Exotic Brittle Star Interactions With Native Octocoral Epizoites: An Endemic Benthic Ctenophore in Peril?

Widespread and large populations of the exotic eastern Pacic ophiuroid brittle star Opthiothela mirabilis now occur in southeastern Florida, extending the range of this recently introduced species from southern Brazil northward to the eastern Caribbean Sea and Florida. The Florida brittle stars, representing two lineages, are epibionts on shallow (3-18 m depth), tropical/subtropical plexaurid (e.g., Eunicea spp., Muricea elongata) and gorgoniid (Antillogorgia spp.) octocorals. The scope of this study includes recent distributional records of O. mirabilis in south Florida, eld abundances in relation to the cohabiting endemic ctenophore Coeloplana waltoni, behavioral observations of the ophiuroid, ctenophore and the predatory amphipod Caprella penantis, as well as a laboratory experiment testing the effects of the alien ophiuroid on the native ctenophore. Individuals of O. mirabilis have been collected near St. Lucie Inlet, extending its northern-most range by about 110 km since 2019. Two years of eld sampling have demonstrated signicant declines of the native, benthic ctenophore with increasing abundances of the exotic ophiuroid. Evidence suggests that the ophiuroid is negatively affecting the abundances of the ctenophore through interference competition, greatly aided by its abrasive armature of calcareous spines, plates and hooks. Sporadic and intense predation by a caprellid amphipod (Caprella penantis) also probably contributes to the ctenophore’s decline, but to a lesser extent than that caused by the ophiuroid. Adding to the risk of extinction of C. waltoni is its narrow requirement of living octocorals as hosts and restricted distribution in southeast Florida and the Bahamas.

Marine organisms exhibiting complex body structures often host numerous symbiotic metazoan associates, thus promoting local and regional biodiversity. These associations are realized across a broad range of seascape depths, from the shallow littoral zone (Bayer, 1961;Goh et al., 1999;Sánchez, 2016) to mesophotic (Maggione et al., 2020) and deep continental margin environments (Buhl-Mortensen et al. 2010). Early on, Bayer (1961) recognized the great variety of organisms associated with tropical western Atlantic octocorals -zooxanthellae (dino agellates); parasitic copepods, cirripeds, pycnogonids, and gastropods; epizootic cnidarians, ctenophores, polychaete worms, porcellanid and spider crabs, hippolytid shrimp, gastropod molluscs, and ophiuroid echinoderms. Our knowledge of the diversity of octocoral communities, including octocoral microbes (viruses, bacteria), fungi, protozoans, invertebrates and shes has been greatly augmented recently (e.g., Van  The focus of this study is on the alien ophiuroid Ophiothela mirabilis and its interactions with the native benthic ctenophore Coeloplana waltoni, both epizoites on species of octocorals. We provide recent data on the occurrence and abundance of the non-native ophiuroid on octocorals across approximately 78 km of coastline, from Hollywood Beach to West Palm Beach, Florida. A seven-day laboratory experiment simulating ophiuroid-ctenophore interactions was performed to further investigate this competitive association and the cause of natural variations in the epizoite relative abundances. Attention is also directed to the recent occurrence of the amphipod, Caprella penantis, a sporadic co-inhabitant of octocorals. The abundances and behaviors of the ophiuroid and caprellid suggest competitive and predatory interactions, respectively, leading to negative impacts affecting the abundance of the benthic ctenophore. This study offers an opportunity to address questions relating to extinction research in the earliest stages of invasion (Sax and Gaines, 2008). A time-lag to extinction may exist for the benthic ctenophore under competition by the introduced ophiuroid and intense predation by the caprellid. To understand this, the ophiuroid invasion needs to be tracked to follow how the ophiuroid changes the abundance of the ctenophore. Interference competition by the ophiuroid may transform the ctenophore's microhabitat su ciently to endanger their presence. Finally, we propose a mechanism of ctenophore exclusion by the incidental and frequent abrasive activities of the heavily armored O. mirabilis. Coeloplana waltoni Glynn, Bayer, Renegar, 2014 is a native south Florida benthic ctenophore (Glynn et al. 2014). Caprella cf. penantis Leach, 1814, an octocoral epizoite, belongs to a complex of cryptic, amphipod species and is probably not cosmopolitan in distribution (Cabezas et  Sampling to determine the spatial pattern of occurrence of O. mirabilis involved observations at several sites off coastal Ft. Lauderdale, at locations north and south of Port Everglades. Ten to 40 plexaurid and ten to 40 gorgoniid colonies were examined at 22 sites from April to December 2020. Occurrence and abundances at survey sites were estimated according to the following qualitative scale: rare, 0 to 1-2 ind. on < 5% of colonies; moderately common, 5-10 ind. on 5-20% of colonies; abundant, 20+ ind. on several branches of over 20% of colonies. At 16 of the sites located around Port Everglades, prevalence was calculated quantitatively (percent of colonies with O. mirabilis present) and prevalence interpolated using an inverse distance weighted method (IDW) in QGIS (Bartier & Keller 1996). Geographic positions were obtained by means of GPS.

Behavioral observations
The feeding behavior of individuals of O. mirabilis were viewed in the eld during the day and night. Competitive and predatory observations were made within 2 to 3 hours after collection, microscopically (8-35X) under subdued lighting in the laboratory on octocoral branches with their normal complement of epizoites. Interactions between individuals of O. mirabilis and their octocoral hosts also were observed in the laboratory during feeding and non-feeding periods.

Experimental procedure
The plexaurid octocoral host species employed in the laboratory experiment were Euniceasuccinea (Pallas, 1766) and Eunicea tourneforti Milne Edwards & Haime, 1857. These octocoral species were the only two found consistently hosting both the non-native ophiuroid Ophiothela mirabilis Verrill, 1867 and the endemic benthic ctenophore Coeloplana waltoni Glynn, Bayer & Renegar, 2014. The common beige color morph of O,mirabilis was employed in this experiment (Glynn et al. 2020). The experimental animals were collected off Lauderdale by the Sea, Florida (26.1920°N; 80.0964°W) on 27 October 2020 at 4-5 meters depth. The basal end of each lateral branch was trimmed until the branch measured 10 cm in length. A total of 10 separate 4 litercontainers, each with a vertically oriented octocoral branch with (experimental) or without (control) individuals of the non-native brittle star were randomly placed in a linear array with respect to the light source. All octocoral branches contained ctenophore epizoites with initial densities: experimental (with brittle stars), mean C. waltoni density (no. ind. 10 cm -1 branch) = 7.2, range = 6-9; controls (without brittle stars), mean C. waltoni density = 7.6, range = 6-10. Initial experimentalO. mirabilis density, mean = 5.0, range = 3-6; control = 0. The water was gently aerated and stirred continuously with magnetic stirring bars, creating a ow rate of approximately 8 cm s -1 . One liter of water with naturally present plankton was changed out daily in each container. Both brittle stars and ctenophores were observed feeding on plankton during daylight hours. Ctenophores were occasionally observed feeding at night. The experiment was conducted for 7 days; data on epizoite densities and location on octocoral hosts were collected daily. Caprellids were absent from the experiment.

Statistical analyses
Two statistical models were t to densities of the three species (O. mirabilis, C. waltoni, C. penantis) inhabiting the 10 cm octocoral branches (406 total branches collected over 16 surveys) using the R package glmmTMB (Brooks et al., 2017). The rst model, to analyze trends in species counts over the survey period, employed a zero-in ated rst order negative binomial (NB1) response structure (ϕ = 6.79) with survey as a random effect (a branch random effect was dropped due to zero estimated variance).
The second model, to examine the relationship between the abundance of C. waltoni and co-occurring species on octocorals, was t using a standard NB1 response structure (ϕ = 9.53) and survey random effect. Fixed effects were selected by AIC from models t by maximum likelihood, with the nal models t by restricted maximum likelihood. Visual examination of simulated scaled residuals indicated assumptions were met su ciently for both models. Inference on parameters was performed by employing tests with denominator degrees of freedom estimated using the inner-outer method (Pinheiro and Bates, 2000).
Laboratory experimental results of the effects of the ophiuroid O. mirabilis on the benthic ctenophore C. waltoni were tested according to the following procedure. A generalized linear mixed model with Poisson response was t by maximum likelihood to C. waltoni counts across seven days of observation under experimental (O. mirabilis present; n = 5) and control (O. mirabilis absent; n = 5) conditions. Correlation between repeated measurements was modeled with a heterogeneous AR(1) structure. Type III Wald tests with Kenward-Roger degrees of freedom indicated that reductions in counts over time by treatment were best modeled as second-degree polynomials; there was no evidence that time or treatment effects differed between octocoral host species. Studentized residual plots indicated model assumptions were met su ciently. To test the overall effect of O. mirabilis on C. waltoni abundance over time, a nested reduced model without treatment effects was t as above and compared to the full model using a drop in deviance Chi-square test. Models were t using SAS/STAT software, version 9.4, PROC GLIMMIX.

Field observations
In the 9-month sampling surveys, O. mirabilis was present on plexaurid and/or gorgoniid colonies at 18 of 22 sites, including the northernmost site surveyed near West Palm Beach (Fig. 1). Abundance varied spatially, with two hotspot locations identi ed, on the inner reef (10 m depth) offshore of Port Everglades and on the nearshore ridge complex (5 m depth) at Hollywood/Hallandale Beach. Individuals of O. mirabilis were found on 100% of colonies at three sites, were abundant at a further three, moderately common on plexaurid colonies at seven sites, moderately common on gorgoniid colonies at four sites and rare at four sites. Quantitative assessment around Port Everglades revealed that 24.37% (±34.32% SD) of octocoral colonies hosted O.mirabilis, 29.6% (±42.38% SD) of gorgoniids and 19.16% (±19.16% SD) of plexaurids.  (Calder, 2013), was attached to the substrate under the bridge in the intracoastal waterway. Circulation at this location was strong, with a twice-daily tidal in ux of ocean water.
Initial eld evidence suggesting that interspeci c interactions between brittle stars and ctenophores might be responsible for abundance differences (including absences) was evident from sequential sampling at the Dania Beach study site. Across 16 quantitative surveys over a 19 month period, mean densities of individuals of C. waltoni ranged from 0 to 30.9 individuals per 10 cm octocoral branch length, mean densities of individuals of O. mirabilis ranged from 0 to 30.5, and mean densities of individuals of C.
penantis ranged from 0 to 10 (Fig. 3). Based on the rst statistical model (correlation based R 2 = 0.48), there was strong evidence that trends in epizoite densities inhabiting octocoral hosts differed by species across the survey period (p < 0.0001). This was driven speci cally by a decline in C. waltoni over time (p = 0.0006), which exhibited an average reduction in population size of 9.2% every 30 days (95% CI: 4.1% to 14.0%). Overall trends in either direction were not signi cant for O. mirabilis (p = 0.31) or C. penantis (p = 0.073).

Laboratory experiment
After accounting for the overall trend in total numbers of C. waltoni, the second statistical model (correlation-based R 2 = 0.51) revealed that an association between the abundance of C. waltoni and counts of O. mirabilis was still quite evident (p = 0.0007). For each additional O. mirabilis individual present per 10 cm branch length, the abundance of C. waltoni displayed an expected 3.8% decline (95% CI: 1.6% to 6.0%). In contrast, the abundance of C. waltoni showed evidence of a second order relationship with counts of C. penantis (p = 0.016), exhibiting a positive association within the limited range of individuals observed. There was no evidence to suggest that these relationships varied over time.
Based on analysis of the laboratory experiment there was strong evidence that the presence of O. mirabilis negatively impacted the density of C. waltoni ( = 18.9, p = 0.0003, Fig. 4). Mean (se) C. waltoni density in the control group began at 7.6 (0.75) and declined to 5.0 (0.55) individuals/10 cm branch length by day seven, while in the brittle star group mean density declined from 7.2 (0.49) to 1.0 (0.63) individuals/10 cm branch length. Ctenophores on octocoral branches remained roughly evenly spaced in the control group, but clearly clumped at the ends of branches, separated from the ophiuroids, in the experimental treatment.

Behavioral observations
Interspeci c behaviors were documented to gain insight into interactions possibly affecting epizoite abundances. Field observations of O. mirabilis revealed frequent suspension feeding during day and night with arms greatly extended and waving (as if searching) in passing currents. Occasionally unidenti ed items were captured, conveyed to the mouth and presumably ingested. Identical activities were observed in the laboratory during feeding trials when offered Artemia nauplii.
Individuals of C. waltoni were nearly always inactive, rmly attached to coenosarc surfaces, and often near polyp bases with tentacles fully contracted (Table 1). When individuals of C. waltoni were contacted by O. mirabilis, this often caused the ctenophore to move. In one instance, contact by O. mirabilis caused rapid contraction with the ctenophore assuming a crinkled body shape (see Fig. 3I in Glynn et al. 2014).
In two collections with individuals of the caprellid (Caprella penantis) present, 16.2% (5 of 31) and 100% (1 of 1) of the individuals of C. waltoni in the samples were attacked and eaten; no ctenophores were attacked in two collections ( Table 1). The caprellid was not necessarily common, but when present effected intense predation on the ctenophore. The predatory caprellids in the 17 March sample moved quickly along the octocoral branch, ostensibly in a search mode; as soon as a ctenophore was encountered it was grasped, shredded, and consumed. In this same sample, two caprellids separately attacked an individual of O. mirabilis, but failed to penetrate the epidermis; each caprellid then moved further along the branch after a few seconds.
When present at moderate and high densities the ophiuroids were generally evenly spaced along the octocoral branch. However, in two high density collections (September, November, Table 1) the ophiuroids were densely crowded with arms and discs overlapping. Ctenophores were not present in or near these aggregations. waltoni this is probably in part a re ection of the patchy occurrence and the sampling protocol for these species; the identical octocoral patches, colonies, and branches were not repeatedly sampled at the Dania Beach study site. The occurrence of individuals of Caprella penantis was sporadic, observed in only four collections. Aquarium observations revealed this amphipod to be highly mobile, moving quickly over a given octocoral branch, and often swimming and moving between branches. It is likely that this predatory amphipod naturally moves frequently among octocoral branches within and among colonies in search of prey.

Discussion
Ophiuroids were not present on dead octocoral branches (Glynn et al. 2020). Their absence could offer a refuge for C. waltoni, however, the ctenophore was absent from dead branches as well. Dead branches are usually quickly colonized by algae and hydroids, a likely inhospitable habitat for mobile epizoites. Live octocoral host tissues bearing nematocysts, sclerites, and chemical defenses (Pawlik et al., 1987;Harvell and Fenical 1989;Yoffe et al., 2012) would provide an advantage in predator avoidance, thus offering a suitable microhabitat.
Since octocorals engage heavily in suspension feeding, capturing particulate organic matter (Lasker, 1981) and microzooplankton (Glynn et al. 2018a), the interference of polyp expansion by individuals of O. mirabilis was suggested to be a potentially signi cant threat against satisfying the trophic requirements of octocoral hosts (Araújo et al. 2018;Ferry et al. 2020), although a recent study suggests this may not be consequential (Glynn et al., 2021). From the current study --which has demonstrated an inverse relationship in the abundance of ophiuroids and ctenophores in eld collections, and decreasing ctenophore abundance in the presence of ophiuroids in a laboratory experiment --it appears that the major effect of the exotic ophiuroid is on the displacement, by interference competition, of the cohabiting benthic ctenophore. This sort of negative impact on biodiversity is the predominant effect of marine invasive species in European seas (Katsanevakis et al. 2014). It is not yet known if O. mirabilis will negatively impact the abundances of octocoral hosts, but if this occurs then local biodiversity would be greatly diminished. Invasive azooxanthellate sun corals are displacing ecosystem engineers (e.g., reefbuilding corals and zoantharians) in Brazil, having a depressing effect on biodiversity, nearly two decades after their rst sightings (Miranda et al., 2016;Luz and Kitahara, 2017). It is too early to determine the extent to which the ophiuroid will continue to affect C. waltoni, and perhaps begin to impact octocorals and other epizoites.
Several studies have described the kleptocommensal and kleptoparasitic trophic relationships of hydroids and other invertebrate epizoites associated with sponges, octocorals, bryozoans, and polychaete worms (e.g., Gotto, 1969;Puce et al., 2008). Bavestrello et al. (1996) observed caprellids seizing previously captured nauplii from hydroid polyps; a nudibranch grazing preferentially on hydroid polyps that previously ingested zooplankton, an example of the consumption of a prey item plus its ingested prey, was termed kleptopredation by Willis et al. (2017). The symbiotic status of the benthic ctenophore and its octocoral host is not clear since we have observed the ctenophore purloining food (unidenti ed organic detritus) captured by its host's polyps, and octocoral polyps have been observed removing captured sh eggs from the tentacles of ctenophore epizoites (Glynn et al. 2018a). Further study is necessary to determine the signi cance of these trophic resources, the relative amounts garnered by the symbiotic partners, and how eld conditions affect these interactions.
The alacrity with which individuals of Caprella penantis attacked, shredded and consumed relatively large fragments of benthic ctenophores suggests that this crustacean species is an occasional, if not frequent, predator of individuals of C. waltoni. No defensive responses by C.waltoni, such as rapid tentacle extension (Glynn et al., 2018b), were observed beyond slow eeing movements, which were ineffective. From feeding observations conducted in northern Florida, Gulf of Mexico, Caine (1974) noted that individuals of C. penantis depended primarily on lter feeding and substrate scraping. Contrary to the observations of Caine (1974) and Paz-Rios et al. (2014) on the relative ineffectiveness of caprellids as predators, we observed that all individuals engaged in feeding moved quickly over octocorals, occasionally swimming, potentially in search of ctenophore prey. This is the rst documented record of a ctenophore species consumed by a caprellid (Saunders, 1966 If the recent decline in abundance of C. waltoni, and local disappearance at some sites, is due primarily to O. mirabilis, then it is necessary to understand how this occurs. The ctenophores are exceedingly well camou aged and typically quiescent when present on their octocoral hosts. Once a ctenophore has found a favorable perch for the extension of its ' shing' tentacles, movement is infrequent. Our observations indicate that brittle star contact with individuals of C. waltoni is an irritation, often causing the ctenophores to move. We hypothesize that ctenophores on the move are more susceptible to dislodgment from their octocoral hosts. Dislodgment could occur during turbulent sea conditions, which are frequent off the southeast coast of Florida, or by shes and other water column predators. A notable report of a caprellid infestation and predation on gorgonian sea fans in North Sulawesi, Indonesia, revealed massive consumption of coenenchyme (octocoral tissues including epidermis, mesoglea and gastrodermis) and skeletal fragments resulting in total colony mortality in three species (Scinto et al., 2008). Octocoral colonies of Melithaea sp. were most severely affected, irreparably damaged within a week. The caprellid responsible for the sea fan mortality was Metaprotella sandalensis, a common species on Indo-Paci c coral reefs. The morphology of the feeding appendages (molar process and mandibular palp) are characteristic of caprellid predators and scrapers (Caine, 1977; Guerra-García and Tierno de Figueroa, 2009). Mean caprellid infestation densities were 8.5 ind cm -1 branch, equivalent to 85 individuals over a 10 cm-long branch (Scinto et al., 2008; C. Cerrano, pers. comm.). These densities are an order of magnitude higher than those observed in south Florida (Table 1).
Such a severe predation event triggering the death of entire gorgonian colonies, would also result in the loss of the associated symbiont community.

Concluding remarks
The ophiuroid invasion of octocorals in southeast Florida is very recent, making it di cult to predict longterm effects. The apparent ongoing displacement of an endemic benthic ctenophore is the only negative effect thus far observed. No obvious signs of impairment to octocoral hosts are evident (Glynn et al., 2021). In light of the ophiuroids establishment in southeastern Brazil, over a period of about 20 years, its rapid and wide-ranging dispersal and possible repeated introductions, it is highly probable that it will continue to spread and become established throughout the western Atlantic warm water region where octocorals and other suitable substrata occur. Since the invasion of O. mirabilis in the western Atlantic region is so recent and still on-going, we urgently encourage continuing surveys to document its occurrence, abundance, and interactions with native species. Coastal and estuarine habitats with suitable environmental conditions in need of surveys and study are the Tropical Northwestern Atlantic province (wider Caribbean Sea) and the Warm Temperate Southwestern (southeastern Brazil) and Northwestern Atlantic (Northern Gulf of Mexico, Carolinian) provinces. Behavioral and experimental studies of alien brittle star interactions vis-a-vis native epizoites are especially encouraged and may shed light on control measures should these be needed. Cryptogenic C.penantis should also receive attention considering its predatory impacts on C. waltoni and other octocoral epizoites.  Square concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. This map has been provided by the authors.