Louisiana State University LSU Digital Commons

LSU Master's Theses Graduate School

July 2020

Trophic Interactions and Habitat Quality of Invasive Lionfish in the Gulf of Mexico

Hanna Bauer

Follow this and additional works at: https://digitalcommons.lsu.edu/gradschool_theses

Recommended Citation Bauer, Hanna, "Trophic Interactions and Habitat Quality of Invasive Lionfish in the Gulf of Mexico" (2020). LSU Master's Theses. 5185. https://digitalcommons.lsu.edu/gradschool_theses/5185

This Thesis is brought to you for free and open access by the Graduate School at LSU Digital Commons. It has been accepted for inclusion in LSU Master's Theses by an authorized graduate school editor of LSU Digital Commons. For more information, please contact [email protected].

TROPHIC INTERACTIONS AND HABITAT QUALITY OF INVASIVE LIONFISH IN THE GULF OF MEXICO

A Thesis

Submitted to the Graduate Faculty of Louisiana State University and Agricultural and Mechanical College in partial fulfillment of the requirements for the degree of Master of Science

in

Department of Oceanography and Coastal Science

by Hanna Bauer B.S. University of Portland, 2015 August 2020 Acknowledgments

I would like to acknowledge my advisor Dr. Cassandra Glaspie, as well as my committee members Dr. Michael Dance, Dr. Dubravko Justic, and Dr. Karen Maruska. Additionally, Dr. Michael Polito, Dr. Hayat Bennadji, Thomas Blanchard, and Dr. Steve Midway provided input and guidance. I sincerely thank my lab members Neal Kolonay, Jill Tupitza, Carolyn Swize, Emily Shallow, and Taylor Clemment for all their help and support with this project. Finally, I would like to thank Louisiana State University for the opportunity to pursue my education.

ii TABLE OF CONTENTS

Acknowledgments ...... ii

List of Tables ...... iv

List of Figures ...... v

Abstract ...... vi

Chapter 1. Invasive Lionfish Size and Reproductive Capability Across a Depth Gradient in the Florida Keys ...... 7 INTRODUCTION ...... 7 METHODS ...... 12 RESULTS ...... 16 DISCUSSION ...... 18 REFERENCES ...... 25

Chapter 2. Trophic Interactions of Invasive Lionfish Across a Depth in the Florida Keys ...... 31 INTRODUCTION ...... 31 METHODS ...... 34 RESULTS ...... 38 DISCUSSION ...... 46 REFERENCES ...... 54

Chapter 3. Habitat Modeling for Invasive Lionfish in the Gulf of Mexico Using a Bioenergetics-Based Growth Model ...... 59 INTRODUCTION ...... 59 METHODS ...... 62 RESULTS ...... 70 DISCUSSION ...... 76 REFERENCES ...... 84

Conclusion ...... 91

Vita ...... 93

iii LIST OF TABLES

1.1. Frequency of each lionfish sex by season and depth, including the female to male ratio ...... 13

2.1. MANOVA results of δ13C and δ15N values as dependent variables, and depth, season, and the depth-season interaction as independent variables...... 42

2.2. Isotopic niche metrics for assessing species isotopic niches by season and depth. ... 44

2.3. Trophic niche overlap estimates for combined and individual ellipse areas for season-depth groupings ...... 45

3.1. Bioenergetic parameters and variables used ...... 68

3.2. Gamma regression model output of depth, season, sex, length on lionfish muscle caloric density ...... 71

iv LIST OF FIGURES

1.1. Map of sample collection sites near Marathon Key, Florida ...... 13

1.2. Distribution of depths that lionfish sample were collected from in the Florida Keys 14

1.3. Lionfish standard length by sex, where undetermined sex fish are immature, and the depth is indicated by the color...... 16

1.4. Female GSI by season and depth ...... 17

2.1. Teleost families in stomach contents ...... 39

2.2. Percent number of each prey category in deep and shallow lionfish stomachs, both fall and spring seasons combined...... 40

2.3. δ13C and δ15N biplot for all lionfish samples and standard ellipse areas ...... 42

2.4. Mean estimates with standard error for isotope values of the total mean, fall samples, and spring samples ...... 43

2.5. Bayesian estimates of standard ellipse areas (with 95% confidence intervals shown for season-depth groupings...... 45

3.1. Map of natural and artificial structured habitats off the coast of the United States in the Gulf of Mexico...... 66

3.2. Gulf of Mexico regions identified for this study with fall bottom temperatures ...... 70

3.3. Correlation between female lionfish GSI values and caloric density with 95% confidence intervals ...... 72

3.4. Lionfish GRP value estimated at each identified structured habitat for four seasons 73

3.5. Growth Rate Potentials for lionfish by region in the Gulf of Mexico ...... 74

3.6. USGS reported lionfish sightings in the Gulf of Mexico and model GRP output values at identified lionfish habitat ...... 75

v

ABSTRACT

Lionfish (Pterois volitans and Pterois miles) are one of the most successful marine invaders of all time and pose a threat to native species that inhabit coral reefs, as well as overall reef health. Current management efforts in the invaded Atlantic region revolve around spearfishing to remove lionfish, often limited to 30 m depth. There is evidence that lionfish may seek refuge from fishing pressure in deeper habitat and replenish shallow sites, undermining this management strategy. To investigate the ecological implications of deep lionfish, size, reproductive capability, and diet were examined across a depth gradient for lionfish in the

Florida Keys. It was found that size weakly increased with increasing depth and that shallow lionfish had higher reproductive potential. Sex ratios were female-biased in shallow sites.

Lionfish are eating a range of teleosts and decapod species, including those of economic importance, and deeper lionfish are eating at a higher trophic level than shallow fish. These results were used to inform the parameters of a bioenergetic-based growth model for lionfish in the Gulf of Mexico. Prey energy density was modified to account for deeper (>30 m) lionfish’s increased consumption of teleost prey compared to shallow counterparts. Bottom temperatures at the locations of potential structured habitat were used to calculate the growth rate potential

(GRP) of lionfish regionally and seasonally. There are regional differences in potential habitat that are not reflected in recorded lionfish sightings, and seasonal temperature changes do not limit lionfish growth in the GOM. A GRP model can be a useful tool to identify areas of lionfish growth and inform where management efforts can be focused for removals, particularly for lionfish at depth. This model could also be extended outside of the GOM to provide a management tool for the entirety of the lionfish’s invaded range.

vi Chapter 1. Invasive Lionfish Size and Reproductive Capability Across a Depth Gradient in the Florida Keys

1.1. Introduction

Invasive species are of serious concern in marine ecosystems and can have significant ecological and economic impacts (Molnar et al. 2008). Lionfishes, both Red Lionfish Pterois volitans (Linnaeus 1785) and Devil Firefish Pterois miles (Bennett 1828), are extremely successful invasive species that arrived in Atlantic waters in the mid 1980’s, likely from the aquarium trade (Semmens et al. 2004). Native to the Indopacific, lionfish can now be found from

New England to Brazil, at densities far greater than those found in their native range (Darling et al. 2011; Kulbicki et al. 2012). They are members of the Scorpaenidae family, subfamily

Pteroinae. While both Red Lionfish and Devil Firefish are present in the invasive range, Red

Lionfish account for 93% of lionfish collected in the Western Atlantic (Hamner et al. 2007).

Because genetic work is needed to confirm the exact species, and the ecological impact is not distinguishable between the two, hereafter both firefish and red lionfish will be grouped together and referred to as “lionfish” (Freshwater et al. 2009).

Many coral reefs and structured habitats like artificial reefs and oil rigs in the Atlantic are now densely populated by lionfish. Lionfish, like many successful invasive species, have a broad generalist diet, and are voracious predators, allowing them to opportunistically exploit a wide range of prey (Muñoz et al. 2011; Peake et. al 2018; Malpica-Cruz et al. 2019). This appetite has had detrimental effects on local prey populations. On shallow Bahamian coral reefs, lionfish presence has caused decline of up to 69% in fish biomass of 42 prey species (Green et al. 2012), and a decrease in fish recruitment of 49 species by 79% over a five-week duration (Albins and

Hixon 2008). Besides direct predation, lionfish can also negatively affect native populations

1 through resource and habitat competition (O’Farrell et al. 2015; Marshak et al. 2018; Bos et al.

2018; Raymond et al. 2015). Lionfish larvae are now at densities similar to native fish and can reduce available resources to those native species (Sponaugle et al. 2019).

Lionfish are in part successful at establishing themselves in the invaded range due to their reproductive and growth characteristics (Morris and Akins 2009). Invasive lionfish are capable of spawning year-round, and in the Western Atlantic, females can spawn every 2-4 days (Morris and Akins 2009; Gardner et al. 2015; Fogg et al. 2017). Male lionfish reach maturity at 100 mm, while females mature at 166-190 mm (Morris and Akins 2009; Gardner et al. 2015; Fogg et al.

2017). In the Caribbean, lionfish reach maximum length 2-4 times faster than comparable-sized native predators (Edwards et al. 2014). These life history characteristics have allowed lionfish populations to rapidly expand and make controlling the invader difficult.

Efforts to control lionfish populations have largely focused on culling fish using sling hand spears or pole spears, often wielded by recreational divers. Lionfish fishing derbies have been held to reduce local populations; however, these are typically limited to recreational diving limits of 30 m depth (Barbour et al. 2011). These targeted removals can reduce local lionfish biomass and density (de Leon et al. 2013) and are particularly effective at high priority management sites (Akins 2012) and on rugose reefs (Bejarano et al. 2015). Indeed, there may be site-specific effectiveness with lionfish removals: one study in Puerto Rico found lionfish removals resulted in a reduction in lionfish across all sites, although the recovery of lionfish populations varied by location (Harms-Tuohy et al. 2018). Complete lionfish removal may not be needed to see benefits to the reef. Using a predictive model with field validation on Bahamian reefs, it was found that depending on the site, a 25-92% reduction in lionfish density was needed

2 to protect prey species, and partial removals resulted in prey biomass increase similar to that seen in complete removal (Green et al. 2014).

While culling has reduced lionfish numbers locally on some shallow reefs (Harms-Tuohy et al. 2018; de Leon et al. 2013; Cote et al. 2014), the overall effectiveness of this method is still under examination. Removal by SCUBA is often limited to 30 m without additional training, yet lionfish are categorized as depth generalists (Stefanoudis et al. 2019). Lionfish have been found at 55 m in Puerto Rico (Bejarano et al. 2014), 100 m in the Bahamas (Lesser and Slattery 2011),

112 m in the Northern Gulf of Mexico (GOM; Nuttall et al. 2014), and some of the deepest lionfish have been observed at 240 m in Honduras and 300 m in Bermuda (Gress et al. 2017).

Lionfish exist at greater abundances and biomasses on deeper reefs in Bermuda (Stefanoudis et al. 2019) due to warmer water temperatures at depth (Gress et al. 2017, Coates et al. 2013).

Mesophotic reefs are those located at 30-150 m depth and are of interest ecologically as potential thermal refugia for their threatened shallow-water coral species (Lesser et al. 2009).

Mesophotic lionfish may also be consuming more or different prey species than shallow counterparts; there is evidence that mesophotic lionfish consume key herbivores, the removal of which can lead to increased algal cover on reefs and subsequent coral decline (Lesser and

Slattery 2011; Stefanoudis et al. 2019). In Bermuda, mesophotic lionfish distribution appears to be correlated to prey fish density and prey fish biomass, both of which co-varied with seawater temperature (Goodbody-Gringley et al. 2019). Mesophotic sites often are less affected by overfishing and other anthropogenic impacts, so these sites may offer more prey fish as well as a relief from direct spearfishing of lionfish (Tyler et al. 2009). Invasive lionfish may harm the health of these deeper reefs and decrease their resilience to anthropogenic stressors.

3 Lionfish may undergo an ontogenetic migration to deeper waters, similar to patterns observed in other reef species (Eggleston et al. 2004; Appeldoorn 2009; Johnson and Swenarton

2016). Juvenile lionfish settle in shallow habitats such as mangroves, sea grass, and sheltered reefs before moving to deeper reefs (Claydon et al. 2012). Tagging studies suggest potential for lionfish to undergo ontogenetic migrations to deeper reefs but have failed to capture such movement directly (Harms-Tuohy et al. 2018; Tamburello and Cote 2014; Akins et al. 2014).

Other studies that offer evidence for ontogenetic migration have found larger lionfish at depth or a lack of mature fish at culled shallow depths (Andradi-Brown et al. 2017b; Andradi-Brown et al. 2017a; Claydon et al. 2012; Johnson and Swenarton 2016). Conversely, trawl surveys suggest that lionfish in the eastern GOM first colonized deeper sites (> 30m) before moving to shallower areas (Switzer et al. 2015).

An ontogenetic migration could be the result of a natural shift to preferred habitat in mature fish, the result of fishing pressure driving larger fish to take refuge in deeper waters, or some combination of the two. Lionfish occur at similar densities on shallow and mesophotic reefs in their native range as they do in the Atlantic, despite a lack of fishing pressure in the native range (Andradi-Brown et al. 2017b). A study comparing a culled shallow (< 30 m) site and an adjacent deep (> 30 m) unculled site reported that mesophotic lionfish were larger, females had a larger proportional mass of reproductive tissue (gonadal mass), and deeper populations had a higher portion of actively spawning females (25%; Andradi-Brown et al.

2017a). This may suggest that mature lionfish prefer mesophotic reefs. Yet there is also evidence that mesophotic reefs offer relief from fishing pressure. The Andradi-Brown et al. (2017a) study compared alert distances between fish from culled shallow reefs, adjacent unculled deep reefs, and unculled shallow reefs. Alert distance was estimated as the distance at which a fish responds

4 to an approaching diver. Unculled mesophotic sites adjacent to culled shallow sites had the same alert distances as their shallow counterparts, while the fish on unculled shallow reefs had shorter alert distances and did not fear the “spearfisher” (Andradi-Brown et al. 2017a). In the Bahamas, lionfish hid deeper on the reefs at times of day when culling typically occurred (Côte et al.

2014). This suggests that larger lionfish migrating to deeper mesophotic reefs to take refuge from adjacent culling pressure, although further investigation and replication is necessary.

Regardless of the drivers of ontogenetic migration, if deeper lionfish are more reproductively active, they can potentially repopulate adjacent shallow sites and undermine culling efforts. There may be evidence that mesophotic lionfish are already repopulating other areas: a biophysical model found that lionfish at 50-300 m depth in the Campeche Bank

(Southern GOM) were the major source of lionfish larvae to the Northern GOM through the loop current (Johnston and Bernard 2017). These reported larvae must remain in the water column a minimum of ten days, with 45-50 days providing maximum connectivity; compare this to previous estimates that lionfish larvae settle between 20 and 35 days, with a mean of 26.2

(Johnston and Bernard 2017; Ahrenholz and Morris 2010).

Fishing pressure has been known to influence fish behavior by selecting specific manners

(Sbragaglia et al. 2018; Januchowski-Hartley et al. 2012; Uusi-Heikkila et al. 2008). This response to fishing may be very refined, as five fish taxa responded differently to snorkelers dressed as “spearfishers'' carrying poles in dark equipment than “snorkelers” in bright colored masks and fins at in the Mediterranean (Sbragaglia et al. 2018). Lionfish in several invaded locations exhibit higher alert distances with increased fishing pressure (Andradi-Brown et al.

2017a; Côte et al. 2014). In addition, Côte et al. (2014) found that lionfish demonstrated higher

5 levels of cryptic behavior during the daytime, when cullings typically take place, compared to dawn.

Additionally, fishing pressure can influence fish population structure and life history characteristics. Increasing fishing pressure can cause a decrease in fish body size and has been associated with maturity at a younger size (Lindfield 2014; Berkeley et al., 2004; Dulvy et al.,

2004; Worm et al. 2009; Rochet 1998). Studies have found that lionfish at culled sites are smaller than those at unculled sites (De Leon et al. 2011; Frazer et al 2012; Andradi-Brown et al.

2017a). In northeastern Florida, where spearfishing pressure is high, older fish are virtually absent (Johnson and Swenarton 2016).

The objectives of this study were to investigate how differences in deep and shallow- dwelling population size structure and reproductive output could inform the potential for ontogenetic migration and fishing-induced changes for lionfish in the Florida Keys. Lionfish from shallow (< 30 m) and deep (> 30 m) habitats were compared in terms of body size and gonad characteristics. Based on previous findings, it was expected that shallow fish would be smaller and have less reproductive mass than deeper fish. Such work has the potential to comment on the effectiveness of shallow culling as a management option for the invasive species.

1.2. Methods

1.2.1. Sample Collection

Seventy-four lionfish specimens were collected from September 2018 to June 2019 near

Marathon Key, Florida (Figure 1.1). Fish were collected from patch reefs of predominantly hard corals with some octocorals and sponges. The depths of collections sites ranged from 13.7 m to

6 57.9 m and were separated into “shallow <30 m” (n=35) and “deep >30 m” (n=39) categories based on the recreational diving depth limit and the distribution of sample depths (Figure 1.2).

Forty-five of these fish were caught using a pole-spear, both in deep and shallow waters, by recreational divers. The remaining 29 fish were captured in passive traps deployed from 34-58 m depth by Florida Fish and Wildlife Conservation Commission. Due to the opportunistic collection of samples, there were no shallow lionfish caught in traps. To account for the temporal variation, a “season” factor was used to split between 21 samples collected September 2018 using pole spears (“fall”), and 53 samples collected between April and June 2019, 30 of which were caught in traps and 23 with pole spears (“spring”) to capture any seasonal variability. All samples were frozen as soon as possible after capture. Two samples were not dissected due to poor condition, only the mass and length were recorded.

Figure 1.1. Map of sample collection sites near Marathon Key, Florida. Fall samples were collected in September 2018, and spring samples were collected in April-June 2019.

7

Figure 1.2. Distribution of depths that lionfish sample were collected from in the Florida Keys.

1.2.2. Dissection

Dissections were performed following the NOAA guide (Green et al. 2012). Standard length (mm) and mass (wet weight, g) were recorded, along with sex and gonad tissue mass (wet weight, g). Freezing gonads has no effect compared on mass and thus no correction was applied

(Fogg et al. 2015). The same individual processed all samples.

Size and reproductive capabilities were used for analysis. Gonadosomatic Index (GSI) is the gonad mass proportional to the body weight of the fish and was selected to represent the reproductive capacity of a fish. It is calculated by dividing the gonad mass by the total mass, then

8 multiplying by 100 for a percentage (Anderson and Gutreuter 1983).

1.2.3. Statistical Analysis

All statistical analysis was done using R program (R Core Team 2020). To evaluate the effects of season, sex, and depth on lionfish length, a three-way Analysis of Variance (ANOVA) was used with log-transformed lionfish standard length (SL) as the dependent variable. Depth (2 levels: deeper than 30m, shallower than 30 m), sex (3 levels: male, female, undetermined), and season (2 levels: fall and spring) were independent variables, with all interactive effects included. Type II sums of squares were used, and p-values were reported. Residuals were visually inspected for homoscedasticity.

As GSI values are proportional data, a beta regression was used with the R package

“betareg” (Cribari-Neto and Zeileis 2010). Only female GSI values were used for analysis, as there is little variation in male and immature gonad weight. The regression model included the same independent predictors used for the length model (sex, season, depth, and interactions), and female GSI values as the independent variable. Significance was detected using p-values from

ANOVA analysis, and least squares means were used to examine the effects within and between season and depth categories using the Lsmeans R package (Russell 2016). Residuals were visually inspected for homoscedasticity.

The frequency of occurrence of males and females in the samples was analyzed using a separate Chi-Squared test for each season. Chi and p-values were reported.

9 1.3. Results

1.3.1. Lionfish size

Male lionfish had a mean standard length of 233 ± 6 mm, females 198 ± 6 mm, and undetermined sex fish 157 ± 7 mm. Shallow lionfish had a mean standard length of 203 ± 7 mm compared to deep lionfish with a mean standard length of 216 ± 6 mm. Sex was a significant predictor of lionfish length (Figure 3, F2,60 = 14.7, p-value <0.0001), and there was a tendency for shallow lionfish to be smaller than deep counterparts (Figure 1.3, F1,60 = 3.13, p-value =

0.08). Season and interactive effects had no significance on lionfish size (p >0.05).

Figure 1.3. Lionfish standard length (mm) by sex, where undetermined sex fish are immature, and the depth is indicated by the color (shallow is less than 30 m depth, deep is greater than 30 m depth).

10 1.3.2. GSI values

Beta regression results showed that depth and season significantly affected female GSI values (p

= 0.03 and < 0.0001, respectively). Shallow female lionfish had a larger mean GSI value (4.43 ±

0.78) than deep female lionfish (1.92 ± 0.54), while fall lionfish had larger GSI values (4.75 ±

0.71) than spring lionfish (1.03 ± 0.72; Figure 1.4). Further analysis showed that GSI values were significantly larger for shallow females in the fall samples only (p= 0.0001), and there was no difference between spring shallow and deep GSI values (p=0.94). The interactive effect of season and depth was not significant (p = 0.07).

Figure 1.4. Female lionfish Gonadosomatic Index (GSI) by season, and the depth is indicated by the color (shallow is less than 30 m depth, deep is greater than 30 m depth).

11 1.3.3. Sex ratios

The frequencies of male, female, and undetermined fish for each season and depth category are reported in Table 1.1. Across all 74 samples, there were 37 females, 30 males, and 7 fish of undetermined sex with a ratio of 1.2 females to males. Fall samples had significantly more females than males in shallow fish (p = 0.03, X2 = 6.11). There were no significant differences in sex frequencies between depths for spring samples (p-value = 0.58, X2 = 0.40).

Table 1.1. Frequency of each lionfish sex by season and depth, including the female to male ratio.

Season Depth Females Males Undetermined F:M Ratio Fall Shallow 11 0 0 1:0 Fall Deep 5 4 1 1.3:1 Spring Shallow 10 10 4 1:1 Spring Deep 11 16 3 0.7:1 Total 37 30 7 1.2:1

1.4. Discussion

1.4.1. Lionfish size

Lionfish size varied by sex, where male lionfish were larger than female fish. These results agree with previous studies that report sexual size dimorphism in lionfish (Fogg et al.

2019; Fogg et al. 2017; Edwards et al. 2014; Dahl et al. 2019). This is common in other fish species as well and is likely due to the energy females must divert from allometric growth to reproductive output (Reznick 1983). Female lionfish in the northern GOM may be particularly fecund compared to other areas in the invaded range, producing over 2 million eggs a year (Fogg et al. 2017). The large amount of energy required for such reproductive feats is subsequently not

12 available for growth and can drive differences in somatic growth between the sexes. There was no seasonal difference in lionfish length.

This weak trend of larger lionfish at deeper reefs in the Florida Keys could suggest a natural ontogenetic migration to preferred habitat, or a behavioral response to seek refuge from spearfishing on deeper reefs. Many reef fish undergo ontogenetic migrations from near-shore, shallow areas that offer reprieve from predation to deeper sites with better habitat (Slattery et al.

2011). Habitat surveys and suitability modeling could be conducted in the area to assess which locations provide more favorable conditions for lionfish. Alternatively, fishing pressure can modify fish behavior and could pressure mature lionfish to seek refuge at deeper sites

(Sbragaglia et al. 2018; Januchowski-Hartley et al. 2012; Uusi-Heikkila et al. 2008; Andradi-

Brown 2017a; Côte et al. 2014). There is certainly pressure on lionfish in the Keys: 1,893 lionfish were removed in organized derbies in 2018 and 1,992 were removed in 2019 (Division of Marine Fisheries Division, Florida Fish and Wildlife Conservation Commission). This is in addition to year-round recreational spearfishing by locals and dive shops. Further testing of this hypothesis could involve broad surveys of lionfish abundance and response to divers across depths, and fish tagging studies.

Size appeared to differ between deep and shallow sites but did not reach statistical significance (p = 0.08). There is some evidence that lionfish in the Keys are larger at depth, but the results lack full statistical significance. It is certainly a weaker trend than seen in several previous findings, including a meta-analysis that showed that mesophotic lionfish were larger than shallow counterparts in three of the five countries in the Atlantic (Andradi-Brown et al.

2017b). The differences observed in that study were affected by time since invasion and presence or absence of culling. A study by Andradi-Brown et al. directly compared adjacent shallow and

13 mesophotic reefs systems and found larger fish, in terms of both length and weight, at mesophotic depths in Utila, Honduras (2017a). Time since invasion does not explain the discrepancy between the Keys and Utila, as lionfish appeared in both Utila and the Florida Keys in 2009 (Schofield 2010; Reef Environmental Education Foundation 2009). Lionfish culling occurs at both locations as well, including organized derbies. It is possible that the intensity of the culling varies between the two popular recreational dive locations, as Honduras requires a license to spear lionfish and Florida does not.

Notably, our samples did not capture as many small lionfish as Andradi-Brown et al.

(2017a) did across all depth ranges, most likely due to sampling method bias. The samples used in this study were largely collected by recreational divers, not in the context of a scientific study.

It could be possible that shallow sites do in fact have smaller fish that were not reported due to this bias. The samples used in this study reported a narrower range of total lengths than previous studies with larger sample sizes. Here, males had a TL range of 191-371 mm, females 176-261 mm, and undetermined sex 189-253 mm, compared to previous work for the southeast region of the GOM (n= 6152) where males ranged 148-434 mm, females 86-353 mm, and unsexed or unknown sex 62-426 mm (Fogg 2013). The samples used in this study do not include smaller individual fish, and it is unsurprising if our sample population did not accurately capture smaller lionfish present. Larger lionfish are easier to detect and are speared at a higher rate and the majority of the samples we received were from recreational spearfishermen (Green et al. 2013).

Still, exploratory analysis found no size differences between the speared samples, n=45, and the deep trap sample, n=29 (analyzed using a linear model and ANOVA, p =0.12). The traps may be biased as well: if not deployed on favorable habitat the traps may select only the larger, bolder

14 fish that can survive in less favorable conditions. Future work could focus on trap development to minimize any size bias, and train recreational divers to find and target smaller lionfish.

It seems likely that an emerging fishery area like the Florida Keys would have a significant number of smaller lionfish present. Size and age structure of a population can be truncated by fishing pressure (Lindfield 2014; Berkeley et al. 2004; Dulvy et al. 2004). The first derby was held in the Keys in 2010 and removed 664 lionfish. Since then, large scale derbies have occurred multiple times a year in addition to unreported removals by recreational spearfishers. A larger, geographically broader survey of lionfish populations in the Keys is needed to explore the effects of spearfishing on lionfish population structure.

1.4.2. GSI values

There were seasonal and depth-related differences between female lionfish GSI values.

Fall lionfish had larger GSI values than spring samples, and shallow lionfish had larger GSI values than deeper fish. The differences seen in GSI values between fall and spring samples are likely due to seasonal differences in spawning. Female lionfish in the northern GOM are capable of spawning year-round with elevated GSI values in May-October, peaking in August (Fogg et al. 2017). Spring samples included samples from April-June 2019 and captured lionfish outside of the elevated GSI season. GSI values can be sensitive to small changes in temperature and are elevated at sea surface temperature above 22 ℃ (Gardner et al. 2015; Fogg et al. 2017).

The difference in GSI is also contrary to results in Utila that found that mesophotic female lionfish (40-72 m depth) had larger gonad to body weight ratios than shallow (0-25 m depth; Andradi-Brown et al. 2017a). The lionfish in this study came from a narrower depth range

(14-58 m) and could have failed to capture meaningful data at deeper sites.

15 Our results could also be explained by differences in fishing intensity. Predation that selects for larger, mature individuals has been shown to select for a population that matures at an earlier age and has a higher reproductive output (Boukal 2008; Reznick et al. 1990; Farley et al.

2014). Spearfishing could exert such an influence on lionfish, particularly on populations in shallow areas that experience higher fishing pressure. In fact, invasive lionfish have higher reproductive output and mature at smaller sizes than in their native range, potential evidence that selection for faster growth and higher reproductive investment are underway (Fogg et al. 2017).

In the northern GOM specifically, lionfish reached 50% maturity at smaller sizes relative to other invaded areas (Fogg et al. 2017). Directly comparing the age of maturity between culled and unculled sites alongside culling intensity data could inform such speculation.

Still, some speculate that this increased reproductive output in lionfish is the result of higher densities in the invaded range (Donaldson et al. 2011). P. volitans engage in courtship before spawning, where males lead the females to the surface to spawn, thus a higher population density would offer more opportunities to encounter a mate and increase overall recruitment

(Fishelson 1975). Perhaps regional differences in lionfish density could partially drive reproductive output through this mechanism.

Diet can also influence fish reproductive output (Luquet and Watanabe 1986). In the northern GOM peak months of spawning frequency correlated to recruitment of potential juvenile prey species to structured habitat, thus seasonal prey availability could impact temporal comparisons, or differences in structured habitat available in deep and shallow sites (Fogg et al.

2017).

Unlike Andradi-Brown et al. (2017a), this analysis did not take into account reproductive phases of the gonads, although the reproductive phase was macroscopically observed and

16 reported. Previous studies have shown that macroscopic identification is only accurate for spawning capable females, not other reproductive phases (Fogg et al. 2017). Further microscopic identification of the reproductive phase could further refine the reproductive characteristics of these populations.

1.4.3. Sex ratios

Sex ratios between fall-deep and fall-shallow were disproportional with more females in shallow sites than deep. While the sample size is small (ndeep=10, nshallow=11), it is an interesting piece of support that culling may be changing the population of invasive lionfish in the Florida

Keys: if male lionfish are larger, and spearfishing selectively removes larger fish, then a population under high fishing pressure may shift to higher ratios of females.

Given that shallow female lionfish had larger GSI values, the higher proportion of females on shallow reefs indicates that these may be reproductive hot spots in the area. This is, again, contradictory to findings in other areas that deeper reefs could harbor more reproductively active fish that could repopulate shallow locations (Andradi-Brown et. al 2017a). For lionfish management this is encouraging for the Florida Keys; shallow lionfish are much more accessible for spearfishers than deeper fish. Further studies should investigate differences in fishing pressure, habitat quality, prey availability, and larval connectivity between shallow and mesophotic reefs, as it may be site-specific, and drivers of these discrepancies should be identified.

A previous study in the area with samples from 2012-2015 found a sex ratio of 1.0 : 0.99

M:F, and that it was not significantly different than 1:1 (n = 4527, Fogg et al. 2017). Selection pressure in the Keys area could have altered the population sex structure as size-related selection pressure altered sex ratios in other fisheries (Farley et al. 2014; Halvorsen et al. 2017).

17 In conclusion, lionfish populations in the Florida Keys may be experiencing morphological and reproductive-capacity changes to its population as a result of fishing pressure.

There may be evidence of an ontogenetic migration to deeper reefs in this region, however shallow reefs have higher reproductive potential. Controlling invasive lionfish populations will remain a difficult task for managers but improving the removal of smaller fish in shallow sites and improving collection methods of mesophotic lionfish could improve the situation. Currently there are limited traps in use to capture lionfish although several entities are looking at its design and efficiency, such as Florida Fish and Wildlife Conservation Commission. Such traps could be used to target deeper lionfish populations. Lionfish are likely to remain a conservation issue for years to come and better understanding the population’s migrations, reproductive connectivity, and response to fishing pressure will be crucial to tailoring control efforts regionally.

18 References

Ahrenholz, D. W., & Morris, J. A. (2010). Larval duration of the lionfish, Pterois volitans along the Bahamian Archipelago. Environmental Biology of Fishes, 88(4), 305-309. doi:10.1007/s10641- 010-9647-4

Akins, J. L., Morris, J. A., & Green, S. J. (2014). In situ tagging technique for fishes provides insight into growth and movement of invasive lionfish. Ecology and Evolution, 4(19), 3768-3777. doi:10.1002/ece3.1171

Albins, M. A., & Hixon, M. A. (2008). Invasive Indo-Pacific lionfish Pterois volitans reduce recruitment of Atlantic coral-reef fishes. Marine Ecology Progress Series, 367, 233-238. doi:10.3354/meps07620

Andradi-Brown, D. A., Grey, R., Hendrix, A., Hitchner, D., Hunt, C. L., Gress, E., . . . Exton, D. A. (2017). Depth-dependent effects of culling-do mesophotic lionfish populations undermine current management? Royal Society Open Science, 4(5), 15. doi:10.1098/rsos.170027

Andradi-Brown, D. A., Vermeij, M. J. A., Slattery, M., Lesser, M., Bejarano, I., Appeldoorn, R., . . . Exton, D. A. (2017). Large-scale invasion of western Atlantic mesophotic reefs by lionfish potentially undermines culling-based management. Biological Invasions, 19(3), 939-954. doi:10.1007/s10530-016-1358-0

Appeldoorn, R. S., Aguilar-Perera, A., Bouwmeester, B. L. K., Dennis, G. D., Hill, R. L., Merten, W., . . . Williams, S. J. (2009). Movement of fishes (Grunts: Haemulidae) across the coral reef seascape: A review of scales, patterns and processes. Caribbean Journal of Science, 45(2-3), 304- 316.

Barbour, A. B., Allen, M. S., Frazer, T. K., & Sherman, K. D. (2011). Evaluating the Potential Efficacy of Invasive Lionfish (Pterois volitans) Removals. Plos One, 6(5), 7. doi:10.1371/journal.pone.0019666

Bejarano, S., Lohr, K., Hamilton, S., & Manfrino, C. (2015). Relationships of invasive lionfish with topographic complexity, groupers, and native prey fishes in Little Cayman. Marine Biology, 162(2), 253-266. doi:10.1007/s00227-014-2595-3

Benstead, J. P., March, J. G., Fry, B., Ewel, K. C., & Pringle, C. M. (2006). Testing isosource: Stable isotope analysis of a tropical fishery with diverse organic matter sources. Ecology, 87(2), 326- 333. doi:10.1890/05-0721

Berkeley, S. A., Hixon, M. A., Larson, R. J., & Love, M. S. (2004). Fisheries Sustainability via protection of age structure and spatial distribution of fish populations. Fisheries, 29(8), 23-32. doi:10.1577/1548-8446(2004)29[23:fsvpoa]2.0.co;2

19 Bos, A. R., Grubich, J. R., & Sanad, A. M. (2018). Growth, site fidelity, and grouper interactions of the Red Sea lionfish Pterois miles (Scorpaenidae) in its native habitat. Marine Biology, 165(10). doi:10.1007/s00227-018-3436-6

Boukal, D. S., Berec, L., & Krivan, V. (2008). Does Sex-Selective Predation Stabilize or Destabilize Predator-Prey Dynamics? Plos One, 3(7), 10. doi:10.1371/journal.pone.0002687

Claydon, J. A. B., Calosso, M. C., & Traiger, S. B. (2012). Progression of invasive lionfish in seagrass, mangrove and reef habitats. Marine Ecology Progress Series, 448, 119-129. doi:10.3354/meps09534

Cote, I. M., Darling, E. S., Malpica-Cruz, L., Smith, N. S., Green, S. J., Curtis-Quick, J., & Layman, C. (2014). What Doesn't Kill You Makes You Wary? Effect of Repeated Culling on the Behaviour of an Invasive Predator. Plos One, 9(4), 6. doi:10.1371/journal.pone.0094248

Cribari-Neto F, Zeileis A (2010). “Beta Regression in R.” Journal of Statistical Software, 34(2), 1–24. http://www.jstatsoft.org/v34/i02/

Dahl, K. A., Edwards, M. A., & Patterson, W. F. (2019). Density-dependent condition and growth of invasive lionfish in the northern Gulf of Mexico. Marine Ecology Progress Series, 623, 145-159. doi:10.3354/meps13028

Darling, E. S., Green, S. J., O'Leary, J. K., & Cote, I. M. (2011). Indo-Pacific lionfish are larger and more abundant on invaded reefs: a comparison of Kenyan and Bahamian lionfish populations. Biological Invasions, 13(9), 2045-2051. doi:10.1007/s10530-011-0020-0 de Leon, R., Vane, K., Bertuol, P., Chamberland, V. C., Simal, F., Imms, E., & Vermeij, M. J. A. (2013). Effectiveness of lionfish removal efforts in the southern Caribbean. Endangered Species Research, 22(2), 175-182. doi:10.3354/esr00542

Donaldson, T.J., D. Benavente, and R. Diaz. 2011. Why are lionfishes (Pterois, Scorpaenidae) so rare in their native range? Proceedings of the Gulf and Caribbean Fisheries Institute 63:352—359.

Dulvy, N. K., Polunin, N. V. C., Mill, A. C., & Graham, N. A. J. (2004). Size structural change in lightly exploited coral reef fish communities: evidence for weak indirect effects. Canadian Journal of Fisheries and Aquatic Sciences, 61(3), 466-475. doi:10.1139/f03-169

Edwards, M. A., Frazer, T. K., & Jacoby, C. A. (2014). Age and growth of invasive lionfish (Pterois spp.) in the Caribbean Sea, with implications for management. Bulletin of Marine Science, 90(4), 953-966. doi:10.5343/bms.2014.1022

Eggleston, D. B., Dahlgren, C. P., & Johnson, E. G. (2004). Fish density, diversity, and size-structure within multiple back reef habitats of Key West National Wildlife Refuge. Bulletin of Marine Science, 75(2), 175-204.

20 Fishelson, L. 1975. Ethology and reproduction of the pteroid fishes found in the Gulf of Aqaba (Red Sea), especially Dendrochirus brachypterus (Cuvier) (Pteroidae, Teleostei). Publ. Staz. Zool. Napoli 39, Supplement: 635-656.

Fogg, Alexander Q., "Life History of the Non-Native Invasive Red Lionfish (Pterois volitans) in the Northern Gulf of Mexico" (2017). Master's Theses. 282. https://aquila.usm.edu/masters_theses/282

Fogg, A. Q., Brown-Peterson, N. J., & Peterson, M. S. (2017). Reproductive life history characteristics of invasive red lionfish (Pterois volitans) in the northern Gulf of Mexico. Bulletin of Marine Science, 93(3), 791-813. doi:10.5343/bms.2016.1095

Fogg, A. Q., Evans, J. T., Peterson, M. S., Brown-Peterson, N. J., Hoffmayer, E. R., & Ingram, G. W. (2019). Comparison of age and growth parameters of invasive red lionfish (Pterois volitans) across the northern Gulf of Mexico. Fishery Bulletin, 117(3), 125-139. doi:10.7755/fb.117.3.1

Frazer, T. K., Jacoby, C. A., Edwards, M. A., Barry, S. C., & Manfrino, C. M. (2012). Coping with the Lionfish Invasion: Can Targeted Removals Yield Beneficial Effects? Reviews in Fisheries Science, 20(4), 185-191. doi:10.1080/10641262.2012.700655

Freshwater, D. W., Hines, A., Parham, S., Wilbur, A., Sabaoun, M., Woodhead, J., . . . Paris, C. B. (2009). Mitochondrial control region sequence analyses indicate dispersal from the US East Coast as the source of the invasive Indo-Pacific lionfish Pterois volitans in the Bahamas. Marine Biology, 156(6), 1213-1221. doi:10.1007/s00227-009-1163-8

Gardner, P. G., Frazer, T. K., Jacoby, C. A., & Yanong, R. P. E. (2015). Reproductive biology of invasive lionfish (Pterois spp.). Frontiers in Marine Science, 2, 10. doi:10.3389/fmars.2015.00007

Goodbody-Gringley, G., Eddy, C., Pitt, J. M., Chequer, A. D., & Smith, S. R. (2019). Ecological Drivers of Invasive Lionfish (Pterois volitans and Pterois miles) Distribution Across Mesophotic Reefs in Bermuda. Frontiers in Marine Science, 6, 12. doi:10.3389/fmars.2019.00258

Green, S. J., Akins, J. L., Maljkovic, A., & Cote, I. M. (2012). Invasive Lionfish Drive Atlantic Coral Reef Fish Declines. Plos One, 7(3), 3. doi:10.1371/journal.pone.0032596

Green, S. J., Dulvy, N. K., Brooks, A. M. L., Akins, J. L., Cooper, A. B., Miller, S., & Cote, I. M. (2014). Linking removal targets to the ecological effects of invaders: a predictive model and field test. Ecological Applications, 24(6), 1311-1322. doi:10.1890/13-0979.1

Gress, E., Andradi-Brown, D. A., Woodall, L., Schofield, P. J., Stanley, K., & Rogers, A. D. (2017). Lionfish (Pterois spp.) invade the upper-bathyal zone in the western Atlantic. Peerj, 5, 15. doi:10.7717/peerj.3683

Halvorsen, K. T., Sordalen, T. K., Vollestad, L. A., Skiftesvik, A. B., Espeland, S. H., & Olsen, E. M. (2017). Sex- and size-selective harvesting of corkwing wrasse (Symphodus melops)-a cleaner

21 fish used in salmonid aquaculture. Ices Journal of Marine Science, 74(3), 660-669. doi:10.1093/icesjms/fsw221

Hamner, R. M., Freshwater, D. W., & Whitfield, P. E. (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. doi:10.1111/j.1095-8649.2007.01575.x

Harms-Tuohy, C. A., Appeldoorn, R. S., & Craig, M. T. (2018). The effectiveness of small-scale lionfish removals as a management strategy: effort, impacts and the response of native prey and piscivores. Management of Biological Invasions, 9(2), 149-162. doi:10.3391/mbi.2018.9.2.08

Januchowski-Hartley, F. A., Graham, N. A. J., Feary, D. A., Morove, T., & Cinner, J. E. (2011). Fear of Fishers: Human Predation Explains Behavioral Changes in Coral Reef Fishes. Plos One, 6(8). doi:10.1371/journal.pone.0022761

Johnson, E. G., & Swenarton, M. K. (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, 27. doi:10.7717/peerj.2730

Johnston, M. W., & Bernard, A. M. (2017). A bank divided: quantifying a spatial and temporal connectivity break between the Campeche Bank and the northeastern Gulf of Mexico. Marine Biology, 164(1). doi:10.1007/s00227-016-3038-0

Kulbicki, M., Beets, J., Chabanet, P., Cure, K., Darling, E., Floeter, S. R., . . . Wantiez, L. (2012). Distributions of Indo-Pacific lionfishes Pterois spp. in their native ranges: implications for the Atlantic invasion. Marine Ecology Progress Series, 446, 189-205. doi:10.3354/meps09442

Lesser, M. P., & Slattery, M. (2011). Phase shift to algal dominated communities at mesophotic depths associated with lionfish (Pterois volitans) invasion on a Bahamian coral reef. Biological Invasions, 13(8), 1855-1868. doi:10.1007/s10530-011-0005-z

Lesser, M. P., Slattery, M., & Leichter, J. J. (2009). Ecology of mesophotic coral reefs. Journal of Experimental Marine Biology and Ecology, 375(1-2), 1-8. doi:10.1016/j.jembe.2009.05.009

Lindfield, S. J., McIlwain, J. L., & Harvey, E. S. (2014). Depth Refuge and the Impacts of SCUBA Spearfishing on Coral Reef Fishes. Plos One, 9(3), 12. doi:10.1371/journal.pone.0092628

Luquet, P., & Watanabe, T. (1986). INTERACTION NUTRITION-REPRODUCTION IN FISH. Fish Physiology and Biochemistry, 2(1-4), 121-129. doi:10.1007/bf02264080

Malpica-Cruz, L., Green, S. J., & Cote, I. M. (2019). Temporal and ontogenetic changes in the trophic signature of an invasive marine predator. Hydrobiologia, 839(1), 71-86. doi:10.1007/s10750- 019-03996-2

22 Marshak, A. R., Heck, K. L., & Jud, Z. R. (2018). Ecological interactions between Gulf of Mexico snappers (Teleostei: Lutjanidae) and invasive red lionfish (Pterois volitans). Plos One, 13(11). doi:10.1371/journal.pone.0206749

Molnar, J. L., Gamboa, R. L., Revenga, C., & Spalding, M. D. (2008). Assessing the global threat of invasive species to marine biodiversity. Frontiers in Ecology and the Environment, 6(9), 485- 492. doi:10.1890/070064

Morris, J. A., & Akins, J. L. (2009). Feeding ecology of invasive lionfish (Pterois volitans) in the Bahamian archipelago. Environmental Biology of Fishes, 86(3), 389-398. doi:10.1007/s10641- 009-9538-8

Munoz, R. C., Currin, C. A., & Whitfield, P. E. (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-U494. doi:10.3354/meps09154

Nuttall, M. F., Johnston, M. A., Eckert, R. J., Embesi, J. A., Hickerson, E. L., & Schmahl, G. P. (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(2), 111-115. doi:10.3391/bir.2014.3.2.09

O'Farrell, S., Bearhop, S., McGill, R. A. R., Dahlgren, C. P., Brumbaugh, D. R., & Mumby, P. J. (2014). Habitat and body size effects on the isotopic niche space of invasive lionfish and endangered Nassau grouper. Ecosphere, 5(10). doi:10.1890/es14-00126.1

Peake, J., Bogdanoff, A. K., Layman, C. A., Castillo, B., Reale-Munroe, K., Chapman, J., . . . Morris, J. A. (2018). Feeding ecology of invasive lionfish (Pterois volitans and Pterois miles) in the temperate and tropical western Atlantic. Biological Invasions, 20(9), 2567-2597. doi:10.1007/s10530-018-1720-5

R Core Team (2020). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/.

Raymond, W. W., Albins, M. A., & Pusack, T. J. (2015). Competitive interactions for shelter between invasive Pacific red lionfish and native Nassau grouper. Environmental Biology of Fishes, 98(1), 57-65. doi:10.1007/s10641-014-0236-9

Reef Environmental Education Foundation. (January 12, 2009). Lionfish Arrive in the Florida Keys [Press release]. Retrieved from https://www.reef.org/news/press-releases/lionfish-arrive-florida- keys

Reznick, D. (1983). THE STRUCTURE OF GUPPY LIFE HISTORIES - THE TRADEOFF BETWEEN GROWTH AND REPRODUCTION. Ecology, 64(4), 862-873. doi:10.2307/1937209

23 Reznick, D. A., Bryga, H., & Endler, J. A. (1990). EXPERIMENTALLY INDUCED LIFE-HISTORY EVOLUTION IN A NATURAL-POPULATION. Nature, 346(6282), 357-359. doi:10.1038/346357a0

Rochet, M. J. (1998). Short-term effects of fishing on life history traits of fishes. Ices Journal of Marine Science, 55(3), 371-391. doi:10.1006/jmsc.1997.0324

Sbragaglia, V., Morroni, L., Bramanti, L., Weitzmann, B., Arlinghaus, R., & Azzurro, E. (2018). Spearfishing modulates flight initiation distance of fishes: the effects of protection, individual size, and bearing a speargun. Ices Journal of Marine Science, 75(5), 1779-1789. doi:10.1093/icesjms/fsy059

Semmens, B. X., Buhle, E. R., Salomon, A. K., & Pattengill-Semmens, C. V. (2004). A hotspot of non- native marine fishes: evidence for the aquarium trade as an invasion pathway. Marine Ecology Progress Series, 266, 239-244. doi:10.3354/meps266239

Slattery, M., Lesser, M. P., Brazeau, D., Stokes, M. D., & Leichter, J. J. (2011). Connectivity and stability of mesophotic coral reefs. Journal of Experimental Marine Biology and Ecology, 408(1- 2), 32-41. doi:10.1016/j.jembe.2011.07.024

Sponaugle, S., Gleiber, M. R., Shulzitski, K., & Cowen, R. K. (2019). There's a new kid in town: lionfish invasion of the plankton. Biological Invasions, 21(10), 3013-3018. doi:10.1007/s10530- 019-02070-1

Stefanoudis, P. V., Gress, E., Pitta, J. M., Smith, S. R., Kincaid, T., Rivers, M., . . . Rogers, A. D. (2019). Depth-Dependent Structuring of Reef Fish Assemblages From the Shallows to the Rariphotic Zone. Frontiers in Marine Science, 6, 16. doi:10.3389/fmars.2019.00307

Tamburello, N., & Cote, I. M. (2015). Movement ecology of Indo-Pacific lionfish on Caribbean coral reefs and its implications for invasion dynamics. Biological Invasions, 17(6), 1639-1653. doi:10.1007/s10530-014-0822-y

Tyler, E. H. M., Speight, M. R., Henderson, P., & Manica, A. (2009). Evidence for a depth refuge effect in artisanal coral reef fisheries. Biological Conservation, 142(3), 652-667. doi:10.1016/j.biocon.2008.11.017

Uusi-Heikkila, S., Wolter, C., Klefoth, T., & Arlinghaus, R. (2008). A behavioral perspective on fishing- induced evolution. Trends in Ecology & Evolution, 23(8), 419-421. doi:10.1016/j.tree.2008.04.006

24 Chapter 2. Trophic Interactions of Invasive Lionfish Across a Depth in the Florida Keys

2.1. Introduction

Lionfish, both Red Lionfish Pterois volitans and Devil Firefish Pterois miles (hereafter lionfish), are an extremely successful invasive species that likely came to Florida waters in the mid 1980’s from the aquarium trade (Schofield 2009; Whitfield et al. 2002). Native to the

Indopacific, lionfish can now be found throughout the Western Atlantic and Caribbean

(Betancur-R et al. 2011). They are members of the Scorpaenidae family, subfamily Pteroinae.

Since their invasion, there has been a significant amount of research on the ecological impacts of invasive lionfish to local reefs. On coral reefs, lionfish presence has caused decline of up to 69% in biomass of prey fish species (Green et al. 2012), and a decrease in fish recruitment by 79% (Albins and Hixon 2008). Bioenergetic modeling for invasive lionfish estimated that, at

60% maximum consumption, lionfish would consume over 900 kg of prey per hectare per year on Bahamian reefs (Cerino et al. 2013). Additionally, invasive species can indirectly impact trophic systems through resource competition with native species (Curtis et al. 2017; Marshak et al. 2018; O’Farrell et al. 2014).

Invasive lionfish’s diet has been well-studied. A meta-analysis found lionfish are generalist carnivores with at least 167 known vertebrate and invertebrate prey species in the

Atlantic (Peake et al. 2018). Fish species account for the majority of prey found in stomach content analysis (70.5% frequency), followed by shrimp (29.5% frequency), crabs (5.4% frequency), and a small portion of other invertebrates (Peake et al. 2018). However, diet can vary by maturity and appears to be dependent on prey abundance (Muñoz et al. 2011; Eddy et al.

2016; Malpica-Cruz et al. 2019; Goodbody-Gringley et al. 2019). Atlantic lionfish ontogenically

25 transition from a shrimp-based diet to a fish-dominated diet (Peake et al. 2018; Sancho et al.

2018; Morris and Akins 2009). In Southeast Florida, the region of this study, lionfish diet consists largely of small fish and crustaceans (Sancho 2018). In Bermuda, lionfish diet varies with prey abundance, and there they eat relatively more crustaceans than other regions (Eddy et al. 2016). A trait-based analysis found that solitary fish resting on or close to the reef with small, shallow bodies are most vulnerable to lionfish predation (Green and Côte 2014).

Compared to their native range, Atlantic lionfish spend similar amounts of time hunting, however, Atlantic lionfish prey on a wider range of prey species than Pacific counterparts (Cure et. al 2012). Additionally, prey size was nearly double on invaded reefs compared to native ones

(Cure et. al 2012). Invasive lionfish eat at a higher trophic level than the native Nassau grouper, despite a smaller body size (O’Farrell et al. 2014). The authors attribute this to prey naïveté and/or lack of competition for larger prey (Cure et al. 2012). The differences in foraging behaviors allow invasive lionfish to reach greater sizes and occur at higher abundances than those found in their native range (Darling et al. 2011).

Efforts to control lionfish populations have largely focused on shallow culling using pole spears. While culling has reduced lionfish numbers locally on some shallow reefs (Harms-Tuohy et al. 2018; de Leon et al. 2013, for example), the overall effectiveness of this method is still under examination (Andradi-Brown et al. 2017a; Andradi-Brown et al. 2017b). Most spearfishermen on SCUBA are limited to 30 m depth, yet lionfish have been observed as deep as

300 m in Bermuda and 240 m in Honduras (Gress et al. 2017). There is growing evidence that culling is causing a learned avoidance of spearfishermen and migration to adjacent mesophotic reefs deeper than 30 m (Côte et al. 2014; Andradi-Brown et al. 2017a).

26 Limited studies exist on how lionfish diet may change with depth. The meta-study by

Peake et al. (2018) did not examine depth and the majority of samples used came from depths shallower than 30 m, save for data from the Northern GOM and North Carolina. Deep mesophotic lionfish may also be consuming more or different prey species; there is evidence that mesophotic lionfish consume key herbivores such as parrotfish, which can lead to increased algal cover on reefs and subsequent coral decline (Stefanoudis et al. 2019; Cure et al. 2012; Lesser and

Slattery 2011). In Bermuda, mesophotic lionfish distribution appears to be correlated to prey fish density and prey fish biomass, both of which co-varied with seawater temperature (Goodbody-

Gringley et al 2019). Mesophotic sites are often less affected by overfishing and other anthropogenic impacts; thus, deep sites may offer more prey fish as well as a relief from direct spearfishing pressure on lionfish (Tyler et al. 2009). However, one may expect that deeper lionfish have a less diverse diet than shallow lionfish, as lionfish diet varies with prey abundance

(Muñoz et al. 2011; Eddy et al. 2016; Malpica-Cruz et al. 2019; Goodbody-Gringley et al. 2019), and prey species richness is highest from 0-30 m depth (Asher et al. 2017; Kahng et al. 2010).

Stomach content analysis is commonly used in fisheries research and has been applied to invasive lionfish (Sancho et al. 2018; Eddy et al. 2016; Morris and Akins 2009; Layman and

Allgeier 2012; Dahl and Patterson 2014). It can provide specific prey identification, but it is only a snapshot of what the fish are eating immediately before capture. Stable isotope methods can also be used for a broader understanding of invasive lionfish diet over the course of weeks to months, depending on the turnover of the tissue (Cocheret de la Morinière et al. 2003). δ13C is an indicator of the basal resource use of the sampled organism and δ15N is a reflection of the relative trophic level of the prey consumed (Benstead et al. 2006). Lionfish δ13C and δ15N values have documented isotopic niche shifts between habitats, fish sizes, and available prey (Malpica-

27 Cruz et al. 2019; Curtis et al. 2017; Muñoz et al. 2011). Together, these methods provide insight into the variable diet of invasive lionfish.

This study will use stomach content and stable isotope analysis to further describe invasive lionfish diet in the Florida Keys archipelago in south Florida, and specifically examine any differences between deep and shallow fish. Stomach contents will be used to assess which types of organisms lionfish are eating and the frequency at which they occur in lionfish diet.

Stable isotope analysis will be used to assess what shallow and deep lionfish are eating on a broader scale and to compare any trophic overlap between deep and shallow fish. We expect that deep lionfish will be eating more fish than invertebrate species, thus at a higher trophic level, and have a less diverse diet than shallow lionfish.

2.2. Methods

2.2.1. Collections

The same 74 lionfish samples used in Chapter 1 were used for the stomach content analysis and stable isotope analysis for this study. The specimens were collected from depths ranging from 13.7 to 57.9 m and were separated into “shallow <30 m” (n=35) and “deep >30 m”

(n=39) categories based on the recreational diving depth limit and the distribution of sample depths. Of these, 21 were collected in September 2018, categorized as “fall”. The remaining 53 samples were classified as “spring” samples: 22 specimens collected in April 2019, 3 specimens collected in May 2019, and 28 were collected in June of 2019. Twenty-nine of the 74 fish were captured in traps deployed by Florida Fish and Wildlife Conservation Commission at depths greater than 34 m; the remainder of the fish were caught using a pole-spear at both deep and shallow depths. The sampling sites were all within 25 km of each other in the Florida Keys

28 (Chapter 1, Figure 1). Specimens were frozen after collection and during transportation or shipment to Baton Rouge, Louisiana. Four samples were not used for isotope analysis because of poor tissue condition.

2.2.2. Stomach Content Analysis

Prior to dissection, fish standard length (SL) and mass were recorded. Fish were dissected according to the NOAA guide (Green et al. 2012). The stomachs were removed, and the contents examined. Prey objects were identified to the family level whenever possible. Those contents which were partially digested were sorted into “teleost,” “cephalopod,” “decapod,” or

“unknown” categories. When possible, prey were identified to the family level using A Field

Guide to Coastal Fishes by Val Kells and Kent Carpenter (2011) or SEAMAP’s “Picture Guide to Shelf Invertebrates of the Northern Gulf of Mexico” (Perry and Larsen 2004).

The frequency of occurrence and percentage by number of prey were reported for four broad prey categories: teleost, decapod, cephalopod and unknown. Frequency of occurrence

(%F) is the proportion of stomachs in which a given prey item is present, excluding empty stomachs. Percentage by number (%N) reports the proportion of a specific prey item relative to the overall number of prey items. %F and %N were reported for all stomachs pooled, for deep and shallow lionfish, and for fall and spring lionfish. A generalized linear model with a Poisson distribution was used with prey frequency as the dependent variable, and depth (2 levels: deeper than 30m, shallower than 30 m), season (2 levels: fall and spring), prey types (4 levels: teleost, decapod, cephalopod, and unknown), and all interactive effects as dependent variables. A chi- squared test was used to select predictive terms from the full model, and we ran a new generalized linear model with a Poisson distribution and only those terms selected by the chi- squared test. P-values were reported, and residuals examined for homoscedasticity.

29 2.2.3. Stable Isotope Analysis

To capture longer term and broader diet data, stable isotope analysis was used. A 1 cm3 piece of muscle tissue was removed posterior to the dorsal fin of 70 lionfish samples (Layman and Allgeier 2012). Samples were freeze-dried and ground until homogenous. For each muscle sample, 0.5 mg was placed into individual tin cups and flash-combusted using a Thermo-

Finnigan element analyzer. Samples were analyzed for δ13C and δ15N values using the Thermo-

Fisher Delta Plus XL continuous flow stable isotope mass spectrometer at the Stable Isotope

Ecology Laboratory at Louisiana State University. Raw δ values were normalized and the sample precision was <0.1.

The following equation is used to express stable isotope abundances as δ notions with per mil (‰) units:

δX = [(Rsample / RStandard) - 1] x 1000

13 15 13 12 15 14 Where X is C or N, and R is the ratio of C/ C or N/ N, respectively. RStandard values were

13 15 based on Vienna PeeDee Belemnite (VPDB) for δ C and atmospheric N2 (air) for δ N.

Two linear regressions were used to test the effects of fish size (standard length) on both

δ13C and δ15N values, collectively and for each depth category, to determine if the data should be separated into groups based on lionfish size for analysis. Previous studies have found 13C and 15N enrichment with increasing body size and have analyzed stable isotope values using size classes

(O’Farrell et al. 2014; Curtis et al. 2017; Malpica-Cruz et al. 2019). However, this study found no significant relationship between lionfish length and 13C or 15N ratios, therefore we did not include lionfish size in the final analysis of stable isotope data.

30 To satisfy the assumptions of normal distribution, δ13C and δ15N values were tested using a Shapiro-Wilk normality test. A multivariate analysis of variance (MANOVA) was used to assess differences in δ13C and δ15N with respect to depth, season, and an interactive effect.

Independent variables were depth, season, and the interaction between depth and season. These factors were selected to capture any seasonal or depth-related shifts in diet. Because baseline particulate organic matter (POM), dissolve organic matter (DOM), and lower trophic isotopic baseline samples were not collected, the effect of season on isotope values was examined in order to control for seasonal shifts in isotopic baseline values, as well as seasonal variation in diet. If significance was detected in the MANOVA results, an Analysis of Variance (ANOVA) was used with either δ13C or δ15N as the dependent variable and the significant effect(s) (depth, season, and a depth-season interaction) as independent variables. P-values were reported, and residuals visually inspected for homoscedasticity.

Isotope ranges and ellipse metrics were calculated using the package Stable Isotope

Bayesian Ellipses in R (SIBER) in R statistical program (Jackson et al. 2011; R Core

Development Team 2020). Standard ellipses represent the bivariate standard deviation for isotope values, and comparison standard ellipse areas (SEA) can be used to identify differences in trophic niche size among groups. SEA was calculated for four groups: fall-deep, fall-shallow, spring-deep, and spring-shallow. From SEA, SEAC was calculated to correct for small sample size, then used to estimate each group’s isotopic niche (Jackson et al 2012). Additional niche metrics were calculated, including carbon range (CR), nitrogen range (NR), and mean centroid distance (CD) according to Layman et al. (2007). The CR and NR show the most enriched and most depleted values of the respective isotope ratio, i.e. the minimum and maximum. The CD value is the mean Euclidian distance between the individual δ13C and δ15N coordinate and the

31 mean δ13C and δ15N values for that group. This metric can represent the mean trophic diversity within that group. These metrics were used to compare the four season-depth groupings, which were identified as significant by the MANOVA results. Using the approach of Jackson et al.

(2012), Bayesian estimates of SEA (SEAb) were generated to obtain confidence intervals for isotopic niche areas to compare between groups. Two chains with 20,000 interactions were used to generate results, with a burn-in of 1000 and thinning of 10.

Niche overlaps were calculated using Bayesian estimates in the SIBER package. Area of overlap was calculated with 100 draws for each of the season-depth combinations according to the methods of Jackson et al. (2011). First, the pairwise niche overlap was calculated by dividing the area of the overlap by the sum of the two ellipse areas with the overlap area subtracted, repeated for each iteration. The mean was taken from the resulting 100 values. Overlap values larger than 0.6 were considered significant, representing a 60% overlapping SEAC (Guzzo et al.

2013). Because the mean niche overlap method failed to capture the overlap for each individual ellipse, the overlap area was divided by each ellipse total area separately and the mean was calculated.

2.3. Results

2.3.1. Stomach contents

There were 10 teleost families, 3 decapod families, and a single cephalopod identified in lionfish stomach contents. A summary of teleost families can be seen in Figure 2.1. Aside from unidentifiable teleost items, Sciaenids (n=8) were the most common teleost prey item, followed by Lutjanids (n =4, specifically red snapper Lutjanus campechanus). The decapod families identified were Caridea (n=17), Galatheidae (n=1), Portunidae (n=1). For all samples, teleosts

32 occurred at a 70.0% frequency, with decapod prey present in 65.0% of stomachs, cephalopods in

5.0%, and unknown prey items in 30.0% (%F). By number, teleosts accounted for 48.9% of prey items, decapods 37.8%, cephalopods 2.2%, and unknown items 15.6% in total lionfish stomachs

(%N). Overall, there was a stomach vacuity rate, the rate of stomachs with no prey items, of

58%, and notably the deep samples caught with traps (n=29) had an 83% vacuity rate.

Figure 2.1. Teleost families found in stomach contents.

For deep samples, teleost occurred in 80% of stomachs, decapods were in 20%, and unknown items were in 10% (%F). By %N, 84.2% of deep prey items were teleosts, 10.5% were decapods, and 5.3% were unknown for deep samples. No cephalopods were found in deep stomachs. In shallow samples, decapods occurred at 76.2% frequency, teleosts at 66.7%, cephalopods at 4.8%, and unknown prey items at 23.8% (%F). By percent number, teleost made up 48.1% of prey items in shallow stomachs, with decapods comprising 38.9%, cephalopods

33 1.9%, and unknown items 11.1% (%N). While the 29 deep samples from traps largely had empty stomachs likely due to barotrauma (vacuity rate = 83% for spring-deep samples), there was a higher portion of decapods in shallow samples relative to deep, according to a comparison of the

%N (Figure 2.2).

Fall samples had 87.5% teleosts, 12.5% decapods, and 12.5% unknown prey items in terms of %F. By %N, teleosts made up 87.5% prey items, decapods 6.25%, and unknown items 6.25% in the fall collections. Spring samples had 65.2% teleosts, 73.9% decapods, 4.3% cephalopods, and 17.4% unknown items by %F. In terms of %N, teleosts comprised 49.1% of prey items, decapods 38.6%, cephalopods 1.8%, and unknown 8.8% in spring samples.

Figure 2.2. Percent number (%N) of each prey category in deep and shallow lionfish stomachs, both fall and spring seasons combined.

34 Season, depth, and prey type were found to be predictive for frequency of prey occurrence and included in the generalized final model. No interactive effects were significant (p> 0.05) and therefore not included in the final model. Cephalopod prey (p= 0.005) and unknown prey

(p=0.006) were consumed less frequently than decapod prey (included in the intercept). The frequency at which teleost prey were consumed did not significantly differ from the frequency at which decapod prey were consumed (p= 0.53). Spring season had a significantly higher frequency of prey consumption than the fall season (p=0.0002), and deeper fish had significantly lower frequency of prey consumption than shallower fish (p=0.001).

2.3.2. Stable Isotope Analysis

δ13C and δ15N values were measured for 70 lionfish samples (Figure 2.3). The Shapiro-Wilk test indicated that δ13C and δ15N values were both normally distributed (W=0.98, p = 0.43).

Depth, season, and the interaction of the two all had significant impact on both isotope values

(Table 2.1). Further analysis of δ13C and δ15N values separately included all three of these effects. δ13C values were lower for fall lionfish (-15.54 ± 0.22 ‰) compared to spring fish (-

14.39 ± 0.16 ‰, F1,65=24.88, p <0.00001), and differed between fall-deep, fall-shallow, spring-

13 deep, and spring-shallow (Figure 4, F1,65=4.86, p =0.049). Depth was not significant for δ C values (F1,65=1.75, p = 0.19). δ15N values were higher in the fall (9.76 ± 0.07‰) compared to spring (9.50 ±0.07 ‰, F1,65=9.47, p = 0.003) and higher in deeper lionfish (9.77 ± 0.07 ‰) than shallow lionfish (9.36 ±0.06 ‰, F1,65=22.77, p <0.00001). The interaction effect between depth

15 and season was not significant for δ N values (F1,65=1.23, p = 0.27).

35 Table 2.1. MANOVA results of δ13C and δ15N values as dependent variables, and depth, season, and the depth-season interaction as independent variables. Pillai F-value DF (numerator) DF (denominator) Pr(>F)

Depth 0.27 11.72 2 65 4.5 ⨯ 10-5

Season 0.36 17.98 2 65 6.1 ⨯ 10-7

Depth:Season 0.09 3.16 2 65 0.05

Figure 2.3. δ13C and δ15N biplot for all lionfish samples, where points represent measured values (circles represent fall, triangles spring, red deep, and black shallow) and ellipses represent standard ellipse areas (red is fall-deep samples, black is fall-shallow, green is spring-deep, blue is spring-shallow).

36

Figure 2.4. Mean estimates with standard error for isotope values of the total mean (black), fall samples (orange), and spring samples (green). Triangles are deep, squares are shallow.

For all lionfish samples, δ13C values ranged from -19.80 to -11.61‰ and δ15N values ranged from 8.53 to 11.03‰. Spring-deep samples occupied a larger ellipse, with larger CR

(5.43) and NR (1.90) values compared to all other groupings, and a larger CD (1.13, Table 2.2,

Figure 2.3). Spring-shallow lionfish also had larger CR (3.65), NR (1.21), and CD (0.83) values compared to fall samples, albeit by a smaller margin than spring-deep samples. For both seasons, deep samples had larger CR and NR values.

37 Table 2.2. Isotopic niche metrics for assessing species isotopic niches by season and depth. (NR is the range of δ15N values, CR is the range of δ13C values, CD is the mean centroid distance, and

SEAC is the standard ellipse area corrected for sample size.)

Season Depth NR CR CD SEAC

Fall Deep 0.82 3.13 0.69 0.58

Fall Shallow 0.78 2.34 0.72 0.65

Spring Deep 1.90 5.43 1.14 1.84

Spring Shallow 1.21 3.65 0.83 0.95

Isotopic niche areas (SEAC) were notably larger for spring-deep lionfish (1.84) than all other season-depth combinations (Table 2.2). The Bayesian estimate of isotopic niche (SEAb) showed that spring-deep fish occupied a larger niche than both fall deep and fall shallow samples (Figure

2.5).

There were no significant pairwise trophic niche overlaps (<0.6) between any of the season- depth groupings (Table 2.3). However, the comparisons of overlap by each individual ellipse showed that there was noteworthy overlap. The overlap between fall-deep and spring-deep overlapped 78% of the fall-deep ellipse, fall-shallow and spring-deep overlapped the fall-shallow ellipse 73%, and spring-shallow and spring-deep overlapped the spring-shallow ellipse 81%.

38

Fall-shallow Fall-deep Summer-shallow Summer-deep Figure 2.5. Bayesian estimates of standard ellipse areas (SEAb) with 95% confidence intervals shown (gray boxes) for season-depth groupings. Red x represents the estimated

SEAC values.

Table 2.3. Trophic niche overlap estimates for combined and individual ellipse areas for season- depth groupings.

Pairwise niche Niche overlap Niche overlap Ellipse 1 Ellipse 2 overlap to Ellipse 1 to Ellipse 2

Fall-Shallow Spring-Shallow 0.20 0.32 0.33 Fall-Deep Spring-Deep 0.40 0.78 0.44 Fall-Shallow Fall-Deep 0.17 0.36 0.24 Fall-Shallow Spring-Deep 0.24 0.73 0.27 Spring-Shallow Fall-Deep 0.21 0.43 0.27 Spring-Shallow Spring-Deep 0.27 0.81 0.29

39 2.4. Discussion

2.4.1. Stomach contents

Lionfish in the Florida Keys ate teleost, decapod, and cephalopod prey. Teleost and decapods were consumed at similar frequencies, while cephalopods occurred at a significantly lower rate

(only one found in stomach contents). Compared to the nearby Biscayne Bay, lionfish in this study ate more teleosts both in terms of %F and %N (Sancho et al. 2018). A larger meta-analysis of lionfish stomach contents from 10 locations in the invaded range found similar proportions of teleost, decapod, and cephalopod numbers to our findings (58.8, 33.6, 0 %N, respectively)

(Peake et al. 2018). Overall, the stomach contents reported in this study are similar to previous work and we do not report any novel prey species.

Of the identifiable teleost items, Sciaenidae (12.9 %F) was the most common prey family, followed by Lutjanidae (6.5 %F), specifically Red Snapper, Lutjanus campechanus. Both

Sciaenids and Lutjanids are fished recreationally in the Florida Keys. These occurred less frequently in Biscayne Bay stomach contents than in our results. Sciaenid prey was present in a meta-analysis of over 8000 lionfish stomachs (0.1 %F), and Lutjanid prey occurred at a higher frequency in our study than the meta-analysis (1.6 %F, Peake et al. 2018).

Sciaenids include croaker and drum species and are known for their diverse ability to produce sounds (Ramcharitar et al. 2006). Lionfish are also soniferous and could use sound to organize observed social behaviors (Beattie et al. 2017; Layman and Judd 2012). Freshwater fish species in the Cyprinidae family have been documented associating sound with food sources

(Holt and Johnson 2011). It is conceivable that lionfish may be targeting soniferous fish in the

Florida Keys, but further investigation is necessary.

40 Red snapper were only present in spring samples. Lionfish predation of juvenile red snapper is a concern for this economically important species for commercial and recreational fisheries.

The fishery has shown signs of overfishing in the population structure and growth rates (Nieland et al. 2007). Further stress to the population recovery via predation by lionfish should be of concern to fisheries managers.

The methods of collection could influence the stomach content results. It is expected that deeper samples caught in traps had high vacuity rates, as barotrauma while bringing fish to the surface often results in regurgitation of stomach contents (author observation). Spring-shallow, fall-shallow, and fall-deep samples were all collected via spearfisherman, but the time of day of collection is unknown. It is possible that the time of collection could influence the number of prey items present. Lionfish have more prey items in their stomach during crepuscular periods, so it is possible that the time of collection could influence these results (Morris and Akins 2009).

Stomach content analyses do not provide information as to the effectiveness of lionfish foraging ability. Further studies in experimental and in situ settings could inform how lionfish foraging ability changes with depth, and collection methods should be taken into account when analyzing stomach content data.

Shallow lionfish diet consisted of a higher proportion of decapods, although the effect was not significant. This finding, however, is consistent with the stable isotope data that shows deeper lionfish are eating at a higher trophic level, discussed in the sections below.

There were a large portion of unidentifiable prey items found in lionfish stomachs that had been digested beyond recognition. Further resolution of species could be accomplished by using

DNA barcoding techniques (Dahl et al. 2018). Future studies in the Florida Keys could

41 incorporate refined techniques along with reef community surveys to identify what lionfish are eating relative to prey availability and compare this information across depths.

2.4.2. Isotope values

There was no effect of lionfish length on either δ13C or δ15N values. Previous studies in lionfish have found a significant effect of fish size on isotopic values within communities

(Malpica-Cruz et al. 2019; Curtis el al. 2017; Dahl and Peterson 2014) and used this to group samples for analysis and support an ontogenetic shift in lionfish diet. It is possible that this study had too narrow a range of lionfish length (TL range: 176-371 mm) to support such a relationship.

Other studies have found no effect of lionfish size on isotope values, or an uneven effect, such as increasing δ13C values with increasing size but no change in δ15N (O’Farrell et al. 2014).

There was seasonal variation with isotope values, presumably due to a change in baseline isotope values or prey availability. In δ13C values, fall samples had significantly lower values than spring samples. δ15N values also varied by season, with higher values in fall samples than spring. However, the Florida Keys experience insignificant temporal differences in particulate organic matter (POM) δ13C and δ15N values (Lamb and Stewart 2018). It is possible that lionfish in the Florida Keys eat more carnivorous prey in the fall than spring, reflected in higher δ15N values (Vander Zanden and Rasmussen 2001), although a study over multiple years is needed to confirm this. Seasonal changes in lionfish diet have been recorded throughout Florida, where mature lionfish consumed more fish prey in winter months on natural reefs (Sancho et al. 2018;

Dahl and Peterson 2014). Conversely, on artificial reefs lionfish ate more invertebrates in the winter (Dahl and Peterson 2014)

The significant interaction between depth, season, and the interaction on δ13C appears to be driven by the difference in δ13C values in fall samples compared to the larger and higher values

42 for spring deep samples, considering that both spring groupings greatly overlapped and that depth alone did not have a significant effect on δ13C values.

Depth was not a significant predictor of δ13C values, however deep samples had significantly higher δ15N values than shallow lionfish. Higher δ15N values indicate that deep lionfish are eating at a higher trophic level than shallow counterparts (Peterson and Fry 1987, Vander

Zanden and Rasmussen 2001). If lionfish are eating more carnivores than herbivores, it is likely they are eating more teleosts. The majority of decapod species found in the stomach contents were caridea, a shrimp family consisting of mainly herbivore or omnivores. This trophic shift to higher δ15N values is likely a reflection of deeper lionfish eating more teleosts. The majority of carnivorous prey items in lionfish diet are teleosts (Peake et al. 2018). Other studies have found an ontogenetic shift in lionfish diet from decapods to teleosts (Peake et al. 2018; Sancho et al.

2018; Morris and Akins 2009), although there was not significant size difference in lionfish between depths (Chapter 1), nor were δ15N values correlated to lionfish body size.

Deep lionfish could have higher δ15N signatures because they are eating less herbivores.

Herbivores decrease with increasing depth (Brokovich et al. 2010; Kahng et al. 2010). Grazers like parrot fish and damsel fish are typically found shallower than 30 m (Liddell and Avery

2000). This would contradict the work of Lesser and Slattery (2011) who found that lionfish ate more herbivores at depth. However, local lionfish diet is based on the prey species abundance and site-specific surveys are needed for further insight (Corey et al. 2016; Muñoz et al. 2011;

Malpica-Cruz et al 2019). Deeper sites may have higher teleost numbers as mesophotic reefs are subject to less anthropogenic stressors, such as fishing pressure, snorkeling, and diving (Polunin and Roberts 1993).

43 A previous study attributed variation in δ13C values between depths to be driven by regional differences, especially those driven by variation of particulate organic matter (POM) exportation to benthos (Stasko et al. 2018). However, this study looked at a depth range of 20-500 m, far greater than the scale of this study, and took place in the arctic. Together with the stomach content results, deeper lionfish appear to be eating at a higher trophic level regardless of shifting isotopic baseline values.

2.4.3. Isotope niche

Spring-deep samples had the largest ellipse area and range of isotope values compared to the three other groups. Part of this variation can be explained by the wide range in dates and sites these samples came from. Of the 27 samples in this group, 22 specimens were collected in April

2019, 3 specimens in May 2019, and 2 were collected in June of 2019. The fish were collected from a relatively small geographical range, sites were at most 25 km from each other (See

Chapter 1, Figure 1). In contrast, all other samples (fall-deep, fall-shallow, spring-shallow) were collected within a four-day span from the same reef. Nonetheless, another study covered 18 sites between June and August and reported a lower carbon range than the values found in the spring- deep samples from the current study (Muñoz et al. 2011). It is possible that these spring-deep sites captured a wider variety of specialized diets in individuals than other groups in this study and previous studies. If many fish within a group consume a narrow isotopic range, but a different range than other individuals in the group, the overall isotopic ranges will be larger. In contrast, a group where each individual consumes a broad range of prey may still have a narrower range, as the values will be averaged. Unfortunately, this group also had a high stomach vacuity rate and these isotope values cannot be compared to prey items found in stomach contents.

44 Still, spring deep lionfish also had larger isotopic niche metrics compared to fall samples, despite a narrower sampling period and area. Some seasonal variation could be explained by baseline shifts and by availability of prey items seasonally. In the adjacent Florida Keys National

Marine Sanctuary, larval supply varied significantly seasonally and by location; larval supply was also positively related to recruitment of at least one damselfish species (Grorud and

Sponaugle 2009). Larval supply of 55 families in the area peaks in late winter and early spring, providing a great amount of diverse juvenile prey fish for the lionfish collected in the late spring

(April-June; D’Alessandro et al. 2007).

Overall, the lionfish in this study had large isotope ranges compared to other lionfish studies, particularly δ13C values. Lower δ13C values indicate the ultimate basal source is from phytoplankton, while higher values indicate benthic algae sources, such as seagrass beds and shallow reef crests (France 1995). A study in North Carolina reported a narrower range of δ13C values, –16.0 to –17.2‰ compared to our -19.80 to -11.61‰ but similar ranges for δ15N (Muñoz et al. 2011). In another south Florida study found a range of δ13C values more similar to these results, -16.32 to -13.13‰, although still with a narrower range of values (Curtis et al. 2017).

Lionfish in the Florida Keys appear to be utilizing a range of both planktonic and benthic basal sources.

The wide range of isotopic values for lionfish indicates a diverse diet and could be due to individual specialization within a population, where individual δ13C values were highly variable.

Lionfish from spring-deep sites showed a wide range of δ15N values as well, suggesting that they consume a range of prey at high and low trophic levels. This group constituted a larger portion of total samples (29 of 74), and perhaps this larger sample size better captures the range of individual diets. Individual lionfish were more specialized in diet than native snapper species in

45 the Bahamas, possibly the result of high lionfish site fidelity and the dominant prey species at that site (Layman and Allgeier 2012). We expected smaller ranges with deeper fish to reflect the decreasing prey species richness associated with increasing depth (Asher et al. 2016; Kahgn et al.

2010), but it appears we have captured a range of specialized diets for individual fish that was not reflected in the other single-day collections used in this study. While both deep and shallow spring lionfish appeared to have more specialized individuals present, there was no difference in diet diversity between depths based on niche metrics.

While the pairwise niche overlap comparison did not indicate significant overlap between overall niche areas, further analysis did show that for several individual season-depth groups overlap was significant. The spring-deep ellipse overlapped >60% of the fall-deep, fall-shallow, and spring-shallow ellipses. The spring-deep ellipse itself was much larger than the other ellipses, as discussed above. Lionfish trophic niches are likely more specific to the site from which the fish is collected than the specific season or depth.

This work does not support an ontogenetic shifts in lionfish diet to more fish prey, however which could be the result of a narrow range of lionfish lengths (Peake et al. 2018, Sancho et al.

2018, Morris and Akins 2009). Otolith information could be used to further investigate an effect of age on lionfish diet, perhaps at a better resolution.

This study used stomach content and stable isotope analysis to further describe invasive lionfish diet in the Florida Keys. Lionfish are eating a range of teleosts and decapod species, including those of economic importance. Stable isotope data appears to confirm that lionfish exhibit individual specialization. Deeper lionfish are eating at a higher trophic level than shallow fish, and combined with the stomach content data, it appears that deeper lionfish are eating more

46 teleost prey. Understanding lionfish diet on a local scale and across environmental gradients is useful for assessing the environmental impacts and managing this invasive species.

47 References

Albins, M. A., & Hixon, M. A. (2008). Invasive Indo-Pacific lionfish Pterois volitans reduce recruitment of Atlantic coral-reef fishes. Marine Ecology Progress Series, 367, 233-238. doi:10.3354/meps07620

Andradi-Brown, D. A., Grey, R., Hendrix, A., Hitchner, D., Hunt, C. L., Gress, E., . . . Exton, D. A. (2017). Depth-dependent effects of culling-do mesophotic lionfish populations undermine current management? Royal Society Open Science, 4(5), 15. doi:10.1098/rsos.170027

Andradi-Brown, D. A., Vermeij, M. J. A., Slattery, M., Lesser, M., Bejarano, I., Appeldoorn, R., . . . Exton, D. A. (2017). Large-scale invasion of western Atlantic mesophotic reefs by lionfish potentially undermines culling-based management. Biological Invasions, 19(3), 939-954. doi:10.1007/s10530-016-1358-0

Asher, J., Williams, I. D., & Harvey, E. S. (2017). Mesophotic Depth Gradients Impact Reef Fish Assemblage Composition and Functional Group Partitioning in the Main Hawaiian Islands. Frontiers in Marine Science, 4, 18. doi:10.3389/fmars.2017.00098

Beattie, M., Nowacek, D. P., Bogdanoff, A. K., Akins, L., & Morris, J. A. (2017). The roar of the lionfishes Pterois volitans and Pterois miles. Journal of Fish Biology, 90(6), 2488-2495. doi:10.1111/jfb.13321

Benstead, J. P., March, J. G., Fry, B., Ewel, K. C., & Pringle, C. M. (2006). Testing isosource: Stable isotope analysis of a tropical fishery with diverse organic matter sources. Ecology, 87(2), 326- 333. doi:10.1890/05-0721

Brokovich, E., Ayalon, I., Einbinder, S., Segev, N., Shaked, Y., Genin, A., . . . Kiflawi, M. (2010). Grazing pressure on coral reefs decreases across a wide depth gradient in the Gulf of Aqaba, Red Sea. Marine Ecology Progress Series, 399, 69-80. doi:10.3354/meps08354

Cerino, D., Overton, A. S., Rice, J. A., & Morris, J. A. (2013). Bioenergetics and Trophic Impacts of the Invasive Indo-Pacific Lionfish. Transactions of the American Fisheries Society, 142(6), 1522- 1534. doi:10.1080/00028487.2013.811098

Cote, I. M., Darling, E. S., Malpica-Cruz, L., Smith, N. S., Green, S. J., Curtis-Quick, J., & Layman, C. (2014). What Doesn't Kill You Makes You Wary? Effect of Repeated Culling on the Behaviour of an Invasive Predator. Plos One, 9(4), 6. doi:10.1371/journal.pone.0094248

Cure, K., Benkwitt, C. E., Kindinger, T. L., Pickering, E. A., Pusack, T. J., McIlwain, J. L., & Hixon, M. A. (2012). Comparative behavior of red lionfish Pterois volitans on native Pacific versus invaded Atlantic coral reefs. Marine Ecology Progress Series, 467, 181-192. doi:10.3354/meps09942

Curtis, J. S., Wall, K. R., Albins, M. A., & Stallings, C. D. (2017). Diet shifts in a native mesopredator across a range of invasive lionfish biomass. Marine Ecology Progress Series, 573, 215-228. doi:10.3354/meps12164

48

D'Alessandro, E., Sponaugle, S., & Lee, T. (2007). Patterns and processes of larval fish supply to the coral reefs of the upper Florida Keys. Marine Ecology Progress Series, 331, 85-100. doi:10.3354/meps331085

Dahl, K. A., Patterson, W. F., Robertson, A., & Ortmann, A. C. (2018). DNA barcoding significantly improves resolution of invasive lionfish diet in the Northern Gulf of Mexico (vol 19, pg 1917, 2017). Biological Invasions, 20(12), 3659-3659. doi:10.1007/s10530-018-1784-2

Darling, E. S., Green, S. J., O'Leary, J. K., & Cote, I. M. (2011). Indo-Pacific lionfish are larger and more abundant on invaded reefs: a comparison of Kenyan and Bahamian lionfish populations. Biological Invasions, 13(9), 2045-2051. doi:10.1007/s10530-011-0020-0 de la Moriniere, E. C., Pollux, B. J. A., Nagelkerken, I., Hemminga, M. A., Huiskes, A. H. L., & van der Velde, G. (2003). Ontogenetic dietary changes of coral reef fishes in the mangrove-seagrass-reef continuum: stable isotopes and gut-content analysis. Marine Ecology Progress Series, 246, 279- 289. de Leon, R., Vane, K., Bertuol, P., Chamberland, V. C., Simal, F., Imms, E., & Vermeij, M. J. A. (2013). Effectiveness of lionfish removal efforts in the southern Caribbean. Endangered Species Research, 22(2), 175-182. doi:10.3354/esr00542

Eddy, C., Pitt, J., Morris, J. A., Smith, S., Goodbody-Gringley, G., & Bernal, D. (2016). Diet of invasive lionfish (Pterois volitans and P-miles) in Bermuda. Marine Ecology Progress Series, 558, 193- 206. doi:10.3354/meps11838

France, R. L. (1995). Differentiation between littoral and pelagic food webs in lakes using stable carbon isotopes. Limnology and Oceanography, 40(7), 1310-1313. doi:10.4319/lo.1995.40.7.1310

Goodbody-Gringley, G., Eddy, C., Pitt, J. M., Chequer, A. D., & Smith, S. R. (2019). Ecological Drivers of Invasive Lionfish (Pterois volitans and Pterois miles) Distribution Across Mesophotic Reefs in Bermuda. Frontiers in Marine Science, 6, 12. doi:10.3389/fmars.2019.00258

Green, S. J., Akins, J. L., Maljkovic, A., & Cote, I. M. (2012). Invasive Lionfish Drive Atlantic Coral Reef Fish Declines. Plos One, 7(3), 3. doi:10.1371/journal.pone.0032596

Green, S. J., Dulvy, N. K., Brooks, A. M. L., Akins, J. L., Cooper, A. B., Miller, S., & Cote, I. M. (2014). Linking removal targets to the ecological effects of invaders: a predictive model and field test. Ecological Applications, 24(6), 1311-1322. doi:10.1890/13-0979.1

Gress, E., Andradi-Brown, D. A., Woodall, L., Schofield, P. J., Stanley, K., & Rogers, A. D. (2017). Lionfish (Pterois spp.) invade the upper-bathyal zone in the western Atlantic. Peerj, 5, 15. doi:10.7717/peerj.3683

49 Grorud-Colvert, K., & Sponaugle, S. (2009). Larval supply and juvenile recruitment of coral reef fishes to marine reserves and non-reserves of the upper Florida Keys, USA. Marine Biology, 156(3), 277-288. doi:10.1007/s00227-008-1082-0

Guzzo, M. M., Haffner, G. D., Legler, N. D., Rush, S. A., & Fisk, A. T. (2013). Fifty years later: trophic ecology and niche overlap of a native and non-indigenous fish species in the western basin of Lake Erie. Biological Invasions, 15(8), 1695-1711. doi:10.1007/s10530-012-0401-z

Harms-Tuohy, C. A., Appeldoorn, R. S., & Craig, M. T. (2018). The effectiveness of small-scale lionfish removals as a management strategy: effort, impacts and the response of native prey and piscivores. Management of Biological Invasions, 9(2), 149-162. doi:10.3391/mbi.2018.9.2.08

Holt, D. E., & Johnston, C. E. (2011). Can you hear the dinner bell? Response of cyprinid fishes to environmental acoustic cues. Animal Behaviour, 82(3), 529-534. doi:10.1016/j.anbehav.2011.06.004

Jackson, A. L., Inger, R., Parnell, A. C., & Bearhop, S. (2011). Comparing isotopic niche widths among and within communities: SIBER - Stable Isotope Bayesian Ellipses in R. Journal of Animal Ecology, 80(3), 595-602. doi:10.1111/j.1365-2656.2011.01806.x

Jennings, S., Pinnegar, J. K., Polunin, N. V. C., & Boon, T. W. (2001). Weak cross-species relationships between body size and trophic level belie powerful size-based trophic structuring in fish communities. Journal of Animal Ecology, 70(6), 934-944. doi:10.1046/j.0021- 8790.2001.00552.x

Kahng, S. E., Garcia-Sais, J. R., Spalding, H. L., Brokovich, E., Wagner, D., Weil, E., . . . Toonen, R. J. (2010). Community ecology of mesophotic coral reef ecosystems. Coral Reefs, 29(2), 255-275. doi:10.1007/s00338-010-0593-6

Kells, Valerie A. and Carpenter, Kent E., "A Field Guide to Coastal Fishes: From Maine to Texas" (2011). Biological Sciences Faculty Books. 5. https://digitalcommons.odu.edu/biology_books/5

Lamb, K., Swart, P.K., 2008. The carbon and nitrogen isotopic values of particulate organic material from the Florida Keys: a temporal and spatial study. Coral Reefs.. doi:10.1007/s00338-007- 0336-5 Layman, C. A., & Allgeier, J. E. (2012). Characterizing trophic ecology of generalist consumers: a case study of the invasive lionfish in The Bahamas. Marine Ecology Progress Series, 448, 131-141. doi:10.3354/meps09511

Lesser, M. P., & Slattery, M. (2011). Phase shift to algal dominated communities at mesophotic depths associated with lionfish (Pterois volitans) invasion on a Bahamian coral reef. Biological Invasions, 13(8), 1855-1868. doi:10.1007/s10530-011-0005-z

Malpica-Cruz, L., Green, S. J., & Cote, I. M. (2019). Temporal and ontogenetic changes in the trophic signature of an invasive marine predator. Hydrobiologia, 839(1), 71-86. doi:10.1007/s10750- 019-03996-2

50

Marshak, A. R., Heck, K. L., & Jud, Z. R. (2018). Ecological interactions between Gulf of Mexico snappers (Teleostei: Lutjanidae) and invasive red lionfish (Pterois volitans). Plos One, 13(11). doi:10.1371/journal.pone.0206749

Morris, J. A., & Akins, J. L. (2009). Feeding ecology of invasive lionfish (Pterois volitans) in the Bahamian archipelago. Environmental Biology of Fishes, 86(3), 389-398. doi:10.1007/s10641- 009-9538-8

Munoz, R. C., Currin, C. A., & Whitfield, P. E. (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-U494. doi:10.3354/meps09154

Nieland DL, Wilson CA, Fischer AJ (2007) Declining size at age among red snapper in the northern Gulf of Mexico off Louisiana, USA: recovery or collapse? Am Fish Soc Symp 60:329–335.

O'Farrell, S., Bearhop, S., McGill, R. A. R., Dahlgren, C. P., Brumbaugh, D. R., & Mumby, P. J. (2014). Habitat and body size effects on the isotopic niche space of invasive lionfish and endangered Nassau grouper. Ecosphere, 5(10). doi:10.1890/es14-00126.1

Peake, J., Bogdanoff, A. K., Layman, C. A., Castillo, B., Reale-Munroe, K., Chapman, J., . . . Morris, J. A. (2018). Feeding ecology of invasive lionfish (Pterois volitans and Pterois miles) in the temperate and tropical western Atlantic. Biological Invasions, 20(9), 2567-2597. doi:10.1007/s10530-018-1720-5

Peterson, B. J., & Fry, B. (1987). STABLE ISOTOPES IN ECOSYSTEM STUDIES. Annual Review of Ecology and Systematics, 18, 293-320. doi:10.1146/annurev.ecolsys.18.1.293

Pimiento, C., Nifong, J. C., Hunter, M. E., Monaco, E., & Silliman, B. R. (2015). Habitat use patterns of the invasive red lionfish Pterois volitans: a comparison between mangrove and reef systems in San Salvador, Bahamas. Marine Ecology-an Evolutionary Perspective, 36(1), 28-37. doi:10.1111/maec.12114

Polunin, N. V. C., & Roberts, C. M. (1993). GREATER BIOMASS AND VALUE OF TARGET CORAL-REEF FISHES IN 2 SMALL CARIBBEAN MARINE RESERVES. Marine Ecology Progress Series, 100(1-2), 167-176. doi:10.3354/meps100167

R Core Team (2020). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/.

Ramcharitar, J. U., Higgs, D. M., & Popper, A. Q. (2006). Audition in sciaenid fishes with different swim bladder-inner ear configurations. Journal of the Acoustical Society of America, 119(1), 439-443. doi:10.1121/1.2139068

51 Schofield, Pamela. (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. 10.3391/ai.2009.4.3.5.

Sancho, G., Kingsley-Smith, P. R., Morris, J. A., Toline, C. A., McDonough, V., & Doty, S. M. (2018). Invasive Lionfish (Pterois volitans/miles) feeding ecology in Biscayne National Park, Florida, USA. Biological Invasions, 20(9), 2343-2361. doi:10.1007/s10530-018-1705-4

Stasko, A. D., Bluhm, B. A., Michel, C., Archambault, P., Majewski, A., Reist, J. D., . . . Power, M. (2018). Benthic-pelagic trophic coupling in an Arctic marine food web along vertical water mass and organic matter gradients. Marine Ecology Progress Series, 594, 1-19. doi:10.3354/meps12582

Stefanoudis, P. V., Gress, E., Pitta, J. M., Smith, S. R., Kincaid, T., Rivers, M., . . . Rogers, A. D. (2019). Depth-Dependent Structuring of Reef Fish Assemblages From the Shallows to the Rariphotic Zone. Frontiers in Marine Science, 6, 16. doi:10.3389/fmars.2019.00307

Tyler, E. H. M., Speight, M. R., Henderson, P., & Manica, A. (2009). Evidence for a depth refuge effect in artisanal coral reef fisheries. Biological Conservation, 142(3), 652-667. doi:10.1016/j.biocon.2008.11.017

Vander Zanden, M. J., & Rasmussen, J. B. (2001). Variation in delta N-15 and delta C-13 trophic fractionation: Implications for aquatic food web studies. Limnology and Oceanography, 46(8), 2061-2066.

Whitfield, P. E., Gardner, T., Vives, S. P., Gilligan, M. R., Courtenay, W. R., Ray, G. C., & Hare, J. A. (2002). Biological invasion of the Indo-Pacific lionfish Pterois volitans along the Atlantic coast of North America. Marine Ecology Progress Series, 235, 289-297. doi:10.3354/meps235289

52 Chapter 3. Habitat Modeling for Invasive Lionfish in the Gulf of Mexico Using a Bioenergetics-Based Growth Model

3.1. Introduction

Understanding habitat quality spatially and temporally is important to predict a species's distribution. In marine systems, studying and predicting preferred habitat is important for management. It can inform a range of topics, including how species will respond to climate change (Beaugrand 2009), which habitat is preferred by protected species (Lagasse et al. 2015), the impact and future distribution of invasive species (Marras et al. 2015; Evangelista et al.

2016), and establishment of fisheries objectives (Hinz et al. 2003; Le Pape et al. 2014; Cogan et al. 2009).

Fish growth can be used to identify important habitats for species (reviewed by Able

1999). It assumes that healthier fish grow larger and faster due to more favorable abiotic and biotic conditions (Brandt et al. 1992; Able 1999; Searcy et al. 2007). Using spatially explicit models, we can estimate the growth of a species based on environmental conditions (Brandt and

Mason 2003; Brandt et al. 1992). The Growth Rate Potential (GRP) is one such modeling framework that measures the growth of a fish in the given physical and biological conditions. It is specific to each species and the size of the fish. This allows us to estimate suitable habitat based on the physiological requirements of a species. When coupled with environmental conditions on a spatial grid, the temperature, dissolved oxygen, and prey density values can be used to calculate the growth of a fish in that area using bioenergetics and, when available, foraging models (Brandt et al. 1992). The output from such modeling can be used to explore how environmental changes like warming, hypoxia, and other disturbances will affect fish habitat quantity and quality.

53 GRP modeling has been applied to predict potential habitat for invasive species. In the

Great Lakes, bioenergetic modeling has informed when and where invasive big-headed and

Asian carp populations could inhabit (Anderson et al. 2015; Cooke and Hill 2010). In particular,

GRP modeling can be especially useful for locating cryptic invasive species or those in hard-to- reach habitats. This modeling could inform hotspots for invasive species growth and serve as a tool to inform removal or monitoring efforts. One such invasive that may benefit from the application of GRP techniques is the invasive lionfish.

Invasive lionfish (Pterois volitans and Pterois miles) were first seen in the southern Gulf of Mexico (GOM) in 2009 (Aguilar-Perera et al. 2010). Since their establishment, their population has expanded throughout the GOM. Lionfish are habitat generalists and can survive at a range of depths and temperatures. They have been found in their invaded range on natural and artificial reefs, mangrove habitats, seagrass beds, mud flats, estuaries, and artificial structures such as oil rigs, shipwrecks, and debris (Cure et al. 2014; Schofield et al. 2014). Lionfish have a broad thermal tolerance (14°C to 32.5°C; Cerino et al. 2013), and therefore can occupy a range of depths, from 1 m to over 300 m (Gress et al. 2017).

Throughout their invaded range lionfish are typically associated with complex structures or regions with high rugosity (de Leon 2013; Jud and Layman 2012; Jud 2011; Hunt et al. 2019), with few exceptions (Anton et al. 2014; Cure et al. 2014). Structured habitat may facilitate recruitment and colonization in hard-bottom habitats and seagrass beds (Smith and Shurin 2010).

In the GOM, there are several types of structured bottom habitat present that could serve as lionfish habitat. Natural coral reefs occur at both deep and shallow depths in the GOM, although most are found in the eastern region (Gil-Agudelo et al. 2020). All the Gulf States implement artificial reef programs to increase reef habitat, either by deploying new structures or reutilizing

54 existing ones, such as oil rigs. Lionfish have been observed at all of these habitats, sometimes at very high densities (Schofield et al. 2014; Dahl and Patterson 2014). A habitat quality model for lionfish should take into account the spatial distribution of such structures.

Currently, lionfish populations are largely controlled by shallow culling with sling or pole spears. Often this is carried out by recreational divers with 30 m depth limits unless additional certification is obtained. There is growing evidence that lionfish control efforts through shallow culling programs may be undermined by deeper lionfish populations (Andradi-

Brown et al. 2017a; Andradi-Brown et al. 2017b; Johnston and Bernard 2017). Lionfish at depths greater than 30 m are reported to be larger and more fecund (Andradi-Brown et al. 2017a;

Andradi-Brown et al. 2017b; Switzer et al. 2015). These uncontrolled deeper lionfish have the potential to repopulate culled shallow sites (Johnston and Bernard 2017). Deeper mesophotic reefs often act as refugia for corals and other reef species, to which the presence of lionfish could threaten (Slattery et al. 2011). Habitat modeling could help inform which deep regions are most likely to be impacted by lionfish.

In addition to differences in size and fecundity between depths, lionfish diet may vary along a depth gradient (Lesser and Slattery 2011; Eddy et al. 2016), with deeper fish consuming more teleost prey in the Florida Keys (Chapter 2). Diet can impact fish condition, with implications for caloric density (Bureau et al. 2002). Bioenergetic models are particularly sensitive to changes in energy density, both that of predator and prey (Beauchamp et al. 1989;

Cerino et al. 2013).

In this study, we used a bioenergetics-based GRP model to investigate potential lionfish habitat in the GOM based on lionfish physiology, diet, and environmental conditions. By modifying an existing bioenergetic model for lionfish developed by Cerino et al. (2013), this

55 study used measurements of bottom temperature to estimate the growth rate potential of lionfish in areas where they are likely to be found: near structures such as reefs, wrecks, and oil platforms. Season and regional difference in potential lionfish habitat were also compared, as these may have implications for management decisions. We also examined potential differences in the energy density of both lionfish and prey in shallow and deep environments and incorporated this information in the bioenergetics model. Finally, we mapped locations of potentially high lionfish growth that can be used to establish lionfish monitoring and removal programs.

3.2. Methods

3.2.1. Calorimetry

Before building a bioenergetics-based GRP model for invasive lionfish, energy density of lionfish from deep and shallow reefs was determined using bomb calorimetry. Thirty-eight lionfish samples from the Florida Keys were used in this study. Samples were obtained from spearfishers and the Florida Wildlife Conservation Commission. Nine samples were collected in

September 2018 and twenty-eight samples were collected in April-June 2019. Of these spring samples, 12 were captured using passive traps. The other 25 fish were captured using pole spears. Eighteen samples were collected from sites shallower than 30 m, and nineteen sites were deeper than 30 m. A map of site locations can be seen in Figure 1 in Chapter 1.

Lionfish were frozen and stored until dissection. Standard length, mass, gonad mass, and sex were recorded for each fish. A full description of these data can be found in Chapter 1. A small portion of muscle tissue below the posterior to the dorsal fin was collected for energy density analysis and refrozen until processed.

56 Muscle tissue was thawed and dried for at least 48 hours at 60° C. It was then ground into a powder using a mortar and pestle, which was cleaned with ethanol and lint free wipes in between samples. Samples were not compressed into pellets, as adequate combustion was obtained from the powder. Samples were weighed to 0.0001 g, and although 0.6 g was the desired amount for each combustion, sample mass ranged from 0.1388 to 0.9458 g, with a mean of 0.5845 g.

Caloric density was measured using a Paar Instrument 1341 Plain Jacket Calorimeter.

Benzoic acid tablets were used to calibrate the calorimeter at regular intervals throughout the trials. Recordings were corrected for nitric acid formation according to procedures in the manual, but preliminary analysis showed that sulfur content was negligible and therefore measurements were not corrected for sulfuric acid formation. Final data were converted from calories per gram to kilojoules per gram (kJ/G) and QA/QC was completed by using linear regressions and

Analysis of Variance (ANOVA) tests of energy density as a function of sample mass, operator, and date of analysis. There were no significant effects (p < 0.05) of mass of sample combusted, operator, or date of analysis on measured caloric density.

3.2.2. Calorimetry data analysis

Bomb calorimetry data was checked for normality using a Shapiro-Wilks normality test.

A gamma regression was used as the data contains only positive values and this produced a better fit than a normal distribution when assessing the residuals. Preliminary analysis showed that no interactive effects were significant, and none were included in the final model. The model included the additive effects of fish sex (3 levels: female, male, undetermined), season of collection (2 levels: fall and spring), the depth at which the fish was collected as factors (2 levels: deep > 30 m and shallow < 30 m), and fish standard length (mm) as independent variables, and

57 energy density as the dependent variable. After residuals were checked for homoscedasticity, a three-way ANOVA was used to compare effects. P- and F-values were reported.

To test the relationship between reproductive capabilities and energy density of muscle tissue, a Spearman correlation was used. While we would not expect causality between the GSI and energy density, we might expect a correlation between the two. Correlation coefficients

(rho) and p-values were reported.

3.2.3. Model Input Data

The area of study includes the entire GOM, the North Caribbean Sea, and a small part of the western Atlantic, from 98.0 W° to 75.0 W° longitude and from 15.0 N° and 35.0 N° latitudes.

Temperature data on a 1/10° grid was collected through the NOAA Gulf Region Climatology website (Boyer et al. 2011). The data was originally collected by the World Ocean Database using 2,300,610 profiles between 1772 and 2008. Months were pooled into seasons according to the following: winter (January-March), spring (April-June), summer (July-September), and fall

(October-December). We assumed that temperature was constant for the entire cell and characteristic of that season. Bathymetry was collected from the mean values of the grid square of the ETOPO1 Arc-Min Global Relief model (Amante and Eakins 2009).

To identify potential structures with the rugosity needed for lionfish, locations of natural reefs, artificial reefs, oil platforms, shipwrecks, and other obstructions were identified in the

GOM. Natural reefs were estimated using the University of Colorado at Boulder’s dbSeabed information integration system for marine substrates (Buczkowski et al. 2006). It contains gridded bottom type data for six bottom types: carbonate, gravel, mud, oyster beds, rock, and sand from 377,000 observations. In this case, areas with 20% percent cover of rock or greater

58 were used to estimate coral reef coverage. The boundaries of these reefs were mapped as a polygon shapefile.

Other structures were gathered from various sources and combined into a database.

Artificial reef data were collected from the following state agencies: Texas Parks and Wildlife,

Florida Fish and Wildlife Conservation Commission, Alabama Marine Resources Division,

Louisiana Department of Wildlife and Fisheries, and Mississippi Marine Resources Division.

The data contained the latitude and longitude, site name, depth, and deployment date. Records of oil rigs and platforms in the GOM are managed by The Bureau of Ocean Energy Management

(BOEM) and available publicly. These include information on the platform type, location, depth, installation and removal date. The Coast Survey's Automated Wreck and Obstruction

Information System (AWOIS) was used to collect records on structural features in the GOM that included latitude, longitude, and brief historic and descriptive details. Any structure that could pose a barrier to navigation is included in the AWOIS database, including wrecks, rocks, and other miscellaneous objects.

The final structured habitat database contained 306 natural reef polygons, 5,970 artificial reef locations, 7,304 oil platforms, and 17,679 shipwrecks and obstructions (Figure 3.1). Bottom temperature and bathymetry data were extracted at each of the points or polygons in the structured habitat database. The input for the GRP model was bottom water temperature at the location of each structure in the GOM.

59

Figure 3.1. Map of natural and artificial structured habitats off the coast of the United States in the Gulf of Mexico.

The bioenergetic model for lionfish developed by Cerino et. al (2013) was used to compute the GRP. GRP is calculated according to Equation 1:

GRP = (C - (R + S + F + U)) * (EDprey / EDpred) (1)

60 Where C is consumption, R is respiration, S is specific dynamic action, F is egestion, U is excretion, EDprey is energy density of the prey species, and EDpred is energy density of the predator, in this case lionfish. All parameters and variables for the bioenergetic model can be seen in Table 3.1.

The mean energy density that was calculated from the lionfish samples was used for

EDpred. Prey energy density for benthic fish and shrimp were obtained from Campbell et al.

(2011). Results in Chapter 2 show that deeper lionfish eat at a higher trophic level and tend to eat more teleost species. For sites < 30 m depth, the energy density for shrimp was used (3260.6

J/g), and for sites > 30 m depth the energy density for benthic fish was used (4142.4 J/g).

Due to the relatively small sample size and limited collection area of lionfish lengths and weights reported in Chapter 1, literature values were used to parameterize the model. A study using over 2,400 lionfish from the GOM estimated a mean age of 1.4 years (Fogg et al. 2019).

The total length was calculated using this age with the corresponding von Bertalanffy growth curve, then weight was calculated using the published weight-length relationships (Fogg et al.

2018). The final result was a total length of 253.9 mm and weight of 254.5 g. This represents fish of both sexes and is specific to lionfish collected within the GOM.

Temperature was selected as the main driver of the bioenergetics model. Dissolved oxygen was assumed to be at optimal levels for growth, and consumption in the model was not limited by prey density. GRP was only calculated at specific locations in the structured habitat database. The output GRP, grams of lionfish muscle growth per day per fish, was used to identify areas of high lionfish growth potential, and therefore high habitat quality.

61 Table 3.1. Bioenergetic parameters and variables used. All values are from Cerino et al. (2013) with the exception of EDpreyD and EDpreyS, which are from Campbell et al. (2011), and Edpred which was estimated in this chapter.

Symbol Description Value Unit

-1 EDpreyD Energy density of deep prey 4142.4 J∙g

-1 EDpreyS Energy density of shallow prey 3260.6 J∙g

-1 Edpred Energy density of lionfish muscle tissue 35500 J∙g

CA Intercept for a 1 g fish at CTO 0.603 g∙ g-1∙ day-1

CB Coefficient for mass dependence -0.46 -

CQ Q10 value for consumption 4 -

CTO Optimum temperature for consumption 29.8 °C

CTM Max temperature for consumption 34.5 °C

RA Intercept for a 1 g fish at RTO 0.0085 g O2 ∙ g-1∙ day-1

RB Slope of allometric respiration function -0.28 -

RQ Q10 value for respiration 2.33 -

RTO Optimum temperature for respiration 32 °C

RTM Max temperature for respiration 34.5 °C

ACT Activity multiplier 1.8 -

SDA Specific Dynamic Action coefficient 0.2 -

FA Proportion of food consumed egested 0.2 -

UA proportion of food consumed excreted 0.06 -

Wpred Lionfish mass 254.5 g

Lpred Lionfish total length 253.9 mm

62 3.2.4. Statistical Analysis of Model Outputs

The model calculated GRP values for each site for each of the four seasons. Spatial patterns in GRP were examined by dividing the GOM into five regions: “west” (-97.72° W to -

95.00° W and 25.00° N to 30.00° N), “west central” (-95.00° W to -92.25° W and 26.00° N to

30.00° N), “east central” (-92.25° W to -88.00° W and 27.96° N to 31.00° N), “northeast” (-

88.00° W to -82.50° W and 27.96° N to 31.00° N), and “southeast” (-85.00° W to -80.152° W and 24.25° N to 27.96° N), as seen in Figure 3.2. These regions were based on the three biogeographical criteria identified by Back and Odaya (2001), and the “central” region was further divided due to the high concentration of habitat in the region. GRP estimates for points outside of these designated regions (e.g. on the east coast of Florida) were not used in this analysis.

A linear regression was constructed with square-root-transformed GRP as the dependent variable and season and region as independent variables. A square root transformation improved the fit of the data according to visual assessment of the qqplot and interactive effects were not significant in the full model, so they were not included in the final model. A two-way ANOVA was used to compare effects with a Type II sum of squares. P and F-values were reported, and residuals assessed for homoscedasticity visually. Tukey’s Honest Significant Difference (Tukey

HSD) test was used to identify significant differences in factor groups based on ANOVA results.

An alpha value < 0.05 was considered significant. All analyses were performed in R, the “car” package was used for the ANOVA analysis, and “agricolae” was used for the Tukey HSD analysis (R Core Team 2020; Fox 2019; De Mendiburu 2020).

63 The graphical results of the model output were compared visually to a map of reported lionfish sightings from United State Geological Survey (USGS) Nonindigenous Aquatic Species

(NAS) database (USGS 2020).

Figure 3.2. Gulf of Mexico regions identified for this study with fall bottom temperatures

in °C.

3.3. Results

3.3.1. Energy Density

The mean energy density of all muscle samples was 35.69 ± 0.50 kJ/g. The data was not normally distributed according to the Shapiro-Wilks normality test (p= 0.001). Log, square root,

64 cubed root, and double square root transformations were used on the calorimetry data in all models but did not greatly improve the heteroscedasticity of the residuals. Therefore, no transformations were used in final models. Summer lionfish had lower muscle caloric density than fall lionfish (p = 0.04). No other effects (depth, sex, or length) had a significant impact on lionfish caloric density (Table 3.2). The Spearman correlation revealed a positive coefficient correlation of 0.40 between the GSI values and caloric density (rho= 0.40, p= 0.10, Figure 3.3).

Table 3.2. Gamma regression model output of depth, season, sex, length (mm) on lionfish muscle caloric density. Intercept represents female fish at shallow depths in the fall. Estimate Std. Error t-value Pr(>t)

Intercept 9.06 0.09 98.98 <0.0001

Deep (>30 m) -0.04 0.03 -1.44 0.16

Summer -0.08 0.04 -2.12 0.04

Male 0.05 0.04 1.27 0.21

Undetermined sex 0.05 0.05 0.90 0.37

Length 0.00 0.00 0.57 0.58

65

Figure 3.3. Correlation between female lionfish GSI values and caloric density with 95% confidence intervals (rho= 0.40, p= 0.10).

3.3.2. Model Output

Our model successfully calculated a total of 168,874 GRP values for four seasons on each of the identified sites of structure lionfish habitat (Figure 3.4). GRP values ranged from 6.14 ⨯ 10-5 g ∙g-1 ∙day-1 to 5.35 ⨯ 10-4 g ∙g-1 ∙day-1 with a mean of 4.65 ⨯ 10-4 g ∙g-1 ∙day-1. There were no negative GRP values. Generally, all seasons had higher GRP values in deeper habitat. The majority of the identified structures lie in the west central and east central regions on the continental shelf.

66

Figure 3.4. Lionfish GRP values (mg ∙g-1 ∙day-1, lower gradient) estimated at each identified structured habitat for four seasons; bottom temperature is in heat colors (right gradient, °C).

Region had a significant effect on lionfish GRP values (F4, 63496= 2564.59, p < 0.0001).

Season did not have a significant effect on GRP (F3, 63496= 0.48, p = 0.70). The mean GRP values for each region can be seen in Figure 3.5. The southeast region had the largest mean GRP (4.85

⨯ 10-4 ± 0.003 g ∙g-1 ∙day-1) and northeast had the smallest mean GRP (3.92 ⨯ 10-4 ± 0.003 g ∙g-1

67 ∙day-1). The west-central and east-central had similar mean GRP values (4.71 ⨯ 10-4 ± 0.003 g ∙g-

1 ∙day-1 and 4.70 ⨯ 10-4 ± 0.003 g ∙g-1 ∙day-1, respectively), and the mean GRP of the western region was 4.51 ⨯ 10-4 (± 0.002) g ∙g-1 ∙day-1. The post-hoc Tukey HSD test revealed all regions were significantly different from each other, with the exception of west central and east central.

A comparison of model output GRP values to recorded lionfish sightings can be seen in

Figure 3.6. While the model appears to have captured the lionfish observed nearshore off of central Florida and the offshore observations near Louisiana, it tended to overpredict the potential habitat off the coast of Texas, particularly in south Texas. It also appears that potential habitat is missing west of central Florida.

Figure 3.5. Growth Rate Potentials for lionfish by region in the Gulf of Mexico, with Tukey HSD significance groupings.

68 USGS Survey

Figure 3.6. USGS reported lionfish sightings in the Gulf of Mexico (top) and model GRP output values (mg ∙g-1 ∙day-1) at identified lionfish habitat (bottom, fall season).

69 3.4. Discussion

3.4.1. Bomb Calorimetry

The calorimetry results found that fish size, sex, depth, and season did not have a significant effect on the caloric density of lionfish muscle tissues. There was a weak correlation between female GSI values and caloric density, which is unexpected given that other reef fish use muscle tissue lipid reserves for gonad development (Jorgensen et al. 1997; Craig et al. 2000;

Schwartzkopf and Cowan 2016). It is possible that lionfish experience longer lag times between

GSI increase and caloric density decrease, or that lionfish use energy stores from other tissues, such as the liver, for spawning and higher muscle energy density confers higher reproductive fitness. In an invasive freshwater fish, Cichla kelberi, somatic energy stores in the muscle increased before and after the reproductive period (Espinola et al. 2012). Further sampling is needed, however, to draw any firm conclusions as the correlation was weak.

Notably, since lionfish energetic density was not dependent on fish length, the original bioenergetic model may consider amending the EDpred parameter (Cerino et al. 2013). It is currently described as a function of fish weight. If anything, these results indicate that connecting the caloric density to GSI values or reproductive stage may yield a more accurate estimate. Other studies have found no correlation between fish length and caloric density (Gracia et al. 2010), or minimal correlation (Burril et al. 2018). It is worth noting, as discussed in Chapter 1, the lionfish sampled represented a relatively narrow range of lengths and few small fish were analyzed.

Other fish species did exhibit seasonal trends in caloric density. Red snappers and red drum exhibit temporal variation on caloric density of muscle tissues due to seasonal spawning

(Schwartzkopf and Cowan 2016; Craig et al. 2000). Energy reserves can decrease before and during spawning (Bulow et al. 1978). Perhaps because lionfish in the southeast GOM exhibit

70 year-round asynchronous spawning, these seasonal trends in muscle energy reserves are less pronounced (Fogg et al. 2017). Year-round samples of lionfish muscle tissue would be needed to further explore this trend.

In this sampling of lionfish muscle tissue from the Florida Keys, there were no differences in the muscle energy reserves of deep and shallow lionfish. Energy reserves can increase with increased feeding rates and fat storage (Bureau et al. 2002). These results may indicate that despite the differences in diet composition indicated in Chapter 2, lionfish across depths are feeding and storing fat at similar rates. An ontogenetic shift to deeper habitat could be driven by habitat preference, as larger, heavier fish were found at deeper sites in Utila (Andradi-Brown et al. 2017a). This is not reflected in the muscle tissue lipid reserves of this study. These preliminary findings suggest that shallow lionfish are as fit as deeper lionfish, and perhaps ontogenetic shifts are driven by other influences, such as fishing pressure at shallow sites or density-dependent factors.

3.4.2. Model output

There were more structures on the continental shelf and near the coast than offshore. Oil platforms only drill on the shelf, artificial reef programs typically focus on shallower (<100 m) depths for maximum ecological and recreational outcomes, and shipwrecks and obstructions are the result of anthropogenic activity and more likely to be located near the coast. Regionally, there are differences in the type and quantity of structures. Notably, there are no oil platforms off the coast of Florida, and the structures are most concentrated in the west central and east central region. Florida has the most artificial reef habitat with over 3000 artificial reefs in the state, although the reefs on the Atlantic coast of the state were excluded from this analysis. Texas, by contrast, had the fewest artificial reefs with 92. Shipwrecks and obstructions were fairly evenly

71 distributed throughout the regions. The majority of natural reefs were identified in the west central and east central region; erroneously, there were no natural reefs found off the coast of

Florida.

Based on trends on the mapped GRP values (Figure 3), deep sites (>30 m) had larger GRP values, which can be expected given the model assumption that deeper lionfish are eating more teleosts which have larger energy density values. If these assumptions hold true, this indicates that deep habitat could harbor large, fast growing lionfish populations. These differences are likely overstated in this model, however, because although deeper lionfish may eat more teleosts, it is unlikely their diet is entirely teleost prey. The same is true for shallow lionfish; it is unlikely the fish are only eating decapod species. Prey energy density is the most sensitive parameter in the lionfish bioenergetic model, so even small differences in diet composition could have significant impacts on lionfish growth (Cerino et al. 2013). Further investigation into the prey density and prey community composition at these sites could inform these findings and present more exact growth estimates.

There were relatively small changes in the bottom water temperature during the four seasons used in this model, and season did not have an impact on lionfish GRP results. The bottom water in the GOM appears to be prime habitat for lionfish year-round, even in colder months. This is unfortunate for the ecological health of the GOM, as colder winter months can prevent lionfish colonization in other regions. Lionfish are sometimes found north of North Carolina in the

Atlantic, however a population is not established there due to the cold winter temperatures

(Schofield et al. 2009; Kimball et al. 2004).

The GRP values generated show ample habitat for lionfish to grow in the GOM, with some regional variation. GRP values were highest in the southeast region, which includes southern and

72 central Florida. This region includes the southern-most points included in the analysis and has warmer water temperature compared to other regions. Additionally, lionfish established in south

Florida at least one year before they were seen in the GOM (Ruttenberg et al. 2012; Schofield

2010). The model has correctly identified this area as suitable habitat for lionfish growth. In contrast, the adjacent northeast region had the lowest GRP values. Compared to the USGS data, the model has captured the lack of growth in northern Florida, particularly on the eastern side of the panhandle. While there is structured habitat located there, growth rates remain relatively low.

This is likely driven by lower water temperatures in the region. There are some areas of higher growth, however, on the western side of the panhandle that are corroborated by USGS sightings data. This area has many artificial reef habitat and perhaps other factors besides temperature, such as high prey density, are driving high growth rates there.

The west central and east central regions, representing the coasts from approximately

Galveston, Texas to Mobile, Alabama, had statistically similar GRP values. This combined area has the most identified habitat locations, as most of the oil operations occur in this region and could harbor a large lionfish population. The USGS sightings data does indicate that lionfish occur on habitat further offshore, whereas the model has estimated a larger area for potential lionfish habitat. Several factors could influence this. Firstly, the model described here does not account for salinity. While lionfish can tolerate very low salinities, it is more likely that habitats with salinities greater than 30 ppt are preferable (Schofield et al. 2015; Evangelista et al. 2016;

Johnston and Purkins 2011). The salinity of this region is highly influenced by freshwater input from the Mississippi river and has lower salinities. The discrepancy between the sparse USGS sightings in the region and model predictions could also be explained by the difficulty of accessing these locations. Substantially fewer divers dive in this region compared to Florida due

73 to the difficult conditions, the frequently low visibility, and depth limitations. Most of the diving in the region takes place on the Flower Garden Banks reef, near the edge of the shelf, which matches the location of most USGS sightings in the area. Lionfish that have been observed on the oil rigs in particular tend to be at or near the bottom of the structure, further increasing the difficulty of assessing the population there (Pers. Comm.). Other studies have documented deeper lionfish using trawl surveys, but that method is not possible on many of the structured sites identified here (Schwitzer et al. 2015). Remotely operated vehicles (ROVs) could be used to collect more accurate surveys of these sites.

The west region included much of the coast of Texas to the U.S.-Mexico border. The model predicted higher lionfish growth at these sites than has been observed in the USGS database.

This discrepancy could also be attributed to observation bias described above, as these sites are deep for recreational diving. The model has identified this as a potential location for further monitoring.

Notably, these GRP values are for lionfish of total length of 253.9 mm and weight of 254.5 g.

Lionfish size can vary with depth, typically larger fish are found deeper (Andradi-Brown et al.

2017a; Switzer et al. 2015; Andradi-Brown et al. 2017b). Future models could select size class based on the depth of the site to ensure more accurate results.

There is significant interest in controlling lionfish populations in the GOM. The GOM is important economically and ecologically for many fish species. Commercial fisheries in the

GOM were worth an estimated $850 million in 2018, and 1.8 million recreational anglers visited the GOM that year (NMFS 2020). Throughout their invasive range, lionfish can reduce native fish biomass, recruitment, and competitiveness for resources (Green et al. 2012; Albins and

Hixon 2008; O’Farrell et al. 2015; Marshak et al. 2018; Bos et al. 2018; Raymond et al. 2015),

74 with consequences for threatened species (Rocha et al. 2015; O’Farrell et al. 2014; Raymond et al. 2015) and key commercial and recreational fish species (Chapter 2; Muñoz et al. 2011; Peake et al. 2018). As a result, there is substantial interest in controlling local lionfish population size and slowing the spread of this successful predator.

Lionfish are great candidates to use growth as a proxy for habitat preference, as their distribution is not influenced by potential predators (Hackerott et al. 2013; Anton et al. 2014) and they exhibit high site fidelity (Tamburello and Cote 2015; Jud and Layman 2012). Additionally, this work can inform where culling should take place to yield the highest biomass removal.

Especially in sites that are hard to access, such as mesophotic sites, modeling can target areas with higher probabilities of occupation to reduce resources needed to locate invasive lionfish.

This model made several assumptions that could impact the results. Firstly, it was assumed that deeper lionfish are eating teleost species and shallow lionfish are eating shrimp species. As discussed previously, these are likely exaggerated differences.

Another assumption was that salinity, dissolved oxygen, and prey density were all favorable for growth. Salinity does vary throughout the GOM due to freshwater input, and while it may not inhibit lionfish from inhabiting an area, its long-term physiological effects are unknown.

However, in one study testing the effects of salinities ranging from 5-34 ppt all experimental lionfish increased mass, regardless of salinity. Dissolved oxygen levels in the GOM could impact lionfish growth, as there are large, annual hypoxia events in GOM off the coast of Louisiana

(Rabalais et al. 2002). Prolonged exposure to hypoxic conditions has been shown to negatively impact fish reproduction and growth, or even lead to mortality (reviewed by Wu 2002). The model did not account for such effects.

75 Prey density was also not considered in the model inputs, thus was considered to be optimal for growth, nor were any foraging parameters used. The calculation of lionfish hunting ability, including hunting ability under various environmental conditions such as reduced light at depth, would provide a more accurate estimation of lionfish consumption and growth. A previous study found that lionfish consumed more prey and attack rates were higher under white and blue lights compared to red, although handling time did not vary between light treatments (South et al.

2017).

Finally, this study did not take into account any density dependent effects of lionfish habitat selection, which previous studies have indicated that growth rates are lower at sites with high lionfish density (Benkwitt 2013).

Temperature was the main environmental driver of growth in this model. While the environmental and biological factors discussed above can influence lionfish growth, other studies have associated lionfish growth and distribution with temperature. In study that sought to link growth and habitat preference, water temperature explained 40% of variation in juvenile otolith growth in Atlantic croaker, Micropogonias undulatus (Searcy et al. 2007). In Bermuda, mesophotic lionfish distribution appears to be correlated to prey fish density and prey fish biomass, both of which co-varied with seawater temperature (Goodbody-Gringley et al. 2019).

Some bioenergetic models model consumption solely based on growth (Bartell et al. 1986).

Future models could include the other factors discussed and compare whether these additions improve model accuracy.

The bioenergetic model used in this study has several shortcomings, including the use aquarium-raised lionfish from the Philippines and unrealistic, high-energy density prey (Cerino et al. 2013). Mummichog (Fundulus hetericlitus) live in salt marshes, not marine reefs and

76 structures where lionfish are found. Invasive lionfish likely grew from a small founding population and are very similar genetically (Betancur et al. 2011), thus may vary from those native fish kept in the aquarium trade. Additionally, a higher prey density has a large influence on the final GRP output of the model. Realistic prey estimates, therefore, are extremely important for useful results.

Overall, our model outputs provide useful insight into areas in the GOM where lionfish could find suitable habitat. There are regional differences in potential habitat that are not reflected in lionfish sightings, and seasonal temperature changes do not limit lionfish growth in the GOM. A

GRP model can be a useful tool to identify areas of lionfish growth and inform where management efforts can be focused for removals, particularly for lionfish at depth. This model could also be extended outside of the GOM to provide a management tool for the entirety of the lionfish’s invaded range.

77 References

Aguilar-Perera, A., Quijano-Puerto, L., & Hernandez-Landa, R. C. (2017). Lionfish invaded the mesophotic coral ecosystem of the Parque Nacional Arrecife Alacranes, Southern Gulf of Mexico. Marine Biodiversity, 47(1), 15-16. doi:10.1007/s12526-016-0536-8

Albins, M. A., & Hixon, M. A. (2008). Invasive Indo-Pacific lionfish Pterois volitans reduce recruitment of Atlantic coral-reef fishes. Marine Ecology Progress Series, 367, 233-238. doi:10.3354/meps07620

Amante, C. and B.W. Eakins, 2009. ETOPO1 1 Arc-Minute Global Relief Model: Procedures, Data Sources and Analysis. NOAA Technical Memorandum NESDIS NGDC-24. National Geophysical Data Center, NOAA. doi:10.7289/V5C8276M

Anderson, K. R., Chapman, D. C., Wynne, T. T., Masagounder, K., & Paukert, C. P. (2015). Suitability of Lake Erie for bigheaded carps based on bioenergetic models and remote sensing. Journal of Great Lakes Research, 41(2), 358-366. doi:10.1016/j.jglr.2015.03.029

Andradi-Brown, D. A., Grey, R., Hendrix, A., Hitchner, D., Hunt, C. L., Gress, E., . . . Exton, D. A. (2017). Depth-dependent effects of culling-do mesophotic lionfish populations undermine current management? Royal Society Open Science, 4(5), 15. doi:10.1098/rsos.170027

Andradi-Brown, D. A., Vermeij, M. J. A., Slattery, M., Lesser, M., Bejarano, I., Appeldoorn, R., . . . Exton, D. A. (2017). Large-scale invasion of western Atlantic mesophotic reefs by lionfish potentially undermines culling-based management. Biological Invasions, 19(3), 939-954. doi:10.1007/s10530-016-1358-0

Anton, A., Simpson, M. S., & Vu, I. (2014). Environmental and Biotic Correlates to Lionfish Invasion Success in Bahamian Coral Reefs. Plos One, 9(9), 10. doi:10.1371/journal.pone.0106229

Azevedo, P. A., Bureau, D. P., Leeson, S., & Cho, C. Y. (2002). Growth and efficiency of feed usage by Atlantic salmon (Salmo salar) fed diets with different dietary protein: Energy ratios at two feeding levels. Fisheries Science, 68(4), 878-888. doi:10.1046/j.1444-2906.2002.00506.x

Bartell, S. M., Breck, J. E., Gardner, R. H., & Brenkert, A. L. (1986). INDIVIDUAL PARAMETER PERTURBATION AND ERROR ANALYSIS OF FISH BIOENERGETICS MODELS. Canadian Journal of Fisheries and Aquatic Sciences, 43(1), 160-168. doi:10.1139/f86-018

Beauchamp, D. A., Stewart, D. J., & Thomas, G. L. (1989). CORROBORATION OF A BIOENERGETICS MODEL FOR SOCKEYE-SALMON. Transactions of the American Fisheries Society, 118(6), 597-607. doi:10.1577/1548-8659(1989)118<0597:coabmf>2.3.co;2

Beaugrand, G. (2009). Decadal changes in climate and ecosystems in the North Atlantic Ocean and adjacent seas. Deep-Sea Research Part Ii-Topical Studies in Oceanography, 56(8-10), 656-673. doi:10.1016/j.dsr2.2008.12.022

78 Benkwitt, C. E. (2013). Density-Dependent Growth in Invasive Lionfish (Pterois volitans). Plos One, 8(6), 5. doi:10.1371/journal.pone.0066995

Betancur-R, R., Hines, A., Acero, A., Orti, G., Wilbur, A. E., & Freshwater, D. W. (2011). Reconstructing the lionfish invasion: insights into Greater Caribbean biogeography. Journal of Biogeography, 38(7), 1281-1293. doi:10.1111/j.1365-2699.2011.02496.x

Bos, A. R., Grubich, J. R., & Sanad, A. M. (2018). Growth, site fidelity, and grouper interactions of the Red Sea lionfish Pterois miles (Scorpaenidae) in its native habitat. Marine Biology, 165(10). doi:10.1007/s00227-018-3436-6

Brandt, S. B., & Kirsch, J. (1993a). SPATIALLY EXPLICIT MODELS OF STRIPED BASS GROWTH-POTENTIAL IN CHESAPEAKE BAY. Transactions of the American Fisheries Society, 122(5), 845-869. doi:10.1577/1548-8659(1993)122<0845:semosb>2.3.co;2

Brandt, S. B., & Kirsch, J. (1993b). SPATIALLY EXPLICIT MODELS OF STRIPED BASS GROWTH-POTENTIAL IN CHESAPEAKE BAY. Transactions of the American Fisheries Society, 122(5), 845-869. doi:10.1577/1548-8659(1993)122<0845:semosb>2.3.co;2

Brandt, S. B., & Mason, D. M. (2003). Effect of nutrient loading on Atlantic menhaden (Brevoortia tyrannus) growth rate potential in the Patuxent River. Estuaries, 26(2A), 298-309. doi:10.1007/bf02695968

Burril, S. E., von Biela, V. R., Hillgruber, N., & Zimmerman, C. E. (2018). Energy allocation and feeding ecology of juvenile chum salmon (Oncorhynchus keta) during transition from freshwater to saltwater. Polar Biology, 41(7), 1447-1461. doi:10.1007/s00300-018-2297-2

Campbell, M. D., Rose, K., Boswell, K., & Cowan, J. (2011). Individual-based modeling of an artificial reef fish community: Effects of habitat quantity and degree of refuge. Ecological Modelling, 222(23-24), 3895-3909. doi:10.1016/j.ecolmodel.2011.10.009

Cerino, D., Overton, A. S., Rice, J. A., & Morris, J. A. (2013). Bioenergetics and Trophic Impacts of the Invasive Indo-Pacific Lionfish. Transactions of the American Fisheries Society, 142(6), 1522- 1534. doi:10.1080/00028487.2013.811098

Cogan, C. B., Todd, B. J., Lawton, P., & Noji, T. T. (2009). The role of marine habitat mapping in ecosystem-based management. Ices Journal of Marine Science, 66(9), 2033-2042. doi:10.1093/icesjms/fsp214

Cooke, S. L., & Hill, W. R. (2010). Can filter-feeding Asian carp invade the Laurentian Great Lakes? A bioenergetic modelling exercise. Freshwater Biology, 55(10), 2138-2152. doi:10.1111/j.1365- 2427.2010.02474.x

Craig, S. R., MacKenzie, D. S., Jones, G., & Gatlin, D. M. (2000). Seasonal changes in the reproductive condition and body composition of free-ranging red drum, Sciaenops ocellatus. Aquaculture, 190(1-2), 89-102. doi:10.1016/s0044-8486(00)00386-0

79

Cure, K., Benkwitt, C. E., Kindinger, T. L., Pickering, E. A., Pusack, T. J., McIlwain, J. L., & Hixon, M. A. (2012). Comparative behavior of red lionfish Pterois volitans on native Pacific versus invaded Atlantic coral reefs. Marine Ecology Progress Series, 467, 181-192. doi:10.3354/meps09942

Dahl, K. A., & Patterson, W. F. (2014). Habitat-Specific Density and Diet of Rapidly Expanding Invasive Red Lionfish, Pterois volitans, Populations in the Northern Gulf of Mexico. Plos One, 9(8), 13. doi:10.1371/journal.pone.0105852 de Leon, R., Vane, K., Bertuol, P., Chamberland, V. C., Simal, F., Imms, E., & Vermeij, M. J. A. (2013). Effectiveness of lionfish removal efforts in the southern Caribbean. Endangered Species Research, 22(2), 175-182. doi:10.3354/esr00542 de Mendiburu, F. (2020). agricolae: Statistical Procedures for Agricultural Research. R package version 1.3-2. https://CRAN.R-project.org/package=agricolae

Eddy, C., Pitt, J., Morris, J. A., Smith, S., Goodbody-Gringley, G., & Bernal, D. (2016). Diet of invasive lionfish (Pterois volitans and P-miles) in Bermuda. Marine Ecology Progress Series, 558, 193- 206. doi:10.3354/meps11838

Eggleston, D. B., Dahlgren, C. P., & Johnson, E. G. (2004). Fish density, diversity, and size-structure within multiple back reef habitats of Key West National Wildlife Refuge. Bulletin of Marine Science, 75(2), 175-204.

Espinola, L. A., Julio, H. F., & Benedito, E. (2012). Invasive non-native species of fish in upper Parana river Basin, Brazil: variations of caloric content in Cichla kelberi. Neotropical Ichthyology, 10(2), 401-408. doi:10.1590/s1679-62252012005000002

Evangelista, P. H., Young, N. E., Schofield, P. J., & Jarnevich, C. S. (2016). Modeling suitable habitat of invasive red lionfish Pterois volitans (Linnaeus, 1758) in North and South America's coastal waters. Aquatic Invasions, 11(3), 313-326. doi:10.3391/ai.2016.11.3.09

Fogg, A. Q., Evans, J. T., Peterson, M. S., Brown-Peterson, N. J., Hoffmayer, E. R., & Ingram, G. W. (2019). Comparison of age and growth parameters of invasive red lionfish (Pterois volitans) across the northern Gulf of Mexico. Fishery Bulletin, 117(3), 125-139. doi:10.7755/fb.117.3.1

Garcia, D. A., Benedito, E., & Takeda, A. M. (2010). Caloric Density of Loricariichthys platymetopon (Osteichthyes, Loricariidae) in the Upper Parana River Floodplain, Brazil. Brazilian Archives of Biology and Technology, 53(5), 1109-1118. doi:10.1590/s1516-89132010000500015

Gil-Agudelo, D. L., Cintra-Buenrostro, C. E., Brenner, J., Gonzalez-Diaz, P., Kiene, W., Lustic, C., & Perez-Espana, H. (2020). Coral Reefs in the Gulf of Mexico Large Marine Ecosystem: Conservation Status, Challenges, and Opportunities. Frontiers in Marine Science, 6, 20. doi:10.3389/fmars.2019.00807

80 Goodbody-Gringley, G., Eddy, C., Pitt, J. M., Chequer, A. D., & Smith, S. R. (2019). Ecological Drivers of Invasive Lionfish (Pterois volitans and Pterois miles) Distribution Across Mesophotic Reefs in Bermuda. Frontiers in Marine Science, 6, 12. doi:10.3389/fmars.2019.00258

Green, S. J., Akins, J. L., Maljkovic, A., & Cote, I. M. (2012). Invasive Lionfish Drive Atlantic Coral Reef Fish Declines. Plos One, 7(3), 3. doi:10.1371/journal.pone.0032596

Gress, E., Andradi-Brown, D. A., Woodall, L., Schofield, P. J., Stanley, K., & Rogers, A. D. (2017). Lionfish (Pterois spp.) invade the upper-bathyal zone in the western Atlantic. Peerj, 5, 15. doi:10.7717/peerj.3683

Hackerott, S., Valdivia, A., Green, S. J., Cote, I. M., Cox, C. E., Akins, L., . . . Bruno, J. F. (2013). Native Predators Do Not Influence Invasion Success of Pacific Lionfish on Caribbean Reefs. Plos One, 8(7). doi:10.1371/journal.pone.0068259

Hinz, H., Kaiser, M. J., Bergmann, M., Rogers, S. I., & Armstrong, M. J. (2003). Ecological relevance of temporal stability in regional fish catches. Journal of Fish Biology, 63(5), 1219-1234. doi:10.1046/j.1095-8649.2003.00244.x

Hoey, J., McCormick, M. I., & Hoey, A. S. (2007). Influence of depth on sex-specific energy allocation patterns in a tropical reef fish. Coral Reefs, 26(3), 603-613. doi:10.1007/s00338-007-0246-6

Hunt, C. L., Kelly, G. R., Windmill, H., Curtis-Quick, J., Conlon, H., Bodmer, M. D. V., . . . Exton, D. A. (2019). Aggregating behaviour in invasive Caribbean lionfish is driven by habitat complexity. Scientific Reports, 9, 9. doi:10.1038/s41598-018-37459-w

Johnston, M. W., & Bernard, A. M. (2017). A bank divided: quantifying a spatial and temporal connectivity break between the Campeche Bank and the northeastern Gulf of Mexico. Marine Biology, 164(1). doi:10.1007/s00227-016-3038-0

Johnston MW, Purkis SJ. (2011). Spatial analysis of the invasion of lionfish in the western Atlantic and Caribbean. Mar Pollut Bull.;62(6):1218-1226. doi:10.1016/j.marpolbul.2011.03.028

Jorgensen, E. H., Johansen, S. J. S., & Jobling, M. (1997). Seasonal patterns of growth, lipid deposition and lipid depletion in anadromous Arctic charr. Journal of Fish Biology, 51(2), 312-326. doi:10.1006/jfbi.1997.0436

Jud, Z. R., & Layman, C. A. (2012). Site fidelity and movement patterns of invasive lionfish, Pterois spp., in a Florida estuary. Journal of Experimental Marine Biology and Ecology, 414, 69-74. doi:10.1016/j.jembe.2012.01.015

Jud, Z. R., Layman, C. A., Lee, J. A., & Arrington, D. A. (2011). Recent invasion of a Florida (USA) estuarine system by lionfish Pterois volitans/P. miles. Aquatic Biology, 13(1), 21-26. doi:10.3354/ab00351

81 Kimball, M. E., Miller, J. M., Whitfield, P. E., & Hare, J. A. (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. doi:10.3354/meps283269

Lagasse, C. R., Knudby, A., Curtis, J., Finney, J. L., & Cox, S. P. (2015). Spatial analyses reveal conservation benefits for cold-water corals and sponges from small changes in a trawl fishery footprint. Marine Ecology Progress Series, 528, 161-172. doi:10.3354/meps11271

Le Pape, O., Delavenne, J., & Vaz, S. (2014). Quantitative mapping of fish habitat: A useful tool to design spatialised management measures and marine protected area with fishery objectives. Ocean & Coastal Management, 87, 8-19. doi:10.1016/j.ocecoaman.2013.10.018

Lesser, M. P., & Slattery, M. (2011). Phase shift to algal dominated communities at mesophotic depths associated with lionfish (Pterois volitans) invasion on a Bahamian coral reef. Biological Invasions, 13(8), 1855-1868. doi:10.1007/s10530-011-0005-z

Malpica-Cruz, L., Green, S. J., & Cote, I. M. (2019). Temporal and ontogenetic changes in the trophic signature of an invasive marine predator. Hydrobiologia, 839(1), 71-86. doi:10.1007/s10750- 019-03996-2

Marras, S., Cucco, A., Antognarelli, F., Azzurro, E., Milazzo, M., Bariche, M., . . . Domenici, P. (2015). Predicting future thermal habitat suitability of competing native and invasive fish species: from metabolic scope to oceanographic modelling. Conservation Physiology, 3. doi:10.1093/conphys/cou059

Marshak, A. R., Heck, K. L., & Jud, Z. R. (2018). Ecological interactions between Gulf of Mexico snappers (Teleostei: Lutjanidae) and invasive red lionfish (Pterois volitans). Plos One, 13(11). doi:10.1371/journal.pone.0206749

Munoz, R. C., Currin, C. A., & Whitfield, P. E. (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-U494. doi:10.3354/meps09154

Nuttall, M. F., Johnston, M. A., Eckert, R. J., Embesi, J. A., Hickerson, E. L., & Schmahl, G. P. (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(2), 111-115. doi:10.3391/bir.2014.3.2.09

O'Farrell, S., Bearhop, S., McGill, R. A. R., Dahlgren, C. P., Brumbaugh, D. R., & Mumby, P. J. (2014). Habitat and body size effects on the isotopic niche space of invasive lionfish and endangered Nassau grouper. Ecosphere, 5(10). doi:10.1890/es14-00126.1

Peake, J., Bogdanoff, A. K., Layman, C. A., Castillo, B., Reale-Munroe, K., Chapman, J., . . . Morris, J. A. (2018). Feeding ecology of invasive lionfish (Pterois volitans and Pterois miles) in the temperate and tropical western Atlantic. Biological Invasions, 20(9), 2567-2597. doi:10.1007/s10530-018-1720-5

82

Rabalais, N. N., Turner, R. E., & Wiseman, W. J. (2002). Gulf of Mexico hypoxia, aka "The dead zone". Annual Review of Ecology and Systematics, 33, 235-263. doi:10.1146/annurev.ecolsys.33.010802.150513

Raymond, W. W., Albins, M. A., & Pusack, T. J. (2015). Competitive interactions for shelter between invasive Pacific red lionfish and native Nassau grouper. Environmental Biology of Fishes, 98(1), 57-65. doi:10.1007/s10641-014-0236-9

Rocha, L. A., Rocha, C. R., Baldwin, C. C., Weigt, L. A., & McField, M. (2015). Invasive lionfish preying on critically endangered reef fish. Coral Reefs, 34(3), 803-806. doi:10.1007/s00338-015- 1293-z

Ruttenberg, B. I., Schofield, P. J., Akins, J. L., Acosta, A., Feeley, M. W., Blondeau, J., . . . Ault, J. S. (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(4), 1051-1059. doi:10.5343/bms.2011.1108

Schofield, Pamela. (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. 10.3391/ai.2009.4.3.5.

Schwartzkopf, Brittany & Cowan, James. (2016). Seasonal and sex differences in energy reserves of red snapper Lutjanus campechanus on natural and artificial reefs in the northwestern Gulf of Mexico. Fisheries Science. 83. 10.1007/s12562-016-1037-1.

Searcy, S. P., Eggleston, D. B., & Hare, J. A. (2007). Is growth a reliable indicator of habitat quality and essential fish habitat for a juvenile estuarine fish? Canadian Journal of Fisheries and Aquatic Sciences, 64(4), 681-691. doi:10.1139/f07-038

Slattery, M., Lesser, M. P., Brazeau, D., Stokes, M. D., & Leichter, J. J. (2011). Connectivity and stability of mesophotic coral reefs. Journal of Experimental Marine Biology and Ecology, 408(1- 2), 32-41. doi:10.1016/j.jembe.2011.07.024

Smith, N.S., Shurin, J.B. (2010). Artificial structures facilitate lionfish invasion in marginal Atlantic habitats. Proc Gulf Caribb Fish Inst 63: 342−344

South, J., Dick, J. T. A., McCard, M., Barrios-O'Neill, D., & Anton, A. (2017). Predicting predatory impact of juvenile invasive lionfish (Pterois volitans) on a crustacean prey using functional response analysis: effects of temperature, habitat complexity and light regimes. Environmental Biology of Fishes, 100(10), 1155-1165. doi:10.1007/s10641-017-0633-y

Switzer, T. S., Tremain, D. M., Keenan, S. F., Stafford, C. J., Parks, S. L., & McMichael, R. H. (2015). Temporal and Spatial Dynamics of the Lionfish Invasion in the Eastern Gulf of Mexico:

83 Perspectives from a Broadscale Trawl Survey. Marine and Coastal Fisheries, 7(1), 8. doi:10.1080/19425120.2014.987888

Tamburello, N., & Cote, I. M. (2015). Movement ecology of Indo-Pacific lionfish on Caribbean coral reefs and its implications for invasion dynamics. Biological Invasions, 17(6), 1639-1653. doi:10.1007/s10530-014-0822-y

Wu, R. S. S. (2002). Hypoxia: from molecular responses to ecosystem responses. Marine Pollution Bulletin, 45(1-12), 35-45. doi:10.1016/s0025-326x(02)00061-9

84 Conclusion

Lionfish are well established in the Atlantic and remain of great ecological concern. Future management directions are crucial to mitigating the damage of this invasive species, and complete eradication remains impossible. Better understanding the impacts of current management strategies, the trophic interactions of lionfish, and the preferred habitat of the fish are essential to making effective, efficient, and informed decisions for lionfish control programs.

This study was concerned with lionfish across a depth gradient in particular. There is currently a discrepancy between the area we are managing via spearfishing, and the actual depth tolerance and distribution of invasive lionfish.

I found that lionfish size weakly increased with increasing depth and that shallow lionfish had higher reproductive potential compared to deeper lionfish in the Florida Keys. Sex ratios were female-biased in shallow sites. It appeared the shallow spearfishing pressure in the Keys had begun to alter the population structure of this emerging fishery.

As to lionfish diet across depths, lionfish are eating a range of teleosts and decapod species, including those of economic importance, and deeper lionfish are eating at a higher trophic level than shallow fish. It was concluded that lionfish are eating more teleost at depth in the Florida

Keys, a finding with ecological implications in the region and beyond.

A bioenergetic-based growth model for lionfish in the Gulf of Mexico was developed using the diet information gathered in this study. Prey energy density was modified to account for deeper (>30 m) lionfish’s increased consumption of teleost prey compared to shallow counterparts and bottom temperatures were used to calculate the growth rate potential (GRP) of lionfish regionally and seasonally at the locations of potential structured habitat. There were regional differences in potential habitat that are not reflected in recorded lionfish sightings.

85 Seasonal temperature changes do not limit lionfish growth in the GOM, implying that winter temperature do not significantly hinder lionfish invasion or growth in the GOM. A GRP model can be a useful tool to identify areas of lionfish growth and inform where management efforts can be focused for removals, particularly for lionfish at depth. These deep sites can be logistically difficult to access and modeling provides a relatively cost-effective tool to assess potential hot spots for growth and targeted removal efforts. This model could also be extended outside of the GOM to provide a management tool for the entirety of the lionfish’s invaded range.

86 VITA

Hanna Bauer was born in Portland, Oregon and attended the University of Portland for her bachelor’s degree in biology. As an undergraduate student, she was an active member of the

Honor’s Program and spent a year abroad studying art and philosophy in Salzburg, Austria.

Upon graduating, she worked as a field botanist, sea turtle patroller, outreach coordinator, research assistant, and science teacher across the U.S. and Central America before attending

Louisiana State University for her master’s degree in Oceanography and Coastal Science. She will graduate in August 2020 and begin a fellowship with the National Academy of Science,

Engineering, and Math’s Gulf Research Program to better understand how to implement science into effective policy.

87