**4. Discussion**

We found that there were no significant effects of rugosity, hard coral cover, soft coral cover, or algal cover on total reef fish and functional feeding guild abundance across our sites. This is different from previous studies that found a greater abundance and diversity of fishes on reefs that have higher hard coral cover and structural complexity [29]. There are several factors that could explain this unexpected finding. First, some studies have found that reef fish abundance generally increases with depth [54]. Second, the shallowwater nearshore habitats of the middle Florida Keys experience a great deal of disturbance, such as nutrient loading or extreme temperatures [55,56], so that there is a possibility of a delayed response of the reef fish community to decades of continuous perturbations [57]. Alternatively, it may be that our methods were not sensitive enough to detect an effect; for example, the acclimation period may not have been long enough, or the artificial structure not large enough to influence fish distribution.

We found that the natural structures were used by fish significantly more than the artificial structures, even though these were designed to mimic the size and shape of the available natural structures. Previous studies have also found that artificial reef structures (ARs) built with numerous distinct holes and crevices were unsuccessful overall [58]. Such studies have also suggested that artificial structures are used mostly at night, a possibility that we did not investigate with our artificial structures [59]. Evidence has also suggested that fish are attracted to the auditory and chemical cues of living hard corals, so the lack of cues from the artificial structures might explain the low rates of use [60,61].

Unsurprisingly, the abundance of fishes on each reef (video transect survey) accounted for more than half of the variance found in the structures' associations. There were more fish using the structures when there were more fish available on the reef. Generally, reef fish used natural structures significantly more than artificial ones, but, surprisingly, the types of natural (sea rod, dead, diseased, and healthy hard corals) or artificial (control, soft, boulder, and elkhorn) structures did not influence fishes' use. Natural structures, on average, had similar heights and total surface area to our artificial structures, which suggests that differences in observations of use were not strictly due to space competition. However, the natural structures were covered with living organisms and algae and provided more food than our artificial structures. The small amount of biota could explain why the artificial structures were used significantly less than natural structures [9,14]. Moreover, these differences in the resources available on artificial and natural structures could affect reef fish habitat specialists differently from habitat generalists (Table S2) [29]. We found that artificial structures were occupied more on low-rugosity sites than on high-rugosity sites. This supports previous observations that artificial structures have their greatest impact in environments with fewer structures [62,63].

Further characterization of structure use by different functional feeding guilds revealed other important patterns not apparent from the analysis of use by all reef fishes combined. Herbivore observations on natural and artificial structures were lower on shallow reefs with high rugosity, but they significantly preferred diseased corals over sea rods. This may be due to the surfaces on the coral heads being newly opened for turf algal colonization, a preferred foraging substrate for all parrotfish species, which commonly inhabit these reefs [51]. Alternatively, the newly available coral tissue could be providing nutritional benefits or a greater concentration of autotrophic organisms that attracted parrotfish foragers [64]. Herbivores on artificial structures were not significantly different from on natural structures, perhaps because after 2 weeks, we observed that the artificial structures had accumulated enough biofilm to be a suitable foraging substrate (Figure 3). The functionally important role that parrotfish play on Caribbean coral reefs is well understood [65,66], and our study suggests that with the flattening of reefs, the reefs of the future may see a decline in parrotfish abundance.

Piscivores rarely used either natural or artificial structures and were unrelated to any of the three component scores of reef substrate. This is unsurprising, due to previous findings that piscivore abundance is associated more with prey availability than any habitat characteristic [67,68]. In contrast, omnivores used natural structures significantly more than artificial structures and decreased structure use on shallow reefs with high rugosity, a response that has been predicted in recent models [31]. However, omnivores' preference for low-rugosity reefs contradicts previous literature that has found that all feeding groups tend to be positively associated with increased complexity [11–13,69]. We expected that rugosity and hard coral cover would be correlated [25], but found that rugosity was not a function of hard coral cover. Presently, the reefs in the Florida Keys are composed of scattered boulder corals, abundant soft corals, and limestone ledges. Additionally, with the recent outbreak of stony coral tissue loss disease, there has been a decrease in live hard coral cover but a lingering presence of dead hard coral structures [70]. With time, overall rugosity and the presence of scattered boulder corals will continue decreasing as dead coral heads begin eroding away.

Similarly, invertivores used natural structures significantly more than artificial structures, but in contrast to the herbivores and omnivores, they increased on shallow reefs with greater rugosity. This suggests that grunts may be the one feeding guild most impacted by the loss of structural complexity and the flattening of the reef. This relationship is best explained by the reef component score associated with rugosity rather than the component score associated with hard coral cover, which suggests that it is physical habitat that matters more to grunts than the health status of the coral. Invertivores often depend on finer-scale shelters for their prey species to occupy and would explain why they, and their prey, would be negatively impacted by structural loss [71]. This negative response of invertivores to reef decline has been predicted as a response to climate change according to predictive climate change models [31].

All guilds had a predominant species that did not mirror the pattern of the rest of the guild. Bicolor damselfish (*Stegastes partitus*), striped parrotfish (*Scarus iseri*), bluestriped grunts (*Haemulon sciurus*), white grunts (*Haemulon plumierii*), schoolmaster snapper (*Lutjanus apodus*), and the bluehead wrasses (*Thalassoma bifasciatum*) had a disproportionate representation within their associated functional feeding guilds, but their patterns did not parallel the overall effect observed within their guild. Their presence may be due to their being habitat generalists, equally at home in hard coral- or soft coral-dominated reefs [67,72]. The reduced structure use by the bicolor damselfish, striped parrotfish, or bluehead wrasses could be explained by the higher abundance of piscivores, although we did not observe an increase in piscivore presence, even though these habitats had an abundance of prey, which should have driven their numbers up [67,68,71,72].

Our results suggest that use of structures, both natural and artificial structures, differs among reef fish functional feeding guilds. If the low-rugosity reefs with low hard coral cover are representative of the future reefs of the middle Florida Keys, we would predict a shift in reef fish community with increasing proportions of herbivores and omnivores, and a decreasing proportion of invertivores. These results can be used as a predictive model for reef fish community responses to changes in reef composition and may be useful in the design of future marine protected areas needed to preserve feeding guilds critical to the recovery of hard corals. Future studies should examine how the relative abundance of fish functional feeding guilds changes in response to this transition from hard-coral-dominated to soft-coral-dominated reefs in the middle Florida Keys.

**Supplementary Materials:** The following are available online at https://www.mdpi.com/article/ 10.3390/oceans2030036/s1. Table S1: Species included in this study, their functional feeding guild, their presence (Y or N) at artificial and natural structures, and their habitat use. Artificial structures = soft coral mimic (SC), control structure (C), boulder coral mimic (BC), and elkhorn coral mimic (EC). Natural structures = sea rods (SR), dead corals (DC), infected diseased corals (IC), and healthy corals (HC). Fish are separated into generalist or specialist habitat use classifications [29]. Table S2: Percent presence of generalist versus specialist species observed for the four artificial and four natural shelter types for each functional feeding guild. Artificial structures = soft coral mimic (SC), control structure (C), boulder coral mimic (BC), and elkhorn coral mimic (EC). Natural structures = sea rods (SR), dead corals (DC), infected diseased corals (IC), and healthy corals (HC).

**Author Contributions:** Conceptualization, K.N., K.S. (Kylie Smith) and M.C.; methodology, K.N., K.S. (Kylie Smith) and M.C.; formal analysis, K.N., T.F. and M.C.; investigation, K.N.; resources, K.N., K.S. (Kylie Smith) and M.C.; data curation, K.N., T.F., K.M. and K.S. (Kelsey Sox); writing—original draft preparation, K.N., T.F. and M.C.; writing—review and editing, K.N., K.M., K.S. (Kelsey Sox) and K.S. (Kylie Smith); supervision, K.N. and M.C.; funding acquisition, K.N. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the American Museum of Natural History Lerner-Gray Fund for Marine Research, the American Society of Ichthyologists and Herpetologists Raney Award, Clemson University's Creative Inquiry Initiative, the Explorer's Club Mamont Grant, the South Carolina Space Grant Consortium Kathryn D. Sullivan Earth and Marine Science Fellowship (# 2014105), and Clemson University.

**Institutional Review Board Statement:** This study was conducted according the guidelines of the Florida Keys National Marine Sanctuary (permit numbers FKNMS-2017-032, approved on 6 January 2017, and permit number FKNMS-2018-119, approved on 10 January 2018).

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Data is archived at Clemson University and is available upon request to Michael Childress (mchildr@clemson.edu).

**Acknowledgments:** We thank Reanna Jeanes, Morgan Gardner, Riley Garvey, Rachel Radick, and Emma Crowfoot for assistance in data collection and processing. We also thank the two anonymous referees and the special issue editor for their great assistance in improving the manuscript.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or in the decision to publish the results.
