Importance of coastal vegetated habitats for tropical marine fishes in the Red Sea

Marine vegetated habitats such as seagrass, mangroves, and macroalgae are common along tropical coastlines globally and provide habitats for a diversity of fishes, including juvenile fish and species found on coral reefs. Understanding the use of these habitats by different fish species and life stages is fundamental to spatial planning, fisheries management, and conservation. While previous studies have focused on the habitat potential of seagrass, macroalgae, or mangroves for coral reef fish independently, it is important to consider their combined roles, as tropical seascapes are often mosaics of such habitats. In this study, we evaluated habitat associations across life stages for fish species in coral reefs, seagrasses, macroalgae, and mangroves in Saudi Arabia’s central Red Sea. Through in situ visual surveys, we documented 36% of local coral reef fish species in one or more vegetated habitats, with the fraction of coral reef species utilizing macroalgae or seagrass much greater than that of mangroves (29%, 18%, and 6%, respectively). Mangroves hosted mainly juvenile fish (83% of observed population) and macroalgae hosted the largest proportion of herbivores (39% of observed population), suggesting that each environment offers different combinations of resources (food and shelter), and a mix of these habitats could support biological connectivity across a seascape. Species targeted by local fisheries made up 37% of the species documented in vegetated habitats. The use of multiple habitat types by juvenile and commercially important fishes in the Red Sea underscores the need for a holistic approach to habitat protection.


Introduction
Coral reefs support diverse fish communities, which are critical to sustaining the ecosystem functions and services provided by many tropical coastlines (Allgeier et al. 2015;Woodhead et al. 2019). Many coastal environments where coral reefs occur consist of a mosaic of different habitat types, including vegetated habitats such as mangrove forests, seagrass meadows, and macroalgae beds. While coral reef fish species are naturally associated with reefs, many of these species are known to use vegetated habitats for a variety of reasons including shelter, foraging, and nursery habitat at some point in their lives (Hemingson and Bellwood 2020;Nagelkerken et al. 2001;Whitfield 2017;Sambrook et al. 2019). The linkage of these habitats to coral reefs has even been shown to enhance the biomass and diversity of nearby coral reef fish populations (Dorenbosch et al. 2005;Mumby et al. 2004;Nagelkerken et al. 2017). By contributing to coral reef fish communities, these nonreef habitats also support coastal fisheries. Fish which utilize macroalgae habitats in tropical seascapes comprise 24% of catches in small-scale tropical fisheries (Wilson et al. 2022), and over half of fish species most strongly associated with tropical macroalgae habitats are targeted by fisheries (Fulton et al. 2020). Similarly, seagrass meadows provide nursery habitat to more than one-fifth of the largest 25 fisheries in the world (Unsworth et al. 2018). Mangrove habitat area has been found to be a good predictor of overall fishery landings (Carrasquilla-Henao and Juanes 2017) and can sustain coral reef fisheries production through their nursery functions, even on degraded reefs following the loss of structural complexity (Rogers and Mumby 2019). Through linkages to coral reefs and by providing habitats for species targeted by fisheries, coastal vegetated habitats play a crucial role for reef fishes and fisheries around the world.
While these non-reef habitats are widely known to host coral reef fishes, the degree to which coral reef fishes use mangroves, seagrasses, and macroalgae varies regionally. For example, mangroves in the Caribbean generally harbor higher densities of juvenile fishes than other non-reef habitats, while in the Indo-Pacific it is seagrass beds in which early life stage fishes are more commonly observed (Igulu et al. 2014;Sambrook et al. 2019). Macroalgae meadows often have a major overlap in fish species with local coral reefs and seagrasses; however, the structure of macroalgae canopies (i.e., canopy density, height, and cover) dictates the habitat quality these ecosystems afford to fish populations (Fulton et al. 2020). Macroalgae beds have also been found to contain more juveniles than neighboring seagrass beds or coral reefs along the Brazilian coastline ) and host a greater number of coral reef fish species than adjacent seagrass and mangroves in Papua New Guinea (Sambrook et al. 2020). In contrast, seagrass hosted more juvenile fish than mangroves or macroalgae in Tanzania (Dorenbosch et al. 2005). Factors such as biogeographic and evolutionary history, tidal cycles, and the availability and configuration of habitats have been suggested as potential drivers of these regional differences (Hemingson and Bellwood 2020;Igulu et al. 2014;Sambrook et al. 2019), highlighting the need to understand local spatial patterns, connectivity, and habitat use to effectively manage coastal resources. Despite this regional variability, studies on the use of non-reef habitats by coral reef fish are unevenly distributed, both geographically and across habitat types. A review by Sambrook et al. (2019) found that more than half of studies on multi-habitat use by coral reef fish were conducted in the western tropical Atlantic, despite being much smaller in size than the Indo-Pacific. These investigations each focused predominantly on one or two non-reef habitats, with only 17 out of 107 studies including three or more such habitats in one study area. Of these, only one study with three or more non-reef habitats was conducted outside of the western tropical Atlantic, underscoring the major disparity in our understanding of the importance of multi-habitat seascapes around the world.
The Red Sea contains one of the longest continuous living reef systems in the world, extending along 4000 km of the basin's coastline (Berumen et al. 2019). Like many tropical seas, the coastline of the Red Sea is home to several different macrophytes including seagrass, mangroves, and macroalgae growing in a marine mosaic along with coral reefs (Saderne et al. 2019;Duarte et al. 2018). Though the presence of these habitats is well-known, how Red Sea reef fish use these habitats is poorly studied. Previous research from Egyptian coastal habitats has documented several coral reef fish species and high abundances of juvenile fish inhabiting mangroves and seagrass beds, suggesting that both may be key juvenile habitats for some species (Abu El-Regal and Ibrahim 2014; Ashworth et al. 2006). McMahon et al. 2012 used otolith chemistry to directly demonstrate connectivity between coastal vegetation and reefs by populations of Red Sea snappers (Lutjanus ehrenbergii) in Saudi Arabia, showing that there were biological linkages between these habitats. These studies demonstrate connectivity between some Red Sea reef fishes and coastal vegetated habitats, but did not explore the relative importance or functional roles of various vegetated habitats to local reef fish communities. Usage of the array of Red Sea vegetated habitats by the wider coral reef fish community remains to be explored.
The degree to which fishes move between different habitats is also dependent on the spatial arrangement of ecosystems. For example, in both the Caribbean and Mozambique, coral reefs closest to shore or to seagrass and mangroves experienced the greatest enrichment of fish species which use these ecosystems as juveniles (Nagelkerken et al. 2017;Dorenbosch et al. 2007;Berkström et al. 2020). The abundance of juvenile fishes using seagrasses as nursery habitats in Curaçao was also found to be higher when seagrass was close to mangroves (Nagelkerken et al. 2001), emphasizing the complex relationships between habitats, as well as the need to understand local seascape arrangements when assessing the roles of different habitats.
In this study, we used visual surveys to investigate the fish communities present in coral reefs and three major macrophyte habitats (macroalgae, seagrass, and mangroves) in the central Red Sea. The goal was to understand the relative use of these various habitats by fish species in this region, particularly the overlap in fish communities between coral reefs and vegetated habitats. By comparing community composition and life history stages of fish found within these four habitats, we explored the varying degrees to which coral reefs and vegetated habitats support the diverse fish community of Red Sea coastal ecosystems. Our aims were to understand (1) which fish species (both reef-associated and non-reef-associated) use coral reef, mangroves, macroalgae, and seagrass as habitats, (2) how fish from different trophic groups and life history stages use each habitat, and (3) how many species targeted by fisheries use vegetated habitats.

Study site and fish surveys
This study was conducted across coastal habitats close to King Abdullah University of Science and Technology in the central Red Sea, Saudi Arabia ( Fig. 1). Much of the coast is fringed by coral reefs, with mangroves, seagrass, and macroalgae beds growing in adjacent beaches, shallow sandflats, lagoons, and reef flats (Fig. 1). Coral reefs surveyed in this study occur at varying distances from the shore, ranging from approximately 2.5-25 km from shore. While survey sites in this study were chosen due to proximity to the authors' working facilities, these habitats are representative of coastal seascapes along the Saudi Arabian coast of the Red Sea (El Shaffai 2011;Tesfamichael and Pauly 2016;Duarte et al. 2018;Roberts et al. 1992).
Macroalgae beds are found along shallow (approx. 1 m deep) inshore reef crests and flats with canopies dominated by Sargassum spp., yet contain high densities of other macroalgae including Polycladia myrica and species of Turbinaria and Padina. Macroalgae beds adjacent to coral reefs in the Red Sea are known to fluctuate seasonally (Afeworki et al. 2013) and their high biomass at the time of surveys is not believed to be a result of a coral-algal phase shift, but rather a common and persistent habitat seen in this region of the Red Sea (Roberts et al. 1992). Seagrass meadows occur on shallow (< 2 m deep) reef flats or coastal embayments, and contain several species (Thalassia hemprichii, or mixed meadows of predominantly Halophila stipulacea with Halodule uninervis and Halophila ovalis). Meadows in this area are extensive yet patchy, with areas of bare sand interspersed between seagrass patches running along hundreds of meters of reef flats or coastlines. While different seagrass species vary in size and shape, which may influence their habitat quality for fishes, survey data from all seagrass species were grouped together to understand which fishes used seagrass habitat in this area. It was outside of the scope of this paper to compare the relative habitat use of different seagrass species, and surveys were grouped together to understand the habitat functions of the mix of seagrass species found in many Red Sea shallow, coastal ecosystems.
Monospecific mangrove stands of Avicennia marina (the dominant species in this region) grow in narrow fringes in shallow water (< 1 m deep) along portions of the coastline, and typical tidal amplitudes in the central Red Sea range from daily fluctuations of < 25 cm (Churchill et al. 2019) to seasonal fluctuations of approximately 40 cm (Pugh and Abualnaja 2015). Fishing activities in this area occur predominantly on coral reefs (particularly those closest to shore), but some fishing is done in mangrove and reef flat environments.
Visual census of fishes was carried out in these four habitats along belt transects. Coral reef surveys were conducted by either snorkeling or SCUBA, while macroalgae and seagrass surveys were undertaken exclusively by snorkeling due to shallower water depth. Underwater surveys of mangrove habitat were not possible due to exceedingly shallow depth (30-60 cm) and horizontal turbidity (visibility typically < 2 m). For this reason, surveys were performed by walking slowly along the outer fringe of the mangrove pneumatophore stands, as fish encountered could be routinely identified from above the water (sensu Heupel and Bennett 2007). All surveys were conducted on days with suitable water conditions as follows: calm days with little to no wind for mangrove surveys, good underwater visibility for seagrass, macroalgae, and coral reefs surveys, and favorable sun angle to avoid casting a disruptive shadow in mangroves.
In each of the four habitats, all non-cryptic fish within belt transects (transect area described below) were identified, counted, and size-estimated (total length) to the nearest centimeter. Fishes were placed into size bins (0-5 cm, 5-10 cm, 10-15 cm, 15-20 cm, 20-25 cm, 25-30 cm, 30-35 cm, 35-40 cm, and > 40 cm) for all analyses to minimize potential observer bias. This method likely excludes many cryptic species which are very small and/or hide out of view in small crevices in each habitat. Coral reefs were surveyed at depths ranging from 1 to 10 m between 2014 and 2019. These data came from a multi-year survey effort spanning a large number of reefs in the area and used to document the broader coral reef fish community over time and space. This large data set was included in order to compare overlap in fish communities between vegetated habitats and a larger diversity of reefs, rather than the small number of reefs surveyed at the time that vegetated habitats were surveyed. The complete list of sites and survey metadata are provided in Table S 2. At each reef site, three replicate belt transects with dimensions of either 3 m × 20 m, 3 m × 25 m, or 3 m × 50 m were conducted. Transect sizes varied as they were carried out for different projects over 5 years. Macroalgae sites were surveyed in May and November 2018, whereby three replicate belt transects of 3 m × 50 m were conducted at each site. Fishes were counted 2 m out from the edge of the macroalgae canopies and 1 m into them, after which point their high density of algae precluded accurate fish identification. Seagrass and mangroves were surveyed in November and December 2019 and May 2021. At each seagrass site, two or three replicate belt transects of 4 m × 50 m were conducted, depending on the extent of the seagrass patch. At seagrass sites, surveyors maintained direction through the densest portion of the seagrass meadow, adjusting this direction as necessary to remain in suitable, continuous seagrass-covered area. Mangrove sites were visually surveyed by observers carefully and slowly walking along the edge of dense mangrove pneumatophores with two to four replicate belt transects of 3 m × 50 m at each site. Fishes were counted 2 m out from the edge of pneumatophores and 1 m into them, after which point their high density precluded accurate fish identification. Therefore, the fishes recorded represent those in direct proximity to mangrove habitat. To account for the difference in transect dimensions across habitats, fish abundance was standardized by the total area of each belt transect, resulting in the density of fish per unit area (individuals/m 2 ).
Surveys were pooled by habitat (coral reef, macroalgae, seagrass, or mangrove) regardless of the depth, habitat morphology, or month of survey in order to assess overall use of these habitats by the larger fish community. Habitats were surveyed at different times of the year to capture different seasonal environmental conditions and potentially different recruitment pulses or intensities. Information on seasonal recruitment pulses of many species in this area were unavailable at the time of this study, so surveys were done during different seasons to sample as representatively as possible and capture the presence of species at various times of the year. Similarly, data from coral reef surveys conducted at different depths or positions on the reef were pooled to gain a holistic understanding of which species use coral reefs generally in this area, and thereby which species are shared between reefs and non-reef habitats.
Different survey methods between habitats (i.e., counting fish in water via snorkeling/diving in seagrass, macroalgae, and coral reef habitats vs. counting fish from above water via walking in mangrove habitats) were employed to best suit each environment, and were dictated by the water depth and visibility. Snorkeling or diving was not possible in the mangroves in this region, as the mangrove forests in the central Red Sea exist in shallow (< 60 cm) water with small tidal ranges. This, combined with calm conditions and limited water movement along mangrove fringes, allowed for better visibility from above water than horizontal visibility within water (which would limit other survey methods such as camera surveys or attempts at snorkeling surveys). Visibility from above water was suitable to accurately see and identify fish along the perimeter and within the mangrove pneumatophores.
In total, 42 surveys (one to three transect replicates each, 124 transects in total) were conducted on 10 coral reefs, 11 surveys (three transect replicates each, 33 transects in total) in 6 macroalgae sites, 6 surveys (three to five transect replicates each, 21 transects in total) in 3 seagrass sites, and 9 surveys (two to four transect replicates each, 27 transects in total) in the 5 mangrove sites (Table S2).
Biomass of fish was estimated using the average of the size bin in which each fish was placed (cm) and speciesspecific a and b constants from FishBase (listed in Table S1) (Froese and Pauly 2021), which was also standardized per unit area (g/m 2 ). Here we use the length-weight relationship W = aL b , where W is fish weight (g), L is the length of the fish (cm), and a and b are parameters defined in Froese (2006).

Size class, life history stage, trophic, and fisheries classifications
Histograms of the size classes of fish species were used to visualize the size distribution of species in each habitat. Mean lengths of fish in each habitat were calculated by assigning each fish the average length of the size bin in which it was placed (e.g., 2.5 cm for a fish in the 0-5 cm bin). The maturation length is unknown for many of the species observed in this study, so individual fish recorded in the surveys were classified as one of three life history (juveniles, subadults, or adults) based on proportions of the species' maximum known total length (TL) (taken from FishBase, listed in Table S1). Following Nagelkerken and Van Der Velde (2002) and Berkström et al. (2013), fish were classified as either juvenile (< 1/3 of species' maximum TL), subadult (1/3-2/3 of species' maximum TL), or adult (> 2/3 of species' maximum TL). Fish were also categorized by broad trophic groups according to the known diets of adult fish of each species (Froese and Pauly 2021). While fish species may have different diets at different life history stages, information on changes in diet through life stages was unavailable for most of our study species, and the diets of adult fishes were used to define the trophic categories of all individuals of one species. Trophic categories used were carnivore, herbivore, detritivore, planktivore, and omnivore, and were considered mutually exclusive. Species that did not have trophic classifications listed on FishBase at the time of the study were classified based on the trophic groups of a regionally-relevant congeneric. Trophic classifications for each species are listed in Table S 1. To compare relative proportion of life history stages and trophic groups of fish occupying each habitat, the percentages of individuals in each category (trophic group types or life history stages) that made up each survey transect were averaged for each habitat. The abundances of fish in different life history stages that made each survey transect were also averaged for each habitat.
Fish were classified as "fishery species" based on available literature of fisheries in Saudi Arabia. A species was classified as a "fishery species" if it was described as a "fishery" or "commercial" species, or observed or surveyed in fish markets on the Red Sea coast of Saudi Arabia according to: Shellem et al. (2021), Pombo-Ayora et al. (2020), Kattan et al. (2017), Al-Rashada et al. (2021), Spaet and Berumen (2015), Khalil et al. (2017), Idris et al. (2015) and Tesfamichael and Pauly (2016). Species classified as "fishery species" are listed in Table S1.

Statistical analyses and data visualization
All analyses and visualization of data were carried out in R version 4.1.0 (R Core Team 2022) or PRIMER v7. Surveys from all sampling times in each habitat were combined to include species which use each habitat through different times of the year. Non-metric multidimensional scaling (nMDS) using Bray-Curtis distance and transformed with a Wisconsin double standardization was performed on the abundance data of fish species in each transect to visualize data using the R package "vegan" (Oksanen et al. 2020). nMDS plotting was used to visually inspect relative similarities between fish assemblages in coral reefs and vegetated habitats, based on how closely the communities in different habitats clustered together.
Generalized linear models (GLM) were fitted to test relationships between biomass or abundance of fish on coral reefs and the distance of each reef from shore. GLMs using a Gamma distribution with a log link were generated with the 'glm' function in R. The relationship between biomass or abundance and distance from shore was compared between (a) coral reef fish that also used vegetated habitats, and (b) all coral reef fish species. SIMPER (similarity percentages) analysis using the Bray-Curtis measure of similarity was performed using the SIMPER analysis tool in PRIMER v7, and was used to determine dominant taxa (species that contribute most to within-group similarity at every site) as well as average dissimilarity between pairs of habitat categories.
Due to differences in survey methods between mangroves and fully submerged habitats, as well as differences in transect dimensions between habitats, we refrained from making any statistical comparison of biomass and abundance between habitats. While all fish survey data are standardized by area, differences in methods may contribute to different observation rates and confound true differences in these metrics between habitats. These values are presented, but the statistical significance of any differences was not explored. Rather, we focus in each habitat on the presence of species, as well as relative proportions of size classes and trophic groupings, to make comparisons between fish communities across habitat types.

Fish populations and species overlap across habitats
A total of 54,764 individual fish representing 187 species from 45 families were observed across the four habitat types. Seventy-eight species were found in all vegetated habitats combined. Table 1 lists the number of species and individuals observed, as well as the mean abundance and biomass of fish in each habitat. Of the species recorded on coral reefs (subsequently referred to as "coral reef species"), 36% (61 species) were also observed in at least one of the three vegetated habitats ( Fig. 2A). Furthermore, 14% (23 species) of coral reef species were recorded in more than one vegetated habitat, and 4% (7 species) were seen across all three vegetated habitats. Macroalgae hosted the largest number of coral reef species (50 species), while 30 coral reef species were observed in seagrass and 11 in mangroves. In addition to sharing the greatest number of species, coral reef fish communities (particularly on nearshore reefs) were more closely clustered to macroalgae fish communities in multivariate space (nMDS analysis) than they were to other vegetated habitats (Fig. 2B). Macroalgae and seagrass fish communities overlapped in multivariate space, while mangrove communities were more distinct from fish communities in all other habitats (Fig. 2B). Nearshore and offshore coral reef communities showed some distinction from each other, with nearshore reef communities clustering more closely to macroalgae and seagrass communities than did those on offshore reefs (Fig. 2B).
Mangrove fish communities showed high dissimilarity to other habitats (as determined by SIMPER analysis), and were more dissimilar to communities on offshore coral reefs than to those in any other habitat ( Table 2). The two habitats with the lowest dissimilarity were macroalgae and seagrass (78.0% dissimilarity). Dominant taxa (as determined by SIMPER analysis) differ between habitats, with planktivores and herbivores contributing the most to average similarity on Fig. 2 A Venn diagram of fish species richness across four habitat types (coral reef, macroalgae, seagrass, and mangroves). In total, coral reefs hosted 170 species, while macroalgae, seagrass, and mangroves harbored 52, 36, and 24 species, respectively. B Nonmetric multidimensional scaling plot of fish surveys across four habitat types: coral reefs separated by nearshore reefs (red triangles) and offshore reefs (purple triangles), macroalgae (yellow circles), seagrass (green squares), and mangroves (brown diamonds). Each data point represents the abundance of fish communities within one survey transect. Vectors of 7 species found across all habitat types are shown over the nMDS plot coral reefs, and carnivores and herbivores being dominant in vegetated habitats (Table 3).
Some of the seven species seen across all four habitat types showed differences in size distribution across habitats (Fig. 3). In mangroves, 75% of the snapper Lutjanus ehrenbergii were juveniles (0-10 cm, while the maximum juvenile size of this species is 11.7 cm) and 24% were smaller than 5 cm. In contrast, all L. ehrenbergii individuals observed outside of mangroves were larger than 5 cm. Lutjanus ehrenbergii individuals in macroalgae and seagrass ranged from 5 to 20 cm, while those in coral reefs ranged from 5 to 35 cm, revealing a progressively increasing maximum size from the mangroves to macroalgae and seagrass, then to coral reefs (Fig. 3D). Similarly, mangroves hosted smaller sizes of Lethrinus harak (majority 5-10 cm) and Parupeneus forsskali (majority 0-5 cm), while larger individuals of both species were seen in seagrass, macroalgae, and coral reefs (Fig. 3 E&F). Acanthurus gahhm and Abudefduf vaigiensis in seagrasses and mangroves had smaller average sizes than their conspecifics in macroalgae and coral reefs, and all A. vaigiensis individuals in the seagrass and mangroves were 0-5 cm (Fig. 3A, B). However, Siganus rivulatus found in the mangroves were on average larger than their conspecifics in coral reefs, macroalgae, or seagrass habitats (Fig. 3G).

Cross-shelf patterns
On coral reefs, there was a significant decrease of both abundance and biomass of fishes also found in vegetated habitats with increasing distance of the reef to shore (p < 0.001, p < 0.001, respectively) ( Fig. 5A, B, Table S3). This differs from the overall trend for abundance and biomass of all fish on coral reefs, where abundance is significantly positively correlated with distance from shore (p < 0.001) and biomass does not show a significant trend with distance from shore (p = 0.26) (Fig. 5C, D). Vegetated habitats predominantly occur close to the shore, on inshore reef flats, in shallow coastal waters, and along beaches.

Contribution of vegetated and coral reef habitats to reef-based fisheries
Marine vegetation hosted numerous species targeted by fisheries in Saudi Arabia. Twenty-nine fishery species were observed in at least one of the three vegetated habitats (Table S1). Coral reefs and macroalgae hosted the greatest number of fishery species (47 and 20 species, respectively), Siganus rivulatus. Blue lines indicate mean length of individuals in each habitat type, with the mean ± SE length labeled in each histogram followed by seagrass (13 species) and mangroves (11 species) ( Table 1). In total, 37% of species observed in vegetated habitats were fishery species. Many of these fishery species used multiple vegetated habitats, including 5 species (A. gahhm, L. harak, L. ehrenbergii, P. forsskali, and S. rivulatus) which were observed in all habitats surveyed. Three fishery species were found only in vegetated habitats and not on coral reefs (Himantura uarnak in mangroves,  Abundance and biomass of fish on coral reefs at varying distances from shore. Each point in scatter plots represents the total abundance or biomass of fish in one coral reef survey transect. Scatterplots show the abundance and biomass of species on coral reefs which were also observed in vegetated habitats (A, B, respectively), and abundance and biomass of all species seen on coral reefs (C, D, respectively). A best-fit line from generalized linear models (GLM) (red line) with 95% confidence intervals (blue shaded area) has been fitted to each scatterplot. p-values are given for each GLM. Note difference in y-axis scale between abundances in A and C Epinephelus malabaricus in macroalgae, and Gerres oyena in seagrass and mangroves) (Table S1).

Coastal vegetation as important reef fish habitat
This study shows that non-reef habitats dominated by marine vegetation provide habitats for reef-associated and fisheries-targeted fish species. Macroalgae canopies and seagrass meadows in particular hosted a large number of Red Sea coral reef fish species, far surpassing mangroves in the richness of reef fish species which they contained. This study contributes to our understanding of the varied use of nonreef habitats by fish in both a region (Red Sea) and marine vegetation type (macroalgae) underrepresented in the literature. Previous studies exploring non-reef habitat use by reef fishes surveyed macroalgae less frequently than mangroves or seagrass, and Indo-Pacific mangroves less frequently than mangroves in the tropical Atlantic (Sambrook et al. 2019). Our surveys of these three major macrophyte habitats, including understudied macroalgae habitats and Indo-Pacific mangroves, contributes to a gap in the knowledge of how these vegetated environments support tropical marine fish communities. Ecology of coral reefs and other marine organisms has historically been understudied in the Red Sea when compared with other major coral reef ecosystems (Berumen et al. 2013), and our study adds to the understanding of how coral reef fish use the seascape of this region.
Fish communities in macroalgae habitats displayed extensive overlap with those from nearby coral reefs, suggesting that macroalgae is an important and unique resource for many reef fish species in this area. This aligns with results of Sambrook et al. (2020) which also found macroalgae to host greater numbers of coral reef fish than adjacent seagrass or mangrove habitats in Papua New Guinea. Macroalgal beds surveyed in our study consisted of large canopies of predominantly Sargassum spp., forming dense structures growing on outer reef flats, providing three-dimensional structure within which fish might seek refuge and food. Over a third of the fish recorded in macroalgae habitat were herbivores, including browsing species that feed on canopyforming Sargassum (such as Naso elegans, N. unicornis, Siganus luridus, and S. rivulatus), as well as grazers and parrotfish that likely target understory algae and epiphytes. Macroalgae canopies also hosted a large number of carnivorous fish. These habitats support high levels of secondary production through their epifauna (Fulton et al. 2019), and this assortment of food production may be contributing to the diversity of reef fish seen in macroalgae habitats in this study. In both seagrass and macroalgae habitats, fish may also be taking refuge in the underlying substrate. A review of fish in macroalgae habitats found that meadows with more structurally complex hard substrates harbored more juvenile fish and species typical of coral reefs (Fulton et al. 2020). Seagrass and macroalgae in this study area grew on a combination of sand and rocky substrate, and the holes, crevices, and other structures of the carbonate platforms underlying these habitats provide extra habitat space in addition to the macrophytes' canopies, which may be a reason why they host so many coral reef fishes.
Macroalgae in tropical coral reef ecosystems are often regarded as an indicator of environmental degradation, where fast-growing macroalgae outcompetes and prevents coral settlement, particularly following mass-mortality bleaching events (McManus and Polsenberg 2004;Meltvedt and Jadot 2014). However, many canopy-forming macroalgae habitats occur naturally, even on pristine reefs (Bruno et al. 2014), and were shown in our study to be habitats for coral reef fishes. Macroalgae may therefore be a significant habitat for Red Sea coral reef fish communities which offers unique functions and resources to fish at different life history stages. As climate change and warming ocean temperatures are expected to alter global distributions of macroalgae (Munguia-Vega et al. 2018;Tanaka et al. 2012), exploring the connection between macroalgae and coral reefs will help us understand the resources available to coral reef fishes in the Anthropocene.
In comparison to the taller and denser Sargassum canopies, seagrass species in the study area (including T. hemprichii, H. stipulacea, H. uninervis, and H. ovalis) have blades generally shorter than 15 cm, offering far less threedimensional structure in which fish can hide from predation than the macroalgal canopies which can grow over 50 cm in height. Despite this, seagrasses in this study supported 18% of species found on coral reefs. While this habitat may not offer the same refuge from predation as larger macroalgae, fishes can still experience reduced predation in seagrasses . Seagrasses of similar blade sizes and genera have been shown to provide habitats for juvenile and coral reef fishes in the Egyptian Red Sea (Ashworth et al. 2006) and in other parts of the world (Bradley et al. 2019;Eggertsen et al. 2022;Dorenbosch et al. 2005), indicating that these smaller macrophytes can still serve foraging or sheltering functions. Seagrass meadows in this study are also likely close enough to reefs that fish can make regular foraging trips to take advantage of food resources in the seagrass before returning to the reef or other habitats for shelter. Other larger species of seagrass (such as Enhalus acoroides, which can have blades up to 2 m long) are present in the Red Sea (El Shaffai 2011) but were not observed in this study. It is likely that seagrass beds comprised of these larger species would provide a better physical framework and refuge for fish, as was seen in both southern Japan and Zanzibar Island, Tanzania, where meadows of taller E. acoroides seagrass hosted a greater diversity and abundance of fishes than smaller T. hemprichii in the same region (Nakamura and Sano 2004;Gullström et al. 2008).
Many of the taxonomic groups seen across vegetated habitats in this study align with the fish communities seen in such habitats globally. Families found in macroalgae habitats in this study (e.g., Labridae, Lethrinidae, Lutjanidae, Serranidae, and Siganidae) were seen in other macroalgae habitats around the world (Fulton et al. 2020;Tano et al. 2017;Sambrook et al. 2020) and contain species targeted by fisheries in both this and other regions (Fulton et al. 2020). Seagrass and mangroves are important nursery habitats for species like Sphyraena barracuda in the Caribbean and Florida (Jones et al. 2010;Nagelkerken et al. 2001) and L. harak and Lutjanus fulviflamma in Tanzania (Kimirei et al. 2013), and these species were seen in both seagrass and mangroves in this study, reinforcing the use of these habitats in the life cycles of species throughout the tropics. The fish documented in this study not only reveal the local patterns of habitat use in the central Red Sea, but reflect patterns seen along coastlines throughout the tropics and highlights the reliance of a diversity of taxa on the presence of these vegetated habitats.

Use of mangroves by Red Sea fishes
Mangroves in some parts of the world are known to be habitats rich in coral reef fishes (Igulu et al. 2014). In our study area, however, they displayed the least overlap with coral reef fish assemblages, hosting the lowest diversity of coral reef fish species. In contrast, Caribbean mangroves were found to be nursery habitats for many species present on neighboring coral reefs (Mumby et al. 2004), and to harbor a greater density of juvenile fishes than seagrass beds (Nagelkerken and Van Der Velde 2002). This geographical difference is likely due in part to both the species of mangroves present and the local tidal regime. Avicennia marina is the dominant mangrove in the Red Sea (Ortega et al. 2020) and has pneumatophores which have a straight trunk architecture protruding from the sediment. This structure is quite unlike Caribbean species such as Rhizophora mangle or Laguncularia racemosa, which possess larger and more complex prop root structures with buttressing architecture. Avicennia marina root systems have a smaller volume of interstitial space between pneumatophores than the branching, prop-root structures of other mangroves species (Rhizophora mucronata and Bruguiera gymnorhiza) (Vorsatz et al. 2021), which limits the size of organisms which can enter these small spaces. While this prevents larger fish from entering the mangroves, juvenile fishes may benefit greatly during times of optimal inundation, when access is viable for smaller-bodied fishes but restricted for larger predatory fishes. Avicennia marina has been identified as a habitat for many juvenile fishes in Australian estuaries (Hindell and Jenkins 2004;Sheaves et al. 2016), and may also here be acting as a valuable juvenile habitat for some species in the Red Sea.
Compounding the limited interstitial volume of A. marina roots, central Red Sea mangroves are rarely deeply inundated as they grow in shallow water with a small tidal amplitude. Changes in sea surface height across seasons is greater than the daily tidal range in this area and is driven by wind movement, with maximum sea surface height occurring during the north-east monsoon (winter). Limited inundation restricts both the size of fish which can enter and the epibionts which can grow in this habitat, altering access to and food resources within the mangrove pneumatophores. Previous studies comparing the Caribbean (with typical tidal amplitudes < 1 m) and the Indo-Pacific (with typical tidal amplitudes ~ 2-4 m) saw that areas with smaller tidal amplitudes had a larger number of reef species or greater density of juvenile fishes in mangrove habitats, as the lower variation in Caribbean mangrove water levels meant that mangroves remained inundated even at low tide (Igulu et al. 2014;Hemingson and Bellwood 2020). Our study found fewer reef fish species using mangroves in an area with a small tidal range, but follows the same principle that the use of and access to mangroves by reef fish is governed by levels of inundation. Weak inundation of Red Sea mangroves frequently leads to the shallower or more onshore portions of the mangroves drying out, eliminating access to the habitat's benthic and epiphytic food resources and regularly impairing its ability to provide adequate shelter from larger predatory fish and shorebirds.
Our results reflect the value of mangrove habitat for juvenile fishes, as mangroves predominantly hosted this life history stage. The majority of juveniles were classified as carnivores, which could be the result of this habitat and the surrounding sandflats hosting important prey (small fish and invertebrates) for small predatory fishes. As these fishes age and grow, they may undertake ontogenetic migration to alternative habitat, as the combination of shallow water and mangrove root architecture limits the size and species of fish that can use this habitat. Mangroves hosted a limited number of coral reef fish species and were used by several species exclusively associated with mangroves, suggesting that mangroves harbor a unique community and could be serving a specialized role for some fish species (possibly as nursery habitats, shelter from avian predation, or feeding grounds). Two coral reef species (Taeniura lymma and A. bifasciatus) were seen in mangroves and no other vegetated habitat, suggesting that while seagrass and macroalgae contain a greater number of reef species, the degradation of mangroves would still remove habitats for a unique assemblage of fishes not seen in other vegetated habitats.

Potential for ontogenetic shifts through habitats
While it was not possible to track the individual movements of fishes throughout their lives in this study (e.g., using telemetry or genetic tools), it is likely that the diversity and proximity of vegetated habitats allow fish to move through different habitats as they age and grow. Lutjanus ehrenbergii was one species observed across all four habitats, with the smallest size classes of juveniles inhabiting the mangroves (and no other habitats) and increasingly larger individuals inhabiting macroalgae and then seagrass and coral reefs. Individuals of other species seen across all four habitats (A. vaigiensis, L. harak, and P. forsskali) had smallest average length in the mangroves and became increasingly larger in other habitats. This could be the results of ontogenetic shifts in habitat use by these species, and may mean that mangroves act as juvenile habitats for multiple coral reef species. Lutjanus ehrenbergii is known to make ontogenetic shifts between inshore seagrass and mangroves and coral reefs in the central Red Sea (McMahon et al. 2012). The shift in size classes seen in our study by multiple species may therefore reflect this previously observed ontogenetic migration for some coral reef fish species in the Red Sea.
In contrast, the average size of S. rivulatus (which was also observed in all vegetated habitats) was greater in the mangroves than in other habitats, and no juveniles of this species were observed in mangroves. This indicates that they do not use the mangroves as nursery habitats, and may instead be using them as feeding grounds when they are older. This highlights the variety of functions that mangrove habitats provide for different species.
As information on the seasonality of juvenile fish recruitment is sparse in this region (particularly in non-reef habitats) (Robitzch and Berumen 2020), it is possible that the surveys in this study captured recruitment pulses for certain species and missed them for others. This could affect the relative proportions of juveniles documented in each habitat. There were differences in juvenile abundances between surveys in May and November in this study, but because these surveys were not done at consistent sites (some sites were only sampled in one season or time point), we hesitate to make any conclusions about the timing of recruitment at any of these sites. In the future, more frequent observations throughout the year could resolve seasonal patterns of habitat use and further our understanding of the connectivity between and nursery functions of tropical coastal habitats.

Potential for connections between inshore reefs and coastal vegetation
Nearshore reefs had greater abundances of species that were found in vegetated habitats and had more similar communities to marine vegetation than did offshore reefs, which could be the result of connections between reefs and coastal vegetation. Seascape connectivity between reefs, seagrass, and mangroves has a large effect on the fish assemblages of coral reefs (Olds et al. 2012), and the degree of connectivity between such habitats can be limited by distance. In the Caribbean, enhancement of fish species which used mangroves and seagrass as juvenile habitats was seen on reefs < 4 km from mangroves and seagrass (Nagelkerken et al. 2017), and were reduced or absent on reefs > 9 km from seagrass and mangroves (Dorenbosch et al. 2007). In Mozambique, the abundance and biomass of species which used seagrass as nursery were significantly correlated to distance from seagrass habitats, with the abundance of these species dropping sharply on reefs > 8 km away from seagrass . Abundance of all fish on coral reefs in this study increased with distance from shore, possibly due to lower fishing pressures, suggesting that the decrease in vegetationassociated fishes was not due to a general decrease in coral reef fish populations. This could mean that vegetated habitats in the Red Sea provide nursery or habitat function to reef fish on coral reefs closest to the coast, and any potential enrichment to reef fish communities by vegetated habitats is spatially limited.
Further experimentation, observations, or fish tracking would be required in the future to understand the extent of fish connectivity between these habitats. Other factors besides habitat proximity influence rates of exchange across habitats. Hydrodynamic circulation (i.e., flow of water currents) can transport fish (particularly in their larval phase) across seascapes (Simpson et al. 2013) and can facilitate biological connections between habitats at great distances from each other (Wang et al. 2019). The degree to which fish use habitat patches can also be dictated by the quality of each habitat (i.e., size and physical structure) (van Lier et al. 2018;Staveley et al. 2017), further indicating that proximity between two habitats does not guarantee connections between them. Since survey data were pooled by habitat in this study, it was not possible to examine the different levels of connectivity between reef zones or vegetated habitats with different morphologies or abundance of macrophyte species. In the future, characterizing the broader environment by measuring water circulation and habitat patch characteristics would give more insight into how fish move across tropical seascapes.

Fisheries potential of marine vegetation
Seagrass, mangroves, and macroalgae in this study hosted commercially important Red Sea fish species, providing evidence for their relevance to local fisheries. Marine vegetated habitats support valuable fisheries around the world, either through replenishing other habitats with fish or through fishing directly in such habitats (Fulton et al. 2020;Unsworth et al. 2018;Carrasquilla-Henao and Juanes 2017). Some of the fishery species found in vegetated habitats are non-trivial components of local fisheries revenue. For example, the grouper E. malabaricus (found only in macroalgae in this study) and endangered emperor Lethrinus mahsena (IUCN 2022) (found in coral macroalgae, and seagrass) made up 9% and 3.29%, respectively, of revenue in central Red Sea fish markets (Shellem et al. 2021). Macroalgae in this study hosted more fishery species than seagrass or mangroves and had the greatest overlap with coral reef fish assemblages, highlighting their potential value to reef-associated fisheries. Many of the fish living in vegetated habitats, particularly mangroves, are juveniles with low biomass and thereby low current value to fisheries. However, as these juveniles could be replenishing the adult population of fish on coral reefs (where most of the fishing in this area takes place), the standing biomass of fish in these habitats may be less important than their supporting role to the fish community across this marine habitat mosaic. This study did not measure food resource availability or experimentally test how each habitat provided nursery functions, so we cannot pinpoint the degree to which various habitat characteristics draw fish to each respective habitat. Collecting information on habitat arrangement and quality (i.e., distance to other habitat patches and the density of macrophytes in each habitat) and correlating it to reef fish use in each habitat patch would further reveal how these habitats attract and host coral reef fishes. Further research into what these habitats provide for coral reef fishes is needed in order to better understand the links between reef fish and these vegetated habitats.

Limitations and future research
Comparisons between the fish communities in each habitat are limited in this study by variations in survey methods. Survey transect sizes differed between habitats, as well as between some reef sites as a result of using a long-term data set with surveys from multiple projects. Differences in transect size affect the detection of fish with different sizes, speeds, and behaviors (Pais and Cabral 2018), so it was not possible to directly compare the abundance or biomass of fishes between habitats. This study also pooled survey data from habitats with different morphologies, depths, and other habitat characteristics. While this allowed for a broad understanding of the diversity of species which make use of these representative habitats along the Red Sea coastline, it was not possible to identify recruitment pulses or test for the effects of habitat quality on fish presence or abundance. This study gives a holistic view of how reef and non-reef associated fishes use the diversity of coastal habitats in the central Red Sea, and future study designs should include more regular and standardized sampling efforts to quantify the seasonality and relative use of each habitat by Red Sea fishes.

Conclusions
This study revealed how multiple habitat types support Red Sea coral reef fish communities and species targeted by fisheries. Observations in multiple habitat types provided information on how fishes use habitats through a whole seascape, showing the importance of surveys across multiple, understudied habitats. The species, life history stages, and trophic groups of coral reef fishes in each habitat vary, suggesting that each habitat attracts different kinds of fish with different resources or types of shelter. Preserving spatial connectivity and seascape mosaics through marine spatial planning will therefore be important for supporting productive, resilient reef communities and fisheries. included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.