, M. J. Olascoaga, M. L. Imhoff, F. Muller-Karger, and M. A. Roffer. 2015. Variability of the Deepwater Horizon surface oil spill extent and its relationship to varying ocean currents and extreme weather conditions. Pages 1–22 in M. Ehrhardt, editor. Mathematical modelling and numerical simulation of oil pollution problems. Springer International Publishing.

Gonzalez, J. M., M. C. Portillo, P. Belda-Ferre, and A. Mira. 2012. Amplification by PCR artificially reduces the proportion of the rare biosphere in microbial communities. PLoS ONE 7:e29973.

Goossens, B., L. P. Waits, and P. Taberlet. 1998. Plucked hair samples as a source of DNA: Reliability of dinucleotide microsatellite genotyping. Molecular Ecology 7:1237–1241.

Graham, W. M., R. H. Condon, R. H. Carmichael, I. D’Ambra, H. K. Patterson, L. J. Linn, and F. J. Hernandez Jr. 2010. Oil carbon entered the coastal planktonic food web during the Deepwater Horizon oil spill. Environmental Research Letters 5:045301.

Green, S., J. Akins, and I. Côté. 2011. Foraging behaviour and prey consumption in the Indo-Pacific lionfish on Bahamian coral reefs. Marine Ecology Progress Series 433:159–167.

Green, S. J. 2012. Predation by invasive indo-pacific lionfish on Atlantic coral reef fishes: Patterns, processes, and consequences. Simon Fraser University.

Green, S. J., J. L. Akins, A. Maljković, and I. M. Côté. 2012. Invasive Lionfish Drive Atlantic Coral Reef Fish Declines. PLoS ONE 7:e32596.

Green, S. J., and I. M. Côté. 2009. Record densities of Indo-Pacific lionfish on Bahamian coral reefs. Coral Reefs 28:107.

Green, S. J., and I. M. Côté. 2014. Trait-based diet selection: Prey behaviour and morphology predict vulnerability to predation in reef fish communities. Journal of Animal Ecology 83:1451–1460.

223

Green, S. J., N. K. Dulvy, A. M. L. Brooks, J. L. Akins, A. B. Cooper, S. Miller, and I. M. Côté. 2014. Linking removal targets to the ecological effects of invaders: a predictive model and field test. Ecological Applications 24:1311–1322.

Grosholz, E. D., G. M. Ruiz, C. A. Dean, K. A. Shirley, L. John, P. G. Connors, S. Ecology, and N. May. 2000. The impacts of a nonindigenous marine predator in a California bay. Ecology 81:1206–1224.

Guichoux, E., L. Lagache, S. Wagner, P. Chaumeil, P. Leger, O. Lepais, C. Lepoittevin, T. Malausa, E. Revardel, F. Salin, and R. J. Petit. 2011. Current trends in microsatellite genotyping. Molecular Ecology Resources 11:591–611.

Gutowsky, L. F. G., and M. G. Fox. 2012. Intra-population variability of life-history traits and growth during range expansion of the invasive round goby, Neogobius melanostomus. Fisheries Management and Ecology 19:78–88.

Gwinn, D. C., M. S. Allen, and M. W. Rogers. 2010. Evaluation of procedures to reduce bias in fish growth parameter estimates resulting from size-selective sampling. Fisheries Research 105:75–79.

Hackerott, S., A. Valdivia, S. J. Green, I. M. Côté, C. E. Cox, L. Akins, C. A. Layman, W. F. Precht, and J. F. Bruno. 2013. Native predators do not influence invasion success of Pacific lionfish on Caribbean reefs. PLoS ONE 8:e68259.

Hacunda, J. S. 1981. Trophic relationships among demersal fishes in. Fishery Bulletin 79:179–186.

Hamner, R. M., D. W. Freshwater, and P. E. Whitfield. 2007. Mitochondrial cytochrome b analysis reveals two invasive lionfish species with strong founder effects in the western Atlantic. Journal of Fish Biology 71:214–222.

Handy, S. M., J. R. Deeds, N. V. Ivanova, P. D. N. Hebert, R. H. Hanner, A. Ormos, L. A. Weigt, M. M. Moore, and H. F. Yancy. 2011. A single-laboratory validated method for the generation of DNA barcodes for the identification of fish for regulatory compliance. Journal of AOAC International 94:201–210.

Harris, H. E., A. Q. Fogg, R. P. E. Yanong, S. Frasca, T. Cody, T. B. Waltzek, and W. F. Patterson III. 2018. First report of an emerging ulcerative skin disease in invasive lionfish (FA209). http://edis.ifas.ufl.edu/fa209.

Harwood, J. D., S. W. Phillips, K. D. Sunderland, and W. O. C. Symondson. 2001. Secondary predation: Quantification of food chain errors in an aphid-spider- carabid system using monoclonal antibodies. Molecular Ecology 10:2049–2057.

Hebert, P. D. N., A. Cywinska, S. L. Ball, and J. R. deWaard. 2003. Biological identifications through DNA barcodes. Proceedings of the Royal Society B: Biological Sciences 270:313–321.

224

Hebert, P. D. N., T. R. Gregory, and V. Savolainen. 2005. The promise of DNA barcoding for taxonomy. Systematic Biology 54:852–859.

Hinkle, D. E., W. Wiersma, and S. G. Jurs. 2003. Applied Statistics for the Behavioral Sciences. 5th edition. Houhgton Mifflin, Boston, MA.

Hixon, M. A., and M. H. Carr. 1997. Synergistic predation, density dependence, and population regulation in marine fish. Science 277:946–949.

Hoese, H., and R. Moore. 1998. Fishes of the Gulf of Mexico. 2nd edition. Texas A&M University, College Station, TX.

Hoffman, J. I., and W. Amos. 2005. Microsatellite genotyping errors: Detection approaches, common sources and consequences for paternal exclusion. Molecular Ecology 14:599–612.

Holway, D. A., and A. V. Suarez. 1999. Animal behavior: An essential component of invasion biology. Trends in Ecology and Evolution 14:328–330.

Hornstra, H. M., and A. Herrel. 2004. Gas bladder movement in lionfishes: A novel mechanism for control of pitch. Journal of Morphology 260:1.

Hyslop, E. J. 1980. Stomach contents analysis-a review of methods and their application. Journal of Fish Biology 17:411–429.

Imamura, H., and M. Yabe. 1996. Larval record of a red firefish, Pterois volitans, from Northwestern Australia (Pisces: Scorpaeniformes). Bulletin of the Faculty of Fisheries, Hokkaido University 47:41–46.

Ingeman, K. 2016. Lionfish cause increased mortality rates and drive local extirpation of native prey. Marine Ecology Progress Series 558:235–245.

Ingeman, K., and M. Webster. 2015. Native prey mortality increases but remains density-dependent following lionfish invasion. Marine Ecology Progress Series 531:241–252.

Ivanova, N. V., T. S. Zemlak, R. H. Hanner, and P. D. N. Hebert. 2007. Universal primer cocktails for fish DNA barcoding. Molecular Ecology Notes 7:544–548.

Jenkins, T. M., S. Diehl, K. W. Kratz, and S. D. Cooper. 1999. Effects of population density on individual growth of brown trout in streams. Ecology 80:941–956.

Jo, H., J. A. Gim, K. S. Jeong, H. S. Kim, and G. J. Joo. 2014. Application of DNA barcoding for identification of freshwater carnivorous fish diets: Is number of prey items dependent on size class for Micropterus salmoides? Ecology and Evolution 4:219–229.

225

Johnson, E. G., and M. K. Swenarton. 2016. Age, growth and population structure of invasive lionfish (Pterois volitans/miles) in northeast Florida using a length- based, age-structured population model. PeerJ 4:e2730.

Johnson, J., C. E. Bird, M. A. Johnston, A. Q. Fogg, and J. D. Hogan. 2016. Regional genetic structure and genetic founder effects in the invasive lionfish: Comparing the Gulf of Mexico, Caribbean and North Atlantic. Marine Biology 163:216.

Jones, G. P. 1987. Competitive interactions among adults and juveniles in a coral reef fish. Ecology 68:1534–1547.

Juanes, F. 2003. The allometry of cannibalism in piscivorous fishes. Canadian Journal of Fisheries and Aquatic Sciences 60:594–602.

Jud, Z. R., and C. A. Layman. 2012. Site fidelity and movement patterns of invasive lionfish, Pterois spp., in a Florida estuary. Journal of Experimental Marine Biology and Ecology 414–415:69–74.

Jud, Z. R., C. A. Layman, J. A. Lee, and D. A. Arrington. 2011. Recent invasion of a Florida (USA) estuarine system by lionfish Pterois volitans/P. miles. Aquatic Biology 13:21–26.

Kearse, M., R. Moir, A. Wilson, S. Stones-Havas, M. Cheung, S. Sturrock, S. Buxton, A. Cooper, S. Markowitz, C. Duran, T. Thierer, B. Ashton, P. Meintjes, and A. Drummond. 2012. Geneious Basic: An integrated and extendable desktop software platform for the organization and analysis of sequence data. Bioinformatics 28:1647–1649.

Kells, V., and K. Carpenter. 2011. A field guide to coastal fishes: From Maine to Texas. Johns Hopkins University Press, Baltimore, MD.

Kelly, C. J., P. L. Connolly, and J. J. Bracken. 1999. Age estimation, growth, maturity, and distribution of the bluemouth rockfish Helicolenus d. dactylopterus (Delaroche 1809) from the Rockall Trough. ICES Journal of Marine Science 56:61–74.

Kimball, M. E., J. M. Miller, P. E. Whitfield, and J. A. Hare. 2004. Thermal tolerance and potential distribution of invasive lionfish (Pterois volitans/miles complex) on the east coast of the United States. Marine Ecology Progress Series 283:269–278.

Kimura, M., and J. F. Crow. 1964. The number of alleles that can be maintained in a finite population. Genetics 49:725–738.

Kolar, C. S., and D. M. Lodge. 2001. Progress in invasion biology: Predicting invaders. Trends in Ecology & Evolution 16:199–204.

Kulbicki, M., J. Beets, P. Chabanet, K. Cure, E. Darling, S. Floeter, R. Galzin, A. Green, M. Harmelin-Vivien, M. Hixon, Y. Letourneur, T. de Loma, T. McClanahan, J.

226

McIlwain, G. MouTham, R. Myers, J. O’Leary, S. Planes, L. Vigliola, and L. Wantiez. 2012. Distributions of Indo-Pacific lionfishes Pterois spp. in their native ranges: implications for the Atlantic invasion. Marine Ecology Progress Series 446:189–205.

Kvitrud, M. A., S. D. Riemer, R. F. Brown, M. R. Bellinger, and M. A. Banks. 2005. Pacific harbor seals (Phoca vitulina) and salmon: Genetics presents hard numbers for elucidating predator-prey dynamics. Marine Biology 147:1459–1466.

Layman, C. A., and J. E. Allgeier. 2012. Characterizing trophic ecology of generalist consumers: A case study of the invasive lionfish in the Bahamas. Marine Ecology Progress Series 448:131–141.

Lee, T. N., C. Rooth, E. Williams, M. McGowan, A. F. Szmant, and M. E. Clarke. 1992. Influence of Florida Current, gyres and wind-driven circulation on transport of larvae and recruitment in the Florida Keys coral reefs. Continental Shelf Research 12:971–1002.

Lee, T. N., and E. Williams. 1999. Mean distribution and seasonal variability of coastal currents and temperature in the Florida Keys with implications for larval recruitment. Bulletin of Marine Science 64:35–56.

Legler, N. D., T. B. Johnson, D. D. Heath, and S. A. Ludsin. 2010. Water temperature and prey size effects on the rate of digestion of larval and early juvenile fish. Transactions of the American Fisheries Society 139:868–875. de León, R., K. Vane, P. Bertuol, V. Chamberland, F. Simal, E. Imms, and M. Vermeij. 2013. Effectiveness of lionfish removal efforts in the southern Caribbean. Endangered Species Research 22:175–182.

Lesser, M. P., and M. Slattery. 2011. Phase shift to algal dominated communities at mesophotic depths associated with lionfish (Pterois volitans) invasion on a Bahamian coral reef. Biological Invasions 13:1855–1868.

Levine, J. M. 2000. Species diversity and biological invasions: Relating local process to community pattern. Science 288:852–854.

Lindeman, K. C., R. Pugliese, G. T. Waugh, and J. S. Ault. 2000. Developmental patterns within a multispecies reef fishery: Management applications for essential fish habitats and protected areas. Bulletin of Marine Science 66:929–956.

Lingo, M. E., and S. T. Szedlmayer. 2006. The influence of habitat complexity on reef fish communities in the northeastern Gulf of Mexico. Environmental Biology of Fishes 76:71–80.

Lockwood, J. L., M. F. Hoopes, and M. P. Marchetti. 2007. Invasion Ecology. Wiley- Blackwell, West Sussex.

227

Lodge, D. M. 1993. Biological invasions: Lessons for ecology. Trends in Ecology and Evolution 8:133–137.

Lorenzen, K. 1996. The relationship between body weight and natural mortality in juvenile and adult fish: A comparison of natural ecosystems and aquaculture. Journal of Fish Biology 49:627–647.

Lorenzen, K., and K. Enberg. 2002. Density-dependent growth as a key mechanism in the regulation of fish populations: Evidence from among-population comparisons. Proceedings of the Royal Society of London B: Biological Sciences 269:49–54.

Mack, R. N., D. Simberloff, W. M. Lonsdale, H. Evans, M. Clout, and F. A. Bazzaz. 2000. Biotic invasions: Causes, epidemiology, global consequences, and control. Ecological Applications 10:689–710.

Maljković, A., T. E. Van Leeuwen, and S. N. Cove. 2008. Predation on the invasive red lionfish, Pterois volitans (Pisces: Scorpaenidae), by native groupers in the Bahamas. Coral Reefs 27:501–501.

McCawley, J. R., and J. H. Cowan. 2007. Seasonal and size specific diet and prey demand of red snapper on Alabama artificial reefs. Red Snapper Ecology and Fisheries in the U.S. Gulf of Mexico 60:77–104.

McDonald, P. S., G. C. Jensen, and D. A. Armstrong. 2001. The competitive and predatory impacts of the nonindigenous crab Carcinus maenas (L.) on early benthic phase Dungeness crab Cancer magister Dana. Journal of Experimental Marine Biology and Ecology 258:39–54.

Meirmans, P. G., and P. H. Van Tienderen. 2004. GENOTYPE and GENODIVE: Two programs for the analysis of genetic diversity of asexual organisms. Molecular Ecology Notes 4:792–794.

La Mesa, M., G. Scarcella, F. Grati, and G. Fabi. 2010. Age and growth of the black scorpionfish, Scorpaena porcus (Pisces: Scorpaenidae) from artificial structures and natural reefs in the Adriatic Sea. Scientia Marina 74:677–685.

Meusnier, I., G. A. Singer, J.-F. Landry, D. A. Hickey, P. D. Hebert, and M. Hajibabaei. 2008. A universal DNA mini-barcode for biodiversity analysis. BMC Genomics 9:214.

Molnar, J. L., R. L. Gamboa, C. Revenga, and M. D. Spalding. 2008. Assessing the global threat of invasive species to marine biodiversity. Frontiers in Ecology and the Environment 6:485–492.

Moncreiff, C. A., and M. J. Sullivan. 2001. Trophic importance of epiphytic algae in subtropical seagrass beds: Evidence from multiple stable isotope analyses. Marine Ecology Progress Series 215:93–106.

228

Mooney, H. A., and E. E. Cleland. 2001. The evolutionary impact of invasive species. Proceedings of the National Academy of Sciences 98:5446–5451.

Moran, Z., D. J. Orth, J. D. Schmitt, E. M. Hallerman, and R. Aguilar. 2015. Effectiveness of DNA barcoding for identifying piscine prey items in stomach contents of piscivorous catfishes. Environmental Biology of Fishes 99:161–167.

Morris, J. 2009. The biology and ecology of invasive Indo-Pacific lionfish. North Carolina State University.

Morris, J. A., and J. L. Akins. 2009. Feeding ecology of invasive lionfish (Pterois volitans) in the Bahamian archipelago. Environmental Biology of Fishes 86:389– 398.

Morris, J. A., K. W. Shertzer, and J. A. Rice. 2011a. A stage-based matrix population model of invasive lionfish with implications for control. Biological Invasions 13:7– 12.

Morris, J. A., C. V. Sullivan, and J. J. Govoni. 2011b. Oogenesis and spawn formation in the invasive lionfish, Pterois miles and Pterois volitans. Scientia Marina 75:147– 154.

Morris, J. A., A. Thomas, A. L. Rhyne, N. Breen, L. Akins, and B. Nash. 2011c. Nutritional properties of the invasive lionfish: A delicious and nutritious approach for controlling the invasion. AACL Bioflux 4:21–26.

Morris, J. A., and P. Whitfield. 2009. Biology, ecology, control and management of the invasive Indo-Pacific lionfish: An updated integrated assessment. NOAA Technical Memorandum NOS NCCOS 99:57.

Morris, J., J. Akins, and A. Barse. 2009. Biology and ecology of the invasive lionfishes, Pterois miles and Pterois volitans. Annual Proceedings of the Gulf and Caribbean Fisheries Institute 61:1–6.

Morte, S., M. J. Redon, and A. Sanz-Brau. 2001. Diet of Scorpaena porcus and Scorpaena notata (Pisces : Scorpaenidae) in the western Mediterranean. Cahiers de Biologie Marineiologie marine 42:333–344.

Moyle, P. B., and T. Light. 1996. Biological invasions of fresh water: Empirical rules and assembly theory. Biological Conservation 78:149–161.

Moyle, P. B., and M. P. Marchetti. 2006. Predicting invasion success: Freshwater fishes in California as a model. BioScience 56:515.

Mumby, P. J., A. R. Harborne, and D. R. Brumbaugh. 2011. Grouper as a natural biocontrol of invasive lionfish. PLoS ONE 6:e21510.

229

Muñoz, R. C., C. A. Currin, and P. E. Whitfield. 2011. Diet of invasive lionfish on hard bottom reefs of the Southeast USA: Insights from stomach contents and stable isotopes. Marine Ecology Progress Series 432:181–193.

Murawski, S. A., W. T. Hogarth, E. B. Peebles, and L. Barbeiri. 2014. Prevalence of external skin lesions and polycyclic aromatic hydrocarbon concentrations in Gulf of Mexico fishes, post-Deepwater Horizon. Transactions of the American Fisheries Society 143:1084–1097.

Nakagawa, S., and H. Schielzeth. 2013. A general and simple method for obtaining R 2 from generalized linear mixed-effects models. Methods in Ecology and Evolution 4:133–142.

Nei, M. 1987. Molecular evolutionary genetics. Page Tempe AZ Arizona State University. Columbia University Press, New York.

NOAA-NRDA. 2015. Chapter 4. Injury to natural resources. In: Final Programmatic Damage Assessment and Restoration Plan and Final Programmatic Environmental Im - pact Statement. Silver Springs, MD.

Norberg, M. 2015. The ecology of tomtate, Haemulon aurolineatum, in the northern Gulf of Mexico and effects of the Deepwater Horizon oil spill. University of South Alabama.

Nuttall, M. 2014. Lionfish (Pterois volitans [Linnaeus, 1758] and P. miles [Bennett, 1828]) records within mesophotic depth ranges on natural banks in the northwestern Gulf of Mexico. BioInvasions Records 3:111–115.

O’Rorke, R., S. Lavery, S. Chow, H. Takeyama, P. Tsai, L. E. Beckley, P. A. Thompson, A. M. Waite, and A. G. Jeffs. 2012. Determining the diet of larvae of western Rock Lobster (Panulirus cygnus) using high-throughput DNA sequencing techniques. PLoS ONE 7:e42757.

Olden, J. D., N. LeRoy Poff, M. R. Douglas, M. E. Douglas, and K. D. Fausch. 2004. Ecological and evolutionary consequences of biotic homogenization. Trends in Ecology & Evolution 19:18–24.

Oliveira, M. L., and J. M. B. Duarte. 2013. Amplifiability of mitochondrial, microsatellite and amelogenin DNA loci from fecal samples of red brocket deer Mazama americana (Cetartiodactyla, Cervidae). Genetics and Molecular Research 12:44– 52.

Van Oosterhout, C., W. F. Huthinson, D. P. M. Wills, and P. Shipley. 2004. Micro- checker: Software for identifying and correcting genotyping errors in microsatellite data. Molecular Ecology Notes 4:535–538.

Paine, R. T. 1966. Food web complexity and species diversity. The American Naturalist 100:65–75.

230

Patterson, W.F., J.P. Chanton, D.J. Hollander, E.A. Goddard, B.K. Barnett, and J.T. Tarnecki. 2019. The utility of stable and radio isotopes in fish tissues as biogeochemical tracers of marine oil spill food web effects. In: S.A. Murawski, C. Ainsworth, S. Gilbert, D. Hollander, C.B. Paris, T. Schulter, and D. Wetzel, Eds., Scenarios and Responses to Future Deep Oil Spills –Fighting the Next War. Springer, New York.

Patterson, W. F., M. A. Dance, and D. T. Addis. 2009. Development of a remotely operated vehicle based methodology to estimate fish community structure at artificial reef sites in the northern Gulf of Mexico. Proceedings of the 61st Gulf and Caibbean Fisheries Institute 61:263–270.

Patterson, W. F., J. H. Tarnecki, D. T. Addis, and L. R. Barbieri. 2014. Reef fish community structure at natural versus artificial reefs in the northern Gulf of Mexico. Proceedings of the 66th Gulf and Caribbean Fisheries Institute 66:4–8.

Patterson, W. F., C. A. Wilson, S. J. Bentley, J. H. Cowan Jr., T. Henwood, Y. C. Allen, and T. A. Dufrene. 2005. Delineating juvenile red snapper habitat on the Northern Gulf of Mexico Continental Shelf. American Fisheries Society Symposium 41:277–288.

Pereira, L. S., A. A. Agostinho, and K. O. Winemiller. 2017. Revisiting cannibalism in fishes. Reviews in Fish Biology and Fisheries 27:499–513.

Perkol-Finkel, S., N. Shashar, and Y. Benayahu. 2006. Can artificial reefs mimic natural reef communities? The roles of structural features and age. Marine Environmental Research 61:121–135.

Persson, L., P. Byström, and E. Wahlström. 2000. Cannibalism and competition in Eurasian perch: Population dynamics of an ontogenetic omnivore. Ecology 81:1058–1071.

Pimentel, D., R. Zuniga, and D. Morrison. 2005. Update on the environmental and economic costs associated with alien-invasive species in the United States. Ecological Economics 52:273–288.

Pinhiero, J., D. Bates, S. DebRoy, and D. Sarkar. 2016. Nlme: linear and nonlinear mixed effects models. R Core Team.

Polis, G. A. 1981. The evolution and dynamics of intraspecific predation. Annual Review of Ecology and Systematics 12:225–251.

Post, D. M. 2002. Using stable isotopes to estimate trophic position: Models, methods, and assumptions. Ecology 83:703.

Post, D. M., C. a Layman, D. A. Arrington, G. Takimoto, J. Quattrochi, and C. G. Montaña. 2007. Getting to the fat of the matter: Models, methods and

231

assumptions for dealing with lipids in stable isotope analyses. Oecologia 152:179–189.

Post, J. R., E. A. Parkinson, and N. T. Johnston. 1999. Density-dependent processes in structured fish populations: Interaction strengths in whole-lake experiments. Ecological Monographs 69:155–175.

Potts, J. C., D. Berrane, and J. A. Morris, Jr. 2010. Age and growth of lionfish from the Western North Atlantic. Annual Proceedings of the Gulf and Caribbean Fisheries Institute 63:314.

Primmer, C. R., M. T. Koskinen, and J. Piironen. 2000. The one that did not get away: Individual assignment using microsatellite data detects a case of fishing competition fraud. Proceedings of the Royal Society of London. Series B: Biological Sciences 267:1699–1704.

Pusack, T. J., C. E. Benkwitt, K. Cure, and T. L. Kindinger. 2016. Invasive red lionfish (Pterois volitans) grow faster in the Atlantic Ocean than in their native Pacific range. Environmental Biology of Fishes 99:571–579.

R Development Core Team. 2016. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing.

Ratnasingham, S., and P. Hebert. 2007. BOLD: The Barcode of Life Data System (http://www.barcodinglife.org). Molecular Ecology Notes 7:355–364.

Raymond, W. W., M. a. Albins, and T. J. Pusack. 2015. Competitive interactions for shelter between invasive Pacific red lionfish and native Nassau grouper. Environmental Biology of Fishes 98:57–65.

Renshaw, M. A., M. Giresi, and J. O. Adams. 2013. Microsatellite fragment analysis using the ABI Prism® 377 DNA sequencer. Methods in Molecular Biology 1006:181–196.

Ricker, W. E. 1954. Stock and recruitment. Journal of the Fisheries Research Board of Canada 11:559–623.

Rilov, G. 2009. Predator-prey interactions of marine invaders. Pages 261–285 Biological Invasions in Marine Ecosystems. Springer Berlin Heidelberg, Berlin, Heidelberg.

Robertson, A., A. Garcia, H. Quintana, T. Smith, B. II, K. Reale-Munroe, J. Gulli, D. Olsen, J. Hooe-Rollman, E. Jester, B. Klimek, and S. Plakas. 2013. Invasive lionfish (Pterois volitans): A potential human health threat for ciguatera fish poisoning in tropical waters. Marine Drugs 12:88–97.

232

Rose, K. A., J. H. Cowan, K. O. Winemiller, R. A. Myers, and R. Hilborn. 2001. Compensatory density dependence in fish populations: Importance, controversy, understanding and prognosis. Fish and Fisheries 2:293–327.

Rudolf, V. H. W. 2008. Impact of cannibalism on predator prey dynamics: Size- structured interactions and apparent mutualism. Ecology 89:1650–1660.

Ruiz, G. M., J. T. Carlton, E. D. Grosholz, and A. H. Hines. 1997. Global invasions of marine and estuarine habitats by non-indigenous species: Mechanisms, extent, and consequences. American Zoologist 37:621–632.

Ruttenberg, B. I., P. J. Schofield, J. L. Akins, A. Acosta, M. W. Feeley, J. Blondeau, S. G. Smith, and J. S. Ault. 2012. Rapid invasion of Indo-Pacific lionfishes (Pterois volitans and Pterois miles ) in the Florida Keys, USA: Evidence from multiple pre- and post-invasion data sets. Bulletin of Marine Science 88:1051–1059.

Sakai, A. K., F. W. Allendorf, J. S. Holt, D. M. Lodge, J. Molofsky, K. A. With, S. Baughman, R. J. Cabin, J. E. Cohen, N. C. Ellstrand, D. E. McCauley, P. O’Neil, I. M. Parker, J. N. Thompson, and S. G. Weller. 2001. The population biology of invasive species. Annual Review of Ecology and Systematics 32:305–332.

Scarborough Bull, A., and J. Kendall. 1994. An indication of the process: Offshore platforms as artificial reefs in the Gulf of Mexico. Bulletin of Marine Science 55:1086–1098.

Schmider, E., M. Ziegler, E. Danay, L. Beyer, and M. Bühner. 2010. Is it really robust? Methodology 6:147–151.

Schneider, P. M., K. Bender, W. R. Mayr, W. Parson, B. Hoste, R. Decorte, J. Cordonnier, D. Vanek, N. Morling, M. Karjalainen, C. M. P. Carlotti, M. Sabatier, C. Hohoff, H. Schmitter, W. Pflug, R. Wenzel, D. Patzelt, R. Lessig, P. Dobrowolski, G. O’Donnell, L. Garafano, M. Dobosz, P. De Knijff, B. Mevag, R. Pawlowski, L. Gusmão, M. C. Vide, A. A. Alonso, O. G. Fernández, P. S. Nicolás, A. Kihlgreen, W. Bär, V. Meier, A. Teyssier, R. Coquoz, C. Brandt, U. Germann, P. Gill, J. Hallett, and M. Greenhalgh. 2004. STR analysis of artificially degraded DNA - Results of a collaborative European exercise. Forensic Science International 139:123–134.

Schofield, P. 2010. Update on geographic spread of invasive lionfishes (Pterois volitans [Linnaeus, 1758] and P. miles [Bennett, 1828]) in the Western North Atlantic Ocean, Caribbean Sea and Gulf of Mexico. Aquatic Invasions 5:S117–S122.

Schofield, P. J. 2009. Geographic extent and chronology of the invasion of non-native lionfish (Pterois volitans [Linnaeus 1758] and P. miles [Bennett 1828]) in the Western North Atlantic and Caribbean Sea. Aquatic Invasions 4:473–479.

Schofield, P. J., J. N. Langston, J. A. Morris Jr., and P. Fuller. 2014. Pterois volitans/miles Fact Sheet. https://dx.doi.org/10.3133/fs20143032.

233

Schooley, J. D., A. P. Karam, B. R. Kesner, P. C. Marsh, C. A. Pacey, and D. J. Thornbrugh. 2008. Detection of larval remains after consumption by fishes. Transactions of the American Fisheries Society 137:1044–1049.

Schultz, T. F., C. K. Fitzpatrick, D. Wilson Freshwater, and J. A. Morris. 2013. Characterization of 18 polymorphic microsatellite loci from invasive lionfish (Pterois volitans and P. miles). Conservation Genetics Resources 5:599–601.

Semmens, B., E. Buhle, A. Salomon, and C. Pattengill-Semmens. 2004. A hotspot of non-native marine fishes: Evidence for the aquarium trade as an invasion pathway. Marine Ecology Progress Series 266:239–244.

Shaw, P. W., G. J. Pierce, and P. R. Boyle. 1999. Subtle population structuring within a highly vagile marine invertebrate, the veined squid Loligo forbesi, demonstrated with microsatellite DNA markers. Molecular Ecology 8:407–417.

Sheppard, S. K., and J. D. Harwood. 2005. Advances in molecular ecology: Tracking trophic links through predator-prey food-webs. Functional Ecology 19:751–762.

Sih, A., D. I. Bolnick, B. Luttbeg, J. L. Orrock, S. D. Peacor, L. M. Pintor, E. Preisser, J. S. Rehage, and J. R. Vonesh. 2010. Predator-prey naïveté, antipredator behavior, and the ecology of predator invasions. Oikos 119:610–621.

Sikkel, P. C., L. J. Tuttle, K. Cure, A. M. Coile, and M. A. Hixon. 2014. Low susceptibility of invasive red lionfish (Pterois volitans) to a generalist ectoparasite in both its introduced and native ranges. PLoS ONE 9:e95854.

Simberloff, D., and L. Gibbons. 2004. Now you see them, now you don’t! - Population crashes of established introduced species. Biological Invasions 6:161–172.

Smith, C., and P. Reay. 1991. Cannibalism in teleost fish. Reviews in Fish Biology and Fisheries 1:41–64.

Smith, N. S., and J. B. Shurin. 2010. Artificial structures facilitate lionfish invasion in marginal Atlantic habitats. Proceedings of the 63rd Gulf and Caribbean Fisheries Institute 63:342–344.

Smith, T., and R. L. Smith. 2001. Ecology and field biology. 6th edition. Benjamin Cummings, San Francisco.

Soulsbury, C. D., G. Iossa, K. J. Edwards, P. J. Baker, and S. Harris. 2007. Allelic dropout from a high-quality DNA source. Conservation Genetics 8:733–738.

Sousa, L. L., R. Xavier, V. Costa, N. E. Humphries, C. Trueman, R. Rosa, D. W. Sims, and N. Queiroz. 2016. DNA barcoding identifies a cosmopolitan diet in the ocean sunfish. Scientific Reports 6:28762.

234

Stachowicz, J. J. ., H. Fried, R. W. . Osman, and R. B. . Whitlatch. 2002. Biodiversity, invasion resistance, and marine ecosystem function: Reconciling pattern and process. Ecology 83:2575–2590.

Stevens, P. W., S. L. Fox, and C. L. Montague. 2006. The interplay between mangroves and saltmarshes at the transition between temperate and subtropical climate in Florida. Wetlands Ecology and Management 14:435–444.

Strelcheck, A. J., J. H. Cowan, and A. Shah. 2005. Influence of reef location on artificial- reef fish assemblages in the northcentral Gulf of Mexico. Bulletin of Marine Science 77:425–440.

Sundqvist, A. K., H. Ellegren, and C. Vilà. 2008. Wolf or dog? Genetic identification of predators from saliva collected around bite wounds on prey. Conservation Genetics 9:1275–1279.

Symondson, W. O. C. 2002. Molecular identification of prey in predator diets. Molecular Ecology 11:627–641.

Taberlet, P. 1996. Reliable genotyping of samples with very low DNA quantities using PCR. Nucleic Acids Research 24:3189–3194.

Tamburello, N., and I. M. Côté. 2015. Movement ecology of Indo-Pacific lionfish on Caribbean coral reefs and its implications for invasion dynamics. Biological Invasions 17:1639–1653.

Tanaka, M., T. Goto, M. Tomiyama, and H. Sudo. 1989. Immigration, settlement and mortality of flounder (Paralichthys olivaceus) larvae and juveniles in a nursery ground, Shijiki bay, Japan. Netherlands Journal of Sea Research 24:57–67.

Tarnecki, J. H., and W. F. Patterson. 2015. Changes in red snapper diet and trophic ecology following the Deepwater Horizon Oil Spill. Marine and Coastal Fisheries 7:135–147.

Teletchea, F., J. Bernillon, M. Duffraisse, V. Laudet, and C. Hänni. 2008. Molecular identification of vertebrate species by oligonucleotide microarray in food and forensic samples. Journal of Applied Ecology 45:967–975.

Thompson, M. J., W. W. Schroeder, N. W. Phillips, and B. D. Graham. 1999. Ecology of live bottom habitats of the northeastern Gulf of Mexico : A community profile. U.S. Dept. of the Interior, U.S. Geological Survey, Biological Resources Division, USGS/BRD/CR--l999-000l and Minerals Management Service, Gulf of Mexico OCS Region, New Orleans, LA.

Toledo-Hernández, C., X. Vélez-Zuazo, C. P. Ruiz-Diaz, A. R. Patricio, P. Mège, M. Navarro, A. M. Sabat, R. Betancur-R, and R. Papa. 2014. Population ecology and genetics of the invasive lionfish in Puerto Rico. Aquatic Invasions 9:227–237.

235

Turpin, R., and S. Bortone. 2002. Pre- and post-hurricane assessment of artificial reefs: Evidence for potential use as refugia in a fishery management strategy. ICES Journal of Marine Science 59:S74–S82.

Tuttle, L. J., P. C. Sikkel, K. Cure, and M. A. Hixon. 2017. Parasite-mediated enemy release and low biotic resistance may facilitate invasion of Atlantic coral reefs by Pacific red lionfish (Pterois volitans). Biological Invasions 19:563–575.

Valdez-Moreno, M., C. Quintal-Lizama, R. Gómez-Lozano, and M. del C. García-Rivas. 2012. Monitoring an alien invasion: DNA barcoding and the identification of lionfish and their prey on coral reefs of the Mexican Caribbean. PLoS ONE 7:e36636.

Valentini, A., F. Pompanon, and P. Taberlet. 2009. DNA barcoding for ecologists. Trends in Ecology & Evolution 24:110–117.

Vestheim, H., and S. N. Jarman. 2008. Blocking primers to enhance PCR amplification of rare sequences in mixed samples – A case study on prey DNA in Antarctic krill stomachs. Frontiers in Zoology 5:12.

Villaseñor-Derbez, J. C., and R. Herrera-Pérez. 2014. Brief description of prey selectivity and ontogenetic changes in the diet of the invasive lionfish Pterois volitans (Actinopterygii, Scorpaenidae) in the Mexican Caribbean. Pan-American Journal of Aquatic Sciences 9:131–135.

Ward, R. D., R. Hanner, and P. D. N. Hebert. 2009. The campaign to DNA barcode all fishes, FISH-BOL. Journal of Fish Biology 74:329–356.

Ward, R. D., T. S. Zemlak, B. H. Innes, P. R. Last, and P. D. . Hebert. 2005. DNA barcoding Australia’s fish species. Philosophical Transactions of the Royal Society B: Biological Sciences 360:1847–1857.

Whitehead, A., B. Dubansky, C. Bodinier, T. I. Garcia, S. Miles, C. Pilley, V. Raghunathan, J. L. Roach, N. Walker, R. B. Walter, C. D. Rice, and F. Galvez. 2012. Genomic and physiological footprint of the Deepwater Horizon oil spill on resident marsh fishes. Proceedings of the National Academy of Sciences 109:20298–20302.

Whitfield, P. E., J. A. Hare, A. W. David, S. L. Harter, R. C. Muñoz, and C. M. Addison. 2006. Abundance estimates of the Indo-Pacific lionfish Pterois volitans/miles complex in the Western North Atlantic. Biological Invasions 9:53–64.

Whitfield, P., T. Gardner, S. Vives, M. Gilligan, W. Courtenay Ray, G. Ray, and J. Hare. 2002. Biological invasion of the Indo-Pacific lionfish Pterois volitans along the Atlantic coast of North America. Marine Ecology Progress Series 235:289–297.

Williams, R., S. Gero, L. Bejder, J. Calambokidis, S. D. Kraus, D. Lusseau, A. J. Read, and J. Robbins. 2011. Underestimating the damage: Interpreting cetacean

236

carcass recoveries in the context of the Deepwater Horizon/BP incident. Conservation Letters 4:228–233.

Wilson, K. L., B. G. Matthias, A. B. Barbour, R. N. M. Ahrens, T. Tuten, and M. S. Allen. 2015. Combining samples from multiple gears helps to avoid fishy growth curves. North American Journal of Fisheries Management 35:1121–1131.

Workman, I., A. Shah, D. Foster, and B. Hataway. 2002. Habitat preferences and site fidelity of juvenile red snapper (Lutjanus campechanus). ICES Journal of Marine Science 59:43–50.

Zar, J. H. 2010. Biostatistical Analysis. 5th edition. Prentice Hall, New Jersey, USA.

237

BIOGRAPHICAL SKETCH

Kristen Dahl was born in 1985 in Gainesville, Florida to Robert (Bob) and Rayann

Dahl. At an early age her family moved to the Gulf Coast and eventually settled in

Navarre, Florida. For as long as she can remember, her favorite activities have been centered around nature, and she spent countless days at the beach, canoeing down area rivers, and exploring the ponds around her house. A love of learning and wildlife was fostered by Kristen’s parents and grandparents. She has fond memories of gardening and beekeeping with Mema and Papa, Velma and Ray Collier, in rural central

Florida, as well as hiking and bird watching with her Grandma and Grandpa, May and

Ed Dahl, in Miami, Florida and Beech Mountain, North Carolina. Throughout high school, Kristen was most interested in the sciences, including biology and meteorology.

A marine biology course offered at her high school introduced her to tropical reef environments in the Florida Keys which would have a lasting impact on her.

Kristen graduated from Navarre High School in 2003 and entered into the marine biology program at the University of West Florida in Pensacola, Florida. A directed study by Drs. Phil Darby and Will Patterson at UWF introduced her firsthand to the world of academic research, where she studied stable isotope composition of wild caught and hatchery raised Florida apple snails and published the work in a peer-reviewed journal.

After receiving her bachelor’s degree in marine biology, Kristen worked as a field technician for Dr. Darby in the Florida Everglades, and then for Florida Fish and Wildlife

Conservation Commission as a biologist. After two years performing fisheries surveys and sampling thousands of fish on the boat docks of Destin, Florida, Kristen came to pursue a doctoral degree in marine science at the University of South Alabama and

Dauphin Island Sea Lab under Dr. Will Patterson. Kristen began studying invasive

238

lionfish populations off the Florida Panhandle, and in 2015, she and her advisor were invited to participate in an ecosystem modeling workshop held by the University of

Florida that aimed to simulate the lionfish invasion and the effects of lionfish harvest.

The next year, Kristen learned that her advisor would accept a faculty position at the

University of Florida, and she had to opportunity to transfer and finish out her degree in

Gainesville. During this time, Kristen honed quantitative and statistical research skills, trained as a scientific diver, and explored the rich “Nature Coast” of Florida.

Upon completion of her Doctor of Philosophy degree, Kristen plans to continue working with Dr. Will Patterson in a post-doctoral position. She looks forward to expanding the subject matter of her research, working ultimately towards a position in natural resource management or academia.

239