A zooplankton sample was collected off Pohnpei Island, Micronesia (6°16'N, 162°09'E). A cDNA mitochondrial COI
c oxidase subunit I
) gene library was constructed from the sample, and 1,336 inserts containing the mitochondrial COI
gene were randomly sequenced [DDBJ: AB332438
]. A cDNA rather than a gDNA library was constructed to remove pseudogene sequences from the analysis [4
]. The mismatch distribution of these 1,336 sequences revealed a high frequency of very small (<0.03) genetic distance sequence pairs (Figure ). These sequence pairs with very small genetic distances were assumed to have originated from the same species (discussed below). A second peak was observed around a distance of about 0.14 (from 0.13 to 0.16), and most of these counts were comparisons between two phylogroups in the Copepoda clade (Figure , Clades 1 and 2). The frequencies between these peaks were very low. The minimum frequency (106 counts) was observed in the range between 0.12 and 0.13. Based on this observation, we set the criterion that if the genetic distance of two sequences was greater than 0.12, the sequences were derived from different species. If the genetic distance of two sequences was less than 0.12, then we considered the sequences to be derived from the same species. The genetic distances of the mitochondrial COI
gene sequence have been reported from various animal taxa (mainly Vertebrata and Arthropoda), and the general ranges for the intra- and interspecies distances are 0.0001-0.05 and 0.04-0.21, respectively (Kimura two-parameter model) [5
]. Although it is not a conclusive value for animal species definition, we have tentatively taken a genetic distance of 0.12 as the boundary between intra- and interspecies distances in this study, and this value was in the range of interspecific genetic distance reported previously [5
]. The rarefaction curve was estimated using the criterion of a genetic distance of 0.12 (Figure ) using DOTUR [6
]. Although the number of observed OTUs is still growing in 1,300 sequenced colonies, the rate of increase of the curve decreased gradually after around 100 sequenced colonies. Figure shows the relationships between species richness estimated by Chao1, rarefaction, and percentage of sequence differences used for the estimation. The figure shows gradual changes of Chao1 and rarefaction around sequence differences of 0.12. As the distance of 0.12 is not a conclusive value for species definition, caution is required in the further use of this value.
Mismatch distributions of the pairwise genetic distances for the 1,336 mitochondrial COI sequences. A total of 891,780 frequencies calculated from the 1,336 sequences are plotted in the figure.
Figure 2 Unrooted neighbour-joining tree of the 1,336 mitochondrial COI gene sequence. Numbers beside internal branches indicate bootstrap values (>90%) obtained for 1,000 replicates (indicated for major branches only). Each dot represents a single mitochondrial (more ...)
Rarefaction analysis of the 1,336 mitochondrial COI gene sequences.
Relationship between species richness estimated by Chao1, rarefaction, and the percentage sequence difference used for these estimations.
We conducted BLASTN searches against the GenBank non-redundant database using as queries all sequences derived from the analysis. Among the sequences, those that fulfilled the criteria (BLAST score and similarity greater than 100 and 83%, respectively) were assigned to 11 taxonomic groups (Figure ). Several of the assigned sequences showed very high degrees of nucleotide similarity to known species, including Copepoda (Candacia longimana, Cosmocalanus darwinii, Neocalanus robustior, Rhincalanus rostrifrons), Euphausiacea (Stylocheiron carinatum), Mollusca (Clio pyramidata, Sthenoteuthis oualaniensis, Strombus mutabilis, Strombus wilsoni), and Vertebrata (Coryphaena hippurus; Table ). The very high degrees of similarity indicated that these species were actually collected in our sampling. Among them, one vertebrate species, Coryphaena hippurus, and two benthic gastropod species, Strombus mutabilis and Strombus wilsoni, were sampled as non-holozooplanktonic animals in the dispersal life history phase as pelagic larvae. These observations indicated that application of this analysis enables the estimation of larval dispersal, which is difficult to achieve based on morphological observations.
BLASTN search results for sequences that showed more than 98% similarity to subject sequences
Figure shows an unrooted neighbour-joining tree of the 1,336 zooplankton COI
sequences. Overall, each taxonomic group formed a single cluster including Gastropoda, Chaetognatha, Euphausiacea, Decapoda, Vertebrata, Copepoda, and Cephalopoda. There were also two cases in which the taxonomic assignment did not work well. The first was the occurrence of Hexapoda in various clusters, which rarely occurs in the ocean environment, except pleustonic insects of the genus Halobates
. The second was the difficulty of assignment of taxonomic groups due to low BLAST scores and similarities (coloured grey in Figure ). The most plausible reason for these ambiguities is the paucity of mitochondrial COI
sequences for some taxa in the DNA database. In general, the mitochondrial COI
gene sequences in the DNA database are biased among taxa, and this bias was assumed to be the main reason for the occurrence of Hexapoda in our analysis. The most efficient solution for these problems will be the expansion of zooplankton DNA barcode, and it is hoped that the progress of the Barcode of Life project [7
] in collaboration with the Census of Marine Zooplankton will fill these gaps.
To our knowledge, the Discovery SOND cruise [8
] is the only other attempt to date to estimate the species richness of a whole zooplankton community collected at a single site. In this series of studies [9
], a total of 618 species of zooplankton were identified and counted in samples collected around the Canary Islands (Table ). The extrapolated species richness (Chao1 [20
]) of the present study was estimated as 188.90 (95% confidence interval, 156.79-255.60) using DOTUR [6
]. Our results cannot be directly compared with the SOND cruise data because of differences in sampling effort between the two studies. In the SOND cruise, two primary types of sampling equipment were used: Isaacs-Kidd midwater trawl and N113H.
Comparison of species that occurred in the present study and the SOND cruise
Furthermore, about 76 vertically stratified zooplankton samples that were collected above 1,000 m were combined to estimate the occurrence of species [19
]. In contrast, the present study was conducted based on a single sample collected from a depth of 721 m to the surface. These sampling effort differences may have accounted for the differences in species richness between the SOND cruise and the present study. In addition, the lower species richness in the present study may have been due to our experimental design. In the present study, after construction of the cDNA library from mRNA, the mitochondrial COI
genes were amplified with "universal (LCO1490 [21
])" and polyT primers. It is possible that some of the mitochondrial COI
gene sequences may not have been amplified due to primer mismatch for some species. Although the single-gene zooplankton community analysis approach is an efficient means of collecting sequence information, given technical difficulties due to primer mismatch, further studies and the development of novel methodologies are required to gain a complete understanding of zooplankton diversity.