Eric W. Vetter, Marine Sciences, Hawaii Pacific University, 45-045 Kamehameha Highway, Kaneohe, HI 96734, USA. E-mail: email@example.com
Submarine canyons are important sources of habitat heterogeneity on the slopes of continents and islands, but the study of canyon ecology has been largely restricted to continental margins. Here we use visual and video surveys from 36 submersible dives to evaluate the role of canyons as abundance and diversity hotspots for megafauna in the Hawaiian Archipelago, an island chain embedded in an oligotrophic ocean. We surveyed megafauna in canyon and slope settings at depths of 350–1500 m along the margins of four islands: the low ‘islands’ of Nihoa and Maro Reef, and the high islands of Oahu and Moloka’i. Megafaunal communities in canyons differed significantly from those in nearby slope habitats at all depths. Highly mobile fishes and invertebrates were consistently more abundant in canyons than on nearby slopes at the same depth off all islands, suggesting that canyons may be important sources of larvae for surrounding habitats. In the few cases where megafaunal abundances were similar or higher on the slope, the differences were typically driven by higher slope abundance of sessile suspension feeders or animals with limited mobility, i.e. by organisms which are likely to have difficulty with high currents and sediment transport in canyons. Megafaunal species richness and diversity generally trended higher within canyons, especially for the highly mobile taxa. Canyons contained 41 megafaunal species never observed on the slope, and increased estimated regional species richness by 25–30 species, indicating that canyons enhanced beta and gamma (regional) biodiversity. An expected trend of greater enhancement of diversity and abundance in canyons on the margins of high versus low oceanic islands was not observed, although megafauna were generally more abundant in both canyon and slope habitats on the high islands (Oahu and Moloka’i). We conclude that submarine canyons on both low and high islands in the Hawaiian Archipelago may provide keystone structures, enhancing megafaunal abundance, providing source populations for the open slope, and enhancing local and regional species diversity.
The role of habitat heterogeneity in supporting biodiversity is well established both theoretically (Tilman 1999) and empirically (Gilinsky 1984; Freemark & Merriam 1986). Habitat heterogeneity may enhance biodiversity by providing refugia from predation (Gilinsky 1984; Vetter 1998), enhanced or alternative food resources (Garrison 1991), spawning areas (Drazen et al. 2003), stress gradients (e.g.Levin 2003) and substratum diversity (van Rensburg et al. 2002). Increased habitat heterogeneity resulting from anthropogenic habitat fragmentation may decrease biodiversity on the landscape scale by reducing connectivity, population sizes and resource availability (Saunders et al. 1991), and by increasing vulnerability to predators and/or parasites (Robinson et al. 1995). Habitat fragmentation and resultant diversity declines are common in terrestrial ecosystems but appear thus far to be relatively rare in marine ecosystems (e.g.Polunin 2008); thus, habitat heterogeneity in the oceans is expected to enhance biodiversity on landscape scales.
The concept of ‘keystone structures’ has recently been introduced for terrestrial ecosystems to describe habitat features that provide essential shelter and resources for particular species or assemblages, in the process contributing fundamentally to habitat heterogeneity (Tews et al. 2004). For example, in the open ocean, seamounts may provide keystone structures by rising into productive surface waters, yielding ‘benthic islands’ of enhanced food availability (Haury et al. 2000) and habitat diversity (e.g. hard and soft substrates, zones of flow enhancement, deep coral beds; Levin et al. 1994; Fock et al. 2002) in an otherwise pelagic ‘landscape’. Along the slopes of continental margins and islands, submarine canyons are recurrent sources of habitat heterogeneity that may also serve as ‘keystone structures’.
Enhanced food availability may also allow canyons to play important roles in the feeding and reproduction of a broad range of benthic and demersal species. Because food-rich patches are critical for recruitment success in many fish and at least some invertebrate stocks, submarine canyons may provide important habitats for various life stages of benthic and demersal fishes and invertebrates along continental margins (Vetter & Dayton 1998, 1999; Company et al. 2008; Vetter, in preparation). Enhanced food availability in canyons may be especially important for allowing demersal fish and benthic invertebrates to reproduce in otherwise relatively oligotrophic regions, such as the margins of oceanic islands, including Hawaii (Yool et al. 2007). Canyons thus may harbor source populations in a ‘source-sink system’ in which dense, but localized, concentrations of breeding individuals broadcast larvae out to the surrounding slope, enhancing local and regional species diversity (Snelgrove & Smith 2002; Rex et al. 2005). Canyons could also facilitate speciation by providing distinct, relatively isolated habitats along more continuous slopes (Levin et al. 2001).
The role of canyons as biomass and diversity hotspots may be enhanced on the margins of oceanic islands embedded in low-productivity open-ocean ecosystems, such as in the Hawaiian Archipelago. Even though oceanic islands and associated canyons are common, the importance of canyons as hotspots of enhanced abundance and biodiversity on island slopes remains virtually unstudied. Here we investigate the ecological role of submarine canyons on oceanic islands by comparing canyon and slope megafaunal communities within the Main and Northwest Hawaiian Islands. Specifically, we test the following hypotheses:
1 Megafaunal abundance and local (alpha) diversity are enhanced in Hawaiian canyons compared to nearby slope habitats.
2 Such canyon enhancement is especially pronounced for the mobile megafauna due to their greater capacity to deal with the strong currents and sediment transport within canyons.
3 Canyon enhancement is greater on the margins of high versus low oceanic islands (because of greater export of terrestrial and nearshore production from high islands).
4 Oceanic canyons harbor faunal assemblages distinct from the open slope, enhancing regional (or gamma) biodiversity.
Material and Methods
We used the Hawaii Undersea Research Laboratory (HURL) Pisces IV and V submersibles to make 36 dives in submarine canyons and nearby slope regions within the Hawaiian Archipelago. Here we report megafaunal diversity and abundance data from video and visual transects on the islands of Oahu and Moloka'i in the main Hawaiian Islands, and on Nihoa Island and Maro Reef in the Northwest Hawaiian Islands (NWHI). The environmental settings for the canyons studied range from high islands with relatively large terrestrial and marine sources of organic matter (Oahu and Moloka’i) through a low dry island with limited capacity for terrestrial detrital input (Nihoa) to an extensive submerged atoll (Maro Reef), considered functionally to be similar to a small, low ‘island.’
The canyon system studied on Oahu (maximum island elevation of 1600 m) is located offshore of Kaneohe Bay on Oahu’s eastern shore (Fig. 1). Kaneohe Bay’s watershed is bordered by cliffs rising to 500–850 m and covers approximately 47 km2 (Smith et al. 1981) with annual rainfall of 100–150 cm (Chave 1973). The Kaneohe Canyon rises from a depth of 1900 m and bifurcates at about 1050 m, terminating to the North near Kaneohe Bay’s Ship Channel and to the south off the Sampan Channel. An extremely high northeast swell (∼7 m) during our Oahu cruise forced a relocation of the control (non-canyon) slope site from an area adjacent to the canyon to the leeward Oahu coast. Off Moloka’i, two canyon and slope sites were studied off the eastern end of the island’s north shore between Kalaupapa Peninsula and Papalaua Valley (Fig. 1). This coast is dominated by 600–800 m sea cliffs with valleys carved by annual precipitation that ranges from 200 to 400 cm (Culliney 2006). The terrestrial environment here resembles that of East Oahu in being a wet and lushly vegetated coast; however, in contrast to Kaneohe, there is no shallow embayment supporting high macroalgal production. Additionally, detritus originating from terrestrial and nearshore production is shared among the many canyons along this coast (Fig. 1).
Within the NWHI, we collected data from canyons and adjacent slopes on the south sides of Nihoa Island and Maro Reef (submerged at high tide) (Fig. 1). Both canyon systems originate on the edges of large carbonate platforms (or atolls) at depths of 200 m and descend to depths of ∼2000 m. Nihoa Island is the largest of the NWHIs, with an area of 70 hectares (∼3 orders of magnitude smaller than Oahu and Moloka’i). It rises to 273 m with steep terrain and is dry and sparsely vegetated relative to the Main Hawaiian Islands. Nihoa is surrounded by 575 km2 of coral-reef habitat on the remnant of an inactive volcanic cone (http://coris.noaa.gov/about/eco_essays/nwhi/nihoa.html). Maro Reef is an extensive coral reef ecosystem covering approximately 1,856 km2 with a small subaereal portion during low tide (http://ccmaserver.nos.noaa.gov/ecosystems/coralreef/nwhi/maro.html). Study of these canyon systems allowed us to explore the importance of island type (i.e. large high islands versus low ‘islands’) to the offshore submarine-canyon environment.
Data collection and analyses
Around Nihoa Island, the distribution of megafaunal fish and invertebrates was evaluated using videographic surveys employing the methods of Vetter & Dayton (1998). For video surveys, the submersible maintained an altitude of ∼1 m above the seafloor. Megafauna were then counted in non-overlapping frame grabs of the lower two-thirds of the video frame, with each frame grab covering ∼5 m2 of seafloor. Low megafaunal densities forced a change in protocol to the use of direct visual censusing of larger areas for the remaining three islands. The different sampling method used at Nihoa prevented us from including the data from Nihoa Island in some analyses because very different spatial scales and numbers of individuals were sampled. Visual transects were conducted on Maro Reef, Oahu and Moloka’i using a standardized protocol in which the pilot maintained a pre-determined heading at constant speed of 2 knots and elevation of 2 m above the bottom. The same light combinations were used for each transect and observers at the port and starboard windows counted into a voice recorder all animals visible in two non-overlapping fields on the port and starboard of the submersible; together, the two fields of view spanned a swath of about 15 m wide. When obstacles to navigation were encountered (such as canyon walls) transects were suspended until the standard survey protocols could be re-established. Only transects with a duration of 3 min or longer were included in the analyses. Measured visibility varied little between the canyon and slope transects at each island-habitat combination with the exception of low visibility within the canyon at 350 m off Oahu. Transects were conducted between 10 m above and below the target depths for 3–12 min with longer transects divided into replicate 3-min segments. Thus, a 3-min visual survey covering ∼2700 m2 was our basic unit of sampling. The canyon habitats generally had a limited width of flat bottom appropriate for transecting, so for most canyon-slope comparisons at a specific depth, more transects were completed on the slope. All sites were studied at depths of 350, 650 and 1000 m. We also studied 1500-m depths off Oahu and 1200-m depths off Nihoa. Still photography, video, and manipulator collections, as well as the HURL image archive, were used to confirm identifications of animals from visual and photographic surveys (cf.Smith 1985).
Because of greater sampling effort on the slope, species richness analyses were conducted by normalizing total species observed or numbers of unique species observed by transect duration (Oahu, Moloka’i, Maro Reef), number of video frames (Nihoa), or number of individuals (i.e. rarefaction). Percent effort (Table 2) is reported as the number of transect minutes (or video frames) in the canyon divided by the number of minutes from the corresponding slope, so that values less than 100% indicated greater effort on the slope. Relative richness (in Table 2) is the ratio of normalized richness from the canyon divided by that of the slope, resulting in values >100% when canyon normalized richness exceeded that of the slope. The same technique was used for reporting relative uniqueness. Rarefaction curves were plotted using Hurlbert’s (1971) modification of Sanders (1968) rarefaction for diversity comparisons at the landscape scale (e.g. comparing canyon versus slopes by island) to (i) control for the effects of differing sampling effort and faunal densities, and (ii) allow the Nihoa data, which sampled much smaller areas using video transects, to be compared with data from the other locations where visual transecting was used.
Table 2. Species richness and number of species observed only in either the slope or canyon habitat by island and depth.
Depth, depth in meters; Habitat, Canyon/Slope. Effort indicates the number of minutes of transecting or, for Nihoa, the number of video frames analyzed for canyon habitats relative to the corresponding slope. Effort <100% indicates greater sampling effort on the slope. Total species represents the species count for each island–depth–habitat combination with the left value from the canyon(s) and the right from the slope(s). Relative richness is the number of species observed in the canyon relative to the slope after normalization for effort (number of minutes or frames). Unique species are those observed only in the canyon or on the slope for each island–depth combination. Relative uniqueness is the ratio of unique canyon taxa relative to unique slope taxa after normalization for effort.
SPSS version 13 (Macintosh) was used for ANOVA and Mann–Whitney tests. Complex interactions involving both depth and habitat (canyon versus slope) precluded use of multifactor ANOVA for the examination of general trends. Statistical hypothesis testing was thus limited to comparisons of species richness, Shannon diversity, and abundance between canyon and slope habitats within a given island and depth. To maintain an experiment-wise alpha level of 0.05, Holm’s modification of the sequential Bonferroni corrections was used (with the Dunn–Sidak correction) (Holm 1979). When data violated the parametric assumptions of normality and/or homoscedasticity, natural log transformations were used. When such transformation failed to produce normal, homoscedastic distributions, the non-parametric Mann–Whitney test was used. The Mann–Whitney test assumes similar distributions but is robust to outliers which were resistant to correction by transformation in some of our data. PRIMER version 6 was used for diversity analyses with Shannon H’ to base e and rarefaction. Species accumulation curves were based on presence–absence data. Mean species accumulation curves (i.e. the average of an infinite number of random permutations of ordering) for successive pooling of habitat types (canyon or slope) across islands were calculated using the method of Ugland et al. (2003), and total species richness was estimated using the Chao 1 and Bootstrap indices as outlined in Magurran (2004) using PRIMER 6 software.
As expected, multifactor analysis revealed strong interactions between all combinations of the three factors examined: island, depth, and habitat (canyon versus slope). This precluded use of factorial ANOVA to examine general trends in megafaunal abundance and diversity. Nevertheless, simple pairwise comparisons (maintaining experiment-wise error rates at 0.05) resulted in clear patterns of megafaunal abundance and diversity in canyons compared to slope habitats at most island-depth combinations (Figs 2–5).
Patterns of abundance
The abundance of total megafauna was not predominantly greater in either canyons or on slopes, with three significant differences for particular island–depth combinations favoring canyons, and four favoring slopes (Fig. 2). However, the highly mobile megafauna (teleosts, chondrychthyes, decapods, and cephalopods) showed enhancement in canyons compared to sessile or moderately mobile organisms. For highly mobile megafauna, seven out of nine comparisons revealed significantly higher canyon abundance with the remaining two comparisons exhibiting non-significant differences (Fig. 2). For example, restricting analyses to highly mobile taxa off Oahu at 1000 m resulted in a change from higher slope abundance (P < 0.01, where the slope was dominated by sessile sea pens) to no difference, and at 1500 m a change from greater abundance on the slope (P < 0.01, octocorals dominant) to greater abundance in the canyon (P < 0.05). Similar shifts were observed off Moloka’i and Maro Reef. In general (6/7 cases), mobile megafauna were two to four times more abundant in canyons than at similar depths on slopes (Fig. 2). The emergent general pattern from these analyses is one of greater megafaunal abundance in the canyons except where suspension and/or deposit feeders abounded on the slope (Table 1, Video S1).
Table 1. Patterns in abundance, species richness and Shannon diversity between canyon and slope habitats by different taxonomic or mobility groupings.
Fish and crustaceans, which dominated the mobile megafauna category, typically were more abundant in canyons. Specifically, in five of the nine island–depth contrasts, fish were significantly more abundant in canyons, and crustacea were significantly more abundant in canyons in six of nine comparisons (they were never significantly more abundant on slopes; Table 1). Cephalopods were observed more often in canyons off Oahu and Maro Reef and were equally abundant in both habitats off Moloka’i. When considered alone, suspension feeders were significantly more abundant on the slope in four of the nine pairwise comparisons and more abundant in canyons at three of the island–depth combinations (Table 1). The data from Nihoa Island collected from video could not be compared statistically with the other three locations (see Material and Methods) and lacked sufficient numbers of transects to provide reasonable statistical power for standard comparisons; however, when normalized to numbers of animals seen per 100 frames, there were more than twice as many animals observed in canyons as on the slope (Fig. 5).
The data from both the Main and Northwest Hawaiian Islands (Figs 2 and 5) strongly support our Hypothesis 2 that the abundance of mobile megafauna is especially enhanced in canyons relative to slope habitats. However, total megafaunal abundance was not generally enhanced in canyons, disproving the abundance component of Hypothesis 1. When higher overall megafaunal abundance occurred on slopes, this difference was driven by greater numbers of sessile suspension feeders or echinoderms on the slope (Table 1). The prediction of greater megafaunal enhancement in canyons on high islands versus low islands (Hypothesis 3) was not supported, with both Maro Reef and Nihoa (low-relief settings) yielding relatively higher megafaunal abundance in canyons (Figs 2 and 5). We caution that the differences in transecting methods prevents direct comparisons (e.g. in absolute abundance) between Nihoa and the other islands, but it is clear that megafaunal abundance patterns on Nihoa show greater relative enhancement of the canyon megafauna than on Oahu or Moloka’i.
Patterns of diversity
The hypothesis that local (or alpha) diversity would be greater in canyons than on the slope was evaluated using comparisons of species richness per normalized area (species counts per transect or frame) and the Shannon Index (H′). Higher megafaunal species richness was observed in the canyons in seven of the nine contrasts, with five of these differences being statistically significant (Fig. 3). Species richness on the slope significantly exceeded that in the canyon only at 350 m off Moloka’i (Fig. 3), largely due to greater numbers of suspension-feeding species (Table 1). A similar pattern was seen using Shannon diversity, with the diversity index higher in the canyons in six of nine contrasts (Fig. 4). Two of those differences were non-significant, but along with the absence of significantly greater diversity at any of the slope sites, they contribute to an overall pattern of greater diversity in canyons. When only highly mobile megafauna were considered, there were no island-depth combinations for which species richness or Shannon diversity were significantly higher on the slope (Figs 3 and 4). The only contrast in which the slope species richness non-significantly exceeded that in the canyon was off Oahu at 1000 m (Fig. 3, Table 2). Other changes in outcomes when restricting analyses to highly mobile taxa included the greater richness and Shannon diversity observed in the canyons off Maro Reef at 1000 m becoming statistically significant, although off Moloka’i formerly highly significant differences favoring the canyons at 650 and 1000 m became non-significant (Fig. 3). In conclusion, data from the Main Hawaiian Islands and Maro Reef generally showed higher species richness and Shannon diversity in canyons for total and mobile megafauna; no general canyon versus slope pattern was evident for suspension feeders when examined alone (Table 1).
Off Nihoa island, species richness was greater in the canyon at 350 m and 650 m, but greater on the slope at 1000 m. After normalizing for sampling effort, species richness in the canyon exceeded that of the slope for all depths (Table 2). When only mobile megafauna were considered, more species were observed in the Nihoa canyon at all depths (Fig. 5).
Species richness and the number of unique species at each island–depth combination are presented in Table 2. Sampling effort in canyons exceeded that on slopes by 10% or more only twice, whereas effort on the slope exceeded that in canyons in nine canyon–slope contrasts. This was a result of the limited extent of the canyon habitat for a given island and depth. The total number of species observed in the canyons was greater in all three instances where canyon effort exceeded that on the slope. The same was true for five of the nine comparisons when effort on the slope greatly exceeded that in the canyon. When corrected for effort, the relative species richness in the canyons greatly exceeded that on the slope in nine of 12 comparisons and was essentially equal in the remaining three. When the analysis is limited to species unique to canyons or to the slope, more unique species were observed in the canyon in nine of 12 contrasts without correcting for effort. After correcting for effort, relative uniqueness was greater in the canyons for each contrast and much greater (38% or more) in all but one comparison (Table 2).
While striking, the greater numbers of unique species in canyons could be driven largely by rare taxa. Examination of the dominant taxa for each island–depth comparison revealed that many of the dominants in one habitat are absent or rare in the other (Table 3). For example, off Oahu at 350 m, there were no shared species among the five dominant taxa observed in both habitats. Similarly, at 350 m off Moloka’i, the second through fifth most dominant taxa diverged strongly between habitats. In a comparison of the five most abundant species at 650 m, two taxa were shared off Oahu, three off Moloka’i, and one off Maro Reef and Nihoa (Table 3). More dominant taxa were shared between the slope and canyon at 1000 m, but this may have been partially due to lumping several species together as shrimp, sea pens, or anemones. The only comparison at 1500 m (Oahu) revealed that only one of the five most abundant taxa was shared between habitats. These data demonstrate that the canyon and slope habitats diverge strongly both in the species assemblages they support and in the dominant species. Taken together these data strongly support Hypothesis 4, i.e. that oceanic canyons harbor faunal assemblages distinct from the open slope, increasing both the variety of habitat types (beta diversity) and regional (gamma) diversity (Ricklefs & Miller 1999).
Table 3. Dominant megafaunal taxa (% of abundance) in canyons and slopes at our study sites. See notes below table.
Analyses of species accumulation and richness at the full canyon or slope scale with all depths combined (i.e. at the ‘landscape’ scale) also indicated higher species richness, and distinct faunal assemblages, in canyons compared to the open slope. The mean species accumulation curve for all taxa in canyons remained substantially (∼20 species) above that for the slope sites (Fig. 6A). Both curves very nearly reach an asymptote, suggesting that both canyon and slope systems were well sampled (Fig. 6A). Estimated total species richness was also greater for canyons by 10–15 species, depending on the index used (Fig. 6B). Finally, species accumulation curves considering only species unique to canyons or to slopes revealed considerably more species unique to canyons than to slopes (Fig. 6C). Forty-one species were only observed in canyons, suggesting that canyon and slope species lists show substantial non overlap, and that canyons are contributing substantially to beta and gamma (regional) diversity. The enhancement of regional diversity resulting from the presence of canyons is indicated by comparing species accumulation curves (Fig. 6A) and species richness estimates (Fig. 6B) for combined canyon plus slope data with those for the slope alone. Both approaches suggest that the presence of canyon habitats in the Hawaiian Archipelago increases the regional megafaunal species pool by 25–37 species above the level supported by slope habitats alone.
Rarefaction curves plotted at the landscape scale (i.e. by island and habitat with all depths pooled) provided additional insights into the influence of canyons on megafaunal species diversity. The initial portions of canyon curves (i.e. at levels below 50 individuals) fell below slope curves, reflecting greater species dominance in canyon habitats at all islands but Moloka’i, where stronger dominance was observed on the slope (Fig. 7). By levels of 200–700 individuals, however, canyon curves had crossed the slope curves, indicating higher species richness in canyon systems in agreement with the Chao 1 and Bootstrap species-richness estimators (Fig. 6B). When canyon and slope data were pooled for individual island margins, the pooled curve invariably fell above the slope curve for each island, indicating that the presence of the canyon habitat enhanced regional megafaunal diversity (Fig. 7).
In summary, analyses of species biodiversity at a variety of scales suggest that local megafaunal diversity is enhanced in canyons compared to nearby slope habitats, and that the canyons support faunal assemblages distinct from the open slope, enhancing beta and gamma (regional) biodiversity. These results again strongly support our Hypothesis 4.
Contrary to our Hypothesis 1, we found no general pattern in megafaunal abundance between the canyon and slope habitats when all taxa were considered; however, the greater abundance of mobile megafauna in canyons strongly supported our second hypothesis that animals capable of dealing with physical disturbance would benefit most from canyon conditions (e.g. higher food availability). Evidence for disturbance in canyons observed in this study included large ripple marks in canyon sediments, coarser sediments, large patches of terrestrial detritus deep in canyons off Moloka’i (Videos S2 and S3), large numbers of algal aggregates moving down-canyon off Oahu (observed from 350 to 1500 m (Videos S4 and S5)), and accumulations of green algae at 350 m in the canyon off Maro Reef (F. De Leo, E. Vetter and C. Smith unpublished data). The enhancement of mobile megafauna, especially fish, in Hawaiian canyons is similar to the findings of enhanced mobile megafauna in other canyon systems, including numerous canyons on the California margin (Vetter & Dayton 1999; E. Vetter unpublished data), in the Western Mediterranean (Stefanescu et al.1994), and in Pribilof Canyon, Alaska (Brodeur 2001).
Elevated abundance of scavenging fish and crustaceans in canyons in the Main and Northwest Hawaiian Islands suggests that those habitats experience greater food availability, as has been observed in many canyons on continental margins (Greene et al. 1988; Vetter & Dayton 1999; Hooker et al. 2002; Cotté & Simard 2005). Because food-rich patches are critical for recruitment success in many fish stocks, submarine canyons may provide important habitats for food-limited life stages of benthic and demersal fishes (Stefanescu et al. 1994; Vetter & Dayton 1998, 1999; Yoklavich et al. 2000; Brodeur 2001; Vetter in preparation). Along the California and Mediterranean coasts, submarine canyons are regularly targeted by commercial and recreational fisherman exploiting rockfish, rattails, other bottom fishes and invertebrates (C. Smith and E. Vetter unpublished observations; Company et al. 2008). The enhanced abundance of fish in Hawaiian canyons documented in this study, together with observations of fishing effort targeting Moloka’i canyons (C. Kelly personal communication), suggests that canyons may provide critical habitat for bottom fishes in the Hawaiian Archipelago as well (e.g. lutjanid snappers and synaphobranchid eels).
The majority of fishes and crustaceans observed in deep-sea canyon and slope habitats are considered to be scavengers that may benefit from funneling of nekton carcasses (food-falls) in canyons. For example, in a baited camera study, King et al. (2008) reported similar numbers of scavengers in the Nazaré canyon off Portugal as on the Porcupine Abyssal plain, which is found beneath much more productive waters. They found that scavengers arrived later and remained longer near bait at a single camera deployment at the abyssal plain (4437 m) than in the canyon and cautiously suggested that this could indicate a greater food supply in the canyons. De Leo, Smith, and Vetter (in preparation) often caught greater numbers and biomass of scavengers (mostly Heterocarpus shrimp and lysianassid amphipods) in traps deployed in canyons at the Hawaiian canyons studied here. There is thus a recurrent trend of greater abundance of mobile megafauna in canyons than on nearby slopes of equivalent depths. Increased food availability and/or habitat diversity are the likely causes of increased numbers of mobile animals in canyons; however, many factors such location, distance from shore, orientation to currents, overlying production regime, etc., can result in differences between and within canyons. One goal of continued research in submarine canyons should be to determine which characteristics of canyons, such as focusing of detritus originating in shallow water, increased current flows, and concentration of vertically migrating organisms (Video S6), are particularly important for enhancement of megafaunal abundance. Ultimately, a classification system integrating geological, biological oceanographic, and hydrographic conditions in canyons could prove useful to predicting patterns of megafaunal enhancement in submarine canyon.
Our hypothesis that megafaunal diversity is enhanced in Hawaiian canyons compared to nearby slope habitats was supported by a strong trend of elevated species richness and Shannon diversity in canyons. These greater values could result from far-ranging organisms remaining longer in canyons because of more food or desirable habitat features; larger population sizes resulting in a greater likelihood of inclusion in surveys; more canyon specialists; and the use of canyons as nurseries or spawning grounds. Canyon assemblages were observed to be more uniformly speciose, whereas the slopes frequently had large tracts with few species or individuals interrupted by patches of plenty. At some slope sites, it was not unusual to travel 100 m or more observing little other than sediment and then to encounter rocky outcrops with high abundance and species richness (e.g. an outcrop with fish, crabs, and shrimp which was encountered after traveling >500 m without seeing any megafauna; Video S7). These rocky habitats appear comparable to other habitat oases such as isolated trees in the African savanna (Milton & Dean 1995). Hecker (1994) attributed pronounced patchiness of benthos on the slope of Cape Hatteras to a combination of habitat heterogeneity, enhanced nutrient availability, and a fauna largely composed of sedentary organisms. We found increased abundance and diversity of highly mobile animals in canyons where hard and soft substrate were both virtually always visible from the submersibles, and greater patchiness in animal distributions on the slope where small islands of hard substrate were often widely separated in a ‘sea’ of sediments. These results are similar to those of Hargrave et al. (2004) where current speed and substrate composition were found to be major determinants of biodiversity and the dominant trophic modes observed in the Gully submarine canyon in the NW Atlantic.
We observed aggregations of suspension feeders (mostly cnidarians and ophiuroids) in both canyon and slope habitats; however, large and dense aggregations of sessile organisms in sedimentary environments were only seen on the slope (especially sea-pens, Pennatulidae). Lower abundance and/or diversity of sedentary and sessile organisms in some canyons or portions of canyon systems may be due to periodic disturbances such as sediment slumps and turbidity currents, which can be relatively frequent in canyons (Shepard & Marshall 1973; Inman et al. 1976; Gardner 1989). While we did not directly observe major disturbance events in the Hawaiian canyon, these events tend to be highly episodic in nature (Shepard & Marshall 1973; Inman et al. 1976; Gardner 1989); we did see evidence of sediment scour and accumulations of terrestrial and shallow water detritus in many parts of the Hawaiian canyons, suggesting periodic down-canyon transport and associated physical disturbance. The large numbers of megafaunal fishes and crustaceans we observed in canyons should be able to avoid or overcome canyon-associated slumps and currents, allowing them to benefit from organic material transported from shallower depths with these events as in Mediterranean canyons (Company et al. 2008).
The hypothesis that oceanic canyons function as hotspots of biodiversity by harboring faunal assemblages distinct from the slope was supported by the long list of species observed only in canyons (Table 4). This together with (i) the smaller list of species only observed on the slope and (ii) the enhancement of species richness and rarefaction diversity when canyon and slope habitats were pooled (Figs 6 and 7) demonstrates that submarine canyons on islands in oligotrophic oceans may contribute substantially to both beta and gamma diversity. The higher abundance and species richness of fish and crustaceans often found in the Hawaiian canyons suggest that canyons may also harbor larval-source populations and provide a critical habitat for a variety of mobile species. Canyons are thus likely to provide keystone structures resulting in hotspots of biodiversity, meriting special attention in coastal management and environmental protection from human impacts resulting from bottom trawling, waste disposal, and dredge or mine spoil dumping on island margins (Smith et al. 2008).
Table 4. List of shared and unique taxa combining all submarine canyons and slopes studied off Main and Northwest Hawaiian Islands during this study.
We thank R. Martini, J. Drazen, C. Berini, A. Halberg, N. Rothe, B. Kivi, D. Vardeh, T. Kirby, and the Pisces IV and V crew of the Hawaii Undersea Research Lab for expert operations of the Pisces IV and V submersibles within the challenging environment of the submarine canyons and the captain and crew of the R/V Ka’imikai-o-Kanaloa. The suggestions of two anonymous reviewers substantially improved the final manuscript. This work was supported by grants from NOAA Ocean Exploration Office Grant # NA03OAR4600109 and by the Hawaii Undersea Research Laboratory. We also thank Capes-Fulbright for a fellowship to FDL. This is contribution no. 7858 from the School of Ocean and Earth Science and Technology, University of Hawaii at Manoa.