As coral populations decline across the Caribbean, it is becoming increasingly important to understand the forces that inhibit coral survivorship and recovery. Predation by corallivores, such as the short coral snail Coralliophila abbreviata,are one threat to the health of reefs worldwide, but understanding of the factors controlling corallivore populations, and therefore corallivore predation pressure, remains limited. To examine the extent to which bottom-up (i.e., coral prey) and top-down (i.e., predators) forces relate to C. abbreviata distributions, we surveyed C. abbreviata abundance, percent coral cover, and the abundance of potential snail predators across six protected and six unprotected reefs in the Florida Keys. We found that C. abbreviata abundance was lower in protected areas, which had more diverse predator assemblages, and that across all sites snail abundance generally increased with coral cover. C. abbreviata abundance had strong, negative relationships with two gastropod predators in particular - the Caribbean spiny lobster (Panulirus argus) and the grunt black margate (Anisotremus surinamensis). Further, we found the size of C. abbreviata was also related to reef protection status, with larger C. abbreviata on average in protected areas, suggesting these gape-limited predators may alter size distributions by targeting small snails. Combined, these results provide preliminary evidence that marine protection in the Florida Keys may be a mechanism available to preserve critical trophic interactions that indirectly promote coral success via control of local populations of the common corallivore C. abbreviata.

Study design

To examine how bottom-up and top-down factors are related to local abundances of Coralliophila abbreviata, we surveyed 12 coral reefs that spanned the reef tract of the Florida Keys National Marine Sanctuary, USA. Although all reefs within the sanctuary are protected by some level of regulation and management, our survey sites included 6 reefs specifically designated as no-take zones called Sanctuary Preservation Areas (hereafter called “SPAs” or “protected” sites) where fishing, harvesting, or possessing any marine life is prohibited (National Marine Sanctuaries 2015). Fishing is allowed within the other 6 non-SPA sites surveyed (hereafter called “non-SPAs or “unprotected” sites), which previous studies suggest have lower fish abundances than protected sites (Kramer and Heck 2007, Bartholomew et al. 2008). All sites were part of the outer barrier reef system and surveys were conducted on the main portion of the shallow reef structure (<30 meters in depth), including reef flats, edges, and spur and groove formations. All surveys were conducted between June – July 2015.

All survey methods were performed in accordance with relevant guidelines and regulations. The permit FKNMS-2015-064 was obtained from Florida Keys National Marine Sanctuary and the special activity license SAL-14-1580-SR was obtained from the Florida Fish and Wildlife Conservation Commission to conduct this research.

Corallivore surveys

Local abundances of C. abbreviata were assessed using visual SCUBA surveys. At each site, two divers examined all living coral colonies located within 12-15, 20x2 m belt transects spaced approximately 15 m apart for the presence and abundance of C. abbreviata. The exact number of transects surveyed varied based on the size of the site and logistical constraints. On every coral present within each transect, divers visually examined and inspected the margins between living coral tissue and dead coral skeleton or the seafloor for the presence of snails, as C. abbreviata often inhabit these margins (Miller 2001, Williams and Miller 2005, Shaver et al. 2017). Inspection included lightly touching the margins of dead coral skeleton or seafloor to feel for snails, as small tufts of algae can sometimes be mistaken for snails. All C. abbreviata were counted and removed, and most were transported to land for biometric measurements. Counts of snails were pooled for each site and standardized for the number of transects surveyed.

On land, we measured C. abbreviata shell length, shell thickness, and live snail biomass (i.e., including shell). Shell length was measured as the distance from the apex of the shell to the bottom of the operculum, while shell thickness was measured as the thickness of the operculum’s outer lip 5 mm into the aperture. Both metrics were measured using digital calipers. Biomass of live snails was measured using a digital scale.

Bottom-up factors: hard corals

Relative percent coral cover measurements at each site were determined by collecting and analyzing benthic cover photos. To do this, we photographed 40, 0.5-m² plots located within 13-15, 20x1-m belt transects spaced at least 5 meters apart. The number of transects per site varied based on the size of the reef. All photos were taken at the same height above the seafloor using a PVC apparatus attached to an underwater camera that pointed at the seafloor. All photographs were analyzed for the percent cover and abundance of all coral species using Microsoft Powerpoint and Excel 2016 software with 25 random points on each slide to identify coral species and estimate cover. Each point was coded to the species level for corals. If points fell on top of objects that were above the seafloor (e.g., queen conch shells), points were moved to the closest area showing benthic cover.

Top-down factors: predators

To examine how much of the variation in C. abbreviata abundance could be explained by predator abundance, we conducted surveys of known gastropod-consuming fishes and other suspected predators of C. abbreviata (Baums et al. 2003a, Johnston and Miller 2006, Humann and Deloach 2014), as actual data on of C. abbreviata predators are limited (Baums et al. 2003b, Johnston and Miller 2006). We used existing scientific literature and books (Claro 2001) to identify potential predators if they met one of three requirements: (1) previous research suggested they may be C. abbreviata predators, (2) they had been observed preying on Coralliophila spp., or (3) they were taxa that specialize in feeding on gastropods. The following species were thus included in surveys as potential C. abbreviata predators: Caribbean spiny lobster (Panulirus argus: Baums et al. 2003a, Humann and Deloach 2014), Family Labridae: slippery dick (Halichoeres bivittatus: Randall 1967), puddingwife (Halichoeres radiatus: Randall 1967), hogfish (Lachnolaimus maximus: Randall 1967, Baums et al. 2003a), Spanish hogfish (Bodianus rufus: Randall 1967); Family Haemulidae: white grunt (Haemulon plumineri: Randall 1967), bluestriped grunt (Haemulon sciurus: Randall 1967), Caesar grunt (Haemulon carbonarium: Randall 1967), Spanish grunt (Haemulon macrostomum: Randall 1967), sailor’s choice (Haemulon parra: Randall 1967), porkfish (Anisotremus virginicuss: Randall 1967), black margate (Anisotremus surinamensis: Randall 1967); and Family Diodontidae: porcupinefish (Diodon hystrixs: Randall 1967: Randall 1967). Queen triggerfish (Family Balistidae, Balistes vetulas: Randall 1967) and balloonfish (Family Diodontidae, Diodon holocanthus: Randall 1967) were also surveyed but were not observed in any site and were therefore excluded from further analyses. Due to logistical constraints, we were restricted from surveying some cryptic species that have also been cited in the literature as potential C. abbreviatapredators (e.g., octopi: Baums et al. 2003a; the snapping shrimp, Synalpheus fritzmuelleri: Goldberg 1971).

Fish surveys were conducted using the Stationary Point Count method (Ayotte et al. 2011), where two divers are stationary along a 30-meter transect at 7.5 and 22.5-meters and each surveys a 15-meter diameter circle over a given time interval. At each site, we conducted 10 fish surveys spaced at least 20 meters apart, allowing for a 5-minute acclimation period after which all species identified as potential predators of C. abbreviata were counted for 5 minutes. Surveys of the spiny lobster P. argus were conducted using timed-diver survey methods, where 1-2 SCUBA divers systematically swam over the reef site (where fish and snail surveys were conducted) and all P. argus observed during a 1-hour total period were recorded. If there were multiple divers, the time was split accordingly. These surveys took place before the 2-day recreational “mini-season” for spiny lobster from 29-30 July 2015 to avoid artificial differences in lobster abundance within and outside of SPAs. Underwater visibility was consistently high during this time period, allowing divers to conduct a clear visual census in each survey area. All surveys were conducted during the day between 1000 hrs and 1600 hrs due to lack of resources for nighttime surveying.

Statistical Analysis

Corallivore abundance as a function of protection, coral cover, and predator abundance - To examine the role of both bottom-up (available coral food resources) and top-down (presence of predators) forces and the role of site protection status in affecting local C. abbreviata abundance, we created a general linear model with a negative binomial distribution to describe snail abundance with survey size as an offset using the R packages MASS and foreign. We used total relative coral cover and total abundance of all potential snail predators surveyed in these models. Total cover of all coral species was included in these models, as this survey and previous surveys in the same sites have observed C. abbreviata preying on a large variety of coral species, with very few coral species where C. abbreviata feeding has never been observed. A negative binomial distribution was used due to overdispersion with Poisson models. We used protection, coral cover, and predator abundance as fixed effects in the initial model and then created additional plausible models by removing parameters and adding potential two-way interactions, ultimately selecting the most parsimonious model with the lowest AIC value. All analyses were conducted in R version 4.0.0 with the tidyverse suite of packages (Wickham et al. 2019).

Corallivore size as a function of protection, coral cover, and predator abundance – All biometric measurements taken of C. abbreviata (e.g., shell length, thickness, and live snail biomass) covaried to some degree, and therefore conducted further analyses using only measurements of snail shell length. This is consistent with previous studies examining the how different coral species prey affect C. abbreviata life history using measurements of size in shell length (Johnston and Miller 2007). Additionally, shell length is the measurement most likely to affect gape-limited gastropod predators. We examined the role of site protection, coral cover, and predator abundance on C. abbreviata size distributions using a general linear model with a Gaussian distribution and identity link to describe snail shell length. All variables were included as fixed effects in the initial snail shell length model. We then created alternate plausible models by removing parameters and adding potential two-way interactions, ultimately selecting the most parsimonious model with the lowest AIC value.

Predator abundance and species richness between protected and unprotected reefs -We additionally examined whether there were alternative characteristics of the predator community that might be driving snail distributions in and out of protected areas. To test whether the abundance and species richness of potential C. abbreviata predators varied due to site protection status, we performed Welch two-sample t-tests with Shapiro tests to check for normality and F tests to check for equal variances. Predator abundance log-transformed to better meet assumptions of normality.

Corallivore abundance as a function of specific potential snail predators - The predators surveyed were a group of species thought to be most likely to prey on C. abbreviata, and the identity of actual predators was therefore unknown. Given that the global models above assessed total predator abundance in each site and it is unlikely that all species surveyed are predators of C. abbreviata, and that a major goal of this study was to identify C. abbreviata predators, we examined whether C. abbreviata abundance appeared to be related to any of the potential predator species surveyed. To examine these relationships, we created general linear models with negative binomial distributions to describe snail abundance as a function of each species’ abundance with survey size (for snail surveys) as an offset. A Bonferroni correction was applied to adjust for multiple comparisons of individual predator species (n=13), with a new alpha value of P < 0.0038.