Reef-associated crustacean fauna: biodiversity estimates using semi-quantitative sampling and DNA barcoding
- 1.5k Downloads
The cryptofauna associated with coral reefs accounts for a major part of the biodiversity in these ecosystems but has been largely overlooked in biodiversity estimates because the organisms are hard to collect and identify. We combine a semi-quantitative sampling design and a DNA barcoding approach to provide metrics for the diversity of reef-associated crustacean. Twenty-two similar-sized dead heads of Pocillopora were sampled at 10 m depth from five central Pacific Ocean localities (four atolls in the Northern Line Islands and in Moorea, French Polynesia). All crustaceans were removed, and partial cytochrome oxidase subunit I was sequenced from 403 individuals, yielding 135 distinct taxa using a species-level criterion of 5% similarity. Most crustacean species were rare; 44% of the OTUs were represented by a single individual, and an additional 33% were represented by several specimens found only in one of the five localities. The Northern Line Islands and Moorea shared only 11 OTUs. Total numbers estimated by species richness statistics (Chao1 and ACE) suggest at least 90 species of crustaceans in Moorea and 150 in the Northern Line Islands for this habitat type. However, rarefaction curves for each region failed to approach an asymptote, and Chao1 and ACE estimators did not stabilize after sampling eight heads in Moorea, so even these diversity figures are underestimates. Nevertheless, even this modest sampling effort from a very limited habitat resulted in surprisingly high species numbers.
KeywordsCoral reefs Biodiversity Crustaceans Pacific Ocean DNA barcoding
To date, approximately 250,000 marine species have been described (Reaka-Kudla 1997; Groombridge and Jenkins 2000, Bouchet 2006). Our assumptions about total marine biodiversity (described and undescribed) are based on broad estimates from different methods. Examples include extrapolations from quantitative marine samples from specific ecosystems (e.g., the deep sea; Grassle and Maciolek 1992), extrapolations from the described fauna from better known regions or groups (e.g., European seas or Brachyura; Bouchet 2006), or comparisons with estimates of terrestrial biodiversity (e.g., tropical rain forests; Reaka-Kudla 1997). These have led to estimates of total marine species diversity spanning three orders of magnitude, from 5 × 105 (May 1994) to 108 (Grassle and Maciolek 1992) [see Bouchet (2006) for more examples of biodiversity estimates]. These extrapolations have created much controversy (Lambshead and Boucher 2003) as there is no easy and straightforward way of estimating global marine biodiversity reliably, leaving the door open for much conjecture and debate.
Uncertainty about the biodiversity of coral reefs is especially high. Although coral reefs represent less than 0.2% of the area of the ocean, they are the most diverse of all marine ecosystems on a per area basis, and perhaps absolutely as well, the deep sea being the other major repository of marine biodiversity (Sala and Knowlton 2006). Reaka-Kudla (1997) divided coral reef biodiversity into three main components: fishes, reef-building organisms and cryptofauna. For some groups, we have a relatively good understanding of the patterns of diversity and endemism because they are easily collected, and taxonomic expertise is long-standing (especially corals, fishes and some macroinvertebrates) (Veron 1995; Roberts et al. 2002; Karlson et al. 2004). Some biodiversity knowledge can be accessed from databases such as Fishbase (http://www.fishbase.org/search.php), Hexacorallia (http://www.kgs.ku.edu/Hexacoral) and Indo-Pacific Marine Mollusks (http://data.acnatsci.org/obis/find_mollusk.html), or printed resources such as the Systema Brachyurorum (Ng et al. 2008). However, most coral reef diversity is made up of small, cryptic species and species from poorly known groups. Thus, we do not know to even the nearest order of magnitude how many species are associated with coral reef ecosystems, with published estimates ranging from ~1 to 10 million species (Reaka-Kudla 1997; Small et al. 1998).
Coral reefs are also one of the most endangered marine ecosystems (Knowlton 2001; Bellwood et al. 2004), and dramatic declines in corals and fishes have been well documented (Gardner et al. 2003; Pandolfi et al. 2003; Bruno and Selig 2007; Knowlton and Jackson 2008). However, most reef biodiversity lies outside these two groups, and we have almost no understanding of how reef degradation threatens this biodiversity. The best example we have is the study by Idjadi and Edmunds (2006) who documented positive relationships between various aspects of reef condition (most notably topographic complexity) and the generic diversity of associated invertebrates.
Lately, an increasing awareness of these problems in the scientific community has led to several large-scale initiatives to inventory coral reef biodiversity. These include the investigation of the mollusk fauna in New Caledonia (Bouchet et al. 2002), the marine biodiversity survey of Guam and the Marianas (Paulay 2003), the Santo 2006 expedition in Vanuatu (http://www.santo2006.org), the Moorea Biocode Project (http://bscit.berkeley.edu/biocode), and the Census of Marine Life—Census of Coral Reefs (http://www.creefs.org) survey of French Frigate Shoals (northwestern Hawaiian Islands) in 2006 and in Australia in 2008. These expeditions have put a special emphasis on small and understudied organisms, particularly invertebrate, algal and microbial species.
The current rate of species description using traditional methods is extremely slow, and identifications based on traditional keys typically require specialized expertise. These two bottlenecks have severely limited our understanding of coral reef biodiversity. However, the revolution in molecular genetics has dramatically changed the potential for reef scientists to make progress in this area. DNA barcoding, in particular, has the potential to speed the identification of described species (Hebert et al. 2003), but its use to estimate species numbers, regardless of their formal taxonomic state, is probably even more important for understanding biodiversity patterns and trends. Especially, noteworthy is its ability to detect cryptic species, which are common on coral reefs and are difficult to detect using traditional taxonomic methods (Knowlton 1993, 2000). DNA barcoding has stimulated intense debate, primarily about its reliability at the species level, and it is accepted that a single molecular divergence cutoff for species delimitation is not defensible (Meyer and Paulay 2005). However, studies on crustaceans based on broad datasets have shown the utility of the 5′ end of the mitochondrial cytochrome oxidase subunit I in species delimitation (Lefébure et al. 2006; Costa et al. 2007).
This study focused on the crustacean fauna inhabiting coral reef interstices at five central Pacific Ocean localities, regions of moderate reef diversity well to the east of the coral triangle epicenter of reef diversity (Myers et al. 2000; Hughes et al. 2002; Hoeksema 2007). In order to avoid biases and inaccuracies associated with nonquantitative sampling strategies, the crustacean fauna was sampled from similar-sized and structured reef units—dead heads of Pocillopora coral—and extrapolation techniques were used to standardize richness data. Furthermore, sampling was restricted to a single depth and type of reef exposure. This allowed estimation of the crustacean species richness in this habitat, as well as the exploration of diversity patterns and estimation of the number of species still to be documented.
Materials and methods
Similar-sized dead heads of Pocillopora verrucosa (height + width + depth ≈ 90 cm ± 22%) were collected from four atolls in the Northern Line Islands in August 2005 (Kingman, Palmyra, Tabuaeran and Kiritimati) and in Moorea, Society Islands, French Polynesia, in August 2006. The coral heads were selected to minimize successional and environmental differences. All were collected from a depth of 10 m on the fore-reef. They were colonized by encrusting flora and fauna so that bare skeleton was obscured, but still remained attached to the reef at the base (so that proximity to the surface of the reef was standardized), and the branching structure of the coral was still present.
The heads were gently broken from the bottom with a hammer and chisel and quickly placed in a 20 l bucket underwater. No mesh or covering was used for this study, but we did not observe any animals escaping (most associates initially cling tightly to dead submerged coral heads if the head itself remains intact). All macroinvertebrates (>~5 mm in size) encountered were extracted from the head, shipboard in the Northern Line Islands and in the laboratory at Moorea, as follows: Each branch of the coral head was detached with a hammer and chisel and examined closely for motile invertebrates. The remaining rubble was placed in a bucket of seawater. When all the branches and the base were broken apart and examined, the fragments were then broken up in smaller pieces (~5 cm) and examined a second time for remaining creatures. We did not attempt to extract boring organisms. The seawater, in which the coral head and later on the coral fragments were kept, was sieved through a 2 mm sieve.
Decapods and stomatopods were sorted to morphospecies, and abundances of each recorded. Minute, often postlarval, decapods and peracarids were not sorted to morphospecies, but all were set aside for sequence-based identification. Each morphospecies was identified in the field to the lowest taxonomic rank possible [with identification confirmed or modified afterwards by Gustav Paulay and several specialists (see Acknowledgements)]. One to three exemplars were photographed, and up to five individuals of each morphospecies were processed for sequencing from each coral head. For larger organisms, a tissue sample was collected (most commonly a leg) for DNA analysis and frozen at −80°C, and the individual was then preserved in 95% ethanol and vouchered at Florida Museum of Natural History (FLMNH). For smaller specimens, the entire organism was frozen at −80°C, sequenced, and thus no vouchers were taken (≈15% of all organisms sampled). A single head typically took an entire day to process (collection of the coral, extraction of the associated fauna and vouchering and preservation of the specimens). This same procedure was applied for each new head sampled.
Extraction, PCR amplification and sequencing
Total genomic DNA was extracted from each specimen using DNeasy 96 Blood and Tissue kit (Qiagen) according to the manufacturer’s protocol. DNA was eluted in a final volume of 50 μl. A 658 base-pair (bp) fragment of the mitochondrial cytochrome oxidase subunit I gene (COI) was amplified using the primers LCO1490 and HCO2198 (Folmer et al. 1994). Twenty-five μl PCR amplifications were performed with 2 μl of DNA extract, 10 pM of each PCR primer and Ready-To-Go PCR beads (Amersham Pharmacia Biotech), each containing 1.5 U Taq polymerase, 10 mM Tris–HCl at pH 9, 50 mM KCl, 1.5 mM MgCl2, 200 μM of each dNTP and stabilizers including bovine serum albumin. The PCR conditions consisted of 1 min at 94°C followed by 5 cycles of 40 s at 94°C, 40 s at 45°C, 60 s at 72°C; followed by 35 cycles of 40 s at 94°C, 40 s at 51°C, 60 s at 72°C; followed by 5 min at 72°C. Successful PCRs, where a single fragment was amplified, were purified using the QIAquick PCR Purification Kit (Qiagen). When several fragments were obtained (low annealing temperatures made it easier to amplify from a broad taxonomic array of organisms, but sometimes resulted in the amplification of pseudogenes), the PCR product was run on an agarose gel (2%) containing EtBr, and the target fragment was excised from the gel and purified using QIAquick Gel Extraction Kit (Qiagen). Automated sequencing was performed directly on purified PCR products using ABI BigDye terminator V3.1. Sequence reactions were purified using Millipore 96-well plates loaded with Sephadex G-50 and run on an ABI 3130xl genetic analyzer (Applied Biosystems). Products were sequenced in both directions using LCO1490 and HCO2198.
Sequences were assembled and edited using Sequencher v. 4.5 (Gene Codes, Ann Arbor, MI). Each unique sequence served as a blast query to the GenBank database to identify the most similar sequence in GenBank to the queried sequence. This allowed us to remove problematic sequences, which stemmed from contamination problems (noncrustacean sequences) or sequences showing a mismatch with the initial description of the organism sampled (e.g. a crab sequence from an animal initially recorded as a shrimp). A few putative pseudogenes were also removed based on the presence of stop codons or reading frame shifts. The rest of the sequences were subsequently aligned using MacClade 4 (Maddison and Maddison 2000) and submitted to GenBank (accession numbers: GQ260847–GQ260981).
In order to cluster the sequences into Operational Taxonomic Units (OTUs), nucleotide sequence divergences were calculated with DNADIST of the Phylip package (Felsenstein 1989) using the Kimura-two-parameter (K2P) model. The pairwise distances served as an input to DOTUR (Schloss and Handelsman 2005). In order to choose the sequence dissimilarity threshold used for species discrimination, a step function analysis was run, testing the number of OTUs found as a function of the value of the threshold for crustaceans found in Moorea, the Line Islands and the combined data from Moorea and the Line Islands.
The furthest neighbor clustering algorithm, with a 5% dissimilarity for definition of Operational Taxonomic Units (OTUs), was used for clustering sequences into OTUs, generating rarefaction curves, and calculating the species richness estimators ACE and Chao 1 (Hortal et al. 2006) for both the total crustacean fauna and for the decapods only. Chao1 and ACE are abundance-based nonparametric estimators of species richness that work by examining the number of species in a sample observed more than once (2 times or up to 10 times for Chao1 and ACE, respectively) relative to the number of species that are observed just once. The advantage of using these estimators is that the estimated diversity of samples can be compared, even when the true diversity of the total population is not known. In the absence of complete inventories, nonparametric estimators have been shown to perform better than most other methods, such as observed species richness, species–area curves or asymptotic estimators. Because they depend on total species abundances, they seem to be quite robust despite variations in sample grain size, and when data on abundances are available, ACE and Chao1 show the most precision. However, the precision of abundance-based estimators is dependent on sample coverage (but see Hortal et al. 2006) and both estimators give a lower bound to species richness, thus producing conservative estimates.
Abundance for each OTU was given by the number of similar sequences (using a 5% threshold) plus the numbers recorded in the field for the abundance of unsequenced individuals assigned to each morphospecies. This method potentially underestimates real diversity, as we may have missed some cryptic diversity, but we minimized this problem by assigning individuals to different morphospecies if there were any doubts and by sequencing multiple individuals for abundant and difficult morphospecies. In no case did we recover cryptic genetic lineages within assumed morphospecies, indicating that field identification was effective within each of these locations.
EstimateS (Colwell 2005) was used to test the performance of the diversity estimators according to the number of heads sampled using the Moorea data set. The computation of both Chao 1 and ACE estimates was done for each of one through eight heads sampled, with a randomized order of samples without replacement for 100 runs. A sample-based rarefaction curve was also computed.
Comparison with existing data sets
In addition to blasting against GenBank, we compared our results to the marine invertebrate barcode database of the Moorea Biocode Project (MBP—http://bscit.berkeley.edu/biocode/) in order to evaluate the amount of overlap between the two studies. For the MBP, crustacean species were collected by hand at 1–30 m, from fore-reef and lagoonal habitats, during the summer of 2006 as part of a pilot study to collect genetic vouchers of every species on the island of Moorea, French Polynesia. These data are available in the Barcode of Life Database at www.boldsystems.org in the public project ‘MBMIA’, Moorea Biocode Marine Inverts A-Crustaceans.
Amplification and sequencing success
From 22 dead Pocillopora heads, 403 usable COI sequences were generated from 500 individuals, for an overall success rate of 80.6%; by comparison, the Guelph barcoding center had a success rate of 70.2% (515 of 734) for Moorean crustaceans processed during the MBP pilot project. Of the 97 organisms that were not sequenced, almost half (N = 46, 9.2% total) were not successfully amplified and a third (N = 33, 6.6% total) produced unreadable sequences (e.g., mixed signal). There was a strong phylogenetic bias, in that 37% of the Caridea failed to sequence. Pseudogenes accounted for one-tenth (N = 11, 2.2% total) of the unusable sequences, mainly within the genus Petrolisthes. The remainder (N = 7, 1.4%) was removed because of apparent contamination problems (i.e. the sequence did not match the taxonomic group from which the specimen originated).
Diversity and taxonomic distributions
N Line Islands
No. of heads
No. tissue samples
No. of OTUs
N Line Islands
Semi-quantitative sampling and diversity estimators
The standardized, semi-quantitative sampling and molecular techniques applied in this work allowed us to estimate the diversity of the crustacean fauna in a comparable and reproducible way and with greater speed and precision compared to that obtained by previous evaluations based on conventional sampling methods. Conventional collection methods used on reefs, such as hand collecting, play a very important role in investigating and cataloguing the biologic diversity and allow for taxonomic inventories and species discovery. However, these methods are poorly adapted to estimate the likely global diversity of coral reefs because they are difficult to standardize, which makes it difficult to compare and combine independent estimates. They are also usually biased in favor of the larger, more numerous and easily studied plants and animals (but see Bouchet et al. 2002). Quantitative sampling methods permit more rigorous biodiversity estimates and comparisons of biodiversity among habitats and sites. Thus, they also have the potential to more accurately evaluate the extent to which biodiversity is being lost as a function of anthropogenic reef degradation.
It is extremely challenging to apply quantitative sampling on coral reefs because of their heterogeneous, rigid and complex structure; this makes them unsuitable for quantified sieving, as is used for soft sediments (Markmann and Tautz 2005). Because of the extraordinary diversity of coral reefs, exhaustive inventories of reef-associated fauna also still remain impractical. Thus, developing standardized sampling methods that work for reefs is a high priority.
We focused on the crustacean fauna inhabiting dead Pocillopora heads in the Northern Line Islands and in Moorea (French Polynesia). In our sampling, crustaceans accounted for a substantial fraction (30–40%) of the macrofaunal diversity encountered in this habitat, but other groups remain to be analyzed. Moreover, our sampling technique does not allow for a thorough sampling of the microcrustaceans, such as amphipods, which are too small to be easily detected on this heterogeneous surface; quantitative sampling of the microfaunal diversity requires alternative approaches such as the use of a mesh bag for the rubble collection, smaller sizes of sieves and dissecting microscopes for the extraction of the organisms. Therefore, even the crustacean diversity reported here is clearly a substantial underestimate for the coral heads sampled, which are themselves a tiny part of the coral reef habitat as a whole.
This study also highlighted the need to sample more heads of Pocillopora in order to have a more precise estimate of the number of species, even for the taxa and habitat type analyzed. In any community, species richness estimates are always tied to sampling effort (Hughes et al. 2001). In this study, the rarefaction curves did not approach a constant value, even for this relatively restricted habitat type (fore-reef, 10 m) (Fig. 4). ACE and especially Chao I estimates performed better, but only for the highest numbers of heads sampled.
This study was based on the use of molecular sequences as a way to determine species diversity. We are aware that species delimitation based on a single molecular marker and the use of a molecular threshold may raise questions about the validity of these results. However, no single approach can provide a definitive conclusion on species boundaries, and challenges exist for any taxonomic and DNA-based methods. Lefébure et al. (2006) have demonstrated that artifacts for species delimitation due to the use of a sole maternally inherited molecule have little effect in crustaceans. Moreover, intraspecific and congeneric divergences in COI for crustaceans, as shown by Fig. 1, overlap only weakly (Lefébure et al. 2006). Costa et al. (2007), studying the ability of DNA barcodes to provide species level identifications in crustaceans, found that sequence divergence among congeneric species averaged 17.16% whereas intraspecific variation averaged 0.46%, and therefore, that species recognition was straightforward in 95% of the cases. DNA barcoding seems to be a useful tool in the discrimination of crustacean species and in diversity studies in this group. DNA sequence-based identification is especially useful and informative for juveniles and microcrustaceans.
The one-to-one correspondence between adult decapod morphospecies identified on the basis of field appearance and genetic OTUs is encouraging. This indicates that field taxonomic methods at a single locality can be very effective in delineating species, even in those taxa where sibling species are common. However, DNA-based identification is likely to reveal more complex patterns in among-site comparisons, where allopatric sibling species complexes are often encountered (e.g., Meyer and Paulay 2005). These were also evident in our data set; for example, individuals identified as Perinia tumida from the Society and Line Islands are deeply divergent. Moreover, standard sequence data (DNA barcodes) permit research teams to cross-correlate morphospecies identified by different workers, which is particularly useful for species, currently without names.
It is now possible to bring state of the art sequencing technology to reefs located in the heart of marine biodiversity. Therefore, DNA barcodes could be used as a real-time guide to traditional surveys.
Comparisons with other studies
Past studies of the cryptofauna inhabiting Pocillopora heads have mainly focused on organisms associated with live corals, and crustaceans are also the main component of this fauna. In living Pocillopora damicornis, examining 30 to 66 heads, more than 50 species of decapods have been recorded for the Gulf of Panama (Abele 1976; Abele and Patton 1976), 101 species of crustaceans for the Great Barrier Reef (Austin and Austin 1980) and 54 species of crustaceans for Western Australia (Black and Prince 1983). Live corals harbor fewer species than dead corals as they mostly host obligate symbionts, whereas species known from a variety of reef habitats can be found on dead Pocillopora. For example, during our sampling campaigns, seven heads of live Pocillopora were also examined from three localities of the Northern Line Islands (Palmyra, Tabuaeran, and Kiritimati). Using the same molecular techniques as described above, we found that the live Pocillopora heads harbored at least 28 species of crustaceans (all decapods), compared to 69 spp. of crustaceans and 60 spp. of decapods from dead heads from the same three islands. Of the crustaceans from living Pocillopora, 16 also were found in dead Pocillopora heads.
One of the striking results of this study is the high proportion of rare species: 44% of all species were singletons, and an additional 33% of species were sampled several times but only from one island. Even the overlap between the crustacean fauna inhabiting the dead Pocillopora heads and the crustacean fauna recorded by the Moorea Biocode Project was low (43%), especially considering the intensity of the sampling effort for the latter (~6 weeks of collecting macro-invertebrates at 48 collecting stations). This pattern has been found before with other reef-associated cryptofaunal groups such as isopods (Kensley 1998) and mollusks (Bouchet et al. 2002); in the latter study, 32% of the species collected in a survey of mollusk diversity in New Caledonia were collected at a single station, and 20% of the species were represented by single specimens. Therefore, the bulk of the reef cryptofauna diversity is made up of low-abundance species. Similarly, singletons in tropical arthropods surveys averaged 32% (Coddington et al. 2009), highlighting the need to increase sampling size to obtain robust and reliable species richness estimates.
None of the species sampled in this study matched COI sequences in GenBank [the highest sequence identity being 91%, indicating that they are probably different species within species complexes (e.g. Knowlton et al. 1993)], and underscoring the still limited availability of DNA barcoding data for identifying coral reef invertebrates. The building of a comprehensive COI barcode database for marine invertebrates is a challenge that several large-scale projects are undertaking (e.g., the Marine Barcode of Life and the Moorea Biocode Project). The assemblage of DNA barcodes will allow for an effective identification system that will be very useful in understanding the structure of reef diversity.
Biodiversity in coral reefs is exceptionally high. As an illustrative example, after centuries of taxonomic inventories in one of the most intensively and comprehensively inventoried regions of the world, only 212 species of true crabs (Brachyura) have been listed for European Seas (Bouchet 2006). When we examined 22 dead Pocillopora heads, sampled from a single depth in five localities from the central Pacific, we found 65 OTUs that could be identified as brachyurans among the 403 sequences distinguished at the 5% level. This represents ~30% of the recorded brachyuran diversity in Europe, and ~1% of the global brachyuran fauna [6,793 recognized species (Ng et al. 2008)]. The implied challenges for documenting the diversity of species on coral reefs are enormous. Bouchet (2006) estimated that, at the current rate of species description, it would take another 250–1,000 years to complete the inventory of marine biodiversity. Clearly new approaches are needed that are repeatable and cross-comparable, not only because the diversity of reefs is so staggering, but also because we have no idea of how biodiversity on reefs is threatened due to rapid rates of reef degradation. In this effort, standardized sampling coupled with molecular analyses will play a key role.
We thank Melissa Roth, Machel Malay, Vaiarii Terorotua and Sea McKeon for their help in the field. We are grateful to Dirk Steinke and Paul Hebert from the University of Guelph for their assistance with DNA barcoding. We also thank Arthur Anker, Rob Lasley, Enrique MacPherson, Sea McKeon, Cory Pittman and Joseph Poupin for their help with identifications. We gratefully acknowledge the Alfred P. Sloan Foundation through their support of the Census of Marine Life CReefs project, the Gordon and Betty Moore Foundation for their support of the Moorea Biocode Project, and the Moore Family Foundation, the National Geographic Society and other donors for their support of the Northern Line Islands expedition.
This article is distributed under the terms of the Creative Commons Attribution Noncommercial License which permits any noncommercial use, distribution, and reproduction in any medium, provided the original author(s) and source are credited.
- Bouchet P (2006) The magnitude of marine biodiversity. In: Duarte C (ed) The exploration of marine biodiversity: scientific and technological challenges. Fundación BBVA, Bilbao, Spain, pp 31–64Google Scholar
- Chao A (1984) Non-parametric estimation of the number of classes in a population. Scand J Stat 11:265–270Google Scholar
- Colwell RK (2005) EstimateS: statistical estimation of species richness and shared species from samples. Version 7.5. User’s guide and application published at: http://purl.oclc.org/estimates
- Felsenstein J (1989) PHYLIP—phylogeny inference package (version 3.2). Cladistics 5:164–166Google Scholar
- Groombridge B, Jenkins MD (2000) Global biodiversity: earth’s living resources in the 21st century. World Conservation Monitoring Centre, Cambridge, United KingdomGoogle Scholar
- Maddison WP, Maddison DR (2000) MacClade. Sinauer Associates, Sunderland, MAGoogle Scholar
- Meyer CP, Paulay G (2005) DNA Barcoding: error rates based on comprehensive sampling. PLoS Biol 3:2229–2238Google Scholar
- Ng PKL, Guinot D, Davie PJF (2008) Systema Brachyurorum: part I. An annotated checklist of extant brachyuran crabs of the world. Raffles B Zool 17:1–286Google Scholar
- Paulay G (2003) Marine biodiversity of Guam and the Marianas: overview. Micronesica 35–36:3–25Google Scholar
- Reaka-Kudla M (1997) The global biodiversity of coral reefs: a comparison with rain forests. In: Reaka-Kudla M, Wilson DE, Wilson EO (eds) Biodiversity II: understanding and protecting our biological resources. Joseph Henry Press, Washington, D.C., pp 83–108Google Scholar
- Small A, Adey A, Spoon D (1998) Are current estimates of coral reef biodiversity too low? The view through the window of a microcosm. Atoll Res Bull 458:1–20Google Scholar
- Veron JEN (1995) Corals in space and time. University of New South Wales Press, Townsville, QldGoogle Scholar