A survey of fishes associated with Hawaiian deep-water Halimeda kanaloana (Bryopsidales: Halimedaceae) and Avrainvillea sp. (Bryopsidales: Udoteaceae) meadows.

The invasive macroalgal species Avrainvillea sp. and native species Halimeda kanaloana form expansive meadows that extend to depths of 80 m or more in the waters off of O'ahu and Maui, respectively. Despite their wide depth distribution, comparatively little is known about the biota associated with these macroalgal species. Our primary goals were to provide baseline information on the fish fauna associated with these deep-water macroalgal meadows and to compare the abundance and diversity of fishes between the meadow interior and sandy perimeters. Because both species form structurally complex three-dimensional canopies, we hypothesized that they would support a greater abundance and diversity of fishes when compared to surrounding sandy areas. We surveyed the fish fauna associated with these meadows using visual surveys and collections made with clove-oil anesthetic. Using these techniques, we recorded a total of 49 species from 25 families for H. kanaloana meadows and surrounding sandy areas, and 28 species from 19 families for Avrainvillea sp. habitats. Percent endemism was 28.6% and 10.7%, respectively. Wrasses (Family Labridae) were the most speciose taxon in both habitats (11 and six species, respectively), followed by gobies for H. kanaloana (six species). The wrasse Oxycheilinus bimaculatus and cardinalfish Apogonichthys perdix were the most frequently-occurring species within the H. kanaloana and Avrainvillea canopies, respectively. Obligate herbivores and food-fish species were rare in both habitats. Surprisingly, the density and abundance of small epibenthic fishes were greater in open sand than in the meadow canopy. In addition, species richness was also higher in open sand for Avrainvillea sp. We hypothesize that the dense holdfasts and rhizoids present within the meadow canopy may impede benthic-dwelling or bioturbator species, which accounted for 86% and 57% of individuals collected in sand adjacent to H. kanaloana and Avrainvillea sp. habitats, respectively. Of the 65 unique species recorded in this study, 16 (25%) were detected in clove oil stations alone, illustrating the utility of clove-oil anesthetic in assessing the diversity and abundance of small-bodied epibenthic fishes.

Introduction

Macroalgal meadows constitute important habitats for reef- and nearshore fish species. Many are important grazing areas for herbivorous fishes (Lobel & Ogden, 1981) and may also serve as spawning sites for recreationally important food fishes such as parrotfish and wrasses (Colin & Bell, 1991). Macroalgae may also serve a key role in ontogenetic habitat shifts in post-settlement fish (Eggleston, 1995; Dahlgren & Eggleston, 2000). Relative to surrounding habitats, which are often sandy and of low-relief, macroalgal meadows constitute highly-complex and rugose habitats which may afford protection to juvenile and small-bodied reef fish species alike.

Until recently, most detailed studies of reef fish diversity have been limited to 30 m or less (e.g., Randall, 1998; Greenfield, 2003), which is the generally accepted limit for conventional SCUBA diving. However, with the recent advent of Closed Circuit Rebreathers (CCR) and mixed-gas diving technology, properly-trained researchers are now able to work at depths of 100 m or more (e.g., Pyle, 2000; Lesser, Slattery & Leichter, 2009; Khang et al., 2010; Kane, Kosaki & Wagner, 2014; Kosaki et al., 2016; Simon et al., 2016). Recent work has focused on the mesophotic zone, which extends from 30 to 150 m (Hinderstein et al., 2010). Much of this work has focused on coral reef habitats and their associated fauna. Few studies have investigated the fauna of mesophotic macroalgae, despite the fact that several meadow-forming species occur at depths of 50 m or more in tropical and subtropical waters (Huisman, Abbott & Smith, 2007; Spalding, 2012; Pyle et al., 2016).

In this paper, we describe the fish fauna associated with two deep-water macroalgal species, Avrainvillea sp. and Halimeda kanaloana, from Hawaiian waters. Avrainvillea sp. and H. kanaloana are siphonous green macroalgae which form predominantly monospecific meadows over large areas of sandy substrate from shallow (1 m) to deep (>80 m) waters in the Main Hawaiian Islands (Spalding, 2012). Halimeda kanaloana is a calcified alga native to Hawai‘i with multiple branched axes up to 30 cm in height (Verbruggen et al., 2006). Avrainvillea sp., an invasive species that was previously misidentified as A. amadelpha (Wade, Tang & Sherwood, 2015), forms dense, mat-like beds with bladed canopies approximately 10 cm in height (Spalding, 2012). Avrainvillea sp. first appeared off Kahe Point, O‘ahu in 1981, and subsequently spread to Maunalua Bay, O‘ahu (Brostoff, 1989), where it outcompeted native algae and seagrasses (Peyton, 2009; Abbott & Huisman, 2004). Both species have been reported to support a greater diversity of epibenthic or infaunal invertebrates when compared to surrounding sandy habitats (Fukunaga, 2008; Magalhaes & Bailey-Brock, 2014); however, little is currently known about the role these assemblages may play in the creation/loss of unique habitat for fishes. Because Avrainvillea sp. and H. kanaloana occupy similar habitats in Hawai‘i (sandy substrate in moderate to low wave environments), they offer an opportunity to determine the effects of canopy type on the composition of the associated fish fauna.

Since little information is currently available for these habitats, our primary goal was to provide baseline information on the fish fauna associated these meadows. As part of this goal, we sought to document the most common and abundant species in both habitats. We also sought to identify any commercially- or recreationally important fish species so that resource managers can determine if these meadows merit additional study or protection. Based on surveys from shallow-water meadows (e.g., Lobel & Ogden, 1981; Francini-Filho et al., 2010), we hypothesized that deepwater H. kanaloana and Avrainvillea sp. meadows might be important feeding areas for herbivorous species such as surgeonfishes or parrotfishes, many of which are prized food-fish species in Hawai‘i. In addition, we also attempted to calculate the level of endemism in both macroalgal habitats. Given that deep reefs in the Northwestern Hawaiian Islands (NWHI) support a greater number of endemic species than their shallow-water counterparts (Kane, Kosaki & Wagner, 2014), we hypothesized that deep water macroalgal meadows might likewise support a greater proportion of endemic species than shallow habitats in the Main Hawaiian Islands (MHI).

Our final goal was to compare the abundance and diversity of small-bodied epibenthic fishes between open sand and meadow canopy subhabitats. Because both macroalgal species form structurally complex three-dimensional canopies, we hypothesized that they would support a greater abundance and diversity of fishes when compared to surrounding sandy habitats, which are typically of low relief and complexity. This hypothesis is supported by the work of Chittaro (2004), who found that fish abundance and species richness in Tague Bay, St. Croix were positively correlated with H. incrassata habitats and negatively correlated with open sand and pavement. Likewise, Ornellas & Coutinho (1998) found that fish diversity in sublittoral areas of Cabo Frio Island (Brazil) was greater in Sargassum furcatum beds than in surrounding sandy habitats. Thus, it seems likely that H. kanaloana and Avrainvillea sp. meadows should support a greater diversity and abundance of fishes than adjacent sandy areas.

Materials and Methods

We utilized CCR technical diving to survey fish assemblages during a total of 20 dives (four in Avrainvillea sp. habitats and 16 in H. kanaloana habitats). Avrainvillea sp. habitats were surveyed at a single site off west O‘ahu and H. kanaloana habitats were surveyed at three sites off of south and west Maui (Fig. 1). All surveys were conducted between June 14th, 2005 and June 12th, 2006. Initial surveys consisted of visual censuses supplemented by collections made with pole spears. All other surveys were conducted using tandem visual surveys combined with collections using clove-oil anesthetic in order to better assess the numbers of small epibenthic fishes (Fig. 2). For these collections, an anesthetic solution of 10% clove oil in 90% ethanol was aspirated from a squirt bottle beneath a 1.5 m weighted plastic tarp. The tarp was placed haphazardly over either sand or canopy. The solution was allowed to work for a period of 10 min, during which the diver conducted visual- and photographic surveys of larger-bodied fish species (see below). After the 10 min had elapsed, the divers removed the weighted tarp and used fine-mesh nets to collect the fishes, which were photographed and preserved in 10% formalin for subsequent identification. Because fishes caught in the clove oil stations were collected from a known area (1.5 m), we were able to compare the species richness and abundance (average number of species and individuals per 1.5 m collection) between sand and canopy sub-habitats.

Figure 1

Map of the Main Hawaiian Islands showing the locations of the four survey sites.

Avrainvillea sp. meadows were surveyed off west O‘ahu. Halimeda kanaloana meadows were surveyed off south and west Maui.

Figure 2

Survey techniques used in Halimeda kanaloana (A–D) and Avrainvillea sp. meadows

Large-bodied species were surveyed visually (A) or photographically (B). Unidentified species were collected with small spears (C). Small-bodied epibenthic fishes were surveyed by injecting a clove-oil solution under a 1.5 m weighted tarp (D). The anesthetized fishes were collected with fine-mesh nets and preserved for subsequent identification. Diver conducting a visual survey of Avrainvillea sp. meadow (E). Note the denser canopy.

Conventional transect-based visual surveys were impractical given the limited bottom-time, mobility, and task-load of diver teams. Instead, large-bodied fishes were surveyed using a stationary point count (SPC) in which a diver recorded all fishes that resided or passed through a visually estimated 10 m cylinder centered on the diver’s location (the weighted tarp). The survey time for each SPC was typically 10 min. Additional species were collected or surveyed via opportunistic spearfishing and photographs. All fish collections were performed in accordance with a University of Hawai‘i Institutional Animal Care and Use Protocol (# 06-058). Permission to use clove oil anesthetic was granted by the Hawai‘i Department of Land and Natural Resources (Permit #s PRO 2006-28 and PRO 2007-47).

Data analysis

All specimens were identified to species or lowest-possible taxon and classified as endemic (restricted to the Hawaiian Islands, Midway, and Johnston Atoll) or non-endemic using available references (e.g., Randall, 2007; Mundy, 2005). For both clove oil collections and visual surveys, we recorded the abundance (N) of each species as well as the habitat (meadow-forming species) and sub-habitat (canopy or sand & rubble) where each species was observed or collected. We used this information to calculate the species richness (S) and a Shannon–Wiener diversity index (log e; H′) for each sub-habitat. We calculated percentage occurrence (%Occ) as the proportion of collections made within each sub-habitat in which a species was recorded, and percentage relative abundance (%RA) as the number of individuals of a species recorded within a sub-habitat divided by the total number of individuals that were collected or observed in that sub-habitat. We follow Randall (1996) in calculating the percent endemism by dividing the number of endemic species by the total number of species present in a particular habitat. We also used available references (Randall, 1996; Randall, 2007) and personal observations to identify species that reside-upon or feed within the sediment as benthic associates or bioturbators (B) in order to determine if abundance of these species differed between sub-habitats. For clove oil collections, we estimated the density (D) of fishes collected by dividing the number of individuals of each species by the total area of the substrate that was sampled. We compared the median abundance (N), density ( D), and species richness (S) within each sub-habitat using the Mood Median Test, as the resulting data did not meet the assumptions necessary to use parametric statistical tests.

Results

We conducted a total of 14 visual surveys and 51 tandem surveys (visual surveys + collections using clove oil anesthetic) in Avrainvillea sp. and H. kanaloana habitats (Table 1).

Table 1

Summary of visual surveys and tandem (visual+clove-oil) collections by habitat type and location.

	Avrainvillea sp.		Halimeda kanaloana
Survey Type	Canopy	Sand/Rubble	Canopy	Sand/Rubble
Visual Only	2	0	8	4
Clove Oil + Visual	8	3	28	12
Depth Range	37–47 m		11–40 m
Dates	10/26/2005–6/12/2006		5/15/06–5/23/06
Collection Locations
	Makaha, O‘ahu		Kahekili Beach Park, Maui 20°56′14.33″N 156°41′40.54″W
	21°26′50.22″N 158°12′42.24″W		Honokowai Beach Park, Maui 20°57′21.00″N 156°41′19.96″W
			Makena Beach Park, Maui 20°37′42.35″N 156°27′15.04″W

Percent occurrence and relative abundance of fishes in H. kanaloana meadows

A total of 49 species from 25 families were recorded from H. kanaloana meadows and surrounding sandy areas (Table 2). Overall species richness and diversity were nominally greater in canopy (S = 31 species, H′ = 2.523) than in open sand (S = 29 species, H′ = 2.064). Wrasses (Family Labridae) were the most speciose taxon within both sub-habitats followed by Gobies. Wrasses were also the most abundant taxon within meadow canopy, accounting for 54.8% of individuals collected or observed. In contrast, gobies comprised only 15.9% of individuals surveyed in the H. kanaloana canopy. The most abundant and frequently-occurring species within meadow canopy was the Two-spot wrasse, Oxycheilinus bimaculatus. Other commonly-occurring species included the goby, Gnatholepis spp., and the wrasses Pseudojuloides cerasinus and Cymolutes praetextatus. The latter species was previously unknown east of the Marshall Islands, and thus constitutes a new record for Hawai‘i (see Randall, Langston & Severns, 2006).

Table 2

Checklist of fishes associated with deep-water Halimeda kanaloana meadows based on clove oil collections and visual surveys.

Sub-habitats are listed as Meadow Canopy (directly within vegetation) and Sand/Rubble (blow-outs and sandy & meadow perimeters). Endemic species are indicated by “E” whereas those which rest directly upon- or feed within the substrate are indicated by “B”. CO(N) and V(N) indicate the numbers of each species collected or surveyed in clove oil collections or visual surveys, respectively. All other abbreviations are described in the methods.

		Meadow canopy				Sand/Rubble
Species	E, B	CO(N)	V(N)	% Occ	% RA	CO(N)	V(N)	% Occ	% RA
Myliobatidae
Aetobatus narinari							1	6.3	0.2
Manta birostris							1	6.3	0.2
Congridae
Conger cinereus		1		2.8	0.3
Synodontidae
Synodus spp.	B		4	8.3	1.2
Aulostomidae
Aulostomus chinensis			2	2.8	0.6
Fistulariidae
Fistularia commersonii			6	8.3	1.8
Apogonidae
Foa brachygramma	B		5	5.6	1.5	2		6.3	0.4
Pristiapogon kallopterus	B		1	2.8	0.3		30	6.3	5.9
Carangidae
Caranx melampygus			1	2.8	0.3
Lutjanidae
Aprion virescens							1	6.3	0.2
Mulllidae
Mulloidichthys spp.	B						1	6.3	0.2
Scorpaenidae
Pterois sphex	E					1		6.3	0.2
Scorpaenopsis diabolus	B		1	2.8	0.3
Chaetodontidae
Chaetodon miliaris	E						1	6.3	0.2
Heniochus diphreutes							1	6.3	0.2
Pomacanthidae
Centropyge fisheri	E		2	2.8	0.6
Pomacentridae
Dascyllus albisella	E					2	50	12.5	10.2
Labridae
Cheilio inermis		1	7	5.6	2.4		2	6.3	0.4
Cymolutes lecluse	E, B	1	8	11.1	2.7
Cymolutes praetextatus	B		10	13.9	3.0
Iniistius baldwini	B						1	6.3	0.2
Iniistius umbrilatus	E, B						1	6.3	0.2
Novaculichthys taeniourus	B		7	5.6	2.1
Oxycheilinus bimaculatus		10	104	41.7	34.1	5		12.5	1.0
Pseudojuloides cerasinus		2	24	16.7	7.8	1	13	12.5	2.7
Pseudocheilinus evanidus							1	6.3	0.2
Pseudocheilinus tetrataenia							1	6.3	0.2
Stethojulis balteata	E, B		9	13.9	1.2
Callionymidae
Callionymus decoratus	E, B		2	2.8	0.6
Synchiropus corallinus	B					2		6.3	0.4
Synchiropus rosulentus	E, B					6		6.3	1.2
Synchiropus spp.	B					1	79	12.5	15.7
Pinguipedidae
Parapercis schauinslandii	B		27	8.3	8.1	4	2	25.0	1.2
Gobiidae
Eviota susanae	E, B					1		6.3	0.2
Gnatholepis spp.	B	21	1	33.3	6.6	13	1	37.5	2.7
Opua nephodes	E, B	1	28	11.1	8.7	15	151	31.3	32.5
Priolepis eugenius	E, B					1		6.3	0.2
Priolepis farcimen	E, B					1		6.3	0.2
Psilogobius mainlandi	E, B	2		5.6	0.6	11	96	31.3	21.0
Microdesmidae
Gunnelichthys curiosus	B		4	8.3	1.2		2	6.3	0.4
Acanthuridae
Acanthurus blochii			3	2.8	0.9
Bothidae
Bothus pantherinus	B		4	11.1	1.2		5	25.0	1.0
Balistidae
Rhinecanthus aculeatus	B		2	2.8	0.6
Monacanthidae
Aluterus scriptus			1	2.8	0.3
Ostraciidae
Ostracion meleagris			2	5.6	0.6
Tetraodontidae
Arothron hispidus	B		4	5.6	1.2
Canthigaster coronata	B		2	5.6	0.6
Canthigaster jactator	E, B		23	19.4	6.9		3	6.3	0.6
Diodontidae
Diodon hystrix	B		1	2.8	0.3

In contrast to H. kanaloana canopy, adjacent sandy areas were numerically dominated by goby species (Total %RA = 56.9%), whereas wrasses only accounted for 4.9% of individuals surveyed within this sub-habitat. The three most commonly-occurring species were gobies: Gnatholepis spp., Opua nephodes, and Psilogobius mainlandi.

Percent occurrence and relative abundance of fishes in Avrainvillea sp. meadows

A total of 28 species from 19 families were recorded from Avrainvillea sp. meadows and adjacent sandy habitats (Table 3). Overall species richness and diversity were nominally greater in open sand (S = 19 species, H′ = 2.74) than in meadow canopy (S = 13 species, H′ = 1.871). Wrasses were the most speciose and abundant taxon in both sub-habitats. All other taxa were represented by two species or fewer.

Table 3

Checklist of fishes found in association with deep-water Avrainvillea sp. meadows based on clove-oil collections and visual surveys.

All abbreviations follow Table 2.

		Meadow canopy				Sand/Rubble
Species	E, B	CO(N)	V(N)	% Occ	% RA	CO(N)	V(N)	% Occ	% RA
Muraenidae
Gymnothorax spp.		1	1	20.0	2.2
Serranidae
Plectranthias nanus	B					1		33.3	2.3
Pseudanthias bicolor						2		66.7	4.7
Apogonidae
Apogonichthys perdix		9		60.0	10.0	2		33.3	4.7
Carangidae
Caranx lugubris			1	10.0	1.1
Lutjanidae
Aprion virescens			1	10.0	1.1
Mulllidae
Parupeneus multifasciatus	B						5	33.3	11.6
Scorpaenidae
Iracundus signifer	B					2		66.7	4.7
Sebastapistes fowleri	B	1		10.0	1.1	2		66.7	4.7
Chaetodontidae
Chaetodon kleinii						1		33.3	2.3
Pomacentridae
Chromis leucura						1		33.3	2.3
Labridae
Bodianus bilunulatus albotaeniatus	E		1	10.0	1.1
Cirrhilabrus jordani	E		21	10.0	23.3
Oxycheilinus bimaculatus		4		30.0	4.4	2	3	100.0	11.6
Pseudocheilinus evanidus		3		30.0	3.3		4	33.3	9.3
Pseudocheilinus octotaenia						1		33.3	2.3
Pseudojuloides cerasinus		1	14	20.0	16.7
Callionymidae
Synchiropus corallinus	B					1		33.3	2.3
Pinguipedidae
Parapercis schauinslandii	B					6		66.7	14.0
Gobiidae
Gnatholepis spp.	B						2	33.3	4.7
Acanthuridae
Naso caesius			30	10.0	33.3
Bothidae
Bothus pantherinus	B						1	33.3	2.3
Soleidae
Aseraggodes borehami	E, B					1		33.3	2.3
Balistidae
Xanthichthys auromarginatus			1	10.0	1.1
Monacanthidae
Cantherhines dumerilii	B						2	33.3	4.7
Aluterus scriptus							3	33.3	7.0
Tetraodontidae
Arothron hispidus	B		1	10.0	1.1
Canthigaster coronata	B						1	33.3	2.3

The most commonly occurring species within meadow canopy was the cardinalfish Apogonichthys perdix. This species was recorded only from clove oil collections and was never observed in visual surveys. Other common species found in Avrainvillea sp. canopy include the wrasses O. bimaculatus and Pseudocheilinus evanidus. The two most abundant species were the unicorn fish Naso caesius and Hawaiian Flame Wrasse, Cirrhilabrus jordani, however, these species were recorded from a single collection each.

Only three collections were made in sandy habitats adjacent to Avrainvillea sp. meadows. Overall, wrasses were the most abundant taxon in open sand (%RA = 23.3%). The most frequently-occurring species was the wrasse O. bimaculatus, followed by the sandperch Parapercis schauinslandii, which was also the most abundant species.

Abundance, species richness, and diversity of epi-benthic fishes from clove oil anesthetic collections

A total of 105 individuals from 19 species were collected from H. kanaloana habitats using clove oil anesthetic (Table 4). The eyebar goby, Gnatholepis anjerensis and wrasse O. bimaculatus, were the most abundant and frequently-collected species in H. kanaloana canopy. In contrast, the cloud goby, Opua nephodes, and Hawaiian shrimp goby Psilogobius mainlandi were the most abundant fishes in in open sand, and were rarely collected from meadow canopy.

Table 4

Abundance and density of fishes from 1.5 m clove oil collections within Halimeda kanaloana canopy (n = 28) and surrounding sand & rubble (n = 12) sub-habitats.

Median abundance (fish per 1.5 m collection, N), species richness (S), and a Shannon–Wiener diversity index (H′) are included at bottom. All other abbreviations follow Table 2. Note that the abundance of vagile species (*) may be underestimated as these species tend to swim away when the tarp is being deployed.

Species	E, B	Meadow (N)	Meadow density (fish per m2)	Sand (N)	Sand Density (fish per m2)
Congridae
Conger cinereus*		1	0.0238	0	0
Apogonidae
Foa brachygramma*	B	0	0	2	0.1111
Scorpaenidae
Unidentified*		1	0.0238	0	0
Pterois sphex*	E	0	0	1	0.0556
Pomacentridae
Dascyllus albisella*	E	0	0	2	0.1111
Labridae
Cheilio inermis*		1	0.0238	0	0
Cymolutes lecluse*	E, B	1	0.0238	0	0
Oxycheilinus bimaculatus*		10	0.2381	5	0.2778
Pseudojuloides cerasinus*		2	0.0476	1	0.0556
Callionymidae
Synchiropus corallinus	B	0	0	2	0.1111
Synchiropus rosulentus	E, B	0	0	7	0.3889
Pinguipedidae
Parapercis schauinslandii*	B	0	0	4	0.2222
Gobiidae
Eviota susanae	B, E	0	0	1	0.0556
Gnatholepis anjerensis	B	20	0.4762	12	0.6667
Gnatholepis cauerensis	E, B	1	0.0238	1	0.0556
Opua nephodes	E, B	1	0.0238	15	0.8333
Priolepis eugenius	E, B	0	0	1	0.0556
Priolepis farcimen	E, B	0	0	1	0.0556
Psilogobius mainlandi	E, B	1	0.0238	11	0.6111
Median Abundance (Nx)		1		4
Median Species Richness (Sx)		1		1
Total Species Richness (S)		10		15
Diversity (H′)		1.5013		2.2474
Average Species Density m2			0.0489		0.1930
Median Density of Fishes m2			0.6667		2.6667
Average Density of Fishes m2			0.9286		3.6667

The median abundance and density of fishes was significantly higher in open sand (N = 4.00 individuals/collection, D = 2.67 fish per m2) when compared to the meadow canopy (N = 1.00, D = 0.67; Chi-Square = 13.41, DF = 1, P = 0.000 for both tests). Total species richness was also nominally higher in open sand (15 vs. 10 species); however, median species richness did not differ significantly between sub-habitat types (Chi-Square = 0.63, DF = 1, P = 0.426). We calculated Shannon Weiner Diversity Index values of 2.2474 and 1.5013, respectively for open sand and canopy.

A total 16 species and 42 individuals were collected from Avrainvillea sp. sub-habitats using clove oil anesthetic (Table 5). The cardinalfish Apogonichthys perdix was the most abundant species collected from Avrainvillea sp. canopy whereas the sandperch Parapercis schauinslandii was most abundant species collected in open sand.

Table 5

Abundance and density of fishes from 1.5 m clove oil collections within Avrainvillea sp. canopy (n = 8) and surrounding sand & rubble (n = 3) sub-habitats.

Median abundance (fish per 1.5 m collection, N), species richness (S), and a Shannon–Wiener diversity index (H′) are included at bottom. All other abbreviations follow Table 2. Note that the abundance of vagile species (*) may be underestimated as these species tend to swim away when the tarp is being deployed.

Species	E, B	Meadow (N)	Meadow density (fish per m2)	Sand (N)	Sand density (fish per m2)
Muraenidae
Gymnothorax spp.		1	0.0833	0	0
Serranidae
Plectranthias nanus	B	0	0	1	0.2222
Pseudanthias bicolor		0	0	2	0.4444
Apogonidae
Apogonichthys perdix		9	0.7500	2	0.4444
Scorpaenidae
Iracundus signifer	B	0	0	2	0.4444
Scorpaenopsis fowleri	B	1	0.0833	2	0.4444
Unident		0	0	1	0.2222
Chaetodontidae
Chaetodon kleinii		0	0	1	0.2222
Pomacentridae
Chromis leucura		0	0	1	0.2222
Labridae
Oxycheilinus bimaculatus		4	0.3333	2	0.4444
Pseudocheilinus evanidus		3	0.2500	0	0
Pseudocheilinus octotaenia		0	0	1	0.2222
Pseudojuloides cerasinus		1	0.0833	0	0
Callionymidae
Synchiropus corallinus	B	0	0	1	0.2222
Pinguipedidae
Parapercis schauinslandii	B	0	0	6	1.3333
Soleidae
Aseraggodes borehami	E, B	0	0	1	0.2222
Median Abundance (Nx)		2.50		7.00
Median Species Richness (Sx)		2		6
Total Species Richness (S)		6		13
Diversity (H′)		1.4383		2.3667
Average Species Density m2			0.0990		0.3194
Median Density of Fishes m2			1.6667		4.6667
Average Density of Fishes m2			1.5833		5.1111

As with collections made in H. kanaloana meadows, fishes from Avrainvillea sp. collections were more abundant and densely-distributed within open sand (N = 7.00 individuals/collection, D = 4.67 fish per m2) when compared to meadow canopy (N = 2.50, D = 1.67; Chi-Square = 11.00, DF = 1, P = 0.001 for both tests). Median Species Richness was also significantly greater in open sand (S = 6.00 species/collection) when compared to meadow canopy (S = 2.00; Chi-Square = 11.00, DF = 1, P = 0.001). The Shannon Weiner Diversity Index was likewise higher in open sand (H′ = 2.3667) vs. meadow canopy (H′ = 1.4383).

Discussion

Deep-water Avrainvillea sp. and H. kanaloana meadows form complex three-dimensional habitats in an otherwise two-dimensional sandy environments. This structure provides habitat and shelter for numerous fish species. Of the two habitat types, H. kanaloana was most diverse supporting a total of 49 species from 25 fish families. In contrast, Avrainvillea sp. meadows contained 28 species from 19 families, though it is likely that the lower numbers for this species may be due to the smaller sample size for this species. Within-canopy diversity (H′) was nominally greater for H. kanaloana (2.52) than Avrainvillea sp. (1.87). In comparison, Friedlander & Parrish (1998) estimated that fish diversity on a shallow Kauai reef ranged between 1.72 and 2.54. Therefore, it would appear that the diversity of these deep-water macroalgal meadows is similar to that of shallow Hawai‘i reefs.

Wrasses (Family Labridae) were the most speciose taxon in both habitats (11 and six species, respectively), followed by gobies for H. kanaloana (six species). The wrasse Oxycheilinus bimaculatus and cardinalfish Apogonichthys perdix were the most frequently-occurring species within the H. kanaloana and Avrainvillea sp. canopies, respectively. These species were considerably less common and abundant in open sand, indicating a strong habitat preference for the meadow canopy. Other common species that showed strong associations with meadow canopy include the toby, Canthigaster jactator (H. kanaloana) and the wrasses Cymolutes lecluse and C. praetextatus (H. kanaloana) and Pseudojuloides cerasinus (both species). With the exception of A. perdix, which is nocturnal, we observed each of these species actively foraging within the meadow canopy on numerous occasions. Thus, meadow canopy appears to be an essential habitat for these species.

Most of the other common species were epibenthic sand-dwellers (e.g., gobies, dragonettes, sandperch, and flatfishes). Many of these species also occur in sandy areas near coral reefs (see Greenfield, 2003). In most cases, these species were actually more abundant in open sand rather than canopy (Tables 2 and 3), suggesting that their association with the meadows is incidental rather than a result of any inherent habitat preference.

Percent endemism

We estimate the percent endemism for fishes living in H. kanaloana and Avrainvillea sp. habitats (including adjacent sandy areas) to be 28.6% and 10.7%, respectively. Given that approximately 25% of Hawaii’s fish species are considered to be endemic (Randall, 2007), H. kanaloana meadows harbor a slightly greater proportion of endemic species than would be expected by chance. In comparison, approximately 46% of fishes surveyed within mesophotic depths in the NWHI are endemic (Kane, Kosaki & Wagner, 2014), and endemism may reach 100% in some areas (Kosaki et al., 2016). Thus, compared to mesophotic habitats in the NWHI, these macroalgal meadows have proportionally fewer endemic species. We speculate that the nominally higher endemism documented for H. kanaloana habitats could be a result of coevolution; H. kanaloana is native to Hawai‘i, with the largest meadows reported from the Maui-Nui complex (Spalding, 2012). It is possible that some of these endemic fish species may have evolved a commensal relationship with the alga. Although only five of the 14 endemic species regularly reside in H. kanaloana meadows, it is possible that some of the sand-dwelling endemics may occasionally dart into the canopy to avoid predators or may opportunistically exploit the new habitat created when blowouts (barren areas) are formed by the scour of winter swells. In contrast, Avrainvillea sp. is an invasive species which was first reported off Kahe Point, O‘ahu in 1981 (Brostoff, 1989). This species supports only three endemic species, with two (C. jordani and Bodianus albotaeniatus) recorded from canopy and one (Aseraggodes borehami) recorded from open sand. Given its supposed recent arrival, it is possible that endemic fish species have had less time to adapt to this unique habitat.

Abundance of Herbivores and Food fish

Contrary to our hypothesis that H. kanaloana and Avrainvillea sp. meadows may serve as foraging grounds for herbivorous fish species, we found that obligate herbivores were quite rare in both meadow types and were represented by only two species: the angelfish Centropyge fisheri (H. kanaloana) and the surgeonfish Acanthurus blochii (Avrainvillea sp.). Neither species was common or abundant, nor do they feed on macroalgae. Centropyge fisheri is an obligate herbivore (Thresher & Colin, 1986) that likely feeds on turf algae, whereas A. blochii typically feeds primarily on benthic algae on reefs or algal films covering sand (Randall, 2005). Surprisingly, parrotfishes (Family Scaridae) and Blennies (Family Bleniidae) were completely absent from the deepwater meadows, despite the fact that we have observed them in shallow-water (<1 m) Avrainvillea sp. meadows (R Langston, pers. obs., 2006). Six species (Rhinecanthus aculeatus, Aluterus scriptus, Ostracion meleagris, Arothron hispidus, Canthigaster jactator, and C. coronata) from H. kanaloana meadows and two from Avrainvillea sp. meadows (A. hispidus and C. coronata) are reported to be omnivorous, and occasionally consume algae. However, given their small size and limited abundance, it is unlikely that they consume significant amounts of macroalgal biomass. These results corroborate the work of Spalding (2012) who found little evidence of feeding scars on plants collected from deep-water H. kanaloana meadows. The absence of herbivores in these habitats may, in part, be due to their extreme depths. Several studies (e.g., Brokovich et al., 2010; Fukunaga et al., 2016; Larkum, Drew & Crossett, 1967; Thresher & Colin, 1986) report that herbivorous fish species are rare below 30 m. In addition, the lack of herbivory may be due to the low digestibility of the algae; both Avrainvillea sp. and Halimeda sp. contain numerous compounds which may deter herbivory (Meyer et al., 1994; Hay et al., 1990; Paul & Alstyne, 1992). In addition, the partial calcification of H. kanaloana may serve as an added impediment to herbivory (Schupp & Paul, 1994).

Commercially- or recreationally-important food-fish species were likewise rare in surveys of either meadow habitat. Three species each were recorded from H. kanaloana (Caranx melampygus, Aprion virescens, and Mulloidichthys sp.) and Avrainvillea sp. habitats (Caranx lugubris, Parupeneus multifasciatus, and A. virescens), however, each of these records was based on a single individual (Tables 2 and 3). In contrast, our surveys do indicate that deep-water Avrainvillea sp. meadows may be an important habitat for the flame wrasse, C. jordani, which is important in the Hawai‘i aquarium fish trade. According to Walsh et al. (2003), commercial fishers in the state reported catches of 13,919 C. jordani between the years 1976 and 2003. They estimated total wholesale value for the catch to be $133,116. Based on our review of two online fish sellers (liveaquaria.com and petsolutions.com- July 2016), C. jordani retails for $130 (juvenile female) to $300 (adult male) each. Assuming the extraordinary price of C. jordani is a result of high demand among aquarium enthusiasts, it is possible that aquarium fishers may eventually target deep-water Avrainvillea sp. meadows as a potential source for this species.

Our clove oil surveys detected significantly higher abundances and densities of epibenthic fishes in open sand when compared to meadow canopy for both H. kanaloana and Avrainvillea sp. Median species richness was also significantly higher in sand vs. meadow canopy for Avrainvillea sp. , and diversity (H′) was nominally higher in open sand for both species. These results are surprising given that several studies (e.g., Chittaro, 2004; Omena & Creed, 2004; Ornellas & Coutinho, 1998) have documented that diversity and abundance of fishes and invertebrates are usually highest within meadow canopy. Moreover, two recent surveys of the invertebrate fauna of H. kanaloana and Avrainvillea sp. meadows further support this hypothesis. Magalhaes & Bailey-Brock (2014) found that infaunal polychaetes were considerably more abundant and diverse in shallow Avrainvillea sp. meadows, when compared to adjacent sandy areas. Fukunaga (2008) likewise found that polychaetes were more abundant and diverse within H. kanaloana canopy. She also reported that epibenthic invertebrates were more diverse and speciose within meadows, but that abundance did not differ significantly between the two sub-habitats.

It is possible that the greater abundance of fishes in open sand may, in part, reflect a sampling bias. Fish inhabiting dense canopy are more likely to be overlooked by visual surveys when compared to individuals occurring on bare sand. Similarly, fish anesthetized in clove oil stations are more easily recovered in bare sediment than in dense canopy (this is particularly true for Avrainvillea sp. meadows given their dense, bladed morphology). Alternatively, it is possible that the greater abundance of some species on open sand may be related to differences in habitat preference or sediment composition between vegetated and non-vegetated areas. For example, within Halimeda kanaloana meadows, the Hawaiian shrimp goby, P. mainlandi was collected almost exclusively on bare sediment. This species lives in burrows constructed by the snapping shrimps, Alpheus rapax and A. rapicida (Randall, 2007). In a concomitant study of the epibenthic invertebrates from the same H. kanaloana meadows, Fukunaga (2008) recorded a total of 131 A. rapax in sand patches and only three individuals within the meadow canopy. Given this, it is not surprising that P. mainlandi was also rare within meadow canopy; without the presence of its invertebrate symbiont, the goby would have a difficult time finding shelter. We believe that the dearth of A. rapax burrows within meadow canopy may be related to sub-surface algal growth. Each H. kanaloana plant has a large, bulbous holdfast over 8 cm in length that penetrates deeply into the substrate, forming a network of stringy rhizoids that extend out into the surrounding sediment (Verbruggen et al., 2006). In addition, most H. kanaloana meadows contain several hundred plants per m−2 (Spalding, 2012), thus forming a dense concentration of rhizoids and holdfasts in the sediment that may be difficult for burrow-forming species, such as A. rapax, to penetrate. In contrast, Avrainvillea sp. has a dense and spongy holdfast that forms a terraced, penetrating mat over the sediment (Huisman, Abbott & Smith, 2007). These mats sequester fine sediments under their holdfast, forming anoxic mounds of sediment (Littler, Littler & Brooks, 2005). It is possible that these structures may likewise negatively impact benthic-dwelling and bioturbator speces. Other published studies support this hypothesis. For example, Milazzo et al. (2004) found that numbers of Bucchich’s goby, Gobius bucchichi increased significantly in plots where the algal biomass was experimentally reduced. This species feeds primarily on benthic mollusks and prefers open sandy habitats (Fasola et al., 1997). Thus, it is not surprising that its abundance would increase when more suitable habitat was made available through the algal removal experiments. Neira, Levin & Grosholz (2005) likewise documented a similar shift in the invertebrate macrofauna for tidal flats invaded by a hybrid cordgrass. They found that the densities of macroinvertebrates were 75% lower in the vegetated flats, when compared to un-vegetated areas, and that species richness was also lower in the vegetated areas. Although we did not experimentally manipulate the algal biomass in this study, it does appear that benthic-dwelling and bioturbator species are numerically more abundant in open sand. Within H. kanaloana habitats (Table 4), these species accounted for 86% of individuals collected within open sand vs. 61% collected meadow canopy. A similar relationship is evident for Avrainvillea sp. collections (Table 5); 57% of individuals collected from open sand were from benthic-dwelling or bioturbator species, whereas only five percent of those collected from canopy were classified as such. Thus, we suggest that the presence of surface- and subsurface algal growths may negatively impact the abundance and diversity of small epi-benthic fishes by reducing the amount of favorable habitat and feeding areas available to these species.

Advantage of clove oil collections

The use of clove oil anesthetic greatly enhanced our ability to estimate the fish diversity within H. kanaloana and Avrainvillea sp. meadows. Of the 65 unique species recorded in this study, 16 (25%) were detected only in clove oil stations alone. Thus, the use of clove oil anesthetic increased our overall estimate of species richness by 32.7%. It also enabled us to more accurately estimate the abundance and species richness of small-bodied fishes (gobies, scorpion fishes, cardinalfishes, dragonettes, and small wrasses), many of which are difficult to identify to species without the use of a dissecting scope. In some cases, species collected with clove oil but missed by visual surveys were also quite common. For instance, the cardinalfish Apogonichthys perdix was found to be the most frequently-occurring species in Avrainvillea sp. canopy (Table 3) though, due to its cryptic coloration and nocturnal nature, it was never observed in visual surveys. In a related study, Fukunaga (2008) measured the diversity and abundance of epibenthic invertebrates within H. kanaloana meadows using both visual surveys and clove-oil collections. She recorded 15 species of invertebrates in the visual surveys and 20 additional species in the clove-oil surveys. In this case, the use of clove oil anesthetic increased her ability to detect small epibenthic invertebrates by 133%. Together, these data highlight the utility of anesthetics (or ichthyocides) in estimating the diversity and abundance of small-bodied fishes and epibenthic invertebrates.

Data Set for Hawaiian Deepwater Macroalgal Fish Survey

Collections data for Hawaiian Deepwater Macroalgal Survey paper. Data includes collection location, date, depth, and habitat type. Fishes are identified to the lowest taxon possible. Survey types include visual surveys as well as tandem surveys consisting of visual surveys combined with collections using clove oil anesthetic to collect small epibenthic fishes.

Click here for additional data file.