CRAMP Rapid Assessment. Coral Results
Coral Community Structure
The average coral cover in the Main Hawaiian Islands is 21.7 ± 1.6% (±SE
n=152). The most dominant species are shown in descending order in Table 1. A total of 21 species of corals were recorded from transects statewide.
The six most abundant species were used in a Correspondence Analysis (CA) to determine coral community structure (Figure 1). Three main gradients are apparent. Sites dominated by Porites compressa group in the lower right quadrant of Figure 1, while those dominated by Pocillopora meandrina cluster towards the lower left of the ordination. Those communities with a high percentage of Montipora flabellata cluster away from other sites, while the sites dominated by the species, Porites lobata, congregate in the center of the linear cluster.
Figure 1: Multivariate Correspondence Analysis of coral community structure
To link dominant coral species to multivariate environmental patterns, a biota and environmental matching procedure (BIOENV) in PRIMER was applied. This procedure uses rank correlations to produce the optimal combination of environmental factors that best explains the variation in the biological data. The environmental variables included: latitude, age of islands, precipitation, distance from stream, watershed size, wave height maximum, wave direction, depth, rugosity, population within 5km, sand, organic matter, CaCO3 and select grain-sizes. The coral community among sites was best explained by the environmental variables silt/clay, latitude, rugosity, maximum wave height and wave direction. These five variables combined to produce the highest matching coefficient (0.34) in a range of 0-1.
A multiple regression was used to determine the best model for predicting coral cover regardless of depth class. The regression model using coral cover as the response variable was significant among the stations (R2 (adj.) =49.1%, p<0.001). The variation in coral cover is best explained by rugosity (t= 8.4, p<0.001), population within 5 km (t=-3.4, p=0.001), depth (t=3.0, p=0.003), distance from a perennial stream (t=-2.8, p=0.006), wave direction (t=-2.7, p=0.009) and maximum wave height (t=-2.3, p=0.023).
A general linear multiple regression model was developed for sites with a depth <5 m. The variables used in this model include: rugosity, biomass, richness, wave max, organics and total fish. Variables used in explaining differences in coral cover at deep sites >5 m included: sand, human population, rugosity, fish biomass, latitude and coral diversity. Wave energy is only important in shallow water (p<0.001). This is correlated with a statistically significant increase in coral cover with depth (p=0.004). Sites<10 m in depth have an average total coral cover of 17.4% ±15.3 (SD), while deeper sites (>10 m) average 27.8% ±24.1 (SD). Species with the strongest skeletal strengths, Montipora flabellata, Pocillopora meandrina, and Porites lobata (Rodgers et al. 2003) have higher mean cover in shallower waters (Figure 2).
Figure 2: Percent cover of dominant Hawaiian coral species at shallow (<10 m) and deep (>10 m) depths ±sd.
The regression model using coral richness as the response variable included the environmental variables: rugosity, depth, organics, distance from a stream, legal protection status, maximum wave height and direction and population within 5 km. This model is statistically significant (R2 (adj.)=23.5%, p<0.001). The variation in coral richness is best explained by organics (t=-4.6, p<0.001), wave direction (t=-3.9, p=0.01), population within 5 km (t=-3.8, p<0.001), distance from a stream (t=-2.8, p=0.006) and maximum wave height (t=-2.3, p=0.025).
Indirect Measures of Coral Cover
Bivariate linear regression was used to predict coral cover based on Chaetodon abundance. The test was statistically significant (p<0.001) with an r2 of 16.3%. A simple linear regression of only corallivorous Chaetodons had a slightly stronger correlation than when using all species of butterflyfishes in the model (r2=19.9%, p<0.001). Of the 152 stations sampled, 60% (91 stations) had no corallivorous Chaetodons present. The other stations ranged from 1 to 5 fishes. Other statistically significant (a=0.05) simple indicators of coral cover include rugosity (r2=35.5%), fish biomass (r2=12.3%), total number of fishes (r2=10.1%) and depth (r2=8.7%).
To determine species tolerances to sedimentation and wave regimes, relative
percent coral cover of each taxon was related with the mean percent of
silt/clay and the maximum wave height at each station (Table 2). This
allowed examination of the width of the environmental gradient in relation
to the niche of the coral species. The relationship may also be due to other
confounding variables not examined. Thresholds were extremely low for
Montipora flabellata (2%) and Pocillopora meandrina (9%) (Figure
Figure 3: Pocillopora meandrina (relative cover %) vs. silt/clay (%).
Porites compressa (55%) (Figure 4) and Montipora capitata (62%) inhabit areas with high levels of fine material (<63 um; silt/clay).
Figure 4: Porites compressa (relative cover %) vs. percent silt/clay (%).
Porites compressa (Figure 5) and Montipora capitata are not found at the maximum wave height of 12 m (40 ft) while all other coral species can tolerate this wave regime (Figure 6, Table 2).
Figure 5: Porites compressa (relative cover %) vs. wave height maximum (m).
Figure 6: Montipora patula (relative cover %) vs. maximum wave height (m).
Kruskal-Wallis tests were used to determine whether statistically significant differences occurred between the Windward and the Leeward sides of the islands. Fig. 7 shows which sites are affected by these Windward/Leeward differences.
Figure 7: Map of Main Hawaiian Islands with 55 sites showing general wave direction and Windward/Leeward influences.
Statistically higher values on Leeward sides include carbonates, fine
grain-sizes, Porites lobata, and watershed area. Variables that were
significantly higher on Windward sides include organics, large grain-size,
Montipora flabellata, and small fishes (Table 3). Higher rain and
wave regimes are well documented on Windward sides of islands.
Multiple regression analysis using coral cover as the response is not significant for wave direction and maximum wave height when only sheltered sites are included in the model. When examining Windward sites alone using coral cover as the response, maximum wave height (p=0.03) and direction (p=0.01) are significant (R2 (adj.) =25.1%), although Leeward sites are not significant (R2 (adj.) =10.8%) for wave parameters. The model used the same variables determined to be most suitable in regression analysis for coral cover.
The species richness model explained over 98% of the variability at Leeward sites where wave parameters were significant (R2 (adj.) =97.9%). The regression model included the environmental variables: rugosity, depth, organics, distance from a stream, legal protection status, maximum wave height and direction and human population within 5 km.
Coral Community Structure
Historically, surveys have been conducted at specific sites on small spatial scales to answer specific questions. These surveys are usually conducted in areas with relatively large coral populations, leading to high estimates of coral cover. This statewide survey provides a more accurate representation of the mean total coral cover (22%) and species abundances on a large scale (Table 1). Hawaiian reefs are predominately Porites reefs, with P. compressa and P. lobata comprising nearly half of the total coral cover in the MHI. Of the 42 species documented from the State of Hawai’i, half were recorded on the transects (21). Species not documented were most likely due to small colony size, difficult field identification, NWHI endemism, species occurring at greater depths, photographic resolution or cryptic species.
Coral communities correspond to wave disturbance and population. A spatial gradient of coral communities is evident from the multivariate analysis. Reefs in Hawai’i have been characterized by a single species of coral that can dominate certain sites. Three species of coral were shown to be influential. The species abundance gradients accompany wave energy patterns. Montipora flabellata are found in highest abundance on north facing, windward shores of islands. Pocillopora meandrina can be indicative of high wave energy environments, while Porites compressa most often inhabits calmer waters. The multivariate environmental patterns most closely linked to coral species are sedimentation, latitude, rugosity, maximum wave height, and wave direction.
Coral Cover and Species Richness
Rugosity and depth have a positive correlation with both coral cover and species richness while maximum wave height, wave direction, population within 5 km and distance from a stream have an inverse relationship. Organics are also correlated with species richness. In all statistical analyses, wave regimes were found to strongly influence coral communities.
From regression analysis, wave energy is statistically significant in shallow waters but less important in explaining coral cover at deeper sites where lower wave energy exists. Coral cover (<10 m, 17.4%±15.3%, >10 m, 27.8%±24.1%) is also stratified by depth.
Prior research has demonstrated depth stratification of coral assemblage characteristics (Dollar 1982). The validity of this model in verifying these established processes provides support of the power to detect true trends and patterns. By substantiating this stratification, it lends credence to other significant correlations, while establishing these relationships on a statewide scale.
Wave energy has been reported to be the primary forcing function in determining coral reef communities (Dollar 1982; Grigg 1983). Coral cover is higher in deeper waters, reflecting lower wave disturbance. The significance of depth in explaining coral cover is analogous to stratification of vegetation by elevation, the most pronounced environmental gradient in terrestrial ecology. The rise in coral cover with increasing depth is partially a function of decreasing wave energy. Research conducted in the eastern Pacific (Glynn 1976) suggests that physical factors control shallow environments, while biological factors are the forcing function in deeper waters.
Coral zonation patterns also reflect their morphology and skeletal strength. This distribution may have evolved as an adaptive response of coral species to disturbance by waves (Rodgers 2001). Species with highly branched morphology, low skeletal strength and high fracture rates, reside in regions with little wave exposure, such as in bays and near sheltered shorelines. Species with lobate or encrusting forms tend to inhabit regions with high wave energy.
Sediment associated organics and fine particles of silt and clay are also correlated with coral species abundance. This is also partially structured by wave energy. Winnowing of fine grain particles in high energy regimes selects for larger, coarser grain sizes while smaller organic particles can remain in areas with little wave disturbance (Te 2001). High organics in sediments may result from human impacts in terrestrial environments. Where fine sediment overwhelms the system, sedimentation rather than wave energy, becomes the dominant forcing function on community structure.
Mean wave direction is also important in explaining coral abundance and species richness. This is directly related to the maximum wave height in Hawai’i. Distinct and consistent directional wave patterns prevail throughout the year (Fig. 7). A storm surf gradient exists along Hawaiian shorelines, increasing in a clockwise direction. Larger winter swells arrive on the north shores of the islands, originating from the North Pacific Swell, while less exposed south shores receive lower energy from South and Trade Wind Swells (Juvik and Juvik 1998). These long period swells are influential in biostratification of species, spatial heterogeneity and structuring of coral reef communities. Anomalous changes in wave direction can significantly impact coral communities.
Although wave energy plays a dominant role in the structuring of coral reefs, other factors also explain the variation in coral communities, particularly in sheltered bays, harbors and shorelines and in deeper waters. Coral reefs are complex, interrelated systems influenced by numerous physical, biological and chemical factors that continually interact.
Rugosity explained a large percentage of the variation in coral cover (r2=38.2%), as well as in fish abundance (r2=24.7%). Areas indicative of high rugosity (>1.5) provide stable attachment sites for coral recruits, thus increasing vertical relief. In comparison, unstable habitats consisting of sand, rubble or silt have relatively low spatial complexity (Birkeland et al. 1981).
Human factors can also be important in the structuring of coral reef communities. Impacts affecting reefs such as sedimentation, eutrophication, introduced species, overfishing and coastal development are usually a direct result of increased human population. Regression analysis indicated that sites in close proximity to high human population and perennial streams had lower coral cover and species richness. Although technological advances in transportation and close geographic proximity of the MHI allows access to most areas, higher activity is found closer to population centers. A large percent of Hawai‘i’s reefs are easily accessible to the human population, located within close proximity of major urban centers of resident and tourist concentration.
Streams in Hawai’i have a history of alteration and diversion. Water quality reflects the resident population and adjacent watershed uses. Physically tied to the ocean, streams affect the marine ecosystem. Of the 366 perennial streams in the state of Hawai‘i, 55 had been significantly altered by 1978 through channel realignment, lining or filling of channels, clearing of riparian vegetation or elevation or extension of the culvert or revetment (Timbol et al. 1978). Modifications have been made to over 150 km of stream channels. Lined channels are the most common type of modification, comprising over 40% of stream channel alterations, with over 90% of these located on the island of O‘ahu (Timbol et al. 1978). Water has been diverted from over half of all perennial streams for irrigation and other uses in drier Leeward areas.
By 1978, only 51 of these 366 streams were considered “physically pristine,” none of which occurred on O‘ahu. Only 95 streams were considered of “high ecological quality” and therefore designated for pristine-preservation use, including streams from all islands, with the exception of O‘ahu (Timbol et al. 1978). No “biologically pristine” streams were reported (Timbol et al. 1978). Every perennial stream sampled on every island had at least one introduced species. Few intact streams remain today and the resultant impact to the nearshore biota has undoubtedly been significant.
Indirect Measures of Coral Cover
Factors that were found to be significantly correlated with coral abundance in multiple regression analyses were investigated to determine whether any single indirect measure could be substituted as a proxy for coral cover.
Reese (1981) supports the monitoring of abundance and territory size of obligate, corallivorous butterflyfishes to monitor the “health” of coral reefs. In support of this hypothesis, a statistically significant correlation was found in this study (r2=19.9%). A stronger correlation is probable with increased transect length. Few butterflyfishes were recorded on each transect (1-5) and were present on only 40% of the transects. This absence is most likely explained by the survey method selected. Designed as a rapid assessment to allow greater spatial coverage, it limits the sample size and accurate representation of observations. One 25 m belt transect cannot encompass the variability in fish populations at a station. Many butterflyfishes have large home ranges and may not be encountered using this abbreviated method. Although butterflyfishes were absent from the majority of stations, the regression analysis showed a statistically significant, positive correlation between corallivorous Chaetodons and total coral cover, explaining approximately 20% of the variation.
A weaker correlation between all fish species and coral cover was also found to be statistically significant (abundance r2=10.1%, biomass r2=12.3%). Corals provide food, shelter and protection for fishes by increasing vertical relief. Friedlander et al. (2003) found a strong correlation between habitat complexity and fish communities.
Although rugosity, depth, fish abundance and biomass have a statistically significant relationship with coral cover, no one factor can substitute as a proxy. Substitutions are recommended only with coefficients of determination >95% (Clarke and Warwick 2001). The structuring of coral reefs involves complex interactions; therefore each factor alone is a weak predictor of coral cover, explaining only a portion of its variability.
Silt thresholds for Montipora flabellata and Pocillopora meandrina were found to be very low. The occurrence of these species at low silt levels may be a function of its distribution rather than its tolerance level, although it strongly suggests that these species do not typically occur in environments where silt dominates. These two species do occur in high to moderate wave energy environments where finer particles are winnowed out. For example, Pocillopora meandrina is found in shallow waters with strong currents where fine sediment can also be swept away. Both M. flabellata and P. meandrina were found to tolerate maximum wave heights of 12 m.
Montipora capitata and Porites compressa are found in areas with high levels of silt suggesting a high tolerance to sedimentation. Field studies on the Great Barrier Reef found corals surviving in extremely turbid zones with sediment input levels at approximately 140 mg/l (Woolfe and Larcombe 1998). Morphological plasticity may partially explain the high tolerance to silt. Foliose or lobate forms foster sediment accumulation while branching, vertical morphologies are less prone to sediment retention. The plate form of M. capitata is an example of morphological change in response to wave action that occurs in high energy environments, while the branching form of this species occurs more commonly in low wave energy environments. Encrusting or lobate forms of corals dominate high wave energy environments while branching, more delicate forms are correlated with low wave energy environment. These species were absent from areas with maximum wave heights of 12 m.
Porites lobata, detritivores and very fine sands occurred more frequently on the Leeward sides of the islands, while Montipora flabellata, maximum wave height, precipitation, small fishes and large grain sediments were more prevalent on the Windward sides of islands.
Higher precipitation on Windward sides of high islands has been well documented. Having verified this established relationship with this dataset provides stronger evidence for other documented correlations.
The Leeward sides of the islands have statistically higher cover of Porites lobata (37.6% vs. 25.3% p=0.03) while the Windward sides have higher total cover of Montipora flabellata (7.6% vs. 0.1% p=<0.001). No other coral species distribution differed significantly between sides of islands. Since the Windward sides of the islands have higher mean wave heights than the Leeward sides, encrusting corals such as M. flabellata occur more frequently. Windward and Leeward sites have significantly different wave regimes that can affect many factors of biotic and abiotic communities. Sorting of sediment grain sizes occur as a result of wave energy that increases erosion. Higher wave energy on north-facing shores reduces fine particles. Consequently, coarser sediments remain on Windward sides, while Leeward shores have significantly higher percentages of finer particles. Fishes in the smallest size class (<5 cm) are found in statistically higher abundances on the Windward sides (Figure 8).
Figure 8: Differences in fish size classes between Windward and Leeward sides of islands (±sd).
Even with the Kāne’ohe Bay sites removed, due to large numbers of small scarids, the positive correlation is still statistically significant. This demonstrates how a large sample size can overshadow the effects of anomalies to elucidate relationships and trends.
Birkeland, C., Rowley, D, Randall, R.H. 1981. Coral recruitment patterns at Guam. Proceedings of the Fourth International Coral Reef Symposium 2: 339-344.
Clarke, K.R., and Warwick, R.M. 2001. Change in marine communities: an approach to statistical analysis and interpretation. 2nd edition. PRIMER-E.: Plymouth, United Kingdom.
Dollar, S.J. 1982. Wave stress and coral community structure in Hawaii. Coral Reefs 1:71-81.
Friedlander, A., Brown, E. K, Jokiel, P. L., Smith, W. R., and Rodgers, K.S. 2003. Effects of habitat, wave exposure, and marine protected area status on coral reef fish assemblages in the Hawaiian archipelago. Coral Reefs 22: 291-305.
Glynn P.W. 1976. Some physical and biological determinants of coral community structure in the eastern Pacific. Ecological Monographs 46:431-456.
Grigg, R. W. 1983. Community structure, succession and development of coral reefs in Hawai‘i. Marine Ecology Progress Series 11: 1-14.
Juvik, S.P. and Juvik, J.O.. Atlas of Hawai’i 3rd ed. 1998. University of Hawaii Press. Honolulu, Hawaii. 333 pp.
Reese, E.S. 1981. Predation on corals by the family Chaetodontidae: implications for conservation and management of coral reef ecosystems. Bulletin of Marine Science 31: 594-604.
Rodgers, K.S. 2001. A Quantitative Evaluation of Trampling Effects on Hawai’i’s Coral Reefs. Thesis Dept. of Geography. University of Hawai’i. 163 pp.
Rodgers, K.S., Cox, E.F., and Newtson, C. 2003. Effects of mechanical fracturing and experimental trampling on Hawaiian corals. Environmental Management 31:377-384.
Te, F.T. 2001. Response of Hawaiian Scleractinian Corals to Different Levels of Terrestrial and Carbonate Sediment. Dissertation. Department of Zoology, University of Hawaii. 286 pp.
Timbol, A.S. and Maciolek, J.A. (1978) Part A: Stream channel modification in Hawai‘i. Part A: Statewide inventory of streams; habitat factors and associated biota. Fish and Wildlife. OBS-78/16. USFW National Stream Alteration Team, Columbia, Missouri, 157 pp.
Woolfe, K.J. and Larcombe, P. 1988. Terrigenous sediment accumulation as a regional control on the distribution of reef carbonates. Special Publications of the International Association of Sedimentology 25: 295-310.
Last Update: 04/21/2008
By: Lea Hollingsworth
Hawai‘i Coral Reef Assessment & Monitoring Program
Hawai‘i Institute of Marine Biology
P.O. Box 1346
Kāne‘ohe, HI 96744