Restored Coastal Habitat Can “Reel In” Juvenile Sportfish: Population and Community Responses in the Indian River Lagoon, Florida, USA

Abstract

Coastal habitats are declining worldwide, which has impacted economically important fisheries, especially in the Indian River Lagoon, Florida. As a result, extensive intertidal oyster reef and living shoreline restoration projects have been implemented. Restoration can also theoretically benefit predator populations, but this relationship is understudied. Here, the impact of habitat restoration on juvenile predatory fish (i.e., sportfish) populations (abundance) and communities (species richness, diversity, and assemblage) was assessed prior to and following oyster reef restoration and living shoreline stabilization for up to three years, and incorporated the influence of 17 environmental predictor variables. Juvenile sportfish abundance and richness (n = 11) were variable over time but collectively higher on restored oyster reefs compared to controls, and similar between control and stabilized shorelines. Sportfish abundance was best described by a combination of biotic features of the site (e.g., reef height and benthic substrate cover), prey abundance, decreasing distance to the nearest ocean inlet and dissolved oxygen. Results suggest future restoration site selection should emphasize adequate dissolved oxygen (~6 mg/L), oyster densities above 50/m² and reef height above 55 mm, and minimum shoreline vegetation coverage of 50% to support macrofaunal prey and subsequently attract sportfish. These findings can help natural resource managers better use habitat restoration as a tool for enhancing fish populations in the future.

1. Introduction

Biodiversity is being lost at an alarming rate, largely due to habitat alteration, which has cascading impacts on ecosystems and the services they provide [1,2,3,4]. Coastal ecosystems outpace other imperiled habitats for rates of loss, as more than 40% of the human population resides within 100 km of the coast, generating multiple synergistic stressors that negatively impact these systems [5]. To combat these challenges and help mitigate biodiversity losses, there has been a growing trend toward restoring habitat to improve degraded ecosystems and thus their services [6], particularly in coastal environments [7]. Restoration studies generally focus on the organisms being restored but may also monitor animal responses to restoration. However, the majority of studies assess lower trophic level arthropods and fishes, or occasionally birds [8], while the response of higher trophic level predators (e.g., sportfish) remains relatively understudied [9].

Predator populations play an important role in regulating ecosystem health, yet are disproportionately impacted by anthropogenic threats that act synergistically, such as overharvesting and habitat loss [10]. These species are generally the first to be lost from a system and their low functional redundancy can lead to repercussions that extend throughout the ecosystem [10], resulting in a phenomenon known as “trophic downgrading” [11]. As such, predators can serve as “sentinels” to reflect ecosystem decline [12]. Fortunately, predators have also demonstrated the potential to recolonize formerly occupied habitat [13]. In a similar way, predators can potentially serve as indicators of ecosystem recovery, making them ideal candidates to reflect restoration success. Theoretically, habitat restoration provides refugia for prey and foraging opportunities for predators, thereby benefitting predator populations, but empirical evidence of this trophic enhancement is lacking.

Few coastal habitat restoration studies have focused on the responses of predatory fishes [9,14]. Restoration of wild fishery stocks is of interest to combat global fishing pressure and subsequent declines, particularly in higher trophic levels [15]. Habitat restoration provides an opportunity to alleviate some population declines by creating a nursery habitat, thereby benefitting juveniles that ultimately recruit into the fishery. Both intertidal and subtidal oyster reefs have been restored to combat severe habitat losses (up to 90% in some locations), where beyond the targeted habitat improvements, there is often an objective to enhance finfish fisheries [16,17,18]. Indeed, restoration has improved habitats for oyster reef dependent fishes [18], enhancing populations and biomass by up to 289 g/m² annually [17,19], which results in estimated economic benefits for local commercial fisheries of up to USD 4123/ha annually in the southeast United States [20]. In addition, living shoreline stabilization projects that improve nearshore habitats by stabilizing the shoreline from erosion through native vegetation and often a breakwater to attenuate wave energy, have enhanced fish populations [14] within three years post stabilization [21,22].

Estuarine systems support the majority of U.S. saltwater recreational fishing catch (55%), particularly for sportfish [23]. Sportfish are defined, here, as higher trophic level fishes that are targeted by anglers. They are particularly important in Florida, earning it the reputation as the “Sportfishing Capital of the World,” as the state supports the majority of the total U.S. recreational fishing effort, harvest, number of anglers [23], and related sales (e.g., USD 10.9 billion in 2016) [24]. Therefore, sportfish provide significant ecological and economic value. Although few studies have targeted the response of sportfish, several have found that more sportfish are associated with restored subtidal reefs relative to degraded reefs [25,26,27]. In addition, adjacent habitat can influence their presence at restored reefs based on the proximity of functionally redundant habitats [28]. Shoreline restoration has also improved sportfish densities and species richness, relative to impacted control shorelines [14,29]. Therefore, there is potential to enhance these species through habitat restoration, which is of interest to anglers and natural resource managers.

In this study, we assessed sportfish abundance and diversity before and after intertidal oyster reef restoration and living shoreline stabilization, to determine how sportfish populations and community dynamics respond to habitat restoration. We hypothesize that if restored areas improve habitat and subsequent prey base, then local sportfish would be attracted to the sites; thus, abundances and diversity would be enhanced following restoration relative to pre sampling. We predict sportfish abundance and diversity will be (1) higher at restored and natural habitats compared to degraded habitats, specifically with restored reefs intermediate to natural and degraded oyster reefs, and (2) increase at restored sites over time. Further, we predict that (3) habitat features (e.g., intermediate reef height and oyster density at oyster reefs and a high percent cover of vegetation at living shorelines) would correspond to higher sportfish abundances and diversity, and (4) sportfish communities will differ between oyster reef and living shoreline habitats based on differences in benthic composition.

2. Materials and Methods

2.1. Study Site

Sportfish are vital to the Indian River Lagoon (IRL) due to their role in the food web, in their position as a top consumer, and appeal to recreational anglers. The IRL is an estuarine system covering 40 percent (251 km) of Florida’s Atlantic coastline. It supports one of the most biodiverse estuaries in the United States, where a dynamic biogeographic transition zone blends species of temperate and subtropical climates [30,31,32,33]. The IRL supports a USD 7.6 billion economic impact on the region, of which the fishing of living resources contributes USD 48 million annually [34]. However, the IRL is considered an imperiled ecosystem, largely due to declining water quality from nutrient enrichment, leading to a significant decline in the areal coverage of seagrass and oyster reef habitat [35,36]. This habitat loss leads to concomitant declines in ecologically and economically important fishes [37], as the IRL commercial finfish industry declined in pounds landed (−57%) and value (−37%) between 1990 and 2015 [34].

Mosquito Lagoon is the northernmost subbasin of the IRL, with limited tidal exchange via the Ponce de Leon Inlet at its northern extent (Figure 1). Mosquito Lagoon is characterized by a shallow depth (average 1 m), salinity ranging from 18 to 45 ppt, and temperature ranging from 14.1 to 37.7 °C [33]. Oyster reef (Crassostrea virginica) habitat has declined by 24% within Mosquito Lagoon due to various stressors and disturbances, largely due to oyster displacement from boat wakes, leading to oyster death and piles of dead shell [38,39]. To reverse this benthic habitat loss, over 80 oyster reefs and 8 living shorelines have been restored over the past decade [38,39,40]. Thus, as part of this effort, 8 additional oyster reefs and 4 living shorelines were restored and stabilized in Mosquito Lagoon (Figure 1).

2.2. Restoration Methods

Oyster reef restoration involved grading dead shell mounds to low intertidal elevation and then anchoring stabilized oyster shell (i.e., “oyster mats”) to the site to provide a substrate for oyster recruitment [38,39]. Oyster mats are made of 36 vertically oriented, dead, disarticulated oyster shells secured to 1 cm² Vexar^® extruded polyethylene mesh with cable ties through a drilled hole in the umbo of the shell. Oyster mats are anchored to the site with four concrete weights attached to neighboring mats with 150-pound test cable ties. Oyster larvae recruit to the shells and form a restored reef as they grow over time [39], showing similar biogeochemical characteristics (e.g., dissolved organic carbon, and total carbon and nitrogen) to live reefs within one year [40]. Restoration of four sites occurred in June 2017 and four additional sites in June 2018. These restored reefs were paired with four dead reefs, serving as negative controls (hereafter referred to as “dead” reefs), and four live, natural reefs serving as positive controls (hereafter referred to as “live” reefs) (Figure 1A). As oyster reef decline in Mosquito Lagoon is largely attributed to boat wake displacement of oysters, dead reefs in need of restoration are often located along busy boating channels [39]. Dead reefs that were good candidates for restoration were selected for height above mean high water, slope, and proximity to boating channels. Suitable proximate control reefs were then identified, with similar attributes in depth and adjacent habitat.

Living shoreline stabilization was conducted to reduce shoreline erosion and improve habitat quality using a shoreward oyster breakwater and vegetation plantings. The oyster breakwater consisted of stacked oyster mesh bags (two bags high) at the mean low-water line to attenuate wave energy [41]. Oyster bags consisted of Naltex^® plastic mesh bags filled with ~19 L of disarticulated, recycled oyster shells collected from restaurants and quarantined outdoors for over six months. Individual bags were 1 m long by 0.4 m wide and weighed ~18 kg, and were hand knotted at each end and attached to neighboring bags with cable ties [41]. Shoreline vegetation was planted behind the breakwater to stabilize the shoreline and mimic vegetation of natural shorelines in this area with marsh grass Spartina alterniflora (three plants/meter), followed landward by mangroves (Rhizophora mangle, Laguncularia racemosa and Avicennia germinans; two plants/m) [41]. There were four shorelines stabilized in July 2017 and paired with three natural, control shorelines (i.e., one control site was shared by two neighboring stabilized shorelines; Figure 1B). One stabilized and control site were dropped from monitoring following damage incurred by Hurricane Irma in September 2017 and were not included in analyses, leaving three stabilized and two control shorelines for the remainder of the study.

2.3. Field Collections

All restoration and control sites were sampled prior to and following restoration/stabilization. To assess the development of restored sites relative to positive and negative controls over time, each oyster reef was monitored for changes in number of live oysters and reef height quarterly. Oyster counts were performed on five randomly selected 0.25 m² quadrats and then averaged to estimate oyster density, while reef height (mm) was measured at five haphazardly placed points in ten quadrats and averaged to estimate vertical accretion above the sediment [39]. These have proven to be suitable metrics to monitor restoration success by implying structural complexity and quantifying fine scale accretion of the low relief intertidal reefs found in Mosquito Lagoon [39,40]. Living shorelines were monitored for changes in percent cover of (1) seagrass (i.e., natural presence of Halodule wrightii and not experimentally transplanted), (2) vegetation (e.g., S. alterniflora, R. mangle) and (3) bare substrate averaged across 0.25 m² quadrats every 1 m along four 3-meter long subtidal to intertidal transects per site [41]. Seagrass and vegetation percent cover were combined for analyses to represent non-bare substrate.

All fish sampling was conducted during daylight hours 2 weeks prior to restoration and following restoration at 1, 2, 4, 6 and 8 weeks, and every 12 weeks thereafter for three years at living shorelines and 2017 restored reefs, and for two years at 2018 restored reefs. These sportfish species are generally transient in nature, but some have demonstrated small ranges and a primary home site (<250 m), which likely reflects foraging and refugia behavior, in a similar heterogeneous coastal habitat in Florida [42]. Efforts were made to sample fishes at all sites of the respective habitat within the same 12–36 h window to reduce temporal sampling variation. Fish were sampled using a 21.3-meter long by 1.8-m deep center bag seine with 3.175 mm mesh. The seine net was pulled parallel to the shore for approximately 12 m to maintain consistency across sites (mean site length: 13.4 m) immediately seaward (~0.5 m) of the reef margin to avoid gear disruption from the oysters and maintain consistency in shallow, subtidal sampling. The abundance of all fish species captured in each sample was recorded to quantify prey fish (e.g., gobies Gobiidae, anchovies Engraulidae). In addition, prey macroinvertebrate (e.g., mud crabs, Panopeidae; porcelain crab Petrolisthes armatus; brown shrimp Farfantepenaeus duorarum) abundance data were collected using lift nets (n = 6/site) and averaged per site. These nets consisted of a 0.25 m² PVC quadrat lined with 1.5 mm mesh and an oyster mat (oyster reefs) or shell bag (living shorelines) placed in the center to weigh the net down and mimic the site substrate and soaked for one week before retrieval [43,44]. These data were not the focus of this study but were incorporated as predictors of sportfish abundance in the models described below.

Abiotic data from the individual sites were collected at the time of sampling using a YSI ProDSS multiparameter probe placed ~0.5 m beneath the water surface, which included water temperature (°C), dissolved oxygen (“DO”, mg/L), salinity (ppt) and secchi depth (m). In addition, air temperature (°C), barometric pressure (mm Hg), and wind speed (kph) were included as potential variables to influence temporal trends in sportfish distribution. These were collected from the nearest National Oceanic and Atmospheric Administration (NOAA) National Center for Environmental Information station located approximately 15 km north from the center of our study area at New Smyrna Beach Municipal Airport (29.05417°, −80.94833°). Local tide table data at Packwood Place, Florida and lunar data were obtained from NOAA Tides and Currents (www.tidesandcurrents.noaa.gov, accessed on 2 November 2020).

On each sampling date, all collected sportfish were transferred to a sorting tub filled with natural seawater. Fish were identified and enumerated in the field, while unidentified specimens were humanely euthanized in 90% ethanol and stored in a −20 °C freezer until identification in the lab occurred in accordance with the University of Central Florida Institutional Animal Care and Use Committee protocol (IACUC Permit # 16-15 W).

2.4. Data Analysis

Sportfish abundance in response to restoration was assessed using generalized linear models (GLM) with a negative binomial distribution and log link function to weigh the influence of the predictors site type (restored vs. controls), biotic factors (substrate composition and macrofaunal abundance) and abiotic factors (listed above) using the MASS package in R statistical software (Table S1) [45]. All predictors were checked for collinearity prior to modeling using the Variable Inflation Factor in the R cars package. Lunar cycle and tide, and the combination of lunar cycle, water temperature and season, were collinear for oyster reefs. For living shorelines, lunar cycle and water temperature, and bare and vegetated substrate percent covers, were collinear, so these were not modeled together. Differences in catch per unit effort, here defined as the abundance of sportfish per number of samples at each site type, were assessed using a Kruskal–Wallis test, followed by Wilcoxon Rank Sum tests.

The development of GLMs was based on hypothesis testing of predicted influential variables on sportfish abundance and variable collinearity results. These a priori predictions focused on site type and benthic site metrics (e.g., oyster density). They were also modeled in combination with abiotic factors thought to influence fish movement and life history, which was also guided by exploratory examination of the data. For example, these site habitat metrics were additively combined with biotic variables such as prey macrofauna abundance at the site, and abiotic features considered to reflect site variation (e.g., DO, salinity). These variables were also modeled individually and compared to additive models to consider their influence alone. Although the control sites remained the same, 2017 restored oyster reefs and controls were analyzed separately from 2018 restored reefs and controls so as not to inflate the values for restored reefs and maintain consistency in the sampling timeline for their respective restoration year (i.e., this kept the analyses at 4 restored, 4 live and 4 dead reefs for each year, versus a combined 8 restored reefs with those same controls).

Since the response variable is count data which are often zero inflated by nature (e.g., 53.4% of the samples here caught zero fish), the standard negative binomial distribution was compared against zero inflated Poisson and zero inflated negative binomial models using the Akaike Information Criterion (AIC) in the R bbmle package. AIC weights imply relative probability, with the highest weight, and lowest AICc score and delta AIC, suggesting the most probable model with the most likely influential predictors. Both the standard binomial and zero inflated negative binomial reduced overdispersion, but analyses proceeded with the standard negative binomial GLM, as it had the highest AICc weight and lowest AICc score compared to the zero inflated models. Generalized linear mixed effect models (GLMM) were also performed, to examine the influence of restoration timeline and site number as random effects, with model comparison and weighting conducted in the R lme4 package. Since the GLMMs performed poorly with some of the highest AICc scores, these random effects were not utilized in other models.

Diversity data were compared between habitat types (i.e., oyster reefs vs. living shoreline) and separately within habitats (i.e., across restoration and control sites) using the R vegan package. The Shannon–Wiener index (H’) was used to estimate species diversity, as it accounts for both abundance and evenness (i.e., rare and numerically dominant species). Subsequently, these values were converted to effective number of species by transforming this entropic index into the observed value of diversity (exp[H’]) [46]. The R indicspecies package was used to identify indicator species at the respective habitats (oyster vs. shoreline) and sites (restored vs. controls); these are species that can reflect aspects of community composition, such as those that are more common to one treatment versus another [47]. The indicspecies package proposes an improvement upon the IndVal index, as it allows a species to be an indicator of several types of sites (or cluster of sites), rather than only one site type, as in the IndVal index. This improvement enables a characterization of the whole suite of habitats of which one species is an indicator, and not only the sites where it is present [47]. In addition, multivariate analyses employed nonmetric multidimensional scaling (NMDS) using Bray–Curtis similarity to quantify the similarities in community composition between habitats and site types. The influence of biotic and abiotic variables was assessed using an environmental fit test. Finally, permutational multivariate analysis of variance (PERMANOVA) was used to test for differences among habitats and sites using the adonis function in the R vegan package.

3. Results

In total, 1143 individuals across 11 sportfish species were captured at the study sites from 2017–2020 (

Table 1. Total abundance of each sportfish species associated with living shoreline and oyster reef restored and control sites. Catch per unit effort (CPUE) is calculated as abundance per number of samples collected per site type. CPUE for dead and live oyster sites here represents all samples for all three years and restored oyster sites are shown by restore year (i.e., 2017 restored reefs = three years, restored 2018 reefs = two years).

Species		Living Shoreline			Oyster Reefs					Total
Common Name	Scientific Name	Control	Stabilized	Total	Dead	Live	Restore 2017	Restore 2018	Total
Bonefish	Albula vulpes	0	0	0	0	1	1	0	2	2
Barracuda	Sphyraena barracuda	1	0	1	0	0	1	0	1	2
Crevalle Jack	Caranx hippos	0	1	1	0	1	0	1	2	3
Horse-eye Jack	Caranx latus	1	0	1	1	1	0	0	2	3
Common Snook	Centropomus undecimalis	16	20	36	2	6	2	1	11	47
Spotted Seatrout	Cynoscion nebulosus	440	264	704	0	4	0	0	4	708
Gray/Mangrove Snapper	Lutjanus griseus	35	62	97	41	46	62	43	192	289
Lane Snapper	Lutjanus synagris	0	0	0	5	18	9	44	76	76
Black Drum	Pogonias cromis	2	4	6	0	0	0	0	0	6
Red Drum	Sciaenops ocellatus	3	1	4	2	0	0	0	2	6
Permit	Trachinotus falcatus	0	1	1	0	0	0	0	0	1
Total		498	353	851	51	77	75	89	292	1143
CPUE		16.6	7.4	10.9	0.6	0.9	1.2	1.9	1.1
Richness		7	7	9	5	7	5	4	9	11

). Oyster reefs accounted for 292 individuals of nine species, while living shorelines accounted for 851 individuals of nine species. Four species, common snook Centropomus undecimalis, spotted seatrout Cynoscion nebulosus, mangrove/grey snapper Lutjanus griseus, and lane snapper Lutjanus synagris accounted for 98% of the sportfish captured and were all juveniles according to their length [48,49,50], with some L. griseus reaching the subadult phase (~100–250 mm) [51]. The most common species was C. nebulosus (n = 708), followed by L. griseus (n = 289), accounting for 61.9% and 25.3% of all sportfish, respectively (Table 1). A few rare species were caught on three or fewer occasions (Bonefish, Albula vulpes; Barracuda Sphyraena barracuda, Crevalle Jack Caranx hippos, and Horse-eye Jack C. latus).

3.1. Oyster Reefs

Juvenile sportfish abundance at oyster reefs varied over time, including seasonal oscillations, and showed similar patterns between live and 2017 restored reefs (Figure 2A). Abundance at 2018 restored reefs experienced a pulse at three months post restoration. Dead reefs had lower mean abundance on all sampling events but 3 (of 17) occasions in 2017 and 1 (of 12) occasion at 2018 reefs (Figure 2A). Abundance was collectively higher at 2018 (n = 88) restored reefs than 2017 (n = 76), despite 2017 reefs being restored a year prior (Table 1). Similarly, catch per unit effort was highest at 2018 restored reefs (Kruskal–Wallis p = 0.02, df = 3, H = 9.7), followed by 2017 restored, live, and then dead reefs (Table 1; Figure 2B).

These abundances corresponded to changes in reef topography following restoration. Restored oyster reefs increased in live oyster density and reef height within the first year and over time as the reef progressed from a dead to live reef, reaching similar densities and thickness to live reefs by 1 and 2 years, respectively [39]. These metrics remained consistently high at live and restored reefs, while dead reefs remained consistently low throughout the study (

Table 2. Oyster reef and living shoreline metrics with related mean sportfish abundance and their top GLM predictor variables (Mean ± SE). Dead and live oyster sites represent all samples for all three years and restored oyster sites are shown by restore year.

Oyster Reefs
Site Type	Sportfish CPUE	Oyster Density	Reef Height (mm)	DO (mg/L)	Invert CPUE	Inlet Distance (km)
Total	1.1 +/− 0.1	59.5 +/− 4.9	52.8 +/− 2.5	6.3 +/− 0.1	66.2 +/− 3.3	16.8 +/− 0.5
Dead (−control)	0.6 +/− 0.1	4.0 +/− 0.9	22.8 +/− 3.6	6.5 +/− 0.1	57.5 +/− 5.3	17.6 +/− 1.2
Live (+control)	0.9 +/− 0.2	106.1 +/− 9.0	81.9 +/− 3.4	6.5 +/− 0.1	72.1 +/− 4.9	17.6 +/− 1.0
Restore 2017	1.1 +/− 0.3	62.5 +/− 9.7	51.2 +/− 5.2	6.6 +/− 0.1	58.5 +/− 7.8	16.05 +/− 0.8
Restore 2018	1.9 +/− 0.5	65.2 +/− 9.1	54.1 +/− 3.4	5.5 +/− 0.2	80.3 +/− 9.2	15.2 +/− 0.8
Living Shoreline
Site Type	Sportfish CPUE	Vegetation % Cover	Bare % Cover	DO (mg/L)	Invert CPUE	Prey Fish CPUE
Total	10.9 +/− 2.1	37.2 +/− 2.5	62.8 +/− 2.5	6.2 +/− 0.1	93.5 +/− 8.1	438.3 +/− 49.8
Control	17.8 +/− 6.3	42.1 +/− 4.1	57.9 +/− 4.1	6.4 +/− 0.3	82.6 +/− 10.1	487.8 +/− 88.0
Stabilized	8.4 +/− 3.6	34.0 +/− 3.0	66.0 +/− 3.0	6.0 +/− 0.2	97.6 +/− 10.5	405.2 +/− 59.0

,

Table A1. List of the top predictor variables (Mean ± SE), identified by GLMs, listed by site type and site number for oyster reefs and living shorelines. Top predictor variables were identified as oyster density (number of oysters/m²), reef height (mm), prey macroinvertabrate abundance, and prey fish abundance (i.e., all fishes, except sportfish), and percent vegetation cover (shorelines).

Oyster Reefs
Site Number	Sportfish CPUE	Oyster Density	Reef Height (mm)	DO (mg/L)	Invert CPUE	Inlet Distance (km)
Total	1.1 +/− 0.1	59.5 +/− 4.9	52.8 +/− 2.49	6.3 +/− 0.1	66.2 +/− 3.3	16.8 +/− 0.5
Dead (−control)	0.6 +/− 0.1	4.0 +/− 0.9	22.8 +/− 3.6	6.5 +/− 0.1	57.5 +/− 5.3	17.6 +/− 1.2
Dead 1	1.0 +/− 0.3	5.8 +/− 2.3	25.1 +/− 2.0	6.0 +/− 0.3	57.4 +/− 12.3	14.3
Dead 2	0.6 +/− 0.3	0.07 +/− 0.04	18.1 +/− 0.8	6.7 +/− 0.2	37.8 +/− 4.2	17.6
Dead 3	0.5 +/− 0.2	9.5+/− 1.8	32.6 +/− 2.0	6.5 +/− 0.3	68.1 +/− 11.4	19.4
Dead 4	0.2 +/− 0.1	0.4 +/− 0.2	16.8 +/− 1.2	6.6 +/− 0.3	66.6 +/− 11.9	19.1
Live (+control)	0.9 +/− 0.2	106.1 +/− 9.0	81.9 +/− 3.4	6.5 +/− 0.1	72.1 +/− 4.9	17.6 +/− 1.0
Live 1	1.8 +/− 0.6	160.1 +/− 20.3	92.7 +/− 6.5	5.9 +/− 0.3	50.8 +/− 8.7	14.6
Live 2	0.6 +/− 0.2	89.0 +/− 11.7	76.8 +/− 7.1	6.7 +/− 0.3	71.3 +/− 9.7	18.1
Live 3	0.2 +/− 0.1	96.9 +/− 18.2	85.6 +/− 6.7	6.5 +/− 0.2	80.3 +/− 7.6	18.7
Live 4	0.8 +/− 0.3	78.5 +/− 12.0	72.9 +/− 6.2	6.8 +/− 0.3	86.1 +/− 11.1	19.0
Restore 2017	1.1 +/− 0.3	62.5 +/− 9.7	51.2 +/− 5.2	6.6 +/− 0.1	58.5 +/− 7.8	16.05 +/− 0.8
Restore 1	2.8 +/− 0.8	103.5 +/− 24.0	69.4 +/− 13.0	6.2 +/− 0.3	65.0 +/− 18.6	14.5
Restore 2	0.4 +/− 0.2	48.6 +/− 15.1	49.0 +/− 7.8	7.0 +/− 0.2	65.9 +/− 14.3	17.5
Restore 3	0.5 +/− 0.3	20.1 +/− 6.3	33.4 +/− 6.4	6.9 +/− 0.3	37.9 +/− 9.8	17.5
Restore 4	0.8 +/− 0.2	69.3 +/− 16.4	48.5 +/− 9.5	6.2 +/− 0.3	61.9 +/− 17	14.7
Restore 2018	1.9 +/− 0.5	65.2 +/− 9.1	54.1 +/− 3.4	5.5 +/− 0.2	80.3 +/− 9.2	15.2 +/− 0.8
Restore 5	0.7 +/− 0.3	45.1 +/− 13.0	48.6 +/− 5.6	5.8 +/− 0.4	81.3 +/− 19.9	17.6
Restore 6	1.2 +/− 0.5	58.0 +/− 11.7	58.1 +/− 5.6	5.9 +/− 0.4	80.0 +/− 21.4	14.9
Restore 7	2.8 +/− 0.8	78.3 +/− 23.3	52.9 +/− 7.2	5.2 +/− 0.4	71.5 +/− 13.4	14.3
Restore 8	2.8 +/− 1.6	79.4 +/− 22.3	56.9 +/− 9.3	5.4 +/− 0.3	89.1 +/− 21.8	14.1
Living Shoreline
Site Number	Sportfish CPUE	Vegetation % Cover	Bare % Cover	DO (mg/L)	Invert CPUE	Prey Fish CPUE
Total	10.9 +/− 2.1	37.2 +/− 2.5	62.8 +/− 2.5	6.2 +/− 0.1	93.5 +/− 8.1	438.3 +/− 49.8
Control	17.8 +/− 6.3	42.1 +/− 4.1	57.9 +/− 4.1	6.4 +/− 0.3	82.6 +/− 10.1	487.8 +/− 88.0
Control 1	4.9 +/− 2.1	27.2 +/− 3.5	72.8 +/− 3.5	6.6 +/− 0.4	90.2 +/− 12.8	353.6 +/− 73.0
Control 2	30.0 +/− 10.5	57.0 +/− 3.9	43.0 +/− 3.9	6.2 +/− 0.4	74.9 +/− 15.8	622.1 +/− 155.8
Stabilized	8.4 +/− 3.6	34.0 +/− 3.0	66.0 +/− 3.0	6.0 +/− 0.2	97.6 +/− 10.5	405.2 +/− 59.0
Stabilized 1	11.6 +/− 7.7	16.3 +/− 3.1	83.7 +/− 3.1	6.5 +/− 0.4	106.7 +/− 16.3	435.3 +/− 101.3
Stabilized 2	8.3 +/− 5.5	44.2 +/− 4.1	55.8 +/− 4.1	5.8 +/− 0.3	92.3 +/− 12.2	467.4 +/− 115.4
Stabilized 3	2.6 +/− 0.9	41.4 +/− 3.7	58.6 +/− 3.7	5.8 +/− 0.3	86.5 +/− 10.0	312.9 +/− 91.1

). One live site (Live 1) became somewhat degraded over time but was still included in sampling and analyses. One restored site (Restore 3) was damaged by Hurricane Dorian in 2019 and was dropped from fisheries data collection and analyses after 24 months. Sportfish abundance and live oyster density showed weak correlations, but indicate highest sportfish abundances were at ~80–100 oysters/m² (Figure 2C). Sportfish abundance and reef height similarly showed weak correlations, with bimodal peaks in abundance at ~40 mm and 75–100 mm (Figure 2D).

Correspondingly, site type and oyster reef metrics alone did not predict sportfish abundance well. However, when included with other variables of prey abundance and abiotics, these predictors were among the best performing models (

Table 3. List of top weighted generalized linear models (with negative binomial distribution and log link) of sportfish abundance, listed by habitat restoration category, with Akaike information criterion (AIC). Models, AICc scores, delta AIC, degrees of freedom and AIC model weight are in order of model rank. Top predictor variables include oyster density (number of oysters/m²), reef height (mm), prey macroinvertabrate abundance, and prey fish abundance (i.e., all fishes, except sportfish) and percent vegetation cover (shorelines). Significant predictors from the top model were included along with their coefficient estimates and significance values. Note: α < 0.1 for oyster density and DO at 2017 reefs.

Model (GLM)	AICc	Delta AIC	df	Weight	Significant Predictor	Coefficient Estimates	Pr (z)
2017 Oyster Reefs
Abundance~Oyster Density + Reef Height + Invert Abundance + Inlet Distance + DO	245.1	0	7	0.65	Inlet Distance	−0.21	0.03
Abundance~Oyster Density + Reef Height + Inlet Distance + Invert Abundance	246.5	1.4	6	0.32	Oyster Density	0.01	0.08
					DO	−0.26	0.07
2018 Oyster Reefs
Abundance~Oyster Density + Reef Height + Inlet Distance + Invert Abundance + DO	204.6	0	7	0.86	Inlet Distance	−0.3	0.004
					DO	−0.5	0.004
Living Shorelines
Abundance~Vegetation/Bare + Fish Abundance + Invert Abundance	268.4	0	5	1	Fish Abundance	0.003	<0.01
					Invert Abundance	0.014	<0.01

,

Table A2. List of all generalized linear models (with negative binomial distribution and log link), with Akaike information criterion (AIC). Generalized linear mixed effects models scored low in AIC but were listed here for comparison. Models, AICc scores, delta AIC, degrees of freedom and AIC model weight are in order of model rank.

2017 Oyster Reefs
Model (GLM)	AICc	deltaAIC	df	Weight
Abundance~Oyster Density + Reef Height + Invert Abundance + Inlet Distance + DO	245.1	0	7	0.65
Abundance~Oyster Density + Reef Height + Inlet Distance + Invert Abundance	246.5	1.4	6	0.32
Abundance~Oyster Density + Reef Height + Fish Abundance + Invert Abundance	251.2	6.1	6	0.031
Abundance~Oyster Density + Reef Height + Salinity + DO	266.9	21.8	6	<0.001
Abundance~Oyster Density + Reef Height + DO + Inlet Distance + Season	271.1	26	9	<0.001
Abundance~Reef Height + Inlet Distance + Tide	284.1	39	7	<0.001
Abundance~Oyster Density + Reef Height + Inlet Distance + Tide	286.3	41.2	8	<0.001
Abundance~Reef Height + Inlet Distance + DO	286.3	41.2	5	<0.001
Abundance~Oyster Density + Reef Height + Inlet Distance + Fish Abundance + DO	286.4	41.3	7	<0.001
Abundance~Oyster Density + Reef Height + Inlet Distance + DO	288.4	43.3	6	<0.001
Abundance~Oyster Density + Reef Height+Inlet Distance + Tide + Site Type	289.6	44.5	10	<0.001
Abundance~Oyster Density + Reef Height + Inlet Distance + Tide	290.6	45.5	5	<0.001
Abundance~Oyster Density + Reef Height + Invert Abundance + Inlet Distance + DO + Site Type	292.1	47	8	<0.001
Abundance~Oyster Density + Inlet Distance + Tide	303.6	58.5	7	<0.001
Abunance~Reef Height	305.5	60.4	3	<0.001
Abundance~Oyster Density + Site Type	308	62.9	5	<0.001
Abundance~Oyster Density	329	83.9	3	<0.001
Abundance~Invert Abund + Inlet Distance + DO	383.2	138.2	6	<0.001
Abundance~Water Temp + DO + Salinity + Secchi + Wind + Baro Pressure + Air Temp	443	197.9	9	<0.001
Abundance~Inlet Distance	507.6	262.6	3	<0.001
Abundance~Fish Abundance + Inver Abundance + DO	511.3	266.2	6	<0.001
Abundance~ DO	522.7	277.6	3	<0.001
Abundance~1 (Null Model)	527.2	282.1	2	<0.001
Abundance~Site Type	527.9	282.8	4	<0.001
Abundance~Time (months)	529.2	284.1	3	<0.001
Model (GLMM)	AICc	deltaAIC	df	weight
Abundance~Oyster Density + DO + Water Temp + Salinity + Inlet Distance + (1|Month Time)	544.2	389.3	7	<0.001
Abundance~DO + Inlet Distance + Fish Abundance + (1|Month Time)	584.7	429.5	6	<0.001
Abundance~Oyster Density + DO + Water Temp + Salinity + Inlet Distance + (1|Site Number)	590.6	435.7	6	<0.001
2018 Oyster Reefs
2018 Oyster Reefs
Model (GLM)	AICc	delta AIC	df	weight
Abundance~Oyster Density + Reef Height + Inlet Distance + Invert Abundance + DO	204.6	0	7	0.86
Abundance~Oyster Density + Reef Height + Fish Abundance + Invert Abundance	209.6	5	6	0.071
Abundance~Oyster Density + Reef Height + Inlet Distance + Invert Abundance + DO + Site Type	209.7	5	9	0.069
Abundance~Oyster Density + Reef Height + Inlet Distance + Invert Abundance	222.6	18	6	<0.001
Abundance~Oyster Density + Reef Height + Inlet Distance + DO	232	27.4	6	<0.001
Abundance~Oyster Density + Tide + Inlet Distance	232.6	27.9	7	<0.001
Abundance~Reef Height + DO + Inlet Distance	233.1	28.5	5	<0.001
Abundance~Oyster Density + Reef Height + DO + Inlet Distance + Season	233.4	28.8	9	<0.001
Abundance~Oyster Density + Reef Height + Inlet Distance + Fish Abudance + DO	233.6	28.9	7	<0.001
Abundance~Oyster Density + Reef Height + Inlet Distance + Tide	234.5	29.9	8	<0.001
Abundance~Oyster Density + Reef Height + Inlet Distance + Water Temp + Tide + Site Type	236.2	31.6	11	<0.001
Abundance~Oyster Density	242.9	38.3	3	<0.001
Abundance~Reef Height	242.9	38.3	3	<0.001
Abundance~Oyster Density + Site Type	245.6	41	5	<0.001
Abundance~ Reef Height + Tide + Inlet Distance	248.2	43.6	7	<0.001
Abundance~Invert Abundance + Inlet Distance	317.4	112.8	6	<0.001
Abundance~Water Temp + DO+Salinity + Secchi + Wind + Barometric Pressure	354.1	149.5	9	<0.001
Abundance~Fish Abundance + Invert Abundance	372.9	168.3	6	<0.001
Abundance~DO	389.1	184.5	3	<0.001
Abundance~Inlet Distance	398.6	193.9	2	<0.001
Abundance~Time (months)	399.9	195.3	3	<0.001
Abundance~Site Type	407.3	202.7	3	<0.001
Abundance~1 (Null Model)	412	207.4	2	<0.001
Model (GLMM)	AICc	delta AIC	df	weight
Abundance~Air Temp + Inlet Distance + Barometric Pressure + (1|Month)	373.5	168.8	6	<0.001
Abundance~Air Temp + Inlet Distance + Barometric Pressure + (1|Site Number)	373.5	168.9	6	<0.001
Living Shorelines
Living Shorelines
Model (GLM)	AICc	delta AIC	df	weight
Abundance~Vegetation/Bare + Fish Abundance + Invert Abundance	268.4	0	5	1
Abundance~Vegetation Cover	335.2	66.8	3	<0.001
Abundance~Bare Cover	335.2	66.8	3	<0.001
Abundance~Fish Abund + Invert Abund + DO + Site Type	376.7	108.3	6	<0.001
Abundance~Fish Abund + Invert Abund + Site Type	413.3	144.9	5	<0.001
Abundance~Lunar + Invert Abund + Fish Abund + Inlet Distance	416.1	147.7	8	<0.001
Abundance~ Fish Abund + Invert Abund	418.5	150.2	4	<0.001
Abundance~Fish Abund + Invert Abund + Lunar	422.3	153.9	7	<0.001
Abundance~Water Temp + DO + Salinity + Inlet Dist + Site Type	423.1	154.7	7	<0.001
Abundance~Water Temp + DO + Salinity + Baro Pressure + Inlet Distance + Air Temp + Wind	426.3	157.9	9	<0.001
Abundance~Inlet Distance + DO + Salinity	440.6	172.3	5	<0.001
Abundance~ DO	446.5	178.2	3	<0.001
Abundance~Invert Abundance	447.5	179.1	3	<0.001
Abundance~Water Temp + Air Temp + Secchi + Inlet Distance	462.4	194.1	6	<0.001
Abundance~Fish Abundance	462.6	194.2	3	<0.001
Abundance~Site Type	485.7	217.3	3	<0.001
Abundance~1 (Null Model)	487.6	219.2	2	<0.001
GLMMs
Abundance~Wind + Water Temp + Inlet Distance + (1|Site Number)	437.2	209.3	6	<0.001
Abundance~Wind + Water Temp + Inlet Distance + (1|Month)	437.2	209.3	6	<0.001

). Sportfish abundance for 2017 and 2018 reefs was best predicted by oyster density and reef height, distance to the Ponce de Leon Inlet, macroinvertebrate abundance and DO (AICw = 0.65, R² = 0.12 for 2017; AICw = 0.86, R² = 0.1 for 2018; Table 3, Table A2). The next plausible model predicting sportfish abundance included these variables with the exception of dissolved oxygen (Table 3, Table A2). The relationship between these variables and sportfish abundance was evident, as invertebrate abundance [44] and prey fish abundance [43,52] were similar between live and restored reefs and generally higher than dead reefs (Table A1). In addition, sportfish demonstrated increasing abundance (CPUE) with proximity to Ponce de Leon Inlet (linear model [LM] overall p < 0.001, F = 30.96, Figure 2E), across all reef types.

There was some variation in sportfish abundance and these top predictor variables across sites (Table A1; Figure S1), but displayed similar trends. Sportfish abundance (mean ± SE 1.1 ± 0.1) was highest at Live 1, and Restore 1, 7 and 8 (Table A1; Figure S1). The highest abundances at dead reefs were at Dead 1, although at significantly lower abundances. Invertebrate abundance (CPUE 66.2 ± 3.3) was similar across sites, but was highest at live and 2018 restored reefs (sites 5–8) (Table A1). Oyster densities (59.5/m² ± 4.9) were similarly highest at live sites and Restore 1, 4, 7 and 8. Reef heights (52.8 mm ± 2.5) were highest at live sites, followed by Restore 1, 4 and all 2018 restored reefs. Distance to the nearest inlet (16.8 km ± 0.5) was shortest at Dead 1, Live 1, and Restore 1, 4, 6–8. Lastly, DO was similar across sites (6.3 mg/L ± 0.1) and slightly below average at Dead 1, Live 1, Restore 1 and all 2018 restore sites. The 2018 restored reefs had lower DO (~3.9 mg/L) one week post restoration and gradually improved to that of the other sites by 6 months post restoration.

Mean sportfish species richness at oyster reefs varied over time (Figure 2F), with an overall higher effective number of species at restored reefs (mean ± SE 2017: 1.09 ± 0.03; 2018: 1.10 ± 0.05), compared to live (1.03 ± 0.03) and dead reefs (1.02 ± 0.02), where dead and 2018 restored reefs differed significantly (Wilcoxon Rank Sum p = 0.01) (Figure 2G). Community composition (NMDS) did not differ among oyster reef sites, as there was considerable overlap in species captured at controls and restoration sites. The two most common sportfish species showed site preferences, as L. griseus and L. synagris were more abundant at 2017 and 2018 restored reefs, respectively, compared to dead and live reefs (Table 1).

3.2. Living Shorelines

Sportfish abundance at living shoreline habitats was initially different between control (i.e., natural, relatively high quality non-eroded shoreline) and stabilized shorelines but converged by three months post stabilization and remained similar over time (Figure 3A). However, CPUE was higher at control shorelines (mean ± SE 16.6 ± 5.9 SE) relative to stabilized shorelines (7.4 ± 2.4) (linear model (LM) p = 0.04, F = 2.742; Table 1; Figure 3B). C. nebulosus experienced an initial settlement pulse that influenced this trend, but the CPUE at living shorelines was comparable between treatments when excluding this two-month species specific pulse (LM p = 0.14, F = 2.22).

Sportfish abundance at living shoreline sites was best predicted by fish and invertebrate abundances and benthic habitat cover (AICw = 1.0). Since percent vegetation and bare covers were complementary (i.e., their combined values = 100), their weights in AICc were equal; therefore, they were modeled alone and represented the next most informative models, but were excluded from subsequent additive GLMs. When percent substrate cover is excluded from AICc, the most influential model also included dissolved oxygen and site type (Table A2). Although sportfish could be found in each substrate, the abundance trends reflect differences at the habitat level, where percent vegetation cover was higher at control sites (mean ± SE, 42.1 ± 4.1) than stabilized (34.0 ± 3.0) (Table 3). At control sites, sportfish abundance was positively correlated (R² = 0.14) with vegetation (particularly ~50% coverage), while at stabilized sites most fishes were caught at <50% vegetation cover and abundance weakly correlated with bare substrate (R² = 0.03). Bare substrate was initially more prevalent at stabilized sites and especially so at Stabilized 1, where sportfish abundance was higher among this site type (Table 3, Figure 3C). There was site variation in these variables among living shorelines, with the highest sportfish abundance at Control 2 and Stabilized 1, and the highest vegetation percent cover at Control 2 and Stabilized 2 and 3 shorelines (Table A1, Figure 3D). Invertebrate abundance (93.5 ± 8.1) was highest at Stabilized 1 and prey fish abundance (438.3 ± 49.8) was highest at Control 2 and Stabilized 2 (Table A1).

Species richness between control and stabilized shorelines was similar over time (Figure 3E) and the overall effective number of species did not differ between control (1.30 ± 0.1) and stabilized shorelines (1.25 ± 0.1) (Figure 3F). Community composition also did not differ between site types (ANOSIM p = 0.17, R = 0.039). Inlet distance was an important variable related to sportfish species composition (Env Fit p = 0.001, r² = 0.27); however, not in GLM living shoreline sportfish abundance (Table 3). Among living shoreline sites, C. undecimalis (LM p = 0.02, F = 0.02) and L. griseus (LM p = 0.03, F = 0.24) were more abundant at stabilized sites compared to controls, and C. nebulosus was more abundant at control sites (LM p = 0.001, F = 2.61) (Table 1).

3.3. Across Habitats

Although overall abundance differed between oyster reefs (n = 292) and living shorelines (n = 851), both habitats showed variable abundance over time, with similar seasonal patterns of highest abundances in summer and lowest in the winter. Some species showed habitat preferences, with Black Drum (Pogonias cromis) exclusively and Spotted Seatrout (C. nebulosus) mostly associating with living shorelines. Alternatively, Lane Snapper (L. synagris) was found exclusively at oyster reefs (Table 1). Similarly, C. undecimalis, P. cromis and C. nebulosus were indicator species for living shorelines, as was L. synagris for oyster reefs (

Table 4. Results of the indicator species test to illuminate differences in species assemblages at Living Shoreline and Oyster Reef habitats, where A represents exclusivity of that species to the habitat grouping and B denotes the proportion of sites the species is found in a habitat. Values closer to 1 indicate a higher association with that habitat grouping.

Habitat	A	B	Stat	p-Value
Living Shorelines
Cynoscion nebulosus	0.99	0.63	0.79	0.0001
Centropomus undecimalis	0.80	0.33	0.52	0.0002
Pogonias cromis	1.00	0.08	0.28	0.0238
Oyster Reefs
Lutjanus synagris	1.00	0.16	0.40	0.0036

).

As such, living shorelines and oyster reefs in Mosquito Lagoon supported overlapping, yet different, sportfish community assemblages (ANOSIM p = 0.001, R = 0.36; Figure 4), which were only assessed for 2017 sites, as these were restored in the same year and would minimize any temporal variation. The effective number of species was significantly higher at living shorelines (1.02 ± 0.002 mean ± SE) than oyster reefs (1.00 ± 0.003) (Kruskal–Wallis p < 0.001). Water temperature (Env Fit p = 0.004, r² = 0.11), secchi depth (p = 0.004, r² = 0.14), air temperature (p = 0.005, r² = 0.10) and inlet distance (p = 0.001, r² = 0.37) were influential abiotic variables related to sportfish communities (Figure 4).

4. Discussion

This study provides a comprehensive examination of juvenile sportfish responses to oyster reef restoration and living shoreline stabilization, incorporating 17 abiotic and biotic factors into this assessment. Many studies have examined the response of resident fishes and invertebrates to restoration, while relatively few have targeted more transient and higher trophic species. Sportfish observations in former studies were generally in low abundances, e.g., [25,26,28], while a recent targeted sportfish study allowed for direct comparisons, which found them to have comparable densities between natural and restored coastal wetlands [14]. Furthermore, restoration benefits impact species to varying degrees and may take time to accrue, as the majority of increased gross fish production does not occur for up to two years at oyster reefs [17] and three years at living shorelines [22], similar to the findings of this study.

Overall, results suggest juvenile sportfish abundance and diversity responded positively to restored habitats, with the biotic factors of the site (e.g., oyster density and percent vegetative cover) and prey abundance representing important predictors of sportfish abundance. Oyster reefs and coastal shorelines are well known for their provision of habitat for macrofauna (e.g., crustaceans and demersal fishes) [53,54], as evident by their effect in this study. In addition, abiotic factors such as DO and proximity to the nearest ocean inlet were influential predictors, suggesting restoration location is crucial to the success of a restored habitat to enhance sportfish populations. This sentiment towards site selection has been echoed by others [28], along with a call to incorporate more biotic interactions in restoration studies [7], such as those examined here, to fill an important knowledge gap in coastal restoration studies.

4.1. Oyster Reefs

Previous studies of sportfish at restored reefs tend to examine subtidal reefs [25,26,28], while the impact of intertidal reefs, where fish can only access the site during high tide, is less studied. These oyster reefs can increase in height until they reach a maximum threshold set by the tide range [39]. In the intertidal Mosquito Lagoon system, sportfish abundance was predicted to be highest at live oyster reefs, lowest at dead reefs and intermediate at restored reefs. However, restored sites outperformed other site types, as sportfish abundance was approximately equal at live and 2017 restored reefs and highest at 2018 restored reefs. This relatively higher abundance could be due to: (1) the creation of a new habitat that was previously degraded or inaccessible to foraging, (2) the initial disturbance caused by the act of restoration could stir up sediments and nutrients, thereby attracting prey and ultimately sportfish, and (3) the restored sites provide an intermediate level of habitat heterogeneity and complexity that sportfish can better utilize.

Restored reefs that supported higher sportfish abundances increased in both height and oyster density to that of live reefs by ~20 months and 12 months at 2017 and 2018 restored reefs, respectively [39], thus lending support to hypothesis 1. Restored 2018 sites experienced an initial pulse of sportfish abundance; however, this peaked at 4–6 weeks post restoration then decreased, which possibly indicates a settlement pulse. Alternatively, sites restored in 2017 had similar abundance to live reefs from the onset. As such, it is unlikely that hypothesis 2 is driving the sportfish response in this system, and the creation of a newly accessible habitat of intermediate complexity is driving observed patterns of abundance.

If hypothesis 3 drove the trend in sportfish abundance, then the highest abundances would occur at sites with intermediate oyster density and reef height. There is some support for this, as sportfish were collectively more abundant at restored reefs where these mean reef metrics were indeed intermediate to controls, with intermediate levels observed during the first year of monitoring (Table 3, Table A1). However, their distribution was less correlative to these metrics and demonstrated multimodal abundance peaks relative to oyster density and reef height (Figure 2C,D). Alternatively, sportfish have been detected in higher abundances at unstructured bottom compared to oyster reef habitat [25,26], which could be attributed to the larger body size of those fishes and their transient nature. Sampling gear in the current study targeted juveniles and detected lower sportfish abundances at dead reefs (i.e., reefs with lowest oyster density and reef heights). The size range of juvenile sportfish can presumably allow them to access heterogenous habitat structures, which may subsequently lend to the range of reef metrics in which they were found, but this relationship requires further exploration. Furthermore, habitat complexity can be challenging to quantify and can be measured by various methods. Additional techniques (e.g., rugosity, bathymetry) could contribute to a better indication of habitat heterogeneity and its influence on sportfish abundance and diversity [55,56,57].

Dissolved oxygen (DO) also influenced sportfish abundance, although only in conjunction with the other variables and was largely consistent at ~6 mg/L across sites. Dead 1 and Live 1 were slightly below average, while the restored 2018 sites were lower, which is likely due to a dip in DO for several months post restoration before increasing to that of the other sites. Changes in dissolved oxygen are of global concern [58], which can negatively impact recreationally and commercially important fishes [59]. These data suggest that in coastal estuary systems such as Mosquito Lagoon, restoration sites should be selected for initially suitable DO, or a location where there is an opportunity to improve DO. This could occur through vegetative plantings, oyster enhancement for the water filtration of excess nutrients and phytoplankton, or where land based conservation actions can limit nutrient runoff that may result in eutrophication and ensuing algal blooms, and potentially fish kills [60].

Restored reefs in this study are located near or along boating channels, as boat wakes are a significant source of oyster reef decline in Mosquito Lagoon [38,39]. Oyster reefs with higher abundances of sportfish were located closer to Ponce de Leon Inlet, or experienced greater tidal currents. When inlet distance was considered alone, it was not a strong predictor of sportfish abundance (Table A2). However, when combined with other metrics of the sites (e.g., oyster density and prey abundance), it proved to be influential, presumably acting in conjunction with these other variables. Proximity to the inlet and its subsequent tidal flow could facilitate the movement of adult fishes along migratory pathways [61] and larval settlement of juvenile fishes. Similarly, this greater tidal exchange created by inlet proximity may facilitate bivalve and macrofauna larval dispersal and ultimately settlement at a site, which can also supply food for these sessile and resident species, thus encouraging the recovery of the habitat and benthic community.

Macrofauna have previously shown improvements following restoration [19,28,62,63], similar to that in this system [43,44,52]. Macroinvertebrates here were found in highest abundances at ~75–100 mm reef height and 75–150 oysters/m² [43], and were an important predictor of sportfish abundance. Sportfish occupied a similar habitat range; therefore, the correlation between these metrics and macroinvertebrates could be a mechanism to explain sportfish habitat use and, therefore, abundance at restored sites, thus supporting hypotheses 1 and 3. In addition, proximity to natural habitat that acts as a source population for oyster spat and macrofauna can facilitate the establishment of a restored site [64]. Proximity of restored oyster reefs to other coastal benthic habitats (namely, seagrass) has shown potential to be functionally equivalent for transient piscivorous fishes, while oyster reefs placed near mudflats augmented juvenile fish abundances [28]. Likewise, results here suggest that, along with the recovery of habitat features at the site, restoration location is essential in maximizing restoration success to enhance sportfish populations. Therefore, it would be prudent to continue to employ restoration methods in areas with high current, which often experience high vessel traffic that can lead to oyster decline, to both mitigate oyster loss and support sportfish and their prey. These efforts may also experience greater success if presented with complementary boater education to inform stakeholders that recreationally important fish populations could benefit from more mindful boating behavior [39].

Restored and control oyster reefs in this study did not differ in their ability to support a diverse sportfish assemblage, as community composition overlapped between controls and restored sites. This is unsurprising given the relatively few species of sportfish, which is further reduced when looking at diversity within habitats rather than between. Like abundance, species richness experienced temporal variation, which could be due to the influence of seasonal environmental variables and sportfish life histories. Live reefs supported the highest overall species richness, while diversity (i.e., effective number of species) was higher at restored reefs than at controls and was likely driven by higher abundances at those sites. The species in this system showed some surprising preferences. For example, C. nebulosus are known to forage near oyster reefs and have been shown to benefit from restored reefs [27], but were found in very low abundances there. Similarly, P. cromis that are known to associate with both seagrass and oyster reefs and forage on bivalves were not observed at oyster reefs. Alternatively, L. griseus were more abundant at oyster reefs and L. synagris were found exclusively at oyster reefs, with a preference for restored 2018 sites (57.9% of their total catch). These are important findings to consider if using habitat restoration as part of a species specific management or recovery plan. For example, management for L. synagris could emphasize sites with sufficient current, since they were found predominantly at 2018 restored sites which experienced greater current and were located closer to the inlet.

4.2. Living Shorelines

Few studies have examined the impacts of restored shorelines on fish populations, particularly for sportfish. Recently, fish abundance, richness and diversity were found to be similar between natural and restored marsh shorelines along the Florida west coast, and higher relative to impacted shorelines [57]. Specifically, sportfish densities at those sites were comparable between natural and restored shorelines and higher than impacted shorelines, which included five of the same species highlighted in the current study (C. nebulosus, C. undecimalis, L. griseus, P. cromis, Sciaenops ocellatus) [14]. Higher abundances and species richness of fishes have also been observed three years post construction of sill living shorelines (i.e., a hybrid technique which incorporates offshore rock material with shoreline marsh plantings), compared to unvegetated shoreline and natural controls, including some juvenile species observed in our study (C. nebulosus, L. griseus, S. ocellatus) [22]. Similarly, two shared sportfish species (C. nebulosus, S. ocellatus) were enhanced at living shorelines with oyster reef breakwaters, relative to natural shorelines [21].

Here, living shorelines supported a greater abundance of sportfish compared to oyster reefs. This was likely driven by benthic vegetation and an influx, putatively a settlement pulse of C. nebulosus in July 2017, which decreased in abundance after three months post stabilization. Mean sportfish abundance was lower at stabilized sites prior to stabilization but converged with control (i.e., natural) shorelines around three months and remained similar throughout the study. The total abundance and CPUE of sportfish were higher at control shorelines compared to stabilized, but would have been comparable if it were not for the settlement pulse of C. nebulosus. This suggests stabilized shorelines have the ability to support sportfish populations in ways that are comparable to natural shoreline habitats, and in a relatively shorter timeframe than previously reported. Previous research found that restored shorelines may take longer to accrue sportfish, which may be due to the rate of recovery and establishment of submerged aquatic vegetation at stabilized sites [21,22,29]. For example, in a west Florida study, sportfish densities were comparable between restored and natural marsh shorelines six to nineteen years post restoration, although a restoration timeline was not examined directly [14].

It is not uncommon to observe single species dominance at some sites [14,29], or fluctuations such as those observed here for C. nebulosus, which likely reflect fish life histories [49]. These episodic fluctuations are an important consideration for assessing the success of restored sites to enhance fisheries, as these “boom and bust” years can influence the presence and absence of species over the course of a study. For example, C. nebulosus were in highest abundances in living shorelines during the first year of the study (95% of their catch) and less common the following year. Similarly, L. synagris were uncommon at oyster reefs in 2017 (~8% of their catch) compared to 2018 (92%). The current study ran consecutively for three years and captured these population dynamics, but caution should be exerted when assessing species population trends over relatively short study durations.

Living shoreline sportfish abundance was influenced by biotic variables reflective of the site types. Living shoreline sites are located farther south in the lagoon, where tidal fluctuations and currents are relatively minimal. However, some of these sites contain patches of H. wrightii seagrass seaward of the oyster breakwaters, which provides an important nursery habitat for many macrofauna [54,65]. The coverage of this seagrass contributed to the higher vegetation cover at control sites and better correlated with sportfish abundance (R² = 0.14) than at stabilized sites (R² = 0.03). Perhaps this reflects a difference in seagrass quality or vegetation density at control sites that better supported sportfish and their prey. Other shoreline vegetative cover (e.g., mangroves) can also provide structure for juvenile sportfish [54] and data here indicate highest sportfish abundances were present at 25–50% cover at stabilized and control sites, respectively. Vegetation that is too dense likely precludes sportfish from accessing and utilizing a site, as evident in the lack of sportfish catch at vegetative cover >75%, or perhaps our gear was limited in sampling higher vegetation densities. Therefore, results in the current study suggest restored sites can support more sportfish as they continue to develop vegetation to ~50% coverage to resemble that of control sites.

Interestingly, juvenile sportfish also associated with bare substrate at control and stabilized sites, while they are more prevalent at the latter. Common snook (C. undecimalis) utilized a mosaic of restored coastal wetland habitats in west Florida but were more abundant in restored creeks relative to marshes and ponds [29], suggesting sportfish also benefit from open subtidal habitat. These two substrates were important predictors of sportfish abundance at living shorelines here, which offer habitat heterogeneity through a mosaic of vegetation, seagrass, and bare substrate that fish appear to utilize [22,29,57]. Juveniles captured in this study may prefer the protection of vegetated and seagrass substrate at living shorelines, while bare substrate may allow for movement of subadult fishes related to ontogenetic habitat shifts e.g., [29,66].

Macrofaunal abundance (both fish and macroinvertebrates) was also an important predictor of sportfish abundance in living shorelines (Table 3), suggesting this habitat provides a healthy prey population to support foraging sportfish. This has also been demonstrated with an increase in crustacean abundance at sill living shorelines relative to controls [21,22]. Highest prey fish abundances were found at ~25% and 75% vegetation cover, and macroinvertebrates were in highest abundances at ~30% and 40% vegetation cover for stabilized and control shorelines, respectively [44,52]. The abiotic variable that ranked highest with biotic variables in additive shoreline models was dissolved oxygen (DO), which may be related to the reduced tidal flow in this area (Table A2). DO was largely consistent across sites, with Stabilized 2 and 3 sites slightly below average, but falling within adequate levels (5.8 ± 0.3 mg/L). These data further suggest that restoration site location is important when a management goal is to support relatively mobile sportfish and should aim to mimic conditions at natural shorelines.

Sportfish community composition, species richness and diversity (i.e., effective number of species) were similar between control and stabilized living shoreline sites, with seven species being found at each. Similarly, these metrics did not differ between natural and restored coastal wetland shorelines among large and small bodied fishes in Tampa Bay, Florida [57]. As described above and demonstrated by the indicator species analysis, some species (e.g., P. cromis) showed a preference toward living shoreline habitat, and within this habitat C. nebulosus were more abundant at control sites (62.5%) and C. undecimalis were fairly evenly distributed between site types. There was an a priori expectation of catching more Red Drum S. ocellatus (n = 6), as they are a popular recreational sportfish in the region, but perhaps the juvenile size class does not utilize these habitats as much as expected, or a larger gear type (e.g., beach seine, gill net) would better quantify their relative abundance [14]. Sportfish have shown species preferences for shorelines of different quality [14,29]; therefore, the habitat diversity of living shorelines offers valuable options for sportfish species to use and, thus, enhance their populations.

4.3. Across Habitats

Living shorelines and oyster reefs provide different habitat opportunities within the Mosquito Lagoon system; shorelines receive little tidal exchange and are characterized by a soft bottom interspersed with intertidal and submerged aquatic vegetation, whereas oyster reef sites are characterized by hard substrate and some mangroves with significant tidal exchange. Despite these differences, these habitats supported similar sportfish abundance (when excluding the shoreline settlement pulse in the first year) and richness/diversity. However, they differed slightly in community composition, which may be related to species’ nursery or juvenile habitat preferences. This is an important consideration for fisheries managers who are interested in using habitat restoration to meet species specific sportfish enhancement goals, whereas more generalist species (e.g., L. griseus) may broadly benefit from restoration.

Juvenile sportfish abundance at both habitats was poorly predicted by biotic and abiotic variables alone, but rather, were best predicted by additive models of habitat metrics (e.g., oyster density and substrate cover), macrofauna prey, inlet distance and DO. Therefore, consideration should be given to oyster site locations that are 16 km or less from the inlet and can support a reef height of ~55–90 mm and oyster density of 60+/m², and shoreline locations that can provide ~50% vegetation cover, with both habitats providing DO around 6 mg/L. These criteria also support a healthy prey population, which contributes to sportfish populations and communities. Dissolved oxygen is particularly noteworthy, as water quality across the Indian River Lagoon system has been under scrutiny due to declines in recent decades [35,36]. In addition, proximate source populations should be examined to encourage the restoration success of benthic species and, ultimately, sportfish. It is particularly important to identify characteristics that can support fisheries in light of documented shifts in the occurrence of prey fish communities in this system [32,33]. Future studies could manipulate habitat complexity, e.g., [55,57], to determine optimal habitat heterogeneity to benefit sportfish. In addition, studies could employ acoustic telemetry to determine the habitat use and movement of fishes related to control and restored sites at different coastal habitats, similar to [27], before and after restoration to better quantify changes.

5. Conclusions

This study provided a baseline understanding of what factors are most influential for restoration site selection in two coastal habitats, with the goal of supporting juvenile sportfish, by examining variables that described prey abundances, habitat, and environmental conditions. It demonstrated that restoration has promise as a management tool to support juvenile sportfish, as restored/stabilized sites exceeded the predicted abundance and diversity of sportfish in both habitats. Correlations among these factors and mobile fishes are difficult to quantify due to their transient nature and combined predictor effects, however, this study identified potential parameters to enhance sportfish abundance through restoration. These results are encouraging for habitat restoration to support juvenile life stages of important predatory fishes, as their survival ultimately leads to recruitment into the adult population that is not only sought after by anglers but serves a greater top down ecological role.