An annotated list of fish parasites (Isopoda, Copepoda, Monogenea, Digenea, Cestoda, Nematoda) collected from Snappers and Bream (Lutjanidae, Nemipteridae, Caesionidae) in New Caledonia confirms high parasite biodiversity on coral reef fish

Background Coral reefs are areas of maximum biodiversity, but the parasites of coral reef fishes, and especially their species richness, are not well known. Over an 8-year period, parasites were collected from 24 species of Lutjanidae, Nemipteridae and Caesionidae off New Caledonia, South Pacific. Results Host-parasite and parasite-host lists are provided, with a total of 207 host-parasite combinations and 58 parasite species identified at the species level, with 27 new host records. Results are presented for isopods, copepods, monogeneans, digeneans, cestodes and nematodes. When results are restricted to well-sampled reef fish species (sample size > 30), the number of host-parasite combinations is 20–25 per fish species, and the number of parasites identified at the species level is 9–13 per fish species. Lutjanids include reef-associated fish and deeper sea fish from the outer slopes of the coral reef: fish from both milieus were compared. Surprisingly, parasite biodiversity was higher in deeper sea fish than in reef fish (host-parasite combinations: 12.50 vs 10.13, number of species per fish 3.75 vs 3.00); however, we identified four biases which diminish the validity of this comparison. Finally, these results and previously published results allow us to propose a generalization of parasite biodiversity for four major families of reef-associated fishes (Lutjanidae, Nemipteridae, Serranidae and Lethrinidae): well-sampled fish have a mean of 20 host-parasite combinations per fish species, and the number of parasites identified at the species level is 10 per fish species. Conclusions Since all precautions have been taken to minimize taxon numbers, it is safe to affirm than the number of fish parasites is at least ten times the number of fish species in coral reefs, for species of similar size or larger than the species in the four families studied; this is a major improvement to our estimate of biodiversity in coral reefs. Our results suggest that extinction of a coral reef fish species would eventually result in the coextinction of at least ten species of parasites.


Background
Parasites probably constitute the least known component of biodiversity in coral reefs, which are considered some of the most diverse ecosystems on the planet [1]. An early evaluation of parasite biodiversity of fish of the Great Barrier Reef (GBR) in Australia proposed a number of 20,000 parasites (all groups included) in the 1,000 fish species believed to exist in the area at this time; however, this evaluation, published as short papers [2,3] was based on very limited data. More reliable estimates are available for only two groups, the digeneans and monogeneans. Estimates were 2,270 digenean species in the 1,300 fish species of the GBR [4] and 2,000 monogenean species on the 1,000 fish species recorded around Heron Island, in the southern GBR [5].
An eight-year program allowed us to investigate the biodiversity of fish parasites off New Caledonia (South Pacific), the largest coral lagoon of the world. A compilation of available literature including a number of papers produced by this program [6] concluded that only 2% of fish parasite biodiversity was known in New Caledonia. Two subsequent comprehensive papers provided abundant, previously unpublished data and a compilation of already published information on two families of fish, the Serranidae (groupers) and the Lethrinidae (emperors) [7,8]. In this paper, we provide information about the parasites of the Lutjanidae, Nemipteridae and Caesionidae and compare our results with those already published for the other families.

Results
Results are presented as a host-parasite list (Appendix 1), a parasite-host list (Appendix 2) and a list of material deposited (Appendix 3). The number of host-parasite combinations (HPCs) and the number of species-level identified parasitehost-parasite combinations (SLIP-HPCs) found in each fish species are given in Table 1.

Comments on each group
For brevity, in this section references to our own published papers on these fish families (available in Table 2 and Appendix 2) are kept to a minimum. For parasites, the "minimized number of taxa" is a cautious minimized evaluation in which all unidentified taxa in a group are counted as a single taxon [8].

Fish
In this paper, we group results from three families of fish, namely the Lutjanidae, Nemipteridae and Caesionidae. Clearly, most of the results concern the Lutjanidae but we included the two other families because they are closely related [9][10][11]. Modern molecular phylogenies are available for the Lutjanidae [12][13][14][15] and confirm the close relationship of the Lutjanidae and Caesionidae.
According to the most recent survey [16], the Lutjanidae, Caesionidae and Nemipteridae include, respectively, 17, 4, and 5 genera and 108, 22 and 66 species, with a total of 26 genera, 196 species. The numbers of species in New Caledonia [17] are, respectively, 43, 13, and 16, with a total 72 species. In this work, we report parasitological results from 18 lutjanid species, 1 caesionid and 6 nemipterids; the total, 25 species, represents 34% of the species reported from New Caledonia, and 13% of the world number of species for the three families.
Diets of lutjanids and nemipterids off New Caledonia mainly comprise fish, crustaceans and occasionally molluscs [18], all of which can serve as intermediate hosts for parasites such as nematodes, digeneans and cestodes.
Most fishes included in this study are reef-dwelling; however, we also include several lutjanids (two species of Etelis and three species of Pristipomoides) which are deeper water fishes, collected from the outer slope of the barrier reef of New Caledonia [19]. These fishes provide data for a comparison of the parasitic fauna of coral-associated and deeper sea fishes.

Isopoda
Adult isopods were rare and belonged to three families: Aegidae, Corallanidae and Cymothoidae. The single aegid, Aega musorstom, was found on a deep water lutjanid. Two cymothoids (Anilocra gigantea and An. longicauda) were found only on deep water lutjanids, but the single corallanid, Argathona macronema, was found on a coral dwelling lutjanid.
An. gigantea was already known from New Caledonia and was recorded from the branchial region of the deep water lutjanid Etelis carbunculus off "Banc de la Torche, au sud-est de la Nouvelle Calédonie" [26]. It was also recorded from the Pacific Ocean from the gills of Epinephelus sp. and Pr. flavipinnis, off Suva reefs, Suva, Fiji [27] and from the Indian Ocean from an unidentified host [28][29][30]. We found this species again on Et. carbunculus, but Et. coruscans and Pr. filamentosus are new host records. Interestingly, we did not collect this species from the branchial region or from the gills of the host fish, as reported by previous authors but on the anterior part of the body just behind the head. A female specimen of An. gigantea attached behind the head of Pr. filamentosus is illustrated by a colour photograph (Figure 1).
An. longicauda was already known from the Indian and the Pacific Oceans [30]. It was previously recorded from Swains Reefs, Great Barrier Reef, Marion Reef, Australian Coral Sea, North West Shelf of Western Australia, Krakatua, Indonesia [31], Singapore and Poulo Condor, Vietnam [29,31], Ragay Gulf, Pasacao and Maribuyoc Bay, Bohol Island, Philippines [27]. This species has been reported from Plectorhynchus goldmani, Diagramma picta and Priacanthus sp. [31]. Pr. argyrogrammicus is a new host record and New Caledonia is a new geographical record.
Aega musorstom was already known from New Caledonia, in the vicinity of Western New Caledonia including the Coral Sea region of the Chesterfield Archipelago and the Loyalty Islands, at depths from 475 to 615 m [32]. Only one fish association was noted, "Synagonopi sp. 1" probably a species of Synagrops (Acropomatidae) [32]. Pr. filamentosus is a new host record.
Argathona macronema was already known from New Caledonia [7,33]. It was previously reported from Lutjanus quinquelineatus 12 Etelis coruscans 18 Pristipomoides argyrogrammicus  Larval isopods belonged to the Gnathiidae. Gnathiids, found as praniza larvae, were collected on 6 species of nemipterids and lutjanids (5 reef-dwelling, 1 deep water). In New Caledonia, larval gnathiids were found on most fish families examined (serranids, lethrinids, lutjanids, nemipterids and many others). Adult isopods were found on serranids and lutjanids but not on lethrinids and nemipterids [7,8]. The biodiversity of larval gnathiids is hard to evaluate [34,35], but it is likely that several species are involved.

Copepoda
Fourteen taxa, including 6 identified at the species level, were found. Seven species of Hatschekia were distinguished but only two are known species, the other five ( Figure 2) are not formally described. A total of 21 undescribed Hatschekia species has now been listed from New Caledonian fish ( [7,8]; this paper). Hatschekia tanysoma was originally described from Kuwait Bay, from Lu. fulviflamma [36] and is reported here from the Pacific for the first time, but from the same host. In contrast H. clava was described from Heron Island from material collected from Lu. carponotatus (Richardson) (as Lu. chrysotaenia) [37].
The copepods belonged to five families, namely Caligidae, Dissonidae, Hatschekiidae, Lernaeopodidae and Pennellidae. Larvae and premetamorphic adults belonging to the Pennellidae were found on the deep-sea lutjanid Et. coruscans. The only adult member of this family found during eight years of sampling was a single female of Lernaeolophus sultanus ( Figure 3) found on Pr. filamentosus. Pennellids are known to utilise two different hosts during their life cycle, either two different fish hosts or a pelagic mollusc and a final fish host [38]. However, the life cycle of no Lernaeolophus species has ever been elucidated so it is not possible to confirm whether the developmental stages found on Pr. filamentosus are those of L. sultanus. L. sultanus exhibits the lowest host specificity of any copepod parasite, occurring on 16 different host fishes in the Mediterranean [39].

Monogenea
The minimized number of taxa for polyopisthocotyleans was 2 and for monopisthocotyleans was 23.
Polyopisthocotyleans were represented by Allomicrocotyla sp. on the deep-sea lutjanid Et. coruscans, and several records of unidentified microcotylids or other polyopisthocotylean families from coral-associated lutjanids and nemipterids. Polyopisthocotyleans on reef lutjanids were rare; similarly, polyopisthocotyleans were rare on Lu. griseus in the Gulf of Mexico [43]. The rarity of polyopisthocotyleans from 4 species of lutjanids off Heron Island on the Great Barrier Reef is apparent from the literature [44] and from unpublished observations by I.D. Whittington at the same locality.
Ancyrocephalids included a series of species of the recently described genus Haliotrematoides Kritsky, Yang & Sun, 2009, several species of Euryhaliotrema Kritsky & Boeger, 2002, which are still undescribed, and we also have records of various ancyrocephalids not attributed to a genus. Clearly, the lutjanids harbour an impressive ancyrocephalid radiation, with probably several species  on each fish species; some of these species seem to be strictly species-specific, others are found on up to 5 different host species [45]. These ancyrocephalids might be a threat for cultured snappers [46].
Diplectanids included several species of Diplectanum and one species of Calydiscoides. Calydiscoides species are numerous in lethrinids [8]; only one species was found in New Caledonia in the families studied here, in the nemipterid Pentapodus aureofasciatus, but none in members of Scolopsis and Nemipterus, which are known to harbour sometimes numerous species of Calydiscoides in other localities [44,47,48].
Four species are attributed here to Diplectanum. D. opakapaka Yamaguti, 1968 was described from Pr. microlepis and Aphareus rutilans off Hawaii [49] and D. curvivagina Yamaguti, 1968 was described from Pr. sieboldii and Pr. auricilla off Hawaii [49] and have rarely been recorded since [50]. D. fusiformis Oliver & Paperna, 1984, described from Lu. kasmira off Kenya [51] was recorded again from its type-host and Lu. fulvus in French Polynesia and Hawaii [52,53]. D. spirale Nagibina, 1976 described from Lu. fulviflamma [54] has apparently not been recorded since its original description. In the absence of a comparative examination of the type specimens, we prefer to keep our identifications of these four species as "cf." and we do not comment on their generic attribution.
Gyrodactylids were represented by a single specimen found on Macolor niger, which is a relatively difficult fish to catch and unfortunately additional specimens could not be obtained. It is the first gyrodactylid collected from a lutjanid. It should be noted that several coralassociated lutjanids were soaked for collection of skin monogeneans such as capsalids and digeneans such as transversotrematids, but no other gyrodactylid specimens were recovered.
Capsalid systematics is currently under reinvestigation (e.g. [55][56][57]) after Perkins et al. [58] demonstrated that the current morphological classification has limited congruence with a phylogenetic hypothesis based on three unlinked nuclear genes. Apparently homoplastic morphological features were highlighted throughout the molecular phylogeny [58]. This has entailed a reluctance to assign some taxa to genera until appropriate characters and generic and subfamilial definitions are better refined within a phylogenetic context. Hence 'identifications' of four taxa are herein provided only as Capsalidae sp. 6, 7, 13 and 17, differentiated phylogenetically by Perkins [59] using nuclear and mitochondrial markers, but from morphological 'identifications' made by I.D. Whittington.
Of the Capsalidae we report here, four from the gills were identified to species (Benedenia elongata from three lutjanid species; Lagenivaginopseudobenedenia etelis from Et. coruscans; Pseudonitzschia uku from Aprion virescens; Trilobiodiscus lutiani from Lu. argentimaculatus). Two capsalid species recovered from gills of Macolor niger and Ne. furcosus remained unidentified. A species assigned only as a Metabenedeniella sp. was recovered from the pectoral fins (see [59]; probably a new species, I.D. Whittington unpublished) and the four capsalid species assigned only as Capsalidae sp. 6, 7, 13 and 17 [59] were recovered from the gills, fins, body washings, branchiostegal membranes and the head of their hosts (for details, see Appendix 1). Additional external sites, rarely examined in this study, received careful scrutiny only when I.D. Whittington visited Nouméa in October/November 2008. During his visit, thorough fish necropsies (four specimens of Lu. vitta; six specimens of Ne. furcosus; one specimen of Lu. kasmira [which was totally uninfected by capsalids]; two specimens of Pr. argyrogrammicus; one specimen of Lu. argentimaculatus) paid particular attention to external microhabitats known to support capsalids (e.g. [60]) followed by freshwater bathing of the same tissues to ensure recovery of parasites that may be cryptic due to camouflage or transparency (e.g. [61]). Dissections of fish at other times, when microhabitats other than gills that may harbour capsalids remained unstudied, seem likely to have appreciably underestimated the diversity of capsalid monogeneans from the caesionids, lutjanids and nemipterids examined.
Other factors in this study that contributed to an inability to assign capsalids to genera from gills, pelvic, pectoral and anal fins, body washings, branchiostegal membranes and head included small numbers of specimens recovered and the juvenile status of many capsalid individuals from Ne. furcosus. Adult specimens of B. lutjani from Lu. carponotatus from Heron Island on the Great Barrier Reef preferentially inhabit the branchiostegal membranes, and the pelvic fins was the site where protandrous parasites that possess a vagina may become inseminated [62,63]. Discovery of juvenile specimens of Capsalidae sp. 13 on pelvic and anal fins and branchiostegal membranes and small, recently matured adults on the branchiostegal membranes of Ne. furcosus suggests a similar migration and habitat partitioning for this taxon on the nemipterid.
Several of the capsalids reported in this study represent new records. Metabenedeniella species are previously reported from oplegnathids, serranids and haemulids [64]. Our report of a Metabenedeniella sp. from a lutjanid (Appendix 1) represents a new fish family and a new geographic record for this capsalid genus. Benedenia elongata was described as Pseudobenedenia elongata from a priacanthid, Priacanthus boops, and two lutjanids, Pr. sieboldii and Arnillo auricilla (now Pr. auricilla, see [65]), off Hawaii [49]. Although two specimens of Pr. auricilla were studied in the present investigation, B. elongata was not recorded. It has, however, been recorded from three new host lutjanid species, Et. carbunculus, Et. coruscans and Pr. argyrogrammicus, each of which represents a new host and geographic record for B. elongata. Like deep water lutjanids, priacanthids can also occur in deep water and it seems as though B. elongata has relatively low host specificity among several deeper water fish species in these two families.
Lagenivaginopseudobenedenia etelis was originally described from Et. carbunculus off Hawaii by Yamaguti [66]. While 16 specimens of Et. carbunculus from New Caledonia were studied, La. etelis was not reported but we did record it from Et. coruscans, a new host and geographic record for this taxon (Appendix 1). As suggested for B. elongata, La. etelis may also exhibit relatively low host specificity and infect several deep-sea lutjanids. Further sampling may indicate whether La. etelis is specific to species of Etelis or whether this capsalid can also infect Pristipomoides species.
As far as we are aware, there have been no published reports of Trilobiodiscus lutiani since its original description [67] although it does occur on the type host, Lu. argentimaculatus in north Queensland (I.D. Whittington, unpublished). The present report of T. lutiani from New Caledonia is a new geographic record (Appendix 1).
For the same sampling limitations presented above, it is possible that specimens of Anoplodiscidae known from external surfaces of nemipterids on the Great Barrier Reef ( [68]; I.D. Whittington, unpublished) may have been overlooked in the present study.

Digenea
The total minimized number of taxa was 33, with 21 species identified at the species level (SLIPs).
The dominant digenean family is the Cryptogonimidae. Members of this family and of the Acanthocolpidae and the Didymozoidae utilise fishes as second intermediate hosts [69], indicating that this component of the lutjanid (and related families) diet is a major source of its digenean fauna. The other digenean families utilise a wide range of invertebrates, often crustaceans, as second intermediate hosts, and occasionally lepocreadiids and opecoelids also use fishes [69]. Some hemiuroids (e.g. Lecithochirium) have interpolated a third intermediate host, a fish, into their life-cycle [70]. Considering the importance of fishes in the diet and infection of lutjanids it is, perhaps, surprising that we found no members of the common family Bucephalidae, which also utilises fishes as second intermediate hosts.
Both acanthocolpid species are reported from Aprion virescens, and are known from the Western Pacific, from Hawaii to the Great Barrier Reef [74,75,80,81].
The Opecoelidae is a difficult group, with many similar species described. Of the four SLIPs one, Hamacreadium mutabile, is a cosmopolitan parasite reported in many lutjanid species in the Atlantic, Indian and Pacific Oceans [82]. This is one of the few opecoelids known to utilise fishes as its second intermediate host [83]. Of the other species, Neolebouria blatta is reported only from New Caledonia, N. lineatus only from southern Western Australia, and Macvicaria jagannathi from the Bay of Bengal [84][85][86][87][88].
The Transversotrematidae includes a single species, which could be identified at the species level. Morphologically and biologically this form agrees with the variable species Transversotrema borboleta, reported from chaetodontids and lutjanids (including Lu. kasmira) from the northern and southern GBR [89]. The species includes 3 genotypes which are not partitioned to different host families, but only genotype 'G2' is reported in Lu. kasmira.
Didymozoidae include several records of unidentified juveniles from coral and deep-sea lutjanids and nemipterids; juvenile didymozoids are found in the intestine of most marine tropical fish [7,8,90,91] and the present records are not surprising. Adult didymozoids include a relatively abundant long filiform form found under the scales of the deep water Et. carbunculus.
Two of the other SLIPs are widespread in the Atlantic and Indo-West Pacific Region, i.e., Ectenurus trachuri, Prosogonotrema bilabiatum (see [92,93]). E. trachuri is mostly reported in carangids, but P. bilabiatum is a common parasite of lutjanids. Allobacciger macrorchis (also known as Monorcheides macrorchis) is reported mainly in the Indian Ocean, but also Japan [77,94,95]. Lepidapedoides kalikali is known only from Hawaii, Japan and Palau [74,96]. Tergestia magna is reported from emmelichthyid fishes from the waters off southern Australia [97].

Cestoda Bothriocephalidea and Tetraphyllidea
For these two cestode orders, only larvae were found, and species identification is not possible.
Bothriocephalideans were represented by larvae found in four species of fishes. They were especially abundant and highly prevalent in Ne. furcosus. White, flattish larvae, about 1 cm in length, were found in the abdominal cavity and in almost all organs of this fish, including the intestinal lumen. Sometimes larvae were visible when the fish was caught, because they were protruding in the region around the anus, with half of their body buried in the flesh. We interpret these larvae as bothriocephalideans due to the lack of morphological characters.
Tetraphyllideans include small forms found in the intestinal lumen of various fishes, both from coralassociated and deeper water fishes. It is possible that these include lecanicephalideans. A detailed morphological examination was not performed.

Cestoda Trypanorhyncha
Only larvae were found. All these trypanorhynchs have their adults parasitic in sharks and are probably transmitted to their final host by predation. Seven species of larval trypanorhynch cestodes, all from coral-associated fish, were identified at the species level on the basis of the armature of their tentacles.
Trypanoselachoidans [98] included four species, found as larvae in cysts in the abdominal cavity. The otobothriid Otobothrium mugilis was found only once in the nemipterid Nemipterus furcosus. The lacistorhynchids Callitetrarhynchus gracilis, Floriceps minacanthus and Pseudolacistorhynchus heroniensis were found in a lutjanid and a nemipterid; in New Caledonia, C. gracilis has already been recorded from 4 serranids and 1 lethrinid, F. minacanthus from 6 serranids and 1 lethrinid, and Ps. heroniensis from 7 serranids and 2 lethrinids [7,8] and show low host specificity at the larval stage. C. gracilis has already been found in Lu. vitta in Indonesia and F. minacanthus has already been found in Ne. furcosus in Indonesia [99]. Ne. furcosus is a new host record for C. gracilis, Lu. vitta is a new host record for Ps. heroniensis and Ne. furcosus is a new geographical and new host record for O. mugilis.
Trypanobatoidans [98] included three species of the tentaculariid genus Nybelinia, found free in the intestinal lumen of the nemipterid Ne. furcosus. Ny. goreensis has already been found in Ne. furcosus in Indonesia [99] and in New Caledonia in 2 lethrinids [8]. Ne. furcosus is a new geographical and new host record for Ny. indica and for Ny. queenslandensis.
It is likely that this high rate of records in Ne. furcosus is simply a consequence of its higher sampling (see also nematodes). In addition, cysts were repeatedly found in small numbers in the abdominal cavity of several deep water lutjanids. Although there is no direct evidence that these represent trypanorhynchs because they never contained larvae, the cysts (Figure 4), about 1 cm in length, were similar to sterile trypanorhynch cysts. We hypothesize that these larvae have a very long development time, perhaps related to the relatively cold deep water environment, and that the fish collected by us were too young to harbour fully developed larvae; deep water lutjanids have long life spans (30-40 years) [100].

Nematoda
The minimized number of taxa is 17, including 9 species identified at the species level.
The Anisakidae is represented by both larvae and adults. Larvae were encapsulated on the surface of organs or free in the lumen of the intestine; a few were occasionally identified at the genus level (Anisakis sp., Hysterothylacium sp., Terranova sp.), but most were identified only at the family level. Most fish species, whether coral-associated [7,8] or deep-sea, harbour these larvae, sometimes in very high numbers.
Adult anisakids included two newly described species of Raphidascaris (Ichthyascaris), one from the coral- associated nemipterid Ne. furcosus, and one from the deep water lutjanid E. coruscans, and several additional specimens were found in other fish species and could not be identified at the species level.
Cucullanids were found only in deep water lutjanids, with Cucullanus bourdini in two species of Pristipomoides, and Dichelyne etelidis in two species of Etelis.
Unidentified gnathostomatids and physalopterids were occasionally found only in Ne. furcosus and probably illustrate the fact that this fish species was more intensively sampled that others, thus providing a number of records of parasites with low prevalence (see also trypanorhynchs).
Gonad-parasitic philometrids were found as two new species from Lu. vitta, but never in other fish species. Lutjanids and nemipterids are known as hosts of a few species of gonad-parasitic philometrids in other regions, such as the North Pacific, Indian and Atlantic Oceans [101,102].
Capillariids included a newly described species, Pseudocapillaria novaecaledoniensis, from a deep water lutjanid, but no other records were made.
Trichosomoidids included two species of Huffmanela, H. branchialis and Huffmanela sp., from the gills of two nemipterids. Interestingly, both species recorded in 2004, were never found again, despite intensive sampling of Ne. furcosus; their prevalence is probably very low, and their initial discovery in a small sample should be attributed to chance. Similarly, no Huffmanela species was recorded from more than 500 serranids examined [7], but a new species was described later from a serranid [103]. Tissue-dwelling trichosomoidids are characterised by two opposing features which probably balance each other out -very low prevalences and extremely high numbers (millions) of eggs in the few individual fish infected.

Hirudinea, 'Turbellaria' and Acanthocephala
Specimens of these three groups were rarely found.
Juvenile specimens of the piscicolid leech Trachelobdella sp. were found on the gills of a lutjanid and a nemipterid. Leeches of this genus are known from other lutjanids [104,105].
Cysts containing an unknown turbellarian were found rarely on the gills of Ne. furcosus; these were orange, abundant but with a very low prevalence, and were not studied in detail. Parasitic turbellarians are rarely found on coral reef fish [106,107].
An unidentified acanthocephalan was found once in the intestine of the deep water Et. carbunculus; this record constitutes a small addition to our very poor knowledge of the acanthocephalans of New Caledonia fish [6][7][8]108].

A numerical evaluation of parasite biodiversity in lutjanids and nemipterids
In presenting our results, we used the same methods as in previous similar papers of this series on serranids and lethrinids [7,8].
Our results (Appendices 1 and 2) include a number of parasite identifications, but the level of taxonomic identification varies greatly. Table 1 details the number of HPCs found in each fish species, and indicates how many fish specimens were examined; it was compiled by counting each parasitological record (i.e. each line in Appendix 1) as a host-parasite combination (HPC).
The number of HPCs differs from the actual number of different parasite species, for two reasons (a) a parasite species present in several hosts is counted as several HPCs; and (b) HPCs in Table 1 enumerate records which vary widely in systematic precision, and may designate, in a decreasing order of taxonomic precision: -Species-level identified parasites (SLIPs); these have full binomial names, and we do not include 'cf ' identifications within them. -Species-level identified parasites which have not yet received a binomial (such as the numbered copepods Hatschekia sp. 17-23, Diplectanum 'cf.' species, and numbered capsalids). These represent valid, independent species but a comparison of the presence of these species in other localities or in other fish species will not be possible until the parasite species are formally identified, described and names are published. -Parasite species identified at the generic level only, but which probably represent only a single species (examples: several digeneans). -Parasite species identified at the generic level only, but for which we already know that they represent several species (example: several ancyrocephalid monogeneans). -Parasite species identified at the family or higher level, for which we know that abundant biodiversity is hidden within this HPC. This includes unidentifiable larvae such as gnathiid isopods, anisakid nematodes, didymozoid digeneans and tetraphyllidean metacestodes. We estimate that these may represent a total of about 50-100 species.
Only species-level identified parasites with binomial (SLIPs) allow valid comparisons between localities and fish. Table 1 includes all results, but some fishes were only studied superficially and their inclusion in further calculations would severely bias the results; for this reason, Table 3 was constructed only from fish species of which at least several specimens have been studied. Table 3 also provides a comparison with the lethrinids and serranids of New Caledonia, based on previous results [7,8]; this comparison will be discussed below. Of course, we are aware that the number of fish studied is generally too low to provide truly significant results on parasite biodiversity, but at least these results allow comparison with other fish families studied using the same methods in the same area. Caesionids are not included in Table 3 because only a single species was involved.
For lutjanids (Table 3), the total number of HPCs was 131, and the number of different parasite species identified at the species level (SLIPs) was 40. For nemipterids (Table 3), the total number of HPCs was 42, and the number of SLIPs was 15. As usual, in addition, indistinguishable larval taxa probably correspond to a high number (50-100?) of different species, but cannot be differentiated on the basis of morphological studies. Table 3 includes evaluations of these numbers as means per species of fish. The main results for all lutjanids (reef-associated and deeper sea) were that 10.92 HPCs were found per fish species, with 3.33 SLIPs (identified with binomial) per fish species. For nemipterids, the results were 14.00 HPCs and 5.00 SLIPs per fish species.

A comparison between reef-associated and deeper-water lutjanids
Our results provide an opportunity to compare parasite biodiversity in reef-associated and deep water fishes. It is widely accepted that fishes in deeper waters have a lower HPCs, number of host-parasite combinations; SLIPs: number of species-level identified parasites (the same parasite species found in different hosts is counted a single time; hence differences with numbers of SLIP-HPCs in Table 1). Source of data for lutjanids and nemipterids: Table 1 and Appendix 2, this paper; for lethrinids and serranids, Table 3 in [8]. *: species with only anecdotal data (see Table 1) were excluded from these calculations.
parasitic diversity than surface fishes [109][110][111][112]. However, in comparative studies, fish species from the deep-sea are generally from orders (e.g. Gadiformes, Ophidiiformes, Notacanthiformes) which are different from the orders of surface fishes (e.g. Perciformes); in contrast, our study allows us to compare fish from the same family, the Lutjanidae, from both environments. Moreover, collection areas were very close and adjacent, with deeper-water fishes collected from just off the barrier reef along the outer slope, i.e. less than one kilometre away from the barrier reef and the lagoon [19].
Recently, a molecular study demonstrated that monogeneans of groupers tend to widen specificity when they infect fish from the outer slope, in comparison to lagoon fish where they are strictly specific [23]. Table 3 shows that the number of HPCs in reefassociated lutjanids was 10.13 per fish species, compared with 12.50 (123%) in deep-water lutjanids. The number of SLIPs in reef-associated lutjanids was 3.13 per fish species, compared with 3.75 (125%) in deeper water lutjanids. These figures suggest that parasite biodiversity was higher in deeper-water fishes than in reef-associated fishes, a highly unexpected result.
However, we identify four biases which diminish the validity of this comparison: (a) Depth bias. Estimates of parasite biodiversity in deep-sea fish [109][110][111][112] generally concern fish from deeper seas (i.e. 1000 m vs 100-250 m) and from other fish orders than those studied here.
(b) Size bias. Most deeper water fish examined, especially Etelis spp., were large fish, usually in the metre range size, while reef lutjanids were smaller, usually 10-30 cm in length [113]. It is known that parasite biodiversity increases with the size of hosts [8,114,115]. (c) Monogenean bias. Coral-associated lutjanids harbour a number of small ancyrocephalid monogeneans, of which a large proportion has not been described yet [45]. (d) Human bias. The high number of parasites identified at the species level in deeper water fish is probably related to the scientific interest they elicit in scientists. Systematicians like to describe rare parasites from rarely examined fish! No particular direction was given to participants of this study to balance their descriptive effort between reef and deeper water fish.

A comparison with lethrinids and serranids
Data on parasite biodiversity, compiled using the same methods at the same location, are available for two other families [7,8]. Table 4 compares parasite biodiversity in four families of reef-associated fishes, the lutjanids and nemipterids (present study), and the serranids and lethrinids [7,8]. Table 4 also compares results for data compiled from fishes with variable sample size (anecdotal collections excluded) and for fishes with significant sampling (i.e. >25 individuals). For the latter, the numbers of HPCs for lutjanids and nemipterids per fish species are 20.00 and  Table 4 includes results obtained by pooling the data for all four families of fish. Results from 13 species of well-sampled reef-associated fishes (only n > 25 or n > 30 sampled individuals according to family), represented a sampling effort of almost 1,000 fishes (382 serranids, 329 lethrinids, 42 lutjanids, 239 nemipterids, total 992). The sampling effort and person-day time represented by these figures, which include collection, preparation, precise identification of parasite specimens, and curation in recognized collections, has no equivalent in the literature for reef-associated fish and all of their parasites (although similar efforts were devoted to digeneans only [4,116,117]). The number of HPCs is 20.77 per fish species, and the number of SLIPs is 9.62 per fish species. Many precautions were taken in these studies to minimize the number of taxa, and larval taxa (which are difficult to identify) certainly represent a significant additional biodiversity. We thus consider that the total result of ten species of parasites per reef fish species is a strict minimum, and that the real number is probably double or triple this estimate. All numbers here concern only macroparasites; the addition of Myxosporea, if they were known, would significantly enhance all results, with probably 2-3 additional parasite species per fish species [118,119].
Our published [6,81,[120][121][122][123][124] and unpublished data on other families of fish suggest that this number of 10 species of parasites per fish species is generalizable to other families of fish, at least those with similar "average" size (10-40 cm), since parasite biodiversity depends upon the size of fish [4,8].
As found for serranids and lethrinids [7,8], the literature includes very few extensive lists of parasites from lutjanids and nemipterids in the Pacific. The checklist of parasites of Heron Island [44] includes a single lutjanid species in common with the present study, Lu. fulviflamma, with a single parasite record, and no nemipterid species in common. This precludes any biogeographical comparison for lutjanids and nemipterids, as was found for serranids and lethrinids [7,8].

Consequences for coextinction of parasites of coral reef fish
The word coextinction was coined by Stork and Lyal (1993) [125] to express that as a host species becomes extinct, so does one or more species of parasites, and was redefined by Koh et al. (2004) [126] in a slightly broader sense as "the loss of a species (the affiliate) upon the loss of another (the host)".
Knowing that parasite species are more numerous than non-parasitic species [127,128], it follows that coextinctions are more numerous than extinctions [129]. Dobson et al. (2008) [130] estimated that 3-5% of helminths are threatened by extinction in the next 50-100 years. However, Dunn (2009) [129] mentioned that there is no well documented case of the coextinction of a vertebrate parasite. Rózsa (1992) [131] pointed out that even a decrease in numbers within a host population, without the danger of extinction, could jeopardize the survival of certain parasite species. Koh et al. (2004) [126] calculated a risk of extinction of 593 species of monogeneans associated with 746 endangered species of fish. Justine (2007) [132] remarked that such a prediction underestimated the number of parasites in fish in rich ecosystems such as coral reefs. For Moir et al (2010) [133], this discrepancy highlights how biogeographic variation and knowledge gaps in dependant species biodiversity may lead to biased estimates of coextinction risk.
Should we be concerned by the extinction of parasites? This is hard to defend to the general public, because "parasites tend to lack charisma" [129] and many bloodsucking parasites transmit diseases [134]. However, parasites play a major role in the balance of populations and the evolution of their hosts [114,129,130,135], and, as such, are an important and irreplaceable part of biodiversity and ecosystems.
The numerical evaluation of parasitic biodiversity in coral reef fish provided in the present study allows a more precise prediction of the risk of coextinction if a coral reef fish species becomes extinct, or simply has its population decreased [133]. Coral reefs are threatened across the planet [136][137][138][139][140] and special threats exist in New Caledonia [141,142]. Our results suggest that extinction of a coral reef fish species of average size would eventually result in the co-extinction of at least ten species of parasites.

Conclusions
As surprising as it might seem for studies mainly based on lists of parasites and morphological identifications, the present paper and our two previous similar papers [7,8] are pioneering works in the field of biodiversity of parasites of tropical coral reef fishes. Our main discovery of a parasitic biodiversity at least ten times higher than fish biodiversity has potential implications in the evaluation of loss of biodiversity when a coral reef fish species is threatened or becomes extinct.

Methods
Methods used in this paper are essentially the same as for the two previous papers of this series [7,8] and for brevity are not repeated here. For parasite collection, we used two methods targeting two sets of organs, designated as "gills" and "abdominal organs". We generally used the "gut wash" method [143] but in some circumstances we used a simplified method [144]. Full details and possible methodological flaws were discussed previously [7,8]. In addition, we soaked a few fish in saline in order to collect surface monogeneans. A high number (239) of fork-tailed threadfin breams, Nemipterus furcosus, were examined, including specimens examined for research and specimens examined by students during practical courses at the University of New Caledonia.
Parasite specimens, generally collected by J.-L. Justine and his team, and sometimes by visiting colleagues, were forwarded to respective specialists: I. Beveridge Hirudineans were examined by E. Burreson (College of William and Mary, Gloucester Point, Virginia, USA); a few monogeneans were examined by L. Euzet (Sète, France); some anisakids were identified at the generic level by S. Shamsi (Charles Sturt University, Wagga Wagga, NSW, Australia). Gills of several lutjanids, prepared with hot water and formalin, were examined by D. C. Kritsky (Idaho State University, Pocatello, Idaho, USA) and monogeneans were described [45]. Sometimes external isopods were brought by fishermen and provided "anecdotal" collections.
The names of cestode orders and suborders follow Khalil et al. [145] updated by recent work [98,146,147]. Polyopisthocotylean and monopisthocotylean monogeneans are treated as two independent groups, because monophyly of the monogeneans is not established [148][149][150][151][152]. However, since polyopisthocotyleans were rare, results for both groups were often pooled. Monogeneans sometimes included in the Dactylogyridae are here considered as members of the Ancyrocephalidae [153,154]. Many specimens have been deposited in recognized collections (Appendix 3); other specimens under study are still in the collections of the various authors but will be eventually deposited in the collection of the Muséum National d'Histoire Naturelle, Paris, France (MNHN) and/or in other recognized, curated collections.
Research carried out on animals (fish) was performed in accordance with the ethical requirements of the IRD (Institut de Recherche pour le Développement, France) and University of Adelaide Animal Ethics Approval S-020-2008 for work by I.D. Whittington.